The mesotheres are a possible case in point. They were a group of notoungulates, one of the two orders of native hoofed mammal that had survived into the Late Miocene along with the litopterns. Earlier in the Miocene, they had perhaps looked even less remarkable, digging herbivores of a similar size and shape to wombats. But as the epoch wore on, they gradually increased in size, while still retaining the adaptations to digging, which they would likely have used to get at tasty roots, and possibly for constructing burrows in which to live.
Typotheriopsis is one of the more common Late Miocene mesotheres, being known, for example, from northern Argentina. In life, it probably still looked much like a wombat in general shape, as its ancestors had done, but it was considerably larger, being around a metre (3 feet) long and perhaps weighing up to 33 kg (75 lbs) - about the same as a golden retriever. That's obviously large for a burrowing animal, but not necessarily out of the question. It was noted for large, relatively flattened teeth that suggest it may have been primarily a grazer, and for a pit beneath the eyes that may possibly have held some deep-rooted scent gland.
Although also notoungulates, the toxodonts were not especially closely related to the mesotheres, and were much larger animals with teeth that look suited to feeding on tough grasses. Microscopic analysis of the wear patterns on the teeth, however, seems to tell a different story, with relatively little tough material in their diet, so they may well have been feeding on soft forest leaves or the like.
Several species of toxodont are known from the Late Miocene, including the anagrammatic Xotodon. Trigodon was one of the largest, being probably around the size of a bison. Its powerfully built body would have made it resemble a small rhinoceros in life, and its general shape may not be the only point of similarity. This is because the skull has a prominent lump between the eyes that looks very similar to that seen on actual rhinoceros skulls beneath the horn.
Rhino horns are composed entirely of keratin and do not contain any bone as true horns do, so if Trigodon had anything similar, we would not expect that to have been preserved in the fossils. While this means that we can't know how long the horn was, or any details of its shape, it seems almost certain to have been present, if somewhat further back on the head than is typical of living rhinos.
Perhaps the strangest-looking of the notoungulates, however, were a group called the homalodotheres. Thought to be more closely related to the toxodonts than the mesotheres, they looked like neither, having long legs, with the forelegs slightly longer than the hind ones, giving them a distinct sloping back. Their teeth formed a continuous row, rather than having a gap between those at the front and back, as is typical of other notoungulates, and of many modern herbivores. Even stranger, they had claws, rather than hooves and may well have been capable of rearing up on their hind legs.
The homalodotheres had been around for a long time, although we know relatively little about the early forms. The best-known example is probably Homalodotherium itself, which lives during the Middle Miocene, but the group dwindled as the Miocene came to a close. The last known example is the Late Miocene Chasicotherium, which continued a general trend of increasing size and reduction in the number of teeth. It would have been the size of a large bear and doubtless looked particularly impressive when standing upright.
The reason it probably did this was to reach the leaves in trees that it would otherwise have been able to get to from the ground, perhaps using the claws on its forearms to pull branches down. This, of course, parallels an unrelated group of prehistoric animals, the chalicotheres of North America, Eurasia, and Africa. Indeed, animals from the two groups would have looked rather similar with a key difference being that the later chalicotheres knuckle-walked like a gorilla, while it's more likely (but admittedly, not certain) that Chasicotherium and its relatives walked on the ends of their toes as hooved mammals normally do.
The litopterns and notoungulates were not, however, the only large herbivores living in South America during the Late Miocene. The continent was also home to the sloths, which still lived on the ground at this time, and, if not quite as large as the species of later epochs, were still pretty impressive. The sloths were, in fact, very diverse, with a number of different families recognised and all of them living alongside one another as the Miocene drew to a close. While molecular evidence suggests that this must have included the earliest tree sloths, the fossils that we have consist solely of the much larger ground sloths which presumably lived in environments more conducive to preserving their bones.
The two living families of tree sloth represent the last survivors of two branches of sloth that split apart even before the dawn of the Miocene. That to which the two-toed sloths belong included the mylodontid ground sloths such as Nematherium, noted for having especially strong arms, and the relatively long-legged Eionaletherium, which lived along the shoreline of what is now Venezuela and is estimated to have weighed as much as a ton.
Although once thought to be related to the two-toed sloths, the giant megathere ground sloths now seem to be over on the other side of the sloth family tree. Examples from the Late Miocene include Promegatherium from Argentina, which was notably smaller than its more famous later relatives. The nothrotheres were yet another family, notable for including the very last ground sloth to go extinct, a mere 11,000 years ago.
Pronothrotherium was a typical nothrothere ground sloth of the Late Miocene. It resembled many other ground sloths of the day, being about the size of a gorilla, with feet that curved inwards so that it walked, not on the soles of its feet, but on the sides - an adaptation that seems to have evolved independently at least three times among ground sloths. Like other nothrotheres, it had a comparatively narrow skull, suggesting a picky eater rather than something that indiscriminately shovelled food into its mouth.
For sheer strangeness, however, it's hard to beat its close relative, the sea-sloth Thalassocnus. This lived along the Pacific coast, being known mostly from Peru, although it also lived at least as far south as central Chile. Much of this region would have been a desert at the time, so it's unsurprising that it hasn't produced many other fossils of land-dwelling animals, but the sea-sloths apparently got around the lack of terrestrial vegetation by heading out into the surf to feed on water plants. Several species are known, and they became more aquatic as time went on, with the later species specialising entirely on aquatic vegetation and diving to a greater depth to do so.
Thlassocnus was smaller than the great ground sloths, perhaps weighing around 80 kg (180 lbs), closer in size to a chimpanzee than a gorilla. Like many living aquatic mammals, it had unusually compact bones, helping to weigh it down in the water, providing a sort of ballast that would have given it superior control and manoeuvrability while swimming. The shape of its limbs also confirms its aquatic lifestyle, with the arms having wide, clawed paws ideal for paddling and comparatively weak hind limbs. The skull had nostrils comparatively far back, as often seen in aquatic mammals, and a jaw shape that implies thick and muscular lips similar to those of an elephant seal. In addition to paddling, it likely walked along the bottom of shallow coastal waters, feeding on seagrasses in a similar manner to living manatees.
Before leaving Miocene South America, however, I have to turn to those animals with a rather less herbivorous diet...
[Photo by Jonathan Chen, from Wikimedia Commons.]