Phenotypic integration approaches predict a decrease of reproduction rates of Caribbean pine populations in dry tropical areas

Abstract

• Key message: The combination of structural equation modelling and linear mixed-effects models opens a new perspective to investigate trait adaptation syndromes through phenotypic integration prediction at large geographical scales, a necessary step to understand the future of organisms under climate change. In the case ofPinus caribaeaMorelet, reproduction limits the species suitability, decreasing towards southernmost latitudes where dry conditions increase. Context: Caribbean pine is an ecologically and economically important species planted in all the tropical regions of the world, where it shows optimal growth and survival but low reproduction rates. Aims: This study investigates Caribbean pine fitness-related traits, accounting for phenotypic plasticity and local adaptation, to detect co-variation among traits and predict their relationship across the tropics. Methods: I re-analysed earlier data of survival, growth, reproduction, stem quality and development stage from 25 provenances of Caribbean pine planted in 16 trials in the tropical regions in a two-step modelling approach including (i) structural equation modelling (SEM) based on the current knowledge of the species and theoretical expectations coming from other species; (ii) mixed-effects model accounting for trait-relationships as defined by SEM and allowing for trait prediction. Results: Growth, survival and reproduction showed a slight but significant provenance effect indicating population differentiation and a positive co-variation between growth and reproduction, suggesting that trees reached optimal growth before they reproduced. Models predicted low reproduction rates of Caribbean pine across the tropics, decreasing towards southern latitudes where dry conditions increased. Conclusion: This study opens new perspectives to investigate trait adaptation syndromes through phenotypic integration prediction at large geographical scales.