探討Kryptolebias marmoratus個體的攻擊性與勇敢程度、探索傾向以及學習能力之間的相關性:前次打鬥經驗的影響以及荷爾蒙機制

全文

(1)國立臺灣師範大學生命科學系碩士論文. 探討 Kryptolebias marmoratus 個體的攻擊性與勇敢程度、探索傾向以及學習能力之間的相關性:前次打鬥經驗的影響以及荷爾蒙機制 The relationship between aggressiveness and boldness, exploratory tendency and learning performance in Kryptolebias marmoratus: the influence of recent contest experiences and the roles of hormones. 研究生：張靖 Ching Chang 指導教授：許鈺鸚博士 Yuying Hsu Ph. D.. 中華民國一百年八月.

(2)

(3) 誌謝本篇論文能夠順利完成，首先要感謝我的指導教授許鈺鸚博士，在實驗設計上以及遭遇困難時，老師皆提供了許多相當良好的意見與鼓勵。在這四年的碩士研究生涯中，老師平日鼓勵我們多閱讀、多思考，紮實的訓練以及密集的討論，使我獲益良多。還要感謝 Alan 在英文上的潤飾及修改，使得本篇能夠有條理且清楚呈現實驗的設計背景以及分享實驗最終發現的結果。此外我要感謝參與口試的口試委員，包括台灣師範大學生命科學系李壽先博士以及東海大學生命科學系卓逸民博士，兩位對於論文提出許多寶貴的意見與指教。本研究能順利完成，要特別感謝 Dr. Ryan Earley、Stephanie、宗楷、依涵對於實驗中荷爾蒙定量分析的指導以及誠裕在分析荷爾蒙時的大力協助。由衷感謝和我一起進入實驗室的夥伴怡婷，有妳和我一起走過這段趕作業、拚實驗、一起修習學程的日子，我才有完成論文的這一天。我也要感謝實驗室的學長姐時懿、孟馨、光洲在實驗初期給予的寶貴意見；宗楷、依涵在實驗設計以及實行上給予的重要協助；實驗室的重要夥伴韋佑、韻茹、郁雲、家怡、誠裕和明義，還有時常給予協助的學弟妹儆寰、昱儒、暐霖、韋稔，有你們一起在實驗室共同奮鬥，讓我的研究生生涯中充滿歡笑以及許多美好的回憶。謝謝在我身旁一直支持我的致玠，每當我實驗受挫、心情沮喪時總以正向積極的態度鼓勵我。待的時間太久，所要感謝的人太多。我只知道，若沒有你們的陪伴，不會有這篇論文的產生。最後要感謝我的父母以及家人，在背後默默的支持我，不因修業時間逐年增加而給我壓力，這份信任使我得以無後顧之憂地專心投入於我的論文研究。.

(4) 摘要具有攻擊性的個體通常較為勇敢，並傾向主動探索陌生的環境，同時有比較好的學習表現。這種不同個體間呈現出相似之行為間的相關性被稱之為行為群（behavioral syndrome）。前人提出，這些行為之間的相關性可能來自於這些行為受到相似的生理機制的影響；例如，一種荷爾蒙同時調控兩種以上的行為，而使得這些行為呈現彼此相關的現象。本篇測量 Kryptolebias marmortus 的攻擊性與勇敢程度、探索傾向以及學習表現之間的相關性。同時藉由強制給予一次落敗／獲勝經驗改變個體的攻擊性，來探討這些行為與攻擊性之間的相關性是否會受到前次打鬥經驗所影響而改變。本篇研究藉由測量個體的各個行為和睪固酮以及皮質醇濃度之間的相關性，來探討這些行為的相關性是否可能源自於它們皆受到此二荷爾蒙的調控而引起。實驗結果發現，個體的攻擊性與勇敢程度成正相關，而攻擊性和探索傾向以及學習表現之間則沒有顯著的相關性。實驗同時發現，個體的攻擊性與勇敢程度之間的相關性並不會受到前次打鬥經驗所影響。除此之外，個體的攻擊性以及勇敢程度皆與睪固酮基礎值成正相關。這些結果顯示攻擊性－勇敢程度行為群穩定存在於 K. marmortus 個體中，而此二行為的相關性可能源自於它們皆受到睪固酮基礎值所調控。. 關鍵字：行為群、輸贏效應、睪固酮、皮質醇.

(5) Abstract Aggressive individuals are usually also bold, actively explore their environment and perform better in learning tasks. A suite of correlated behaviors reflecting between-individual consistency in behavior across multiple situations is termed a ‘behavioral syndrome’. Researchers have suggested that a possible cause for the correlation between different behaviors might lie in the neuroendocrine system. In this study, I examined the relationship between aggressiveness and each of boldness, exploratory tendency and learning performance in a mangrove killifish, Kryptolebias marmoratus. I tested the stability of the relationships by altering individuals’ aggressiveness with a recent winning or losing experience. I also explored the possibility that the relationships arise because these behaviors are modulated by similar physiological mechanisms (i.e. levels of cortisol and testosterone). The results showed that aggressiveness was positively correlated with boldness but not with exploratory tendency or learning performance. Recent contest experience influenced individuals’ aggressiveness, tendency to explore and learning performance but not their boldness. Contest experience did not, however, influence the relationship between aggressiveness and boldness, both of which were positively correlated with baseline testosterone levels. The results suggest a stable aggressiveness-boldness syndrome in K. marmoratus and that baseline testosterone level might be involved in mediating this relationship.. Key words: behavioral syndrome, winner/loser effects, androgen, glucocorticoid.

(6) Contents Introduction………………………………………..…………………..01 1.1 Animal contest and aggression………………………………………...…..01 1.2 Aggression and correlated behaviors……………………………………...03 1.3 Objectives……………………………………………………………….…08. Materials and methods………………………………………………...10 2.1 Study organism………………………………………………………..…...10 2.2 Experimental design…………………………………………………….....11 2.3 Experimental procedures………………………………………………......12 2.4 Data analysis……………………………………………………………....18. Results………………………………………………………………….20 3.1. Rank correlation between aggressiveness and boldness, exploratory tendency and learning performance for individuals with different contest experiences………………………………………………………………...20 3.2 Aggression (latency to gill display): the effect of recent contest experience and its relationship with testosterone and cortisol levels………….….…...21 3.3 Boldness, exploratory tendency and learning performance: their relationship with aggressiveness and levels of testosterone and cortisol and how they are affected by contest experience………………………………………….....21 3.4 Effect of contest experience on the levels of testosterone and cortisol……23. Discussion………………………………………………………………25 4.1 Relationships between aggressiveness and boldness, tendency to explore and learning performance…………………………………………………25 4.2 The influence of recent contest experience on aggressiveness and boldness, the tendency to explore and learning performance, and on the relationships between them……………………………………………………………...26 4.3 Relationships between hormones and behavioral traits…………………...29 4.4 Conclusions………………………………………………………………..31. References………………………………………………………………32 Tables and figures……………………………………………………...38 Appendix……………………………………………………………….53.

(7) Introduction. 1.1 Animal contest and aggression Animals often compete for limited resources, such as mates (Dickinson 1992), food (Chapman & Kramer 1996), breeding sites (Eckert & Weatherhead 1987) or shelter (Fero & Moore 2008). Fighting can be costly in terms of expenditure of time and energy (Neat et al. 1998) and risk of both physical injury (Briffa & Sneddon 2007) and predation (Brick 1999). It would therefore be advantageous for individuals to evaluate the potential costs and benefits associated with a contest and adjust their fighting behavior accordingly (Smith & Price 1973; Smith & Parker 1976). Empirical studies of animal contests show that individuals‟ aggressiveness does indeed vary with risks and the value of the contested resource. Aggression is usually defined as behavior intended to threaten or harm other individuals (Moyer 1968). Individuals often tend to behave more aggressively (e.g. persist longer or fight at higher intensity) when resources are more valuable. For example, male bowl and doily spiders (Frontinella pyramitela) persisted longer in a fight when the females competed for were bearing more unfertilized eggs (Austad 1983). In red-backed salamanders (Plethodon cinereus), individuals fed with high quality food spent more time showing a threat posture toward intruders than individuals fed with low quality food (Gabor & Jaeger 1995). In addition, individuals can place a different value on a resource according to their internal state: hungry hermit crabs (Pagurus bernhardus) performed cheliped extension more frequently and were more likely to compete for food than others (Laidre & Elwood 2008). Information about potential costs may affect individuals‟ aggressiveness. For example, involvement in a contest usually decreases contestants‟ vigilance (Neat et al. 1998). In this way, individuals tend to. 1.

(8) shift their fighting intensity from higher to lower when under predation risk; male cichlid fish (Nannacara anomala) used lower intensity fighting and took longer to initiate mouth wrestling when a model predator was present (Brick 1999). One contest cost is direct physical damage inflicted by contest opponents, where stronger contestants can probably inflict more damage on their opponents than weaker ones. Individuals of a mangrove killifish, Kryptolebias marmoratus, appear to assess their contest opponent‟s size (which correlates well with fighting ability) and retreat more quickly when fighting against a larger opponent (Hsu et al. 2008). In these ways animals seem to be able to assess the potential costs associated with a fight and seem to avoid behaviors that could incur excess costs.. 1.1.1 Winner and loser effect Recent contest experiences can influence individuals‟ fighting behavior in subsequent contests (Whitehouse 1997; Hsu & Wolf 1999; Hsu et al. 2006). Individuals with prior winning experiences tend to become more aggressive (e.g. more likely to initiate future contests and attacks) and have a higher probability of winning again (winner effect); individuals with prior losing experiences tend to become less aggressive (e.g. less likely to initiate future contest and attacks) and have a higher probability of losing again (loser effect). Winner and loser effects are hypothesized to result from the influence of winning and losing experiences on how individuals assess of their own fighting ability, thereby influencing their assessment of the potential costs of taking part in future contests, which in turn influences their future fighting strategy (Whitehouse 1997). Individuals with a winning/losing experience might perceive themselves to have a better/worse fighting ability and expect a lower/higher potential contest cost. In this way. 2.

(9) individuals could become more/less aggressive, because they are more/less willing to participate in future contests, which could in turn lead to the higher/lower probability of winning again.. 1.2 Aggression and correlated behaviors Numerous studies have discovered that individuals‟ aggressiveness is often correlated with their boldness (the tendency to take risk), exploratory behavior (the willingness to explore their environment) and learning performance. Although these behaviors are correlated, they may, however, each have a different impact on individuals‟ fitness. Smith and Blumstein‟s (2008) review combined information from previous studies to identify the fitness consequences of different aspects of personality across species. They suggested that there was a positive relationship between aggressiveness and reproductive success. There was also a positive relationship between boldness and reproductive success, although boldness incurred a survival cost, whereas exploration had a positive effect on survival. In general, proactive individuals that are usually more aggressive, bolder and have greater willingness to explore novel environments, may do well in competitive or mating contexts, but they may have lower fitness in certain contexts due to their “non-adaptive” bold personality (e.g. expose themselves to environments with predators, increasing their predation risk). Although less aggressive individuals may do poorly in competitive conditions, they may do well in a context where low aggressiveness is favored, for example, when they are giving parental care or where avoiding predators is key, but poorly in competitive situations (Sih et al. 2004). A suite of correlated behaviors reflecting between-individual consistency in behavior across multiple situations is termed a „behavioral syndrome‟ (Sih et al. 2004).. 3.

(10) The concept of a behavioral syndrome can explain behaviors that seem to be non-adaptive in isolation. For instance, individuals showing high activity when predators are present might be doing so because this „non-adaptive‟ behavior is positively correlated with aggressiveness (which enables these individuals to do well in competitive situations). It might also explain the variation in types of behavior; rather than all individuals behaving in the „best‟ way in response to every situation (Sih et al. 2004). Identifying behaviors that are correlated with aggression in an animal will thus enable us to have a better understanding of the costs and benefits associated with aggression in the animal.. 1.2.1 Correlation between aggressiveness and boldness Many studies have revealed a positive relationship between boldness and aggressiveness. For example, in three-spined stickleback (Gasterosteu aculeatus), individuals which were aggressive toward conspecifics were also bolder toward predators (Huntingford 1976). Male field crickets (Gryllus integer) who won the majority of fights with a weight-matched opponent also had a shorter latency to leave their shelters (Kortet & Hedrick 2007).. 1.2.2 Correlation between aggressiveness and exploratory tendency The relationship between the tendency to explore and aggressiveness is also often positive. Fast explorers which visited artificial trees quickly in a novel environment won more fights than slow explorers in great tit (Parus major) (Verbeek et al 1996). More aggressive male field crickets (Gryllus integer) which won the majority of fights with weight-matched opponents had a shorter latency to become active in a novel environment (Kortet & Hedrick 2007). In addition, less aggressive mice (Mus. 4.

(11) musculus domesticus), those with longer latency to attack their opponent, took longer to reach the goal box in an unfamiliar maze (Benus et al 1987).. 1.2.3 Correlation between aggressiveness and learning performance There was a positive relationship between aggressiveness and avoidance learning in guppies (Poecilia reticulate): the individuals that did well in an avoidance learning task spent more time displaying aggressive behavior (Zhuikov 1993). Similar patterns have also been observed in mice such that aggressive mice (which had a shorter latency to attack) were better active shock avoiders than non-aggressive ones (Benus et al. 1989).. 1.2.4 Possible physiological mechanisms underlying the relationships between aggression and the other behaviors One possible cause for different behaviors to correlate with each other is that they may be governed by the same physiological mechanisms (Sih et al. 2004). Sex hormones (androgens) and stress related hormones (glucocorticoids) are good candidates for simultaneously influencing aggression and correlated behaviors. These two hormones have taken the center stage of the possible mechanisms because they play an important role in mediating the aggressive behaviors (Summers et al. 2005; Soma 2006).. Androgens Androgens are steroid hormones which mediate the development of male sex organs and male secondary sex characteristics in vertebrates. Contest behavior can directly affect gonadotropin-releasing hormone neurons (GnRH neurons) to activate. 5.

(12) the hypothalamic-pituitary-gonad (HPG axis) and stimulate the secretion of sex hormones (Francis et al. 1993). Testosterone is the primary androgen in reptiles, birds and mammals, while 11-ketotestosterone is the primary androgen in fish (Borg 1994; Vermeulen et al. 1994). Androgen levels are frequently positively correlated with aggressiveness. For example, castrated males behaved less aggressively (e.g. attacked or bit less frequently) than the control group, but there were no significant differences between the control group and castrated male mice which had been treated with testosterone supplement (Leshner and Moyer 1975). Spotted antbirds (Hylophylax naevioides) treated with testosterone outside the breeding season had a shorter latency to display than the control group (Hau et al. 2000). In a study of great tit (Parus major), males treated with testosterone sung more aggressive calls than control males (Duyse et al. 2002). In research on California mice (Peromyscus californicus), individuals treated with testosterone had a shorter latency to attack their opponents the following day than the control group (Trainor et al. 2004). Androgen levels are also shown to influence exploratory tendency and learning performance. Immature white rats (Rattus norvegicus) given an infusion of testosterone displayed an increase in horizontal and vertical locomotor activity in an open field test, suggesting a positive relationship between testosterone levels and exploratory tendency (Lambadjieva 1999). Testosterone supplement enhanced the performance in a water maze of castrated short-day mice (exposed to a 16:8 night:day cycle) compared with other short-day mice (Pyter et al. 2006). These studies show that the correlation between hormone levels and behaviors is a general phenomenon, and suggest that the relationships between boldness, exploratory tendency, learning ability and aggressiveness might arise because all these behaviors are influenced by. 6.

(13) the same physiological mechanisms (androgens or glucocorticoids).. Glucocorticoids In vertebrates, the stress response is mainly mediated by neuroendocrine system. Contest interactions may activate the hypothalamic-pituitary-adrenal axis (HPA axis) or in fish the hypothalamic-pituitary-interrenal axis (HPI axis). This activates the adrenal cortex or interrenal gland to release glucocorticoids, which can inhibit the function of insulin, decrease the synthesis of glycogen by using glucose or increase blood glucose levels (Genuth 1993). In amphibians, reptiles, birds and rodents the primary glucocorticoid is corticosterone; in fish and primates it is cortisol (Bonga et al. 1995). The correlation between glucocorticoid levels and aggressiveness varies according to the duration of the treatment. For example, aggressive behavior in rainbow trout (Oncorhynchus mykiss) was inhibited by long-term, but not short-term, cortisol treatment (Ø verli et al. 2002). Male rats reared with a female and given corticosterone treatment 10 min before an encounter attacked the intruder with increased frequency (Haller et al. 1997). Rats that received corticosterone treatment 2 or 20 min before an encounter spent more time displaying offensive behaviors than the control group (Mikics et al. 2004). Glucocorticoid levels also correlate with exploratory behavior. In rats, it was found that the administration of corticosterone (short-term; 90 min before tests) could enhance rat exploratory tendency and reduce the instance of a freezing reaction. Repeated administration of corticosterone, however, (long-term; for 25 days) had the opposite effect, decreasing plasma corticosterone concentrations, inhibiting exploratory behavior and increasing freezing responses (Skórzewska et al. 2006).. 7.

(14) Because glucocorticoids are associated with aggression, exploratory tendency and learning performance, they might be one of the mechanisms mediating the relationships between these behaviors.. 1.3 Objectives If a behavioral syndrome exists, then the relationships between the behaviors are stable population traits: when one of an individual‟s behaviors changes because of some specific influence (e.g. fighting with other conspecifics), other correlated behaviors change too, maintaining the stable relationship between them (Smith & Blumstein 2008). I was therefore interested to test whether behaviors correlated with aggression would change when an individual‟s level of aggression was altered by a recent winning or losing experience. If the behaviors correlated with an individual‟s aggressiveness reacted inconsistently to contest experience, the correlation between aggressiveness and other behaviors might break down. I was also interested in testing whether androgens and glucocorticoids play important roles in the relationships between aggression and the correlated behaviors. Studies of the highly aggressive killifish, Kryptolebias marmoratus, have shown that individuals that received a recent winning/losing experience were more/less likely to initiate gill display and attack, and had higher/lower probability of winning a non-escalated contest (Lan 2010). Moreover, pre-fight cortisol/testosterone levels were negatively/positively correlated with aggressive contest behavior; winners with higher pre-fight testosterone levels and lower cortisol levels attacked losers with higher frequency than others (Earley and Hsu 2008). These physiological and behavioral properties of the fish make it ideally suited to my study aims.. 8.

(15) The aims of my study were to test: (1) whether aggression in K. marmoratus correlates with boldness, exploratory tendency and learning performance; (2) whether winning or losing contest experiences influence the relationship between these behaviors and (3) whether cortisol and testosterone levels play important roles in mediating the relationship between these behaviors. If the relationships between aggression and the correlated behaviors are stable phenomena in K. marmoratus, I expect contest experience to have a similar effect on aggression, boldness, exploratory tendency and learning performance so that relationship between them is preserved in individuals with different contest experiences. If testosterone and cortisol levels play important roles in mediating these relationships, then they should have similar relationships with all of the correlated behaviors.. 9.

(16) Materials and methods. 2.1 Study organism The mangrove killifish Kryptolebias marmoratus is an internally self-fertilizing hermaphroditic fish (Harrington 1961; Taylor et al. 2001; Tatarenkov 2009). It is capable of producing fertilized eggs all year round and does not have obvious oviposition cycles (Harrington 1963). Most populations of the fish exist in nature as isogenic, homozygous strains, although outcrossing heterozygous populations have been discovered in Twin Cays, Belize (Taylor et al. 2001). K. marmoratus has an epidermal capillary bed (Grizzle and Thiyagarajah 1987) which enables the fish to breathe cutaneously, be semi-terrestrial and travel between locations by flipping or slithering through wet pebbles and mud (Davis et al. 1990; Taylor 1990; Sayer 2005). Some emersion might be obligatory for the fish, as completely submerged traps often contained dead individuals (Davis et al. 1990). In their natural environment, the fish are often found to hide under a damp cover (e.g. logs, mangrove leaves) or live in land-crab burrows (Davis et al. 1990; Taylor 2008). This study used individuals of five strains of K. marmoratus from various geographical areas (DAN2K: Dangria, Belize, collected in 2000; HON9: Utila, Honduras, collected in 1997; RHL: San Salvador, Bahamas, collected in 2001; SLC8E: St. Lucie County, Florida, USA, collected in 1995; VOL: Daytona Beach, Florida, USA, collected in 1995) which were the F3 to F6 of the fish originally collected from the field by Dr. D. Scott Taylor, FL, USA. Fish were isolated within a week of hatching and kept alone in a 13 x 13 x 9 cm translucent polypropylene plastic maintenance container filled with 550 ml of approximately 25 ppt synthetic sea water (Instant OceanTM powder) and labelled with a unique code for individual. 10.

(17) identification. Four holes were drilled in the lid and one hole close to the upper edges of each container for aeration. Because the fish is capable of respiring through its skin (see above), no extra aeration was needed. Each container was, however, filled only half full with water so that the fish could flip out of the water, stick on the lid/wall of the container and be exposed to the air at will. Maintenance containers were cleaned and water replaced every two weeks. Fish were maintained at 25  2oC on a 14:10 h light: dark cycle, and fed newly hatched brine shrimp (Artemia) nauplii daily.. 2.2 Experimental design I examined the aggressiveness, boldness, exploratory tendency and learning ability of individuals with different contest experiences to investigate how the different contest experiences affected these behaviors and the correlations between them. At the start of the experiment, fish were given a pre-designated, randomlyassigned win (W), loss (L) or no fighting experience (N) followed by four behavioral tests to examine the individuals‟ aggressiveness, boldness, exploratory tendency and spatial learning performance. I systematically arranged for the fish to take the aggressiveness, boldness and exploratory tendency tests in different orders, so as to avoid any systematic sequential effects. A test sequence was assigned to each fish before the experiments began. Because the spatial learning test took much longer than the other tests, I carried it out after the other three tests had been completed. All the fish used in this study had been used in previous research more than a month previously. All treatments had equal number of fish that won and lost their last contests and the outcome of their last contest was subsequently used as a control factor in data analyses. All the behavioral tests were recorded by a SONY handycam. 11.

(18) (DCR-SR100).. 2.3 Experimental procedures Seven days before the experiment, I familiarized the fish with shelters (which were used in later parts of my experiment) by putting one into the fish‟s maintenance container. In the field the fish are often found hiding in a shelter (Davis 1990). Laboratory-hatched individuals placed in a container with a shelter often utilized the shelter within one day and stayed in it for most of the time thereafter. I therefore concluded that the shelter-hiding may be an innate tendency, and used this tendency as the basis of my study. Between 9 and 10am on the first day of the experiment, a water sample was collected from the focal individual for baseline hormone analyses. (See “2.3.1 Hormone collection and extraction” for a description.) Afterwards, the fish received its pre-assigned winning, losing or no-contest experience. (See “2.3.2 Experience training” for details.) After training the fish was replaced in its maintenance container and fed newly hatched brine shrimp. Between 9 and 10 am on the second day of the experiment, another water sample was collected from the fish for post-experience hormone analyses. The fish was placed back in its maintenance container for 15 minutes and fed a small amount (0.5 ml) of brine shrimp. The fish was then exposed to three tests (aggressiveness, boldness, exploratory tendency) consecutively in a pre-determined order. Each test lasted for 30 minutes, but because the fish was allowed to acclimate in the test tank for an hour before the test started, each test took an hour and a half to complete. When the three tests were completed, the fish was transferred to a T-maze for the spatial learning test. Procedures for conducting these tests are described below in sections. 12.

(19) (2.3.3 - 2.3.6). After completing the learning test, the fish was returned to its maintenance container.. 2.3.1 Hormone collection and extraction At 9 am on the first and the second days of the experiment, I removed the focal individual from its maintenance container for an hour to analyze its baseline and post-experience hormone levels, respectively, as mentioned above. The fish was placed in a 400 ml glass beaker filled with 200 ml 25 ppt clean synthetic sea water and housed inside an individual translucent plastic container, visually isolating it in order to reduce the risk of disturbance. The fish stayed in the beaker for an hour. At 10 am, after removing the fish from the beaker, hormones were extracted from the water sample using a C18 solid phase extraction column (Lichrolut RP-18, 500 mg, 3.0 ml; Merck) fitted to a 12-port manifold. Before use, the columns were primed by 2 consecutive washes with 2 ml methanol (MeOH) followed by 2 consecutive washes with 2 ml pure water. Tubing was then fastened to the top of each column and placed into the water sample collected from the focal fish. A vacuum pump was connected to the tubing and the water sample was extracted through the tubing into the column. After processing all the water from the sample, I used 2 consecutive 2 ml washes with pure water to purge the seawater salt from the column and then stored it in a freezer at -80ºc until future processing. Freeze storage of water samples and columns has been determined not to influence steroid concentrations (Ellis et al. 2004). Hormones were eluted from the columns into 6 ml vials by 2 consecutive 2 ml washes with EA (Ethyl Acetate). The 4 ml of eluted solvent was evaporated at 37ºc with nitrogen gas to facilitate efficient evaporation. The resulting hormone powder was then stored at -20ºc until being assayed.. 13.

(20) Both hormones (testosterone and cortisol) were assayed using Cayman Chemicals Inc. EIA kits and following the procedures recommended by the manufacturer. Plates were read at 405 nm on a BioMek microplate reader.. 2.3.2 Experience training On day one, using random-selection procedures, the fish were given a pre-designated randomly assigned experience (W, L or N). The relative standard length (the distance from the tip of the snout to the beginning of the tail fin) is an important factor in determining the likelihood that individuals of K. marmoratus will win a contest (Hsu et al. 2008). To ensure that the focal individuals received their pre-designated losing or wining experiences, I fought them against much larger/ smaller (SL difference >2 mm) standard winners/ losers, which were strong/ weak and had won/ lost several fights with similar-sized opponents. For experience training, a fish was placed in each of two symmetrical compartments (randomly selected) of a standard aquarium (12 × 8 × 20 cm3, containing water 16 cm deep and 2 cm of gravel) divided by an opaque partition, which was removed to start the training. Individuals received their pre-designated experience quickly due to the large size differences and were again separated by the partition soon after the contest was decided. Fish assigned to receive no (N) experience were treated exactly as above except with no opponent in the standard aquarium. After training, the fish were replaced in their maintenance containers and fed newly hatched brine shrimp.. 2.3.3 Aggressiveness test I used a mirror test to quantify the fishes‟ aggressiveness (Earley et al. 2000). The test was carried out in a 16  10  20 cm3 experimental aquarium (Fig. 1),. 14.

(21) containing water 12 cm deep with 2 cm of gravel. Each aquarium had a mirror on one of the narrow sides and, 10 cm away from the mirror, a black opaque removable partition. The focal individual was placed in the part of the aquarium behind the partition, separated from its mirror image, for an hour‟s acclimation. After the acclimation, I removed the partition and recorded the fish‟s behavior for 30 minutes. As my index of aggressiveness, I recorded the time elapsed until the fish first took part in gill display (erected its operculum during display), the most aggressive fish having the shortest latency to gill display during the mirror test. I used this index for aggressiveness because gill display was usually the first aggressive behavior displayed and was therefore probably less influenced by the mirror image than other agonistic behaviors such as the latency to the attack or attack frequency. The mirror test was terminated after 30 minutes even if the focal fish did not display its gills towards its mirror image. After the mirror test, I moved the fish to the next experiment aquarium, to acclimatize for an hour.. 2.3.4 Boldness test Boldness is defined as the tendency to take risk (Wilson 1994). Following Brown et al. (2005), I used the latency to emerge from a small dark plastic shelter as my index of boldness, bolder individuals having a shorter latency to emerge from the shelter. The testing aquarium (Fig. 2) was the same size as that used for the mirror test, containing water 12 cm deep with 2 cm of gravel. The aquarium was fitted with a (6  6  8 cm3) black plastic shelter with a removable sliding door, through which the fish could freely swim when open. I placed the fish to be tested in the black plastic shelter for an hour‟s acclimation, pulled up the door at the beginning of the boldness test and recorded the time elapsed until the fish first emerged. I terminated the test after 30. 15.

(22) minutes even if the fish had not come out.. 2.3.5 Test of tendency to explore I used the time elapsed before the fish started exploring a new environment (with a shelter) as my index to quantify the tendency to explore (Gerlai et al. 1990; Dingemanse and Goede 2004). Fish with a greater tendency to explore novel environments should arrive at the shelter sooner. The experimental aquarium (Fig. 3) was 34  12  26 cm3, containing water 16 cm deep with 2 cm of gravel. I put a new shelter at the end of the wide side and a black opaque partition 6 cm from the opposite end. I marked a small dot 4 cm from the shelter on the surface of the aquarium as the „well explored region‟. The focal individual was placed in the smaller compartment, behind the partition, for an hour to acclimatize to the new experimental settings. At the start of the experiment, I pulled the partition up to let the focal fish freely explore the surroundings and recorded the time elapsed until the fish went into the „well explored region‟ (< 4 cm distance from the shelter). I terminated the experiment after 30 minutes even if the fish did not get close to the shelter.. 2.3.6 Test of spatial learning performance I used a T-maze test to test the fishes‟ spatial learning performance. Fish which found their way into the target end of the T-maze during the test were deemed (on average) to have better spatial learning performance than those that did not. The T-maze (Fig. 4) was a slightly modified version of that used in a previous study by Darland and Dowling (2001), and consisted of a cross piece (70 cm long), with a down piece (35 cm long) joining the mid-point of the cross piece. Both sections were 10 cm wide, 12 cm deep and contained water 5 cm deep. At each end of the. 16.

(23) cross-piece, there was a 22.5  22.5  22.5 cm3 reservoir, with 1 cm of gravel, 3 shelters and water 13 cm deep. As the fish tend to stay in shelter when it is available (see “experimental procedures” above) the reservoir area with its gravel and shelters should be desirable to them. The reservoirs and the shelters inside them were arranged symmetrically, to avoid any possible visual clues to the fish. There was an opaque removable partition at the entrance to each reservoir, controlling whether the fish could see and get into the reservoir area or not and a further removable partition 10 cm away from the end of the down piece of the T-maze (start point). I placed the focal fish at the end of the down piece of the maze, behind the removable partition, to acclimatize for an hour before the test. Before the test, one of the two reservoirs was chosen at random as the target. The fish was then given two training sessions (I and II), in which to learn to go into the target reservoir, and a test to see whether it had learnt from its training, all as described below. a) Training phase I: I pulled up the partition of the pre-assigned reservoir before the test so that the fish could enter it, but not the reservoir opposite. After acclimatization for an hour, I pulled up the partition of the start point so that the focal fish could freely explore the t-maze. After the fish entered the target reservoir, I closed the door (removable partition) so that the fish was restricted to the shelter-rich environment. Five minutes later, I terminated the first training period. If, after 30 minutes, the fish had not moved out of the down piece of the T-maze, I forced it to move to the junction of the T-maze by using a white opaque partition, leaving it to make its own turning decision there. If, after a further 30 minutes, the fish still had not arrived at the target reservoir, I forced it in into the reservoir using an opaque partition as above. After the first training period I placed the fish back at the start point, with. 17.

(24) the start partition closed, to acclimatize for 10 minutes in preparation for the next round of training. b) Training phase II: The procedures in training phase II were the same as in training phase I, except that, if after 30 minutes the fish did not arrive at the target reservoir, even after being forced to the junction of the T-maze if necessary, I terminated the training and determined that the fish had failed the T-maze training. Fish that did enter the reservoir were placed at the start point, with the start partition closed, to acclimatize for 10 minutes in preparation for testing phase. c) Testing phase: I pulled up the partitions at the entrances to both the reservoirs, allowing the fish to choose for itself whether to enter the correct or the incorrect reservoir. At the start of the test phase, I pulled up the partition at the start point allowing the fish to swim freely in the T-maze. If the fish found its own way to the correct reservoir it was deemed to have passed the T-maze test. If the fish did not move out of the start point, I forced it out of using an opaque partition as above. If the fish moved out of the start point but failed to get into the target reservoir in 30 minutes, I defined that the fish had failed the T-maze test.. 2.4 Data analysis A total of 180 individuals (60 each for the winning, losing and no contest experience treatments) were tested for this study. I used Spearman rank correlations to measure the relationship between aggressiveness and boldness and exploratory behavior. A Wilcoxon test was used to measure the difference in aggressiveness between the individuals which had a better learning performance (entered the correct reservoir in testing phase) and those with a worse learning performance (entered the wrong reservoir in testing phase).. 18.

(25) To examine whether the contest experience had any effect on individuals‟ aggressiveness, a non-parametric survival analysis (Cox regression) was employed to analyze the relationship between the three experiences (W, N and L) and the latency to gill display – this because some (18/180) of the individuals did not manifest gill display in the aggressiveness test. Cox regression models were also used to analyze whether the contest experience affected the relationship between aggressiveness and boldness and exploratory behavior. A multiple logistic regression model was used to examine whether contest experience affected the relationship between aggressiveness and learning performance. Baseline and post-experience levels of testosterone and cortisol included in the regression models to examine the relationship between hormones and aggressiveness, boldness, tendency to explore and learning performance. Testosterone and cortisol levels and the latency to gill displays were log transformed to fit the distribution requirements. All regression models controlled for strain type, body size (standard length) and last contest outcome. JMP (v. 5.0.1 SAS Institute Inc., Cary, NC, U.S.A.), a commercial statistical package, was used for the statistical analyses in this study.. 19.

(26) Results. 3.1 Rank correlation between aggressiveness and boldness, exploratory tendency and learning performance for individuals with different contest experiences The latency to gill display and the latency to emerge from a shelter were positively correlated for individuals that received a winning, losing or no recent contest experience (P  0.061, Fig. 5), indicating a stable, positive correlation between aggressiveness and boldness in this fish. Although the latency to gill display also appeared to correlate positively with the latency to explore a novel object for individuals with different contest experiences, the trends never reached statistical significance (P  0.080, Fig. 6). In addition, there was no significant difference in aggression (latency to gill display) between the fish with on average better and those with on average worse learning performance (those that did or did not enter the target reservoir) for any of the contest experience treatments (P  0.331, Fig. 7. Five individuals that failed either training phase I or training phase II were excluded from this analysis). These results indicated that, in K. marmoratus, there is a positive correlation between aggressiveness and boldness, but did not show a significant relationship between aggressiveness and exploratory tendency or between aggressiveness and learning performance.. 20.

(27) 3.2 Aggressiveness (latency to gill display): the effect of recent contest experience and its relationship with testosterone and cortisol levels The Cox regression model indicated that contest experience had a significant effect on an individual‟s latency to gill display (P = 0.023, Table 1); a winning experience shortened while a losing experience prolonged the average time to display, although neither the effect of a winning (P = 0.058) nor of a losing experience (P = 0.386) was significant when compared individually with the control treatment of no recent contest experience. Strain type had a significant effect (P = 0.025) on this measurement; individuals of DAN2K strain took longer to display than individuals of HON9 strain (Fig. 8). Baseline and post-experience levels of testosterone were negatively (P < 0.001) and positively (P = 0.001) related with the latency to gill display, respectively, indicating that individuals with higher levels of baseline testosterone tended to display more quickly (be more aggressive), while individuals with higher post-experience testosterone levels tended to display more slowly (were less aggressive). Neither baseline nor post-experience levels of testosterone were significantly correlated with the latency to gill display (P  0.123).. 3.3 Boldness, exploratory tendency and learning performance: their relationship with aggressiveness and levels of testosterone and cortisol and how they are affected by contest experience. 21.

(28) 3.3.1 Boldness (latency to emerge from a shelter) Cox regression (Table 2) showed a positive relationship between the latency to emerge from a shelter and the latency to gill display (P = 0.002), again confirming a positive relationship between aggressiveness and boldness. Contest experience, however, did not have a significant effect on the latency to emerge from a shelter (P = 0.571). There were no significant interaction effects between contest experience and the latency to gill display on the latency to emerge from a shelter (P = 0.520), indicating that the relationship between aggressiveness and boldness was not influenced by contest experience. Baseline testosterone level negatively related with the latency to emerge from shelter (P = 0.008) showing that individuals with higher baseline testosterone levels tended to emerge from the shelter sooner (be bolder). Strain type had a significant effect (P = 0.011); individuals of RHL and VOL strain took longer to emerge from a shelter than individuals of HON9 strain (Fig. 9).. 3.3.2 Tendency to explore (latency to explore a novel object) Cox regression (Table 3) indicated that there was no significant relationship between the latency to gill display and the latency to explore a novel object (P = 0.133), confirming the lack of a significant relationship between aggressiveness and the tendency to explore. The tendency to explore was, however, significantly influenced by contest experience (P = 0.005): individuals with a losing experience approached the novel object more quickly (P = 0.022). There were no significant interaction effects between contest experience and the latency to gill display on the latency to explore (P = 0.444), indicating that the (lack of) relationship between aggressiveness and the tendency to explore was not influenced by contest experience. Post-experience testosterone levels negatively related with the latency to explore (P =. 22.

(29) 0.029), i.e. individuals with higher post-experience testosterone levels had a higher tendency to explore.. 3.3.3 Learning performance (probability of entering the target reservoir in the testing phase) Logistic regression (Table 4) showed no relationship between the latency to gill display and the probability of entering the target reservoir (P = 0.729). Recent contest experience, however, had a significant effect on this probability (P = 0.004); more specifically, individuals with a losing experience were significantly less likely to enter the target reservoir than individuals with no recent contest experience (P = 0.045). Individuals with winning experiences were more likely than those with no contest experience to enter the target reservoir, and although this difference did not reach significance level (P = 0.175), the contrast between the behavior of the winners and the losers contributed to the significance of contest experience overall. There were no significant interaction effects between contest experience and the latency to gill display on the probability (P = 0.8134), indicating that the (lack of) relationship between aggressiveness and learning ability was not influenced by contest experience. Neither baseline nor post-experience levels of testosterone or cortisol had a significant relationship with this probability (P  0.500). In addition, last contest outcome had a significant effect on the probability; individuals that lost a contest more than one month previously had a higher probability of entering the target reservoir (P = 0.044).. 3.4 Effect of contest experience on the levels of testosterone and cortisol Post-experience testosterone levels, although positively correlated with baseline testosterone levels (P < 0.001), were not affected by the contest experience treatments. 23.

(30) (P = 0.992) (Table 5). Similarly, post-experience cortisol levels were positively correlated with baseline cortisol levels (P < 0.001) but were also not influenced by contest experience treatments (P = 0.379) (Table 6).. 24.

(31) Discussion. My study investigated aggression-related behaviors in K. marmoratus and how recent contest experiences affect both these behaviors and their relationship with hormones. The fish appeared to exhibit a behavioral syndrome including aspects of aggressiveness and boldness. Individuals that behaved aggressively were also bolder than others, and this relationship between aggressiveness and boldness was not influenced by recent contest experience. In addition, both the fishes‟ aggressiveness and their boldness positively correlated with their baseline testosterone levels, which lends support to the possibility that these two behaviors are positively correlated because they are modulated by common physiological mechanisms. Neither the fishes‟ tendency to explore nor their learning performance was significantly correlated with aggressiveness, although as with aggressiveness, these two behaviors were influenced by recent contest experience. Individuals with a winning experience were more aggressive and had better learning performance but a lower tendency to explore than individuals with a losing experience. The fishes‟ tendency to explore was positively correlated with post-experience testosterone levels, whereas learning performance was not significantly related with levels of testosterone or cortisol.. 4.1 Relationships between aggressiveness and boldness, tendency to explore and learning performance There was a positive relationship between aggressiveness and boldness. Individuals that had a shorter latency to gill display emerged from the shelter more quickly. This result is consistent with other studies in different species, suggesting a. 25.

(32) relatively stable aggressiveness-boldness relationship across taxa (Kortet & Hedrick 2007; Moretz et al 2007; Pintor et al 2008). There was no significant relationship between aggressiveness and either the tendency to explore or learning performance. This result differs from previous studies which found that, in general, more aggressive individuals were quicker to explore their environment than others (Benus et al 1987; Verbeek et al 2006), whereas less aggressive individuals took longer to explore their environment (van Oortmersse et al 1985; Benus et al 1987). It is therefore possible that the relationship between aggressiveness and both exploratory tendency and learning performance might vary with the study animal. There might also be a confounding effect between shelter-hiding tendency and exploratory tendency. Being in a novel environment is stressful, which might cause the individuals to flee or freeze in response. In my study, I provided a shelter in the area to be explored. Timid fish might therefore have approached the shelter to hide in it, rather than bold fish approaching the shelter to explore it.. 4.2 The influence of recent contest experience on aggressiveness and boldness, the tendency to explore and learning performance, and on the relationships between them. 4.2.1 Effect of contest experience on the different behaviors Contest experience received one day previously significantly changed individuals‟ behavior in mirror tests (of aggressiveness). Individuals with a winning experience had a lower latency to gill display (were more aggressive) than those with. 26.

(33) a losing experience. This result is consistent with previous studies of K. marmoratus (Hsu & Wolf 1999, 2001; Lee 2009; Lu 2010; Huang et al 2010; Lan 2010), which found that prior winning and losing experiences increased and decreased, respectively, individuals‟ aggressiveness (e.g. the probability of initiating gill display, attack) and the probability of winning a non-escalated contest. In my study, contest experience received one day previously did not affect individuals‟ boldness. As well as the possibility that the effect of contest experience on boldness might vary with the study organism, it is possible that individuals with different personalities might react differently to contest experience. In a study of rainbow trout, the authors found that the boldness level of shy individuals increased no matter what contest experience they received, whereas bold individuals became shy after receiving a losing experience (Frost et al 2007). In my study, individuals were randomly assigned to receive a winning/losing/no experience regardless of their original personality, while the boldness levels of the individuals with different contest experience did not differ, suggesting that there were no significant experience effects on boldness rather than the dilution effect of different innate personalities. Contest experience received one day previously significantly changed individuals‟ tendency to explore. Individuals which received a losing experience approached the novel object more quickly. These results might have occurred because the object we provided for the individuals to explore was a shelter. Individuals which had received a losing experience one day previously might have been inclined to hide in the shelter quickly rather than to explore the environment slowly. Recent contest experience received one day previously had a significant effect on learning performance. Individuals which had received a losing experience were more likely to pick the wrong side of the reservoir than others, suggesting that a losing. 27.

(34) experience reduced learning performance. This result might arise because losing increased anxiety levels, which subsequently impeded learning performance. Rats which were housed with cats to induce anxiety had lower locomotor activity in open field tests and worse performance in water maze tasks than the control group (Park et al 2001). Therefore, there appears to be a general trend for stress to reduce learning performance.. 4.2.2 Effect of contest experience on the relationship between the behaviors The different types of experience (W, L & N) had no significant effect on the relationships between aggressiveness and boldness, tendency to explore or learning performance. Aggressiveness and boldness were, however, highly correlated for all three experience types, suggesting a stable aggressiveness-boldness syndrome in K. marmoratus. Recent studies have indicated that predation risk might be related to the relationship between aggressiveness and boldness in sticklebacks (Bell 2005; Bell and Sih 2007): a positive relationship existed only in populations that were under high predation risk, suggesting that stressful conditions (e.g. high predation risk) might promote a tradeoff and cause correlations between aggressiveness and boldness. For instance, when predation risk is high, bolder individuals that tend to wander away from shelter are probably more likely to suffer higher mortality. This disadvantage of being bolder may be compensated by being aggressive toward conspecifics. When predation risk is high, selection against individuals which are both bold and not aggressive may therefore cause a positive relationship between boldness and aggressiveness.. 28.

(35) Although contest experience had a significant effect on aggressiveness but not on boldness, the relationship between aggressiveness and boldness was not significantly affected by their contest experiences (W/N/L). This might be because the relationship between boldness and aggressiveness is stronger than the effect of contest experience on aggressiveness so that we could not detect a significant effect of contest experience on the relationship between them.. 4.3 Relationships between hormones and behavioral traits. 4.3.1 Effect of contest experience on individuals’ post-experience levels of testosterone and cortisol Although winning and losing experiences caused the fish to become respectively more and less aggressive, they did not affect the fishes‟ post-experience levels of testosterone or cortisol. Previous studies of K. marmoratus reached similar conclusions on the effect of contest experiences on hormone levels (Lee 2009; Lu 2010). These results suggest that changes in the levels of these hormones are not the primary mechanisms for winner and loser effects in K. marmoratus. The physiological mechanisms that mediate winner and loser effects in the fish are yet to be uncovered. Since losing experience did cause a deterioration in learning ability (see 4.2.1 above) as might be expected for animals under stress, but did not cause an increase in post-experience cortisol levels, it may be that cortisol does not mediate the stress response in this fish.. 29.

(36) 4.3.2 Relationships between hormones and each of the behaviors There was a significant relationship between baseline testosterone levels and both aggressiveness and boldness. This result lends support to the possibility that the positive relationship between aggressiveness and boldness is caused by their being mediated by similar physiological mechanisms. My finding of a positive relationship between aggressiveness and baseline testosterone levels is similar to that of a previous study of K. marmoratus. Li (2010) found a positive relationship between baseline testosterone and the fish‟s tendency to initiate aggressive acts and its persistence when fighting with a stronger opponent. I also discovered a negative relationship between aggressiveness and post-experience testosterone levels. However, two previous studies (Lee 2009, Lu 2010) of the fish showed that the aggressiveness (e.g. the likelihood of initiating gill displays, attacks and escalate contests) of individuals with a losing experience was positively correlated with post-experience testosterone and cortisol levels. One possible reason for the difference in findings between my and previous studies is different experimental designs. Lee (2009) and Lu (2010) provided their focal individuals with three losing contest experiences whereas the individuals in my study received only one contest experience. Three consecutive losing experiences might have caused a more significant relationship between post-experience hormone levels and contest behaviors than just one experience. I did not find any significant relationship between cortisol levels and aggressiveness in the fish. Earley and Hsu (2008), however, reported a negative relationship between cortisol levels and aggressiveness. Earley and Hsu (2008) measured the hormone levels of both participants in a contest whereas I only measured the hormone levels of a fish displaying against its own reflection in a mirror,. 30.

(37) which could have cause the difference in our findings about the relationship between aggressiveness and cortisol levels.. 4.4 Conclusions Overall, I found a stable aggressiveness-boldness relationship in K. marmoratus: more aggressive individuals were also bolder. And, the relationship between aggressiveness and boldness remained stable regardless of whether the fish received a wining, a losing or no contest experience. The fish‟s tendency to explore and its learning performance, however, were not significantly correlated with its aggressiveness. In addition, baseline testosterone levels positively correlated with individuals‟ aggressiveness and boldness. The fish‟s tendency to explore correlated with post-experience testosterone levels, whereas learning performance did not have significant relationship with testosterone or cortisol levels. These results suggest that baseline testosterone levels might play an important role in mediating the relationship between aggression and boldness in the fish.. 31.

(38) References Austad SN, 1983. A game theoretical interpretation of male combat in the bowl and doily spider (Frontinella pyramitela). Animal Behaviour 31:59-73. Bell AM, 2005. Differences between individuals and populations of three spined stickleback. Journal of Evolutionary Biology 18:464-473. Bell AM, Sih A, 2007. Exposure to predation generates personality in threespined sticklebacks (Gasterosteus aculeatus). Ecology Letters 10:828-834. Benus RF, Bohus B, Koolhaas JM, van Oortmerssen GA, 1989. Behavioural strategies of aggressive and non-aggressive male mice in active shock avoidance. Behavioural Processes 20:1-12. Bonga SEW, Balm PHM, Lamers AE, 1995. The involvement of ACTH and MSH in the stress-response in teleost fish. Netherlands Journal of Zoology 45:103-106. Borg B, 1994. Androgens in teleost fishes. Comparative Biochemistry and Physiology-Part C: Toxicology and Pharmacology 109:219-245. Brick O, 1999. A test of the sequential assessment game: the effect of increased cost of sampling. Behavioral Ecology 10:726-732. Briffa M, Sneddon LU, 2007. Physiological constraints on contest behaviour. Functional Ecology 21:627-637. Brown C, Jones F, Braithwaite V, 2005. In situ examination of boldness-shyness traits in the tropical poeciliid, Brachyraphis episcopi. Animal Behaviour 70:1003-1009. Chapman MR, Kramer DL, 1996. Guarded resources: The effect of intruder number on the tactics and success of defenders and intruders. Animal Behaviour 52:83-94. Darland T, Dowling JE, 2001. Behavioral screening for cocaine sensitivity in mutagenized zebrafish. Proceedings of the National Academy of Sciences of the United States of America 98:11691-11696. Davis WP, Taylor DS, Turner BJ, 1990. Field observations of the ecology and habits of mangrove rivulus (Rivulus marmoratus) in Belize and Florida (Teleostei: Cyprinodontiformes: Rivulidae). Ichthyological Exploration of Freshwaters 1:123-134.. 32.

(39) Dickinson JL, 1992. Scramble competition polygyny in the milkweed leaf beetle Combat, mobility, and the importance of being there. Behavioral Ecology 3:32-41. Dingemanse NJ, de Goede P, 2004. The relation between dominance and exploratory behavior is context-dependent in wild great tits. Behavioral Ecology 15:1023-1030. Earley RL, Hsu Y, 2008. Reciprocity between endocrine state and contest behavior in the killifish, Kryptolebias marmoratus. Hormones and Behavior 53:442-451. Earley RL, Hsu Y, Wolf LL, 2000. The use of standard aggression testing methods to predict combat behaviour and contest outcome in Rivulus marmoratus dyads (Teleostei : Cyprinodontidae). Ethology 106:743-761. Eckert CG, Weatherhead PJ, 1987. Owners, floaters and competitives asymmetries among territorial red-winged blackbirds. Animal Behaviour 35:1317-1323. Ellis T, James JD, Stewart C, Scott AP, 2004. A non-invasive stress assay based upon measurement of free cortisol released into the water by rainbow trout. Journal of Fish Biology 65:1233-1252. Fero K, Moore PA, 2008. Social spacing of crayfish in natural habitats: what role does dominance play? Behavioral Ecology and Sociobiology 62:1119-1125. Francis. RC,. Soma. K,. Fernald. RD,. 1993.. Social. regulation. of. the. brain-pituitary-gonadal axis. Proceedings of the National Academy of Sciences of the United States of America 90:7794-7798 Frost AJ, Winrow-Giffen A, Ashley PJ, Sneddon LU, 2007. Plasticity in animal personality traits: does prior experience alter the degree of boldness? Proceedings of the Royal Society B-Biological Sciences 274:333-339. Gabor CR, Jaeger RG, 1995. Resource quality affects the agonistic behavior of salamanders. Animal Behaviour 49:71-79. Genuth SM, 1993. The endocrine system. In: Physiology (Berne, R.M. & Levy M.N. ed.), pp. 813-1024. Mosby Year Book, St. Louis. Gerlai R, Crusio WE, Csanyi V, 1990. Inheritance of species-specific behaviors in the paradise fish (Macropodus opercularis) - a diallel study. Behavior Genetics 20:487-498. Grizzle JM, Thiyagarajah A, 1987. Skin histology of Rivulus ocellatus-marmoratus Apparent adaptation for aerial respiration. Copeia:237-240.. 33.

(40) Haller A, Albert I, Makara JB, 1997. Acute behavioural effects of corticosterone lack. specificity. show. marked. context-dependency.. Journal. of. Neuroendocrinology 9:515-518. Harrington. RW,. Jr.,. 1961.. Oviparous. hermaphroditic. fish. with. internal. self-fertilization. Science 134:1749-1750. Harrington RW, JR., 1963. Twenty-four-hour rhythms of internal self-fertilization and of oviposition by hermaphrodites of Rivulus marmoratus,. Physiological Zoology 36:325-341. Hau M, Wikelski M, Soma KK, Wingfield JC, 2000. Testosterone and year-round territorial aggression in a tropical bird. General and Comparative Endocrinology 117:20-33. Hsu Y, Earley RL, Wolf LL, 2006. Modulation of aggressive behaviour by fighting experience: mechanisms and contest outcomes. Biological Reviews 81:33-74. Hsu Y, Lee SP, Chen MH, Yang SY, Cheng KC, 2008. Switching assessment strategy during a contest: fighting in killifish Kryptolebias marmoratus. Animal Behaviour 75:1641-1649. Hsu Y, Wolf LL, 1999. The winner and loser effect: integrating multiple experiences. Animal Behaviour 57:903-910. Hsu Y, Wolf LL, 2001. The winner and loser effect: What fighting behaviours are influenced? Animal Behaviour 61:777-786. Huang SP, Yang SY, Hsu Y, 2010. Persistence of winner and loser effects depends on the behaviour measured. Ethology 177:171-180. Huntingford FA, 1976. The relationship between anti-predator behaviour and aggression among conspecifics in the three-spined stickleback, Gasterosteus Aculeatus. Animal Behaviour 24:245-260. Kortet R, Hedrick A, 2007. A behavioural syndrome in the field cricket Gryllus integer: intrasexual aggression is correlated with activity in a novel environment. Biological Journal of the Linnean Society 91:475-482. Laidre ME, Elwood RW, 2008. Motivation matters: cheliped extension displays in the hermit crab, Pagurus bernhardus, are honest signals of hunger. Animal Behaviour 75:2041-2047. Lambadjieva ND, 1999. Influence of testosterone on some behavioral reactions of male immature rats exposed to continual light. Methods and Findings in Experimental and Clinical Pharmacology 21:17-20. 34.

(41) Lan YT, 2010. Does contest experience from more than a month previously in Kryptolebias marmoratus influence winner/loser effect arising from recent contests? Master's thesis. Taipei, Taiwan: National Taiwan Normal University. Lee IH, 2009. The hormonal mechanisms of the loser effect in Kryptolebias marmoratus. Master's thesis. Taipei, Taiwan: National Taiwan Normal University. Leshner AI, Moyer JA, 1975. Androgens and agonistic behavior in mice: Relevance to aggression and irrelevance to avoidance-of-attack. Physiology & Behavior 15: 695-699. Li CY, 2010. Temporal changes in androgen, estrogen, cortisol and 5-HT1A receptor gene expression after wins and losses in Kryptolebias marmoratus. Master's thesis. Taipei, Taiwan: National Taiwan Normal University. Lu, CK, 2010. Winner effect and its relationship with testosterone and cortisol in Kryptolebias marmoratus. Master's thesis. Taipei, Taiwan: National Taiwan Normal University. Mikics E, Kruk MR, Haller J, 2004. Genomic and non-genomic effects of glucocorticoids. on. aggressive. behavior. in. male. rats.. Psychoneuroendocrinology 29:618-635. Moretz JA, Martins EP, Robison BD, 2007. Behavioral syndromes and the evolution of correlated behavior in zebrafish. Behavioral Ecology 18:556-562. Moyer KE, 1968. Kinds of aggression and their physiological basis. Neat FC, Taylor AC, Huntingford FA, 1998. Proximate costs of fighting in male cichlid fish: the role of injuries and energy metabolism. Animal Behaviour 55:875-882. Overli O, Kotzian S, Winberg S, 2002. Effects of cortisol on aggression and locomotor activity in rainbow trout. Hormones and Behavior 42:53-61. Park CR, Campbell AM, Diamond DM, 2001. Chronic psychosocial stress impairs learning and memory and increases sensitivity to yohimbine in adult rats. Biological Psychiatry 50:994-1004. Pintor LM, Sih A, Bauer ML, 2008. Differences in aggression, activity and boldness between native and introduced populations of an invasive crayfish. Oikos 117:1629-1636.. 35.

(42) Pyter LM, Trainor BC, Nelson RJ, 2006. Testosterone and photoperiod interact to affect spatial learning and memory in adult male white-footed mice (Peromyscus leucopus). European Journal of Neuroscience 23:3056-3062. Sayer MDJ, 2005. Adaptations of amphibious fish for surviving life out of water. Fish and Fisheries 6:186-211. Sih A, Bell A, Johnson JC, 2004. Behavioral syndromes: an ecological and evolutionary overview. Trends in Ecology & Evolution 19:372-378. Skorzewska A, Bidzinski A, Lehner M, Turzynska D, Wislowska-Stanek A, Sobolewska A, Szyndler J, Maciejak P, Taracha E, Plaznik A, 2006. The effects of acute and chronic administration of corticosterone on rat behavior in two models of fear responses, plasma corticosterone concentration, and c-Fos expression in the brain structures. Pharmacology Biochemistry and Behavior 85:522-534. Smith BR, Blumstein DT, 2008. Fitness consequences of personality: a meta-analysis. Behavioral Ecology 19:448-455. Smith JM, Parker GA, 1976. The logic of asymmetric contests. Animal Behaviour 24:159-175. Smith JM, Price GR, 1973. The logic of animal conflict. Nature 246:15-18. Soma KK, 2006. Testosterone and aggression: Berthold, birds and beyond. Journal of Neuroendocrinology 18:543-551. Summers CH, Watt MJ, Ling TL, Forster GL, Carpenter RE, Korzan WJ, Lukkes JL, Overli O, 2005. Glucocorticoid interaction with aggression in non-mammalian vertebrates: Reciprocal action. European Journal of Pharmacology 526:21-35. Tatarenkov A, Lima SMQ, Taylor DS, Avise JC, 2009. Long-term retention of self-fertilization in a fish clade. Proceedings of the National Academy of Sciences of the United States of America 106:14456-14459. Taylor DS, 1990. Adaptive specializations of the Cyprinodont fish Rivulus marmoratus. Florida Scientist, 53, 239-248. Taylor DS, Fisher MT, Turner BJ, 2001. Homozygosity and heterozygosity in three populations of Rivulus marmoratus. Environmental Biology of Fishes 61:455-459. Taylor DS, Turner BJ, Davis WP, Chapman BB, 2008. Natural history note - A novel terrestrial fish habitat inside emergent logs. American Naturalist 171:263-266.. 36.

(43) Trainor BC, Bird IM, Marler CA, 2004. Opposing hormonal mechanisms of aggression revealed through short-lived testosterone manipulations and multiple winning experiences. Hormones and Behavior 45:115-121. Van Duyse E, Pinxten R, Eens M, 2002. Effects of testosterone on song, aggression, and nestling feeding behavior in male great tits, Parus major. Hormones and Behavior 41:178-186. Verbeek MEM, Boon A, Drent PJ, 1996. Exploration, aggressive behavior and dominance in pair-wise confrontations of juvenile male great tits. Behaviour 133:945-963. Vermeulen GJ, Lambert JG, van der Looy MJ and Goos H J, 1994. Gas chromatographic-mass spectrometric (GC-MS) analysis of gonadal steroids in plasma of the male African catfish, Clarias gariepinus: effects of castration or treatment with gonadotropin releasing hormone analogue. General and Comparative Endocrinology 96:288-297. Whitehouse MEA, 1997. Experience influences male-male contests in the spider Argyrodes antipodiana (Theridiidae:Araneae). Animal Behaviour 53:913-923. Wilson DS, Clark AB, Coleman K, Dearstyne T, 1994. Shyness and boldness in humans and other animals. Trends in Ecology & Evolution 9:442-446. Zhuikov AY, 1993. Avoidance-learning and aggression in guppies. Animal Behaviour 45:825-826.. 37.

(44) Tables Table 1. Cox regression modeling the relationship between contest experience (W, N & L) and hormone (testosterone and cortisol) levels (log transformed) and aggressiveness (latency to gill display log transformed), controlling for strain, body size and last fighting outcome. Cox regression models the hazard (instantaneous risk) of the event (gill display); a higher hazard of gill display (positive coefficient) indicates a shorter latency to gill display, ie higher aggression. (T: testosterone, CORT: cortisol; * P < 0.05) Latency to gill display Variable. b  SE. Contest experience. df. 2. P value. 2. 7.54. 0.023*. W1. 0.37  0.19. 1. 3.61. 0.058. L1. -0.17  0.20. 1. 0.75. 0.386. Baseline T. 0.71  0.20. 1. 13.16. < 0.001*. Post T. -0.68  0.21. 1. 10.76. 0.001*. Baseline CORT. 0.12  0.08. 1. 2.38. 0.123. Post CORT. 0.03  0.07. 1. 0.23. 0.634. 4. 11.16. 0.025*. Strain Body Size. -0.11  0.05. 1. 3.82. 0.051. Last L2. 0.00  0.16. 1. < 0.001. 0.991. 1. An indicator variable for the focal individuals that received a pre-designated winning. or losing experience; the baseline group comprised the individuals that received no experience. 2. An indicator variable for the focal individuals that lost more than one month. previously; the baseline group comprised the individuals that won a contest more than one month previously.. 38.

(45) Table 2. Cox regressions modeling the relationship between aggressiveness (latency to gill display log transformed), contest experience (W, L & N), the interaction between aggressiveness and contest experience and hormone (testosterone and cortisol) levels (log transformed) and boldness (latency to emerge from a shelter), controlling for strain, body size and last contest outcome. Cox regression models the hazard (instantaneous risk) of an event (emerging from the shelter); a higher hazard of emerging from the shelter (positive coefficient) indicates a shorter latency to emerge from the shelter (ie higher boldness). (T: testosterone; CORT: cortisol; * P < 0.05) Latency to emerge from shelter P value b  SE 2 9.29 0.002* -0.19  0.06 1.12 0.571 0.60 0.438 -0.15  0.19 0.06 0.803 0.05  0.19. Variable df Latency to gill display 1 Contest experience 2 1 W 1 L1 1 Latency to gill display  1.31 0.520 2 Contest experience 7.05 Baseline T 1 0.008* 0.15  0.08 0.51 0.476 Post T 1 -0.14  0.19 3.69 0.055 Baseline CORT 1 0.15  0.08 0.37 0.545 Post CORT 1 0.04  0.07 13.05 Strain 4 0.011* 0.10 0.750 Body Size 1 -0.02  0.05 2 0.03 0.866 Last L 1 -0.01  0.08 1 An indicator variable for the focal individuals that received a pre-designated winning or losing experience; the baseline group comprised the individuals that received no experience. 2. An indicator variable for the focal individuals that lost more than one month. previously; the baseline group comprised the individuals that won a contest more than one month previously.. 39.