Disentangling adaptive evolutionary radiations and the role of diet in promoting diversification on islands

Although the initial formulation of modern concepts of adaptive radiation arose from consideration of the fossil data, rigorous attempts to identify this phenomenon in the fossil record are largely uncommon. Here I focus on direct evidence of the diet (through tooth-wear patterns) and ecologically-relevant traits of one of the most renowned fossil vertebrates-the Miocene ruminant Hoplitomeryx from the island of Gargano-to deepen our understanding of the most likely causal forces under which adaptive radiations emerge on islands. Results show how accelerated accumulation of species and early-bursts of ecological diversification occur after invading an island, and provide insights on the interplay between diet and demographic (population-density), ecological (competition/food requirements) and abiotic (climate-instability) factors, identified as drivers of adaptive diversification. A pronounced event of overpopulation and a phase of aridity determined most of the rate and magnitude of radiation, and pushed species to expand diets from soft-leafy foods to tougher-harder items. Unexpectedly, results show that herbivorous mammals are restricted to browsing habits on small-islands, even if bursts of ecological diversification and dietary divergence occur. This study deepens our understanding of the mechanisms promoting adaptive radiations, and forces us to reevaluate the role of diet in the origins and evolution of islands mammals.


Dataset
As a consequence of a high incongruence between the taxonomic classifications proposed to date to accommodate species (1,2), and given that the vast majority of the dental (especially upper) elements have not been taken into account for systematic purposes, none of the current classifications for Hoplitomeryx has been adopted for this study. Because the material belongs to fissure fillings, the absolute age of the localities is not at all clear. Hoplitomeryx sp. 2 is the species with more abundant sample and largest chronological distribution (supplementary Table S1).

Long-term patterns of tooth wear
Dental mesowear reflects the cumulative wear imposed on teeth during a relatively long period (months, years) of feeding activity as a consequence of the abrasive elements intrinsic to the plants themselves, and also exogenous grit encroaching on food items (3). All available isolated upper molars and tooth rows containing dental specimens were sampled. Although both the lower and upper dentition of Hoplitomeryx has been studied, only the upper one is here considered for analysis following (3,4). It must be also stressed that the lower dentition shows a higher variability in size than the upper one, with probably 5 or 6 or species considering the lower teeth, while 4 ones according to the upper. Unworn, juvenile/senescent specimens and teeth with poor preservation of occlusal enamel surfaces or taphonomic alterations were discarded from analysis, following Fortelius and Solounias (3). In the original formulation of the method (3), cusp apices were classified by two variables, molar cusp shape (MCS; sharp, rounded or blunt, according to the degree of facet development) and occlusal relief (OR; high or low, depending on the relative difference in height between the tip of the cusp and the intercusp valley). However, MCS and OR are not independent categories, since higher relief cusps tend to be sharper than low relief cusps, and cusps with no relief are always blunt. Therefore, a mesowear score (MS) was here computed for each fossil species (that is, individual MCS and OR scores were converted to a single scoring) according to the five-point scoring system proposed by Rivals et al. (5) as follows: a score 0 is given to teeth with a combination of high relief and sharp cusps; 1 to teeth with high relief and rounded cusps; 2 to teeth with low relief and rounded cusps; 2.5 to teeth with low relief and sharp cusps; and 3 to teeth with low relief and blunt cusps. The fact of treating mesowear as a univariate scoring allows for a rapid and easy way to represent dietary data as a single data point. For comparative purposes, the mesowear scores for extant species were converted from original data of Fortelius and Solounias (3). As a baseline sample, a set of 45 extant ungulate (artiodactyl and perissodactyl) species (and 1764 specimens) with well-known diets compiled by Fortelius and Solounias (3) was used for comparative purpose. The dataset was partitioned into browser (N=9), mixed feeder (N=25) and grazer (N=11) species.

Body mass
Body size is fundamental for nearly every aspect of organismal function and important underlaying the diversity of feeding niches in ruminants (6). In general, large ruminants must feed on diets of low quality because they require large quantities of food, and the more abundant plants and plant parts (e.g., stems or twigs) are generally of 3 lower nutritional quality than less abundant, higher-quality parts (e.g., forbs, leaves or fruits) (6). Only adult specimens, in which the length growth is complete, were considered for analysis. Results were compared to body mass estimations obtained for other endemic ruminants (7).

Molar crown height
Hypsodonty is fundamentally an adaptive response to increasing demands for wear tolerance and functional durability as a consequence of the development of more abrasives in a progressively more open and dry-adapted vegetation (8). Accordingly, it yields information about feeding ecology and habitat structure (especially aridity degree) (9). The mean hypsodonty value was calculated for each Hoplitomeryx species

Statistical tests
Hierarchical, complete-linkage (Ward´s method) cluster analyses based on Euclidean distances ( Fig. 2A), and discriminant Canonical Variate Analyses (CVA) ( Hoplitomeryx species in a as in Fig. 3). Specimens not suitable for metrical study were discarded of the CVA.

Supplementary Tables
Supplementary Table S1. Fissure-fillings (considered as fossil localities) and biozones with dental material of Hoplitomeryx.