Empiricism and Modeling for Marine Fisheries: Advancing an Interdisciplinary Science
- 779 Downloads
Marine fisheries science is a broad field that is fundamentally concerned with sustainability across ecological, economic, and social dimensions. Ensuring the delivery of food, security, equity, and well-being while sustaining ecosystems in the face of rapid change is, by far, the main challenge facing marine fisheries. A tighter integration of modeling and empiricism is needed to confront this challenge. In particular, improved incorporation of empirically grounded and realistic representation of human behaviors into models will greatly enhance our ability to predict likely outcomes under alternative adaptive strategies. Challenges to this integration certainly exist, but many of these can be overcome via improved professional training that reduces cultural rifts between empiricists and modelers and between natural and social sciences, ideally ending the presumption that there is a divide between empiricism and modeling.
Keywordsmarine fisheries modeling interdisciplinarity global change sustainabiity empiricism
Marine fisheries science is a broad field that is fundamentally concerned with sustainability across ecological, economic, and social dimensions. People rely on the ocean for food, livelihoods, and sociocultural well-being (Poe and others 2014; Daw and others 2015). At the same time, fishing can lead to environmental degradation and a loss of benefits that people desire, and this can be difficult to reverse (Hughes and others 2010; Neubauer and others 2013; Levin and others 2016). Marine fisheries science is, therefore, inherently interested in interactions among and within systems: biophysical systems (the ocean, the species, communities, populations), human systems (market forces, supply chains, communities, cultures and institutions), and various scales of governance (Charles 2001). Consequently, solving problems requires expertise and knowledge from physical sciences, biological sciences, social sciences, and humanities.
Given the many fields of knowledge and the applied aspects of fisheries, marine fisheries science has had to integrate empiricism and modeling to differing degrees. Consequently, there are a spectrum of tools and activities, ranging from (1) empirical experiments, observations, and surveys that employ descriptive and inferential statistics; (2) fitting statistical models to empirical data to analyze patterns in empirical data (for example, regressions, multi-variate analysis); (3) fitting parameters of dynamic process models to data using statistical theory (for example, estimating population abundance from data); and (4) simulation and theoretical models that are informed by data but are not formally fit to them. Here, we consider activities in 1 to 2 as being largely empirical and activities in 3 to 4 as being largely modeling.
Ensuring the delivery of food, security, equity, and human well-being while sustaining ecosystems in the face of rapid change is, by far, the main challenge facing marine fisheries. As we write this paper, the Indo-Pacific reefs are experiencing a massive bleaching event, and projected increases in sea surface temperatures are likely to overwhelm the acclimation capacity of corals (Ainsworth and others 2016). Additionally, hypoxia is increasing in magnitude and frequency in coastal oceans (Diaz and Rosenberg 2008), and dissolved oxygen minimum zones are expanding in open oceans (Stramma and others 2008). Changes in temperature are inducing changes in species distributions (Perry and others 2005; Pinsky and others 2013), and changes to the carbonate system are leading to unprecedented acidification that can alter ecological interactions at all levels of marine food webs (Gaylord and others 2015). Meanwhile, demand for marine-derived protein continues to grow and fishing vessels become more powerful and effective at removing fish (Watson and others 2015). Other human activities in oceans are also increasing (Andrews and others 2014), as people rely on oceans for energy development, sea floor mining, shipping, and recreation (Halpern and others 2008).
The pace of change poses severe challenges for management of fishery systems. How will the redistribution of resources caused by climate change affect food security and economies (Allison and others 2009; Barange and others 2014)? How can we predict likely outcomes to guide climate adaptation given that future conditions will be out of the range of our past experience (Schindler and Hilborn 2015)? What is the potential for acclimation and adaptation, and what management strategies might maximize that potential (Hoffmann and Sgro 2011)? How will social-ecological feedbacks promote or prevent adaptation (Ostrom 2009)?
Blending Empiricism and Modeling
Because marine life exists in an environment that is inhospitable to humans, it is largely invisible to us. Resource users are more visible, yet their behavioral drivers and intentions are complex and highly context-dependent. For these reasons, models are needed to understand and predict how populations, food webs, habitats, and human behavior and well-being change through time. Qualitative and quantitative empirical study is also needed to understand the linkages among and within biological, physical, and human components of fisheries systems, and how these linkages change through time. Below we elaborate on the specific needs for blending modeling and empiricism, describe some barriers, and finally identify approaches that promote successful blending. Finally, we review the type of professional training that will lead to practitioners that can work across the empirical-modeling divide.
Needs and Opportunities for Blending Modeling and Empiricism
The challenges outlined above beg for a close linkage between modeling and empirical studies. Climate change is driving fisheries systems to new states (Blanchard and others 2012), so mechanistic understanding of fishery systems and key drivers of change—derived from empirical study—is critical to develop and improve projections in models. This understanding will require advances in both empirical studies and modeling approaches. Data collection and monitoring practices will need to change from the current focus on counting fish, estimating population numbers, and measuring food production to be more inclusive of environmental, ecological, economical, and social variables and how they respond to key drivers, for example, tracking changes in demographic rates or fished species (mortality, growth) and relating these changes to environmental conditions. Scaling these data to predict change in fishery system structure will require model fitting to reveal the functional relationships among these variables. As ecosystems shift to new, previously unobserved states, regular data updates will be needed as part of an adaptive management feedback process. Because much of these data will be automated, models will need to be adapted to handle large data streams, and unconventional data types. Hopefully, large data streams, automation of data handling and evaluation, and inclusion in models will enable models to make predictions at the appropriate timescale for policy decisions (Dunn and others 2016). This type of dynamic ocean management should become more common as data collection and computing power capacity increase (Lewison and others 2015) and potentially lead to models that can make regional and seasonal forecasts based on higher-resolution inputs and analyses.
Forecasting the futures of marine fisheries is challenged by uncertainty regarding the adaptive potential of marine organisms (Merila and Hendry 2014). Modeling and empirical studies can be used together to address the challenges of eco-evolutionary dynamics (Anderson and others 2012), for example, blending niche-based modeling (for example, bioclimate envelope models, Cheung and others 2009) and empirical analysis of reaction norms and phenotypic expressions (for example, Gaggiotti and others 2009). This level of methodological and conceptual understanding is necessary not only to develop new models to project species distribution and traits in future climate, but also to assess the effects of human-induced selection gradients, such as those originated through management and policy actions (for example, Olsen and others 2012).
Another key uncertainty that needs novel blending of empiricism and models is how food webs will respond to environmental change and the extent to which functional redundancy may buffer changes in productivity and catch (Micheli and others 2014). Changes in catch composition and overall productivity will likely be complex and it is expected that there will be winners, losers, and surprises. A more sophisticated and deeper understanding of the community-wide effects of environmental change is beginning to emerge through empirical and modeling studies (Cheung and others 2009; Kroeker and others 2012), but models can be greatly improved by a more realistic merging of the two.
Models in marine fisheries can also be improved by incorporating human behaviors and realistically modeling them (Fulton and others 2011). To this end, it is important to add more knowledge from psychology, behavioral economics, and political science–disciplines that are exceedingly underrepresented in marine fisheries science, despite the importance of this knowledge for achieving sustainable resource systems (Hicks and others 2016). Social psychology is particularly relevant where national governance rules and restrictions (command-and-control) are ineffective such that behavior change needs to come from changed incentives (Grafton and others 2006; Fulton and others 2011; van Putten and others 2012). Psychology has a tradition of building theory based on experimental data but the results of their experiments are rarely incorporated in coupled human-resource behavior marine models (but see McClanahan and others 2016). Additionally, the experimental approaches used by economists are needed for quantitative integrated human-resource models, as well as information on cultural values, resource values (Ban and others 2013; Poe and others 2014; Hicks and others 2015), and how they drive human behavior (McClanahan and Cinner 2012). Analogous to the natural sciences, long(er) monitoring and time series of resource user behavior should better enable understanding the potential of adaptive resource management. Tapping into potential new data sources, such as satellite data, may help provide functional relationships and resource interactions in the human domain.
Challenges for Blending Modeling and Empiricism
Fisheries are complex, multi-dimensional systems that make identifying causal relationships challenging (Sugihara and others 2012). Non-linearities and context-dependent responses are common, but our approaches to identify causal relationships are often overly simplistic and therefore prone to errors. A well-known example of this problem is the failed attempts to relate environmental conditions to population reproductive success. Reproductive success of marine organisms is highly variable because it depends on ideal or optimal conditions for survival, growth, and dispersal during early life history (Hjort 1917; Cushing 1996). The multitude of factors that produce these optimal conditions, coupled with the difficulty of measuring them at scales relevant for early life history survivorship (Stachura and others 2014), means that many apparent causal relations break down as more variables and data are collected (Myers 1998). Also, this addition of mechanistic complexity to models and management has real costs in terms of data collection and the accuracy and precision of models (Punt and others 2014), and may not improve management advice (De Oliveira and Butterworth 2005). Although detailed inclusion of realistic, empirically based demographic rates in fisheries models is possible, it generally requires extensive combinations of long-term data collection, experiments, and meta-analyses of existing data (for example, Rossetto and others 2015). Such detailed information is not yet available for most fishery systems.
There are important cultural rifts between both empiricists/modelers and between natural and social sciences that can be difficult to bridge (Levin and Anderson, in press). The type of holism and efforts to coordinate studies at scales of entire fishery systems (for example, Fulton and others 2014; Daw and others 2015; McClanahan and others 2016) is challenging when professional incentives are often to focus studies that contribute to specific fields of enquiry (Fox and others 2006; van Dalen and Henkens 2012; Bromham and others 2016). Developing the long-term and team perspective is a challenge in the current atmosphere of individualism and rapid and precise contributions. Another key challenge in these teams is that natural scientists often do not fully appreciate the diversity of social science disciplines, approaches, and information that they can bring to the table, and vice versa (Moon and Blackman 2014). That creates an environment where there is an isolated, “social science team” (often consisting of a single person) that is tasked with providing a limited number of inputs, parameters, or model components that are integrated into a larger model, and some of the important aspects of human behavior being marginalized in the process (Cambell 2005). Further, in many disciplines, information derives from narratives, which are not easily translated into a model function or parameter value (Drew and Henne 2006).
A final barrier to developing coupled models of fishery systems is the paucity of information on larger scale financial drivers of fishery systems, specifically information on the markets for marine products. It is increasingly difficult and expensive to obtain market and price information for research purposes due to propriety issues of the often large horizontally and vertically integrated commercial enterprises. This issue applies beyond fisheries to aquaculture, transport, energy, and seafloor mining. If models aim to incorporate dynamic human behavior from all use sectors, the relationships between demand and supply and business behavior are needed to track the changing nature of trading enterprises in a global market (Plagányi and others 2014). Simplistic representation of supply chains in integrated marine models currently represents a big gap in evaluating and modeling the total fishery system and its response to changing climate and external drivers.
A Path Forward—Clear Objectives and Collaborative Modeling
The scope and breadth of change in fishery systems mean that empiricists and modelers can make better strides through more imaginative thinking outside of the realm of current models, current problems, and incremental change. Many things that influence marine fisheries are undergoing rapid change (Ban and others 2010), which means that perception for what is “normal” may have little relevance for the future. In this sense, models quickly outpace the accumulation and understanding of empirical knowledge, particularly mechanistic understanding of drivers and responses. Rather than building ever more complicated versions of existing models in the hopes that we can precisely predict what the future will bring (Schindler and Hilborn 2015), greater attention should be given to scenario planning, where the goal is to identify plausible future states, key uncertainties, and to develop adaptation strategies that are robust to these uncertainties (Peterson and others 2003). The collaborative approach (Figure 1) is essential to enable a robust exploration of objectives, uncertainties, and plausible strategies to cope with change.
Professional Training for Marine Fisheries Science
Marine fisheries science will continue to require individuals that have depth of knowledge in core disciplines, while being fluent in key concepts and principles from other disciplines and approaches. The need for interdisciplinary training in environmental sciences is well known, as are the particular constraints and challenges that such training entails (Graybill and others 2006). Here we focus on professional training that will encourage the blending of empiricism and modeling.
For modelers, the most pressing need is to leave the computer and experience the systems that they are modeling first hand. Modeling necessarily involves making simplifying assumptions, and we notice that individuals that work with the same model on a regular basis tend to take these assumptions as fact. That is, the model’s structural assumptions profoundly shape the way modelers view the real world. Gaining first-hand experience in the real-world challenges these assumptions, as even a week in the field will reveal the true complexity of actual systems (Figure 1). Modelers also need training in natural history as well as the fundamentals of taxonomy, life history theory, and ecology so that they can more critically evaluate their own model predictions and assumptions. Finally, modelers need practice working with scientists, stakeholders, and managers that are not proficient in computer programming and mathematics, to ensure that a diversity of knowledge can be incorporated into the models and their assumptions and outputs are understandable to all who utilize them (Figure 1).
Empiricists will be using or interpreting model outputs (for example, regional climate predictions) so they need to know how to do so critically. Most fundamentally, empiricists should recognize that models are simplifications of reality that are built for a given purpose; therefore a model is judged not by how accurately it resembles reality (for example, the level of detail and breadth of the model) but how useful it is (Starfield 1997). Basic modeling principles should be taught to everyone to enable this critical analysis and to appreciate the types of variables and metrics needed to feed into models. This includes “why do we model, and how do we use models to improve understanding?”; “how and why are simplifying assumptions made?”; “How do we evaluate sensitivity of our models to assumptions”; “how does one evaluate whether a model is appropriate for a question?”; and “what are different types of models, and what are trade-offs across generality, realism, and precision?”
Transformative training programs will end the presumption that there is a divide between empiricism and modeling. For the most part, professionals presently identify as either “quantitative” or “empirical,” with the former having advanced statistical, mathematical, and programming skills and the latter having deep knowledge of experimental and sampling designs, natural history, and detailed knowledge of organisms and human communities. Training should foster a new generation of professionals that routinely work across these approaches. Modeling can be done without high-level mathematics and statistics—algebra, calculus and basic probability theory is sufficient for the vast majority of modeling applications. Thus, the barriers for empiricist-minded trainees are not as high as they might appear. There are few barriers to conducting empirical work, other than patience, willingness to endure unpleasant conditions at times, knowledge of appropriate sampling techniques, strong interpersonal skills (especially for social science), and the need to practice mindful observations of the world around us. We see some notable progress in this regard, but much more remains to be done. This progress may be enhanced through training programs that build upon experiences in transdisciplinary training (Ciannelli and others 2014) that are structured to explicitly encourage shared language, knowledge, and respect for distinct approaches and disciplines.
We thank Steve Carpenter and Monica Turner for inviting us to join the special feature. Tim Essington was supported by National Science Foundation Grant OCE-1154648AM001. Tim McClanahan was funded by the John D. and Catherine T. MacArthur Foundation during this study period.
- Charles A. 2001. Sustainable fishery systems. Oxford: Blackwell Science. p 370p.Google Scholar
- Ciannelli L, Hunsicker M, Beaudreau A, Bailey K, Crowder LB, Finley C, Webb C, Reynolds J, Sagmiller K, Anderies JM, Hawthorne D, Parrish J, Heppell S, Conway F, Chigbu P. 2014. Transdisciplinary graduate education in marine resource science and management. ICES J Mar Sci 71:1047–51.CrossRefGoogle Scholar
- Cushing DH. 1996. Towards a science of recruitment in fish populations. Luhe: Ecology Institute. p 175.Google Scholar
- Drew JA, Henne AP. Conservation biology and traditional ecological knowledge: integrating academic disciplines for better conservation practice. Ecol Soc 2006;11: online.Google Scholar
- Fulton EA, Jones T, Boschetti F, Chapman KL, Little RL, Syme G, Dzidic P, Gorton R, Sporic M, de la Mare W. 2013. Assessing the impact of stakeholder engagement in management strategy evaluation. Int J Econ Manag Eng 3:83–99.Google Scholar
- Gaggiotti OE, Bekkevold D, Jorgensen HB, Foll M, Carvalho GR, Andre C, Ruzzante DE. 2009. Disentangling the effects of evolutionary, demographic, and environmental factors influencing genetic structure of natural populations: Atlantic herring as a case study. Evolution 63:2939–51.CrossRefPubMedGoogle Scholar
- Gaylord B, Kroeker KJ, Sunday JM, Anderson KM, Barry JP, Brown NE, Connell SD, Dupont S, Fabricius KE, Hall-Spencer JM, Klinger T, Milazzo M, Munday PL, Russell BD, Sanford E, Schreiber SJ, Thiyagarajan V, Vaughan MLH, Widdicombe S, Harley CDG. 2015. Ocean acidification through the lens of ecological theory. Ecology 96:3–15.CrossRefPubMedGoogle Scholar
- Grafton RQ, Arnason R, Bjorndal T, Campbell D, Campbell HF, Clark CW, Connor R, Dupont DP, Hannesson R, Hilborn R, Kirkley JE, Kompas T, Lane DE, Munro GR, Pascoe S, Squires D, Steinshamn SI, Turris BR, Weninger Q. 2006. Incentive-based approaches to sustainable fisheries. Can J Fish Aquat Sci 63:699–710.CrossRefGoogle Scholar
- Halpern BS, Walbridge S, Selkoe KA, Kappel CV, Micheli F, D’Agrosa C, Bruno JF, Casey KS, Ebert C, Fox HE, Fujita R, Heinemann D, Lenihan HS, Madin EMP, Perry MT, Selig ER, Spalding M, Steneck R, Watson R. 2008. A global map of human impact on marine ecosystems. Science 319:948–52.CrossRefPubMedGoogle Scholar
- Hjort J. 1917. Fluctuations in the great fisheries of northern Europe viewed in the light of biological research. Rapports et Proces-verbaux des Réunions. Conseil International pour l’Éxploration de la Mer 20:1–228.Google Scholar
- Levin PS, Anderson LE. When good fences make bad neighbors: overcoming disciplinary barriers to improve natural resource management. Coast Manag. (in press)Google Scholar
- Levin PS, Francis TB, Taylor NG. Thirty-two essential questions for understanding the social-ecological system of forage fish: the case of Pacific herring. Ecosyst Health Sustain 2016;2.Google Scholar
- Lewison R, Hobday AJ, Maxwell S, Hazen E, Hartog JR, Dunn DC, Briscoe D, Fossette S, O’Keefe CE, Barnes M, Abecassis M, Bograd S, Bethoney ND, Bailey H, Wiley D, Andrews S, Hazen L, Crowder LB. 2015. Dynamic ocean management: identifying the critical ingredients of dynamic approaches to ocean resource management. Bioscience 65:486–98.CrossRefGoogle Scholar
- McClanahan TR, Cinner JE. 2012. Adapting to a changing environment: confronting the consequences of climate change. New York: Oxford University Press.Google Scholar
- Plagányi EE, van Putten I, Hutton T, Deng RA, Dennis D, Pascoe S, Skewes T, Campbell RA. Integrating indigenous livelihood and lifestyle objectives in managing a natural resource. Proc Natl Acad Sci U S A 2013Google Scholar
- Plagányi EE, van Putten I, Thebaud O, Hobday AJ, Innes J, Lim-Camacho L, Norman-Lopez A, Bustamante RH, Farmery A, Fleming A, Frusher S, Green B, Hoshino E, Jennings S, Pecl G, Pascoe S, Schrobback P, Thomas L. 2014. A quantitative metric to identify critical elements within seafood supply networks. PLoS ONE 9:e91833.CrossRefPubMedPubMedCentralGoogle Scholar
- Walters CJ. 1986. Adaptive management of renewable resources. New York: MacMillan Publishing Company. p 374.Google Scholar