Cover -- Half-title -- Title -- Copyright -- Dedication -- Contents -- Preface and Acknowledgements -- Part I Introduction to the nature of monitoring problems and to rivers -- 1 Why we need well-designed monitoring programs -- 1.1 HUMAN PRESSURES ON FLOWING WATERS -- 1.2 THE NEED FOR THIS BOOK -- 1.3 THE SCOPE, APPROACH AND INTENDED AUDIENCES OF THIS BOOK -- 1.4 THE STRUCTURE OF THE BOOK AND THE PURPOSE OF EACH OF THE CHAPTERS -- 1.5 IMPORTANT ISSUE -- 2 The ecological nature of flowing waters -- 2.1 RIVERS AND THEIR CATCHMENTS -- 2.2 THE BIOTA OF RIVERS AND STREAMS -- 2.3 CONCEPTS OF RIVER STRUCTURE AND FUNCTIONING -- 2.4 ISSUES OF SCALE AND PATCHINESS IN FLOWING WATERS -- 2.5 IMPORTANT ISSUES -- 3 Assessment of perturbation -- 3.1 TYPES OF DISTURBANCE -- 3.2 THE PURPOSES OF MONITORING -- 3.2.1 To assess the ecological state of ecosystems -- 3.2.2 To assess whether regulated performance criteria have been exceeded -- 3.2.3 To detect and assess the impacts of human-generated disturbance(s) -- 3.2.4 To assess the responses to restoration efforts -- 3.3 IMPORTANT ISSUES -- Part II Principles of inference and design -- 4 Inferential issues for monitoring -- 4.1 SAMPLING -- 4.2 UNCERTAINTY AND PROBABILITY -- 4.3 VARIABLES -- 4.4 ESTIMATION -- 4.5 STATISTICAL MODELS -- 4.5.1 Regression models -- 4.5.2 Analysis of variance (ANOVA) models -- 4.5.3 Fitting models -- 4.5.4 Comparing models -- 4.6 ANALYSES OF VARIANCE (ANOVA) -- 4.6.1 Type of factors -- 4.6.2 Partitioning the variance -- 4.7 HYPOTHESIS-TESTING: CLASSICAL APPROACH -- 4.8 HYPOTHESIS-TESTING: THE BAYESIAN APPROACH -- 4.9 ASSUMPTIONS OF STATISTICAL ANALYSES OF MONITORING PROGRAMS -- 4.10 UNIVARIATE AND MULTIVARIATE ANALYSIS -- 4.11 IMPORTANT ISSUES -- 5 The logical bases of monitoring design -- 5.1 CLASSES OF MONITORING -- 5.2 MONITORING TO DETECT HUMAN IMPACTS ON THE ENVIRONMENT.

5.2.1 Detecting change -- 5.2.2 Discriminating impacts from natural changes -- 5.3 BACI DESIGNS -- 5.3.1 Natural dynamics and the duration of monitoring -- 5.3.2 Spatial variation and multiple locations -- 5.3.3 Asymmetry in impact assessment monitoring -- 5.4 SCALES OF IMPACT AND MONITORING -- 5.4.1 Sampling within locations - impacts on status -- 5.4.2 Sampling within locations - impacts on variation -- 5.4.3 Sampling within periods - duration and fluctuations in impact -- 5.4.4 Collecting the samples -- 5.4.5 Other considerations -- 5.5 CAREFUL DESIGN IS THE MOST IMPORTANT STEP -- 5.6 IMPORTANT ISSUES -- 6 Problems in applying designs -- 6.1 A BRIEF HISTORICAL SKETCH -- 6.2 PROBLEMS INHERENT IN THE NATURE OF RIVERS -- 6.2.1 Interdependence between locations -- 6.2.2 Variation in time -- 6.2.3 Logistic and technical issues -- 6.3 PROBLEMS ARISING FROM THE TYPES OF VARIABLES USED -- 6.3.1 Variation and imprecision -- 6.3.2 Physicochemical variables as surrogates for biological variables -- 6.3.3 Univariate biological variables -- 6.3.4 Multivariate response variables -- 6.4 SOCIAL, INSTITUTIONAL AND POLITICAL ISSUES -- 6.4.1 Difficulties caused by different backgrounds -- 6.4.2 Insufficient lead time for pre-impact monitoring -- 6.5 IMPORTANT ISSUES -- 7 Alternative models for impact assessment -- 7.1 BACKGROUND OF APPROACHES -- 7.1.1 BACI -- BACIP -- Intervention Analysis -- 7.1.2 MBACI -- 7.1.3 Beyond-BACI -- 7.2 THESE APPROACHES ARE DIFFERENT! -- 7.2.1 Why it matters -- 7.3 FORMAL SAMPLING AND ANALYTICAL FRAMEWORK -- 7.3.1 The sampling program -- BACIP -- MBACI -- Beyond-BACI -- 7.3.2 The analytical models and formal hypotheses -- BACIP -- MBACI -- A special case: MBACI with a single Before and single After sample -- Beyond-BACI -- 7.3.3 Tests for Impact -- BACI and BACIP -- MBACI -- Beyond-BACI -- 7.4 POWER CONSIDERATIONS -- 7.4.1 BACI and BACIP.

7.4.2 MBACI -- 7.4.3 Beyond-BACI -- 7.5 DETECTING MORE SUBTLE EFFECTS -- 7.6 EXTENT OF IMPACTS -- 7.7 FLEXIBLE ANALYSIS/INFLEXIBLE HYPOTHESIS -- 7.8 IMPORTANT ISSUES -- Part III Applying principles of inference and design -- 8 Applying monitoring designs to flowing waters -- 8.1 SPATIAL VARIATION AND THE LOCATION OF CONTROLS -- 8.1.1 The nature of controls -- 8.1.2 Spatial extent and nature of impact -- 8.1.3 Finding control locations -- Criteria for controls -- The dilemma of the trade-off in similarity and number of controls -- Statistical independence and location of controls -- Ensuring control locations are free of the human impact -- Spatial confounding, environmental differences and location of controls -- The relative significance of problems with controls -- How many control locations are necessary? -- 8.1.4 Subsampling of locations -- 8.1.5 Examples of decision trees for finding and choosing controls -- 8.2 TEMPORAL VARIATION, AND BEFORE AND AFTER SAMPLING -- 8.2.1 Temporal extent and nature of impact -- 8.2.2 Frequency of sampling within Periods -- 8.2.3 Subsamples within Times -- 8.2.4 Statistical independence and sampling through time -- 8.3 DOING THE SAMPLING -- 8.4 A WORKED EXAMPLE - EFFECTS OF LIMING TO DECREASE ACIDITY -- 8.4.1 Background to the problem and preliminary data -- 8.4.2 Selection of control and impact locations -- 8.4.3 Predictions and data collection -- 8.4.4 Results and analysis -- 8.5 IMPORTANT ISSUES -- 9 Inferential uncertainty and multiple lines of evidence -- 9.1 A BRIEF REVISIT OF INFERENTIAL UNCERTAINTY AND PROBABILITY -- 9.2 A LEVELS-OF-EVIDENCE APPROACH -- 9.3 A SUGGESTED STEP-BY-STEP GUIDE TO USING A LEVELS-OF-EVIDENCE APPROACH -- 9.3.1 Defining and quantifying causal criteria -- Strength of association -- Consistency of association -- Specificity of association -- Temporality.

Biological or ecological gradient -- Biological or ecological plausibility -- Experimental evidence -- Analogy -- 9.3.2 Building a 'levels-of-evidence' case for changes associated with the human impact -- 1. Set down the characteristics of the human activity -- 2. Set down the characteristics of the impact location -- 3. Clarify the question(s) -- 4. Decide how an effect will be considered to have been 'detected' -- 5. Decide upon the qualities of studies to be included in the review -- 6. Carry out a broad-ranging review, extracting relevant data -- 9.3.3 Collating common sources of natural variance in the response variables -- 9.3.4 Cataloguing effects -- 9.3.5 Predictions and ways of ruling out alternative explanations -- After impact with data from the impact location only -- 9.3.6 Assessing the predictions -- 9.4 SOME FINAL COMMENTS ON THE PROCESS -- 9.5 IMPORTANT ISSUES -- 10 Variables that are used for monitoring in flowing waters -- 10.1 CONSIDERATIONS FOR CHOOSING VARIABLES -- 10.1.1 Questions addressed by the monitoring program -- 10.1.2 Causality, mechanisms, inference -- 10.1.3 Ecological and socioeconomic significance of change -- 10.1.4 Efficiency -- 10.2 RELATIVE WEIGHTING OF ATTRIBUTES -- 10.3 IMPORTANT ISSUES -- 11 Defining important changes -- 11.1 WHY DO WE NEED TO DEFINE CHANGES IN TERMS OF 'EFFECT SIZES'? -- 11.2 KINDS OF CHANGE, RISKS AND CONSEQUENCES -- 11.3 PRACTICAL STEPS, AND SOME DIFFICULTIES, IN SETTING AN EFFECT SIZE -- 11.3.1 A caricature of how this seems to work for drinking water -- 11.3.2 Quantifying the relationship between the response variable and the potential impact -- 11.3.3 Negotiating about values, risks and consequences -- 11.4 IMPORTANT ISSUES -- 12 Decisions and trade-offs -- 12.1 MAKING STATISTICAL DECISIONS -- 12.2 BALANCING TYPE I AND TYPE II ERRORS -- 12.2.1 Fixed Alpha, adjust n.

12.2.2 Fixed Beta, adjust n and Alpha -- 12.2.3 Scalable decision criteria -- 12.3 COST-BENEFIT ANALYSIS AND DESIGN -- 12.4 FURTHER VARIATIONS ON BALANCED DECISION-MAKING -- 12.5 IMPORTANT ISSUES -- 13 Optimization -- 13.1 WHAT WE MEAN BY OPTIMIZATION -- 13.2 BY NOW YOU SHOULD HAVE… -- 13.3 YOU WILL NEED AN ESTIMATE OF VARIANCE -- 13.3.1 Sources of variance estimates -- 13.4 DEVELOPING AN IDEALIZED SAMPLING SCHEME… -- 13.4.1 Form of output -- 13.5 TRADING OFF -- 13.5.1 Spend more -- 13.5.2 Live with increased risk -- 13.5.3 Maintaining the risk, reducing the cost -- Eliminating variables -- Sampling more cheaply -- 13.5.4 Accepting larger effect sizes -- 13.6 UNCERTAINTY IN OPTIMIZATION -- 13.6.1 Origins of uncertainty -- 13.6.2 Incorporating capacity for readjusting the sampling program -- 13.7 POST-MONITORING 'OPTIMIZATION': IMPLICATIONS FOR DECISION CRITERIA -- 13.8 A WORKED EXAMPLE - LIMING TO DECREASE ACIDITY OF STREAMS -- 13.8.1 Nomination of an important effect size -- 13.8.2 Deciding the relative costs of Type I and II errors -- 13.8.3 Deciding the actual probability of errors -- 13.8.4 Use of pilot data and power analysis to examine the number of locations needed in the monitoring program -- 13.8.5 Trading off costs and risks -- Spend more money? -- Live with increased risks? -- Reduce the cost of sampling? -- 13.8.6 Implications from this example -- 13.9 IMPORTANT ISSUES -- 14 The special case of monitoring attempts at restoration -- 14.1 ISSUES CONCERNING THE STUDY OF ECOLOGICAL RESTORATION -- 14.2 CAN BACI DESIGNS BE APPLIED TO ECOLOGICAL RESTORATION? -- 14.2.1 The real way that restoration differs -- 14.3 ANALYTICAL TECHNIQUES APPLICABLE TO RESTORATION MONITORING -- 14.3.1 The logic of specifying an effect size for recovery -- 14.4 HOW LONG SHOULD WE MONITOR ATTEMPTS AT RESTORATION?.

14.5 THE NEED FOR CLARITY IN DECLARING THE GOALS OF RESTORATION.