The evolution of butterfly wing shape is driven by multiple selective, phylogenetic and developmental influences. In my thesis, I focused on the evolution of wing shape in Papilionidae, a butterfly family presenting a high diversity of wing shapes. Papilionidae are collectively referred to as Swallowtail butterflies, owing to the tails that many species harbour on the hindwings. While this feature is particularly striking and diversified, its evolutionary drivers have never been investigated. Did tails evolve neutrally? What are the selective pressures affecting it? Do forewings and hindwings evolve independently? By combining micro- and macro-evolutionary approaches, my thesis aimed at answering these questions and identifying the main factors affecting the evolution of wing shape, with a particular focus on hindwing tails. Focusing on Iphiclides podalirius, I first tested whether tails deflect birds attacks away from the butterfly body (the deflecting effect hypothesis; Chapter I). I showed that natural wing damages mostly concern hindwings tails and colour-pattern, suggestive of predation attempts; I then conducted a behavioural assay with dummy butterflies, and showed that great tits (Parus major) focus theirs attacks on the tails; finally, quantifying the mechanical properties of fresh wings, I found that the tails are particularly fragile. Altogether, these results support a deflecting effect of hindwing tails, suggesting that predation is an important selective driver of the evolution of tails in butterflies. I then investigated the relative aerodynamic importance of tails in flapping flight (the aerodynamic effect hypothesis; Chapter II), conducting flight analyses of phenotypically altered I. podalirius. I showed that hindwing tails have a significant stabilising impact on flapping flight, suggesting that selection on aerodynamic performance likely affects the evolution of tails. Based on these experimental results, I then quantified the variation of fore- and hindwing wing shape at the macro-evolutionary scale (across the Papilionidae family; Chapter III). I compared the shape diversity and evolutionary rate among the two wings, and tested the link between diversification and phenotypic disparity. I specifically characterized the evolution of the tail at the family level. My results show that hindwings are strikingly more diversified than forewings, suggesting contrasted selective regimes on the two pairs of wings. Forewings might be under stabilizing selection in relation to flight anteromotorism, while hindwings might be submitted to a diversity of selective pressures. Our results on I. podalirius suggest a possible trade-off between attack deflection and aerodynamic effects, promoting the diversity of hindwing shape, and particularly the evolutionary lability of tails and associated colour patterns. Contrary to previous work, my results also suggest a tight coevolution of the two wing pairs, the presence of tails possibly affecting the selection on the forewings. Overall, this study shows that the combination of behavioural ecology and macro-evolutionary studies might shed light on key factors affecting morphological evolution. Altogether, my PhD work has brought some insights on the selection pressures involved in hindwing tail evolution and highlighted the complex links existing between forewings and hindwings evolution, between contrasted selection, developmental constraints and co-evolution.