BEHAVIORAL, ECOLOGICAL AND MORPHOLOGICAL CORRELATES OF FORAGING FOR ARTHROPODS BY THE HUMMINGBIRDS OF A TROPICAL WET

This paper seeks to clarify the ecological and evolutionary impacts of foraging for arthropods vs. nectar in hummingbirds, using data on morphology, sites, and tactics of arthropod foraging, and prey taken by ii species of hummingbirds of La Selva, a Costa Rican lowland wet forest, as well as information on time budgets of male and female hummingbirds from two other sites. Hermit hummingbirds have long, curved bills and are nearly exclusively understory hover-gleaners that take mostly spiders. Most glean from webs but Threnetes ruckeri differs in foraging behavior and takes mostly jumping spiders. Nonhermits have shorter, straighter bills and employ a greater range of foraging tactics and sites, taking a wider variety of prey. In most species flycatching is frequent and flies and wasps are the predominant prey items, but some also take many spiders and ants; in particular, Heliothryx barroti is predominantly a hover-gleaner of the canopy and takes more spiders than flies. In Thalurania colombica, males and females differ in sites and tactics of arthropod foraging during the breeding season but not at other times. Breeding female hummingbirds spend much more time foraging for arthropods than do males in the same times and places, although flower visitation still, constitutes the majority of foraging time. Among La Selva hummingbirds, bill curvature is strongly correlated with the proportion of gleaning vs. hawking, and with the proportion of prey taken from the substrate (spiders, ants) as opposed to volant prey (flies, wasps). Broad wings (low aspect ratio) and high wing disk loading are correlated with hovering, gleaning, and the proportion of spiders and ants in the diet; narrow wings and high wing disk loading, with flycatching. Except for the three species that engage in the most flycatching, females have broader wings than males and in virtually all species, females have higher wing disk loading. However, no parameter of wing morphology shows a clear relationship to strategies of nectar exploitation, probably reflecting the facultative nature of the latter. Hermits and nonhermits differ strongly in their preferred foraging levels in the vegetation, both for nectar and for arthropods, with the former using the understory almost exclusively, the latter using the canopy to a much greater extent, in both forest and second growth. Nevertheless, the two groups do not differ in the proportion of foraging attempts for nectar vs. arthropods, nor in the distribution of foraging through the day, in either habitat. In nearly all species, bills of females are longer than those of males, and this could increase their ability to forage for arthropods at least as much as for nectar. Much of the controversy regarding frequency and energetic importance of arthropod foraging vs. flower visitation reflects the biases inherent in observing each type of foraging or in interpreting data from stomach contents or emetic samples, or from failure to consider the specific tactics of foraging for arthropods. Reports that hummingbirds can exist for extended periods or breed without access to nectar or alternative sugar sources require confirmation, as does a report that availability of arthropods rather than nectar determines hummingbird breeding seasonality in dry areas. The major impact of arthropods in hummingbirds' diets appears to be at the daily, rather than seasonal, level. Data on arthropod prey do not support the notion that hummingbirds visit flowers for insects rather than nectar, and there is no evidence to suggest that their bills are specialized for extracting insects from flowers. Rather, it appears that the bills of hummingbirds are adapted to extracting nectar from particular types of flowers, and that these bill specializations in turn constrain the tactics of arthropod foraging-in particular, the long, curved bills of the hermits might preclude effective flycatching. Selection on wing morphology from flower visitation might well be mostly indirect, via constraints on arthropod foraging tactics imposed by bill morphology, rather than directly through nectar foraging strategies. There is no basis for considering hermits to be more insectivorous than nonhermits; they represent a separate radiation within the Trochilidae, in a different habitat and in association with a different group of flowers, than do several major radiations of the nonhermits.