Although there exists a large and growing body of research pertaining to watershed governance networks (White, 1957; McGinnis, 1999; Schneider, Scholz, Lubell, Mindruta, Edwardson, 2003; Imperial, 2005; Sabatier, Focht, Lubell, Trachtenberg, Vedlitz, Matlock, 2005; Schlager and Blomquist, 2008), no study to date has attempted to undertake a comprehensive network analysis encompassing the legislative, regulatory, partnership, advocacy, and technical assistance networks that are simultaneously in action within the same region to ensure water quality.  An extensive, qualitative case study of the Lake Champlain Basin watershed governance networks identified several key  overlapping action arenas (Ostrom, 2005; Koliba, Meek, and Zia, 2010).  Testing the premise that organizations will likely play overlapping, yet distinct, roles, we suggest that an organization’s network placement is contingent on each organization’s self-understood role, jurisdiction, policy tool preferences, and resource capacity.

Even in mandated networks, organizations will choose their method and level of involvement in network activities (Moynihan, 2009; Provan and Lemaire, 2012). To identify patterns in how organizations utilize a governance network, this study will examine the policy implementation networks in two watersheds of the Lake Champlain Basin (LCB): the Winooski watershed and the Missisquoi watershed.  These watersheds are part of a large and complex system of water quality governance that has been struggling to improve water quality in the LCB, making the implementation network a center of attention for governmental and non-governmental actors.  This will provide a rich and varied network, covering many action arenas and policy subdomains, such as wastewater, stormwater runoff, and river corridor management.

An online survey is used to identify what types of interactions exist between actors in a variety of policy subdomains and action arenas.  Network actors are coded for levels of jurisdiction, sectors and main functions.  Policy tool or task enactments will be identified including: technical assistance provision, regulation and permitting, grants and contracts, public information, and other incentives.  Using these characteristics, the study applies an exponential random graph model (ERGM), which uses the location of observed ties and relationships between the characteristics of nodes that do and do not share ties to predict the likelihood of a tie occurring between any two actors.  Applying this model in the different networks that exist in different policy subdomains and action arenas reveals these likelihoods in each network.  This indicates where different types of organizations are more and less likely to be active participants in networked governance, which, in turn, provides a picture of the detailed structure of governance networks that will aid in understanding how they vary between subdomains and action arenas and provide a basis for further study into how governance networks behave over time.