Abstract

Floral integration may result from the combined effects of pollinator-mediated selection, genetic correlations and abiotic factors. Thus, by sampling a set of populations in the field and examining their variation of floral correlations in relation to pollinators and abiotic factors, we intended to shed light on the ecological factors underlying the evolution of floral integration. In this study, floral integration patterns and the composition of pollinator assemblage were characterized across ten populations of Alstroemeria ligtu to test the following: (1) Whether the patterns of floral integration estimated at population level covary with the composition of local pollinator assemblage, climate and/or geographic distance. (2) Whether the magnitude of floral integration decreases with the number or the morphological variability of pollinators. (3) Whether the behavior of the three most widespread pollinators is affected by floral integration. Our results indicated that populations with similar patterns of integration showed greater similitude in pollinator assemblages, but that this was not related to climate or geographic distance. We also found that the most widespread pollinator, Centris nigerrima, invested lower handling time in populations with higher levels of floral integration. The magnitude of floral integration was not related to taxonomical diversity, yet, unexpectedly, correlated positively with the variability of pollinator proboscis length. We suggest that variation in the magnitude and pattern of floral integration across the species range was related to the composition of pollinator assemblage, and not to pollinator diversity per se, nor to climatic variables. A. ligtu was visited by numerous pollinators, but floral integration patterns are likely shaped by only a subset of them, such as C. nigerrima, which is sensitive to variations in floral architecture.