Returns to emptied cache sites by Clark's nutcrackers, Nucifraga columbiana: A puzzle revisited

The cache recovery behaviour of Clark's nutcrackers was investigated with a new, more refined technique in which nutcrackers were presented with small clusters of test holes to search during recovery sessions. The first experiment evaluated the usefulness of the cluster technique. The birds performed extremely well, making many fewer errors than would be expected by chance and avoiding clusters that did not contain cache sites. The second experiment employed the cluster technique to investigate the tendency of nutcrackers to revisit old cache sites emptied earlier in recovery. During the first half of recovery testing, the birds were tested with three cluster types: (1) good clusters: Cache sites with seeds in them; (2) plundered clusters: Cache sites from which the experimenters had removed the seeds; and (3) bad clusters: With no cache sites. The birds performed accurately in both the good and plundered clusters, but avoided bad clusters. During the second half of recovery testing, the birds were presented with good, bad and old clusters. Old clusters included both good clusters that had been visited and emptied, and plundered clusters that had been visited during the first half of testing. The birds were less likely to enter old clusters, and their behaviour was less well-directed towards cache sites in old than in good clusters. These results demonstrate that nutcrackers can discriminate good sites from emptied ones, probably by either 'tagging' the location by remembering it or by forgetting it.