Mechanical defenses in leaves eaten by Costa Rican howling monkeys (Alouatta palliata).

Primate species often eat foods of different physical properties. This may have implications for tooth structure and wear in those species. The purpose of this study was to examine the mechanical defenses of leaves eaten by Alouatta palliata from different social groups at Hacienda La Pacifica in Costa Rica. Leaves were sampled from the home-ranges of groups living in different microhabitats. Specimens were collected during the wet and dry seasons from the same tree, same plant part, and same degree of development as those eaten by the monkeys. The toughness of over 300 leaves was estimated using a scissors test on a Darvell mechanical tester. Toughness values were compared between social groups, seasons, and locations on the leaves using ANOVA. Representative samples of leaves were also sun-dried for subsequent scanning electron microscopy and energy dispersive x-ray (EDX) analyses in an attempt to locate silica on the leaves. Both forms of mechanical defense (toughness and silica) were found to be at work in the plants at La Pacifica. Fracture toughness varied significantly by location within single leaves, indicating that measures of fracture toughness must be standardized by location on food items. Monkeys made some food choices based on fracture toughness by avoiding the toughest parts of leaves and consuming the least tough portions. Intergroup and seasonal differences in the toughness of foods suggest that subtle differences in resource availability can have a significant impact on diet and feeding in Alouatta palliata. Intergroup differences in the incidence of silica on leaves raise the possibility of matching differences in the rates and patterns of tooth wear.

Nested dictionary learning for hierarchical organization of imagery and text

A tree-based dictionary learning model is developed for joint analysis of imagery and associated text. The dictionary learning may be applied directly to the imagery from patches, or to general feature vectors extracted from patches or superpixels (using any existing method for image feature extraction). Each image is associated with a path through the tree (from root to a leaf), and each of the multiple patches in a given image is associated with one node in that path. Nodes near the tree root are shared between multiple paths, representing image characteristics that are common among different types of images. Moving toward the leaves, nodes become specialized, representing details in image classes. If available, words (text) are also jointly modeled, with a path-dependent probability over words. The tree structure is inferred via a nested Dirichlet process, and a retrospective stick-breaking sampler is used to infer the tree depth and width.