Behavioral Ecology Advance Access published online on February 27, 2008
Behavioral Ecology, doi:10.1093/beheco/arn001
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
The role of avoidance learning in an aggressive mimicry system
School of Integrative Biology, University of Queensland, Brisbane, Queensland 4072, Australia
Address correspondence to K.L. Cheney. E-mail: k.cheney{at}uq.edu.au.
Received 4 June 2007; revised 18 December 2007; accepted 30 December 2007.
 |
ABSTRACT |
|---|
 TOP ABSTRACT INTRODUCTIONMETHODSRESULTSDISCUSSIONFUNDINGREFERENCES |
|---|
Batesian mimicry systems are driven by predators that learn to associate a distinct signal with an unpalatable species and then avoid similar signals on future occasions, including those emitted by palatable mimics. Whereas many studies have investigated learning and predator memory in protective mimicry, few studies have considered learning in other mimicry systems. Aggressive mimics are defined as the resemblance of a predatory or parasitic species to another nonthreatening or even attractive species in order to approach and attack prey or to gain other benefits such as pollination or transportation. If signal receivers learn to avoid aggressive mimics and their models or learn to distinguish mimics from their models, then the success of mimics should decline. On coral reefs, juvenile bluestreak cleaner fish (Labroides dimidiatus) are aggressively mimicked by the bluestriped fangblenny (Plagiotremus rhinorhynchos). Instead of removing parasites from larger host fish, fangblennies attack reef fish and remove scales and dermal tissue. In this study, the signal receiver (a damselfish) spatially avoided cleaner fish and their associated fangblenny mimics after aggressive interactions; however, avoidance was not exhibited for longer than 24 h. Damselfish did not appear to discriminate between cleaner fish and mimic fangblenny, however, did discriminate between cleaner fish and nonmimic fangblenny. Avoidance learning was more pronounced when the fangblenny was less similar to the cleaner fish in terms of color. Finally, damselfish learnt to preempt attacks within 1 h of contact with fangblennies. This study provides important insights into the role of both spatial and preemptive avoidance learning in aggressive mimicry.
Key words: aggressive mimics, cleaner wrasse, cleaning symbioses, coral reef fish.
|
INTRODUCTION |
|---|
|
TOPABSTRACTINTRODUCTIONMETHODSRESULTSDISCUSSIONFUNDINGREFERENCES |
|---|
Mimicry systems are described as a classic example of adaptation by natural selection; mimics evolve to resemble another species in order to avoid predation, capture prey, or increase mating opportunities (Wickler 1965
; Zabka and Tembrock 1986). Many mimicry systems involve 3 participants: models, mimics and "signal receivers," or "duped" individuals (Wickler 1968). In protective mimicry systems, signal receivers are predators that learn to associate warning signals emitted by an unpalatable model with a negative experience. These predators then avoid similar signals on future occasions, including those emitted by palatable Batesian mimics (Bates 1862) or defended Müllerian mimics (Müller 1879). In Batesian mimicry, disruption of predator avoidance learning can occur if palatable mimics are sampled too frequently, causing protection of both models and mimics to be reduced (Huheey 1964; Lindström et al. 1997; Pfenning et al. 2001). Furthermore, predators can selectively drive the evolution of mimics to closely resemble their unpalatable models: if predators can differentiate palatable mimics from unpalatable models, then protection of mimics will be lost (Caley and Schluter 2003; but see Holen and Johnstone 2004). Hence, learning, memory, and prey recognition by predators are key factors that drive evolutionary and ecological dynamics of protective mimicry systems (Turner and Speed 1996).
Whereas many studies have examined the process of learning by predators in protective mimicry systems (Turner and Speed 1996; Speed and Turner 1999), there are few theoretical or experimental studies to suggest whether learning is important in other types of mimicry (but see Davies 2000; Kunze and Gumbert 2001; Wong and Schiestl 2002; Wong et al. 2004). Aggressive mimics are defined as the resemblance of a predatory or parasitic species to another nonthreatening or even attractive species in order to approach and attack prey or to gain other benefits such as pollination or transportation (Cott 1940; Wickler 1968). Aggressive mimicry systems may be affected by the ability of the signal receiver to learn avoidance after an attack; if the signal receiver learns to avoid mimics, then the success of mimics will be reduced. If the outcome of aggressive mimicry is nonfatal, there should be an opportunity for the recipient of the mimics' aggression to learn to distinguish between models and mimics or to avoid areas where attacks by mimics are frequent (Cheney and Côté 2005). Here, I classify avoidance by signal receivers in an aggressive mimicry system as spatial, defined as the relocation away from aggressive mimics or from areas where attacks may be frequent, or preemptive, defined as the evasion of direct attacks by mimics, each resulting in the decreased attack success of mimics.
On coral reefs, bluestriped fangblennies (Plagiotremus rhinorhynchos) are aggressive mimics of juvenile bluestreak cleaner fish (Labroides dimidiatus) (Wickler 1966). Instead of removing parasites from client reef fish, fangblennies nip at fish to remove scales and mucus (Wickler 1966; Kuwamara 1981). Bluestriped fangblennies benefit by associating with juvenile cleaner fish in terms of increased access to victims and increased attack rates (Côté and Cheney 2004). However, when the abundance of mimics to models is high, the attack success of mimics declines, indirectly suggesting that reef fish may learn to avoid attacks by aggressive mimics or avoid cleaning stations altogether (Cheney and Côté 2005). Therefore, this study examines whether reef fish 1) spatially avoid cleaner fish models and aggressive fangblennies after attacks by mimics, 2) retain information about attacks by mimics, 3) learn to preempt attacks by aggressive mimics, and 4) can distinguish between models and mimics.
|
METHODS |
|---|
|
TOPABSTRACTINTRODUCTIONMETHODSRESULTSDISCUSSIONFUNDINGREFERENCES |
|---|
Study site and species
The study was conducted at Lizard Island Research Station, Great Barrier Reef, Australia, 23°27'S, 151°55'E, in July 2005 and January and November 2006. All fish were collected with hand and barrier nets from patch reefs at depths of 2–10 m and placed in aquaria with running seawater. Fish were left to acclimatize for 7–14 days before trials began and were fed small pieces of prawn or fish flakes daily.
Bluestriped fangblennies (P. rhinorhynchos) vary in coloration ranging from a "mimic" black body with 1 lateral electric blue stripe extending from snout to tail to a "nonmimic" color. This nonmimic coloration can be an olive, brown, or orange body with 2 light blue–white stripes running laterally (Randall et al. 1997; Côté and Cheney 2005). There appears to be no significant difference in the size of fangblennies in relation to their color (Côté and Cheney 2005; Cheney et al. 2008). Fangblennies with the former color combination closely resemble and associate with juvenile cleaner fish (L. dimidiatus), which are the most widespread and abundant cleaner fish in the Indo-Pacific. Cleaner fish remove ectoparasites, mucus, and diseased tissue from larger "client" reef fish (Feder 1966) and have been shown to significantly reduce ectoparasites loads on their clients (Grutter 1999). Up to 190 fish clients per hour (Grutter 1995) can visit cleaner fish at traditional sites called cleaning stations (Youngbluth 1968). Fangblennies can switch between mimic and nonmimic color forms rapidly (within ca., 5 min) (Moland and Jones. 2004; Côté and Cheney 2005; Cheney et al. 2008), and fangblennies with nonmimic color variations often find protection in schools of fish (Randall et al. 1997; Côté and Cheney 2005). In this study, only black/blue (mimic) and brown/white (nonmimic) individuals were used (Figure 1). These colors were maintained during the observational sessions.
|
Staghorn damselfish (Amblyglyphidodon curacao) were used as the signal receiver in this study. Staghorn damselfish feed on zooplankton and algae, are relatively large damselfish (standard length, 12 cm), and have a small home range (Randall et al. 1997
). This species was chosen due to its high abundance on the reef, its adaptability to aquaria, and because it is attacked by fangblennies in the field (Cheney KL, personal observation). Damselfish were collected from reefs where no fangblennies (Plagiotremus sp.) had been observed during frequent thorough surveys more than 2 years (Cheney KL, unpublished data) and during surveys conducted 1–2 weeks before fish were collected; therefore, it is assumed that they were essentially naive with respect to previous attacks from fangblennies.
Experimental trials
In aquarium trials, damselfish were tested for ways in which they might avoid fangblennies. In 1-m-long aquaria, clear Perspex partitions were placed at 10 cm from each end. Fangblennies and cleaner fish could be placed together at 1 end, or at opposite ends, depending on the treatment. The ends in which the fangblenny and/or cleaner fish were placed were randomly chosen per trial. An individual staghorn damselfish was left in the central portion that was further visually divided into three 26-cm-wide sections using a marker pen (Figure 2). In this central portion, damselfish were given a polyvinyl chloride tube for shelter (20 cm long, 10 cm diameter) in which they could enter to avoid attacks by fangblennies. The damselfish could see both the fangblenny and cleaner fish and could therefore choose to associate with (i.e., be within 26 cm of) the fangblenny and/or cleaner fish or stay in a central, neutral position. The observer was positioned with the experimental fish at eye level and remained motionless in front of the experimental tank for 10 min to allow the fish to settle before the trial began. The experiments were conducted in natural light conditions.
|
To test for both spatial and preemptive avoidance, 3 consecutive observational sessions were conducted.
Observation 1
Before contact with fangblennies, a 30-min observation was conducted to assess whether damselfish would spend significantly more time in 1 section of the aquarium compared with another. At 30-s intervals, the section in which the damselfish was located was noted: adjacent to fangblenny and/or cleaner fish (depending on treatment) or center. Due to the interval nature of the data collected, the proportion of time the damselfish was located in each section of the aquaria was estimated.
Observation 2
The Perspex partitions were then removed to permit physical contact between the fish. A second observation session was conducted for 60 min during which the numbers of successful and unsuccessful attacks by the fangblenny on the damselfish were recorded. A successful attack was considered to be a dart by the fangblenny toward the damselfish and contact made with the body, which usually resulted in a slight jolt by the damselfish (however, after visual inspection, no mark or wound was ever visible on damselfish). An unsuccessful attack was considered to be a dart by the fangblenny toward the damselfish with no contact made. The percentage of total attacks that were successful was calculated in 10-min intervals. If damselfish learned preemptive avoidance, the proportion of successful attacks by the fangblenny was predicted to decline over time. To determine why attacks were unsuccessful, I noted whether the damselfish evaded attacks from the fangblenny (defined as a rapid move away from the fangblenny during its attack) and the number of aggressive chases made by the damselfish toward the fangblenny before an attack and in retaliation to an attack. Whether the damselfish aggressively chased the cleaner fish and time spent cleaning by the cleaner fish were also recorded. A cleaning interaction was defined as any event that involved contact and visual inspection by the cleaner of the body surface of the damselfish, and its duration was determined as the time from when the cleaner approached the damselfish until it departed. Because I was interested in the change of the number of successful attacks and chases over time, 60 min was found to represent an optimum observational period, after which the proportion of attacks and the behavior of the fish did not change significantly with time (Friedman 
< 3.34, P > 0.20).
Observation 3
The Perspex partitions and the fish were replaced as they were at the beginning of the trial. Fish were allowed to settle for 10 min, and the observations were repeated as in observation 1.
Spatial avoidance of fangblennies and/or cleaner fish by the damselfish was tested by calculating the difference in the proportion of time that damselfish spent in each section of the aquarium compared with the first observation. To assess retention of information by damselfish, the third observation session was then repeated 24, 48, and 72 h later.
To test the various hypotheses, 3 versions of the experiment were conducted.
Treatment 1: mimic and cleaner fish (same side) (n = 10)
First, damselfish were tested to determine if they would spatially avoid mimic fangblennies and cleaner fish at a cleaning station after damselfish had contact with both fish. In this treatment, a mimic fangblenny and juvenile cleaner fish were placed on the same side of the aquarium. If damselfish spatially avoided juvenile cleaner fish together with a mimic fangblenny, then the amount of time spent adjacent to the model and mimic would be significantly reduced in the third observation compared with the first. Although 1 end of the aquarium was empty, which may influence the result due to the damselfish's desire to shoal with other fish, using a before and after experimental design controlled for this.
Treatment 2: mimic or cleaner fish (opposite sides) (n = 10)
Second, the ability of damselfish to distinguish mimic from model was tested by placing a mimic fangblenny and juvenile cleaner fish on opposite sides of the aquarium. If damselfish can detect the differences between model and mimic, they should spend significantly more time adjacent to the cleaner fish.
Treatment 3: nonmimic or cleaner fish (opposite sides) (n = 10)
As a control to test for an effect of model–mimic similarity while controlling for species, a nonmimic fangblenny was placed on the opposite side to a juvenile cleaner. Again, if damselfish can distinguish between cleaner fish and fangblenny, they should spend more time adjacent to the cleaner fish. Avoidance of fangblenny should be more pronounced in this treatment compared with treatment 2.
In all treatments, fangblennies ranged in size from 40 to 63 mm (standard length). Fangblennies were size matched with juvenile cleaner fish, and on average, pairs differed by 2.3 ± 1.2 mm. New fish were used for each trial.
|
RESULTS |
|---|
|
TOPABSTRACTINTRODUCTIONMETHODSRESULTSDISCUSSIONFUNDINGREFERENCES |
|---|
Spatial avoidance
Before physical contact with fangblenny or cleaner fish (observation 1), damselfish did not spend significantly more time in any one section of the aquarium in any treatment (Friedman
< 4.16, P > 0.13). After physical contact with fangblennies and cleaners, damselfish in treatment 1, mimic and cleaner treatment (same side), spent significantly less time adjacent to cleaner fish and fangblennies than they did prior to contact (Wilcoxon signed-rank test, Z = –2.81, P = 0.005) and more time in the section adjacent to no fish (Wilcoxon signed-rank test, Z = –2.83, P = 0.005; Figure 3i). In treatment 2, mimic or cleaner treatment (opposite sides), there was no significant difference in time damselfish spent adjacent to fangblenny or cleaner fish but spent significantly more time in the central section (Wilcoxon signed-rank test, Z = –2.29, P = 0.02; Figure 3ii). In treatment 3, nonmimic or cleaner treatment (opposite sides), damselfish spent significantly less time adjacent to fangblennies than they did prior to contact (Wilcoxon signed-rank test, Z = –2.67, P = 0.008; Figure 3iii).
|
Retention of information
The percentage difference in time spent by damselfish adjacent to fangblennies decreased over 24, 48, and 72 h (Friedman test, treatment 1, mimic and cleaner:
=17.91, P < 0.001; treatment 3, nonmimic or cleaner: =27.67, P < 0.001; Figure 4). Forty-eight hours after the initial experiment, there was no longer a difference in the time spent adjacent to fangblenny compared with the first observation (Figure 4). In treatment 2, mimic or cleaner treatment, there was no significant change over time (Friedman =1.67, P = 0.64; Figure 4).
|
Preemptive avoidance
In observation 2, the percentage of successful attacks by mimic fangblennies was greater than for nonmimic fangblennies (mimic 84.1 ± 12.5, nonmimic 70.1 ± 8.5; Mann–Whitney U = 50.5, P = 0.03; Figure 5). The total number of attacks toward damselfish ranged from 0 to 28 attacks per hour (mean ± standard deviation [SD]: 11.1 ± 5.2). The proportion of successful attacks declined for each 10-min interval for both mimic and nonmimic fangblennies (Friedman
< 14.3, P < 0.02; Figure 5), and this decline corresponded to an increase in the percentage of attacks that were evaded by the damselfish (Friedman =15.1, P = 0.01).
|
Damselfish chases and cleaning interactions
In observation 2, the total number of pre- and postattack chases by damselfish ranged from 0 to 30 (preattack) and 0 to 21 (postattack). There was no difference between treatments (Kruskal–Wallis
< 0.39, P = 0.82). However, chases toward fangblennies increased over the 60-min session for both preattack (Friedman =64.6, P < 0.001) and postattack chases (Friedman =50.6, P < 0.001). The total number of chases by damselfish toward nonmimic fangblennies was higher than toward mimics (mean ± SD: nonmimic = 21.6 ± 8.2, mimic = 14.0 ± 6.3; Mann–Whitney U = 33.5, P = 0.002). The number of chases toward the cleaner fish was higher when a mimic fangblenny was present compared with a nonmimic fangblenny; however, this difference was nonsignificant (mean ± SD: mimic = 9.6 ± 4.0, nonmimic = 5.8 ± 2.9; Mann–Whitney U = 57.2, P = 0.06). The time damselfish spent being cleaned by juvenile cleaner fish was low (mimic: 34.5 ± 13.9, nonmimic: 30.0 ± 17.8 s/h), and there was no difference between treatments (analysis of variance F2,29 = 0.19, P = 0.83). There were also no correlations between the amount of cleaning and the total numbers of attacks, successful attacks, or chases by damselfish toward the fangblenny (Pearson's correlations, all r < 0.21, P > 0.46).
|
DISCUSSION |
|---|
|
TOPABSTRACTINTRODUCTIONMETHODSRESULTSDISCUSSIONFUNDINGREFERENCES |
|---|
Both spatial and preemptive avoidance learning exhibited in this cleaner fish–fangblenny aggressive mimicry system could reduce the success of aggressive mimics. Our signal receiver (staghorn damselfish) spatially avoided a cleaner fish and associated mimic fangblenny after contact with the fangblenny; however, it is unclear whether they avoided the fish themselves or the area in which attacks occurred (see Wong and Schiestl 2002
). Reef fish that are attacked frequently due to a high abundance of mimics, or residents that are in close proximity to a mimic's home range, may be more cautious in approaching models or mimics and attempt to distinguish between the 2. They may also avoid particular cleaning stations altogether as reef fish can be "choosy" and visit cleaning stations that provide a high-quality service and avoid cleaning stations where cheating, either by cleaners (Bshary and Schaffer 2002) or by mimics, occurs. Indeed, cleaning stations with mimics present have been shown to receive fewer visitors than those without (Côté and Cheney 2004).
Theory predicts that there should be strong selection on mimics to resemble the model accurately (Sheppard 1958; but see Holen and Johnstone 2004). If signal receivers are able to detect aggressive mimics, then the probability of a successful attack is reduced. In this study, evidence suggests that damselfish were unable to distinguish between cleaner fish models and fangblenny mimics. First, damselfish spent more time in central portion when model and mimics were similar, in terms of color, however, more time adjacent to a cleaner when they were less similar (nonmimic fangblenny). Second, damselfish chased cleaner fish models more when a mimic fangblenny was present compared with when a nonmimic fangblenny was present; however, this result was nonsignificant. Finally, mimic fangblennies were more successful at attacking damselfish compared with nonmimic fangblennies. Therefore, although cleaner mimics may not emit all visual signals used by cleaner fish, such as dancing (Becker et al. 2005), their resemblance appears sufficient to approach fish in order to attack them and, due to their association with cleaner fish, may also benefit from a reduced predation risk (Cheney KL, Bshary R, Grutter AS, unpublished data). Furthermore, although staghorn damselfish may be unable to distinguish models from mimics, visual and detection capabilities in fish vary considerably (Losey et al. 2003); therefore, other reef fish species may perceive models and mimics somewhat differently.
In an artificial system, discrimination learning by bumble bees (Bombus terrestris) of food-deceptive flowers (defined as those that mimic food-rewarding plants in order to attract potential pollinators, however, fail to offer a food reward) was enhanced when olfactory cues were used in addition to visual cues. Learning was greater when the mimic's scent was different to that of its model (Kunze and Gumbert 2001). The importance of different visual signals (e.g., color, dancing, and body shape) and other potential signals used by fish to discriminate between models and mimics requires further investigation.
Staghorn damselfish exhibited spatial avoidance, and therefore information retention, for no longer than 24 h. However, negative reinforcement, which in this case would be an encounter with a fangblenny, may have to occur in order for spatial avoidance to be maintained. Reinforcement of avoidance learning may also depend on whether the signal receiver is site attached, as in this study, or whether they have a large home range as this will affect the likelihood of encountering a mimic. In site-attached species, the likelihood of being attacked may be very high (if they live near a mimic) or very low (if they do not). For nonsite-attached species, frequency of encounter may be generally low or unpredictable; therefore, the interval between encounters may often exceed retention time. In this study, the plastic partition prevented the fangblenny from attacking damselfish and thus for reinforcement of learning to occur. In a recent study, goldfish and trout learned to spatially avoid areas associated with a painful stimulus; however, information retention and response to increased intensity of stimulus varied between the 2 species (Dunlop et al. 2006). Therefore, memory retention of attacks by aggressive mimics may also vary between species of victims.
The proportion of successful attacks by fangblennies declined over time in all treatments as damselfish began to evade attacks. The proportion of successful attacks could also decline through satiation, fatigue, or reduced motivation by the mimic. However, if this occurred, fewer attacks rather than more unsuccessful ones would be expected. There should be a net cost in unsuccessfully attacking a fish in terms of energy expended and risk of injury owing to retaliatory attacks.
In aggressive mimicry, if signal receivers cannot distinguish between models and mimics, then avoidance learning may be context dependent (Holen and Johnstone 2006). For example, the thynnine wasp Newzeleboria cryptoides appears unable to discriminate between chemicals that are emitted by female conspecifics and the aggressive mimic orchid Chiloglottis trapeziformis. However, the wasp will avoid orchid fields after a deception (Wong and Schiestl 2002; Wong et al. 2004). Context learning may also be important in the cleaner fish–fangblenny mimicry system. Reef fish may learn to avoid certain cleaning stations based on landmarks or visit other cleaners such as adult cleaner wrasse or shrimps.
In conclusion, mathematical models investigating population dynamics and evolution of protective mimicry systems have incorporated the rate of learning and memory by predators alongside relative numbers of models and mimics (Huheey 1964; Speed and Turner 1999). This study provides empirical evidence of avoidance learning, retention of information, and the importance of model–mimicry similarity in an aggressive mimicry system. This information will help in developing theoretical models for aggressive mimicry systems that, to date, have not been studied.
|
FUNDING |
|---|
|
TOPABSTRACTINTRODUCTIONMETHODSRESULTSDISCUSSIONFUNDINGREFERENCES |
|---|
Royal Society Postdoctoral Fellowship; Australian Research Council Postdoctoral Fellowship and Discovery Grant (K.L.C.).
|
ACKNOWLEDGEMENTS |
|---|
The author thanks the staff at Lizard Island Research Station for their ongoing logistical support and to A. Grutter for aquaria and fieldwork equipment. J. Becker, B. Smith, R. Jacob, B. Cameron, L. Curtis, M. Eckes, and M. Horton provided valuable assistance in the field. Thanks to I. Côté, A. Goldizen, J. Becker, and C. Jones for comments on the manuscript.
|
REFERENCES |
|---|
|
TOPABSTRACTINTRODUCTIONMETHODSRESULTSDISCUSSIONFUNDINGREFERENCES |
|---|
Bates HW. Contributions to an insect fauna of the Amazon Valley. Lepidoptera: Heliconidae. Trans Linn Soc Lond (1862) 23:495–566.
Becker JHA, Curtis LM, Grutter AS. Cleaner shrimp use a rocking dance to advertise cleaning services to clients. Curr Biol (2005) 15:760–764.[CrossRef][Medline]
Bshary R, Schaffer D. Choosy reef fish select cleaner fish that provide high-quality service. Anim Behav (2002) 63:557–564.[CrossRef][Web of Science]
Caley MJ, Schluter D. Predators favour mimicry in a tropical reef fish. Proc R Soc Lond B Biol Sci (2003) 270:667–672.[Medline]
Cheney KL, Côté IM. Frequency-dependent success of aggressive mimics in a cleaning symbiosis. Proc R Soc Lond B Biol Sci (2005) 272:2635–2639.[Medline]
Cheney KL, Grutter AS, Marshall NJ. Facultative mimicry: cues for colour change and colour accuracy in a coral reef fish. Proc R Soc Lond B Biol Sci (2008) 275:117–122.[Medline]
Côté IM, Cheney KL. Distance-dependent costs and benefits of aggressive mimicry in a cleaning symbiosis. Proc R Soc Lond B Biol Sci (2004) 271:2627–2630.[Medline]
Côté IM, Cheney KL. Choosing when to be a cleaner-fish mimic. Nature (2005) 433:211–212.[CrossRef][Medline]
Cott HB. Adaptive coloration in animals. (1940) London: Methuen.
Davies NB. Cuckoo, cowbirds and other cheats. (2000) Princeton (NJ): Princeton University Press.
Dunlop R, Millsopp S, Laming P. Avoidance learning in goldfish (Carassius auratus) and trout (Oncorhynchus mykiss) and implications for pain perception. Appl Anim Behav Sci (2006) 97:255–271.[CrossRef][Web of Science]
Feder HM. Cleaning symbiosis in the marine environment. In: Symbiosis—Henry SM, ed. (1966) New York: Academic Press. 327–380.
Grutter AS. Relationship between cleaning rates and ectoparasite loads in coral-reef fishes. Mar Ecol Prog Ser (1995) 118:51–58.[CrossRef]
Grutter AS. Cleaner fish really do clean. Nature (1999) 398:672–673.[CrossRef][Web of Science]
Holen ØH, Johnstone RA. The evolution of mimicry under constraints. Am Nat (2004) 164:598–613.[CrossRef][Web of Science][Medline]
Holen ØH, Johnstone RA. Context-dependent discrimination and the evolution of mimicry. Am Nat (2006) 167:377–389.[CrossRef][Medline]
Huheey JE. Studies of warning coloration + mimicry. 4. Mathematical model of model-mimic frequencies. Ecology (1964) 45:185–188.[CrossRef][Web of Science]
Kunze J, Gumbert A. The combined effect of color and odor on flower choice behaviour of bumble bees in flower mimicry systems. Behav Ecol (2001) 12:447–456.
Kuwamara T. Mimicry of the cleaner wrasse Labroides dimidiatus by the blennies Aspidontus taeniatus and Plagiotremus rhynorhynchos. Nank Seib (1981) 23:61–70.
Lindström L, Alatalo RV, Mappes J. Imperfect Batesian mimicry—the effects of the frequency and the distastefulness of the model. Proc R Soc Lond B Biol Sci (1997) 264:149–153.[CrossRef]
Losey GS, McFarland WN, Loew ER, Zamzow JP, Nelson PA, Marshall NJ. Visual biology of Hawaiian coral reef fishes. I. Ocular transmission and visual pigments. Copeia (2003) 2003:433–454.[CrossRef]
Moland E, Jones GP. Experimental confirmation of aggressive mimicry by a coral reef fish. Oecologia (2004) 140:676–683.[Medline]
Müller F. Ituna and Thyridia: a remarkable case of mimicry in butterflies (Meldola R, translator). Proc Entomol Soc Lond (1879) 1879:20–29.
Pfenning DW, Harcombe WR, Pfenning KS. Frequency-dependent Batesian mimicry. Nature (2001) 410:323.[CrossRef][Medline]
Randall JE, Allen GR, Steene R. Fishes of the great barrier reef and coral sea. (1997) Bathurst (Australia): Crawford House.
Sheppard PM. Natural selection and heredity. (1958) London: Hutchinson.
Speed MP, Turner JRG. Learning and memory in mimicry: II. Do we understand the mimicry spectrum? Biol J Linn Soc (1999) 67:281–312.[CrossRef][Web of Science]
Turner JRG, Speed MP. Learning and memory in mimicry. 1. Simulations of laboratory experiments. Philos Trans R Soc Lond B Biol Sci (1996) 351:1157–1170.
Wickler W. Mimicry and evolution of animal communication. Nature (1965) 208:519–521.[CrossRef][Web of Science]
Wickler W. Mimicry in tropical fishes. Philos Trans R Soc Lond B Biol Sci (1966) 251:473–474.
Wickler W. Mimicry in plants and animals. (1968) New York: McGraw-Hill.
Wong BBM, Salzmann C, Schiestl FP. Pollinator attractiveness increases with distance from flowering orchids. Biol Lett (2004) 271:S212–S214.
Wong BBM, Schiestl FP. How an orchid harms its pollinator. Proc R Soc Lond B Biol Sci (2002) 269:1529–1532.[Medline]
Youngbluth MJ. Aspects of the ecology and ethology of the cleaning fish, Labroides phthirophagus. Z Tierpsychol (1968) 25:915–932.
Zabka H, Tembrock G. Mimicry and crypsis—a behavioural approach to classification. Behav Processes (1986) 13:159–176.[CrossRef][Web of Science]
CiteULike 
Connotea 
Del.icio.us What's this?
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||