Morphological evolution and the behavioral organization of agricultural division of labor in the leafcutter ant Atta cephalotes.

A main challenge in the analysis of division of labor in insect societies characterized by worker polymorphism has been identifying the number of physical castes and determining their relationship to task performance. We addressed this question using the extremely polymorphic leafcutter ant Atta cephalotes as a model by measuring 22 morphological characters and applying multivariate Gaussian mixture modeling to define worker size-class boundaries. Our statistical approach discriminated five physical worker size classes (subcastes) and found continuous variation in most morphological characters. Some worker size classes showed patterns of covariance, modularity, and integration in head, mandible, and leg traits, suggesting biomechanical functionality in tasks such as leaf cutting. Task repertoires and act frequencies were recorded to identify how fungal gardening, brood care, leaf-harvesting, and other tasks were distributed across worker size groups. Results showed that small and mid-sized media workers performed more diverse and complex tasks, including leaf harvesting, than workers of other size classes. Minims participated in fungal-gardening and nursing tasks, whereas large medias mainly participated in leafcutting and majors made few direct contributions to fungal cultivation. Results of our integrative analyses demonstrate the existence of an unexpectedly large number of worker size classes that can differ significantly in performance of agricultural tasks. However, overlap in task repertoires of workers at size-class boundaries suggests that not all morphologically distinct groups display discrete behavioral profiles. Morphological groups are instead distinguished by a combination of the tasks that each performs and their relative frequencies. Significance statement: The evolutionary coupling of behavior and morphology is central to understanding division of labor in insect societies. Our work applies robust statistical modelling to offer a novel analysis of worker physical caste categorization based on extensive morphometric sampling. Our method improves upon previous approaches by incorporating measurements of a large number of diverse morphological traits likely involved in task performance to examine worker physical caste evolution. We integrate morphological evolution with behavioral data on relative task performance rates and task repertoire size to describe the morphological and behavioral space of size-variable workers. Our approach provides methods to quantitatively analyze division of labor and behavioral performance, and offers new insights into worker trait modularity and integration and task performance associated with the organization of complex insect societies. [ABSTRACT FROM AUTHOR]