Learning and remembering of spatial patterns by hummingbirds

Hummingbirds' responses to the spatial distribution of their food were examined using arrays of visually identical feeders in the laboratory. Rufous hummingbirds, Selasphorus rufus, learned to visit profitable feeders in square arrays of 64 feeders on a wall in which only half contained food and the only source of information about feeder quality was location. Birds learned quickly and well if profitability was simply patterned. They learned more slowly but eventually reached high performance on complex patterns, especially if feeders were regularly distributed. Memory of pattern accounted for much more of performance on all patterns than either simple movement rules or memory of individual locations. When food distributions were switched to their mirror images after birds had reached asymptotic performance, all birds that had performed well before the switch continued for several trials to visit the feeders they had visited before the switch, lagging in redistributing effort to match the new pattern. It is suggested that in the absence of visible correlates of feeder quality, hummingbirds use coarse-grained memories of the spatial patterning of energetic profitability to guide their foraging. These memories develop with repeated experience of environments, develop more quickly in simply patterned environments and persist despite changes in food distribution. Poorer performance in complex arrays is inconsistent with a fine-grained model in which birds remember point sources of food independently. Persistence of foraging patterns at the cost of significantly reduced gross food intake and severely reduced net intake after pattern switches demonstrates that hummingbirds rely greatly on memory of profitability in guiding foraging; however, their quick recovery to previous performance shows that they quickly modify their memory of places on the basis of new information. These results are compared with some recent field studies of hummingbird foraging in which use of memory could not be reliably demonstrated, or was only one of several mechanisms underlying foraging patterns.