One view of pollination systems is that they tend toward specialization. This view is implicit in many discussions of angiosperm evolution and plant—pollinator coevolution and in the long—standing concept of pollination syndromes. But actual pollination systems often are more generalized and dynamic than these traditions might suggest. To illustrate the range of specialization and generalization in pollinators' use of plants and vice versa, we draw on studies of two floras in the United States, and of members of several plant families and solitary bee genera. We also summarize a recent study of one local flora which suggests that, although the colors of flowers are aggregated in phenotype space, there is no strong association with pollinator types as pollination syndromes would predict. That moderate to substantial generalization often occurs is not surprising on theoretical grounds. Plant generalization is predicted by a simple model as long as temporal and spatial variance in pollinator quality is appreciable, different pollinator species do not fluctuate in unison, and they are similar in their pollination effectiveness. Pollinator generalization is predicted when floral rewards are similar across plant species, travel is costly, constraints of behavior and morphology are minor, and/or pollinator lifespan is long relative to flowering of individual plant species. Recognizing that pollination systems often are generalized has important implications. In ecological predictions of plant reproductive success and population dynamics it is useful to widen the focus beyond flower visitors within the correct" pollination syndrome, and to recognize temporal and spatial fluidity of interactions. Behavioral studies of pollinator foraging choices and information—processing abilities will benefit from understanding the selective advantages of generalization. In studies of floral adaptation, microevolution, and plant speciation one should recognize that selection and gene flow vary in time and space and that the contribution of pollinators to reproductive isolation of plant species may be overstated. In conservation biology, generalized pollination systems imply resilience to linked extinctions, but also the possibility for introduced generalists to displace natives with a net loss of diversity.

Mutually beneficial interactions between flowering plants and animal pollinators represent a critical 'ecosystem service' under threat of anthropogenic extinction. We explored probable patterns of extinction in two large networks of plants and flower visitors by simulating the removal of pollinators and consequent loss of the plants that depend upon them for reproduction. For each network, we removed pollinators at random, systematically from least-linked (most specialized) to most-linked (most generalized), and systematically from most- to least-linked. Plant species diversity declined most rapidly with preferential removal of the most-linked pollinators, but declines were no worse than linear. This relative tolerance to extinction derives from redundancy in pollinators per plant and from nested topology of the networks. Tolerance in pollination networks contrasts with catastrophic declines reported from standard food webs. The discrepancy may be a result of the method used: previous studies removed species from multiple trophic levels based only on their linkage, whereas our preferential removal of pollinators reflects their greater risk of extinction relative to that of plants. In both pollination networks, the most-linked pollinators were bumble-bees and some solitary bees. These animals should receive special attention in efforts to conserve temperate pollination systems.

Anthropogenic climate change is widely expected to drive species extinct by hampering individual survival and reproduction, by reducing the amount and accessibility of suitable habitat, or by eliminating other organisms that are essential to the species in question. Less well appreciated is the likelihood that climate change will directly disrupt or eliminate mutually beneficial (mutualistic) ecological interactions between species even before extinctions occur. We explored the potential disruption of a ubiquitous mutualistic interaction of terrestrial habitats, that between plants and their animal pollinators, via climate change. We used a highly resolved empirical network of interactions between 1420 pollinator and 429 plant species to simulate consequences of the phenological shifts that can be expected with a doubling of atmospheric CO(2). Depending on model assumptions, phenological shifts reduced the floral resources available to 17-50% of all pollinator species, causing as much as half of the ancestral activity period of the animals to fall at times when no food plants were available. Reduced overlap between plants and pollinators also decreased diet breadth of the pollinators. The predicted result of these disruptions is the extinction of pollinators, plants and their crucial interactions.

The pollination syndrome hypothesis as usually articulated does not successfully describe the diversity of floral phenotypes or predict the pollinators of most plant species. Caution is suggested when using pollination syndromes for organizing floral diversity, or for inferring agents of floral adaptation. A fresh look at how traits of flowers and pollinators relate to visitation and pollen transfer is recommended, in order to determine whether axes can be identified that describe floral functional diversity more successfully than the traditional syndromes.

No abstract