- Shopping Bag ( 0 items )
Reviews in Fish Biolgy and Fisheries
An extremely valuable book. . . . [W]e believe this to be a very important text for fisheries ecology and management.
"This book is the next major contribution to the field of fisheries science. Walters and Martell provide fresh and non-intuitive perspectives on a variety of issues. Their merciless slaughtering of cherished sacred cows is supported by sound scholarship, and cogent, well-reasoned arguments. This will become a landmark work."—Jeffrey Hutchings, Dalhousie University
Much of this book is about the derivation, use, and abuse of various mathematical models used to make decisions about how to manage harvested aquatic ecosystems. There is a long tradition of such modeling, and many biologists still look upon that tradition with much puzzlement and even contempt. Anyone who has taken even a bit of time to look at any aquatic ecosystem cannot help having seen that such systems are incredibly complex in their spatial, temporal, and trophic organization. Further, the complexity is not just a matter of structural diversity (lots of kinds of creatures). It also involves dynamic complexity in the form of a rich variety of feedback effects. Changes in the abundance of any creature due to natural or human factors are likely to result in a cascade of changes in the vital statistics (birth, death, growth rates) of other creatures in the food web, which in turn can feed back to impact further changes in the abundance of that creature. In the face of this complexity, it often seems both arrogant and foolish to pretend that we can make any useful predictions about what will happen when people selectively harvest some species that are fun to catch or good to eat, or change ecosystem fertilitythrough deliberate or inadvertent changes in nutrient loading, or alter the physical habitat of an ecosystem.
After much experience in the field, we would be the first to agree that it is indeed impossible to capture fully the rich behaviors of ecosystems in mathematical models, particularly when we try to include unregulated human activities (humans as dynamic predators) in the calculations. But in this chapter, we offer three main arguments about why it is important to keep trying to build useful models. The first, which we will not discuss any further because it is so obvious, is that modeling is a great and perhaps necessary way for scientists to force themselves to think clearly and to put their claims to understanding on the table in the form of specific predictions. The second, which we discuss in the following three sections, is that prediction in some form is required for management choice, i.e., the issue in management-policy design is not whether to model but rather how to go about it. The third, which we discuss in a closing section, is that there are some predictable regularities in the way natural populations and ecosystems respond to human disturbance, so that at least some kinds of useful predictions are not as likely to fail as they may initially appear.
1.1 The Role of Predictive Models
If the people in a fisheries management agency watch some fishery change while asserting that they are powerless to implement regulations that might alter the path of change, then that agency is not really a management agency at all; at best, it is a monitoring agency. The very word "management" implies some capability for making choices among options that might make some difference. That is, management is making choices. But what is involved in making any choice among alternatives? If we can choose either option A or option B, then we must either toss a coin or consciously construct arguments in the form "we believe that the outcome of A will be X while the outcome of B will be Y, and we prefer X to Y." Such sentences contain two kinds of assertions: (1) about the outcome (or range of outcomes, or probabilities of various outcomes) for each choice, i.e., predictions about what will happen in the future, and (2) about management objectives, i.e., which future outcomes would be preferred.
So making choices necessarily involves some method for predicting the future. This means the issue in management decision-making is not whether to model the future somehow (that is inevitable) but, rather, what model to use in making the prediction(s). Here there are two basic choices: to predict using the sometimes wonderful intuitive (and largely subconscious) capabilities of the human mind, or to resort instead to some explicit model or "deductive engine" for piecing together known elements of the prediction in some conscious way.
It is worth noting in passing that scientific research also necessarily involves making predictions, whether or not these predictions are stated as explicit alternative hypotheses about the outcomes of alternative experimental treatments. Even purely "observational" or "natural history" research programs cannot be designed and implemented without making some very strong assumptions (predictions) about where, when, and what variables or factors are worth observing, i.e., are likely to carry useful information about causal relationships. The experimental scientist can escape some responsibility for making specific predictions by constructing treatments (choices) that give clear, qualitatively different predictions about directions of response under alternative hypotheses. And the scientist has another advantage in terms of being able to choose the questions (options) to be addressed without much regard for whether those questions are of general interest to anyone else. So it is perhaps not surprising that scientists are much more likely than managers to make misleading assertions, i.e., "prediction is impossible in complex systems" or "it is not necessary to construct quantitative models in order to make useful predictions." Scientists who make such claims are clearly not the people to provide guidance about policy choices, nor are they likely to have much experience with the agonies of having to make hard choices.
Given that natural ecosystems are very complex and will be "driven" to future change by unpredictable environmental changes as well as human activities, so that we cannot possibly produce good unconditional or "open loop" predictions of future change, how can we hope to manage ecosystems if management choices require prediction? Or how can we hope to compare policy choices until we "understand" all the interactions and external forces that drive change? The answer to these questions is actually quite simple, if we look carefully at the character of the policy predictions required for decision-making: to choose between policy A and policy B, we do not, in fact, require unconditional predictions about the future, or even about most of the causes and patterns of variability that the future will bring. Rather, we need only to be able to predict whether policy A will do better than policy B for a sufficiently wide range of possible futures to make it a "better bet" than policy B. That is, policy predictions need not be about the future in general but, rather, only about those aspects of future change that could be directly impacted by the specific actions/interventions involved in the policy, and even in relation to these changes we generally require only predictions of relative performance. This means, e.g., that when someone asks, "How can you manage the fish when you do not even know how many there are?", we can answer by pointing out that we can compare policy choices for a wide range of possible actual numbers of fish, to find choices that are at least somewhat robust despite the uncertainty about the numbers. Further, we can generally specify policy choices as rules for response to change rather than absolute degree of impact. Consider the following example: suppose policy choice A is to allow a particular, fixed quota of fish to be harvested in perpetuity (i.e., a quota property right), and policy choice B is to allow some fixed proportion of the fish to be harvested each year (this proportion is called the exploitation rate). It is easy to show with practically any population or ecosystem accounting model that policy A is prone to catastrophic failure: under natural variation, the stock is bound to get low enough so that the quota looms larger and larger as a factor of change, driving the stock down faster and faster as the number of remaining fish (and hence the basis for future population growth) declines. On the other hand, policy B has built-in "feedback" to adjust harvests downward during stock declines (and hence help reverse the declines) and to take advantage of higher harvest opportunities when the stock is large. In this example, only a fool would advocate policy A, whether or not we can predict specifically what variation the future will bring.
1.2 The Distinction between Fish Science and Fisheries Science
We can provide useful predictions and advice about some kinds of management choices without resorting to precise, quantitative models that are bound to be incorrect to at least some degree. For example, it is easy to explain in qualitative terms why fixed-quota harvest policies are dangerous compared to feedback policies in which harvests are varied in response to unforeseen change. But most management decisions involve quantitative choices: How many fish should be harvested this year? What sizes of fish should be caught? How large should a protected area be? How many licenses should be issued? How much unregulated fishing effort will occur this year if a given regulation is imposed on catch or size of fish or location of fishing? How much can we harvest without "impairing" the ability of the ecosystem to support other creatures that depend on the ones we harvest?
Somebody has to provide the answers to these difficult questions, i.e., somebody has to do some quantitative modeling and prediction, whether the work is done well and systematically or instead by some seat-of-the-pants calculation. In a way, it has been really unfortunate in the historical development of fisheries management that there has been a general assumption that the right people to answer such questions are fish biologists. There have been no real professional standards in fish or fisheries biology, and a high proportion of us got into the field in the first place because we could do so without a lot of distasteful quantitative training. We were taught to study biological process and pattern from a largely qualitative perspective, and we never expected to be "bean counters." Furthermore, most of us never imagined that many of the questions that we would be asked would not even be about fish at all but would, instead, be about the behavior of people (fishers). This state of affairs is changing rapidly, with recognition that there is a lot more to fisheries science than just studying biological processes and counting fish. But a new pathology is accompanying the change: the top levels of management agencies are dominated by people with the traditional training (and cunning as institutional players) who now have to turn to younger people for help when there is no way to sidestep the difficult quantitative questions. This means that as demands for improved, quantitative management prescriptions have grown in order to deal with more complex management options and trade-offs, key fisheries managers have had to rely more and more on people and methods (modeling) that they do not understand and certainly do not trust. Such specialization of capabilities and functions leads in turn to increased opportunities for misinterpretation and misunderstanding, among all stakeholders involved in management (fishers, managers, scientists, representatives of conservation interest groups, etc.).
1.3 Approaches to Prediction of Policy Impact
Given that predictions are an inevitable part of making management choices, what options does a fishery manager have for making these predictions? Surely there are alternatives to the rather complicated mathematical modeling described in this book; indeed, there are at least five alternative approaches that can be (and have been) used. These approaches are not mutually exclusive; each uses or is derived from at least some components or results of the others.
Appeal to Conventional Wisdom and Dogma
In a surprising variety of decision situations, fisheries managers have ignored empirical data and past experience in favor of essentially dogmatic assumptions about the responses to particular policy options and system disturbances. For example, it is routine to presume that habitat alterations to natural ecosystems always cause reduced productivity (because the organisms are "adapted to" the natural circumstances). Another common assumption is that harvesting always causes a reduction in the abundance of target species, even if/when the harvesting selectively removes individuals that differentially drive away or kill other conspecifics (e.g., cannibalism). When field evidence is found that contradicts such assumptions-e.g., evidence that coho salmon may actually be enhanced by forest harvesting in some watersheds of the Pacific northwest (Holtby 1988; Thedinga et al. 1989; see discussion in Walters 1993)-this evidence is either ignored entirely or is rejected as "nonrepresentative" or "atypical." When this happens, managers are essentially indicating their willingness to behave essentially as though some principles or assumptions were equivalent to religious dogma, i.e., were impervious to scientific invalidation.
A time-honored way of making fisheries management predictions has been by simple trend extrapolation: plot the historical data, and "eyeball" alternative projections forward in time while making some intuitive guess about the likely impact of policy change on the trend. We can, of course, formalize the eyeball part of this approach by using formal time-series analysis models, but that is unlikely to produce a better result (except perhaps in multivariate systems) than the remarkable integrative and pattern-finding abilities of the human eye.
This approach has failed in modern fisheries, for a variety of reasons. (1) It is really only valid for systems that exhibit incremental, slow change; modern fisheries can change very rapidly. (2) It is easy to confuse wishful thinking with good intuition in making predictions about the effects of policy change on trends, and to keep applying small Band-Aids to gaping wounds. (3) It is all too common to use misleading trend indicators, especially catches. In any fishery, catch results from three factors: the area "swept" by fishing, the size of the stock, and the area over which the stock is distributed. So an apparently "healthy" increase in catch can mean either that the stock is healthy, that the fishing effort (the area swept by gear) has increased, or that the range occupied by the stock has decreased. It does not help matters to use catch per effort, since this commonly used trend index can even increase during stock declines due to contractions in the range area used by the fish.
Empirical Models Based on Past Experience and/or Experience with Similar Systems
For many policy issues there is a rich range of historical and spatial comparative data upon which to base predictions about the responses to any particular new circumstances. Some fish stocks (e.g., Pacific herring off British Columbia) have been severely overfished, then allowed to recover, so that we have good information about likely stock response as a function of stock size. There are large data sets on how lakes and coastal areas respond to eutrophication, and strong regression relationships have been found between nutrient loading and performance measures such as chlorophyll concentration, so that the likely response in almost any new case can be "interpolated" from the regressions. For fish populations that are maintained through artificial stocking (hatcheries), there are large data sets on the effects of factors such as time and size of release and stocking density on performance measures such as survival rate and growth.
Unfortunately, most of the important policy issues in fisheries today involve options and performance measures for which there are no historical precedents. We have not yet tried to manage aquatic ecosystems in any holistic way, and in particular, we have not systematically gathered information on the abundances and spatial distributions of the wide variety of organisms (beyond harvested fish) in an expanded view of what would constitute a "healthy" managed ecosystem. Existing reviews of comparative data, e.g., May (1984) and Hall (1999), show mainly a confusing variety of fragmentary patterns.
Excerpted from Fisheries Ecology and Management by Carl J. Walters Steven J. D. Martell Copyright © 2004 by Princeton University Press. Excerpted by permission.
All rights reserved. No part of this excerpt may be reproduced or reprinted without permission in writing from the publisher.
Excerpts are provided by Dial-A-Book Inc. solely for the personal use of visitors to this web site.
LIST OF FIGURES xi
LIST OF TABLES xvii
PART ONE: CHANGING OBJECTIVES AND EMERGING ASSESSMENT METHODS 1
CHAPTER 1: Introduction 3
1.1 The Role of Predictive Models 3
1.2 The Distinction between Fish Science and Fisheries Science 5
1.3 Approaches to Prediction of Policy Impact 6
1.4 Experimental Management 9
1.5 The Ecological Basis of Sustainable Harvesting 12
CHAPTER 2: Trade-Offs in Fisheries Management 20
2.1 Trade-Off Relationships and Policy Choices 22
2.2 Short-Term versus Long-Term Values 25
2.3 Biological Diversity versus Productivity 31
2.4 Economic Efficiency versus Diversity of Employment Opportunities 37
2.5 Allocation of Management-Agency Resources 39
PART TWO: ELEMENTARY CONCEPTS IN POPULATION DYNAMICS AND HARVEST REGULATION 41
CHAPTER 3: Strategic Requirements for Sustainable Fisheries 43
3.1 Harvest Optimization Models 46
3.2 Constructing Feedback Policies 49
3.3 Feedback Policy Implementation 58
3.4 Feedback Policies for Incremental Quota Change 61
3.5 Actively Adaptive Policies 63
CHAPTER 4: Tactics for Effective Harvest Regulation 65
4.1 Tactical Options for Limiting Exploitation Rates 67
4.2 Managing the Risk of Depensatory Effects under Output Control 69
4.3 Tactics for Direct Control of Exploitation Rates 74
4.4 Regulation of Exploitation Rates in Recreational Fisheries 77
4.5 In-Season Adaptive Management Systems 79
4.6 Monitoring Options and Priorities 80
4.7 Maintaining Genetic Diversity and Structure in Harvested Populations 83
PART THREE: USE AND ABUSE OF SINGLE-SPECIES ASSESSMENT MODELS 87
CHAPTER 5: An Overview of Single-Species Assessment Models 89
5.1 Objectives of Single-Species Assessment 89
5.2 State-Observation Components 91
5.3 Estimation Criteria and Measuring Uncertainty 95
5.4 Modeling Options 101
5.5 Using Composition Information 110
5.6 Dealing with Parameters That Aren't 121
CHAPTER 6: Foraging Arena Theory (I)124
6.1 Beverton-Holt Model for Stock-Recruitment 128
6.2 Alternative Models Based on Juvenile Carrying Capacity 132
6.3 Using Foraging Arena Arguments to Derive the Beverton-Holt Model 136
6.4 Implications for Recruitment Research and Prediction 147
CHAPTER 7: Problems in the Assessment of Recruitment Relationships 151
7.1 Which Parameters Matter? 152
7.2 Predicting Reproductive Performance at Low Stock Sizes 158
7.3 Predicting Capacity to Recover from Historical Overfishing 160
7.4 The Errors-in-Variables Bias Problem 162
7.5 The Time-Series Bias Problem 165
7.6 Can Statistical Fisheries Oceanography Save the Day? 173
PART FOUR: MODELING SPATIAL PATTERNS AND DYNAMICS IN FISHERIES 179
CHAPTER 8: Spatial Population Dynamics Models 181
8.1 Life-History Trajectories 182
8.2 Multistage Models 185
8.3 Eulerian Representation 188
8.4 Lagrangian Representation 193
8.5 Policy Gaming with Spatial Models 198
CHAPTER 9: Temporal and Spatial Dynamics of Fishing Effort 200
9.1 Long-Term Capacity 201
9.2 Short-Term Effort Responses 204
9.3 Spatial Allocation of Fishing Effort 210
9.4 Mosaic Closures 223
PART FIVE: FOOD WEB MODELING TO HELP ASSESS IMPACT OF FISHERIES ON ECOLOGICAL SUPPORT FUNCTIONS 229
CHAPTER 10: Foraging Arena Theory (II)231
10.1 Understanding Foraging Arena Theory 232
10.2 Predicting Trophic Flows 236
10.3 Adding Realism (I): Foraging Time Adjustments 240
10.4 Adding Realism (II): Trophic Mediation 244
10.5 Ecosim 246
10.6 Representing Trophic Ontogeny in Ecosim 248
10.7 Single-Species Dynamics from Ecosim Rate Equations 252
10.8 Ecosystem-Scale Variation 254
CHAPTER 11: Options for Ecosystem Modeling 256
11.1 Qualitative Analysis of Dominant Trophic Interactions 259
11.2 Qualitative Analysis of More Complex Linkages 270
11.3 Models That Link Dynamics with Nutrient Cycling Processes 271
11.4 Representation of Mesoscale Spatial-Policy Options 276
11.5 Individual-Based Size-and Space-Structured Models 283
CHAPTER 12: Parameterization of Ecosystem Models 286
12.1 Parameterizing Models 287
12.2 Parameter Estimates from Experimental Data 289
12.3 Estimating Parameters from Mass Balance Snapshots 292
12.4 Challenging Ecosystem Models with Data 300
PART SIX: STRATEGIES FOR ECOSYSTEM MANAGEMENT 311
CHAPTER 13: Marine Enhancement Programs 313
13.1 Things That Can Go Wrong 317
13.2 Critical Steps in Enhancement Program Design 326
13.3 Monitoring and Experimental Requirements 331
CHAPTER 14: Options for Sustainable Ecosystem Management 334
14.1 Alternative Visions of Ecosystem Structure 335
14.2 Moving Toward Sustainable Ecosystem Management 344
APPENDIX: Definitions for Mathematical Symbols 349