the spatial distribution of tree cutting activity by beaver (castor canadensis kuhl) and tree availability were measured at two beaver colonies near thunder bay, ontario. beaver maximize the net rate of energy acquisition (e/t) when they minimize the distance travelled from the lodge, their central place, while obtaining a balanced diet. however, beaver must travel through two mediums to obtain terrestrial forage so minimum distance is a function of the relative costs of swimming and walking. a continuum of relative costs was used to generate two contrasting foraging models: 1) swimming equalled the cost of walking and the foraging path was a straight line from the lodge to a tree; 2) swimming was costless and the foraging path was the shortest distance from water to a tree. the models were compared for relative goodness of fit with the observed foraging pattern using chi-square and linear regression goodness of fit tests. the water costless model was the best fit model and empirically supported the implicit assumption of most workers that the pond is the effective central place. in practical terms 'water costless* was judged to be indistinguishable from a 5-10 times advantage for swimming over walking. although experimental confirmation is required, this advantage was considered too large to be explained by energy or time savings and was therefore not an optimization of e/t. it was tentatively concluded that the pond should be viewed as a refuge from predators which probably constrain the relative availability of terrestrial forage to beaver. a test for optimal foraging, given the predation constraint, showed that beaver were maximizing e/t at northbranch pond. at pinetop pond, a relatively old site, beaver expanded their refuge to obtain patches of relatively high food quality. they could have realized higher e/t with perfect knowledge of their site, but apparently they maximized e/t with a patch use strategy that minimized search time. a review of beaver literature discussed optimization processes at four levels: evolution of body form and function, the individual lifespan, yearly, and seasonally. arguments of particular interest were developed for dispersal (population regulation), central place, and the size-distance relationship.