Data-driven domains such as public administration, health or mobility have adopted a so-called 'data ecosystem' perspective to unify the socio-technical aspects fostering data-driven collaboration. While a data ecosystem is technically able to collect and merge their different datasets, it is yet relatively unable to facilitate meaningful forms of collaboration between actors. Based on previous research on value creation in data ecosystems, we hypothesize that this inability is mainly due to ecosystems not reflecting actor values, i.e. aspects which are important and imply a desirable behavior, often related to goals, objectives, motivations and decision making. This paper therefore proposes a reflective approach to reveal the values in data-driven collaboration by answering the following research questions: What role do values play in the process of developing a data ecosystem? And how can value-led participatory design support data-driven collaboration? We attempt to answer these questions through an exploratory study based on 5 interviews with consortium members of a garden data ecosystem currently in development around a citizen science initiative in Flanders, Belgium. We discovered that the explicit use of values and frictions has the potential to augment the collaboration between actors. This approach can thus be useful to future practitioners who aim to expand the societal impact of their work.