The Dynamics of Shoaling in Zebrafish by Noam Y. Miller A thesis submitted in conformity with the requirements for the degree of Doctor of Philosophy Graduate Department of Psychology University of Toronto © Copyright by Noam Miller 2010 The Dynamics of Shoaling in Zebrafish Noam Miller Doctor of Philosophy Graduate Department of Psychology University of Toronto 2010 Abstract A wide array of species, from ants to humans, live or forage in groups. Shoaling – the formation of groups by fish – confers protection from predation and enhances foraging. However, little is known about the detailed characteristics or the dynamics of shoaling. Shoaling is a complex social interaction and a better understanding of its mechanisms and limitations would permit the study of natural and induced changes on social behavior generally in fish. Here, I present data on the shoaling characteristics of zebrafish (Danio rerio). Novel tracking techniques are used to extract detailed trajectories of all members of a free-swimming shoal of zebrafish. Multiple measures of shoaling – such as distributions of nearest neighbor distances, shoal polarizations, and speeds – are calculated, to better describe the subtleties of the behavior including, for the first time, the high resolution spatio-temporal dynamics of shoaling. In addition, a novel criterion is introduced to determine when and how individual fish or sub-groups leave the shoal. Comparisons are presented between the shoaling characteristics of three populations of zebrafish (LFWT, SFWT, AB) and between days and hours of repeated exposure to the same testing environment, demonstrating the gradual effects of habituation on shoaling. In addition, the effects of manipulating the number of fish in the shoal, hunger levels, and predation threat are also examined, lending empirical support to ecological theories on the adaptive functions of the behavior. Finally, the data are compared to two leading theoretical models of shoaling and a ii novel simulation approach is suggested. The data strongly suggest that various aspects of shoaling in zebrafish are constantly changing, complex, and flexible, representing a dynamic form of social cognition. The study of these characteristics sheds much-needed light on complex social interactions in this popular genetic model organism, which may eventually lead to a better understanding of social behaviors in other species, including our own. iii Table of Contents Page Chapter 1: Introduction 1 1.1 Natural history of zebrafish 3 1.2 Behavioral ecology of shoaling 5 1.3 Existing data on shoaling 7 1.4 Mechanisms of shoaling 11 1.5 Comparative studies 12 1.6 Experimental measures 13 1.7 Dynamics of shoaling 15 1.8 Outline 16 Figures 17 Chapter 2: Methods 19 2.1 General methods 20 2.2 Tracking 22 2.3 Analysis 29 Chapter 3: Coming and Going 35 3.1 Existing measures 35 3.2 The current measure 38 3.3 Results 42 3.4 Discussion 50 Tables and Figures 53 Chapter 4: Characterizing Zebrafish Shoaling 60 4.1 Distributions of basic measures 61 iv 4.2 Correlations between measures 64 4.3 Dynamics of shoaling measures 65 4.4 Discussion 67 Figures 73 Chapter 5: Ecological Considerations 80 5.1 Experiment 3: the effect of N 82 5.2 Experiment 4: the effect of food 86 5.3 Experiment 5: the effect of a predator 91 5.4 Discussion 95 Figures 98 Chapter 6: Testing Models of Shoaling 107 6.1 Methods for testing models against data 109 6.2 Testing existing models 110 6.3 A modest proposal 119 6.4 Discussion 125 Figures 127 Chapter 7: Summary and Conclusions 137 Figures 143 References 146 Appendix A: Equations and Formulae 159 Appendix B: Additional Figures and Tables 163 Copyright Acknowledgements 180 v List of Tables Table 3.1. Cross-correlations and results of PCA on excursion-related measures of shoaling. Table B.1. Results of KS tests on excursion duration by day in Experiment 1. Table B.2. Results of KS tests on excursion type by excursion duration in Experiment 1. Table B.3. Results of KS tests on SGS by testing day for Experiment 1. Table B.4. Results of KS tests on distance measures by day for Experiment 1. Table B.5. Results of KS tests on speed by day for Experiment 1. Table B.6. Results of KS tests on polarization by day for Experiment 1. Table B.7. Results of KS tests on all unimodal measures by day for Experiment 3. Table B.8. Results of KS tests on polarization distributions by day for Experiment 3. vi List of Figures Figure 1.1. Explanation of NND and IID. Figure 3.1. Procedure for calculating the group membership criterion. Figure 3.2. Mean number of excursions per session by day for each of the three strains in Experiment 1 (A) and by hour for Experiment 2 (B). Figure 3.3. Density distributions of excursion duration by day for the SF shoals in Experiment 1 (A) and by hour for Experiment 2 (B). Figure 3.4. Density distributions of the summed group size (SGS) by day for the SF shoals in Experiment 1 (A) and by hour for Experiment 2 (B). Figure 3.5. Density distributions of position in the shoal before the beginning of an excursion, by strain for Experiment 1. Figure 4.1. Density distributions of NND (A), IID (B), speed (C), and polarization (D) by day for Experiment 1. Figure 4.2. Density distributions of NND (A), IID (B), speed (C), and polarization (D) by hour for Experiment 2. Figure 4.3. Density distribution of IID including all fish (blue) and excluding fish on excursions (red) for Experiment 1. Figure 4.4. Density distributions of speed (A) and polarization (B) by strain for Experiment 1. Figure 4.5. Density distributions of bearing to nearest neighbor by strain for Experiment 1. vii Figure 4.6. Modes of polarization distributions by day for all strains in Experiment 1, separated into schooling (open) and shoaling (solid) components. Figure 5.1. The effects of shoal size on zebrafish shoaling. Figure 5.2. Modes of polarization distribution components by shoal size. Figure 5.3. The effects of food and hunger on zebrafish shoaling. Figure 5.4. Density distributions of excursion duration (A), NND (B), speed (C), and polarization (D) for the final, probe, day for all groups of Experiment 4. Figure 5.5. Modes of polarization distribution components for Experiment 4. Figure 5.6. The effects of a predator image on zebrafish shoaling. Figure 5.7. Density distributions of excursion duration (A), NND (B), speed (C), and polarization (D) for the final, probe, day for all groups of Experiment 5. Figure 6.1. The zones in Couzin et al.’s (2002) model. Figure 6.2. Viscido et al.’s (2004) AAR function. Figure 6.3. Error distributions for Couzin et al.’s model for different values of the radius of zoo. Figure 6.4. Error distributions by day for Couzin et al.’s model. Figure 6.5. Error distributions by rank for Couzin et al.’s model. Figure 6.6. Error distributions for Viscido et al.’s model for different values of δm. Figure 6.7. Error distributions by day for Viscido et al.’s model. viii Figure 6.8. Error distributions by rank for Viscido et al.’s model. Figure 6.9. Model suggestion measure distributions. Figure 7.1. Spider-web graphs of all shoaling measures by strain for Experiment 1. Figure 7.2. Spider-web graphs of all shoaling measures by day, for SF fish in Experiment 1. Figure B.1. Experimental setup for Experiments 1, 2, 3, and 4. Figure B.2. Experimental setup for Experiment 5. Figure B.3. Density distributions of excursion durations by strain for Experiment 1. Figure B.4. Density distributions of excursion durations by excursion type for Experiment 1. Figure B.5. Density distributions of individual speed in the 1 sec before (blue) and the first sec of (red) an excursion in Experiment 1. Figure B.6. Density distributions of NND by day for LF fish in Experiment 1. Figure B.7. Density distributions of periodic oscillations in shoaling measures by strain for Experiment 1. ix List of Appendices Appendix A: Equations and Formulae. Appendix B: Additional Figures and Tables. x 1 Chapter 1: Introduction1 Many species, from ants to humans, live or forage in groups (review in Sumpter, 2006). Amongst fishes, shoaling behavior is exhibited by almost half of all species at some point in their lives (Shaw, 1978) and has been proposed to confer multiple advantages in foraging and avoiding predation (see below; Krause & Ruxton, 2002). Shoaling and other group behaviors (herding, flocking) have been intensively studied at the ecological and evolutionary levels (Krause & Ruxton, 2002) but far less is known about the mechanisms underlying these behaviors. This imbalance is at least partially due to the difficulty in recording the detailed individual trajectories required for mechanistic analyses. Several research groups, including our own, have recently succeeded in gathering such data under carefully controlled conditions (e.g. Lukeman et al., 2010; Ballerini et al., 2008a, 2008b; Delcourt et al., 2009; Viscido et al., 2004). Using these detailed data, the studies presented here aim to characterize the shoaling behavior of zebrafish (Danio rerio). Zebrafish have become a favorite behavior genetic model organism (Gerlai, 2003). They are easy to breed and house in large numbers and mature quickly (see below) and have thus been suggested as an efficient complement to other vertebrate models such as rats and mice (Guo, 2004). As such, and due to the wide array of genetic tools available for zebrafish, insights into their behavior can potentially be linked with relative ease to genetic and physiological mechanisms. Though findings on zebrafish behavior are accumulating rapidly (Miklosi & Andrew, 2006), a thorough characterization of zebrafish shoaling is still lacking.