Climatic niche shift predicts thermal trait response in one but not both introductions of the Puerto Rican lizard Anolis cristatellus to Miami, Florida, USA

Global change is predicted to alter environmental conditions for populations in numerous ways; for example, invasive species often experience substantial shifts in climatic conditions during introduction from their native to non-native ranges. Whether these shifts elicit a phenotypic response, and how adaptation and phenotypic plasticity contribute to phenotypic change, are key issues for understanding biological invasions and how populations may respond to local climate change. We combined modeling, field data, and a laboratory experiment to test for changing thermal tolerances during the introduction of the tropical lizard Anolis cristatellus from Puerto Rico to Miami, Florida. Species distribution models and bioclimatic data analyses showed lower minimum temperatures, and greater seasonal and annual variation in temperature for Miami compared to Puerto Rico. Two separate introductions of A. cristatellus occurred in Miami about 12 km apart, one in South Miami and the other on Key Biscayne, an offshore island. As predicted from the shift in the thermal climate and the thermal tolerances of other Anolis species in Miami, laboratory acclimation and field acclimatization showed that the introduced South Miami population of A. cristatellus has diverged from its native-range source population by acquiring low-temperature acclimation ability. By contrast, the introduced Key Biscayne population showed little change compared to its source. Our analyses predicted an adaptive response for introduced populations, but our comparisons to native-range sources provided evidence for thermal plasticity in one introduced population but not the other. The rapid acquisition of thermal plasticity by A. cristatellus in South Miami may be advantageous for its long-term persistence there and expansion of its non-native range. Our results also suggest that the common assumption of no trait variation when modeling non-native species distributions is invalid.