Nature-based flood defence is an innovative design alternative for achieving protection against flooding. Despite significant advancements in science, models and concepts, routine implementation beyond pilot projects remains limited. To better understand why, we have looked into the complexities of nature-based flood defence implementation and its resolutions, modelling decision-making situations using game theory in three nature-based flood defence cases: The Markermeer Dikes, the Afsluitdijk Dam and the Sand Engine. We observe that nature-based flood defence games are of a multi-level and nested nature. While the decision of whether to employ a nature-based flood defence is seemingly made at the project level, this can only happen when it is coherent with the institutional context that is determined at the policy level. A social dilemma is apparent: while a multi-functional nature-based solution is attractive to a coalition of actors, it is not the most beneficial option for individual actors. Hence, they are faced with the dilemma of opting for their maximum benefit or opting for the greater societal benefit which is less favorable to them. This social dilemma can be tackled by making ‘smart moves’, as inspired by the Sand Engine case. The nested nature of the problem requires structural change in the institutional context to enable favourable conditions for nature-based flood defence implementations.

Greening flood protection (GFP) is increasingly recognized as an adaptive and flexible approach to water management that is well suited to addressing uncertain futures associated with climate change. In the last decade, GFP knowledge and policies have developed rapidly, but implementation has been less successful and has run into numerous barriers. In this paper, we address the challenge of realizing green flood protection goals by specifically considering knowledge in the decision-making of a Dutch flood protection project in Lake Markermeer. In this project, an ecological knowledge arrangement and a traditional flood protection knowledge arrangement are compared and their interactions analysed. The analysis provides insight into the specific difficulties of implementing GFP measures and identifies ways to realize GFP goals. The primary challenge is twofold: First, a self-reinforcing cycle of knowledge production and decision-making in the flood protection domain inhibits the introduction of innovative and multifunctional approaches such as GFP; second, the distribution of power is severely unbalanced in terms of ecological enhancement and flood protection, favouring the latter. Implementation of GFP requires structural change and the integration of ecological and flood protection knowledge and policy. Potentially rewarding routes towards this integration are the exploration of shared interests in GFP and the creation of mutual dependency between knowledge arrangements. The case study and the insights it provides show that GFP is far from mainstream practice and that implementation requires serious effort and courage to break with historical practices.