TRAPLINE FORAGING BY BUMBLE BEES .2. DEFINITION AND DETECTION FROM SEQUENCE DATA

Trapline foraging-repeated sequential visits to a series of feeding lo cations-presents interesting problems seldom treated in foraging model s. Work on traplining is hampered by the lack of statistical, operatio nal approaches for detecting its existence and measuring its strength. We propose several statistical procedures, illustrating them with rec ords of interplant flight sequences by bumble bees visiting penstemon flowers. An asymmetry test detects deviations from binomial expectatio n in the directionality of visits between pairs of plants. Several tes ts compare data from one bee to another: frequencies of visits to plan ts and frequencies of departures to particular destinations are compar ed using contingency tables; similarities of repeated sequences within bees are compared to those between bees by means of sequence alignmen t and Mantel tests. We also compared observed movement patterns to tho se generated by null models designed to represent realistic foraging b y ono-traplining bees, examining: temporal patterns of the bee's spati al displacement from its starting point using spectral analysis; the v ariance of return times to particular plants; and the sequence alignme nt of repeated cycles within sequences. We discuss the different indic ations and the relative strengths of these approaches.