Evolution of Horns in Ungulates: Ecology and Paleoecology

Summary (1) The savanna ungulate faunas of the North American Miocene were broadly similar to those of present‐day East Africa in terms of overall morphological and taxonomic diversity. However, the predominant ungulates of the African faunas are bovids, which possess bony horns that are primitively sexually dimorphic in their occurrence. The predominant ungulates of the North American Tertiary were equids, camelids and oreodonts, which all lacked horns. A limited number of horned ruminants were present, but these were largely Miocene immigrants from Eurasia. Horns were also absent from the large‐bodied herbivores in the endemic faunas of South America and Australia. (2) The absence of horns in equids and tylopod artiodactyls is unlikely to be due to genetic insufficiency. Bony horns were present in brontotheres, which were closely related to equids, and in protoceratids, which were closely related to camelids. Nasal horns were present in one oreodont genus. (3) Studies on living ungulates show that a strong correlation exists between habitat type, feeding behaviour, social behaviour and morphology. It is possible to use the morphological remains of extinct ungulates to reconstruct the types of feeding and social behaviour, and to use the distribution of morphologies and body sizes in a community of mammals, in conjunction with geological and paleobotanical evidence, to reconstruct the type of habitat. (4) The importance of the post‐Eocene climatic changes to the history of mammalian evolution is stressed. Continents at higher latitudes have become increasingly seasonal in terms of temperature and rainfall since the equable global conditions of the early Tertiary. Savanna mosaic were the predominant biome in North America by the early Miocene, and in Eurasia by the middle Miocene. Living temperate‐latitude species of ungulates may not be a reliable guide for the assessment of the interrelationship between behaviour and morphology in an evolutionary perspective, as their behaviour may have been recently adapted to a habitat type that has only been in existence since the Pleistocene. (5) The primitive condition in eupecorans and protoceratids is the absence of horns, with the presence of large sabre‐like canines in the males. The first horned members of these divisions had horns in the males only. Small present‐day antelope, where horns may also be present in the females of the species, are probably secondarily small. (6) Horns were acquired independently in ruminant artiodactyls at least three times, and a maximum number of seven times is not unlikely. In each case, horns first appeared at a critical body weight of about 18 kg, and in correlation with a change in habitat from closed to open woodland. (7) Horns in living ruminants are associated with territorial defence by males holding exclusive feeding and reproductive territories in woodland habitats. Such behaviour in present‐day antelope is correlated with a body size of greater than 15 kg and a folivorous diet. It is argued that horns evolved in ruminant artiodactyls on the adoption of this type of territorial behaviour once the critical combination of body size, diet and habitat type had been attained in their evolution from small, essentially frugivorous, forest‐dwelling animals. (8) Perissodactyls never evolved sexually dimorphic bony horns of the type seen in ruminant artiodactyls. This is because their foraging and digestive strategies necessitate a larger daily intake of food. In a woodland habitat they were never able to adopt a feeding area small enough to make exclusive territory maintenance an economical proposition. Territory holding in male perissodactyls is seen, but under the opposite conditions of habitat to territorial behaviour in ruminant artiodactyls. (9) Study of the morphology and paleoecology of oreodonts suggests that they were woodland herd‐forming browsers with exclusively folivorous diets. They probably had some forestomach fermentation, but did not chew the cud. Similar studies of Tertiary camelids suggest that they were predominantly selective browsers eating herbage at a low level in open country and formed mixed‐sex feeding groups. These combinations of feeding and social behaviour suggest a more open structure of the mid‐Tertiary habitat in North America than in Eurasia. (10) Studies of the behaviour and morphology of living members of the Ruminantia, and of the morphology and paleoecology of their fossil ancestors, suggest that they were primitively tree browsers living in closed woodland habitats. Such habitats were abundant in the Old World, but in limited supply in North America during the Oligocene, where the protoceratids were the only ungulates to parallel the eupecoran type of feeding and social behaviour. South America appears to have had an even more open habitat in the Oligocene than North America, and no parallel to the eupecorans was seen amongst the indigenous ungulates. The radiation of the Bovidae into open grassy habitats in the Pliocene may have been dependent on the immigration of grazing equids into the Old World. (11) I conclude that there was a difference in habitat structure between North America and the Old World during the Tertiary. The food resources in North America were more widely dispersed, and this may have been the result of the trees being more widely spaced. A possible causal mechanism for this was the stable land mass of the North American continent during the Tertiary, resulting in a more continental climate, with a more severe effect of the post‐Eocene seasonality on the vegetation. The faunal record of the two continents also implied a greater density of trees in the Old World. (12) Thus most endemic North American ruminants did not evolve horns because, at the critical combination of body size and diet seen in the evolution of horns in the Old World ruminants, the dispersal of the food resources within the vegetation was too great for an effective home range to be maintained as an exclusive territory. (13) Attention is drawn to the dangers of constructing evolutionary stories about living animals without primary reference to the fossil record to see if the hypotheses are upheld, and of assuming that fossil animal communities can be made to fit models of existing communities.