Learning and memory in hybrid migratory songbirds: cognition as a reproductive isolating barrier across seasons

Abstract

Hybrid zones can be used to identify traits that maintain reproductive isolation and contribute to speciation. Cognitive traits may serve as post-mating reproductive isolating barriers, reducing the fitness of hybrids if, for example, misexpression occurs in hybrids and disrupts important neurological mechanisms. We tested this hypothesis in a hybrid zone between two subspecies of Swainson's thrushes (Catharus ustulatus) using two cognitive tests-an associative learning spatial test and neophobia test. We included comparisons across the sexes and seasons (spring migration and winter), testing if hybrid females performed worse than males (as per Haldane's rule) and if birds (regardless of ancestry or sex) performed better during migration, when they are building navigational maps and encountering new environments. We documented reduced cognitive abilities in hybrids, but this result was limited to males and winter. Hybrid females did not perform worse than males in either season. Although season was a significant predictor of performance, contrary to our prediction, all birds learned faster during the winter. The hypothesis that cognitive traits could serve as post-mating isolating barriers is relatively new; this is one of the first tests in a natural hybrid zone and non-food-caching species. We also provide one of the first comparisons of cognitive abilities between seasons. Future neurostructural and neurophysiological work should be used to examine mechanisms underlying our behavioral observations.

Figure 1

Results from associative learning spatial task, showing performance over the testing period for all three ancestry groups in separate seasons for each sex. Score refers to the number of inspections required to obtain the food reward. Top right panel shows hybrid males required more inspections than parental forms during the winter (χ² = 10.00, p = 0.0067). Day 1 is grayed out because it was not included in our formal analyses; birds are expected to perform randomly that day (one-sample t-tests validate this suggestion, see text). Means (± standard errors) are shown.

Figure 2

Results from response to novelty task, showing latency to eat from a feeder. Normal feeder used pre- and post-treatment. Novel feeder used for treatment. Mean ± SE shown. Analyses were run using logged values for latency to eat but untransformed values are shown here for clarity. Results for latency to approach feeder shown in Fig. S3. Birds took significantly longer to eat from novel dishes (treatment) in all cases; hybrids did not require significantly more time to eat in any comparison.