Just Keeping Swimming

Permanent Link:

Material Information

Title: Just Keeping Swimming A Review of the Biological and Social Components of Teleost Fish Shoaling
Physical Description: Book
Language: English
Creator: Strock, Stuart
Publisher: New College of Florida
Place of Publication: Sarasota, Fla.
Creation Date: 2011
Publication Date: 2011


Subjects / Keywords: Fish
Genre: bibliography   ( marcgt )
theses   ( marcgt )
government publication (state, provincial, terriorial, dependent)   ( marcgt )
born-digital   ( sobekcm )
Electronic Thesis or Dissertation


Abstract: Shoaling is a very common behavior among fish species; approximately 50% of all fish species engage in schooling behavior at one or more stages of their life. Extensive research has investigated the nature of this behavior over the past century, but the past 40 years have offered considerable progress in the understanding the evolutionary, sensory and social components of shoaling behavior. This comprehensive review encompasses a wide spectrum of schooling research from the last half century to the present. It begins with an analysis of the evolutionary selection pressures acting for or against the continued expression of shoaling behavior in fishes, and extends to an exploration of the inheritance and dynamic adaptation of shoaling, the role of senses in school cohesion dynamics, and a further analysis of the way fish select their groups, based on an array of acting social pressures. Finally, the research itself is reviewed and possible avenues for future research are considered.
Statement of Responsibility: by Stuart Strock
Thesis: Thesis (B.A.) -- New College of Florida, 2011
Supplements: Accompanying materials: DVD (4)
Bibliography: Includes bibliographical references.
Source of Description: This bibliographic record is available under the Creative Commons CC0 public domain dedication. The New College of Florida, as creator of this bibliographic record, has waived all rights to it worldwide under copyright law, including all related and neighboring rights, to the extent allowed by law.
Local: Faculty Sponsor: Beulig, Alfred

Record Information

Source Institution: New College of Florida
Holding Location: New College of Florida
Rights Management: Applicable rights reserved.
Classification: local - S.T. 2011 S9
System ID: NCFE004460:00001

Permanent Link:

Material Information

Title: Just Keeping Swimming A Review of the Biological and Social Components of Teleost Fish Shoaling
Physical Description: Book
Language: English
Creator: Strock, Stuart
Publisher: New College of Florida
Place of Publication: Sarasota, Fla.
Creation Date: 2011
Publication Date: 2011


Subjects / Keywords: Fish
Genre: bibliography   ( marcgt )
theses   ( marcgt )
government publication (state, provincial, terriorial, dependent)   ( marcgt )
born-digital   ( sobekcm )
Electronic Thesis or Dissertation


Abstract: Shoaling is a very common behavior among fish species; approximately 50% of all fish species engage in schooling behavior at one or more stages of their life. Extensive research has investigated the nature of this behavior over the past century, but the past 40 years have offered considerable progress in the understanding the evolutionary, sensory and social components of shoaling behavior. This comprehensive review encompasses a wide spectrum of schooling research from the last half century to the present. It begins with an analysis of the evolutionary selection pressures acting for or against the continued expression of shoaling behavior in fishes, and extends to an exploration of the inheritance and dynamic adaptation of shoaling, the role of senses in school cohesion dynamics, and a further analysis of the way fish select their groups, based on an array of acting social pressures. Finally, the research itself is reviewed and possible avenues for future research are considered.
Statement of Responsibility: by Stuart Strock
Thesis: Thesis (B.A.) -- New College of Florida, 2011
Supplements: Accompanying materials: DVD (4)
Bibliography: Includes bibliographical references.
Source of Description: This bibliographic record is available under the Creative Commons CC0 public domain dedication. The New College of Florida, as creator of this bibliographic record, has waived all rights to it worldwide under copyright law, including all related and neighboring rights, to the extent allowed by law.
Local: Faculty Sponsor: Beulig, Alfred

Record Information

Source Institution: New College of Florida
Holding Location: New College of Florida
Rights Management: Applicable rights reserved.
Classification: local - S.T. 2011 S9
System ID: NCFE004460:00001

This item is only available as the following downloads:

Full Text


JUST KEEP SWIMMING: A REVIEW OF THE BIOLOGICAL AND SOCIAL COMPONENTS OF TELEOST FISH SHOALING BY STUART STEVEN STROCK JR. A Thesis Submitted to the Division of Natural Sciences New College of Florida In partial fulfillment of the require ments for the degree Bachelor of Arts Under the sponsorship of Dr. Alfred Beulig Jr. Sarasota, Florida January, 2011


J ust Keep Swimming !! ACKNOWLEDGEMENTS I would like to thank everyone who told me to keep g oing when I was ready to give up This thesis is dedicated to you. To my dearest friend, Pat Dees. You led the way, followed your dreams, and encouraged me to follow mine. To Ian Haywood. You knew my frustration and listened to me complain with the patience of a saint. To Alton Labrecque. You helped me with statis tics and sent me funny pictures, that means more than you might think. To David Banks. You were the best roommate I!ve ever had, and you!ve broadened my mind more than anyone else I!ve ever met. To Calvin Gainey. You distracted me at every opportunity, and you just might!ve maintained my sanity that way. To Katie McAuley, for teaching me the ways of awesome and to smash the patriarchy. To Ron Overing III. Thank you for being my friend and teaching me to watch for traps. To Mike Marazzi. I wish I!d met you s ooner, you shifty bastard. To my two great loves, you know who you are. Thank you for everything. To all of you who have ever been my friend and believed in me. To New College of Florida, for shaping me into the person I am today. To my committee, Alfred Beuilg, Leo Demski, Gordon Bauer and Wendy Bashant. And finally to you, Brian Lee. Thank you for being my benefactor, my neighbor, and my friend. Without you, none of this would have happened.


J ust Keep Swimming !!! Table of Contents Acknowledgements ii Abs tract v Introduction 1 Chapter 1: Se lection Pressures For and Against Schooling Behavior 3 A. Primary Benefits of Schooling Behavior 3 i. Vigilance and Predation Protection 3 ii. Predation Protection: The Importance of Conformity 7 iii. Increased Foraging Efficiency 9 B. Primary Costs of Schooling Behavior 14 i. The Oddity Effect 14 ii. Competition 18 C. Secondary Benefit s of Schooling Behavior 24 i. Learning 2 4 ii. Sleep Mitigation 30 D. Secondary Costs of Schooling Behavior 3 2 i. Parasites 3 2 Chapter 2: Inheritance and Ada ptation of Schooling Behaviors 37 A. Genetically Inherited Behavioral Phenotypes 3 7 B. Indirect Genetic Effects 41 C. Epigenetic effects A Theoretical Proposal 4 5 Chapter 3. The Senses and Their Effect on Schooling Dynamics 50


J ust Keep Swimming !# A. The Acoustico Lateralis Sense: The Lateral Line 50 B. The Roles of the Acoustico Lateralis Sen se and Vision in Conjunction 59 C. The Influential Zones: A Threshold of Desir ed Personal Space 64 Chapter 4. The Selfish Herd and Methods of School Choice 67 A. Hamilton!s "Selfish Herd! Model and the Nearest Neighbor Approach Strategy 68 B. Choosing the Right School: Fish Seek C ommonality in Shoal Ma tes 79 i. Choosing the Right School: Visual Homogeny 80 ii. Choosing the Right School: Group Size 85 iii. Choosing the Right School: Olfactory Diet Cu es 89 iv. Choosing the Right School: Familiarity 94 v. Choosing the Right School: Gauging F itness of Conspecifics 98 vi. Choosing the Right School: Sex biased Scho oling 104 Di scussi on 109 A. Cost Benefit Ana lysis 109 B. The Senses 111 C. Adaptive Shoal C hoice 11 3 D. Topics of Interest and Suggestions for Future Re search 114 Appendix A A Look at the Cell Cycle, Steps to a Theory of Somatic Variation in Neuronal Gen omes 117 Appendix B Thoughts on Collective Fitness 124 References 1 25


J ust Keep Swimming # JUST KEEP SWIMMING: A REVIEW OF THE BIOLOGICAL AND SOCIAL COMPONENTS OF FISH SHOALING Stuart Strock New College of Florida, 2011 ABSTRACT Shoaling is a very common behavior among fish species; approximately 50% of all fish spec ies engage in schooling behavior at one or more stages of their life. Extensive research has investigated the nature of this behavior over the past century, but the past 40 years have offered considerable progress in the understanding the evolutionary, sen sory and social components of shoaling behavior. This comprehensive review encompasses a wide spectrum of schooling research from the last half century to the present. It begins with an analysis of the evolutionary selection pressures acting for or against the continued expression of shoaling behavior in fishes, and extends to an exploration of the inheritance and dynamic adaptation of shoaling, the role of senses in school cohesion dynamics, and a further analysis of the way fish select their groups, based on an array of acting social pressures. Finally, the research itself is reviewed and possible avenues for future research are considered. ____________________________ Dr. Alfred Beulig Division of Natural Sciences


"Of shes, such as swim in shoals together are friendly to one another; such as do not so swim are enemies." Aristotle, 350 B.C. Introduction !Shoaling, when applied to sh, is dened as staying together for social reasons. Schooling occurs when these shoaling sh begin swimming in the same direction in a coordinated manner. Therefore, by denition, sh that school also shoal, but not all shoaling sh school. Furthermore, schooling sh do not necessarily always school; schooling can be initiated as a response to a stimulus, usually the presence of a perceived threat (Parrish, Viscido and GrŸnbaum, 2002). These sh spend much of the rest of their time foraging or seeking mating opportunities within the group while shoaling. Fish that do consistently school are primarily those that travel long distances together over the open ocean. Shoaling is exhibited by over 50% of all studied species of sh at some point in their lives, and at least 25% of studied sh shoal for their entire lives (Shaw, 1978). !The rigorous study of shoaling began with Charles Breder in 1926 and his book The Locomotion of Fishes Since the publishing of that text, the study of shoaling behavior has gained the attention of researchers in several countries. Articles related to schooling behavior have been published in numerous academic journals, and many have been translated and distributed to countries all over the world. Outside of academic circles, images of sh schools have become synonymous with the beauty of the ocean. Educational television programs that take their cameras beneath the waves almost always feature shots Just Keep Swimming 1


of colorful, shoaling reef sh. At great personal expense, tourists from every major country in the world travel to tropical reefs to see these sh, and entire coastal economies rely on this tourism to sustain themselves. !Studies of shoaling behavior occur in the eld and the laboratory. Those that occur in the eld are primarily observation experiments, meant to record the responses of shoaling sh to a plethora of stimuli. These studies are expensive and cumbersome, though, as researchers can only observe shoals for as long as their air tanks allow them to stay underwater. As a result, many researchers opt to conduct research on well-known shoaling and schooling species, testing the specic responses of these specimens to specic situations. With advances in computational software in recent years, many studies even forego the use of actual test subjects, instead using what is already known of schooling mechanics to construct predictive mathematical models that simulate different schooling actions based on varying environmental conditions (Breder, 1954; Hensor, Couzin, James and Krause, 2005; Ward, Sumpter, Couzin, Hart and Krause, 2008). !The purpose of this thesis is to provide a comprehensive review of the selection pressures, biological mechanisms and social pressures of sh shoaling, using a broad cross-section of the last 40 years of research to accomplish this goal. It begins with an analysis of the evolutionary selection pressures acting for or against the continued expression of shoaling behavior in shes. What follows is an exploration of the inheritance and dynamic adaptation of shoaling, the role of senses in school cohesion dynamics, and a further analysis of the way sh Just Keep Swimming 2


select their groups, based on an array of acting social pressures. Finally, the research itself is reviewed and possible avenues for future research are considered. Chapter 1. Selection Pressures For and Against Schooling Behavior !Social behaviors have evolved over millions of years and are exhibited by a plethora of organisms. Shoaling behavior is no different. It is not difcult to draw parallels between schooling and herding; as with herds, schools of sh derive many benets from sticking together. Conversely, banding together can also bring about a new set of problems unique to this maintained closeness. Cost-benet analysis of schooling behavior can help to shed light on some of the factors that contributed to the evolutionary perpetuity of sh schooling, as well as the drawbacks that have arisen in the process. A. Primary Benets of Schooling Behavior i. Vigilance and Predation Protection !Arguably the most important benet of schooling is that of protection from predators, as predation often results in the death of individuals and the end of their genetic lines. Schools obtain this benet by three means: diluting the risk of any particular individual"s mortality with sheer numbers, or the #dilution effect" (Foster and Treherne, 1981); increasing chances that a nearby predator will be noticed by at least one school member the #detection effect" or #many eyes" theory (Dehn, 1990); and limiting a predator"s ability to target any particular individual through visible group homogeny, also known as the #confusion effect" (Landeau and Terborgh, 1986). Groups that school cohesively often utilize Just Keep Swimming 3


these effects in conjunction, and reap compounded anti-predatory benets as a result. !Dehn (1990) found that the detection and dilution effects were both important factors in models designed to determine the chance of survival as it related to frequency of vigilance against predators over increasing group size. Previously, models had commonly only considered the detection effect. Dehn presented a model considering the detection effect, and then created a "security model," which considered the detection and dilution effects together. The security model showed that, while dilution and detection are both important, detection becomes less important as group size increases. Detection initially provided a great deal of security to small groups (70% more security than dilution alone in groups of 2), but its added security decreased over group size (18% more security than dilution alone in groups of 5). Dehn noted that as group size increased school members tended to grow less vigilant, but never abandoned vigilance completely. Clearly, vigilance is still essential to the survival of large groups, but its urgency is lessened. !Roberts (1996) reviewed changes in vigilance over varying group size with respect to multiple factors. According to the #encounter effect," or #predation risk hypothesis," larger groups are likely to be spotted more readily by predators. As noted before, however, the dilution effect suggests that an individual is less likely to be captured by a predator when it is part of a larger group. Roberts postulated that individual vigilance may be lower in larger groups due to a lowered overall predation risk. Roberts then addressed the matter of predator detection as it Just Keep Swimming 4


relates to group size, and noted studies that showed an increased likelihood of detecting a predator with larger group sizes. In many cases, this overall increase in predator detection occurred even when individual vigilance levels decreased. It is also worth noting, however, that the studies mentioned in Roberts" review only considered predator detection during a nal, obvious approach to the school. It stands to reason that animals can gain an added benet by noticing a predator prior to an attack, something that future research should consider when investigating predator detection. !Roberts also referred to #collective detection," or the #group vigilance hypothesis," as socially transmitting knowledge of a nearby predator through certain cues. Animals such as birds provide alarm calls. Fish use specic movements to transmit the necessary signals, but it seems more likely that nonvigilant sh, unaware of a nearby predator, will simply follow their nearest neighbors in evasive maneuvers before realizing a predator is nearby (dynamics of animal aggregations are discussed further in the #Inuential Zones" section of Chapter3). Circumstances may be such that even a vigilant group member does not notice a nearby predator. Roberts explained that unsuccessfully vigilant birds were quicker to disperse in response to an alarmed group member than were their non-vigilant cohorts. In those cases, individual vigilance levels could have meant the difference between life and death. !Roberts also described that as individual vigilance declined more time could be spent engaging in other activities. A common observation is that feeding time increases with decreased vigilance. It is peculiar, however, that feeding Just Keep Swimming 5


success and overall intake of food does not often increase, even with increased feeding time. Roberts suggested that this lack of productivity in the face of decreased necessary vigilance indicated that greater foraging efciency is not an advantage of reduced vigilance. !Along with group size, group density has a noticeable effect on vigilance. Roberts evinced that individuals in denser groups glean relevant information more quickly; alarm cues, such as sudden behavioral changes in response to a predator, are easier to notice when in closer proximity to their origin. Roberts explained that when group size was controlled, neighbor distance had an effect on vigilance, but that when neighbor distance was controlled, group size did not have an affect on vigilance. Those ndings also relate to Hamilton"s (1971) "selsh herd" model of group dynamics, which suggested that animals will tend to place themselves in higher-density groups to lower the risk of predation (Hamilton"s model is discussed in detail at the beginning of Chapter 4). Roberts also noted the existence of edge effects, where animals located on the outer edges of an aggregation had been consistently found to be more vigilant. !Finally, Roberts attempted to distinguish the hypothesis of predation risk from the hypothesis of group vigilance, ultimately suggesting that they were not mutually exclusive. The nding that the probability of an individual being captured decreased with larger group sizes may imply that larger groups benet less from a dilution effect than from a simple predator deterrent caused by increased overall vigilance. This in turn leads to fewer successful attacks on those groups. Predators may then adapt to preferentially attacking smaller groups. Roberts Just Keep Swimming 6


encouraged continued study to nd a clear distinction between the predation risk and group vigilance hypotheses, and suggested a study that measures the speed at which groups of different sizes detect and respond to a stimulus that indicates availability of food. An experiment of that type would help to distinguish advantages of group vigilance from the predation risk hypothesis. !Predation is obviously the greatest natural threat that an organism can face, as it often results in death. In order to continue to evolve and adapt, a species must rst continue to survive. Strength in numbers, as seen with the dilution effect, is a very effective strategy for continued survival and sh have been using it for generations. Numbers alone, however, can not be a permanent survival strategy in the scope of the perpetual evolutionary arms race between predator and prey. Fish have adapted by making a predator"s hunting even more difcult, utilizing the confusion effect to hinder a predator"s ability to target an individual. By adopting visual uniformity throughout the group, schools can cut the number of casualties they suffer by predators (McRobert and Bradner, 1998; Milinksi, 1977; Ohguchi, 1981). ii. Predation Protection: The Importance of Conformity !Uniformity, or conformity, represents a common trait in schooling sh and may show itself through individuals" body coloration (McRobert and Bradner, 1998), group behaviors such as quorum decision-making (Ward, et al., 2008), sizes of individuals (Peuhkuri, 1997; Barber, 2003), or any combination of these factors. Variations appear among these factors throughout differing species and ecosystems. Visual uniformity may be less important in species that experience Just Keep Swimming 7


fewer predation pressures. Many shoaling sh, for example, do not exhibit visible group homogeny. These species are often more social; male guppies are visually distinct, and spend much of their time seeking out mating opportunities (Lindstr šm and Ranta, 1993). !Anti-predator benets of visual uniformity arise from the confusion effect (McRobert and Bradner, 1998; Milinksi, 1977; Ohguchi, 1981), which occurs when a predator experiences difculty targeting a single individual in a high concentration of visually identical individuals and thus experiences less predatory success. Cohesive sh schools could possibly gain an added benet against predators perceiving the group from a distance; when visibility is poorer, a tightly formed school may appear to be one larger entity, thereby deterring further investigation by their predators. It follows logically that sh exposed to higher predation pressures will travel in schools of similarly-colored individuals in order to utilize the confusion effect, particularly those species living in waters with higher visibility (The various factors relevant to shoal choice are discussed at length in Chapter 4). !Roccanova (1993) suggested that the bright colors observed in certain species of shoaling sh may have evolved to attract conspecics, specically at times when individuals stray from the group and subject themselves to higher predation risk. Brightly colored individuals may attract the attention, and company, of nearby conspecics, thereby reducing their predation risk by bringing the shoal, and protection, along with them when they stray. Just Keep Swimming 8


!Conformity can show itself in many ways, but its importance is ubiquitous. Many species rely on the confusion effect for continued protection against predators, and maintained uniformity is a crucial component of that effect. Without it, schools would suffer a signicant increase in predation and their tness would suffer a proportional impact. That kind of compromising position would likely cause a drastic change in the ecological standing of the species, and could potentially lead to extinction. iii. Increased Foraging Efciency !The second widely-accepted benet of schooling comes from sharing information or, to be more specic, an awareness of what nearby conspecics are doing. This intra-group vigilance, combined with the many eyes theory, increases the likelihood that a school will locate a food source, thus increasing overall foraging efciency for the group. It comes as no surprise that foraging efciency and predation protection are the two greatest benets of schooling behavior, as energy intake and continued protection are essential to tness. Many studies have been concerned with the extent to which foraging efciency is affected by schooling, as presented in this section. !In 1984, Mittelbach conducted one of the rst experiments to consider individuals" prey-encounter rates in schooling sh, using the bluegill sunsh ( Lepomis macrochirus ). After observing bluegills feeding on amphipods in habitats of dense vegetation and noticing the amphipods sometimes elude capture in the vegetation, Mittelbach hypothesized that groups of bluegills may yield more food by enabling individuals to capture prey that had previously Just Keep Swimming 9


eluded another group member. The feeding rates of single sh and groups of two, four, and six individuals were compared. Experimental groups were placed in a 120.5 cm $ 55.5 cm $ 32 cm tank with 5 cm of vegetation at the bottom, and 300 amphipods per tank (at the lower end of densities reported from nature). !Upon release into the feeding area, a focal individual in each group was monitored for number of amphipods captured. Mittelbach found a signicant effect by group size on individual foraging rates, and mean prey captures per minute reached a maximum at a group size of four (~1.0 captures/min.). Two key trends were observed in this group: group members ushed out prey that they could not capture and these amphipods were eventually captured by the focal sh; when an individual captured an amphipod other group members would be drawn to the area, and these sh were often able to snatch up additional prey. !Curiously, at groups of six sh, prey captures sharply declined to the lowest mean value over all group sizes, ~0.45 captures/min. This observation was strange because, according to Mittelbach, bluegills feed in groups ranging from 2 to 20 individuals. However, groups of six were observed to have a high rate of aggressive interaction, particularly subgroups of the highest densities spent more time chasing one another than foraging. This aggression may have been due to the relatively conned conditions of the test tank. Mittelbach conjectured that the low prey capture rate may have also arisen from the larger group quickly eliminating higher density or more exposed prey patches, followed by relatively low success thereafter. Results of this nature were never previously reported by other studies, and Mittelbach concluded that they were an artifact of his study. Just Keep Swimming 10


Mittelbach ultimately concluded that foraging rates do vary with group size, and, despite the decline at six group members, they eventually would, most likely, increase. !Taking a different approach, Day, MacDonald, Brown, Laland and Reader (2001) found that larger group sizes of guppies could more readily nd a hidden food source when they were able to visually transmit relevant information. Three experiments were carried out to this effect. In the rst experiment, six feeder boxes with small openings were placed at the corners of a rectangular tank; one was located in each top corner, and another two were placed in two of the bottom corners. Only one compartment actually contained food. In this experiment, twelve focal sh were each tested once, in group sizes of two, four, eight, and sixteen guppies. The dependent variables were: time taken for the rst sh to enter the feeder, time taken to feed, time taken for the focal sh to enter the feeder, and time taken for the focal sh to feed. As group size increased, the time taken for each dependent variable linearly decreased (Figure 1). Day, et al. noted that a contributing factor to these data might have come from sh learning that food would be present, but not where it would be located. This would likely cause the Just Keep Swimming 11


Figure 1. Mean latencies SE of each shoal size, for (a) the rst sh to enter the feeder, (b) the rst sh to feed, (c) the focal sh to enter the feeder and (d) the focal sh to feed. N=12 in each case. (LSD: *P<0.05; **P<0.01). (From Day, MacDonald, Brown, Laland and Reader, 2001) Just Keep Swimming 12


sh to be more aware of their shoal mates" behaviors, and use those cues to locate the food for themselves faster. !The second experiment differed from the rst in that there was only one oating feeder located on the other side of an opaque partition, and a small opening at the bottom of the partition. This time group sizes were four, eight, and sixteen. Dependent variables were time taken for each sh to enter the #goal zone" located behind the partition and feed. The results of this experiment were essentially the reverse of the previous experiment; sh in shoals of sixteen took the longest to locate the food (302.6 19.5 s), followed by groups of eight (135.4 13.1 s) and then four (106.3 s 10 s). These data were the opposite of what was expected, but Day, et al. suggested that this was because sh were less inclined to leave the security of larger shoals into an unseen area. A choice to leave a smaller shoal would have less overall detrimental value. The presence of the opaque partition allowed the other members of the shoal no visual cues regarding the fate of any conspecics that passed through the threshold. The research team thus hypothesized that shoaling with the largest possible number of conspecics may generate a positive frequency-dependent transmission of foraging information. This effect was termed conformist transmission. In other words, sh may tend to adopt the behaviors of the majority of their shoal mates. As group size increases, so does the minimum necessary number of shoal mates to affect a decision. !Experiment number three was identical to experiment number two, save that the partition was transparent. It was expected that retained visual contact Just Keep Swimming 13


with any conspecics would more readily coax additional sh into crossing the threshold. In contrast with the results of experiment two, the sh in experiment three showed a decreased latency in locating the feeder and eating as group size increased. Fish in groups of sixteen located food the fastest (166.6 13.2 s), followed by groups of eight (346.5 22.5 s) and four (424.9 19.9 s). These results supported the hypothesis that a lack of visual cues between successful and unsuccessful foragers was responsible for the reverse of the results in experiment one. The team concluded that the conformity hypothesis was supported by the ndings of the experiment, which clearly showed that hidden food sources were located faster with increasing group sizes. !Increased foraging efciency is clearly a major benet gained from aggregation: energy intake is universal across all forms of life. Less time spent foraging means more time to engage in other vital activities, such as mating or resting. Better exploiting a resource patch leads to higher energy reserves and reduces the frequency of necessary foraging episodes. Lasting nutritional health is a key component of continued survival, and for many species of sh living in schools affords the continued possibility to obtain it. B. Primary Costs of Schooling Behavior i. The Oddity Effect !The confusion effect, or the confusion of a predators senses in order to minimize a group"s casualties, is dependent on total uniformity for success Individuals that stand out in the group run a signicantly higher risk of being captured (Landeau and Terborgh,1986; Theodorakis,1989). This is known as the Just Keep Swimming 14


oddity effect (Peuhkuri, 1997). Despite the merits of the confusion effect, the likelihood of total visual homogeny within a group is very small. Juvenile individuals swimming with an overwhelmingly adult group will experience oddity. Oddity is even greater for individuals with different coloration from the rest of the group female guppies are neutrally colored, whereas males are often bright and ashy. In some cases, the group itself is the oddity. For pelagic sh, much of the time is spent navigating empty open ocean, and a group of small sh may be just what a pelagic predator was looking for. Consider a large school of herring, consisting of thousands of sh. Large tuna will attack such groups, and the subsequent urry of activity from both parties may attract more predators. Dolphins take part by swimming near the bottom of the school and blowing pillars of bubbles that frighten the herring, and the dolphins utilize this advantage to #cage in" the herring, causing them to begin swimming in a tight formation and forming a large swirling mass, known as a bait ball. The herring then swim up to escape the threats on all other sides. This attracts predators from above, such as pelicans or seagulls, which will dive into the water and force the sh back down into a tighter ball. With predators of all sizes and strategies surrounding and puncturing the tight formation of herring, it"s only a matter of time before the entire school is gone. This kind of scenario has played out in front of cameras on many occasions, and has been featured in documentaries such as the BBC"s Blue Planet and Planet Earth series. !In 1985, Wolf conducted a eld experiment to examine oddity and its effect on responses in mixed groups of striped parrotsh ( Scarus iserti ), stoplight Just Keep Swimming 15


parrotsh ( Sparisoma viride ), and ocean surgeonsh ( Acanthurus bahianus ), to a model of a common predator, the trumpetsh ( Aulostomus maculatus ). As a control, groups were also shown a model of a non-predatory sh, the French grunt ( Haemulon avolineatum ). Each species responded differently to the models, but there was a consistently more substantial response to the predator model in each prey species. Stoplight parrotsh were signicantly more likely to abandon the group and hide in coral when presented with the predator model. Striped parrotsh preferentially associated with groups in the presence of the predator model. If a focal striped parrotsh was in a group smaller than those typically used for protection, those groups were abandoned in favor of a larger group. The surgeonsh in this study showed mixed responses based on size. Small surgeonsh sought protection in groups, while larger individuals swam away alone. Wolf notes that when smaller individuals left groups, they only did so because they could not keep pace with eeing groups. !Wolf suggested that stoplight parrotsh sought cover in the reef because of their relatively low numbers and bright coloration; both of those factors would increase their oddity and leave them more susceptible to predation. With respect to striped parrotsh, Wolf noted a tendency to roam in large numbers, a trait that may limit their familiarity with available cover. Those traits may have predisposed the striped parrotsh to seek cover within groups of conspecics. Larger surgeonsh tended to leave groups and swim alone based on their larger size, which severely limited the ability of a predator the size of the model to swallow such a sizable individual. Another factor to consider when imagining an attempt Just Keep Swimming 16


to swallow a larger surgeonsh can be found in the sh"s namesake razor-sharp protrusions located on either side of the caudal peduncle, the area where the tail meets the body. Considering the results of this study it is clear that each of the observed species responded to the predator by seeking cover of some kind, but these reactions also indicated attempts to limit predation by associating with conspecics, thereby limiting oddity, or simply escaping altogether. Oddity can present problems to individuals other than predation risk, as evidenced by Peuhkuri (1997). The effects of oddity on foraging behavior were investigated using small and large three-spined sticklebacks ( Gasterosteus aculeatus ) by observing their foraging activity in groups of three, six, and twelve, where the focal individual either differed in size from the all other group members, an assortment of sizes were present, or all sh were the same size. Small sticklebacks did not vary signicantly in foraging time or feeding rate, regardless of their relative size compared to the rest of the group. Large sticklebacks, however, showed signicant variance in both regards. When large sticklebacks were the odd one out, they spent less time foraging than other large focal individuals in mixed, or size-assorted schools. Large sh, when under oddity, also fed at a slower rate than those large individuals in mixed and sizeassorted groups. Large sh in mixed or size-assorted shoals did not show signicant differences in time spent foraging or feeding rate. Fish spent less time foraging and fed at a slower rate across all trials when in the presence of a predator. With respect to the apparent discrepancy in perceived oddity by small and large sticklebacks, Peuhkuri conjectured that large individuals may stand out Just Keep Swimming 17


more among smaller individuals than smaller sh in groups of larger sh, but cited an unfortunate lack of evidence to support the claim. A supplemental explanation for the lack of variance in foraging behaviors of small sticklebacks may come from size-dependent differences in competitive ability. !Oddity represents the greatest potential cost of schooling behavior. As an effect that clearly exposes an individual to predators, and thus death, oddity is an effective mechanism for natural selection. Fish from species that experience high predation and rely heavily on the confusion effect for protection should exhibit fewer odd traits, and genetic anomalies that lead to higher oddity in these species are likely eliminated with predation. Under certain circumstances, however, oddity may serve species that experience lower overall predation. As mentioned earlier, male guppies are brightly colored with unique patterns, and these bright colors are used to attract mates. Larger, brighter individuals display higher tness, attracting more mates. ii. Competition !Peuhkuri (1997) postulated that the small sticklebacks in her study might not have shown any variance in foraging behavior while in groups of larger individuals because a lack of necessary competition. The food supplies in the study were limitless. To elaborate, smaller individuals are presumably less capable of competing against larger individuals. This would likely lead to assortment in schools by competitive ability, and thus by size. Many schools are size-sorted, but it is not clear that this is due to sorting by competitive ability. Thus, in the articial conditions of Peuhkuri"s experiment, the larger individuals Just Keep Swimming 18


surrounding smaller focal individuals should dominate in terms of foraging efciency. However, as mentioned before, resources were limitless and competition was not necessary. As such, smaller sticklebacks did not suffer lower foraging rate. !The rst research to show a marked decline in foraging efciency with increased group size occurred in 1984, as previously noted; Mittelbach found that, by increasing group size in bluegills, individual foraging rates experienced a decline due to an increased frequency of "aggressive interactions" between test sh. Mittelbach notes that wild populations of increasing size are not prone to exhibiting these aggressions during episodes of foraging, and attributes this to an artifact of a laboratory setting. The conditions of that laboratory study did not lend themselves well to simulating natural conditions, as food supplies may not always be superabundant. This variability was noted, and the hypothesis that crowding may increase competition for resources gained attention for some years to come. !Measures were later taken to investigate the effects of group size on competition, both in the presence of a predator and not, when food resources were scarce (Grand and Dill, 1999). Juvenile coho salmon ( Oncorhynchus kisutch ) were exposed to a current, which was capable of carrying food away from a feeder, and their respective proximities to the feeder where a predator sometimes resided behind plexiglass were monitored by a video camera. A wide pillar was placed near the center of the test arena, for use as cover by test sh. Focal sh were tested in multiple group-size conditions: solitary, one companion Just Keep Swimming 19


sh, or three companion sh. Companion sh were isolated from the focal sh in an adjacent plexiglass enclosure, so that focal sh could observe the behaviors of their companions. The feeder pumped single brine shrimp into the test arena every three minutes, for a total of ve shrimp over 15-minute trials. During the trials at 30-second intervals, the position of the focal sh in proximity to the feeder (rated as 1-7 at 10, 20, 30, 40, 50, 60, and 70 cm away) was noted. !Generally, focal sh displayed a variety of strategies during trials. Some remained near the pillar and only abandoned cover to capture an incoming brine shrimp. Others completely ignored cover and remained upstream. Companion sh tended to also remain upstream and capture prey as they entered the foraging area. Grand and Dill noted that it appeared as though focal individuals were alerted to the presence of prey items by the behaviors of the companion sh. Focal individuals consistently captured fewer prey than companion sh, but captured more with increasing group size. Prey capture by focal individuals occurred at closer proximity to the feeder when a predator was not present, as well as with increasing companion group size. Due to a lack of signicant interaction between predator presence and number of companions with respect to prey captures or prey capture distance, Grand and Dill interpreted these results as an indication that the changes in risk-taking behavior observed in the study subjects were a direct result of increased competition for nite resources. Thus it appears that increased intra-group competition for vital resources leads individuals to take greater risks. Just Keep Swimming 20


!Humphries, Metcalfe and Ruxton (1999) further investigated group size vs relative competitive ability using European minnows ( Phoxinus phoxinus ) in a continuous-resource setting. Sixteen minnows were selected from a captive collection, marked with unique patterns of blue dye, and split into eight pairs. The procedure was relatively simple; a number of pairs were moved to a test arena and allowed to settle before being fed 150 pellets. The pellets were dropped one by one through a funneled current which led to a plastic dome to randomize the direction in which the pellet fell into the tank. Trials were arranged so that no pair was used in the same trial type more than once. The rst round of tests were conducted on two groups of eight four pairs each. The second round consisted of four groups of two and two groups of four sh. The third round was the same as the second, but the pairs were switched so each experienced being in groups of both four and two. The last round tested each pair at once for a group of sixteen. !Their results were similar to those of Peuhkuri (1997): when resources were superabundant, group size did not have a signicant effect on relative competitive ability. Humphries, et al. explain, however, that individual size did have an effect on relative competitive ability. Groups of European minnows may not display homogeny of size among individuals, and larger individuals tend to exhibit greater relative competitive ability. If a group is small, for example a pair, and one member is signicantly larger than the other, the two sh have different relative food intake, and thus signicant differences in relative competitive ability. As group size or group density increases, however, individuals are more likely to Just Keep Swimming 21


represent a wider variety of sizes. This diversity alleviates the appearance of the disparity between the competitive abilities a large individual and a relatively small one by adding individuals whose competitive ability falls between the original two. Additions of this kind help to lessen the degree of difference between individuals in sequence and make them more similar to one another. Thus it seems when comparing two members of a larger group, competitive disparities are likely to appear, but one must consider the whole group, which does not show a signicant difference of competitive ability (Figure 2). This was precisely the case in this study; the competitive abilities of two compared individuals often showed signicant differences, but differences in competitive ability were not nearly as pronounced when the entire group was considered. !Competition is a variable cost of schooling; individual species deal with competition in different ways. In order to limit competition as much as possible schools are often sorted by relative competitive ability, which has been associated with relative body size (Peuhkuri, 1997). Individuals in a school with a signicantly lower competitive Just Keep Swimming 22


Figure 2. Plot of a conceptual model to explain the disparity between individual size-effects and group size-effects on relative competitive ability. As group size increases, differences in relative competitive ability become less pronounced and overall intake decreases. (From Humphries, Metcalfe, and Ruxton, 1999) ability are far less likely to regularly get the nutrition they need to survive, and will most likely leave the group to seek out better social conditions. In doing so, however, the sh may become exposed to increased predation and subject itself to higher risk. These sh may be presumed to suffer higher casualties than those that remain with the group, which would prevent the passing of genetic material on to the next generation and thus not be an adaptive trait. Nevertheless, it could be the case that individuals preferentially join neighboring schools while not in the presence of a predator, thereby reducing their personal risk. Further study would help to elucidate a distinction on the matter. Just Keep Swimming 23


C. Secondary Benets of Schooling Behavior i. Learning !Associating with conspecics allows an individual to glean relevant information in a number of scenarios, from the location of food patches (Day, et al., 2001) to the presence of a threat to the group (Roberts, 1996). Fish, like other organisms, learn throughout their entire lives; they learn of new threats, new ways to avoid those threats, new sources of nutrition, and ways to exploit those resources. The ocean is in a constant state of change, and its denizens must adapt to that change or perish. It is necessary, however, that a foundation be laid for the basic behaviors that will be used for the duration of an individual"s life, and this occurs during the early stages of life (Chapman, Ward and Krause, 2008). Difculty arises when studying learning in sh. As a result many studies have only focused on the learning of specic behaviors. !Chapman, et al. (2008) aimed to clarify the extent to which social behaviors are inherited genetically or affected by early life experiences, and used the guppy as a model. Juveniles were reared in low (1-4 individuals) and high densities (7-12 individuals) and later tested for their tendency to shoal and their ability to learn socially a social foraging task. The shoaling tendency test consisted of a basic binary choice experiment where focal sh were placed in a central tank with two chambers and observed for their preference by monitoring how much time they spent in close proximity in this case, within two body lengths to either chamber. One chamber contained three conspecics and the other was empty. Social learning was gauged by placing a focal sh from either density Just Keep Swimming 24


housing into the central chamber of a test maze that had two additional compartments with small #doorways," one of which contained food. Other guppies were trained to know which chamber contained food, and were placed into the central compartment of the test maze to demonstrate for the focal sh, who were naive to the task. During these tests, which lasted 15 minutes, the time to feed was recorded for focal and demonstrator sh. After eight days of trials, focal sh were placed into the maze alone and observed for time taken to enter a compartment, time taken to enter the correct compartment, and time taken to feed. !Shoaling tendency of groups reared in low-density treatments was higher overall than those reared in high-density treatments. Over the course of the maze trials, sh reared at low densities excelled more than those from higher density conditions; low density sh followed demonstrator sh more quickly, improved over time, fed faster, and generally completed the task more frequently than sh reared at high densities. In the solo trials, low density-reared sh again fed more quickly. The study suggested that sh reared at high densities experience higher intra-group aggression and thus a higher cost for schooling. The guppies purportedly learned from this experience and decreased their tendency to shoal or use social cues. This claim may very well be true, as Mittelbach (1984) mentioned that large groups of bluegills spent more time displaying aggressive intra-group behavior than foraging, as detailed earlier in this chapter. Chapman, et al., however, did not make any mention of aggressive behavior among the test subjects or show any data to support the claim. Regardless of this shortcoming, Just Keep Swimming 25


the data still showed that low density-reared sh continued to feed quicker in the absence of demonstrators than high density-reared sh. This indicated that those sh learned better from the social cues of the demonstrators, or at least learned where the food would be located asocially. !What, then, do these data say about schooling"s effect on learning? After all, it would seem they indicate that smaller groups learn better and forage more effectively. Day, et al. (2001) showed evidence that positive frequency-dependent social learning does occur in large groups. The study in question, however, did not take into consideration the presence of any threats to the group"s safety. If a predatory threat had been incorporated into the study, the tendency to shoal would likely increase across the board. Chapman, et al. also suggested that higher density groups experience greater competition, which was shown to indeed be the case earlier in this chapter. It was then not implausible that sh acclimated to higher competition were better competitors themselves, and by that logic sh unfamiliar with competition would be poor competitors. It would seem, then, that the benets to social learning gained by developing in a low conspecic-density setting outweigh the lack of competitive ability developed in those individuals. !Guppies are a shoaling species. How does learning factor into the early lives of pelagic schooling sh? Gallego, Heath and Fryer (1995) attempted to answer this question by studying the schooling of larval herring ( Clupea harengus ) when in the presence of more mature conspecics. More specically, they sought to assess whether or not the presence of advanced conspecics Just Keep Swimming 26


could stimulate premature schooling in pre-metamorphosis herring, a species that exhibits schooling behavior only after metamorphosis. Larval herring were reared in three different tanks with densities of 14, 28, and over 100 individuals. A day after observations began which lasted for 36 days 14 postmetamorphosis juveniles were introduced to the tank of 14 larvae and observations continued. For the duration of the observation period, the tanks were monitored from above by video cameras. !Testing was conducted by moving eight individuals from each rearing tank into a testing tank, with the exception of the mixed juvenile-larva tank, which had two groups of eight. Here, one group consisted of four juveniles and four larvae, and the other consisted of eight larvae. These two groups were chosen to discern if larvae would engage in premature schooling only around the more advanced juveniles, or if they retained the behavior even in the absence of juveniles. During testing, group behaviors were observed for three nine-minute periods: after acclimation to the test tank, after food was introduced to the tank, and an hour after the food was introduced. Fish were considered to be schooling if they were within one body-length distance from another sh, and the durations of these interactions were noted. It was observed that larvae began engaging in schooling behavior, but only when juvenile conspecics were present (Figure 3). Gallego, et al. noted that the larval sh spent less time foraging, then quickly returning to the protection of the school, even to the detriment of their nutritional health. The results showed that premature schooling was induced in larval herring, but whether or not the behavior was actually learned or a simple act of Just Keep Swimming 27


mimicry was not clear. Larvae reared in the presence of juveniles, but tested only among larvae did not show any signs that they learned anything. It is plausible that schooling was not actually learned, and that herring do not retain the behavior until after metamorphosis. Clarication may have been possible if Figure 3. Changes the proportion of observed time spent schooling (RI) and the duration of schooling interactions (DURAT) of larval herring among observational groups and food treatments. Empty circles and tted lines correspond to larvae observed in the absence of food; empty triangles and dotted lines correspond to larvae immediately after the introduction of food; full circles and dashed lines correspond to larvae 1 hour after the introduction of food. (a) and (c) correspond to larvae in the presence of juveniles, and (b) and (d) correspond to all other groups. (From Gallego, Heath and Fryer, 1995) Just Keep Swimming 28


observations continued through the metamorphosis process and beyond. !Levin and Vergara (1987) tested learning in groups of mature schooling ame tailed tetras ( Aphyochorax erithrurus ), a species previously observed by Levin and Vergara to show little to no improvement among individual sh through reversal learning tests. Reversal learning is the process by which a subject learns to make a discrimination and then is made to reverse that discrimination. For example, in this study, groups of ame tailed tetras were tested on their ability to avoid being scooped out of the tank by a large paddle by swimming through a hole on one side of the paddle. The reversal occurred when the paddle"s rotation was reversed, forcing the sh to swim through the hole now located on the other side of the paddle. Whereas their previous experiment showed no improvement, Levin and Vergara this time found that the number of errors decreased within reversals for each of the groups tested. Levin and Vergara concluded that sensory cues were likely the most important factor in their results; sh were able to quickly respond to visual and lateral line information regarding conspecics that successfully passed through the paddle opening and follow suit (The roles of vision and the lateral line in schooling are discussed in Chapter 3). One thing is certain, ame tailed tetras are capable of learning from unique problems and adapting to them. !Thus, there is evidence to support the idea that learning is a benet of schooling behavior. Fish can learn more about their surroundings from paying attention to their shoal mates, as well as strategies for overcoming obstacles. Learning is an incredibly useful adaptive trait of any animal species. It opens the Just Keep Swimming 29


door to consistent behavioral change that need not be limited to a single generation. If a new behavior persists long enough, genetic modications may occur in a population over time to better facilitate the behavior, if it increases tness. For examples one must look no further than to common physical traits such as webbed appendages on semi-terrestrial animals, used for more efcient travel through water. Difculty in studying the ability of schooling sh to learn arises from the necessary specicity of those studies. Learning studies are typically task-oriented and specic, and thus limited in their scope. Overall, though, learning in the context of schooling remains a subject that could benet from more study and attention. ii. Sleep Mitigation !At the time of this writing, the subject of this section is one that is controversial and strictly theoretical. Regardless, it has its place in this review and could stand to receive more academic attention. In 2001, Kavanau drew parallels between daily sensory input and sleep, sh schooling, and avian ocking. Sleep, Kavanau explains, can serve as a state wherein synapses are reinforced and strengthened through repetitive electrical oscillations in the brain, enabled by activity-dependent plasticity. This activity helps to convert short-term memories into long-term memories, which in turn need to be maintained through the electrical oscillations that occur during sleep. Not all organisms sleep, however, and a question arises concerning how memory is developed and conserved in these organisms. Kavanau explains that the selection pressure for sleep may have arisen from an increasing complexity in the lives of more Just Keep Swimming 30


advanced organisms; the upward gradient of sensory acuity and complex behaviors of higher organisms necessitates some type of mitigation in the brain, which may manifest in the form of a sleep regimen. !For those organisms that forego sleep, data concerning memory mitigation is sorely lacking; Kavanau thus posits a theoretical explanation for a lack of sleep in pelagic and #inactively" schooling sh. Inactive schooling is dened as that which occurs in a relatively xed location and doesn"t involve any complex behaviors. Inactive schooling occurs in many small nocturnal reef sh, which are exposed to much more complexity in their daily lives, and may serve as a substitute for traditional sleep. Independent foraging takes place throughout the reef at night, but school members converge before dawn and resume schooling inactively in a small area. Juvenile black tip sharks were found to exhibit this behavior in summer nursery areas of the Terra Ceia Bay in Tampa, Florida (Heupel and Simpfendorfer, 2005). This period of relative inactivity may afford these reef sh the opportunity to enter a rest-like state that may serve the same purpose as sleep. Other reef sh retreat to safer, darker portions of the reef during the day and actually rest or sleep. !For pelagic sh, visual input is largely unremarkable with the only regular exception being schoolmates. Visibility extends to around 50 meters far less at night in any direction. With the lack of complex behaviors and a constant need for water coursing over their gills, sleep is apparently not a necessity for pelagic sh. The repetition and lack of complexity in their daily lives may allow these Just Keep Swimming 31


shes" brains to enter a rest-like state, and this state may be adequate for neural reinforcement. !Inactive schooling allows the species that exhibit it to maintain the protective benets of being in a group while simultaneously allowing them to rest. Theoretically, the grouped sh need only be aware enough to remain vigilant, and with the group vigilance hypothesis, discussed at the beginning of this chapter, these vigilance demands are lessened with higher group sizes. If a large enough group schools inactively, the many eyes theory suggests that any particular individual need not devote much attention to vigilance. If this theory has any validity, inactive schooling could prove a vital benet to gregarious sh. This opens up a whole new avenue of potential study. Tests could be conducted to examine the short and long-term memories of sh that are prevented from inactive schooling, and compared with those of sh allowed to inactively school. The results of these potential studies could offer a wealth of information regarding the formation of memories in animals that do not sleep. D. Secondary Costs of Schooling Behavior Parasites The presence of parasites in a school can make homogeny a much more difcult prospect, as they tend to stand out visually or inuence the behaviors of infected individuals. The Adaptive Manipulation hypothesis states that certain parasites begin life in a source of food for an intermediate host in this case a sh and manipulate the behavior of that intermediate host to increase the probability of reaching their nal host, such as a piscivorous bird (Ward, Hoare, Just Keep Swimming 32


Couzin, Broom, and Krause, 2002). If the goal of the parasite is to make sure the intermediate host is eaten, then it stands to reason that the manipulation involved should increase the host"s oddity. The following section examines trends in parasitism and its effect on school members. !Ward, et al. (2002) conducted a simple survey of wild shoals of banded killish ( Fundulus diaphanus ) and took note of individuals" positions and their degree of parasitism. Fish were collected by laying a grid-net (a broad net with many pouches aligned in a grid pattern) ush with the sand at the bottom of shallow water habitats and quickly pulling the net up after an entire shoal was within the bounds of the net. The grid pattern of the net allowed the surveyors to see the two-dimensional position of each individual in the shoal at the time of capture. In addition to placement, body length and parasite abundance was recorded for each individual, as well as the shoal"s direction of travel. A subsequent study surveyed singular individuals, using the grid-net to conrm their solitary state. !Prevalence of parasites over all individuals captured was 62%, but there was no correlation between body-length and parasite prevalence or abundance. This implied that body-length and parasitism were independent factors in shoal position. That said, median body length and parasite prevalence were greater near the front of shoals. Fish with higher parasite abundance, however, were found at the rear of shoals. Ward noted that in 11 of 14 shoals, parasitized sh were located farther away from the geometrical center of the shoal, but due to a statistical outlier, statistical signicance could not be obtained. The most visually Just Keep Swimming 33


obvious distinction in shoals with many parasitized individuals was a phalanxshaped distribution, as opposed to the normal elliptical shape of banded killish shoals. This was shown by taking the angle between the direction of travel and the tted trendline of individual distribution. In highly parasitized groups, this angle approached 90 degrees. Solitary individuals experienced higher parasite prevalence and abundance. !The results of this study showed a marked likelihood for parasitized individuals to take up forward positions in the shoal. Since little is known about the specic effects of the parasite in question Crassiphiala bulboglossa on banded killish, speculation dominates the discussion of this topic. Larger individuals were more likely to take up forward shoal positions, and forward positioned individuals were more likely to be parasitized. Larger individuals are better competitors, so it would make sense that they take up positions better for foraging. This competitive edge would result in higher foraging efciency, but also increase the likelihood of contracting a parasite. It could also be that the parasite can manipulate its hosts energy reserves, thereby increasing the host"s motivation to feed. This motivation would lead the sh to take up the observed forward shoaling position and thus be more exposed to predation. Conversely, the host may become more motivated to feed, thereby increasing its likelihood of contracting more parasites. Overwhelmed by an abundance of parasites, the host may not be able to keep up with the costs of swimming and foraging with a group. This might account for the increased parasitism observed in solitary Just Keep Swimming 34


killish. At this point, the ability of the host to avoid predation would be minimized, providing predators with a meal too good to pass up. !Barber (2003) carried out a similar study using three-spined sticklebacks in order to determine whether or not sh engage in parasitism-assorted schooling. Six complete schools 366 individuals total were captured and transferred to a laboratory where each individual was measured and examined for external parasites. Three parasites were observed in all: Cryptocotyle lingua Glugea anomala and larvae of Caligus sp. Clear size assorting was found in the captured schools. C. lingua had visibly infected 91% of the total sample, G. anomala infected 6.8% of the sample, and Caligus sp. infected 3.9% of the sample. The likelihood of infection by C. lingua and G. anomala increased with body size of individuals. Infection by Caligus sp. was only marginally signicant with body size. Barber noted that infections of Caligus sp. were aggregated within schools, with 54% of all infections found in a single group made up of 21% of total individuals sampled, and another school of 127 individuals not having a single infection. Barber suggested that this may be due to the parasite being highly visible, which may lead uninfected sh to avoid associating with those infected individuals. !The results do not strongly support the hypothesis that sticklebacks engage in shoal-assortment by level of parasitism, though, as nearly all of the individuals sampled had some degree of parasitism. The schools were size-assorted, but the levels of parasitism within each school were not homogenous. Barber posited one possible explanation for the lack of parasite-assortment within captured Just Keep Swimming 35


schools; lower sh densities may have increased the distance between schools and limited individuals" abilities to re-assort themselves. This suggestion was supported by the relatively few schools that were captured, and the sighting of only one or two schools per hour on average. Further study with regard to this confusion is clearly necessary. !Parasitism is certainly a detrimental factor to the health of an individual, and here we have seen evidence that it can be detrimental to group health. Not only is there the risk of spreading parasites to other group members, but parasites" tendencies to alter the behavior of their hosts puts an entire shoal at risk. Disorganization draws the attention of predators, as disorganization of a protective unit provides predators with easier meals. As such, parasitism has great potential to undermine the rst primary benet of schooling behavior, predation protection. On another level, parasites can affect the appetites of their hosts, encouraging those hosts to abandon their shoals and seek out food in an environment free of competition. This further undermines an individual"s safety from predators, as the individual is left completely alone and open to predation. !Natural selection imposes selection pressures on any number of traits exhibited by an organism. Behaviors are a major part of the daily life in all organisms, and as such are subject to a great many selective pressures. Traits are selected for or against, which ultimately determines whether or not they continue in further generations. Schooling and shoaling are often behaviors that are selected for as they have been exhibited by modern shes for generations and continue to present themselves in a signicant proportion of sh species. As Just Keep Swimming 36


we have seen, there are many pressures that act to reinforce schooling, such as predation protection and heightened foraging efciency. There are also pressures acting against schooling, as seen in competition for resources and the spread of ailments and parasites. Nevertheless, the benets must outweigh the costs, or else schooling would have been selected against. Biotic and abiotic factors are always changing, though, and in order to continue to thrive a species must constantly adapt. Chapter 2. Inheritance and Adaptation of Schooling Behaviors !Schooling is an already well-established set of behaviors among modern shes. The foundations of schooling behaviors have been found to possess an inherited genetic component (Seghers, 1974), but there is a degree of plasticity to these behavioral phenotypes. Aggregation behaviors can be altered based on life experiences, and many laboratory studies strive to nd the degree to which inherited behaviors can be adapted to suit a new set of circumstances. The previous section on learning as a benet of schooling provided some examples of different species" abilities to detect changes in their environment and adapt their lifestyles accordingly. This section shifts the focus toward an evolutionary perspective on schooling by looking at the inherited components of schooling and how they can change over subsequent generations. A. Genetically Inherited Behavioral Phenotypes !In 1974, Seghers sought to determine if tendency to shoal was a heritable trait. He did so by capturing wild guppies from ve different isolated populations and raising these populations for three to four generations, then testing them for Just Keep Swimming 37


shoaling tendency. Each of the tested populations experienced different levels of predation in their natural habitats, and the hypothesis was that subsequent generations of these populations would express different shoaling tendencies based on those predation levels, even though the subsequent generations had never experienced any predation. Shoaling tests were executed by placing groups of ten matured sh from one stock population into a rectangular tank that was lled to a depth of ~3cm and featured a ten-square grid along the bottom. After ten minutes of acclimation time, the positions of each sh were recorded every minute for 30 minutes. The recording process was repeated after another ve hours inside the test tank. Results were calculated as an #index of cohesion," which was the average maximum sh density in the grid over each 30 minute observation session. Each stock exhibited a signicant decrease in the index during the second 30 minute period, but the degree of dissociation was greater in the populations from areas with less predation. The population stocks derived from areas with high predation pressures displayed a much higher tendency to shoal. Seghers claimed that the variation in the study closely resembled that of the natural populations he initially observed. Figure 4 shows these results, which supported the hypothesis that an underlying genetic component correlated with relative predation pressures played a denitive role in the differences in shoaling tendency across different populations. Fish are capable of complex social behaviors, however, and later research expanded upon the foundations laid by Seghers. Just Keep Swimming 38


Figure 4. Index of cohesion for ve population stocks of guppies. Stocks derived from natural populations exposed to a wide range of predation pressures. Vertical lines represent the means of 10 replicates per stock; horizontal lines represent the ranges; rectangles are 95% condence limits. (a) Cohesion during the rst 30 minute observation period. (b) Cohesion during the second 30 minute observation session, ve hours later. (From Seghers, 1974) !Ruzzante and Doyle (1993) sought to understand the relationship between the development of social behaviors and growth rate in medaka ( Oryzias latipes ). Fish were selectively bred for fast or slow growth rate over three generations in the lab, and differences in schooling, social tolerance, and agonistic behavior Just Keep Swimming 39


were compared in the third generation. Agonistic behavior is dened as any social behavior related to ghting, and can include anything from aggression to submission. Specically, agonistic behavior can include subordinance, retreat, and conciliation. Over the three generations of growth rate selection, sh were reared in one of two social regimes: high interaction or low interaction. Generation 1 was divided equally between the two regimes, and sh were selected for fast or slow growth rate in both regimes, producing four total lines. Generation 2 males were mated to unselected females to produce generation 3. Generation 3 sh from each line were then subsequently divided into high and low interaction regimes, producing eight sub-lines. !Each of the eight sub-lines was observed for the following: agonistic behaviors (nips, chases, sudden charges), response to a predator, and social tolerance. Agonistic behavior under conditions of high social interaction was negatively correlated with mean growth rate in the fast growth line; no signicant deviation from the norm was observed in the other lines. In the presence of a predator, sh selected for fast growth rate were more likely to school than slow growth-selected lines. Finally, social tolerance was signicantly more positive in the fast growth line, but only when the selection was conducted under conditions of high social interaction. From these results arose the #Threshold Hypothesis," which suggests that selection for fast growth in high interaction environments may decrease agonistic behaviors, increase tendency to school under duress, and increase social tolerance. Ruzzante and Doyle coined this hypothesis due to a change in the nature or intensity (threshold) of the stimulus necessary to evoke Just Keep Swimming 40


these behaviors. Ruzzante and Doyle also suggest that intraspecic competition and growth rate in wild populations of medaka will likely vary depending on the balance between population density and distribution of food; levels of agonistic and schooling behaviors in wild populations are likely to be the product of the past episodes of selection in any direction. !These studies are a clear indication that micro-scale evolution plays an important role in determining generational differences in social behaviors. Selection for and against particular phenotypes varies widely, even across nearby microcosms, and this variety can cause a proportionate diversity within a single species. Phenotypic change is not limited to inherited genetic traits, though; indirect genetic effects (IGEs) are a result of learning throughout life, and can actually lead to an alteration of an individual"s phenotype. B. Indirect Genetic Effects !Indirect genetic effects occur when a phenotype expressed by a conspecic inuences an individual"s phenotype, and these IGEs represent a mechanism for the inheritance of behavioral reciprocity (Bleakley and Brodie, 2009). In other words, the expressed traits and behaviors of one individual can inuence those of another individual over time, which the affected individual may then pass to its genetic offspring. At the time of this writing the mechanism by which this process occurs is still unclear, but there is a recently proposed hypothesis suggesting that behaviors could diversify through reverse transposition of LINE-1 genomic elements, which inuence gene expression Just Keep Swimming 41


(Singer, McConnell, Marchetto, Coufal, and Gage, 2010). Describing this process requires signicant background explanation, which is located in Appendix B. !The mechanism for genetic inheritance of of social behaviors is largely unknown. Bleakley and Brodie (2009) sought to elucidate the inheritance of cooperation and reciprocity, and used IGEs as the framework for understanding this concept. Groups of common guppies ( Poecilia reticulata ) were bred in specic lines: blue, 1/2 yellow, 1/2, green, R-cobra, and snakeskin varieties. These lines were selected for their difference in expression of anti-predator behaviors. Tests were conducted with sh from the blue and snakeskin strains as focal individuals, and these individuals were tested in the presence of groups consisting entirely of sh from one of the other three #context groups." Focal individuals were tested in a tank with a model of a natural predator and a plant for cover, then observed for the following: time spent oriented toward the model predator (orientation), time spent in close proximity to the model (proximity), number of predator inspections, time spent in extremely fast randomly patterned swimming (agitation), and time spent gathering in groups (schooling). Predator inspection is a cooperative behavior exhibited by guppies: pairs or small groups will swim close to a predator and observe it. This behavior exposes these group members to higher predation risk, but can yield important information regarding the activity and satiation level of the predator. A #coefcient of interaction" was calculated for each test criterion, denoted by % ij This coefcient represented the interaction between the behavior of the focal individual ( i ) and that of the social partners ( j ), and could range from -1 to 1. A negative value would indicate a Just Keep Swimming 42


proportional lack of inuence by social partners on the focal individual, which conversely, a positive value indicated a proportional inuence of individual behaviors by the context group. Bleakley and Brodie hypothesized that antipredator behaviors in focal individuals would reect both direct genetic effects and IGEs; effects of the focal individual"s genetic strain as well as the specic social group behaviors would affect the individuals" choices. !With so many variables being scrutinized, the data set of this study was quite vast. Despite the profusion of data, notable trends were present in the results. Blues spent less time oriented toward the predator model, more time in close proximity to the model, and more time in agitated swimming than did Snakeskins, regardless of the context group. Focal sh of all types tended to engage in inspections more often and spent more time in close proximity to the model when paired with half-green context groups. When in the presence of halfyellow context groups, focal individuals spent more time oriented toward the model. Blue focal individuals shoaled more frequently around 1/2 green groups than 1/2 yellows, and overall spent more time shoaling than snakeskin focal individuals. Bleakley and Brodie noted that blue and 1/2 green strains were typically more active than snakeskin and 1/2 yellows, which may have affected their predisposition toward shoaling in general. The IGEs with the strongest % ij values occurred when the focal individual expressed the same behavior as its own context group. Overall, the snakeskin strain responded differently with respect to predator inspections in the presence of each context group, but the inuence on predator inspections was the strongest on the snakeskins, across all Just Keep Swimming 43


context groups (% = 0.52 0.14). When paired with 1/2 green and 1/2 yellow context groups, the greatest inuences on snakeskin predator inspections were agitation and orientation, respectively. Bleakley and Brodie showed that, despite inuence levels calculated for differing focal behaviors, the value of % ij were consistently two to three times larger than all others in cases where the the focal behavior was the same as that presented by the context group. This suggested that reciprocal trait interactions were of notable importance to the subjects" social behavior. !Ultimately, this study indicated that direct and indirect genetic effects hold weight when considering the inheritance and evolution of social antipredator behaviors. Within the scope of predator density, wild guppy populations have been found to display relatively rapid social evolution (Seghers, 1974). Predation pressures are not the only factor guiding this social evolution, however; Bleakley and Brodie noted that guppies tend to show genetic diversity within a single shoal and that wild guppies had been found to display an inter-shoal migration rate of approximately 5% over 12 days. This type of inter-group migration leads to individuals of different phenotypes being introduced to shoals on a semi-regular basis. Traveling with new groups exposes individuals to new social pressures, and the possibility of new environmental pressures. With this in mind, a high propensity for an individual to pick-up on the social cues and reciprocate cooperative behaviors like predator inspections, indicative of a high % ij value, should be a very important adaptive trait among guppies. If this propensity is genetically heritable, it would seem that direct genetic effects may inuence the Just Keep Swimming 44


efcacy and result of IGEs and lead to dynamic phenotypical adaptation, or the differentiation of expressed traits during an organism"s lifetime. C. Epigenetic effects A Theoretical Proposal !Singer, et al. (2010) constructed a theoretical argument suggesting that LINE-1-mediated retrotransposition could generate neuronal DNA sequence diversity by impacting the transcriptome, which is the set of all RNA molecules in one or more cells within cell population. For background information concerning the cellular processes that facilitate the subject matter of this study, consult Appendix A on page 108. Higher expression of certain genes will lead to a change in the phenotype of an organism. This implies that altering the expression of neuronal genes could lead to changes in behavior, such as adaptations to schooling behavior. Singer, et al. began by explaining that L1 retrotransposition occurs during neurogenesis (the growth and development of nervous tissues), and can lead to a state known as #somatic mosaicism." Somatic mosaics are populations of a specic cell population that display varied gene expression. During early development, while the central nervous system is forming, and later during adult neurogenesis, L1 elements are mobilized. Singer, et al. describe a previous experiment during which L1 elements in rat neuronal stem cells (NSCs) were tagged with a uorescent dye and observed. Fluorescence was only observed following a successful retrotransposition event, and occurred at a low frequency. When the NSCs underwent differentiation, however, a high rate of L1 retrotransposition was observed. Human neural precursor cells (NPCs) were also found to support retrotransposition of L1 elements. Indeed, retrotransposition Just Keep Swimming 45


events in NPCs occurred at rates of up to 80-800 new retrotransposon insertions per cell. L1 elements are not simply sources of increased gene expression, however; they can also be #silenced" by epigenetic effects. !Epigenetics is the study of changes in phenotype or gene expression not caused by changes in the DNA sequence. Epigenetic factors are molecules that bind with proteins which the DNA wraps around for compaction and gene regulation, called histones. Binding with these histones alters the extent to which DNA is wrapped around the histone and thus the availability of the genes in that region of the DNA sequence; more tightly wrapped sections of DNA are #unavailable," making those genes inactive. These epigenetic factors originate from many sources, such as diet, drugs, aging, environmental chemicals, and development. A common epigenetic factor is a methyl group, which is an organic compound featuring a carbon bound to three hydrogens. Methyl groups can bind to histone #tails" and activate or repress genes in the vicinity. Epigenetic factors are not permanent, however; bonds between these factors and histones can be severed. Singer, et al. mention that L1 elements are subject to methylation and histone modication, which silences them. When the epigenetic factors are removed, however, these elements are activated. The upor down-regulation of these L1 transpositions in individual cells will inevitably affect the transcriptome, leading to somatic mosaicism across populations of neurons. Are the overall effects on gene expression for an individual cell random? Despite researchers" efforts to control and standardize genotype and the environment in studies using isogenic (having closely related genotypes) mice populations, various Just Keep Swimming 46


phenotypes had been observed throughout such experiments. This phenotypical diversity implies a source of random variability. !The seeming randomness of phenotype in study populations is known as #intangible variance." Singer, et al. suggested that this intangible variance may have a mechanism in L1 transposition, and mentioned that the modulation of the ring pattern in a single rat neuron had been found to alter the subject"s behavior. This evidence suggested that even subtle effects of L1 transposition could alter behaviors, leading to the broadening of the behavioral spectrum from a single genome. If neural diversity could arise from controlled conditions, what then would be the effect on L1-mediated somatic variance from environmental factors? !Singer, et al. made it clear that no direct link between physiological cell stressors and augmented L1 retrotransposition had, to date, been discovered in vivo (inside a living organism) but that some evidence existed to suggest a stress response, triggered by environmental factors. It was discovered that L1 mRNA and proteins sometimes bunch together into #stress granules," which could serve as a reservoir of these structures that could be activated by stress stimuli in response to environmental factors. It was also suggested that androgenic steroids, such as testosterone, could induce L1 activity. This suggestion gained merit from evidence that steroid hormones play a part in embryonic brain development, stress responses, and behavior. Since stress responses are fairly common in the life of an organism, a regular increase in L1 activity would lead to Just Keep Swimming 47


more frequent variation in the genome of somatic brain cells. The question lies in whether or not these modications could provide an evolutionary advantage. !An important factor to consider when trying to understand if L1 retrotransposition could be an evolutionary advantage is that it has been retained over many generations of all kinds of life. It is possible that a metabolic mechanism to actively prevent the process would be too costly to initiate, but it is equally possible that the genetic diversity in neuronal cells that results from L1 retrotransposition could give rise to new traits and new phenotypes that would be subject to natural selection, and could therefor be favored by it. The traits modied by L1 retrotransposition could provide an evolutionary advantage to new environmental circumstances and, in the case of behaviors, could lead to new, successful strategies for overcoming selection pressures such as predation or changing food sources. If the theory posited by Singer, et al. has truth to it, somatic mosaicism in brain cells would lead to a wider variety of possible emergent phenotypes from a single genome. Adaptability of this kind lends itself well to natural selection, and the idea that a mechanism such as that proposed by Singer, et al. is an enticing one when attempting to consider driving forces of evolutionary pathways. !While Singer, et al. proposed their theory of LINE-1 retrotransposition"s role in creating behavioral diversity based on research conducted on rat and human cells, it is not unreasonable to imagine a similar process occurring in the neurons of sh and other vertebrates. Social and environmental cues have been shown to alter behavioral tendencies in live subjects (Chapman, et al., 2008 and Just Keep Swimming 48


Bleakley and Brodie, 2009, respectively). Perhaps transposition of genetic elements in the brain cells of sh plays a role in behavioral plasticity of that individual. If so, then could this mechanism be controlled for? Could it be manipulated for a degree of control over behavioral tendencies? More research is certainly vital to clarifying and solidifying the details and validity of this new theory. !Behaviors like schooling are made up of inherited traits and adaptive traits. To a certain extent, behaviors are regulated by genetic tendencies, but behaviors are reinforced and modied in large part by the combination of external factors throughout the exhibiting individual"s lifetime. Bleakley and Brodie (2009) produced evidence to indicate that genetic variance has an affect on both behaviors and the ability of an individual to adapt their behaviors. Guppies from different genetic populations were shown to express varying tendencies toward cooperative behavior, and their propensity to modify their phenotype based on social cues was subject to a comparable variability. On a cellular level, the theoretical mechanisms put forward by Singer, et al. (2010) provide enticing new possibilities for explaining stochastic environmental effects on behaviors like schooling. Whether or not the induced somatic changes to neuronal cells are random, they may bring about new phenotypes, which would then be subject to natural selection. The adaptation of schooling behavior works in tandem with other adaptations; an organism"s physiology is also ever-changing. Over thousands of generations, shes" senses have adapted not only to suit their physical surroundings, but their social environment as well. Just Keep Swimming 49


Chapter 3. The Senses and Their Effect on Schooling Dynamics !Senses allow organisms to give shape and meaning to their environment; various stimuli are categorized and archived by the brain, based on the senses used to perceive those stimuli. Just as behaviors evolve, so do sensory organs. A quick glance at any organism can provide clues as to what types of senses that organism employs, and in what proportions. Bats tend to have large ears, which are used to capture as much sound as possible, especially the returning sound waves of their sonar. Other nocturnal animals may have considerably large eyes in proportion to the rest of their head, used for taking in as much light as possible in low-light conditions. Sometimes senses are even entirely removed over the course of evolution. Many species living beyond the reach of light in the deep areas of the ocean no longer have functioning eyes. It was once believed that vision played a much more important role in schooling than other senses (Breder, 1951), but it has been suggested that vision plays a relatively small role in cohesive schooling behavior (Partridge and Pitcher, 1980). Fish and amphibians share a unique sense known as the acoustico-lateralis sense that allows the sh to detect subtle shifts in water currents passing over the body and offers the sh a better understanding of its moving surroundings (&ur' i( -Blake and van Netten, 2006; Liao, 2006). A. The Acoustico-Lateralis Sense: The Lateral Line !The acoustico-lateralis sense is a network organ known as the lateral line. The lateral line network consists of receptors known as neuromasts, each of which is a grouping of hair cells, similar to the sensitive hairs found in the inner Just Keep Swimming 50


ears of vertebrates, surrounded by a gelatinous protrusion, known as a cupula. Neuromasts are often recessed into grooves running along the sides of the sh, but canals with neuromasts are also found on the head, around the eyes and along the jaw. There are two types of neuromast: supercial and canal neuromasts. Supercial neuromasts are free standing on the outside of the lateral line canal, and canal neuromasts are located within the canal itself. Water passes over each neuromast, bending the hair cells. This bending triggers nerves connected to the neuromasts, and the collective data of each stimulated neuromast provides the sh with a #picture" of the surrounding water movement ( &ur' i( -Blake and van Netten, 2006; Liao, 2006). !Kršther, Mogdans, and Bleckmann (2002) sought to better understand the brainstem processing of lateral line information from both supercial and canal neuromasts of goldsh ( Carassius auratus ) specically the area of the brainstem known as the medial octavolateralis nucleus (MON). Drawing on prior research, Kršther, et al. suggested that the MON had at least two #channels," which process local hydrodynamic information, and another which processes more complex water movements, such as those generated by a moving source. Kršther, et al. combined these factors by testing lateral line responses from a vibrating local source in still and running water. Fourteen goldsh were used in the experiment; subjects were held stationary, submerged 1 cm below the surface of the water, with electrodes penetrating the tissue of the brainstem. Brain activity was digitized, monitored and recorded on a computer. Lateral line stimulation was provided by an 8 mm diameter 50 Hz vibrating sphere. The experimenters Just Keep Swimming 51


determined #units" of the lateral line by monitoring the areas of the brainstem that experienced activity when the vibrating sphere was placed at different locations along the sides of the sh. The process was repeated in the presence of a constant current, which began without the presence of the vibrating sphere. Overall, 46 lateral line units were designated and monitored for their activity, specically their neural discharge rates. !In still water, 37 units increased in activity, while 5 units decreased in activity. Another 2 units increased and decreased in activity, depending on where the vibrating sphere was in relation to the subject"s body. Two other units did not change in discharge rate, but their discharges occurred in phase with the pulses of the vibrating sphere. In running water without the vibrating sphere, 31 units appeared to be ow-sensitive, meaning their ongoing discharge rates were signicantly different from those in still water. Seventeen of those units, discharge rate increased, and 11 others decreased discharge rate as water velocity increased. The remaining 3 of those units initially experienced an increase in discharge rate, but then reached a maximum and began to decrease as water velocity increased further. The remaining 15 lateral line units had no response to the running water, and were designated #ow-insensitive." Upon testing unit responses in running water with the vibrating sphere, Kršther, et al. suggested that there were three types of lateral line units based on their responses to running water and the masking of responses to the sphere by the running water. !Type I units were characterized by a response to running water, meaning either increased or decreased discharge activity in running water. The responses Just Keep Swimming 52


of the 27 identied Type I units to the vibrating sphere were masked by running water. Of these, 17 were masked in terms of discharge rate, and 10 were masked in terms of both discharge rate and phase-matching. Type II units, of which there were 7, were those that displayed insensitivity to water ow. Additionally, Type II units were characterized by a lack of masked responses to the vibrating sphere in running water. As in Type I units, Type II units responded with either an increase or decrease in discharge activity. Lastly, Type III units, totaling 7, were also ow-insensitive, but differed from Type II in that responses to the vibrating sphere were masked by running water in terms of discharge rate, phase-matching, or both. More specically, overall activity of Type III units in running water was not different from that in still water, but discharge rate and phase-matching to the vibrating sphere were lower than those in still water. Figure 5 shows the differences between the three types of lateral line units, separated by: sensitivity to running water, masking of discharge rate, and masking of phase-coupling. Kršther, et al. concluded that the ow-sensitive Type I units were most likely responsible for innervating supercial neuromasts, whereas ow-insensitive Type II units would innervate canal neuromasts. Just Keep Swimming 53


Figure 5. Characteristics of type I, type II and type III lateral line units. Box-and-whisker plots are shown representing median values and 10, 25, 75 and 90 percentiles as well as data points below the 10th percentile and above the 90th percentile. (A) Slopes of regression lines tted to ow-response functions. (B) Masking of dipole-evoked discharge rate in running water. Integrals of rate-level functions measured in running water are expressed as percentage integrals of rate-level functions measured in still water. (C) Masking of phase-coupling to the vibrating sphere. Integrals of response functions measured in running water are expressed as percentage integrals of response functions measured in still water. Asterisks indicate statistically signicant differences. The denotation #n.s." signies no signicant difference ( P > 0.05). (From Kršther, Mogdans and Bleckmann, 2002) Just Keep Swimming 54


!If these designations were correct, then supercial neuromasts would likely be necessary for rheotaxis (a term applied to the facing of an oncoming current), and that canal neuromasts help in orienting, this suggests that they are essential locating hydrodynamic sources. These data pertain to tests conducted on goldsh, a species prone to relatively calm water conditions. In other sh species, the lateral line innervation and units may be more or less pronounced, based on the conditions that species thrives in. Nevertheless, the data produced by the studies of Kršther, et al. provided signicant new information regarding the sectionalized nature of the lateral line, and how it is processed in the MON. Differentiation of the signals interpreted by the lateral line is integral to understanding the lateral line"s functioning, but one must also consider the acuity of the lateral line. !The lateral line network can determine not only that there are objects or animals nearby, but also accurately provide information regarding the proximity of a stimulus. Janssen and Corcoran (1998) studied the ability of the lateral line to determine distance to a stimulus, using four mottled sculpin ( Cottus bairdi ). The subjects were blinded by way of surgical eye removal to prevent them from seeing the stimulus, a small oscillating plastic sphere. In this experiment, sh were trained to remain stationary on a glass platform, which was moved to various distances from the stimulus sphere. Furthermore, the sh was trained to respond to the sudden oscillation of the stimulus sphere by ipping sideways and touching its mouth to the sphere. Upon a successful touching, the sh was rewarded with a piece of squid mantle, but only if the sh"s mouth made contact Just Keep Swimming 55


with the sphere. Trials were monitored from below, by video camera, to facilitate the measuring of the target distance and response distance, determined by the distance of the subject from the sphere and the point on that distance line at which the subject"s snout intercepts after the stimulus, respectively (Figure 6). The results of these tests showed clear trends, and demonstrated the lateral line"s ability to judge distances of nearby water disturbances, without the aid of visual cues. Figure 6. Measurements of sculpins responding a vertically oscillating sphere (bead). The view is from below. "Target distance" is the shortest distancs from the surface of the sh"s trunk (origin) to the bead, with the sh in its prestimulus onset position (initial position); this line is the "target line." "Response distance" is the distance along the target line from its origin to where the snout intercepts the target line. (From Janssen and Corcoran, 1998) !The overall success of subjects in touching their mouths to the stimulus sphere was similar in subjects before lateral line blocking (63% of 180 trials) and subjects with one blocked canal in trials where the stimulus was on the side of Just Keep Swimming 56


the intact lateral line canal (66% of 80 trials). Those with blocked canals in trials where the stimulus was on the side of the blocked canal yielded less success (29% of 80 trials). Over all trials, response distance over target distance showed a clear linear progression in subjects before canal blocking, with the linear progression breaking down in subjects after canal blocking (Figure 7). !These data demonstrated a degree of variance, however, indicating that the ability of sculpins to determine distance from a stimulus source via their lateral line uctuated from sh to sh. Janssen and Corcoran noted that the largest sh of the four was the most accurate and the smallest was the least accurate, and suggested that there could be a size effect on the accuracy of the lateral line. This difference in accuracy may have been a product of the increased proportional distance that smaller individuals had to travel to the sphere, or a reduced precision in smaller lateral line trunks. The results showed a reduction in precision with blocked canals, but not a signicant difference in distance determination ability. These data support the hypothesis that the lateral line can accurately determine distance to a source of water disturbance. Just Keep Swimming 57


Figure 7. Response distance versus target distance for stimuli for (A) sh before canals were blocked, both sides pooled (B) unblocked sides of sh after the other side was blocked, and (C) blocked sides of sh. (From Janssen and Corcoran, 1998) Just Keep Swimming 58


!Throughout their lives, sh grow accustomed to particular movement patterns that indicate the movements around them, such as the spasms of a dying sh or the wake caused by a cohort swimming ahead. Being attuned to the movements around them on a tactile level allows schooling sh to travel together even under conditions of zero visibility. The ability to accurately determine one"s proximity to a nearby conspecic would aid signicantly in staying organized in a school, and not bumping into one another. In conditions where visual information is unreliable, such as at night or in dense schools where moving bodies completely overwhelm an individual"s eld of vision. This acuity of the lateral line could be used for offensive purposes, as well, such as locating prey. Lateral line cues are most useful within a xed range, but their usefulness diminishes as the target distance increases. Beyond the effective range of the lateral line system, vision plays a much stronger role. Locating predators, prey, and conspecics is entirely necessary in the daily life any organism. For sh, the tandem use of vision and lateral line is necessary for the level of sensory input needed to effectively school and survive. With this in mind, Partridge and Pitcher (1980) sought to determine the relative importance of both vision and the lateral line in the mechanics of schooling behavior. B. The Roles of the Acoustico-Lateralis Sense and Vision in Conjunction !Partridge and Pitcher tested the relative importance of these senses using saithe ( Pollachius virens ), some of which had been blindfolded with cones of opaque lm afxed to their eye-sockets, or had their lateral lines sectioned. Schools of 20-30 saithe were videotaped swimming around a circular channel Just Keep Swimming 59


while illuminated by a red spotlight which cast shadows of the sh at an angle along the bottom of the tank. This angled shadow-casting allowed the threedimensional placement of each individual to be calculated. Test schools consisted of ve blindfolded sh, ve sh whose lateral lines had been severed on both sides, ve sh with both blinders and sectioned lateral lines, and randomly selected control sh under neither condition. It was found that none of these conditions prevented sh from schooling, but specic facets of schooling were affected. !Several observations were made of the the test schools: orientation of neighbors, nearest neighbor distance, velocity correlations, and response to a startling stimulus. Blind sh and control sh had similar orientations throughout the test period, whereas laterally-sectioned sh were far more likely to be oriented at a sharp angle (approaching 90 degrees) in relation to their school mates. Blind sh had marginally increased nearest-neighbor distances than controls, whereas laterally sectioned sh maintained closer distances to one another. When startled with a visual stimulus from above the tank, blind sh only responded if their nearest neighbors did, and with a high degree of lag. Laterally sectioned sh responded with similar latencies to control sh, but were the only subjects to actually collide with one another, following four out of ve presentations of the fright stimulus. Partridge and Pitcher also observed the sh"s ability to match the velocities of its neighbors. They observed that blinded sh more closely matched the velocities of their rst two nearest neighbors than control sh, but also that laterally sectioned sh did not experience a decline in Just Keep Swimming 60


velocity-matching ability. The lack of effect on velocity-matching from lateral sectioning seems strange, given the results of the other tests. It may be the case that saithe do not need their lateral lines to keep track of their neighbors" relative velocities. The results indicate that vision and lateral line both play important parts in schooling mechanics, at least in the saithe. The lateral line appeared more important for proper school orientation, while vision was mainly used for threat detection and keeping track of the relative positions of school mates. The lateral line network is a complex organ, though, and its scope of capabilities extends beyond tracking the positions of nearest neighbors. As an organ capable of interpreting changes in water movement, it would make sense for the lateral line to aid sh in navigating currents. !Throughout their lives, fresh and saltwater sh alike encounter currents, and navigating these currents is made easier by the lateral line network. Freshwater sh living in rivers and other environments with regular currents spend much of their time swimming against these currents. One of these species, the rainbow trout ( Oncorhynchus mykiss ), became the focus of a study investigating hydrodynamic placement of sh to minimize effort of swimming in turbulent currents (Liao, 2006). These unique #kinematics" were called K‡rm‡n gaiting, which results from the exploitation of residual uid momentum from uctuating water vortices in order to generate thrust. Four trout were used to test the following variables: whether or not sh would alter their K‡rm‡n gait kinematics when either their lateral line or vision was blocked, and whether or not sh in those conditions would change their preferences to reside in different Just Keep Swimming 61


hydrodynamic environments. The lateral line was chemically blocked so inner ear function was not also impeded, and vision was blocked by performing tests in the dark. Fish were exposed to a solution of 0.15 mmol -1 cobalt hexachloride in calcium-free de-ionized water at 15 degrees Celsius in order to block the mechanoreceptors of the lateral line. Tests were conducted with single sh in a test tank, featuring a uniform current, disrupted by a cylinder placed in the center of the current, and sh were monitored for the position of their head, center of mass, and tail tip relative to the midline of the body. The sh were also monitored for their body wavelength, body wave speed, their downstream distance from the cylinder, maximum body curvature, and tail-beat frequency. Four conditions were tested: control sh with functional vision and lateral line, blinded sh with a functional lateral line, sh with vision but a non-functional lateral line, and sh without both vision and a functional lateral line. !Across all conditions, the average tail-beat frequency was not signicantly different Fish with blocked lateral line canals exhibited alteration of all other K‡rm‡n gait variables; they K‡rm‡n gaited farther downstream from the cylinder exhibited longer and more variable body wavelengths than controls, and their body waves traveled faster toward the tail When both vision and the lateral line were blocked, maximum body curvature was lessened. Fish with intact lateral lines had signicantly higher body curvatures in the dark Fish from each condition displayed different preferences for different hydrodynamic locations around the cylinder. Trout with functional lateral lines spent most of their time K‡rm‡n gaiting in the light, but preferred to entrain (hold station in an area with Just Keep Swimming 62


less current) in the dark. Some sh with blocked lateral lines would still K‡rm‡n gait, but others preferred to entrain. Fish with both blocked lateral lines and vision would always entrain. Across all test subjects, the time spent K‡rm‡n gaiting in the light was higher for sh with intact lateral lines (50 of 60 minutes; 83% of the test time) than those without (25 of 60 minutes; 41% of the test time). In the dark, most sh would entrain rather than K‡rm‡n gait (40 of 60 minutes, 67% of test time with intact lateral line; 44 of 60 minutes, 73% of test time without an intact lateral line). These results are all with respect to single sh, but they apply to groupings of these sh, as well. !Groups of sh acclimated to high-current areas display K‡rm‡n gaiting, and, in these cases, the group members around them may be the sources of the exploited water vortices. Without functional lateral lines, these sh encounter difculty in efciently exploiting vortices in order to engage in K‡rm‡n gaiting. Open water species may utilize major oceanic currents for migratory purposes, and this behavior would also benet from the use of the lateral line. Applying the data from the study above, sh in a natural setting would likely be rendered incapable of proper K‡rm‡n gaiting and functional navigation of strong currents without their lateral lines, particularly in the dark where not even visual cues are available. These problems could lead to fragmentation of the group and increased exposure to predators, as well as an unnecessary increase in energy expenditure. !While schooling, sh heavily utilize their vision and acoustico-lateralis senses to maintain cohesion. This is not to say that other types of sensory input Just Keep Swimming 63


are not also used; auditory cues are integral to the lives of sh, as well. Due to unfortunate time constraints, however, this thesis is unable to detail the research concerning the inner ear functions of teleost shes. Fish use their senses to simultaneously prevent fragmentation of the group while avoiding collisions with one another. This balance allows for the high degree of organization seen in high-population schools, like those of smaller pelagic species. Gregarious organisms from sh to birds exhibit these dynamics, and formulas have been derived from the study of these aggregations. These formulas have described a series of zones surrounding each individual of the herd, ock, or school. The #inuential zone" model was explained by Gueron, Levin, and Rubenstein (1996). C. The Inuential Zones Model: A Threshold of Desired Personal Space Gueron, Levin, and Rubenstein (1996) created a simple model for understanding the coordinated movements of an organized or familiar school, which outlines a series of inuence zones that affect an individual"s choice of short-term spatial destination. These zones, described in this section, are found around each individual in a shoal, and are known as the stress zone, neutral zone, attraction zone, and the rear zone. An individual"s path through these zones follows a few assumptions: the integrity of the group is maintained by individual movements, which are affected only by the positions of neighbors in a restricted range, and school populations can be subdivided into "speeders" and "laggers." Speeders are typically adults, which may be searching for food or mating opportunities, and laggers are often juveniles that cannot keep the same pace as adults. Just Keep Swimming 64


!The stress zone is a product of the personal preference for space, which causes an individual to turn and move away from an #intruder" within this personal space. The size of this space may differ from individual to individual, and species to species. Depending on the direction from which the intruder arrives, the focal sh will react differently; the sh will either speed up or slow down to avoid collisions with rear or front-oriented intruders, respectively, or move laterally in the direction opposite of an incoming intruder. In this way, sh in the stress zone repel one another much like two magnetic poles of the same sign repel each other. If there are no intruders in the stress zone, the individual shifts its focus to the neutral zone. !When dealing with the neutral zone, the individual does not respond to neighbors unless the neighbors are all to one side of the individual. If this is found to be the case, the sh will move in the direction of those neighbors to not become separated. Furthermore, if there are no other neighbors in the neutral zone, the sh shifts its focus to the attraction zone. It is often accepted that this attraction zone begins past two body lengths away from the individual. When the nearest neighbors are all to one side in the attraction zone, the individual will increase its speed in order to close the gap. If there are neighbors in the attraction zone on both sides, the individual increases its speed but does not change direction. If there are no neighbors in the attraction zone or neutral zone, the individual focuses on the rear zone. If there are neighbors in the rear zone, the individual will slow and allow them to catch up. Just Keep Swimming 65


!The relative positions of sh in a school can lead to the distinction between two types of schoolers, trailers and leaders. Gueron, Levin, and Rubenstein (1996) dened leaders as sh with no neighbors in the rst three zones, but neighbors in the rear zone. Consequently, trailers are dened as any sh with neighbors in one or more of the rst three zones. The aforementioned speeders and laggers are thus more likely to be leaders and trailers, respectively, and both distinctions come with unique costs and benets. Leaders, for example, are able to make their own choice of directional movement, whereas trailers typically do not. Conversely, trailers often benet from the protection of more central positions within the school, while leaders are always on the periphery. !In the presence of a threat, the inuential zones model is more likely to be strictly adhered to, as individuals in the school will want to remain protected from the threat of predation. The zones may tend to get smaller, however, as individuals struggle to surround themselves with conspecics. The stress zone will get smaller, because threatened sh will not want to avoid closeness when threatened. The neutral zone will shrink, as well, bringing the attraction zone in closer. Fish at or near the periphery will more quickly attempt to close any distance between themselves and the rest of the group. Predation avoidance may also be marked with sudden direction changes, which can shift a school member"s status as a leader or a trailer. If a hard turn is made by the school, the individuals on the anks of the school will become the leaders, while the former leaders assume roles as trailers. Actions such as those may not be the result of cooperative efforts to avoid predation, however; an inherent sense of self Just Keep Swimming 66


preservation, combined with the #rules" of inuential zone model, may be implicit in the unied movements of dense aggregations of sh and other gregarious organisms. This idea of a #selsh herd" does not change the benets or costs inherent in schooling, nor does it change the dynamics of the group itself. The selsh herd serves to explain the mechanics behind the decisions made by school members, including the decision to join a particular school. Chapter 4. The Selsh Herd and Methods of School Choice Schooling behaviors in those species that exhibit them have been selected for over generations of natural selection, and the basic selection pressures for and against them have not changed signicantly. Fish today still face the dangers of predation and starvation. For the most part, staying within a group helps to alleviate the effects of these pressures (See Chapter 1 for details on the costs and benets of shoaling behavior. Fish that spend their entire lives schooling in large groups have undergone physical adaptations to better suit their bodies to the task of maintaining group cohesion (See Chapter 3 for information concerning the senses of sh and their role in the school cohesion). The question remains of how sh decide which schools to join. It is clear that sh typically aggregate with members of their own species, but, as outlined in Chapters 1 and 2, more nuance needs to be considered when the confusion effect is in play. Size-matching, color-matching, and homogenous movement are all viable ways of increasing the benets of the confusion effect. In order to maximize their defensive yields, sh should therefore select school mates that complement their own physical characteristics and maximize the potential gains from predation Just Keep Swimming 67


protection mechanisms. The motivation for self-preservation is inherently selsh, and thus lays the foundation for the selsh herd model, proposed by Hamilton (1971). A. Hamilton!s "Selsh Herd! Model and the Nearest Neighbor Approach Strategy In 1971, Hamilton suggested that gregarious behavior did not evolve through benet to a population or species, such as a school of sh being protected from predators due to a coordinated formation. Instead, the #selsh herd model" states that the foremost reason for aggregation serves to benet the individual. Aggregation itself is described as a form of cover-seeking behavior performed by every individual in the group. Hamilton explains this concept with a scenario depicting a small pond populated by a colony of frogs and a solitary water snake. Examples of the interactions between these prey and their predators are explained rst in one dimension, then expanded to two dimensions. !The frogs of this pond have learned that the snake feeds only at a certain time-frame in any given day, and that it prefers to capture prey in the water. As a result, the frogs leave the pond at that particular time of day, but do not venture far from the rim due to the presence of other, terrestrial predators. The watersnake thus observes the grouped frogs at the pond"s edge, and snatches the nearest one. Having learned to leave the pond during the snake"s active hours, the frogs avoid being captured in the pond. In a one-dimensional sense (i.e. a straight line), the greatest odds of survival arise from reducing the empty space around the self, the #domain of danger." Just Keep Swimming 68


!The domain of danger is dened as half the length of the gap between two neighbors, and represents the area where predatory threat applies to that individual. Depending on where the snake comes out of the water determines which frogs will be nearest to it, and thus targeted for capture. More isolated individuals appear closer, and easier to capture, than clustered groups. Hamilton thus explains that one way of reducing one"s domain of danger is by approaching one"s nearest neighbor, thereby decreasing the length of the domain. Furthermore, the most effective way to reduce this domain is for the frog to hop between two nearby individuals,which reduces the domain on both sides. Unfortunately for the frogs, individuals will not be satised when their neighbor leaps away from them in favor of a denser group. These frogs will then do the same, so that nearly all of the frogs are constantly shifting positions. Eventually, all the shifting will result in dense heaps of frogs, and the individuals that continue to jump back and forth between groups will then bear the greatest risk. The most important aspect of this scenario, though, is that even the selsh avoidance of a predator can lead to aggregation. !Hamilton (1971) supported his suggestion of the selsh herd with a description of real-world observations: "With sh schools observers have noted the apparent uneasiness of the outside sh and their eagerness for an opportunity to bury themselves in the throng and a parallel to this is commonly seen in the behaviour of the hindmost sheep that a sheepdog has driven into an enclosure: such sheep try to butt or to jump their way into the close packed ranks in front. Behaviour of this kind certainly Just Keep Swimming 69


cannot be regarded as showing an unselsh concern for the welfare of the whole group." !When expanded to two dimensions, the model of predation and prey behavior takes on new properties. One such property is that each individual"s domain of danger becomes a polygon. Each point inside this polygon is closer to that individual than any of its neighbors. An imaginary line leading from one individual to its nearest neighbor would bisect the line between them at a right angle. The reason for well-dened domains of danger within the group, as well as on the periphery, is that the school or group may not notice a concealed predator, which waits to strike until a group member ventures too close (Figure 8). The predator may need to wait until the entire group occupies the surrounding area for such an opportunity to present itself. The moment the predator presents itself, the domains begin to shift as group members make an effort to reduce their chance of capture. !The reduction of one"s domain of danger is represented by a decrease in area of the polygonal domain surrounding it (Figure 8). If an individual is relatively isolated, however, its domain may actually increase during an approach to a neighbor. These cases are most often observed in polygons with a low number of sides. If an individual is well-surrounded by group members, its domain of danger will often decrease on an approach to a near neighbor. Upon reaching its nearest neighbor, an individual will likely try to decrease its domain even further by moving to a side of the neighbor that minimizes its domain of danger. Attaining this optimal placement is incredibly difcult in a two-dimensional Just Keep Swimming 70


plane, though. Like with the frogs in the original example, the group members in this scenario would also be simultaneously striving to achieve cover; with the added area to maneuver, the chances of optimal placement are much lower in two dimensions than in one. !Krause and Tegeder (1994) further investigated Hamilton"s nearest neighbor approach strategy using three-spined sticklebacks, hypothesizing that the mechanism behind this behavioral model is such that a sh will follow a path that minimizes the time necessary to approach its nearest neighbor. Just Keep Swimming 71


Figure 8. Domains of danger for a randomly distributed, two-dimensional group of animals when a hidden predator may attack the nearest animal. Thin arrowed lines represent the nearest neighbor of each animal. Thicker arrowed lines represent supposed movements of particular prey. Position of prey is given a lower case letter and the domain obtained from each position is given the corresponding upper case letter. The pattern underlining the domain letter corresponds to the pattern used to indicate the boundary of the appropriate domain. Dashed lines represent the boundaries of domains that come into existence after the rst movement. The inset diagram shows geometrical algorithms for minimum domains attainable by a (upon reaching c ), and by y In the case of y the minimum domain is equal to the triangle of neighbors and is obtained from the orthocenter of that triangle. (From Hamilton, 1971) Just Keep Swimming 72


! Figure 9. Test sh approached a stationary conspecic from various distances and various initial body orientations (measured in degrees). Data were split into three angle classications: (a) 0-50 (N = 75), (b) > 50-100 (N = 31) and (c) > 100-150 (N = 41). Slopes were not signicantly different, but intercepts were (0-50 versus 50-100: F 1 = 32.880, P < 0.001; 50-100 vs 100-150: F 1 = 44.538, P < 0.001; 0-50 vs 100-150: F 1 = 260.114, P < 0.001). (From Krause and Tegeder, 1994) !To do this, three sequential experiments were conducted. The rst experiment simply tested approach times for a single sh approaching another single sh at distances between 3.4 cm and 36 cm. Both sh began trials conned in glass cylinders, with the test sh beginning the trial oriented at an angle between 0 and 150 degrees with respect to the stimulus sh. The overhead lights were quickly icked off then on again to simulate an aerial predator"s shadow crossing over the sh, followed by the release and Just Keep Swimming 73


observation of the test sh. Test sh responded to the stimulus as expected, by closing the distance between themselves and the other sh. When test sh began trials oriented at an angle from the other sh, Krause and Tegeder observed that the test sh always rst turned to face the second sh, followed by closing the distance between them. As the measure of the angle increased, so did the average time spent turning, as well as average total time spent approaching the near neighbor. At a distance of 10 cm, for example, a 90 degree turn constituted approximately 35% of the total approach time. Figure 9 shows that the average approach time increased over increasing distance and angle. Figure 10. The percentage of cases when test sh (T) turned and swam towards the secondary sh (F), plotted against the distance between T and the originally faced sh (N) for the three angles indicated. Each point is based on 20 trials. Arrows give the predicted switch point for 45, 90, and 120 degrees. (From Krause and Tegeder, 1994) !The second experiment expanded upon the rst by including two stimulus sh. Stimulus sh were initially placed equidistant from the test sh in a triangle, with the angle at the vertex measuring 45, 90, or 120 degrees. Test sh were Just Keep Swimming 74


oriented toward one of the stimulus sh, and the distance to this sh was set at 14, 16, 18, 22, and 28 cm. Following the light stimulus, test sh were again released and observed. Krause and Tegeder hypothesized that sh would approach the neighbor that they could reach in the shortest amount of time, even if that meant turning toward the second sh and approaching it. This hypothesis held true in the experiment, as test sh took time to turn and approach their nearer neighbor if it would take less time than swimming straight toward a farther sh. It should be noted that sh did not always turn toward the closer neighbor, but this behavior became more probable as distance increased (Figure 10). !In the nal experiment, the time-minimization hypothesis was tested using 5 free-swimming sh in a test pool. Fish were given time to acclimate and spread out before the light stimulus was introduced. Immediately following the stimulus, sh executed behavior known as a #C-start," a short-term predator evasion reex behavior that involves a sudden sharp turn and rapid acceleration for a brief moment (Eaton and Emberley, 1991). After the C-start sh began aggregating, beginning with the sh located on the periphery of the group. Fish on the periphery tended to begin moving sooner than those in central positions. Fish were designated as peripheral if they were located on a vertex of the smallest possible polygon enclosing the group, and central if they were not at a vertex. For the sake of reducing confusion, only the rst sh to move was actively observed. Taking the initial orientation and distance from nearby sh, the time minimization hypothesis was used to predict the neighbor to be approached with 95.7% accuracy. In 83% of those trials, the sh approached was not simply the Just Keep Swimming 75


closest sh in terms of distance. Krause and Tegeder thus concluded that, when turning time is taken into account, Hamilton"s nearest neighbor approach strategy changes from involving the nearest neighbor in space to involving the nearest neighbor in time, but otherwise holds true. !Simulating the visual presence of a predator clearly elicited a response in line with Hamilton"s selsh herd model. If sh respond to visual information indicating the presence of a predator by approaching their nearest neighbors, then presentation of a simulated predator to senses other than vision should have a similar effect. In 1993, Krause investigated the effects of an olfactory threat-cue, #Schreckstoff," on the aggregation behaviors of shoals of dace ( Leuciscus leuciscus ), and individual minnows ( Phoxinus phoxinus ) introduced into the schools. Schreckstoff, a German word translating literally to "fright material," was rst described by von Frisch in 1938, when it was discovered in the skin of several species from the family Cyprinidae This unique chemical is released when a sh is damaged by a predator, and has been shown to elicit a strong aggregation response in minnows (von Frisch, 1938; Schutz, 1956). !Krause investigated the effects of Schreckstoff on shoals of dace by monitoring the reactions of focal individuals to Schreckstoff in groups of sh habituated to the presence of the compound. Habituation was conducted to minimize the responses of stimulus sh in order to independently observe the reactions of focal individuals. Fourteen sh were introduced to the test tank 20 days before the experiments began. The rst 6 days were an acclimation period, and the last 14 days were spent exposing and habituating the sh to Just Keep Swimming 76


Schreckstoff. Fish were considered to be habituated when reactions to Schreckstoff were reduced to #skitters" and when the distance between subjects did not change signicantly after the introduction of Schreckstoff. Tests were conducted by introducing a naive sh to the habituated school, then introducing concentrated Schreckstoff to the tank after an acclimation period of approximately 20 hours. The test tank was monitored via videocamera beginning 5 minutes before the introduction of Schreckstoff, and ending 15 minutes later. !Before the introduction of Schreckstoff, minnows and dace did not show signicant differences in nearest neighbor distances or their number of nearest neighbors. The test minnows responded to the Schreckstoff by signicantly decreasing nearest neighbor distance and increasing the number of nearest neighbors, whereas the dace responded weakly to the Schreckstoff (Figure 11). Dace occasionally showed brief fright reactions, but ceased after a few seconds. Figure 11 shows that dace exhibited an increase in nearest neighbor distance during the second stage of trials, but no signicant change was observed in median number of near neighbors. After the introduction of Schreckstoff, the median nearest neighbor distance of minnows was signicantly smaller than that of dace. Additionally, the median number of near neighbors was signicantly larger in minnows. Krause speculated that the increased nearest neighbor distance seen in dace was a product of the occasional #skitter" reactions to the Schreckstoff; the skitters may have caused some degree of dispersal between individuals, but not enough to cause the school to draw in again. If the speculation was true, then this study indicated that sh can successfully be Just Keep Swimming 77


acclimated to the presence of Schreckstoff. For the minnows, however, Hamilton"s selsh herd model had held true even in the absence of a visual threat. Olfactory cues could reliably elicit a fright response that adhered to the principles of the selsh herd model. Figure 11. (a) Median nearest neighbor distance and (b) median number of near neighbors before and after the introduction of Schreckstoff for minnow and dace. The 10-min period after the introduction of Schreckstoff is additionally split into two intervals of 5 min. Error bars indicate quartiles across trials ( N = 13). (From Krause, 1993) !The selsh herd model serves as a useful mechanism for predicting the behaviors of gregarious animals when threatened with predation. By extension, Just Keep Swimming 78


the nearest neighbor approach strategy is a reliable tool for threatened animals to minimize their exposure to predators. It may seem tempting to suggest the merits of a group traveling in an evenly distributed formation, such as a lattice. The trouble with this, as Hamilton (1971) pointed out, is that when a predator presents itself and the group members close the distance between themselves and their neighbors, the group will become highly fragmented. When the distance between oneself and any of one"s neighbors is equal, the decision to approach any of these neighbors becomes arbitrary. Lattice formations are not observed in gregarious animals, however, which may be due, in part, to the problem pointed out above. Also, as discussed in chapter 1, subtle differences exist in every individual, which may make them more or less attractive as sources of cover or predatory confusion. How sh determine which schools to associate with has become a major subject of investigation in the scientic community. B. Choosing the Right School: Fish Seek Commonality in Shoal Mates !Throughout the lives of most schooling and shoaling sh, individuals make an active choice to join the specic groups that they subsequently remain with. This choice often takes into account many factors, and these factors tend to follow the paradigms that maximize the potential gains of the confusion effect (Chapter 1). This means that an individual will seek out conspecics that share as many physical characteristics with itself. In addition to visual homogeny, enticing potential shoal mates can be those possessing characteristics that indicate a high degree of tness and carry with them benets to the individual"s tness. For example, an individual detects the scent of a particularly nutritious Just Keep Swimming 79


food type in the vicinity of a particular group, indicating that the group may have recently located a source of the food and could thus lead the individual to the source, as well. The remainder of this chapter explores the preferences of sh species when selecting a new group, and focuses mainly on studies that allowed test sh to actively choose to associate with multiple #stimulus groups." i. Choosing the Right School: Visual Homogeny !As discussed in Chapter 1, many species of sh seek visual uniformity in their schools to avoid oddity, which may attract predators and make individuals easier prey. One of the traits important to uniformity is relative body size; a signicantly larger or smaller individual tends to stand out more than those that t into the group"s common size range. Larger sh in groups of relatively smaller sh were also found to forage less, due in part to the added exposure that comes with minor group fragmentation characteristic of foraging episodes ( Peuhkuri, 1997). More important than body size, though, is body coloration. Whether it"s a solid color or a particular pattern, a single sh that does not match the color scheme of the rest of the school will stand out more than an individual of unique relative size. !The easiest pattern to reproduce is no pattern at all; a grouping of sh with single-color bodies can utilize the confusion effect by creating a swirling mass of yellow, gray, or whichever color the species happens to exhibit. McRobert and Bradner (1998) studied the preferences of white and black mollies ( Poecilia latipinna ) for associating with conspecics of either color. Based on the principles of the confusion effect, it was hypothesized that mollies would prefer to associate Just Keep Swimming 80


with sh of similar coloration to their own. Individuals were tested in a tank with compartments on either end. The central compartment contained two opaque partitions to block line of sight to the opposite side of the tank when the test subject was on one side. Subjects were initially placed in the center of the tank, where both partitions blocked line of sight to either end of the tank. Depending on which assay was being conducted, the end compartments would contain either four individuals of similar or dissimilar coloration to the test subject, or would remain empty. In assay 1, both side compartments were empty. Assay 2 featured four individuals of similar coloration in one compartment, and none in the other. One compartment in assay 3 contained four dissimilarly-colored individuals, and the other was empty. In the nal assay, one compartment contained four dissimilarly-colored mollies, and the other contained four similarly-colored mollies. Trials spanned 10 minutes each, and test subjects were observed and timed for duration spent on either side of the central compartment. Test sh were considered to be associating with a side compartment shoal if they were on that side of the central compartment. !Test mollies of both colors displayed no signicant difference in time spent on either side of the tank during assay 1, when both side compartments were empty. In assay 2, when one compartment contained a dissimilarly colored group, black mollies spent signicantly more time on the side of the stimulus group, indicating that they preferred to be in close proximity to other sh rather than no sh at all. White mollies, however, did not show a particular preference in assay 2. Assay 3 found that both colors of test mollies preferred to associate with Just Keep Swimming 81


mollies of similar coloration when presented with the choice between a similarly colored stimulus shoal or an empty compartment. When faced with a decision between similarly and dissimilarly colored shoals in assay 4, test mollies of both colors showed a signicant preference to associate with similarly colored shoals. The above results strongly supported the hypothesis that the mollies would act to maximize the potential to utilize the confusion effect. Both types of mollies showed signicant preferences for conspecics of similar coloration, but only the black mollies showed a preference to associate with dissimilarly colored conspecics rather than no conspecics at all. This may have been an artifact of the experiment, and replication of the experiment would help to elucidate this confusion. Nevertheless, this experiment presented clear evidence for a relationship between the confusion effect and shoaling preferences. !Not all sh have bodies made up of a single color, though; many tropical sh carry elaborate patterns on their bodies, made up of several bright colors. With so many types of patterns in sh species, a question arises of whether or not sh recognize the ner details of body coloration patterns, and the degree to which subtleties in patterns affect shoal choice. Indeed, Wolf (1985) observed that several brightly colored species in a coral reef environment swam loosely together until frightened, when many of the species converged into compact groups of conspecics (The details of this study can be found in Chapter 1). The species in that study were very visually distinct, but sometimes sh must choose between conspecics with varying degrees of coloration similarity with respect to their own. Just Keep Swimming 82


!Rosenthal and Ryan (2005) conducted an experiment not unlike that McRobert and Bradner (1998); zebrash of various phenotypes (genus Danio ) were presented with a series of choices between striped or stripeless computer generated zebrash (see Figure 12) on small screens placed outside either end of the tank. Five phenotypes of Danios were tested, each with a different level of stripe expression: D. albolineatus D. rerio ( golden strain), D. rerio ( leapord strain), D. rerio and D. nigrofasciatus (Figure 12). Individuals were reared exclusively with members of their own phenotype. The #choice" sections of the tank consisted of the last 10 cm at each end of the tank"s length. Test subjects acclimatized to being in the test tank with the monitors on for 10 minutes, then were observed for another 10 minutes after the presentation of the generated stimulus sh. As in the protocol executed by McRobert and Bradner (1998), Rosenthal and Ryan (2005) recorded the duration each test sh spent in either choice zone. Just Keep Swimming 83


Figure 12. Net preference of each phenotype for stripes ( X SE association time in seconds). A negative value indicates a preference for the stripeless stimulus. Letters above bars indicate signicantly different groups in ANOVA. Representative stills from stimuli used in video playback are shown at left. (From Rosenthal and Ryan, 2005) !Test subjects were observed to associate preferentially with the generated stimulus sh that most resembled their own phenotype. Preference for stripes was seen in the striped wild-types D. rerio and D. nigrofasciatus The stripeless wild-types D. albolineatus and D. rerio gold preferred to shoal with the stripeless animations, though the preference in D. rerio gold was not signicant. The intermediate type, D. rerio leopard somewhat expectantly showed no signicant preference. Figure 12 shows a plot of the preferences observed in each phenotype. The results of this study conrmed the hypothesis that Danios of varying phenotype would prefer to associate with #conspecics" of similar phenotype. These data thus indicated that species of Danios also act to Just Keep Swimming 84


maximize their potential gains from the confusion effect. The problem with this study was that there were only two phenotypic choices for test subjects, while the specimens themselves displayed ve different phenotypes. It might be more costly, but an experiment like this would benet from using live subjects of each phenotype as stimuli. Were that to be the case, more specicity could be achieved in the data; the degree of detail considered by subjects when making active shoaling decisions could be subject to increased scrutiny. This is not to denigrate the results reported by Rosenthal and Ryan, however; the ndings of the study described above did provide further evidence to the link between the confusion effect and active shoal choice by individuals. !As protection from predators is likely the most important benet of aggregation behavior, it stands to reason that maximizing this benet factors heavily into the choices individuals make to travel with particular groups. Indeed, the examples in this section provide strong evidence to this effect; when faced with a binary choice, sh consistently chose to remain in close proximity to groups of similar appearance to their own. The confusion effect is a useful method for avoiding predation, but is less effective singularly than when coupled with the dilution effect. Shoal choice takes on a new element when one considers the inuence of group size on an individual"s active shoal choice. ii. Choosing the Right School: Group Size !Fish benet from the confusion effect by adopting visual homogeny within the group or species, but they can attain an added anti-predatory benet through the dilution effect. The basic principle of the dilution effect is reducing one"s risk Just Keep Swimming 85


of predation by associating with larger groups. The more individuals in a group, the lower an individual"s chance of being captured by a predator that must choose an individual to attack. This choice becomes harder when the ability to distinguish individuals is blurred, which occurs when the dilution effect is exhibited in conjunction with the confusion effect. A panicked school of sh, all the same color and all swimming tightly together, can use these combined effects to confuse the senses of the predator and lessen each individual"s chance of becoming a casualty. If this is indeed the case, then individuals should seek out larger groups with which to associate. !Pritchard, Lawrence, Butlin and Krause (2001) found that zebrash ( Danio rerio ) tended to shoal with larger groups of conspecics when faced with a choice between two groups of different sizes, ranging from 1 to 4 conspecics, with all other factors equal. The secondary goal of this study was to determine the signicance of shoal size vs. shoal activity level in an individual"s shoal preference. As such, only the assays where all factors but group size were equal will be considered here, and the results as they relate to group activity level will be described later in this chapter. Two assays resulted in signicant preference for one stimulus shoal over the other; zebrash signicantly preferred 4 sh over 1, and 4 sh over 2. Over all replicates of these trials, test zebrash signicantly preferred the larger group. Furthermore, group size was found to have a signicant bearing on association across all trials. The results suggest that zebrash prefer to associate with larger groups over smaller groups. The size range for stimulus shoals in this study was rather small, however. An increase in Just Keep Swimming 86


size range for these shoals would almost certainly result in a greater distinction of preference to associate with larger shoals. Lindstršm and Ranta (1993) conducted an experiment to test the shoal preferences of male guppies, Poecilia reticulata based on shoal size and sex ratio of the school. Males were presented with a series of binary choices: 1 female vs. 1 male, 2 females vs. 2 males, and 3 females vs 3 males. While these experiments were intended to conrm that male guppies prefer to shoal with female conspecics, the results also indicate a preference for larger groups by male guppies. Multiple regression analysis showed that, as group size increased, group size accounted for 45% of the variation in how test guppies budgeted their trial time, meaning that 45% of the reason for swimming with either group across all trials was because of the group"s size. Compare that with sex ratio"s contribution, which accounted for only an additional 7% of the variation in time budgeting across all trials. These data show a signicant preference for larger groups by male guppies. Lindstršm and Ranta explicitly cited the role of the dilution effect anti-predatory behavior, and suggested that, while males prefer to associate with females for the sake of increased mating opportunities, they also show a strong predilection for larger groups. Not all shoaling species have been observed to lean toward groups of a particular size, though. !Jenkins and Miller (2006) conducted a simple two-way shoal choice experiment, using the central mudminnow ( Umbra limi ), to determine the extent of shoaling behavior, if any, in the under-studied species. Twenty-two individual mudminnows were tested for shoal preference in three consecutive experiments. Just Keep Swimming 87


Two tanks were placed adjacent to the center tank on either end, and shoals of differing sizes were held in each adjacent tank. In the rst experiment, one tank held 5 conspecics and the other held none. In experiment 2, one tank held 3 conspecics and the other held 7. In the nal experiment, one tank held 3 conspecics and the other held 12. In experiment 1, test sh preferred to spend time near the tank with conspecics rather than the empty tank, suggesting that the species is indeed a shoaling species. In experiments 2 and 3, however, test sh showed no signicant preference for either shoal; they preferred to spend time with either shoal rather than the empty center area. Jenkins and Miller concluded that the central mudminnow is a shoaling species, but that the simple experiment may not have tested enough variables to conclusively claim that central mudminnows display any preference for shoal size. Further investigation of this species should be conducted to elucidate the results of Jenkins and Miller"s study of central mudminnows. !It would seem that not all species of shoaling sh seek out larger groups, as long as a group is available to them. These preferences may uctuate based on the resident predation, and could easily lead to different levels of group size preference even among populations of a single species. For others, though, the principles of the dilution effect encourage individuals to consort with larger groups, and evidence to this effect was described in this section. At the very least, guppies and zebrash have been observed to show preference for larger shoals. Circumstances of shoal choice are not always so obvious, though; sometimes an individual might choose a group simply due to the way they smell. Just Keep Swimming 88


iii. Choosing the Right School: Olfactory Diet Cues !The basis for this section comes from a string of studies suggesting that certain species of sh may opt to follow a shoal if that shoal carries with it the distinct scent of a particularly appealing food source. If these individuals follow such shoals, there is a chance that members of the shoal know where to nd the food in question and can lead naive individuals to these food sources. By indirectly sharing information in these cases, members of a species can avoid malnourishment or potentially starvation, depending on the availability of food in a given area. !Webster, Adams and Laland (2008) sought to investigate the transfer of social information from diet-specic chemical cues and if it affects the selection of diverging groups during shoal ssion, as well as prey patch selection by focal individuals. The subjects of this study were Whitecloud mountain minnows ( Tanichthys albonubes ), a species previously observed to exhibit varying shoal size based on hunger levels (Richardson, 1994). In this study, three experiments were conducted. In the rst experiment, focal individuals were given a binary choice between 5 individuals fed the same diet as themselves, or 5 individuals fed an alternative diet. The partitions isolating these stimulus groups differed from previous binary choice experiments in that they were perforated to allow the transmission of chemical cues. Experiment 2 simulated a shoal ssion event by keeping stimulus shoals in separate mesh cylinders, which were moved and subsequently spatially divided. The aim of this experiment was to determine if subjects could make quick decisions in naturalistic situations based on chemical Just Keep Swimming 89


cues. The nal experiment again presented subjects with a binary choice, this time between a novel, neutral food located adjacent to a stimulus shoal fed on the same diet as themselves and an identical food patch located adjacent to a stimulus shoal that had been fed the alternative diet. It was hypothesized that not only would subjects prefer to shoal with groups that shared their diet, but also that subjects would prefer to feed near those groups. !As expected, focal sh signicantly preferred to associate with stimulus groups that shared their diet. The fact that both groups displayed this tendency indicated that the type of diet the groups were fed did not affect their tendency to group with shoals of the same dietary background. During shoal ssion, sh followed the shoal that shared their diet signicantly more often than not. Fifteen focal sh from group A followed the group A stimulus group, whereas only 3 followed the group B stimulus shoal. Of the focal sh from group B, 14 followed the group B stimulus shoal,and 5 followed the group A stimulus shoal. When deciding which prey patch to feed from, focal sh tended to feed rst from the patch adjacent to the group that shared their diet. From group A, 9 focal individuals fed from the group A food patch, and 4 fed from the group B patch. Thirteen focal individuals from group B rst fed from the group B patch, while only one fed from the group A patch. Notably, focal sh tended to consume signicantly more food from the prey patch associated with their like-diet group. Unlike the other tests, this true of both groups separately. The results of this study consistently support the hypothesis that Whitecloud mountain minnows, Just Keep Swimming 90


when given the choice, would prefer to shoal with groups that shared a regular diet with themselves. !As the study above demonstrated, olfactory diet-cues have a clear bearing on the social preferences of Whitecloud mountain minnows. A more common species, consistently mentioned in behavioral research, is the common guppy ( Poecilia reticulata ). As has already been seen, shoal size and sex-ratio of a shoal weighs heavily on the shoaling choices of guppies. Morrell, Hunt, Croft and Krause (2007) sought to determine whether diet cues play a role in the shoal choices of guppies. To avoid complications from sex-biased schooling, only females were selected for this study. Subjects were divided into groups of 10 and fed one of two diets: bloodworm (a high protein food, but lacking the diversity of a normal guppy diet), or ake food (made up of several foods, but not particularly rich in any in particular). !In the rst of three experiments, focal individuals were exposed to a binary choice between either a group of 4 guppies of the same diet regimen as themselves or a group of 4 guppies with the alternative diet. The second experiment presented two subjects per trial with a Y-shaped maze, where water owed from either branch of the maze into the central corridor. Extracts from either diet were incorporated into the ows, to cause olfactory cues to ow through each branch and into the central corridor where the focal individuals were held and subsequently released. Focal sh were monitored for their initial preference and which compartment they spent more time in. In order to investigate whether the preferences from either previous experiment could be Just Keep Swimming 91


attributed to differences in nutritional value of the diets, experiment 3 was conducted. In this experiment, four groups of ten guppies were assigned to separate holding tanks and fed either diet, two groups of bloodworm and two groups of ake food, until satiation for two weeks. At the end of the two weeks, any growth in the sh was recorded. !In experiment 1, bloodworm-fed sh signicantly preferred shoaling with the group also fed bloodworm. Interestingly, the ake-fed subjects also showed a preference for the bloodworm-fed group. Experiment 2 found bloodworm-fed sh to initially enter the bloodworm scent zone, but no particular preference in akefed sh. Over the rest of the trial times, bloodworm-fed sh preferred to stay in the bloodworm zone, and ake-fed sh again showed no preference. Additionally, the majority of cases showed that both sh entered the preference zones, and a signicant majority of those cases showed that both sh chose to enter the same preference zone. Morrell, et al. thus concluded that the sh were following each other into the preference zones. Experiment 3 found that sh fed either diet increased in mass, but that the sh fed bloodworm experienced more growth than those fed ake food (Figure 13). Just Keep Swimming 92


Figure 13. Results of growth experiment (Experiment 3), showing mean 2 SE weight before and after 2 weeks on either a bloodworm (squares) or ake food (circles) diet. (From Morrell, Hunt, Croft and Krause, 2007) !The results of the above study reveal several things, not the least of which being that guppies clearly prefer bloodworm over ake food. The growth rates recorded in experiment 3 indicate that bloodworm is a much more nutritious food than ake food. With this is mind, the results of experiment appear to show two effects at work. The rst is a corroboration of the results from the work of Webster, et al. (2008); focal sh from the bloodworm group preferred to associate with conspecics that shared their diet. The second effect is that ake-fed guppies preferred to shoal with conspecics giving off olfactory cues from a more nutritious diet. This makes sense, as access to a more nutritious diet would increase the overall health and tness of an individual. As for the peculiar lack of preference in ake-fed sh during experiment 2, Morrell, et al. suggested that the extract of bloodworm may have had a more concentrated odor, and thus Just Keep Swimming 93


pervaded the test arena. The results of this study provided further evidence to the importance of dietary olfactory cues in shoaling decisions, and showed that olfactory cues were important in the shoaling decisions of common guppies. !Joining a group due to olfactory cues may be a temporary affair; a single sh may join an informed group just long enough to learn the location of additional food patches. After all, if predation is low and there is a diminished need to utilize the dilution effect, there is no need to expose oneself to increased competition between shoal mates (Intra-group competition is discussed in detail in Chapter 1). Indeed, in tested groups of banded killish ( Fundulus diaphanus ), 60% of individuals were found to begin swimming alone when food extract was introduced to the tank (Hoare, Couzin, Godin and Krause, 2004). Morrell, et al. (2007) did not only study the role of diet cues in adaptive shoaling decisions; the subject of individual familiarity was also investigated. The role of familiarity with conspecics in choosing a shoal has become the subject of several studies in recent history. iv. Choosing the Right School: Familiarity !When studying highly social sh, such as guppies, one must consider the eeting nature of their shoal afliations. These more social sh tend to come and go, seeking out mating opportunities or other social benets from many different shoals. Within isolated populations there is a good chance that these sh will become repeat visitors, and a degree of familiarity will manifest between certain individuals. This familiarity may factor in when an individual seeks a group for a particular task, such as predation protection or locating patches of food; Just Keep Swimming 94

PAGE 100

familiarity with an individual allows for inside information regarding their reliability for the necessary task. Morrell, et al. (2007) found that bloodworm-fed guppies preferred to associate with shoals including familiar individuals over non-familiar shoals, but that ake-fed guppies actually preferred unfamiliar shoals. A possible explanation for this strange result was that ake-fed sh were aware of the presence of a higher-nutrition food source, and sought the co-presence of unfamiliar individuals for the chance that they may be able to lead the individual to the source of the appealing smell. !The question of familiarity comes with a question of time, as well. If familiarity is attained from exposure to unfamiliar individuals, then how long does it take to acquire this familiarity? Grifths and Magurran (1997) sought to determine the time it takes for individual sh to gain familiarity with one another. Many schooling sh prefer to school with conspecics that are familiar. Female guppies ( Poecilia reticulata ) were placed in a large stock tank together for 4 days until they were then separated into groups of six for the experimental duration of a month. As a control, 36 more females were separated into groups and tested only once 12 days later. During each trial, four females from two randomly chosen groups were placed in separate perforated 1-liter bottles at either end of a test tank and allowed to settle for 15 minutes. At the end of this acclimation period, a single test sh from one of the randomly chosen groups was released into the center of the tank and the time it spent near either of the two stimulus groups (or neither) was recorded. This provided a simple measure of the test Just Keep Swimming 95

PAGE 101

shes" preferences for the sight and scent of familiar or unfamiliar groups. These data were then compared over the course of the experiment"s duration. !Subjects primarily preferred to school with familiar conspecics, and there was a distinct interaction between familiarity and day, indicating that the preference for familiar sh grew over time. The control group, having no previous experience with the experimental protocol, also displayed a marked preference for schooling with familiar conspecics. Time spent schooling with familiar schoolmates was 56.4% in the control test, as opposed to 51.2% in the day-12 test. This similarity indicated that familiarity gained by day 12 was not an artifact of repeated testing. The results of this study led Grifths and Magurran to conclude that it took approximately 12 days for familiarity to be gained between female guppy conspecics, and further concluded that this relatively long period likely leads wild individuals to associate for prolonged periods of time. This kind of behavior is not typically seen in male guppies, but the possibility of it occurring in females is not out of the question (sex-biased shoaling is discussed later in this chapter). !Morrell, Croft, Dyer, Chapman, Kelley, Laland, and Krause (2008) investigated familiarity in guppies in a more practical manner by comparing association patterns and foraging in natural and articial shoals with a novel foraging task. Complete guppy shoals, as well as mixed shoals, were captured from the Arima River in Trinidad. Articial shoals were created from the mixed captures by placing them in a large pool for 24 hours, then removing 80 individuals of similar size and placing them randomly into 10 separate groups. Just Keep Swimming 96

PAGE 102

Once settled, groups both natural and articial were observed for membership of subgroups over 30 minutes. Group sizes in articial shoals were signicantly larger than natural shoals. Small subgroups were more common in natural shoals, and larger groups more common in articial shoals. !Ten minutes after the observation of association patterns, a white plastic cylinder was placed on the water 30 cm from the edge of the pool. The cylinder had an entrance hole, which faced the observer. Next, a pinch of freeze-dried bloodworm was introduced into the cylindrical feeder. Following that, the time it took for the rst sh to approach within 4 body lengths was recorded, then observations continued for 30 minutes. During the observation period, the time was recorded, as well as the identity, each time a sh entered the feeder or fed on the bloodworm. Members of natural shoals approached the feeder and fed much more rapidly than articial shoal members. Overall, at least 3 sh fed in all natural shoals, whereas only 5 articial shoals had even one individual feed. !The results of this study led Morrell, et al. to hypothesize that familiarity must have factored into the clear advantage held by natural shoals. Familiar sh were more likely to follow one another, which was supported by the results that showed natural shoals entering the feeder and feeding in small clusters. Articial shoals did not display this behavior. A possible explanation for this was that members of articial shoals prioritized familiarization over foraging in the apparently safe conditions of the laboratory. Overall, it is not difcult to posit that there was less confusion for familiar groups, as there was no need for an individual to investigate the presence of several strange organisms in an isolated Just Keep Swimming 97

PAGE 103

environment with itself. Further research should be conducted to better discern the cause and nature of the apparent foraging advantage held by natural guppy shoals. !The study of familiarity and its role in adaptive shoal choice is recent, though it does show promise. Research has provided evidence to support the idea that shoaling sh prefer the company of familiar individuals (Morrell, et al., 2007), and may even lead to improved immediate performance during behavioral tasks (Morrell, et al., 2008). This aspect of adaptive shoal choice likely occurs most often in more social, shoaling species like the guppy. These species tend to reside in relatively isolated ecosystems, leading to pockets of smaller populations peppered throughout the environment, which run a high likelihood of encountering one another and intermingling. Pelagic schooling sh, such as many species of mackerel (family Scombridae ), spend long periods of time with the same school, migrating across the world"s oceans and seas. These species have a high degree of school member familiarity, and in the event of school transference, familiarity with any members of the new school would be highly improbable. A lack of familiarity does not always mean an individual won"t join a group, particularly if the group otherwise suits the preferences of the individual. Another preference sh share is one of health; groups that appear more healthy tend to be chosen over groups that do not. v. Choosing the Right School: Gauging Fitness of Conspecics !A healthy school is an effective school. Unhealthy individuals are less likely to be capable of performing tasks vital to survival, such as predator avoidance or Just Keep Swimming 98

PAGE 104

proper foraging. An inability to perform these tasks poses a serious detriment to tness. Many ailments present themselves visually, such as certain parasites and many diseases (Ward, et al., 2002; Barber, 2003). Some researchers have suggested that animal grouping may serve to dilute the risk of parasitization as an extension of the dilution effect (Mooring and Hart, 1992). This does appear to be the case in terrestrial animal aggregations, but in the cases of many sh schools this dilution of risk appears to be minimal. Schools that were found to harbor parasites were often found to harbor them in a majority of their members (Barber, 2003). It was believed that these ndings were because sh who noticed visible parasitization of conspecics would not join those infected schools. Apparent health is not limited to a lack of visible parasitism, however; one can gauge approximate well-being in an individual based on their activity levels. Pritchard, Lawrence, Butlin, and Krause (2001) studied shoal choice in the zebrash ( Danio rerio ) as it was affected by shoal size and activity. Previously, no study had investigated the effect of activity on shoal choice in sh. A simple choice experiment was carried out with a test sh in a center tank and stimulus shoals in separate, adjacent tanks. One of the adjacent tanks contained a reference shoal of 4 individuals contained in either 25 o C or 15 o C water (50% of trials for each temperature), as activity levels were discovered to be affected by water temperature. The other adjacent tank contained an alternative shoal of varying size (1-4 sh) and was kept at a constant 25 o C, as was the center tank. The test sh would be observed for spatial preference (association) for the reference shoal or the alternative shoal. Four of the treatments resulted in Just Keep Swimming 99

PAGE 105

signicant preferences (4 warm vs. 1 warm: preference for larger group; 4 warm vs. 2 warm: preference for larger group; 4 cold vs. 3 warm: preference for warms; 4 cold vs. 4 warm: preference for warms). Test sh tended to prefer to associate with more active shoals, particularly when the more active shoal was a larger shoal. When shoals were of comparable size, the more active shoal was preferred. These results clearly indicated that shoal preferences can be based on the overall activity level of a shoal. !Activity as a property of tness is not limited to vigor; the frequency of certain exhibited behaviors can provide insight into the tness of a conspecic. An individual of higher tness is more likely to engage in risk-taking behavior (such as predator inspection or leaving cover; Bleakley and Brodie, 2009), and it was this "boldness" that caught the attention of Harcourt, Sweetman, Johnstone and Manica (2009). In their study, Harcourt, et al. sought to determine the effect of boldness, if any, on the active shoal choices of three-spined sticklebacks ( Gasterosteus aculeatus ). This investigation would consist of a binary choice experiment between bold and shy sticklebacks, but those classications needed to be assigned to individuals, rst. !Boldness assessments were conducted in long tanks, with weeds for cover at one end and an upward slope at the other. At the top of the slope was a tile, blocked from line of sight to the weeds by a small opaque partition. A bloodworm was placed on the tile and single subjects were scored for the total time spent foraging during trials. In these trials, subjects were considered to be bold if they spent at least 40% of trials foraging, and shy if they spent less than 5% of the Just Keep Swimming 100

PAGE 106

time foraging. After boldness or shyness designations were made, the choice experiment began. A binary choice experiment placed single sh of either shy or bold classication into a central compartment between two side compartments containing groups of 4 shy or 4 bold individuals. Subjects were further divided into categories of hungry or not, based on whether or not they were fed prior to testing. !Subjects of either classication showed a preference to associate with the bold shoal, but there was no trend regarding which shoal was approached rst. Figure 14 illustrates the proportion of time spent near either shoal, based on personality type and hunger level of focal sticklebacks. Bold individuals who were recently fed spent less time with the bold shoal, whereas hungry bold sh spent much more time with the bold shoal (Figure 14). Shy sh showed the opposite proportions of time spent near the bold shoal. In terms of activity, bold individuals exhibited more movement around the tank, crossing the midline signicantly more often than shy individuals. Fish in the bold shoals showed a similar increased activity level over shy shoals. Unexpectedly, the stimulus shoals themselves showed increased activity levels when interacting with bold focal individuals (Figure 15). The sh in this study could clearly tell the difference between bold and shy individuals. Just Keep Swimming 101

PAGE 107

Figure 14. Mean proportion of time SE spent by hungry or fed sticklebacks with the bold shoal. P < 0.05. (From Harcourt, Sweetman, Johnstone and Manica, 2009) Figure 15. Mean activity SE of bold and shy shoals in trials with both bold and shy focal sticklebacks. (From Harcourt, Sweetman, Johnstone and Manica, 2009) !Harcourt, et al. found a preference in three-spined sticklebacks to interact with #bold" conspecics. This distinction of boldness was applied to individuals Just Keep Swimming 102

PAGE 108

who spent less time in cover and explored a holding tank with a hidden food source at the section of the tank farthest away from cover. Harcourt, et al. then suggested after recording the data above that focal sh preferred to associate with shoals of bold conspecics. The problem with this assertion was that sh did not watch each other perform the original boldness task, and therefore could not make judgements based on any exhibited boldness. Harcourt, et al. did, however, note that activity levels of bold focal sh and bold shoals were higher than those of shy individuals. Based on the ndings of Pritchard, et al. (2001), who found that zebrash exhibited a clear preference for more active conspecics, it is more plausible that the sh in this study made their judgements based on activity levels. However, it is still worth noting that the more active sh observed by Harcourt, et al. demonstrated an increased tendency to engage in risky behavior, in this case venturing out of cover alone to explore a new space. !As evidenced by the briefness of this section, the stock of research concerning shes" preferences for visibly t shoal mates is relatively lacking. Pritchard, et al. (2001) directly demonstrated the preferences of zebrash for individuals that exhibit more vigor. Harcourt, et al. (2009) posited similar ndings, but called this increase in activity "boldness." Research concerning the schooling behaviors of sh aficted by parasites has focused primarily on actions and placement of individuals within an established school (Ward, et al., 2002; Barber, 2003). These studies suggest that levels of parasitization within schools indicate active decisions by unafliated sh to not join those aficted groups, based on a tendency for schools to exhibit either high parasitization or virtually none. In order Just Keep Swimming 103

PAGE 109

to conrm these suspicions, however, further studies should address whether or not these adaptive shoaling decisions actually occur. Given the many factors that contribute to an individual"s choice of school, evidence to support that hypothesis would not be surprising. There is still one more major inuencing factor when considering why sh choose to join the schools they join: the ever important sexfactor. vi. Choosing the Right School: Sex-biased Schooling !Depending on the species in question, the sex-ratio of a school can have a signicant impact on whether or not an individual will join that school. Species located in areas with less overall predation can devote more of their time to seeking out mating opportunities. Species like the common guppy ( Poecilia reticulata ) are classic examples; males guppies are often brightly colored with unique patterns, and spend a signicant portion of their time seeking new mates. Females, however, produce broods at approximately monthly intervals and can retain sperm to ensure fertilization at a later date, in case she was unable to mate for a prolonged period (Grifths and Magurran, 1998). These sexual differences between genders is believed to be the source of gender-biased schooling in these species. Lindstršm and Ranta (1993) investigated the social preferences of male guppies ( Poecilia reticulata ) based on shoal size and sex of shoalmates. As guppies have a high tendency to form shoals and specically male guppies spend a great deal of time following females and displaying for the purposes of mating, Lindstršm and Ranta hypothesized that a male"s shoal choice will heavily Just Keep Swimming 104

PAGE 110

depend on whether or not there is a high female-to-male ratio for mating purposes, and a large overall shoal size for predation protection. Thus, the experiment tested the effect of group size versus sex ratio on shoal preferences of male guppies. Focal males were presented with a traditional binary choice experimental setup. Three choice situations were used: one female versus one male, two females versus two males, and three females versus three males. As was expected, males preferred to associate with the female groups. As group size increased, focal males still showed a preference for female groups, but the margin between shoal choice was narrowed. Nevertheless, these results supported the notion that male guppies, when presented with a choice, will signicantly prefer to swim with female-biased shoals. Lindstršm and Ranta note that wild guppy shoals display largely female-biased sex ratios, and concluded that larger groups likely incorporate more females in wild populations. Guppies are not the only species to display gender-biased schools, however. !Females in species such as the guppy may not associate primarily with other females solely for protection from predators. Dadda, Pilastro and Bisazza (2005) bring the issue of #sexual harassment" by male sh into the spotlight. Dadda, et al. described this problem as a detriment to vital behaviors, such as foraging, for females of this species due to the near-constant mating attempts by male mosquitosh. This harassment leads to a necessary increase in vigilance by females, which in turn leads to females converging into largely female shoals. !Three experiments were carried out in order to address three predictions of actions taken by females to reduce harassment. First, it was predicted that Just Keep Swimming 105

PAGE 111

sexually unreceptive females would aggregate into a tighter group when a sexually active male was present. To test this prediction, two size-matched females were placed in a circular test arena with an opaque central cylinder. After the females settled overnight, the opaque cylinder was lifted to reveal a transparent cylinder housing a male mosquitosh, and the distance between the females, as well as the angle between them originating at the center of the tank, were monitored for a 30-minute period. Trials consisted of eight of these observation periods, alternating between having the opaque partition raised or lowered. Twelve female pairs were tested this way, and a further six were tested with a female mosquitosh in the central cylinder. !The second experiment tested a prediction based on a previous study that, if harassed, females would approach other males to instigate male-male competition. This prediction was tested by aligning two adjacent tanks, one of which contained a single female and was partially divided by a partition of ne mesh, to allow for the female to swim to either side of the tank. Aligned with this partial partition was a full partition in the second tank, dividing that tank into two separate compartments. One compartment contained three males, and the other was empty. In the tank with the female was an opaque "start box" that contained a single male, which was released into the main chamber during the trial. The focal female"s position was monitored for 30 minutes before and 30 minutes after the release of the single male. !The third prediction was an extension of the second prediction, positing that females would not only approach other males when harassed, but would Just Keep Swimming 106

PAGE 112

approach larger males, if possible. The same setup was used from experiment 2, but one stimulus compartment contained one small male, and the other contained one large male. Females were again monitored for their positions for 30 minutes before and 30 minutes after the start box was opened. Ten replicates were conducted where a male was released from the start box, and ve replicates were conducted where the box contained no male. !During experiment 1, female pairs swam signicantly closer to one another when a male was visible in the center and the angle between them decreased in size in these cases (23.46 1.05 degrees vs. 41.78 2.64 degrees. Subjects" interpersonal distance tended to decrease over time, as well. When a female was housed in the central cylinder schooling distance increased as well as the schooling angle. In experiment 2, females showed no signicant preference for either compartment prior to the release of the stimulus male. After the male was released, however, females preferred to be near the side of the tank with the three males (percentage of time spent near chamber in front of males: 62.19 0.09%). Females displayed no clear preference in trials where no male was released (48.02% 0.05%). Between replicates, where the three males were housed in alternating sides, there was no observed preference for a particular side of the tank. !As in experiment 2, female subjects in experiment 3 showed no initial preference for either side of the tank. After the male was released, females spent a majority of their time in front of the larger male (68.42 0.03%), while no preference was shown in trials where no male was released (50.67 0.04%). Just Keep Swimming 107

PAGE 113

These observations were most acute during the rst ve minutes after the release of the male (95.72 0.57% of time spent in front of larger male; controls: 53.33 0.10%). As seen in experiment 2, females did not show an independent preference for a particular side. !The results of this study consistently reect the predictions made by Dadda, et al. (2005) prior to experimentation. Signicant preferences were found for females to school closer when a sexually active male was present, approach a group of males when harassed by another, and in those cases to approach larger males, if at all possible. Female mosquitosh were observed to take immediate action in these cases. The females in this study were unreceptive females, as mosquito sh are not always receptive to mating. The measures illustrated in the study above are thus presumed to occur predominantly during periods where females are unreceptive. During more receptive periods, females may choose to reserve these actions for undesirable males, while allowing the advances of more desirable potential mates. Similar actions taken by female guppies would not be unexpected. After all, males spend a signicant amount of their time seeking mating opportunities, and females spend more time in female-biased shoals. Investigation of this kind would be easily applied to guppies, and many other particularly social species. One thing is certain: sexual asymmetry presents several unique problems to the mechanics of a functioning shoal that go beyond simple competition between shoal mates. Just because the survival strategy of a species depends on strength in numbers does not mean that everyone gets along. Just Keep Swimming 108

PAGE 114

Discussion !The research examined in this thesis constitutes a broad cross-section from the topics of behavioral shoaling and schooling that, at the time of this writing, remain subjects of academic scrutiny. A complete assessment of relevant research conducted in recent history could ll volumes, and the purpose of this review was to create an accessible reference point to the breadth of established subjects concerning the evolutionary basis, biological mechanisms and social factors of sh aggregation behaviors. That said, schooling and shoaling are behaviors of deceptive complexity. A. Cost-Benet Analysis First, schooling is primarily a means of predation-protection. Due to the complex group movements seen in many schools of sh, it was not difcult for researchers in the rst half of the last century to presume that this protection was a mutual benet achieved through cooperative swimming. That changed in 1971 when Hamilton published his model for the selsh herd, suggesting that animals joined groups in order to minimize their own personal risk of predation. This #selsh" behavior, when applied to all the individuals in a group, leads to clustering of individuals into groups and accounts for the constant shifting of individuals" positions seen in ocks of birds, schools of sh, and herds of terrestrial mammals. Hamilton"s theory gained such strong endorsements from the academic community that, 40 years later it is still accepted as the motivational basis for schooling by many researchers today. Just Keep Swimming 109

PAGE 115

!Schooling is secondarily used as a method for gleaning vital information from social partners. The group vigilance hypothesis states (Roberts, 1996) that the odds of a group noticing and actively protecting against an incoming danger are increased as the number of individuals in the group increases. This is because if one or more sh spot danger, other individuals will notice changes in those shes" behavior as indicative of the presence of a predator or some other danger and those sh will take action accordingly (Ward, et al., 2008). This group vigilance is not limited to signs of danger, though; sh can also pick up on social cues that suggest the presence of a food source or another resource. For this reason, schooling is also a helpful tool for foraging (Mittelbach, 1984; Day, et al., 2001). This acquisition of information via social observation is directly linked to another benet of schooling behavior: learning (Chapman, et al., 2008; Gallego, et al., 1995). In addition to the locations of predators or prey, sh learn novel behaviors from their school mates, which may help them through situations later in life. !No behavior is perfect, and schooling is no exception. The benets must outweigh the costs, or else the phenomenon of schooling would not have been naturally selected for over the course of history. Nonetheless, there are problems that go along with the protection that schooling offers. Schooling sometimes works in the opposite way it was intended, in terms of maximizing the benets described in this outline. These instances are often due to the oddity effect (Landeau and Terborgh, 1986; Theodorakis, 1989; Peuhkuri, 1997). If a predator attacks a large group, it primarily chooses an individual that stands out among Just Keep Swimming 110

PAGE 116

the crowd, whether that individual is out too far in the open or its appearance sets it apart from the rest of the group. Often, these encounters end with minimal casualties to the group, as the predator will quickly become satiated. Sometimes, the activity of the startled school will attract more predators, and it is in these cases that the entire school faces real danger. When a school is surrounded by hungry predators on all sides, it is not uncommon for the entire school to be wiped out (In the case of a #bait ball"). !Another major drawback of schooling is intragroup competition (Grand and Dill, 1999; Humphries, et al., 1999). Larger individuals are found to be better competitors, and smaller sh can suffer from malnourishment in cases where they are consistently outcompeted. This is a major reason why juveniles often do not swim with schools of adults. Competition poses a threat to tness, but not as much as the transmission of parasites. The particularly dangerous parasites are the ones that use schooling sh as secondary hosts, and inuence the behaviors of affected sh in order to become ingested by their nal hosts (Ward, et al., 2002; Barber, 2003). These parasites are often foodborne, and can infect signicant proportions of a school"s population during periods of foraging. B. The Senses !Sensory information is of critical importance to sh, and the senses of schooling sh have adapted over generations to suit the daily activities of those sh. Sight is often the foremost sense for sh near the surface of the water. Fish use visual cues to place the bulk of the school and keep track of its velocity and direction, as well as spotting predators or food (Breder, 1951). Vision is also used Just Keep Swimming 111

PAGE 117

to glean social information, such as the location of a food source or predator, based on the visual signs given off by conspecics. On a smaller spatial scale, however, the lateral line takes precedence. !The acuity and importance of the lateral line in schooling behavior is well documented ( &ur' i( -Blake and van Netten, 2006; Liao, 2006) and is an area of study in itself. Fish lateral lines trace the locations and velocity of their near neighbors, and provide information which prevents collisions during strenuous activity or low-light conditions. Since it detects even subtle water movements, the lateral line can also sense the pulses from distressed and dying sh. This information can be used by predators to locate an easy meal, or by wary sh to avoid the vicinity of the dying sh. !Olfaction also plays a role in the lives of schooling sh. Cyprinids have been found to secrete a chemical from their skin when they die, known as Schreckstoff. Research has shown this chemical to elicit a stress response in cyprinids, resulting in tighter school cohesion (von Frisch, 1938; Krause, 1993). It is unclear whether other families of sh possess a chemical such as Schreckstoff. Olfactory cues are also used by shoaling sh during adaptive shoal choice episodes. Whitecloud mountain minnows (Webster, et al., 2008) and guppies (Morrell, et al., 2007) have been found to prefer associating with individuals that carry the scent of high-protein food sources (i.e. bloodworm). This is believed to be due to the possibility of learning the location of the food source from the shoal in question. Indeed, learning where to nd high-nutrition foods is an important part the life of any foraging organism. Just Keep Swimming 112

PAGE 118

C. Adaptive Shoal Choice !Depending on the ecosystem, populations of schooling and shoaling sh may consist of several sub-populations. These subdivisions of the greater population often encounter each other, and these meetings allow individuals to transfer to another group. These transfers are not arbitrary, as sh prefer several dening traits in a new group. These inuential shoal choice factors have been the subjects of extensive study. !In terms of predation protection, sh prefer visual homogeny to maximize the confusion effect and larger numbers to maximize the dilution effect. The confusion effect necessitates that school members be of similar size and appearance (McRobert and Bradner, 1998; Rosenthal and Ryan, 2005). Larger groups increase the power of the dilution effect, leading to a decreased probability of any particular sh being captured (Pritchard, et al., 2001; Lindstršm and Ranta, 1993). Most species of social sh seek to increase this effect by preferentially joining larger groups. Central mudminnows, however, were found to have no particular preference (Jenkins and Miller, 2006). !Fish also prefer to associate with visibly healthy shoals. This is not surprising, as healthy individuals are likely to be more active and contribute more to group vigilance (Pritchard, et al., 2001). This heightened level of activity has been called "boldness" (Harcourt, et al., 2009). A lack of visible parasites is also a clear advantage for a school. Visible parasites can disrupt the confusion effect and lead to higher group predation (Barber, 2003), not to mention the possibility of transmitting parasites between group-members. Just Keep Swimming 113

PAGE 119

!Fish have also been observed to show a distinct preference for individuals that are familiar over strangers (Morrell, et al., 2007; Grifths and Magurran, 1997). Individuals with a history together need not spend important time familiarizing themselves with one another, and familiar guppies been observed to perform a novel foraging task with greater success than groups of unfamiliar guppies (Morrell, et al., 2008). Female guppies and mosquitosh preferentially shoal with other females of their species, due to a high frequency of sexual advancements by the males of their species (Grifths and Magurran, 1998; Lindstršm and Ranta, 1993). This suggests a high rate of school delity in these species, but frequency of shoal transfers by these females is not well documented. D. Topics of Interest and Suggestions for Future Research !Wherever possible, it was also the goal of this thesis to introduce new topics of interest, and to encourage further study of subjects that have received little attention. One such topic is the theory that inactive schooling could be a form a sleep mitigation for those species of sh that are in perpetual motion for the purposes of respiration (Kavanau, 2001). If it is indeed the case that these sh spend a portion of their time in a sleep-like state wherein their brains are afforded enough rest to reinforce and refresh their neurons, then further study of this behavior could provide considerable insight into the nature of memory in sh. With enough study, it might even be possible to determine the degree of information retention in the brains of sh. Just Keep Swimming 114

PAGE 120

!The other major topic of interest presented in this thesis is the a theory posited by Singer, et al. (2010). Singer, et al. suggested that LINE-1 retrotransposition could lead to neuronal diversity and somatic mosaicism in populations of neuronal cells. This somatic mosaicism would create unique distributions of neurons, leading to further uniqueness of the connections in the brains of every organism in which this retrotransposition occurs. Of further interest is idea that this retrotransposition can occur as a result of environmental stressors. During times of stress, when the heart rate increases and body temperature is raised slightly, this retrotransposition could occur and alter the genome of neurons. The effects of this altering of the genome are not clearly understood, but it can increase or decrease the expression of genes within a cell"s genome. Affecting the expression of these genes has a very real possibility of altering the expressed behaviors of that organism. Further research addressing the possible effects of induced neuronal somatic mosaicism is strongly recommended. !The effects of nearby predators on schooling behavior are well studied, but could benet from further investigation in the eld. Conditions in the laboratory are often not the same as in the wild, but conditions in the wild are seldom as simple as those in a laboratory. The balance between quantitative and qualitative studies seems to have been tipped in favor of the quantitative in recent years. Conversely, the effects of parasitism on shoal choice show a lack of diverse study. Particularly, continued study should address whether or not sh actively discriminate against joining shoals with visibly parasitized members. This review Just Keep Swimming 115

PAGE 121

would suggest that sh would show such discrimination, but data to support this is not present in the literature. !Morrell, et al. (2008) found that familiar shoals could better perform a novel foraging task than unfamiliar shoals. Conclusions for this apparent advantage were unclear, though it was suggested that familiar shoals more readily explored the test arena because they were already comfortable with one another. This suggestion is not unreasonable, but future studies should address this possibility directly. Finally, there is a severe lack of research concerning man"s inuence on the schooling patterns of sh. In light of recent catastrophes, particularly oil spills, millions of sh are dying and entire regions of the world"s oceans are becoming inhospitable. These changes may not be immediate, but migration patterns are likely to begin shifting to avoid contaminated areas of the oceans. !Schooling and shoaling are deeply complex group behaviors that have evolved and adapted over time. Environmental pressures act on every generation of these shes, augmenting the way they live and manipulate their environments. Comparatively, the investigation of schooling and shoaling are still young. Despite the relative newness of this area of study, there is a richness to the wealth of information gleaned from its studies. Many of the guiding principles for schooling behavior are easily superimposed onto the grouping behaviors of other organisms, which is indicative of the usefulness of such behaviors for any organism that chooses to adopt them. It doesn"t matter if you"re a social buttery or butterysh; for millions of species across this planet, being social is what counts. Just Keep Swimming 116

PAGE 122

Appendix A A Look at the Cell Cycle, Steps to a Theory of Somatic Variation in Neuronal Genomes Brain cells, or neurons, are a specialized cell type and form a network that regulates all bodily functions, responses to external and internal stimuli, and behaviors. Neurons are formed as the result of cell differentiation during early development. Development begins with the merging of two sex cells (gametes), sperm and ova, to form a new cell known as a zygote. This cell divides and duplicates, differentiating into different types of cells (skin cells, brain cells, etc.), eventually forming a complete organism. Cell division persists throughout the lifespan of the host organism for several purposes, namely hair growth and skin, blood, and internal organ renewal. This process of division and replication is known as the cell cycle, and is divided into many phases. !Healthy cells exist in one of three states: the quiescent or senescent state, interphase, and cell division. The quiescent state (abbreviated as G 0 ) is exhibited by newly formed cells and is considered a resting phase. Interphase is divided into three phases: Gap 1(G 1 ), Synthesis (S), and Gap 2 (G 2 ). The G 1 phase is marked by cell growth and ends with a G 1 checkpoint, during which the cell #decides" whether to divide, delay division, or enter a resting state, depending on external and internal conditions. Should the cell proceed to division, the S phase is initiated. During S phase, DNA replication occurs. Immediately before DNA replication occurs, a section of the DNA strand, known as an origin, responds to a specic #replication initiator protein" which then gathers other proteins and Just Keep Swimming 117

PAGE 123

together they #unzip" the double helix structure of the DNA. This unzipping is the breaking of hydrogen bonds along the central line of the DNA strand. The two resulting strands are known as #template strands." Following this, a fragment of DNA or RNA, known as a primer, is created and paired with each template strand, which then allows DNA replication to be carried out by a family of enzymes known as DNA polymerases. These polymerases travel along the DNA strand and match appropriate nucleotides (molecules that make up the structural units of DNA when connected) with the template strand, completing each strand to eventually duplicate the original unied DNA strand. This process thereby creates two copies of the original chromosome, called chromatids. These two chromatids are connected, and the connection point is called the centromere. !Once DNA replication is complete, G 2 phase begins. During G 2 phase, the cell continues to grow until the G 2 checkpoint is reached, wherein the cell #checks" several intracellular factors that ensure the cell is ready to begin the division state. The division state can take several paths, but most commonly proceeds into mitosis (M phase). During mitosis, cell growth stops and the cell is divided into two daughter cells. Mitosis comprises several phases of its own. After interphase (described above), an extra-nuclear cellular structure called a centrosome has replicated, creating a pair of centrosomes, which nucleate (form a nucleus around) a network of #microtubules." Specic #motor proteins" push the two centrosomes apart to either end of the cell along these extending microtubules. This process is known as prophase. In the next phase, prometaphase, the membrane of the cell nucleus, known as the nuclear Just Keep Swimming 118

PAGE 124

envelope, degrades and the array of microtubules from either centrosome invade the nuclear space. While this is occurring, the two chromosomes form structures at their centromeres (the place where each chromatid meets), one attached to each chromatid, called kinetochores. The microtubules then attach to these kinetochores, which activates a chemical process which pulls the chromatids toward either centrosome. This pulling begins metaphase, and aligns the centromeres of the chromosomes in an area equidistant from either centrosome. When this is accomplished, anaphase begins. In anaphase, the proteins binding the sister chromatids together at the centromere are cleaved, and the sister chromatids separate. As the microtubules attached to the kinetochores pull the chromatids toward the centrosomes, the other microtubules push the centrosomes further apart to the edges of the cell, stretching the cell out. Since the microtubules pull one chromatid from each chromosome to each side, the cell has successfully separated identical copies of the original genetic material to separate areas of the cell. This marks the end of anaphase. The nal phase of mitosis is telophase. During telophase, the microtubules continue to push the centrosomes apart, elongating the cell further. The sister chromatids on either side of the cell attach, and remnants of the original degraded nuclear envelope are used by the cell to construct new envelopes around either set of chromosomes. When the nuclear envelope is complete, the chromatids separate again and mitosis is completed. The nal step of cell division is often associated with mitosis, but is considered to be a separate process. It is known as cytokinesis. During cytokinesis, the cell is already stretched, and a furrow forms Just Keep Swimming 119

PAGE 125

around the area of the cell where kinetochore alignment occurred during metaphase. This furrow #pinches" the cell at the center and nally splits the cell into two daughter cells, each with their own nucleus and one centrosome. !The division of cells and expansion of mass is not enough to form an organism, however, as there are many different types of cells in a complete organism. The differentiation of unspecialized cells into other types of cells is necessary to form different tissues and bones. This differentiation is accomplished by a process known as cell signaling, a process by which cells communicate with one another. These signals can come from direct contact with another cell (juxtacrine signaling), short distances (paracrine signaling), and even over long relative inter-cellular distances (endocrine signaling). !Developmental cellular differentiation is instigated by a type of juxtacrine signaling known as #notch signaling." A notch receptor is a #transmembrane protein," meaning it resides in the outer cell membrane, and its structure protrudes a bit outside the cell and inside the cell. The notch receptor protein acts as a signal trigger, which can only be tripped by a specic molecule that interacts with only that receptor type, known as a ligand. Ligands are often also transmembrane proteins, which would explain why direct cell-to-cell contact is necessary for notch signaling. Upon triggering, the intracellular domain (the part of the notch protein residing inside the cell) is broken down and released into the cytoplasm, the thick medium inside of the cell. These remnants of the intracellular domain then enter the nucleus of the cell, where the DNA resides, and alter the gene expression of the cell. This alteration of gene expression causes changes Just Keep Swimming 120

PAGE 126

in the cell"s size, shape, responsiveness to signals, metabolic activity, and membrane potential (a measure of the difference in voltage between the interior and exterior of a cell). Gene expression is described as the process by which information from a gene, a #unit of heredity" encoded in a stretch of the DNA strand, is used to synthesize a #gene product," which can be either a protein or an RNA. RNA is a molecule similar to DNA, with a key structural difference being that it is single stranded and not double stranded. These products are used by the cell and the organism as a whole for numerous normal functions throughout the lifetime of the organism. !While the proteins produced by gene expression are used for any number of specic processes throughout the body, RNA serves more specic purposes, namely in the process of transcription. Put simply, transcription is an intracellular process whereby a copy of DNA is made with an equivalent RNA. Since RNA and DNA are both #nucleic acids," meaning acids that are formed with complementary pairs (base pairs) of nucleotides, the single-stranded RNA copy can serve as a template for new proteins with the original DNA strand as a blueprint. The process of transcription begins in a very specic section of the DNA, known as a promoter. The promoter serves as a binding site for an enzyme known as RNA polymerase. This binding can only occur in conjunction with the binding of specic proteins known as transcription factors. Upon successful binding to the DNA, the DNA strand is unzipped as RNA proceeds over the strand, very much like the process of DNA replication described earlier in this section. A key difference between DNA replication and transcription, however, is Just Keep Swimming 121

PAGE 127

that transcription only copies a specic section of the DNA strand. Since these sections may be repeated throughout the DNA strand (meaning a higher expression of that gene), transcription can occur at multiple places simultaneously on the DNA strand, and subsequent transcription events can occur repeatedly. When more transcription occurs, more RNA copies are produced, leading to higher production of the proteins synthesized by the RNA copies. The presence of higher concentrations of these proteins affect the functions of the cells and whole organism, leading to alteration of the phenotype in that organism. The DNA of a particular organism lends itself toward certain phenotypes, leading to distinction between species. But what if the expression of genes could be increased or decreased throughout the life of an organism, dynamically altering the phenotype of that organism? !Certain sections of DNA are capable of moving, or transposing, themselves to new positions in the genome. These sections are called transposons, sometimes called #jumping genes." Retrotransposons do not just move themselves, they copy and move themselves, which can increase both the expression of the genes in those sequences and the size of the genome in general. This #copy and paste" mechanism occurs rst through transcription of the retrotransposon, then from RNA back into DNA through a process called reverse transcription. Reverse transcription is initiated by the binding of RNA polymerase II and catalyzed by reverse transcriptase. Specic retrotransposons known as LINEs (#Long interspersed repetitive elements," or #Long interspersed nuclear elements") actually code for reverse transcriptase, allowing it to be produced Just Keep Swimming 122

PAGE 128

based on that coding. LINE elements are found in large quantities in the genomes of many animals; LINE-1 (L1) elements, for example, make up approximately 20% of the human genome (Singer, Mcconnell, Marchetto, Coufal, and Gage, 2010). With that in mind, could retrotransposition of these LINE-1 elements in brain cells alter the brain function of an organism? If so, this process may provide a deeper explanation to the inuence of indirect genetic effects (IGEs) on an organism"s phenotype, as seen in experiments on genetically isolated populations of Trinidadian guppies (Bleakley and Brodie, 2009). Just Keep Swimming 123

PAGE 129

Appendix B Thoughts on Collective Fitness !The idea that certain behaviors or actions are taken in order to benet an entire population or species is not typically accepted with reference to most organisms. Exceptions are those species that are believed to possess a "hive mentality," such as colonial insects, which are characterized by individuals that will readily sacrice themselves for the welfare of the group. The concept of hive mentality is a popular one in science ction, wherein it is often taken to the extreme. Creatures like the Xenomorphs in the Alien movie series, or the Borg in the television series Star Trek: The Next Generation are characterized by individuals possessing no autonomy whatsoever, whose actions are governed by a collective consciousness, act only in the interest of the entire population. The difference between animals with a hive mentality and these creatures of ction is the concept of shared consciousness, which allows them to perceive the risks posed to each individual in the population and thereby take action to minimize casualties and carry out a desired task with ruthless efciency. When considering the dangers faced by gregarious animals throughout their lives, the characteristics of these ctional creatures make for an enticing strategy. Whether or not organic life is capable of such a feat has yet to be seen, of course. Just Keep Swimming 124

PAGE 130

References Barber, I. 2003. Parasites and size-assortative schooling in three-spined !sticklebacks. Oikos 101: 331-337. Bleakley, B. H. and Brodie, E. D. III. 2009. Indirect genetic effects inuence !antipredator behavior in guppies: estimates of the coefcient of interaction psi and the inheritance of reciprocity. Evolution 63(7): 1796-1806. Breder, C. M., Jr. 1926. The locomotion of shes. New York, Ofce of the Society !New !York. Breder, C. M., Jr. 1951. Studies on the structure of the sh school. Bull. Am. Mus. !Nat. !Hist. 98: 7-28. Breder, C. M., Jr. 1954. Equations descriptive of sh schools and other animal !aggregations. Ecology 35(3): 361-370. Chapman, B. B., Ward, A. J. W., and Krause, J. 2008. Schooling and learning: !early social environment predicts social learning ability in the guppy, Poecilia reticulata. Anim. Behav. 76(3): 923-929. Dadda, M., Pilastro, A. and Bisazza, A. 2005. Male sexual harassment and !female schooling behaviour in the eastern mosquitosh. Animal Behaviour !70: 463-471. Day, R. L., MacDonald, T., Brown, C., Laland, K. N., and Reader, S. M. 2001. !Interaction between shoal size and conformity in guppy social foraging. Anim. Behav 62: 917-925. Dehn, M.M. 1990. Vigilance for predators: detection and dilution effects. Behavioral Ecology and Sociobiology. 26(5): 337-342. Just Keep Swimming 125

PAGE 131

Foster, W.A. and Treherne, J.E. 1981. Evidence for the dilution effect in the !selsh herd from sh predation on a marine insect. Nature 293(5832): !466-467. Gallego, A., Heath, M. R., and Fryer, R. J. 1995. Premature schooling of larval !herring in the presence of more advanced conspecics. Anim. Behav. 50(2): 333-341. Grand, T. C. and Dill, L. M. 1999. The effect of group size on the foraging !behavior of juvenile coho salmon: reduction of predation risk or increased !competition? Anim. Behav 58: 443-451. Grifths, S. W. and Magurran, A. E. 1998. Sex and schooling behaviour in the !Trinidadian guppy. Animal Behaviour 56: 689-693. Hamilton, W.D. 1971. Geometry for the selsh herd. Journal of Theoretical !Biology. 31: 295-311. Harcourt, J. L., Sweetman, G., Johnstone, R. A. and Manica, A. 2009. Personality !counts: the effect of boldness on shoal choice in three-spined sticklebacks. Animal Behaviour 77: 1501-1505. Hensor, E., Couzin, I. D., James, R. and Krause, J. 2005. Modelling !density-dependent sh shoal distributions in the laboratory and eld. OIKOS 110: 344-352. Heupel, M. R. and Simpfendorfer, C. A. 2005. Quantitative analysis of !aggregation behavior in juvenile blacktip sharks. Marine Biology 147: !1239-1249. Just Keep Swimming 126

PAGE 132

Hoare, D. J., Couzin, I. D., Godin, J.-G. J., and Krause, J. 2004. Context!dependent group size choice in sh. Animal Behaviour 67: 155-164. Humphries, S., Metcalfe, N. B., and Ruxton, G. D. 1999. The effect of group size !on relative competitive ability. Oikos 85(3): 481-486. Jenkins, J. R. and Miller, B. A. 2006. Shoaling behavior in the central mudminnow ( Umbra limi ). Am. Midland. Naturalist 158: 226-232. Kavanau, J. L. 2001. Brain-processing limitations and selective pressures for !sleep, sh schooling and avian ocking. Anim. Behav. 62: 1219-1224. Krause, J. 1993. The effect of #schreckstoff" on the shoaling behaviour of the !minnow: a test of Hamilton"s selsh herd theory. Anim. Behav. 45: !1019-1024. Kršther, S., Mogdans, J., and Bleckmann, H. 2002. Brainstem lateral line !responses to sinusoidal wave stimuli in still and running water. The Journal !of Experim. Biol. 205: 1471-1484. Landeau, L. & Terborgh, J. 1986. Oddity and the #confusion effect" in predation. Animal Behaviour 34: 13721380. Levin, L. E. and Vergara, E. 1987. Reversal learning in groups of the schooling !sh Aphyocharax erithrurus on an avoidance paddle. J. of Comparative !Psych. 10(4): !317-321. Liao, J. C. 2006. The role of the lateral line and vision on body kinematics and !hydrodynamic preference of rainbow trout in turbulent ow. The J. of !Experimental !Bio. 209: 4077-4090. Just Keep Swimming 127

PAGE 133

Lindstršm, K. and Ranta, E. 1993. Social preferences by male guppies, Poecilia !reticulata based on shoal size and sex. Anim. Behav. 46: 1029-1031. McRobert, S. P. and Bradner, J. 1998. The inuence of body coloration on !shoaling preferences in sh. Anim. Behav. 56(3): 611-615. Mittelbach, G. 1984. Group size and feeding rate in bluegills. Copeia !1984!(4): 998-1000. Mooring, M. S. and Hart, B. L. 1992. Animal grouping for protection from !parasites: !selsh herd and encounter-dilution effects. Behaviour 123(3-4): !173-193. Parrish, J. K., Viscido, S. V. and GrŸnbaum, D. 2002. Self-organized sh schools: !an examination of emergent properties. Biological Bulletin 202: 296-305. Partridge, B. L. and Pitcher, T.J. 1980. The sensory basis of sh !schools:relative roles of !lateral line and vision. J. Comp. Physiology. 135: !315-325. Peuhkuri, N. 1997. Size-assortative shoaling in sh: the effect of oddity on !foraging behaviour. Anim. Behav. 54(2): 271-278. Pritchard, V. L., Lawrence, J., Butlin, R. K., and Krause, J. 2001. Shoal choice in !zebrash, Danio rerio : the inuence of shoal size and activity. Anim. Behav. !62(6): 1085-1088. Richardson, J. M. L. 1994. Shoaling in white cloud minnows, Tanichthys albonubes : effects of predation risk and prey hunger. Anim. Behav. 48(3): !727-730. Just Keep Swimming 128

PAGE 134

Roberts, G. 1996. Why individual vigilance declines as group size increases. Animal Behaviour 51: 1077-1086. Roccanova, L.P. 1993. Evolution of bright coloration in schooling sh. Anim. !Behav. 45: 1034. Rosenthal, G. G. and Ryan, M. J. 2005. Assortative preference for stripes in !danios. Animal Behavior 70: 1063-1066. Ruzzante, D. E. and Doyle, R. W. 1993. Evolution of social behavior in a !resource-rich, structured environment: selection experiments with medaka ( Oryzias latipes ). Evolution 47(2): 456-470. Seghers, B. H. 1974. Schooling behavior in the guppy ( Poecilia reticulata ): an !evolutionary response to predation. Evolution 28(3): 486-489. Schutz, F. 1956. Vergleichende untersuchungen Ÿber die schreckreaktion bei !schen und deren verbeitung. Z. Vergl. Physiol. 38: 84-135. Shaw, E. 1978. Schooling shes. American Scientist 66: 166-175. Singer, T., McConnell, M. J., Marchetto, M. C. N., Coufal, N.G., and Gage, F. H. !2010. LINE-1 retrotransposons: mediators of somatic variation in neuronal !genomes? Trends in Neurosciences 33: 345-354. von Frisch, K. 1938. Zur Psychologie des Fisch-Schwarmes. Naturwissenschaften 26: !601-606. Ward, A. J. W., Hoare, D. J., Couzin, I. D., Broom, M., and Krause, J. 2002. The !effects of parasitism and body length on positioning within wild sh shoals. J. of Anim. Ecol. 71(1): 10-14. Just Keep Swimming 129

PAGE 135

Ward, A. J., Sumpter, D. J. T., Couzin, I. D., Hart, P. J. B. and Krause, J. 2008. !Quorum decision-making facilitates information transfer in sh shoals. PNAS 105(19): 6948-6953. Webster, M. M., Adams, E. L., and Laland, K. N. 2008. Diet-specic chemical !cues inuence association preferences and prey patch use in shoaling sh. Anim. Behav 76: 17-23. Wolf, N.G. 1985. Odd sh abandon mixed-species groups when threatened. Behavioral Ecology and Sociobiology 17(1): 47-52. Just Keep Swimming 130