There are two main reasons why this happens. Firstly, there isn't so much food to eat on an island, so smaller animals that eat less will be at an advantage, and able to have more offspring that survive to have offspring of their own. But, while the problem of a limited food supply is more obvious on a smallish island, even on a large continent, the supply is never likely to be inexhaustible. In which case, why do large animals exist in the first place?
That, of course, is where the second reason comes in: predators. Being large is a protection against being eaten; lions and tigers, for instance, rarely eat elephants, rhinos, or hippos. Large predators, being at the precarious pinnacle of the food chain, and present in smaller numbers to start with, find it even harder to survive on small islands than herbivores, and they often simply die out. That removes the need for herbivores to avoid them: there's no real need to be too large for a lion to eat if the biggest thing you'll ever face is a fox, anyway.
So, with no pressure to stay large, and a reasonably strong pressure to become smaller, animals on islands often shrink over the course of thousands or millions of years. Many dramatic examples are known from the Mediterranean, where changing sea levels have caused many islands to form and re-merge over the the last few million years. Perhaps the most famous example are the dwarf elephants that once lived on many Mediterranean islands, but are now long extinct. There were a number of different species of these since the ones living on, say, Malta, had difficulty breeding with those on places like Corsica.
Slightly less well known is the phenomenon of insular gigantism. In this phenomenon, small animals, such as mice, trapped on islands become larger over the course of evolution. This at first seems counter-intuitive: if large animals get smaller on islands, why do small ones get larger? After all, they're all on the same island, so aren't they facing the same pressures?
Well, not necessarily. For one thing, if you're the size of a mouse, an island has to be pretty darn small before you're in much danger of running out of food. Furthermore, while the large predators may have gone, the smaller ones might not have, so there's an advantage to being too large for them to kill easily. Besides which, there are many other reasons why being large might be handy - just as a for instance, larger males may find it easier to compete for mates, and larger females may be able to raise more, or fitter, young.
There is, naturally enough, a limit to this. If nothing else, if you get too large, you'll have the same problems that the big animals do, and have to shrink. There may also simply not be enough time to change too radically from your original form, especially if that would require some major change in your diet. So these "giant" island animals aren't necessarily all that big in the grand scheme of things - just a lot bigger than their mainland relatives.
One Mediterranean island on which we have particular evidence for insular gigantism is Gargano. Today, Gargano is the spur on Italy's boot, the small rounded peninsula that projects into the Adriatic Sea a little way north of the heel. But, in the late Miocene, between about 12 and 4 million years ago, Gargano was an island, separated from the Italian mainland by the high sea levels of the time.
The species living there would not have been entirely in isolation, not least because the island did not exist throughout the entire time - at the very least, it must have rejoined the mainland during the Messinian Salinity Crisis around 5 million years ago, when Mediterranean sea levels dropped dramatically for hundreds of thousands of years. Even so, it became an island again following the devastating Zanclean Flood that brought the Crisis to an end, before finally returning to its current status as a peninsula during the Ice Ages. (The Salinity Crisis is a topic I will return to on this blog at some point - it's a crucial turning point in the history of Europe during the Age of Mammals).
The peninsula today, and the island that preceded it, are a typical "karst" landscape, dominated by limestone. Because limestone is soluble, on sufficiently long timescales, these landscapes are rich in caves, pot-holes, and the like. In particular, on Gargano, there are a number of deep cracks in the rock, that developed during the Miocene or Pliocene and have long since filled in with red clay sediment, and it is these ancient fissures that many of the fossils have been found.
This does, unfortunately, create something of a problem with dating, because the fossils are not lying in their native bedrock, but in fissures that have been extensively re-worked over time. As a result, it isn't terribly clear quite when the unusual animals of the island actually lived there, and there are a number of different theories to account for the precise timing. The best we can do is to use the fossils we do have to at least work out a rough sequence of events, even if we can't quite put dates to them.
Many of the fossils found on this tiny island (it may have been about the size of, say, Rhodes) are known nowhere else in the world. While, of course, there were a number of animals that stayed small, almost every kind of small mammal local to the area developed at least one out-sized species. Particularly common were giant hamsters, Hattomys, and large rats, Mikrotia, were so abundant that the entire collection of fossil species is known simply as the "Mikrotia assemblage". There were also rabbit-like animals, Prolagus, that may, or may not, have been close to modern pikas, although, while large, these weren't really "giants" relative to their modern kin.
The largest dormouse ever known to have lived, Stertomys, also lived on the island. For the most part, we can identify it only from its teeth, and we don't have anything in the way of a complete skeleton. As a result, we don't really know quite how big it was. But with each of its teeth about 3 cm (over an inch) across, we can safely say that, at least for a dormouse, it was gigantic. As a comparison, the common dormouse found in Britain is only about 9 cm long in total, not counting the tail, although the edible dormouse, the largest living species, is about twice that.
Indeed, about the only local animals that never seemed to have evolved larger forms at all were the shrews, which seem to have been quite content to retain their usual tiny size. Interestingly, and supporting the claim about a lack of predators being a factor in insular gigantism, about the only mammalian carnivore found among the fossils is an otter. The species in question wasn't unique to the island, and presumably swam across from the mainland without much difficulty, and only one has actually been found, suggesting that they were hardly common even then. (Update: as noted in the comments below, though, this doesn't mean that predatory birds couldn't also reach the island).
Two types of local animal, however, are particularly interesting. One is Hoplitomeryx, which was a... well, actually there's nothing quite like it around today. It was certainly a ruminant, and probably more related to deer than anything else, although it's a bit hard to tell. Unlike deer, however, it had horns, not antlers. And it had five of them.
Not content with having five horns - an unusual number, to say the least - it also had large sabre-like canines, as musk deer do today. As a herbivore, it presumably used the canines to fight, which is odd, because you would have thought that five horns would have been quite enough to do that effectively. After all, it wasn't even as if it had anything worse than an enraged otter to fend off - any fighting it was doing was most likely for access to mates.
Hoplitomeryx came in a range of different sizes, which may represent different species. Given that the animals apparently started out quite large, some of these species seem clear examples of insular dwarfism, but the largest were the size of an elk - albeit much slimmer. Perhaps the need to aggressively compete for mates overcame the pressure to shrink, in at least some forms. Incidentally, while its difficult to know for sure, both large and small forms seem to have lived on the island at the same time.
One possible explanation is that it wasn't an island at all, but rather an archipelago, with different forms on different islands. The downside of that argument, though, would be that the individual islands would be even smaller, only increasing the evolutionary pressure not to be the size of an elk.
The other especially odd animal from the island is Deinogalerix, the "giant hedgehog". Unlike the giant dormice, we do have fairly complete skeletons of these animals, and so can say just how giant we're talking here. There is some variation in size between the different fossils (again, they're probably different species), but the very largest were apparently about 60 cm (two feet) long, making them the size of a smallish dog.
For a hedgehog, that's pretty huge - although its worth noting that its a good deal smaller than, say, a large porcupine. While living hedgehogs feed almost entirely on invertebrates such as insects, spiders, and earthworms, it's likely that the largest giant hedgehogs could have eaten small mice or shrews, taking over the predator niche left empty by the absence of things like wolves or wildcats.
Disappointingly, while it's commonly called the "giant hedgehog" I have to confess that it's not really a hedgehog at all. It is, to be fair, a member of the hedgehog family, and that's why it's listed as such. But, from a number of features of the skeleton, not least the fact that it has a head about half the length of the rest of its body, it belongs to the half of the hedgehog family that doesn't have the 'true' hedgehogs in it.
Instead, it appears to be a kind of moonrat, an animal that today only lives in Southeast Asia. Moonrats are essentially hedgehogs with unusually long heads and no spines. For the benefit of Americans, they look kind of like opossums, but with longer snouts, and for Europeans... well, if you've not seen an opossum, you may to have to imagine a sort of rat-sized shrew. Or, as in this case, a two foot long shrew (plus tail!)
And all of these strange animals lived, so far as we know, only on one small island off the coast of Italy.
[Picture by "Ghedoghedo", from Wikimedia Commons]