The idea of Pleistocene rewilding, even though it annoys the hell out of me, is so interesting in what it implies about ecosystems.
If we accept that North America's ecosystems are "incomplete" or "impoverished" because of the extinction of Pleistocene megafauna, that implies there is a "complete" state of ecosystems. In the absolute sense, of course ecosystems don't ever have a "complete" state, but is it possible for an ecosystem to be relatively incomplete? What does that even mean?
Could an "incomplete" state of an ecosystem be recognizable without knowing what used to exist in that ecosystem, for comparison? Could a researcher tell that they were in an environment where an animal had gone extinct, without any direct evidence of that animal or knowledge of what it was? Who is to say how many taxa of a kind of creatures "should" be in the ecosystem?
Say we accept, then, that North America's ecosystems after the Pleistocene (but before European colonization, which involved intentional destruction) were "complete," in the sense that researchers couldn't detect any obvious "dysfunction," whatever that means.
But 10,000 years, compared with life's history on the earth, are nothing--- the blink of an eye. There hasn't been very much time for entirely new types of animals to evolve.
So it would imply that ecosystems have a LOT of plasticity and ability to re-arrange to absorb shock, and that animals can quickly expand their ranges and change their niches to adapt to the new state of existence.
...this, in turn, implies something strange about the introduction of new animal species to a continental mainland: that "native" and "non-native" animal species probably won't be distinguishably different in their impacts in the long term, because the ecosystem is chaotic and constantly changing to begin with.
Introducing new animals to islands is a disaster, because it's introducing an animal with a niche that didn't exist before at all, such as terrestrial predators or large herbivores. Introducing plants is a disaster in a small and unpredictable sample of cases.
But in the example of horses in North America, the impact could range from positive (horses used to be here, and their extinction "damaged the ecosystem," therefore horses being introduced "fixes" that damage) to neutral (the ecosystem adapted to not having horses very fast, therefore the ecosystem can likely adapt to having horses again very fast). Saying that horses are invasive seems to require us to believe contradictory things: that the ecosystem has changed so much since the Pleistocene that horses no longer belong, and that ecosystems can't adjust to change quickly.
Then, why indeed should we not introduce camels, or cheetahs, or lions?
Well, this is where "Pleistocene rewilding" gets on my nerves: it sees North America as fundamentally impoverished of animals, and at the same time, somehow treats different species of animal as weirdly interchangeable. We don't know if the American lion was closer to a lion or a tiger, and we don't know some important things like its hunting behavior. The "American cheetah" was not any more closely related to the African cheetah than to the cougar, and might not have been a specialized fast runner like the cheetah.
So this might apply to the horse just as well: the species of horse in Pleistocene times might have been so different from today's horse that they don't have the same role in the ecosystem. Well, is it better to be horseless or horsed?
I don't think that introduced species are inherently bad. This isn't a extreme position. Among plants, very few introduced species actually become invasive, and even some of those considered "invasive" are not actually harming the ecosystem in a way that can be demonstrated. I don't think I would recommend the introduction of a plant purposefully, though...or would I? With climate change occurring rapidly, I am in favor of moving species to areas where they can survive.
One philosophy of biodiversity is that the more biodiverse the ecosystem, the more ability the ecosystem has to absorb shock and adapt to change. Introduced species could have a range of potential to adapt different from native species, and could raise the shock absorption potential of an ecosystem. But they would also disrupt existing relationships and cause a shock to the native species that already exist.
Range expansions are an alternative to extinction for some species. We will probably HAVE to consider introducing species to new areas in the future. Well, imagine in the future we put Zebras in Arkansas, and the Zebras outcompeted the white-tailed deer in that area. Is that good or bad? Both species get to keep existing, but the deer's range is a bit smaller. Is the measure of biodiversity more important in a local area or in the world?