Abstract
1. Human disturbance of natural habitats has increasingly promoted hybridization between previously isolated species. Coffee trees (Coffea canephora, C. liberica and C. arabica) have been introduced into New Caledonia for cultivation since the 1850s. Many plantations have since been abandoned, leaving the cultivars to evolve in natural conditions. This historical situation and the local environmental conditions have led to gene flow between sympatric species and the creation of hybrid zones which can be considered as a natural laboratory for studying gene flow dynamics. Our objectives were to assess the nature of these hybridizations and to model the distribution of habitats favourable for these hybrid zones at a regional scale. 2. We analysed a tri-specific model population with 26 microsatellite markers using a multilocus analysis approach. We applied Bayesian methods to characterize the species identity of founder individuals and identify inter-specific hybrids. A high level of genetic diversity and inter-specific hybridization was observed in the population. To evaluate the presence of other favourable environments in the local area, a geographic information system was used to spatially differentiate the phenomenon and optimize future hybrid collecting missions. We constructed a predictive distribution model based on observation and environmental expertise (climatic parameters, physical parameters and landcover) at four reference sites of the hybrid zone. 3. At all four sites the Coffea species were found to be involved in inter-specific hybridization, with one population, in particular, revealing high genetic diversity. According to the environmental expertise, 40% of the studied region appeared to be favourable for this phenomenon and the model was validated with a prediction accuracy of 79%. 4. Synthesis and applications. The use of microsatellite markers was efficient for assessing the nature and extent of coffee tree hybridization events. The environmental expertise and model allowed comparisons between New Caledonia and Africa ecological niches. This methodology may be recommended for better management of de novo in situ coffee genetic resources and applied to other plants to provide insight into the impact of adaptation to new environment. Combining environmental and species biological parameters would help in analysing modifications in plant reproductive behaviour induced by global climate change.