Although bark eating is quite common among mammals, few species include it as a major item in their diet. The diet of the small squirrel Sundasciurus lowii on Siberut Island, Indonesia, consists mainly of bark. Physical characteristics, nutrient concentrations, and the presence of defense compounds were investigated in bark from 711, 13, and 57 food and nonfood trees, respectively. The results suggest that the squirrels select bark on large, smooth-barked trees with low levels of hydrolysable tannins, whereas concentrations of fat, crude protein, crude fibre, carbohydrate, and calorific value examined in a set of bark samples had no detectable influence on food tree selection. FEEDING ON BARK IS WIDESPREAD AMONG MAMMALS although none has been reported to consume bark to the exclusion of more nutritious items like fruit, herbs, seeds, or grasses. Bark has been recorded as an item in the diet of orangutans (Rijksen 1978), elephants (Hubback 1941, Olivier 1978), rhinoceros (Hubback 1939, Strien 1974), rabbits (Thompson & Worden 1956), squirrels (MacKinnon 1978, Payne 1979, Farentinos et al. 1981), and many species of deer. In the majority of cases, however, no attempt has been made to analyze the physical or chemical features important in bark selection. A study of the ecological separation of three species of diurnal squirrels on Siberut Island, Indonesia (J. Whitten 1981), found that a major food type in the diet of one small species, Sundasciurus lowii (total length, 23 cm), was bark (an average 49% of stomach contents and over 90% of food trees). It was never observed to eat fruit, and stomach analyses revealed that animal remains occupied an average of only 39 percent of the stomachs' volume. S. lowii does not strip bark from large areas in the same way as do common red and grey squirrels, but removes only small flakes, even from trees whose bark would easily strip. Bark is not carried away from the tree and used in nest building, nor has any evidence been found that this behavior is associated with intraspecific aggression. None of the food trees (either trees on which we saw S. lowii feeding or those whose bark showed clear signs of squirrel feeding) produced sap, and no evidence pointed to squirrels' seeking exudates to feed upon (cf. I Received 7 November 1984, revision received 12 December 1985, accepted 8 January 1986. 2 Present address: School for Resource and Environmental Studies, Dalhousie University, Halifax, Nova Scotia, Canada. Lacher et al. 1984). Thus, all bark removal seems to be associated with feeding. This paper describes an analysis of bark and bark eating by S. lowii. METHODS AND MATERIALS Diurnal squirrels were studied on Siberut Island, 100 km west of West Sumatra, for a total of 12 mo during 1976 and 1977. The study area has been described by J. Whitten (1981) and A. Whitten (1982a, b), and the general methods of squirrel observation have been described by J. Whitten (1981). In order to investigate the selection of bark food trees by S. lowii, the following data were recorded from all trees of at least 5 cm DBH in four 0.25ha plots and from all known food trees: species, girth, tree height, canopy height (distance from lowest major bough to topmost leaves), canopy width, bark roughness on a scale of 1 (smooth) to 7 (very rough), thicknesses of the outer bark (that which could be split away from the inner bark) and of inner bark, and ease with which the outer bark could be split away from the inner bark (easy or difficult). In addition, signs of feeding on bark (Fig. 1) and presence of moss, lichen, and climbers were estimated on a scale of 0 (none present) to 3 (very abundant). Similarly, the ease of access for a squirrel from the surrounding vegetation was judged on a scale of 0 (impossible without coming to the ground) to 3 (possible with great ease). Bark samples were dried initially in an oven over an oil lamp for 2 days, after which the samples were taken to the mainland to be dried in an open gas oven for 2 more days. Although we assumed that all signs of bark feeding were made by S. lowii, other species may have contributed. BIOTROPICA 19(2): 107-115 1987 107 This content downloaded from 157.55.39.247 on Wed, 27 Apr 2016 06:49:50 UTC All use subject to http://about.jstor.org/terms