In a mainly experimental science based traditionally on hypothesis testing such as ecology, studying futures may be difficult. However, in the last few decades, predicting the consequences of global changes on the dynamics and function of ecological systems has become a major challenge in ecological research. To study how ecological scientists deal with potential difficulties in studying futures, we adopted a reflexive viewpoint on how scientists address the study of ecological futures. To do so we questioned a panel of ecological scientists on their practical involvement and point of view. Quantitative and qualitative analyses of their responses showed that predictions or predictive models were the dominant theme. Many quantitative models, based on statistical correlations, empirical rules or processes have been developed and their methodological limitations explored by the researchers we interviewed. In a small proportion of studies, qualitative scenarios have been elaborated to explore the range of possible futures. Interviewees emphasized the problem of dealing with ecological complexity and multiple future possibilities. Specificities of futures compared to past or present events were not fully identified. In fact, researchers studying futures mainly adopted a reductionist approach, trying to simplify complex ecological systems. But methods and tools promoted by such an approach to science may not always be appropriate to deal with future ecological complexity. Indeed, an emphasis on prediction prevents ecologists from acknowledging the multiplicity and undetermined nature of futures.