USING STRUCTURAL EQUATION MODELING TO MEASURE COMPLEXITY IN CHICK-A-DEE CALLS OF MOUNTAIN CHICKADEES (POECILE GAMBELI)

Although challenging to define and measure, communicative complexity has been studied in various species and signals. One heavily studied communication signal, the chick-a-dee call, is interesting because of its structural and functional complexity. Recent research on the complexity of the chick-a-dee call in mountain chickadees has revealed novel call characteristics that may not be fully captured by traditional complexity measures like uncertainty statistics. As relatively little is known about the usefulness of structural equation modeling (SEM) in animal communication research, here I provide what I believe is the first exploration of the functionality of SEM to understand signal complexity. SEM is a statistical measure that tests and estimates the relationships between observed variables and unobserved (abstract) constructs. SEM analyses of calls from Colorado (CO) and California (CA) mountain chickadee flocks revealed that specific observed variables seemed to measure call complexity well. Conflicting group differences in complexity were also observed between the two groups. CO flocks had significantly higher complexity in their calls than CA flocks, the opposite of what had been found using uncertainty statistics. Although complexity is defined differently in these two measures, this contrasting evidence stresses the need to refine the definition and measurement of communicative complexity. Additionally, significant relationships in both groups were found between two predictor variables (flock size and temperature), and moderation effects for one of them (temperature). Although more research is needed to define theoretically sound variables to measure abstract constructs like complexity, the unique relationships revealed by SEM in this foundational study show the possibilities available to a variety of researchers studying animal communication.