ADAPTATION TO SINGLE RESOURCES AND THE EVOLUTION OF CROSSBILL (LOXIA)DIVERSITY

I quantitatively test the hypothesis that four taxa or ''types'' (spec ies or subspecies) of Red Crossbills (Loxia curvirostra) in the Pacifi c Northwest have diversified morphologically in bill characters in res ponse to alternative adaptive peaks presented by their food: seeds in conifer cones. Hypothetically, each adaptive peak corresponds to one c onifer species whose seeds are (1) produced regularly from year to yea r, (2) held in cones through late winter when seed is most limiting, a nd (3) protected from depletion by potential noncrossbill competitors. Four such conifers, termed ''key conifers,'' are present (Tsuga heter ophylla, Pseudotsuga menziesii, Pinus ponderosa, and Pinus contorta va r. latifolia). I use data on foraging efficiency for 31 captive crossb ills of four types to determine the optimal bill size and palate struc ture for foraging on the key conifers. As predicted, if each type is a dapted for foraging on a key conifer, the observed morphology of a giv en type is often the predicted optimal morphology for foraging on its respective key conifer. Two of the types have mean bill sizes (bill de pth) equalling their predicted optimal size. For one of the remaining types, the observed differs from the optimum by 0.4 mm; I was unable t o predict an optimal size for the remaining type. Optimal bill size va ries with season. Bill sizes corresponded mom closely to the optima fo r winter (lean period) than for summer. Observed mean width of the pal ate groove, in which crossbills hold conifer seeds while the seeds are being husked, was consistently close to the estimated optimal groove width. Optimal groove width was correlated (r2 = 1.00, n = 4) with see d size (cube root of mass), suggesting optimal groove width is determi ned by seed size. Overall, each crossbill type has either the optimal bill size or optimal husking groove width, or both, for foraging on th eir key conifers. Fitness set analyses indicate that there are substan tial trade-offs in foraging efficiency. The best phenotype for foragin g on one conifer is often only one-half as efficient on other conifers . AlI four fitness sets are concave, implying selection against interm ediate phenotypes. I conclude, first, that reliability of seeds on key conifers during periods of food scarcity is a critical feature in the ecology and evolution of crossbills. Second, optimization of morpholo gical traits occurs even in populations in highly variable environment s. Third, disruptive selection against intermediate phenotypes is like ly. This should maintain, if not reinforce, the distinctiveness of typ es. Fourth, the diversity of cone structure and seed size among key co nifers is ultimately responsible for the diversification of crossbills .