In environmental management, like in other applied operational domains, concrete actions are often at odds with state-of-the-art theoretical knowledge. Institutional settings organizing the way scientific knowledge is applied in environmental management practice play an active role in structuring such knowledge-action discrepancies. Scientific councils of protected areas are key institutions playing this role. Their two main missions are (1) to design, monitor, and evaluate the production of knowledge inside protected areas, and (2) to advise managers as to designing and implementing conservation actions. This article explores a database gathering information on scientific councils in France. The case of this country is exemplary because the existence of scientific councils is systematic in its protected areas, and their structure is streamlined by regulatory requirements. We use this database to investigate the publication records of the members of these scientific councils. This enables us to shed a quantitative light on the "epistemic landscape" (i.e., the kinds of knowledge and knowledge-holders that play a key role in interactions between science and practice) that scientific councils materialize. Our findings suggest that this epistemic landscape is poorly connected to the concerned protected areas. Moreover, some prominent topics in academic research (functional ecology, invasion biology and conservation planning) appear to be neglected in the recollected publication records. Such results prompt the question of whether the composition of SCs should be adjusted to reinforce the role of scientists tackling research questions applied to the PA at issue, and to better reflect scientific priorities in the conservation literature. Though results solely based on bibliometric analysis should be interpreted with due caution, they are useful open debates on how to improve the design of institutions structuring knowing-doing spaces, beyond the specific case of French scientific councils. Keywordsknowing-doing gaps, scientific councils, protected areas, bibliometric analysis, expertise.

1. Introduction Since the 2000s, a growing literature has emphasized that environmental initiatives are plagued by socalled "knowing-doing gaps" or "knowledge implementation gaps", with major detrimental implications [1-3]. These "gaps" refer to the fact that environmental actions in the field are often at odds with state-ofthe-art scientific knowledge, which might translate into wrong-headed or at least suboptimal environmental policies and actions. Although this literature was certainly useful to raise awareness among environmental scientists and practitioners, the term "gap" was arguably ill-chosen. Whereas this term suggests that a void separates knowledge from action, the so-called "knowing-doing gaps" owe their very existence to complex mechanisms hindering knowledge transfer and/or the application of transferred knowledge. These socalled "gaps" are hence better characterized as spaces "in which shared interests, value conflicts, and complex relationships between scientists and publics can interact" [4-6]. In this article, we propose a focussed, but hopefully useful, exploration of some aspects of what we term "the epistemic landscape" of these spaces. We use this phrase to refer to the pattern of knowledge characterizing these spaces, i.e., the kinds of knowledge and knowledge-holders that play a constructive (or destructive) role in interactions between science and actors involved in designing, monitoring and evaluating environmental actions. In general terms, the design and implementation of environmental actions involve complex, multifarious decision-making processes, with multiple actors and institutions playing different, sometimes overlapping, and/or contradictory roles [7-11]. This complex picture is often referred to as "governance". Governance patterns condition the extent to which different forms of knowledge percolate through decision-making processes and shape actions, making them key drivers of the epistemic landscape [6]. Yet, the influence and importance of governance patterns extend beyond epistemic issues: they also wield crucial influence in determining how legitimate [12-13] and operational [14] collective decisions are. A vast literature provides valuable insights into this multifarious influence of governance patterns. However, by highlighting how governance patterns have concomitant effects on the epistemic landscape, legitimacy and operationality, the existing literature tends to overlook its distinct effects on the epistemic domain, potentially ignoring some of its unique peculiarities. To overcome this lacuna in the literature, Mangos et al. [6] proposed a conceptual framework, based on Meinard & Tsoukias's [15] analysis of the concept of rationality of decision support, which is itself anchored in Habermas's theory of communicative action [16]. According to Mangos et al. [6], what we term the "epistemic landscape" (the authors do not use this phrase) is moulded by three constraints (reflecting Meinard & Tsoukias's [15] three conceptions of rationality, which themselves reflect Habermas's three models of rational action [16]): "framing", "initiative", and "governance." The first constraint refers to the fact that, in some decision-making processes, some actors are entitled to filter the kind of information that can be used to inform decision-making. The second one presents the notion that some decision makers impose that decision support should leave them enough room to make decisions they can really consider to be theirs, rather than being entirely or mostly dictated by the decision support. Lastly, Mangos et al. [6] use the term "governance" in a more specific sense than the one in which it is used in the literature more broadly, to refer to the fact that, in some decision-making processes, some actors are entitled to validate or invalidate decisions, and are therefore in a position to discard acts of decision support that point towards decisions they are not liable to validate. In an empirical case-study, Mangos et al. [6] show how these three constraints materialize and mould the epistemic landscape. Similar studies at larger scales are needed to elaborate a general picture of the constraints and their effects on the epistemic landscape, which would be useful to understand important aspects of the structure and functioning of epistemic landscapes. However, by confining their focus to mechanisms endowed with a constraining force, Mangos et al. [6] arguably overlook important phenomena and processes moulding the epistemic landscape in subtler manners. Such phenomena and processes might give prominence to some pieces of knowledge at the expense of others, or suggest courses of actions without enforcing them. In that sense, Mangos et al. 's [6] conceptual model provides only a partial view on the epistemic landscape, which should be complemented by other perspectives that address the phenomena and processes neglected by this model. In this article, our aim is to provide such a view, by analyzing a specific institutional setting designed to organizing interactions between science and decision-making in many conservation projects or institutions: scientific councils (SCs)also termed "scientific boards", "scientific committees", or "research councils". These are interdisciplinary and multi-institutional entities created on a legal or voluntary basis. SCs are made up of experts appointed for an office (between 4 and 6 years) who are willing, without any additional remuneration, to devote part of their time to supporting the management of a protected area [17, 18]. We focus on the case of the scientific councils of natural protected areas