It can be argued some fantasy ecology is in the realm of the fundamentally bizarre. So let's try to fix it a bit.
In all ecosystems energy flows, with a big loss (about 90%) between levels. (Insects tend to be more efficient, for those of you into Evil Bugs, but of course there are other anatomical problems with Big Evil Bugs). Theoretically this would lead us to assume that with food chains long enough like
Grass - Rabbit - Human - Troll - Giant - Dragon
the Dragon would be living on such a minute amount of the available energy it would be in serious trouble. In practice, however, it seems as the fundamental constraint will not be on energy flow but in ecosystem size, assuming the ecosystem is at least marginally productive. (For comparative numbers of ecosystem productivity, check the old posts on Planetary Productivity on this LJ, though the Topic Index). That isn't to say food chains are unlimited in length, to the contrary. I think the chain above is at least one step too long, something should go.
A small enough ecosystem, like that fairly small forest, wouldn't be able to support a dragon population (or giants or maybe even trolls), there simply isn't enough energy and nutrients to go around. But if the ecosystem goes quite unproductive (Productivity: 0 in the terms of the already referred-to posts), energy flow probably still limits us, which is why big, mighty and scary predators of deep dark caves, glaciers and barren deserts are unlikely. So our terrible monsters of a fantasy world are more likely to show up in productive environments. Dragons may be better served on say, a savannah than in the Vast Northern Expanse.
This is of course counterintuitive to the great human fascination of the unknown, where - certain jungles aside - we have inhabited the productive zones and often imagined the terrors to live in the remote unproductive ones, the ones we left for last to explore and live in. Which makes ecology to go into direct conflict with cliché fantasy based upon the very same fascination. And we know who will win. Lets make ecology at least take one or two points even if it loses the game, please?
What however seems directly related to energy and nutrient flow is diversity. Basically, we would tend to have more species in a high-productivity ecosystem. And fewer in the marginal ones.
This means that we really aren't likely to have twenty different cool dangerous predators to choose from (on the Encounter Table or whatever) in a Cold Forest, that sort of diversity is likely only in very high-productive ones, like jungles. So the highest amount of Dragon Species isn't likely to be in the lands of the Boreal Barony, nor will the forests be packed with dozens of different top predators - and this goes for underground ecosystems too. Again, this is somewhat counterintuitive to use because our imagination has dreamt up countless of supernatural critters in our back yard, hobgoblins and goblins and rå and vittra and troll and svartalf and tomte and lord knows what. From a realistic-ecological sense we probably wouldn't have that many types to choose from. Boring to your hack-&-slash hero, maybe, but if he wants diversity in killing beasties he should really go equatorward.
So, what a more realistic fantasy ecology would emphasize is that diversity also applies to "monsters", and thus that the greatest selection of interesting and a bit unexpected challenges for the exploring adventurer should be in productive environments. Like, you know, jungles.
This is not to say that energy considerations aren't important in ecology of a sample species, like say, dragons. For a predator of any kind to be successful it usually has to be big and strong and fast enough to get the prey. The problem with dragons and many other fantasy critters is that they are too big. It is a complete waste of energy to be so big if you are going to eat sheep or humans. You don't need to be the size of a garbage truck to be a top predator if you mainly live on that sort of prey. The energy needed to lug around that huge body combined with the fact that you need to hunt many more of these small things to get by is outright detrimental. Traditionally authors tend to blame this and perhaps also the suspicious absence of top predators on extensive hibernation patterns, but it still means the dragon or whatever is in deep trouble when the feeding period is on.
So I'd like to make a case for smaller top predators in fantasy. Smaller dragons, as an example.
Course, the best way to justify huge predators is that the prey is similarly huge. Which likely means an abundance of very large herbivores. This isn't really a problem... ...unless we have a rather historical world with normal horses and cows, and not a megafauna of some kind. So we'd need an alien ecology, to us, maybe in the world as it is, a world of giant cattle and giant deer, or on an unknown, verdant continent (remember - we need a big ecosystem, an isolated valley might not do) where the Big Dragons survive while the smaller ones live in the world humans live in. Alternately we can argue that the doomed dragon species, big as houses, are remnants of an extinct megafauna, top-predators oversized and by long life spans having generations too slow to adjust size downward evolutionary-wise. Perhaps too slow to be saved.