Résumé
Since its opening 5.33 million years ago, the Gibraltar Strait has always contributed to the Mediterranean fauna and flora. Despite the increasing importance of the phenomenon, ecological determinants underlying colonization success of Atlantic fishes in the Mediterranean Sea have been poorly investigated. Here we reconstruct the recent historical colonization of the whole Mediterranean Sea by Atlantic fishes and we aim to determine where Atlantic fishes preferentially establish and whether some biological traits and ecological factors can be correlated to the colonization success (climate match, position in the water column, maximum body length, propagules, confamilial resistance, depth). A database on Atlantic fish species records from 1810 to 2006 was built and the colonization rate of each introduced species was estimated. Analysis of Variance, Chi squared test and logistic regression were used to investigate the relationships between ecological variables and colonization success. In addition, an index of asymmetry was used to analyse the relative colonization on the two sides of the Mediterranean Sea. Overall 48.33% of Atlantic species introduced in the Mediterranean Sea succeeded in colonizing eastwards. We found that habitat depth of Atlantic species is significantly related to their colonization success due to the obstacle of the shallow depth of the Gibraltar Strait (300 m). It also appears that despite the cyclonic general water circulation on the North Western basin, the northern side is more colonized than the southern one: 70.40% of the studied species colonize the northern side, while only 29.62% colonize the southern one. Two hypotheses may explain this trend: the bottom-up process that enhances the colonization success of Atlantic fishes along the Spanish coast of Alboran Sea owning to its high productivity and the intensiveness of the scientific explorations along the northern Mediterranean side. We conclude that crossing the Gibraltar Strait does not guarantee the colonization success and that species life-history and functional traits are poor predictors. Instead we suggest that environmental factors may determine favourable locations for invasive species installation.