Towards an empirical foundation for marine ecosystem-based management: fishing behavior, fish abundance, and climate change
As policy makers strive to implement marine ecosystem-based management, they are faced with limited empirical support for their decisions. Managers specifically lack empirical information on the environmental dependence of fishery resources (including physical and biological oceanographic features), spatio-temporal patterns of fish abundance and metapopulation dynamics, and the behavioral responses of the harvest sector to spatially explicit policies. Moreover, as management institutions change, patterns of human activity will also change. Managers need an empirical foundation to understand how complex marine systems will evolve in response to policy changes. The specter of a changing climate accentuates the empirical difficulties of ecosystem-based management.
The premise of our proposal is that economic data from fisheries provide an enormous reservoir of information that can be used to help understand bio-physical features of marine systems, guide spatial management of the marine environment, and develop management institutions that are resilient to climate change. Fishery-dependent data from logbooks, landings tickets, observer coverage, and vessel monitoring systems can reveal on each day within a season where each individual vessel is fishing and how much of each species is being caught. Although fisheries scientists and managers have long used fishery-dependent data for stock assessment, data are typically aggregated across vessels to all vessels in the fleet, across space to the entire fishery, and across time to an annual time step. Aggregating fishery-dependent data may be expedient for fitting population models, however, aggregation discards valuable spatio-temporal information that is embedded in disaggregated choices of individual fishermen. In essence, disaggregated data are collected, but severely underutilized.
Figure 1 provides a simple schematic to guide intuition about the empirical work in the project. In the lower right corner, we observe the fishing choices of individual vessels over time and space (IV). These choices reveal information about spatio-temporal fish abundance (III) that is more finely resolved than fishery-independent stock assessment data. Observations on environmental variables (II) then allow the analyst to infer environmental dependence of fish abundance without directly observing abundance. Climate forcing (I) affects environmental variables, but we do not empirically explore these effects. Instead, we use climate models to explore the range of potential impacts on environmental variables. We then trace these potential impacts through the system to abundance and fishing behavior.
Microeconometric techniques, and discrete choice methods in particular, offer a way forward in meeting the empirical challenges of ecosystem management and quantifying the linkages above. Discrete choice models allow researchers to characterize the economic structure of fishing decisions, linking fishing behavior to spatio-temporal patterns of fish abundance (Eales and Wilen 1986; Dupont 1993; Larson, Sutton, and Terry 2000; Curtis and Hicks 2000; Smith 2000; Smith 2002; Hicks and Schnier 2006). This link, in turn, provides an empirical foundation for connecting to fish population dynamics (Smith, Zhang, and Coleman 2008) and modeling spatially explicit policies like marine reserves (Wilen et al. 2002; Smith and Wilen 2003; Smith, Sanchirico, and Wilen 2009). Discrete choice methods ultimately can harness the massive amount of information that is embedded in individual fishing logbooks and observer data to reveal properties of marine ecosystems. In this project, we draw on the industrial organization literature on product differentiation (Berry 1994; Berry, Levinsohn, and Pakes 1995) and an emerging literature on spatial sorting in public economics (Bayer and Timmins 2005; 2007; Timmins and Murdock 2007) to estimate sorting models of observable heterogeneity.
A key insight for our analysis is that choice-specific fixed effects, which are estimated from choice data in a first stage, contain important information about site attributes. These attributes, in turn, reflect knowledge of the fishing fleet about the spatio-temporal pattern of fish abundance (Smith and Zhang 2007). In the second stage, choice-specific fixed effects are decomposed using ordinary least squares or instrumental variables regression. In our case, this decomposition involves quantifying the environmental basis for spatio-temporal abundance. These second stage coefficients are related to the bottom-up correlations computed by Ware and Thomson (2005), but our parameters are informed by an economic model of behavior. As such, they do not implicitly assume that fishing vessels search randomly for fish; our parameters are structural and can directly inform out-of-sample simulations under changing climate. Importantly, our model also makes distinctions among the spatial choices that vessels typically make on a fishing trip.
The project uses the structural dependence of spatio-temporal fish abundance on environmental variables to explore the effects of climate change scenarios on the entire coupled system. By simulating the effects of climate change on environmental variables, the resulting effects on abundance and fishing behavior can be traced back up the system. This allows managers to consider how climate change could influence patterns of fishing behavior and, in turn, how climate-resilient spatial management- such as use of Marine Protected Areas-could be designed.
Working group members
Martin Smith, Nicholas School of the Environment
Susan Lozier, Nicholas School of the Environment
Richard Barber, Nicholas School of the Environment
Dylan McNamara, Department of Physics and Physical Oceanography, UNC Wilmington
Junjie Zhang, School of International Relations and Pacific Studies, University of California at San Diego