Assessment and Recognition of Rivals in Anuran Contests · 2016-03-09 · Assessment and...

ARTICLE IN PRESS

Assessment and Recognition ofRivals in Anuran ContestsM.A. Bee*,1, M.S. Reichertx and J. Tumulty**University of Minnesota, St. Paul, MN, United StatesxUniversity College Cork, Cork, Ireland1Corresponding author: E-mail: [email protected]

Contents

1. Introduction
2 1.1 Rival Assessment 4 1.2 Rival Recognition 5 1.3 A Note on Structure and Nomenclature 6
2. Rival Assessment
6 2.1 History and Context 6 2.2 Assessment of Body Size and the Role of Call Frequency 9
2.2.1 Common Toads
12 2.2.2 Natterjack Toads 13 2.2.3 Wrinkled Toadlets 14 2.2.4 North American Bullfrogs 15 2.2.5 Eastern Gray Treefrogs 17
2.3 Frequency Alteration and the Question of Signal Honesty
18 2.3.1 Northern Cricket Frogs 19 2.3.2 Green Frogs 22 2.3.3 Other Cases of Frequency Alteration 24
2.4 Assessment of Graded Aggressive Signals
25 2.4.1 Northern Cricket Frogs 26 2.4.2 Hourglass Treefrogs and Small-Headed Treefrogs 28 2.4.3 Other Cases of Graded Signal Assessment 30
2.5 Nonacoustic Rival Assessment
32 2.6 Summary of Rival Assessment 34
3. Rival Recognition
35 3.1 History and Context 35
3.1.1 Relative Threat Versus Familiarity
35 3.1.2 Components of Recognition Systems 37
3.2 Neighbor Recognition
38 3.2.1 North American Bullfrogs 38 3.2.2 Golden Rocket Frogs 42 3.2.3 Agile Frogs 43
Advances in the Study of Behavior, Volume 48ISSN 0065-3454http://dx.doi.org/10.1016/bs.asb.2016.01.001

© 2016 Elsevier Inc.All rights reserved. 1 j

http://dx.doi.org/10.1016/bs.asb.2016.01.001

2 M.A. Bee et al.

ARTICLE IN PRESS

3.2.4 Strawberry Poison Frogs
44 3.2.5 Concave-Eared Torrent Frogs 46
3.3 Individual Vocal Distinctiveness
47 3.3.1 North American Bullfrogs 47 3.3.2 Golden Rocket Frogs 52 3.3.3 Agile Frogs 53 3.3.4 Strawberry Poison Frogs 54 3.3.5 Concave-Eared Torrent Frogs 55 3.3.6 Studies of Individual Vocal Distinctiveness in Other Territorial Species 55 3.3.7 Studies of Individual Vocal Distinctiveness in Lek-Breeding Species 56
3.4 Perceptual Basis of Neighbor Recognition
57 3.4.1 North American Bullfrogs 57
3.5 Acquisition of Differential Responses to Neighbors and Strangers
61 3.5.1 North American Bullfrogs 62 3.5.2 Studies of Habituation in Other Territorial Species 67 3.5.3 Studies of Habituation in Lek-Breeding Species 67
3.6 Summary of Rival Recognition
70 4. Future Directions 71
4.1 Function
71 4.2 Evolution 73 4.3 Mechanisms 74 4.4 Conclusion 76
Acknowledgments
76 References 76
1. INTRODUCTION

Animals settle disputes through contests (Archer, 1988; Hardy &Briffa, 2013; Huntingford & Turner, 1987). These disputes occur when an-imals compete to determine priority access to limited resources, such asmates, food, shelter, territories, display areas, and nesting sites. Whilesome animal contests escalate to costly physical fights, in which one orboth contestants may sustain serious injury (Dunn, Jander, Lamas, & Pereira,2015; Piper, Walcott, Mager, & Spilker, 2008), most contests do not.Instead, contests are more frequently settled, and in many instances avoidedaltogether by the animals’ reliance on less-costly communication signals(Bradbury & Vehrencamp, 2011). Signals often have features that predictthe signaler’s likelihood of winning a fight (eg, its size or physical condition).By assessing these features of a rival’s signals and using them to determineone’s own engagement or persistence in a contest, animals may avoid paying

Assessment and Recognition of Rivals in Anuran Contests 3

ARTICLE IN PRESS

the costs of escalation (Clutton-Brock & Albon, 1979). Signals also have fea-tures that encode individual identity. By recognizing past rivals based ontheir signals, animals may avoid paying the costs of repeated contests withthe same individual, in which the most likely outcome may be predictablebased on the outcomes of past contests (Falls & Brooks, 1975). However,there may also be costs to, and limitations on, animals’ abilities to gatherthe information required for both the assessment and recognition of rivals.Thus there is likely to be a diversity of strategic contest behaviors in animals,which calls for studies into the functional, evolutionary, and mechanistic ba-ses for this diversity. Here, we critically review advances made during theprevious 50 years in the study of rival assessment and recognition in a groupthat exemplifies the diversity in animal contest behavior, the anuran am-phibians (frogs and toads).

Anurans exhibit an impressive diversity of social, reproductive, andcommunicative behaviors (Wells, 2007). A rich literature has been devel-oped that reports on studies of contest behavior in male anurans, which pri-marily compete for access to reproductive females. This competition cantake many different forms, ranging from physical combat over females orbreeding resources to the exchange of distinct aggressive vocalizationsused during agonistic interactions (Dyson, Reichert, & Halliday, 2013;Gerhardt & Huber, 2002; Wells, 2007). The types of social and reproductiveinteractions that occur between contestantsdand by extension, howcommunication signals mediate these interactionsdare heavily influencedby the environment. Of particular importance in shaping anuran breedingecology is the temporal and spatial availability of aquatic resources requiredby the complex life cycles of most anurans (Wells, 1977b, 2007).

As will become apparent, contest behavior in anurans must be consideredin light of the diversity of reproductive and communicative behaviors foundin this group. The breeding ecology of anurans generally falls along a con-tinuum between “explosive breeders” and “prolonged breeders” (Wells,1977b, 2007). Explosive breeders have extremely short breeding seasons,lasting only a single day in some species. Males often engage in scramblecompetition, attempting to clasp and reproduce with any unpaired femalethey can find and hold on to. In contrast, prolonged breeders have longerbreeding seasons lasting weeks or months, not days. Two mating systemsare common among prolonged breeders. In “lek-breeding” species, contestsarise as males attempt to exclude each other from calling too closely, possiblyto minimize the risk of acoustic interference. Turnover in the selection and

4 M.A. Bee et al.

ARTICLE IN PRESS

use of specific calling sites is often high, resulting in the relatively short-termdefense of a specific calling site by an individual. Calling sites in leks seldomcontain specific resources that might be required for reproduction. Incontrast to lek-breeding species, males of “territorial” species occupy anddefend fixed areas for periods lasting from several weeks or months tomore than a year. These long-term territories typically contain contestedbreeding resources required for successful reproduction, such as ovipositionsites or tadpole-rearing sites. Thus, across the entire continuum of explosiveto prolonged breeders, contest behavior involves aggressively competingagainst rivals for access to reproductive females, to short-term calling sites,or to long-term territories.

1.1 Rival AssessmentThe study of rival assessment in anurans has involved determining whichcharacteristics give individuals an advantage in contests, and how these char-acteristics might be communicated to rivals and used as the basis for decidingwhether to persist, give up, or escalate. In Section 2, we provide some his-torical context on rival assessment (Section 2.1) and then review the majoradvances that have been made in the study of rival assessment in anurans,along with some additional interesting complications, which can be brieflysummarized as follows.

Assessment of body size and the role of call frequency (Section 2.2). Largerindividuals generally have an advantage in animal contests, and this hasbeen confirmed in many anuran species. Moreover, because the frequencyof vocalizations is often negatively correlated with body size, call frequencyis commonly invoked as an acoustic cue by which rivals assess one anotherwhile avoiding outright physical combat. However, not all species conformto this description, and there are cases in which either body size has littleinfluence on contest success or there is no evidence for frequencyassessment.

Frequency alteration and the question of signal honesty (Section 2.3). In somespecies, males actively lower the acoustic frequencies of their calls duringaggressive interactions. This behavior has received much attention becauseof the possibility that weaker individuals could attempt to fool their oppo-nents by signaling a larger body size than they actually possess. In general theevidence for bluffing is mixed, and other explanations for the role of fre-quency alteration in contests remain viable.

Assessment of graded aggressive signals (Section 2.4). Many anurans producegraded aggressive signals with signal properties that covary with the level of


ARTICLE IN PRESS

contest escalation. There is no conclusive explanation for this behavior, inpart, because studies rarely link variation in graded aggressive signals tothe likelihood of further escalation and subsequent contest success.

Nonacoustic rival assessment (Section 2.5). Although acoustic communicationis the most widely studied signaling modality in anurans, evidence is accumu-lating that signals and cues in other modalities, including visual, vibrational,and tactile, may also be involved in rival assessment in many species.

1.2 Rival RecognitionIn Section 3, we turn to rival recognition. The study of rival recognition inanurans has focused on the so-called “dear enemy” effect (Fisher, 1954), inwhich territorial neighbors exhibit relatively low levels of aggression towardeach other, but maintain high levels of aggression toward unfamiliarstrangers. After providing some historical context in Section 3.1, we reviewthe following advances in the study of rival recognition in anurans.

Neighbor recognition (Section 3.2). Some territorial male frogs exhibit aform of vocally mediated neighbor recognition similar to that observed insongbirds. Neighbors elicit lower levels of aggression than strangers, andneighbors are associated with particular locations. However, there is alsodiversity across species in terms of whether neighbors are treated as dearenemies.

Individual vocal distinctiveness (Section 3.3). Acoustic and statistical analyseshave revealed high levels of individual vocal distinctiveness in anurans thatdefend both long-term territories with breeding resources and short-termcalling sites in leks. There is so far no evidence to suggest that territorialityis associated with greater individuality in anuran vocalizations.

Perceptual basis of neighbor recognition (Section 3.4). Repeated exposure tothe calls of a new neighbor allows territory holders to learn about its individ-ually distinctive vocalizations and its location. Preliminary evidence indicatesa good coevolutionary match between signalers and receivers in that receiverthresholds for behavioral discrimination are related to the patterns and mag-nitudes of variation in signals present within and among individuals.

Acquisition of differential responses to neighbors and strangers (Section 3.5).Early studies of the dear enemy effect suggested a form of learning knownas habituation might be involved in directing lower levels of aggressiontoward neighbors. The aggressive responses of anurans exhibit parametriccharacteristics of habituation. Both long-term and short-term forms ofhabituation occur, and their functional characteristics in contests are relatedto whether a species defends long-term territories or short-term calling sites.

6 M.A. Bee et al.

ARTICLE IN PRESS

1.3 A Note on Structure and NomenclatureWe purposefully structure our review in a case study format to highlight notonly the behavioral and ecological diversity among anurans, but also diver-sity in the methods used to investigate rival assessment and recognition inthis group. This approach requires attention to nomenclature, a topic thathas generated much recent debate in anuran taxonomy (AmphibiaWeb,2015; Frost, 2015; Grant et al., 2006; Pyron & Wiens, 2011). In this article,our use of scientific names follows AmphibiaWeb (2015) because it is moreconservative in using names present in much of the older literature wereview.

2. RIVAL ASSESSMENT

2.1 History and Context
The modern study of animal contest behavior is based on a series of
game theory models that were developed to explain the strategic behaviorof individuals engaged in agonistic interactions (reviewed by Kokko,2013; Riechert, 2013). The earliest models addressed the frequent observa-tion that most animal contests are resolved “peacefully” by display ratherthan fights. Why should animals, which are often well equipped with weap-onry for fighting, ever give up a valuable resource without maximal defense?Game theoretic analyses answered this question in terms of individual selec-tive advantage and the distribution of strategies in the population. A funda-mental conclusion from these analyses is that the resolution of contests byconventional means such as display can be advantageous for both contestantsbecause they avoid the costs of escalated fights (Maynard Smith, 1982b;Maynard Smith & Price, 1973).

Initial game theory models assumed contestants were identical except intheir choice of strategy and thus did not account for other differences incharacteristics that influence the outcome of escalated encounters (MaynardSmith & Price, 1973). Parker (1974) introduced the concept of resource-holding potential (RHP), defined as an individual’s ability to win an escalatedphysical contest, and showed that asymmetries in RHP between contestantscould be used to settle disputes. RHP is often related to body size, withlarger individuals being more likely to win (reviewed by Archer, 1988),but other characteristics may also determine RHP, such as physiological state(Briffa & Sneddon, 2007) or experience (Hsu, Earley, & Wolf, 2006; Rutte,Taborsky, & Brinkhof, 2006). Asymmetries in the value of the resource


ARTICLE IN PRESS

between the two contestants can also be used to resolve contests with victorygoing to the individual that places greater value on the resource (Arnott &Elwood, 2008; Enquist & Leimar, 1987; Maynard Smith & Parker, 1976;Parker, 1974).

Assessment in contests describes the decision-making processes that deter-mine whether an individual persists, withdraws, or escalates. The concept ofassessment is the key to understanding how contests are often resolved bydisplays. If signals contain information about individual RHP, then assess-ment of signals could allow for contests to be resolved based on RHP asym-metries, but without escalation to physical combat. An important predictionof early models of rival assessment is that if contests are resolved by RHPasymmetries, then contests involving more closely matched competitorsshould take longer to resolve (Enquist & Leimar, 1983; Parker, 1974). Alarge number of empirical studies, therefore, focused on testing the hypoth-esis of a negative correlation between RHP asymmetry and the escalationand duration of contests (reviewed by Arnott & Elwood, 2009).

Enquist and Leimar (1983) developed the sequential assessment modelto explain individual strategies for persistence in contests based on theassessment of a competitor’s RHP. Individuals base their decision to persistor withdraw from contests on an assessment of their own RHP relative tothat of their opponent; this and other similar models are referred to asmutual-assessment models. While individuals should have a good estimateof their own absolute RHP, their assessment of relative RHP in the sequen-tial assessment model depends on the absolute RHP of their opponent,which is unknown prior to the contest. In the model, contests proceedby a series of interactions in which information is exchanged, allowingeach individual to assess its opponent and update its estimate of relativeRHP. The uncertainty of this estimate decreases with repeated exchanges,and contests persist until one individual gives up because its estimate of itsrelative RHP is lower than a threshold that is set by the expected costs andbenefits of continued persistence. When contestants are more closelymatched, the assessment of which contestant is superior becomes necessarilymore difficult, and more symmetrical contests therefore take longer toresolve. In contrast to mutual-assessment models, in self-assessment models,individuals do not assess their opponent but rather persist in a contest upto an individual threshold that is determined only by their own RHP(Mesterton-Gibbons, Marden, & Dugatkin, 1996; Payne & Pagel, 1996,1997). In the cumulative-assessment model, animals persist according to anindividual threshold and do not assess their opponent’s RHP, as in the

8 M.A. Bee et al.

ARTICLE IN PRESS

self-assessment model, but the decision to withdraw is based on both theindividual’s own expenditures during the conflict and other “external” costs(Payne, 1998). External costs include costs inflicted by the opponent duringphysical fighting along with indirect costs of engaging in aggressivebehavior, such as predation risk and time lost to other activities, such asmating or foraging (Payne, 1998).

Assuming that individuals can obtain reliable information about theiropponent’s RHP through signals and that the evaluation of relative RHPis cost-free, mutual-assessment is the superior strategy because it allows forinferior individuals to give up more quickly when they are clearly out-matched. However, these assumptions are not always met in natural con-tests. While individuals almost certainly have reliable information on theirown RHP, the evaluation of opponent RHP is susceptible to at least twosources of error. First, even honest signals of RHP are generally only honest“on average,” meaning that the correlation between signal and RHP is lessthan perfect and raising the possibility that contestants misidentify their rela-tive qualities (Searcy & Nowicki, 2005). Second, the information content ofsignals of RHP is likely to degrade due to environmental factors duringtransmission from signaler to receiver (Wiley, 1994). Such effects are partic-ularly relevant in anuran choruses, which are characterized by high levels ofnoise that create severe challenges for communication (Schwartz & Bee,2013; Vélez, Schwartz, & Bee, 2013). While some studies have shownthat loud chorus noise reduces the overall incidence of aggressive signaling(Schwartz & Wells, 1983), whether noise affects the assessment strategyused in contests is an interesting open question for future studies. Together,these sources of error can reduce the value of attempting to assess an oppo-nent’s RHP. In addition, there may be costs associated with gathering infor-mation about opponents; if these costs are high enough then theory predictsthat self-assessment is favored over mutual-assessment (Mesterton-Gibbons& Heap, 2014).

Despite these potential limitations, the sequential assessment model washighly influential because its structure resembles the dynamics of many nat-ural contests, in which exchanges of aggressive signals appear to be used forassessment of opponent RHP, and because many studies found support forthe key prediction of the model that contest duration or escalation shouldbe negatively related to the degree of asymmetry between opponents(Enquist, Leimar, Ljungberg, Mallner, & Segerdahl, 1990; Hack, 1997;Jennions & Backwell, 1996; Leimar, Austad, & Enquist, 1991). However,Taylor and Elwood (2003) pointed out that the negative relationship


ARTICLE IN PRESS

between contest duration and measures of relative RHP is not diagnostic ofmutual-assessment and indeed is also expected to arise under self-assessmentand cumulative-assessment strategies, in which relative RHP is not actuallyassessed by the contestants. Taylor and Elwood (2003) described additionalanalyses that could distinguish among these strategies (see Box 1), althoughthe issue remains controversial (Arnott & Elwood, 2009; Briffa & Elwood,2009; Elwood & Arnott, 2012, 2013). In light of the reevaluation of theprevalence of mutual-assessment and the development of better methodsto distinguish among different assessment strategies, a large number ofstudies have been performed to test which strategy best describes contestbehavior in a wide variety of animal species (Briffa & Elwood, 2000;Egge, Brandt, & Swallow, 2011; Elias, Kasumovic, Punzalan, Andrade, &Mason, 2008; Kemp, Alcock, & Allen, 2006; Moore, Obbard, Reuter,West, & Cook, 2008; Morrell, Backwell, & Metcalfe, 2005). However,as discussed in the following section, most studies of anuran contestswere performed prior to Taylor and Elwood’s (2003) publication andthus potentially misinterpreted negative relationships between size asym-metry and contest duration or escalation as evidence for mutual-assessment(see also Dyson et al., 2013). In the remainder of Section 2, we review pre-vious studies of assessment in anurans with an emphasis on assessment basedon vocalizations during aggressive contests.

2.2 Assessment of Body Size and the Role of Call FrequencyMany studies of rival assessment in anurans aimed to test the hypotheses,largely inspired by studies of common toads (Section 2.2.1), that bodysize is a strong predictor of RHP and that contests are characterized byassessment of the opponent’s call frequency as an honest signal of RHP.These studies, therefore, focused on testing one or more of the followingpredictions in different species: (1) larger males are more successful in con-tests (ie, RHP is determined by body size), (2) there is a relationshipbetween call frequency and body size, and (3) individuals respond differ-ently to playbacks of different call frequencies. We discuss in this sectionseveral cases in which these predictions were fulfilled. However, thereare also several studies that have found negative or conflicting evidence.For instance, in many species residenteintruder asymmetries determinecontest outcome, and body size plays a limited or secondary role (Crump,1988; Given, 1988; Heying, 2001; Kluge, 1981; Sullivan, 1982; Wells,1978).

Box 1 Best Practices in Studies of Rival AssessmentAnurans contributedmuch to early investigations of aggressive behavior, but the-ory has advanced greatly since many of these studies were published. For studiesof anurans to continue to be impactful, it will be important to consider the mostrecent developments in techniques to study rival assessment. Here, we outlinekey methodological considerations for future studies of anuran contests.1. Characterize the determinants of RHP. Even for species in which body

mass is the primary determinant of RHP, other characteristics, includingthe size of specific body parts involved in fighting, physiological condition,and previous experience, are also likely to play a role. Rather than examiningeach possible characteristic related to RHP separately, a fruitful strategywould be to simultaneously measure multiple potential components ofRHP to obtain an index of “multivariate” RHP, as recently applied successfullyin a study of chameleon aggressive behavior (Stuart-Fox, 2006). When inves-tigating RHP, it is important to keep in mind its definition as an individual’sability to win an escalated physical contest (Parker, 1974). Thus the abovecharacteristics should ideally be examined in physical fights, although itwill also be of interest to determine if they also resolve less-escalated con-tests. Assessment of rival RHP involves exchange of signals conveying infor-mation on RHP. Thus it is equally important to characterize the aggressivesignaling system by measuring the relationship between aggressive signalcharacteristics and putative correlates of RHP. A critical step is to confirmthat the signals involved are indeed aggressive signals; many studies haveclaimed an aggressive function for variation in signal characteristics that ismore likely related to competition to attract females (Section 2.4). We recom-mend the criteria of Searcy and Beecher (2009) to confirm a signal’s aggres-sive function: it should increase in aggressive contexts, it should predictfurther escalated behavior by the signaler, and it should elicit responses inreceivers consistent with a function in aggression.

2. Test for assessment. Many studies of anurans have assumed mutual-assessment, and this seems reasonable given how responsive individualsare to variation in aggressive signals. Nonetheless, greater care should betaken to rule out alternative hypotheses of cumulative- or self-assessment(Section 2.1) using the framework described by Arnott and Elwood (2009)for staged and natural interactions and Reichert (2014) for playback tests.To distinguish among these strategies, it is necessary to make separate plotsof the relationship between contest duration and RHP for both winners andlosers of natural or staged contests. A strong positive relationship betweenloser RHP and contest duration is predicted for contests that are resolvedboth by self-assessment and by mutual-assessment. However, undermutual-assessment, the relationship between winner RHP and contest dura-tion is predicted to be strong and negative while under self-assessment thisrelationship is predicted to be weakly positive (Arnott & Elwood, 2009;

10 M.A. Bee et al.

ARTICLE IN PRESS

Box 1 Best Practices in Studies of Rival Assessment (cont'd )Taylor & Elwood, 2003). In addition, in size-matched pairings, contest dura-tion should increase with absolute contestant size under self-assessment(and cumulative-assessment) but should not vary with contestant sizeunder mutual-assessment. Distinguishing between mutual-assessment andcumulative-assessment is more challenging and may require a detailedexamination of the dynamics of contests (Briffa & Elwood, 2009). It is impor-tant to keep in mind that it may not always be possible to clearly distinguishbetween different assessment strategies, which are probably better repre-sented as endpoints on a continuum.

3. Use a multifaceted approach. Diverse methodological approaches areavailable for studying anuran aggressive behavior, but in most speciesonly a single type of experimental design has been utilized. Observationsof natural interactions are critical for describing the social context underwhich aggressive interactions naturally take place and for providing baselinedata to validate the ability of staged interactions to replicate key features ofnatural interactions. However, aggressive interactions are uncommon anddifficult to observe in many anuran species, and this approach can only pro-vide correlative data on components of RHP and the role of signals in con-tests. Staged interactions between contestants in natural or seminaturalenvironments provide a more controlled setting but maintain some realismby using two real-life competitors. This approach has only occasionally beenutilized in studies of anurans (Davies & Halliday, 1978; Fellers, 1979), and maynot be feasible for all species, but should be attempted when possiblebecause it allows for ready quantification of multiple features of contest dy-namics and contestant characteristics (Reichert & Gerhardt, 2011). Playbacktests can be used to study the role of aggressive signals in contests and todetermine the assessment strategy used by contestants (Reichert, 2014).Playback designs are particularly useful for isolating the effects of singlecall characteristics on receiver aggressive responses. The disadvantage ofplayback tests is that even interactive designs (Schwartz, 2001) cannot repli-cate all aspects of contests (eg, physical combat). Playback designs are there-fore best used to test specific hypotheses of aggressive signaling based onobservations from natural or staged interactions and measurements of RHP.Finally, experimental manipulations are a potentially powerful but underutil-ized technique. For instance, if energetic state is thought to be an importantcomponent of RHP, feeding experiments could be used to create energeticasymmetries between otherwise matched competitors. Hormonal manipula-tions could create differences in aggressive motivation (Marler, Chu, &Wilczynski, 1995). If aggressive signaling is hypothesized to convey informa-tion about RHP, then experimental manipulation of the characteristic thatdetermines RHP should alter signal characteristics.


ARTICLE IN PRESS

12 M.A. Bee et al.

ARTICLE IN PRESS

2.2.1 Common ToadsThe common toad, Bufo bufo (Bufonidae), provided one of the first empiricaldemonstrations of assessment in animal contests. This species is an explosivebreeder, in which most mating activity takes place within a time span of onlya few days (Davies & Halliday, 1979). Large aggregations of males formaround breeding ponds, and individual males engage in scramble competi-tion by actively seeking out females that are arriving at the pond. Uponfinding a female, males grasp her in amplexus until she oviposits. Competi-tion for females is intense, and males not only grasp at nearly any object thatcrosses their path, but they also attack already-amplexed pairs and try to takeover the female by forcing her current mate to release its grasp (Davies &Halliday, 1977, 1979). The paired male defends the female by kicking theattacker while producing a series of vocalizations (Davies & Halliday, 1978).

In a series of influential papers, Davies and Halliday investigated maleemale competition in common toads as a mechanism promoting size-assortative mating. In an initial laboratory investigation, they placed a femalein a tank with two males: one large and one small and then observed whichmale was able to maintain amplexus with the female. Larger and smallermales were equally likely to be the first to pair with the female. Followingattacks by the other male, however, small males were often forced off of thefemale. Large males never lost possession of the female to a smaller opponent(Davies & Halliday, 1977). Thus large males have an advantage in contests inthis species and body size is an important component of RHP; indeed in thiscase larger males quite literally had a greater potential to hold on to theresource. Observations of a natural chorus showed that the mean size ofpaired males increased as pairs moved from land toward the spawning sitein the pond, implying that larger males had taken over females from smallermales (Davies & Halliday, 1979).

Davies and Halliday (1978) then explicitly examined how assessmentmight take place during these contests. They noted that attacking maleswere more persistent in their attacks when the defender was smaller andhypothesized that males assessed the defender’s size by attending to its vocal-izations given during the encounter. In particular, they examined the effect offundamental frequency, which was a good candidate signal characteristic forsize assessment because it is determined in large part by the size of the vocalcords and thus indirectly by body size (Martin, 1971, 1972). Larger males pro-duce lower-frequency calls. Davies and Halliday (1978) tested the assessmenthypothesis using a combination of playbacks and staged interactions betweenmales. Either a large or small male was allowed to enter into amplexus with a


ARTICLE IN PRESS

female. This male was mechanically prevented from calling while a playbackspeaker broadcast calls of either a small male or a large male. Medium-sizedmales were allowed to attack this pair. As predicted by the assessment hypoth-esis, males were more likely to attack (Fig. 1A), and more persistent in attack-ing (Fig. 1B), a simulated small male (trials with a high-frequency callplayback) than a simulated large male (trials with a low-frequency call play-back). This experiment provided clear evidence that attackers assessed theiropponent’s RHP using characteristics of signals given during the encounter.Nonetheless, signals were not the only means by which assessment took placeduring these encounters. Regardless of the playback stimulus, males were lesslikely to attack (Fig. 1A), and less persistent in attacking (Fig. 1B), a large de-fender; presumably this was mediated by tactile cues of male size and the forceof the defending male’s counter attack. While there is clear evidence that theattacker assesses the RHP of the defender, whether the defender also assessesthe attacker has not been investigated.

2.2.2 Natterjack ToadsIn the natterjack toad, Bufo calamita (Bufonidae), males defend calling sites,the size and location of which are highly variable within and between nights.In natural agonistic encounters, males usually gave up their calling site when

Figure 1 Effects of call frequency on individual persistence in common toad contests.(Redrawn from Fig. 3 of Davies, N. B., & Halliday, T. R. (1978). Deep croaks and fightingassessment in toads Bufo bufo. Nature, 274(5672), 683e685.) Results are from an exper-iment in which medium-sized males could attack mated pairs with either a large orsmall defending male. The defending male was silenced, and calls of either high fre-quency (gray bars) or low frequency (white bars) were played back through a speaker.(A) Number of attacks on the pair; males were less likely to attack when exposed to low-frequency calls. (B) Percent of time spent attacking during the experiment.

14 M.A. Bee et al.

ARTICLE IN PRESS

attacked by larger intruders and usually repelled smaller intruders (Arak,1983). Call frequency was negatively correlated with body size. Individualsexposed to playbacks of high-frequency calls were more likely to attack theloudspeaker while individuals exposed to playbacks of low-frequency callswere more likely to retreat from the calling site. Thus there is evidencethat males assess the RHP of their opponents using call frequency. However,the role of the subject’s own RHP in assessment is unknown. In a study on adifferent population, Tejedo (1988) observed direct competition for accessto females as in the common toad. Here, larger males were more likely toattack mated pairs, although these attacks were rarely successful in displacingthe defender. Interestingly, Tejedo (1988) also noted a relationship betweenthe operational sex ratio (OSR) and the occurrence of fighting. Somewhatcounterintuitively, fights were more common on nights when the OSR wasless male-biased. This suggests that rival assessment strategies may operatedifferently in different social environments, although this has not been inves-tigated. Nonetheless, the process of rival assessment appears very similar incommon toads and natterjack toads, despite potential differences in the re-sources defended by each species.

2.2.3 Wrinkled ToadletsContests between males of the wrinkled toadlet, Uperoleia rugosa (Myobatra-chidae), also involve calling site defense. In both natural and staged interac-tions, larger males were more likely to win contests (Robertson, 1986). Thiswas the case whether the contest only involved exchange of vocalizations orescalated to physical combat (Fig. 2A). Meanwhile, in staged interactions,residents did not have an advantage over intruders. Dominant frequencywas negatively correlated with mass and body size (tibia length). In playbackexperiments, males that were of a similar size or larger than that of the simu-lated intruder attacked the speaker while males that were smaller than theintruder tended to retreat from the playback. By gradually increasing theamplitude of the playback, the threshold at which males initiated aggressivebehavior could be measured. Males had a lower threshold for aggressiveresponses to the small-male playback. Robertson (1986) also noted that innatural contests, less-escalated contests involved greater mass asymmetriesbetween contestants than contests that escalated to physical fighting. Theimplied conclusion of this finding was that males engaged in mutual-assessment of RHP by attending to one another’s call frequencies, althoughas discussed earlier, comparisons between relative asymmetries and contestescalation are insufficient to distinguish among mutual- and self-assessment

Figure 2 Effects of body size on contest success. (A) Heavier males of the wrinkledtoadlet were more successful in natural contests involving displays only or physicalfighting. (Drawn from data in Table 2 of Robertson, J. G. M. (1986). Male territoriality,fighting and assessment of fighting ability in the Australian frog Uperoleia rugosa. AnimalBehaviour, 34(3), 763e772.) The dotted horizontal line represents the null expectation ofno effect of body size; error bars are exact binomial 95% confidence intervals. (B) Effectsof body mass on success in staged interactions in eastern gray treefrogs. Results areshown separately for interactions that reached four different levels of escalation(ADV, advertisement calls only; AG1, only one of the contestants gave aggressive calls;AG2, both contestants gave aggressive calls; PF, physical fight). There was a weak over-all effect of body mass on contest success, although within contest categories this wasonly significant for AG1 interactions. (Drawn from data in Table 2 of Reichert, M. S., & Ger-hardt, H. C. (2011). The role of body size on the outcome, escalation and duration of con-tests in the grey treefrog, Hyla versicolor. Animal Behaviour, 82(6), 1357e1366.)


ARTICLE IN PRESS

strategies. Nevertheless, the results of the playback test favor the mutual-assessment hypothesis because a male’s response depended both on itsown size and that of the simulated opponent. Robertson (1986) also pointedout that male body condition tended to deteriorate over several nights ofcalling and argued that dominant frequency may thus not be the most reli-able signal of a male’s current condition because it probably does not changein concert. We will revisit the issue of the reliability of frequency as a signalof body size Section 2.3.

2.2.4 North American BullfrogsBullfrogs, Rana catesbeiana (Ranidae), have a prolonged breeding season(Wells, 1977b), and males defend territories that function as female ovipo-sition sites (Howard, 1978a, 1978b). Males respond aggressively to intrudersin their territories by producing distinctive aggressive vocalizations (Bee &Bowling, 2002; Bee, 2002; Wiewandt, 1969). In response to vocalizing

16 M.A. Bee et al.

ARTICLE IN PRESS

males and playbacks of advertisement calls inside their territory, males orienttoward the sound source, produce aggressive calls, and move toward it witha series of conspicuous movements and splashes while attempting to localizethe perceived intruder (Emlen, 1976; Howard, 1978a; Ryan, 1980;Wiewandt, 1969). In observations of natural encounters, Emlen (1968,1976) found that most terminated without any physical aggression whileHoward (1978a), in a study of the same population, found that encountersinvolving physical fights were more common than displays. Large body sizewas an important component of RHP: Emlen (1976) found that 20 of 26aggressive encounters were won by larger males, and Howard (1978a) foundthat 78 of 108 encounters were won by larger males and only 9 by smallermales. Howard (1978a) also showed an advantage of male age but arguedthat size was more important because within age classes larger males weremore likely to win. In addition, contests that escalated to physical fightswere characterized by smaller differences in male size than contests thatwere resolved by display only (Howard, 1978a), conforming to theoreticalpredictions that more closely matched competitors should engage in longerand more costly contests (Enquist & Leimar, 1983; Parker, 1974).

Because body size is a component of RHP in bullfrog contests, Bee (2002)tested whether males assess their opponent’s size based on call frequency.Fundamental frequency is strongly negatively correlated with body size inbullfrogs (R2 ¼ 0.81; Bee & Gerhardt, 2001c) and thus could potentiallyserve as a signal of body size and, hence, RHP. Territorial males were exposedto playbacks of synthetic advertisement calls that were either size matched (ie,they had the same frequency as the subject’s calls) or that had a low or a highfrequency that simulated a large or a small male, respectively (Bee, 2002).Although not explicitly designed as a test of whether contests are resolvedby mutual- or self-assessment, because the relationships between both subjectsize and simulated opponent size and contest escalation were tracked, thisdesign could potentially discriminate between these strategies. Bee (2002)measured several variables associated with male aggression including variousmeasures of calling responses, movements in relation to the playback speaker,and the latencies to give an aggressive response or to habituate to the play-back. Despite this thorough examination of males’ aggressive responses, therewas no evidence that the subject’s size, the simulated opponent’s size, or thesize asymmetry between opponents affected how males responded to play-backs (Fig. 3). The conclusion was that even though size is a component ofRHP in this species and call frequency is significantly correlated with size,males do not use call frequency in the process of assessment during contests.

Figure 3 No effect of size on aggressive responses to playbacks in bullfrogs. Bars areaverages (�SE) of the maximum number of aggressive calls given by a male duringany stimulus period. Averages were calculated separately with respect to the simulatedsize of the stimulus based on its fundamental frequency (S, small; SM, size-matched with the same frequency as the subject male’s calls; L, large), the size classof the subject (S, small; M, medium; L, large), and an index of size asymmetry betweensubject and playback stimulus (�, subject smaller than opponent; 0, subject size-matched to opponent; þ, subject larger than opponent). None of these variableshad significant effects on the maximum number of aggressive calls in response to play-backs. (Redrawn from Fig. 3 of Bee, M. A. (2002). Territorial male bullfrogs (Rana catesbei-ana) do not assess fighting ability based on size-related variation in acoustic signals.Behavioral Ecology, 13(1), 109e124.)


ARTICLE IN PRESS

The reason for the lack of frequency assessment in this species is not clear, butamong the potential explanations discussed by Bee (2002) is that frequencyassessment may have been devalued because males lower the acoustic fre-quency of their signals during aggressive interactions (Bee & Bowling,2002), and thus smaller males may give dishonestly low-frequency calls as abluff of large body size (Section 2.3).

2.2.5 Eastern Gray TreefrogsIn the eastern gray treefrog, Hyla versicolor (Hylidae), residents have anadvantage in natural contests over calling sites (Fellers, 1979). Reichertand Gerhardt (2011) used staged interactions to neutralize the residentadvantage and explicitly tested for assessment of body size in this species.

18 M.A. Bee et al.

ARTICLE IN PRESS

This is one of the few studies of anurans to use analyses that were appropriateto differentiate between self-assessment and mutual-assessment. Males weremoved toward one another while they called from wheeled platforms, andthe outcome and duration of the ensuing encounter was recorded. Contestsranged in escalation from exchanges of advertisement calls, in which bothcall effort and the proportion of overlapping calls increased (Reichert &Gerhardt, 2012, 2013b), to exchanges of aggressive calls (Reichert &Gerhardt, 2013a), to physical combat. There was a weak, but statistically sig-nificant, advantage for larger males when all types of contests were consid-ered (Fig. 2B; Reichert & Gerhardt, 2011). However, larger males did nothave an advantage in physical fights as would be expected if size were acomponent of RHP (Fig. 2B). Indeed, males only had a size advantage incertain less-escalated interactions. Accordingly, there was no evidence forassessment of body size in this species: there were no correlations betweenthe size of either winners or losers and the level of escalation reached orthe duration of various components of the contest (Reichert & Gerhardt,2011). Nevertheless, there was some evidence that individuals assessed thecall frequency of their opponents and that contestants with lower-frequencyaggressive calls were more likely to win (Section 2.3.3; Reichert &Gerhardt, 2013a; Reichert, 2014).

Reichert and Gerhardt (2011) argued that body size is unlikely to pro-vide a substantial advantage for eastern gray treefrogs due to the species’ rela-tively low-impact fighting behavior. They suggested instead that energeticstate may be a more important component of RHP. While some suggestiveevidence has been discovered for this hypothesis in eastern gray treefrogs(Section 2.3.3), and in general energetic costs are an important componentof anuran acoustic competition in the context of mate attraction (Prestwich,1994; Wells, 2001), little attention has been paid to the energetic costs ofaggressive behaviors in anurans. Physiological state may be an importantcomponent of RHP in animals (Briffa & Sneddon, 2007), and this is thusa high-priority area for future investigations of anuran contest behavior.

2.3 Frequency Alteration and the Question of SignalHonesty

One of the major debates in the study of animal aggressive signaling is whatmaintains signal honesty (reviewed by Searcy & Nowicki, 2005). In otherwords, if animals compete over valuable resources and base the decisionto withdraw from a contest on an assessment of signals of RHP producedby their opponent, what is to prevent a contestant from dishonestly signaling


ARTICLE IN PRESS

a larger RHP than it actually possesses (Dawkins & Krebs, 1978; MaynardSmith & Parker, 1976)? Such bluffing could reduce the evolutionary stabilityof the communication system because receivers should no longer pay atten-tion to these dishonest signals (Maynard Smith, 1979). This question hasbeen raised in relation to assessment of RHP by call frequency in anurans.There are biomechanical and energetic limitations on the frequencies thatcan be produced by an individual of a given size (Martin, 1971, 1972);thus small individuals may simply be incapable of producing low-frequencycalls, and call frequency would be an index signal of body size (Bradbury &Vehrencamp, 2011). When this is the case, call frequency is by definition anhonest signal of size, and it is reasonable to expect animals to assess the callfrequency of opponents in contests if size is an important component ofRHP. However, several anuran species are capable of altering the frequencyof their calls. This was first demonstrated in white-lipped frogs, Leptodactylusalbilabris (Leptodactylidae), in which males adjusted their own call fre-quencies to more closely match those of a playback stimulus (Lopez, Narins,Lewis, &Moore, 1988). It is unclear whether the change in call frequency inthis species is related to aggression, but subsequent studies in other speciesshowed that lowering call frequency is an important component of aggres-sive signaling interactions. This raises a serious challenge for understandingaggressive signaling: if receivers assess RHP based on call frequency, do sig-nalers lower their call frequency to bluff their body size in agonistic interac-tions? How would assessment take place in this situation? In this section, wewill discuss how these questions have been addressed in studies of severalanuran species.

2.3.1 Northern Cricket FrogsNorthern cricket frogs, Acris crepitans (Hylidae), are one of the best-studiedanuran species in terms of their aggressive behavior (see also Section 2.4.1).Males in this prolonged-breeding species defend calling sites, and in naturalcontests that escalate to wrestling, larger males win (Wagner, 1989a). Con-tests involve exchanges of advertisement calls and there is no distinctiveaggressive call. The dominant frequency of advertisement calls negativelycorrelates with a male’s body length, and indeed it is the best acoustic pre-dictor of male size (Wagner, 1989c). Furthermore, males were more likelyto maintain calling or to attack a playback speaker when it broadcast a high-frequency stimulus and more likely to cease calling in response to a low-frequency stimulus (Fig. 4A; Wagner, 1989a; but see Burmeister, Ophir,Ryan, & Wilczynski, 2002). Thus all of the ingredients seem to be in place

Figure 4 Relationship between call frequency and male behavior in response to play-backs in northern cricket frogs. (A) Behavioral responses of males to high-frequency(gray bars) and low-frequency (white bars) advertisement-call playbacks. Males tendedto attack the high-frequency stimulus and retreat from the low-frequency stimulus.(Drawn from data in Table 1 of Wagner (1989a). Fighting, assessment, and frequencyalteration in Blanchard’s cricket frog. Behavioral Ecology and Sociobiology, 25(6), 429e436.) (B) Mean � SE change in call frequency (measured as the difference betweenthe male’s call frequency during the playback period and the call frequency during aprestimulus control period) for males that responded differently to playbacks. Malesthat attacked the speaker lowered their call frequency more than males that aban-doned the playback; these in turn lowered their call frequency more than males thattolerated (ie, continued calling without moving) the playback. (Redrawn from Fig. 2 ofBurmeister, S. S., Ophir, A. G., Ryan, M. J., & Wilczynski, W. (2002). Information transfer dur-ing cricket frog contests. Animal Behaviour, 64(5), 715e725.)

20 M.A. Bee et al.

ARTICLE IN PRESS

for call frequency to be used as a means for males to assess one another’sRHP. However, male call frequency is not constant; rather, males oftenlowered their call frequencies in response to playbacks (Wagner, 1989a).Furthermore, the extent to which males lowered their own call frequenciesdepended on the frequency of the playback stimulus; males gave lower-frequency calls in response to lower-frequency playback stimuli (Wagner,1989a). Males also gave lower-frequency calls in response to louder play-backs. Thus frequency is not a static feature in northern cricket frogs. Thisfinding called into question the utility of frequency as a signal of RHP inthis species. Nonetheless, males are clearly responsive to variation in oppo-nent call frequency (Wagner, 1989a, 1989b).

To explain the significance of frequency alteration in northern cricketfrogs, Wagner (1992) tested three hypotheses. The first of these was that fre-quency alteration was related to honest signaling of size. This could comeabout either because male size is better predicted by the lowest frequencya male is capable of producing than by its baseline call frequency or becausethe amount by which males decrease frequency is itself a signal of size. How-ever, this hypothesis was not supported because size actually became less


ARTICLE IN PRESS

predictable when males lowered their dominant frequencies in response toplaybacks, and the magnitude of the frequency decrease was not correlatedwith male size.

The second hypothesis was that the frequency decrease was a signal ofsome component of RHP unrelated to size (termed “size-independent”fighting ability). Males of superior physiological condition or that aremore highly motivated may have a greater RHP, and this may be reflectedin the extent to which males decreased their call frequency. There was someevidence supporting this hypothesis. Specifically, in playback tests in whichthe frequency of the playback signal was decreased from 3.5 kHz to 3.3 kHz,males were more likely to abandon calling than they were in response to aconstant 3.3-kHz call. Thus the decrease in frequency, rather than the abso-lute call frequency of a competitor, was the most salient stimulus for repel-ling subject males. Presumably, therefore, the magnitude of frequencylowering is an important component of RHP assessment. Furthermore,the magnitude by which subjects lowered their call frequencies in responseto playbacks was predictive of their subsequent aggressive behavior. Specif-ically, males that either abandoned their calling site or attacked the playbackspeaker lowered their call frequency while males that continued callingthroughout showed no change. Males that attacked lowered their frequencymore than males that abandoned (Fig. 4B; Burmeister et al., 2002). So, fre-quency alteration may indicate a male’s aggressive motivation and intentionsduring a contest. Importantly, a male’s absolute size did not determine itsresponse, and there were no differences in prestimulus call frequency be-tween individuals that ultimately gave different responses to the playback(Burmeister et al., 2002; Wagner, 1992). Thus the reduction of call fre-quency appears to be the most important variable predicting a male’s aggres-sive behavior, and it is independent of male size. However, it has not beenconfirmed that males that lowered their frequency the most were actuallythe most likely to win in natural contests, and thus the extent to which fre-quency alteration truly reflects male fighting ability is unknown.

The final hypothesis was that males lower their frequency to produce adeceptive signal of size. Small males may benefit by producing lower-frequency signals because they intimidate competitors that otherwise wouldeasily win an escalated contest. However, this hypothesis was rejectedbecause the males that lowered their call frequency the most were alsomost likely to attack. This behavior would not be expected of bluffingmales because such males would be the least likely to be successful in a fightand should therefore avoid physical aggression as much as possible.

22 M.A. Bee et al.

ARTICLE IN PRESS

Nevertheless, males that retreated from playbacks did lower their call fre-quencies before doing so (Wagner, 1992). This may have been a “tacticalbluff,” in which individuals attempt to repel opponents by lowering callfrequency but retreat quickly if the opponent persists. However, loweringcall frequency and then retreating is also consistent with assessment of honestsignals, when there is some uncertainty in the assessment of an opponent(Wagner, 1992).

The role of call frequency in northern cricket frog contests is complex,and there is evidence for multiple stages of assessment based on different fre-quency characteristics. Playback tests indicated that absolute call frequencyis important in some circumstances, and the initial frequency of a male isprobably the best indication of its body size. However, body size is appar-ently not the only component of RHP and assessment of frequencylowering is also an important factor in contests in this species. It is quitelikely that multiple stages of assessment take place in a contest, and thatmultiple components of RHP exist and are signaled in different ways to op-ponents. In addition, some evidence suggests that frequency alteration is nota signal of RHP per se but instead is a signal of aggressive intent (Burmeisteret al., 2002). Furthermore, northern cricket frogs also vary many othercomponents of their calls besides call frequency in aggressive interactions,a topic that we discuss in Section 2.4.1.

2.3.2 Green FrogsThe green frog, Rana clamitans (Ranidae), has a breeding system very similarto that of the closely related bullfrog (R. catesbeiana), and males defendterritories with vocalizations and physical aggression. In this species bothindividual size and status as a territory holder are important determinantsof success in contests (Wells, 1977a, 1978). One playback study suggestedthat males base their aggressive response to an intruder on assessment ofthe ratio of its call frequency components (Ramer, Jenssen, & Hurst,1983). However, Bee and Perrill (1996) found that green frogs, like north-ern cricket frogs, lowered the frequency of their calls during playbackssimulating aggressive encounters. Furthermore, males gave lower-frequency calls to lower-frequency playback stimuli, although the magni-tude of this frequency decrease was small (Bee, Perrill, & Owen, 1999).Males also decreased call frequency in response to louder calls (Owen &Gordon, 2005).

The phenomenon of frequency alteration in green frogs was examined inrelation to Wagner’s (1992) three hypotheses (Bee & Perrill, 1996; Bee,


ARTICLE IN PRESS

Perrill, & Owen, 2000). The hypothesis that frequency lowering is a signalof size was rejected because there was no correlation between size and themagnitude of the frequency decrease and because the correlation betweensize and frequency did not improve for calls given in response to playbackscompared to a prestimulus control period (Bee et al., 2000). There was noevidence to support the hypothesis that frequency lowering is a signal ofsize-independent fighting ability because there was no relationship be-tween the magnitude of the frequency decrease and the likelihood of mov-ing toward or attacking the speaker (Bee et al., 2000). However, propensityto attack is not necessarily related to likelihood of winning the contest, andso even a positive relationship in this assay may be better interpreted as arelationship with motivation rather than size-independent fighting ability(Bee et al., 2000). A more direct test of this hypothesis, by examiningthe relationship between success in contests and the extent to which an in-dividual lowers its call frequency, has not been performed. In addition,there was no relationship between body condition, a possible measure ofsize-independent quality, and the magnitude of frequency lowering (Beeet al., 2000).

Evidence in favor of the bluffing hypothesis was mixed. There was norelationship between body size and the magnitude of frequency lowering,so smaller males did not disproportionately lower their call frequencies.After frequency alteration, call frequency was still a good predictor ofbody size. Nonetheless, Bee et al. (2000) point out that if receivers, whichnormally are more likely to retreat from low-frequency calls, do not updatetheir criteria for evaluating frequency in aggressive situations, they may stillbe more likely to flee from small opponents that lowered their call fre-quency than if the opponent’s call frequency remained static. Indeed, smallmales lowered their call frequency much more to a playback simulating alarge opponent than to a playback simulating a small one. Large malesshowed no such differential response. Frequency alteration may also be aby-product, constrained by the mechanics of call production, of theincreased note duration typical of aggressive calls in this species (Bee &Perrill, 1996). While plausible in green frogs, this would not be a generalexplanation for frequency decreases in species in which the temporal char-acteristics remain largely unchanged. Bee et al. (2000) discuss several otheralternative hypotheses for the cause of frequency decreases in greenfrogs, and the main conclusion is that there is no strong evidence for abluffing function of frequency lowering relative to many other possibleexplanations.

24 M.A. Bee et al.

ARTICLE IN PRESS

2.3.3 Other Cases of Frequency AlterationAs discussed earlier (Section 2.2.4), male bullfrogs lowered call frequency inresponse to aggressive challenges and thus produce a deceptive or at leastpotentially less-informative signal of body size during aggressive interactions(Bee & Bowling, 2002). The magnitude of the frequency decrease in thisspecies was small, although receivers are capable of perceiving smallfrequency differences (Section 3.4.1). In carpenter frogs, Rana virgatipes(Ranidae), males decreased the “secondary” (usually dominant) frequencypeak of advertisement calls following playbacks of advertisement and aggres-sive calls (Given, 1999). In addition, male aggressive calls tend to containmore energy in the primary (lower) frequency peak. The functional signif-icance of frequency lowering was not tested in this species. In the strawberrypoison frog, Oophaga pumilio (Dendrobatidae), males lowered the dominantfrequency of their calls during aggressive interactions (Meuche, Linsenmair,& Pr€ohl, 2012). The magnitude of frequency change was significantly corre-lated with body condition. However, it is unknown whether males in bettercondition have a greater fighting ability in this species. Male painted reedfrogs, Hyperolius marmoratus (Hyperoliidae), gave lower-frequency calls inresponse to louder playbacks of advertisement calls (Grafe, 1995). InAmerican toads, Anaxyrus americanus (Bufonidae), males lowered dominantfrequency during vocal interactions with other males (Howard & Young,1998). Interestingly, the correlation between dominant frequency andbody size increased for interacting males compared to isolated males. How-ever, the function of frequency alteration in American toads is more likelyrelated to mate attraction than aggression (Howard & Young, 1998).

In the eastern gray treefrog, H. versicolor, and in Cope’s gray treefrog,Hyla chrysoscelis (Hylidae), males actively lower the frequencies of theiraggressive calls during contests (Reichert & Gerhardt, 2013a, 2014; Reich-ert, 2013b). In eastern gray treefrogs, the magnitude of the frequencydecrease is related to the level of escalation in contests. For staged interac-tions that escalated to physical fights, the aggressive-call frequencies of thetwo opponents decreased significantly as the interaction escalated (Reichert& Gerhardt, 2013a). A similar, but nonsignificant trend was observed for in-teractions that escalated to exchanges of aggressive calls. Winners also tendedto decrease their call frequencies more than losers. Reichert and Gerhardt(2013a) concluded that aggressive call frequency was a graded signal associ-ated with the level of escalation of contests (Section 2.4), and, because of itsrelationship with contest success, potentially important for RHP assessment.The assessment is probably not of an opponent’s size because the magnitude


ARTICLE IN PRESS

of the frequency decrease was not correlated with body size (Reichert,2013b) and because body size was a poor predictor of fighting ability(Reichert & Gerhardt, 2011). The latter finding also likely rules out the hy-pothesis that frequency lowering is a deceptive signal of size in this species.The frequency decrease could be a size-independent signal of fighting abil-ity. Larger frequency decreases may be more energetically costly (althoughthis has not yet been investigated), or the magnitude of the frequencydecrease may somehow reflect aggressive motivation. In playback tests,aggressive calls with a lower frequency were not more effective at drivingaway test subjects, although these tests presented stimuli with a static call fre-quency (Reichert, 2014). The role of aggressive-call frequency in H. versi-color contests thus remains somewhat uncertain.

2.4 Assessment of Graded Aggressive SignalsMany studies of anuran contest behavior have noted that males give signalswith characteristics that vary with the apparent level of escalation or aggres-siveness of the contest (Grafe, 1995; Schwartz, 1989; Wells, 1989). Theoret-ically, these graded aggressive signals present a challenge because it may notbe favorable for an individual to reveal the likelihood of its subsequentaggressive behavior and because the system is open to bluffing by dishonestsignalers (Hauser & Nelson, 1991; Maynard Smith, 1982a, 1994). Nonethe-less, many anuran species produce aggressive signals that are highly variablenot only in frequency characteristics (Section 2.3), but also in temporal char-acteristics, such as pulse rate and call duration, and these call characteristicsseem to correlate with some measure of the escalation of maleemale inter-actions (Wells, 1988). Two different types of graded aggressive signals havebeen described in the literature. First, some species alter characteristics oftheir advertisement calls with changes in context from mate attraction toaggressive competition, and there is thus no discrete aggressive call (eg,northern cricket frogs; Section 2.4.1). Second, other species have qualita-tively different advertisement and aggressive calls, but there is gradationwithin the aggressive call (eg, hourglass treefrogs; Section 2.4.2).

Wagner (1989b) proposed four questions that should be addressed whenstudying the significance of graded aggressive signals. Unfortunately, most ofthese questions have not been answered in most studies of graded aggressivesignaling in anurans: (1) What is the context in which graded signals vary?Graded aggressive signals are generally defined as such because their charac-teristics vary with intermale distance or nearest neighbor call amplitude(Wells, 1988). However, these variables are not always clearly linked to

26 M.A. Bee et al.

ARTICLE IN PRESS

aggressive escalation and may instead be a function of chorus density orcompetition to attract females. (2) Is variation in graded signals correlatedwith subsequent signaler behavior? In many cases, graded signals aredescribed as signaling a male’s aggressiveness but it is unclear what thismeans. It is theoretically contentious that animals should signal their aggres-siveness in the first place (Hurd, 2006). Few studies have examined whethervariation in an individual’s graded aggressive signals correlates with its sub-sequent aggressive escalation such as initiating a physical attack. (3) Doesgraded signal variation affect receiver behavior? Studies of graded aggressivesignals involving playbacks often show that males respond with more“aggressive” signals to playbacks of signals that have been defined to bemore aggressive, showing that they are responsive to the variation in callcharacteristics. However, in terms of the aggressiveness of the response,this is a somewhat circular argument that does not address the question ofwhat the signal actually means in an aggressive context. An appropriatetest of this question would be to examine whether males are more likelyto attack or retreat from playbacks varying in aggressive signal characteristics.(4) Does variation in the signal correlate with fighting ability? Implicit inmany descriptions of graded aggressive signals is that males assess one anotherbased on the level of aggressiveness in graded aggressive signals, but rarely areattempts made to match male behavior in contests to models of aggressivemotivation, and the relationship between variation in graded aggressive-call characteristics and contest outcome has received little study.

2.4.1 Northern Cricket FrogsDynamic contest behavior in northern cricket frogs, A. crepitans, includesnot only alteration of frequency characteristics (Section 2.3.1), but also oftemporal characteristics. In observations of natural agonistic interactions,males gave advertisement calls with longer call groups and lower rates ofproduction of call groups and of calls within call groups; several characteris-tics of the calls within call groups were also variable (Wagner, 1989c). Manyof these characteristics also varied with nearest neighbor distance and thesound pressure level (SPL) of the neighbor’s calls in both naturally spacedmales and during playback tests (Wagner, 1989b, 1989c). Thus Wagner(1989b) argued that males have graded aggressive signals with a suite of tem-poral call characteristics changing as maleemale competition increases.Furthermore, males gave more “aggressive” calls in response to playbacksof low-frequency advertisement calls. This result contrasts with findingsfrom several other anuran species, in which males were more likely to


ARTICLE IN PRESS

express aggressive behavior in the form of attacking a playback speaker inresponse to high-frequency playbacks (Sections 2.2 and 2.3). Althoughlarger males were more likely to win aggressive interactions, male size didnot correlate with temporal call characteristics (Wagner, 1989b).Thus there is no evidence that the graded signal conveys information aboutmale RHP. Furthermore, in playback tests, neither the absolute size of thesubject nor the size of the subject relative to the simulated competitor had aneffect on the temporal characteristics of the subject’s calls (Wagner, 1989b).Males that responded to playbacks with large changes in their temporal callcharacteristics were more likely to attack or abandon the speaker while malesthat “tolerated” the playback speaker showed few changes in temporal callcharacteristics (Burmeister et al., 2002). Thus temporal signal gradation mayindicate whether a male will tolerate an intruder (but not how it willrespond if it does not tolerate the intruder). Burmeister et al. (2002) arguedthat graded changes in temporal characteristics are an “intentional” signalrelated to resource value; if these signal exchanges do not resolve the contest,then individuals escalate and assess each other’s size. However, it is unclearwhat determines resource value or whether it is assessed in northern cricketfrogs, and there is no evidence that variation in temporal characteristics isrelated to success in such contests.

Burmeister, Wilczynski, and Ryan (1999) used playbacks to explicitlytest the effects of variation in temporal call characteristics on receiverresponse. Here, the “aggressiveness” of a male’s response was defined basedon the call characteristics described by Wagner (1989a, 1989b, 1989c) asbeing associated with more escalated aggressive interactions. Malesresponded with more aggressive calls to playbacks of an “aggressive” stim-ulus that had characteristics typical of males interacting at close range thanto a supposedly “attractive” stimulus that had characteristics of relativelyundisturbed males. This effect depended on the order of stimulus presen-tation: when “aggressive” calls were broadcast first, males responded moreaggressively to this stimulus than to a subsequent “attractive” call playback.When the order was reversed, males responded similarly to both stimuli.Importantly, neither subject size nor temporal parameters of playback stim-uli influenced whether a male attacked the speaker or was repelled by thestimulus (Burmeister, Konieczka, & Wilczynski, 1999; Burmeister et al.,2002; Ophir, Burmeister, & Wilczynski, 1997). Instead, the most impor-tant determinants of a male’s persistence in a contest were social environ-mental variables: at higher densities, males were more likely to tolerateintruders, and males became more likely to attack, and less likely to tolerate

28 M.A. Bee et al.

ARTICLE IN PRESS

an intruder or abandon their calling site, as the breeding season progressed(Burmeister, Konieczka, et al., 1999).

All of these results call into question the standard interpretation ofvariation in characteristics of the graded aggressive call as reflecting aggressivemotivation or male RHP. Although males appear capable of discriminating atleast some variation in graded aggressive calls, their response is largely limitedto altering their own call characteristics. Variation in graded call characteristicsdoes not consistently predict a male’s likelihood of attack, and it is unknownwhether it determines the subsequent winner of the contest. Interestingly,female northern cricket frogs actually prefer calls with temporal characteristicstypical of those of interacting males (ie, those classified as more “aggressive”;Kime, Burmeister, & Ryan, 2004). Thus graded changes in temporal callcharacteristics may be more important for female attraction than agonisticcompetition in this species. These two functions are not necessarily mutuallyexclusive. Indeed, males of many species change their advertisement calls in agraded fashion with changes in male density (Wells & Schwartz, 1984a; Wells& Taigen, 1986). These changes are often interpreted as improving a male’sattractiveness to females but such changes also occur in contexts (eg, close-range maleemale interactions) in which aggressive escalation is most likelyto occur. It may not always be possible to cleanly separate the potentiallydifferent roles of graded call characteristics in aggression versus mate attrac-tion, but from the point of view of studies of aggression, it will be importantto confirm that graded changes in call characteristics are indeed a relevantcomponent of assessment or a determinant of contest success before anaggressive function can be assigned to them.

2.4.2 Hourglass Treefrogs and Small-Headed TreefrogsGraded aggressive calling has been studied in several Neotropical treefrogsin the genus Dendropsophus (Hylidae). The aggressive calls of these speciesdiffer from those of northern cricket frogs in two respects. First, the aggres-sive call is distinctly different from the advertisement call. All gradation takesplace within the aggressive call itself (Reichert, 2013a; Schwartz & Wells,1984a, 1984b, 1985; Wells & Schwartz, 1984b). Second, the most escalatedaggressive calls are the least attractive to females (Schwartz, 1987; Wells &Bard, 1987).

The hourglass treefrog, Dendropsophus ebraccatus, has received the moststudy. Male aggressive calls have a higher pulse rate than advertisement calls(Wells & Schwartz, 1984b). As intermale distance decreases or playbackintensity increases, males increase the duration of the main call note, and


ARTICLE IN PRESS

decrease the mean pulse rate and the number of click-note appendages,among other changes (Fig. 5; Reichert, 2011b; Wells & Schwartz, 1984b).Furthermore, males are responsive to variation in temporal call characteristics.Males gave more aggressive calls and calls with longer introductory notes inresponse to playbacks of longer aggressive calls (Wells, 1989). Males were alsosensitive to the presence of multiple competitors and, similar to northerncricket frogs (Burmeister, Wilczynski, et al., 1999), playback order influencedthe characteristics of the males’ aggressive-call response (Reichert, 2011b).

Females preferred aggressive calls with more click notes, which aretypical of interactions between more distant males, over those with fewerclick notes, which are typical of more close-range interactions (Wells &Bard, 1987). Thus the gradation in male call characteristics was describedas a mechanism by which males trade off the attractiveness of their callswith the need to repel rivals. This is plausible from the point of view of

Figure 5 Graded aggressive calling in hourglass treefrogs. All plots are waveform dis-plays of calls from a recording session with a single male in Gamboa, Panama in 2008(M. S. Reichert, unpublished data). (A) Advertisement call, consisting of a pulsed intro-ductory note, followed after a pause by a single click note. (BeD) Aggressive calls,showing typical gradation (increased introductory note duration and decreased click-note production). The arrangement of calls from top to bottom illustrates the usual di-rection of change in males’ calls in response to playbacks of longer or louder aggressivecalls; hence, the call in (D) would be considered the most “aggressive” (but see text).

30 M.A. Bee et al.

ARTICLE IN PRESS

mate attraction, but the role of the graded aggressive call in agonistic inter-actions remains unclear. No data are available on the relationship betweengraded call characteristics and the likelihood of escalation or success in con-tests in hourglass treefrogs. In the small-headed treefrog, Dendropsophusmicrocephalus, Schwartz (1994) found that subjects responded to playbacksthat matched the duration of their own aggressive calls with more and longeraggressive calls. However, as discussed earlier, this only shows that males arevocally responsive to variation in aggressive-call duration and does notaddress the question of whether this variation is involved in assessment dur-ing aggressive interactions. In hourglass treefrogs, large numbers of aggres-sive calls are given throughout the night, and this behavior is often notobviously related to agonistic competition (Reichert, 2010). Instead,many instances of aggressive calling appear to be important in competitivecall timing interactions to attract females (Reichert, 2011a, 2012). Thisdoes not rule out a function of aggressive calls in actual contests and doesnot provide an obvious explanation for the gradation in aggressive calls.Nonetheless, these data illustrate that the aggressive function of graded signalvariation must be explicitly tested rather than assumed because there aremany potential alternative explanations. For instance, Bond (1989) arguedthat the gradation seen in many aggressive displays is unrelated to commu-nication and instead is caused by a process of behavioral efference, in whichpositive feedback from aggressive behaviors results in increasingly intensesignal expression. This interesting alternative has been discussed, but neverexperimentally verified, in relation to graded aggressive calling in anurans(Reichert & Gerhardt, 2013a; Schwartz, 2001).

2.4.3 Other Cases of Graded Signal AssessmentPainted reed frogs, H. marmoratus, have advertisement calls that gradesmoothly into aggressive calls that tend to be longer and contain morediscrete pulses (Grafe, 1995). In response to louder playbacks, males gavemore aggressive calls, and these calls had a longer duration, more pulses,and a lower dominant frequency than aggressive calls given to quieter play-backs (Fig. 6). Playbacks of aggressive calls were more effective at elicitingthese responses than playbacks of advertisement calls. However, the changein dominant frequency was not related to body size, and no evidence wasgiven that males with certain aggressive-call characteristics were more likelyto escalate further or to win contests. Similar results were found in springpeepers, Pseudacris crucifer (Hylidae) (Schwartz, 1989). This species has anaggressive call that is distinguished from advertisement calls by having pulses.

Figure 6 Graded aggressive calling in painted reed frogs. Mean � SE aggressive-callcharacteristics during a control period with no playback (NP), and in response to play-backs of advertisement calls at four different sound pressure levels (SPL). (A) Percent-age of calls that were aggressive calls, (B) Number of pulses per call, (C) Duration ofcall, (D) Call dominant frequency. Redrawn from data in Fig. 6 of Grafe, T. U. (1995).Graded aggressive calls in the African painted reed frog Hyperolius marmoratus (Hyper-oliidae). Ethology, 101(1), 67e81.


ARTICLE IN PRESS

Males gave more and longer aggressive calls in response to louder playbacksof both advertisement and aggressive calls than to quieter playbacks, and inresponse to longer-duration aggressive call stimuli, relative to shorter calls.Thus they are responsive to at least some variation in the graded signal.Schwartz (1989) argued, based on an analysis of trade-offs between callingrates and call duration, that the increase in aggressive-call duration was ener-getically costly and thus a potential signal of male endurance in aggressivecompetition, but this has not yet been tested experimentally. In green frogs,R. clamitans, males responded to louder playbacks both by increasing thenumber of movements toward the speaker and by giving lower-frequency

32 M.A. Bee et al.

ARTICLE IN PRESS

and shorter-duration calls (Owen & Gordon, 2005). The advertisement(Type I) calls of this species appear to grade smoothly into the aggressive(Type III) call. Whether males are responsive to this gradation has not yetbeen investigated.

2.5 Nonacoustic Rival AssessmentAnuran social interactions are considered to be largely mediated by acousticcommunication, and most studies have focused on their acoustic interac-tions. Nonetheless, an increasing number of studies are demonstrating thatnonacoustic signals and cues are significant components of anuran socialbehavior (Starnberger, Preininger, & H€odl, 2014). This corresponds witha general trend toward studies of complex and multimodal signals in animals(Hebets & Papaj, 2005; Higham & Hebets, 2013; Partan & Marler, 1999).While a great deal of studies have described some form of nonacousticsignaling in anuran aggressive interactions, most of these are descriptiveand few have examined in detail how these signals might be involved inassessment during contests.

Visual signaling is very important in diurnal anurans and even plays a rolein many nocturnal species (H€odl & Amézquita, 2001). Many of the earlystudies of anuran aggressive behavior used both playbacks of acoustic signalsand the presentation of physical model frogs (Emlen, 1968; Given, 1993;Wells, 1978; Wiewandt, 1969). Males often responded by attacking themodels and, in bullfrogs, there is evidence that the posture of an intruderis important. Even in the absence of sound playback, males attacked modelsdisplaying a “high” posture that is often associated with aggressive interac-tions and territory ownership while the model in a “low” posture withmost of its body under the water was largely ignored (Emlen, 1968). Emlen(1968) and Ryan (1980) noted that the bright yellow gular area is exposed bymales in the high posture, although it is unknown if this coloration is a signalinvolved in rival assessment. Yellow throat coloration has also been sug-gested as an aggressive signal in some female dendrobatids, which adoptan upright posture and slowly pulsate their throats at territorial intruders(Durant & Dole, 1975; Sexton, 1960; Test, 1954; Wells, 1980). In straw-berry poison frogs, O. pumilio, males maintain fixed territories and bothacoustic and visual cues are involved in aggressive interactions with intruders(Bunnell, 1973). Males are particularly responsive to the brightness ofopponents. In choice tests in which males were given the opportunity tointeract with opponents differing in brightness, subjects were more likelyto approach brighter opponents, and brighter subjects were generally faster


ARTICLE IN PRESS

to initiate approaches and calling (Crothers, Gering, & Cummings, 2011). Instaged contests in a neutral arena, in which the effects of territorial residencywere removed, brightness was a significant predictor of a male’s likelihood ofinitiating an aggressive interaction, and was a better predictor of thisbehavior than body size measures (Crothers & Cummings, 2015). This studyalso found evidence that contests between males that were more symmetri-cal in brightness were more escalated. The authors argued that brightness isused as a signal of aggressive intent in this species and that brighter males maybe superior competitors (Crothers & Cummings, 2015). However, no datawere presented on whether brighter males were actually more likely to wincontests. While brightness is clearly an important variable in aggressive inter-actions in this species, it cannot be assumed that the increased aggressivenessof brighter males actually confers increased success in contests.

In the red-eyed treefrog, Agalychnis callidryas (Hylidae), contests betweenmales involve acoustic and visual signals, as well as a distinctive tremulationbehavior, in which males shake the branch on which they are perched(Caldwell, Johnston, McDaniel, & Warkentin, 2010). In staged contests,winners produced more and longer tremulations than losers. The vibrationalcomponent of the signal appears to be sufficient to elicit aggressive behavior,as males responded to tremulation displays even when they were not facingtheir opponent. In a playback experiment using a robotic frog to simulatevisual and vibrational signals, males responded to playbacks including a trem-ulation display but not to a static model that provided only visual cues(Caldwell et al., 2010). Whether males assess variation in characteristics ofthe tremulation display was not tested. Anurans are likely to be generallysensitive to vibrations (Narins, 1990), and the use of vibrational signalingin agonistic interactions has received little attention but may be common.A related form of signaling involves the production of water surface waves.In two species of fire-bellied toad, Bombina bombina and Bombina variegata(Bombinatoridae), males float at the surface of the water and use theirhind legs to produce surface waves that appear to demarcate territorialboundaries (Seidel, 1999; Walkowiak & M€unz, 1985). Whether this is actu-ally an agonistic signal is unknown. In both green frogs and bullfrogs, malesmake conspicuous splashes in the water while approaching an intruder dur-ing aggressive interactions (Wells, 1978; Wiewandt, 1969). It is conceivablethat these displays are involved in contest assessment, and presumably largeror more energetic males could produce more intimidating splashes.

Finally, tactile cues are very likely to be important during interactionsthat escalate to physical fighting. If and how males perceive these cues has

34 M.A. Bee et al.

ARTICLE IN PRESS

received very little attention in anurans. Davies and Halliday (1978) noted intheir playback study of common toads that males attacked large defendersless than small defenders, even when the large defender was paired with ahigh-frequency call (see Fig. 1). They suggested that this was possiblybecause males also assessed tactile cues from the kicks of large defenders.Assessment should continue to take place during physical combat (Enquist& Leimar, 1983; Payne, 1998) and should be mediated in part by tactilecues of opponent RHP, but how this functions in anurans is unknown.

2.6 Summary of Rival AssessmentThe use of call frequency as a means to assess body size in anuran contests isliterally a textbook example (Alcock, 2013; Davies, Krebs, & West, 2012) ofrival assessment in animal contests and is supported by convincing evidencefrom several species. However, there are at least as many studies of other spe-cies in which this process cannot fully explain the species’ behavior. In somecases, body size may not be the strongest determinant of RHP. In others,body size may be a major component of RHP, but rivals’ call frequenciesare not assessed. Still other species may use entirely different forms of mutualrival assessment, and the possibility of self-assessment has barely been inves-tigated in anurans. There are presently too few species studied to make broadgeneralizations to explain this diversity in assessment strategies.

An important complication in acoustically mediated rival assessment isthat there is often large within-individual variation in aggressive-call charac-teristics. Frequency alteration and graded aggressive calling are related be-haviors that have both been described in a large number of species. Theseforms of vocal plasticity differ in the hypotheses originally proposed toexplain their function: frequency alteration initially attracted widespreadattention because of the possibility that male anurans lower call frequencyas a bluff of body size while graded aggressive calling was thought to berelated to the level of motivation or escalation of contests. Nonetheless, todate there is little convincing evidence that frequency alteration is a decep-tive signal of body size in anurans, and this behavior has been documentedeven in species like eastern gray treefrogs for which size is not a majorcomponent of RHP. Several studies are more supportive of the hypothesisthat frequency alteration is related to size-independent components ofRHP, although it is unknown what those components are. Likewise, thehypothesis that graded aggressive calling is a signal of aggressive motivationhas rarely been tested directly, and much of the evidence relies on evokedvocal responses to playbacks of graded signals. The studies of northern


ARTICLE IN PRESS

cricket frogs are an exception because they linked variation in graded signalcharacteristics to variation in behaviors that are clearly aggressive: attackingor retreating from a simulated opponent. Even in northern cricket frogs,however, graded signaling may be more important for female attractionthan rival assessment, a possibility that generally has not been addressed inother species.

The finding that many species use nonacoustic or multimodal aggressivesignals further emphasizes the potential diversity of forms of rival assessmentin anuran contests and invites further study into how such diversity arose. Inthe next section, we turn to another form of assessment in anuran contestsbased on learning to recognize a rival’s identity.

3. RIVAL RECOGNITION

3.1 History and Context
A common form of social recognition among contest rivals is reflected
in the dear enemy effect (Fisher, 1954), in which territorial individuals showless aggression toward familiar territorial neighbors than toward strangers.This behavior is hypothesized to function in minimizing the costs ofrepeated aggressive interactions with established neighbors (Falls, 1982;Jaeger, 1981; Temeles, 1994; Wilson, 1975). The dear enemy effect hasbeen documented extensively in territorial male songbirds, which respondmore aggressively to playbacks of the songs of unfamiliar males comparedwith those of their adjacent territorial neighbors (reviews in Falls, 1982;Lambrechts & Dhondt, 1995; Stoddard, 1996). Like male songbirds, themales of many anurans defend long-term breeding or multipurpose terri-tories, regularly produce conspicuous vocalizations, and respond aggressivelyto conspecific males intruding in their territories. Thus vocally mediatedneighbor recognition could be widespread in anurans, but as we will discuss,studies of the behavior in this group are few in number, taxonomicallyrestricted, and plagued by methodological problems.

3.1.1 Relative Threat Versus FamiliarityMost efforts to explain the dear enemy effect rely on the assumption thatstrangers pose a greater relative threat than neighbors to continued territoryownership. The reasoning is that unlike a stable neighbor, which alreadypossesses a territory of its own, a stranger could be a nonterritorial floaterthat is more likely to attempt a territory takeover or insert itself between

36 M.A. Bee et al.

ARTICLE IN PRESS

existing territories (Falls, 1982; Getty, 1987). Based on this reasoning, it iseasy to understand why territory holders might behaviorally discriminate be-tween two classes of individuals (neighbors and strangers) that pose differentlevels of threat to territory ownership. Placing the dear enemy effect in theframework of the Prisoner’s Dilemma (Axelrod & Hamilton, 1981) andassuming that neighbors and strangers differ in their relative threat, Getty(1987) showed that cooperating with one’s neighbors (ie, respecting amutual territorial boundary) can be evolutionarily stable given enoughrepeated interactions between the same neighbors.

In contrast to explanations based on differences in relative threat,Ydenberg, Giraldeau, and Falls (1988, 1989) argued that the asymmetricwar-of-attrition model (Parker & Rubenstein, 1981; Parker, 1984) explainsthe dear enemy effect. This mutual-assessment model (Section 2.1) is basedon repeated interactions between the same individuals, in which individualsadopt roles (likely winner or loser) and place “investment bids” on theirpersistence in a contest. Since strangers know little about their opponents,they are more likely to make role mistakes and have more escalated contests.In contrast, established neighbors are less likely to make role mistakesbecause they know their opponents well from previous contests, and there-fore show less aggression. Ydenberg et al.’s (1988) war-of-attrition modelhas been criticized (Getty, 1989) for having limiting assumptions thatmake it an inappropriate framework for understanding the dear enemyeffect. Getty (1989), in turn, proposed an alternate hypothesis also basedon familiarity, but with less-restrictive assumptions, in which contestants“fight to learn” about opponents. Strangers fight to learn if there is anythingto be gained from further escalation, and neighbors show little aggressiontoward familiar neighbors because they have little left to learn.

Temeles (1994) grouped these models into two general hypotheses: therelative threat hypothesis and the familiarity hypothesis. He suggested thatthe relative threat hypothesis should hold for species that defend breedingterritories or multipurpose territories that contained breeding resources.He reasoned that with these types of territories, neighbors may competeagainst each other for access to mates, but strangers threaten to overtakethe territory and with it the breeding resources and potential future mates.In contrast, in species that defend only feeding territories lacking breedingresources, the threat posed by neighbors and strangersdstealing food fromthe territorydshould be the same. The familiarity hypothesis predicts thatthe dear enemy effect should be present in any territorial system in whichrival neighbors can potentially become familiar with each other regardless


ARTICLE IN PRESS

of territory type. Consistent with the relative threat hypothesis, the dearenemy effect tends to be more often documented in species that defendmultipurpose/breeding territories than in species that defend feeding terri-tories (Temeles, 1994). The data supporting the relative threat hypothesisover the familiarity hypothesis come primarily from birds. No study hastested the predictions of these two hypotheses in anurans.

3.1.2 Components of Recognition SystemsRecognition is an internal process on the part of an animal receiving a signal; itoccurs when a receiver’s perception of a signal matches some internal rep-resentation or template coded by the nervous system (Bee, 2006). Evidencefor recognition comes from behavioral discrimination among signals represent-ing different social categories, such as neighbor versus stranger. The absenceof behavioral discrimination among signals cannot rule out the possibilitythat a receiver recognizes the signals as belonging to different social cate-gories, but that it also treats the categories as biologically equivalent(Beecher, 1991; Beecher & Stoddard, 1990; Stoddard, 1996). Because thebehavioral response of a receiver has fitness consequences for both thesignaler and receiver, biologists often study recognition systems that considerboth individuals (Sherman, Reeve, & Pfennig, 1997). A key componentof all social recognition systems is the requirement that signals exhibit greatervariation among the to-be-discriminated social categories than within thosecategories. In the context of neighbor recognition, the relevant variation oc-curs at the level of individuals, as we would not generally expect the cate-gories of neighbor and stranger to have distinct, group-specific signatures.Thus a requirement for neighbor recognition is that signals are individuallydistinctive. Additionally, receivers must be able to perceive differences insignals associated with different social categories. Any individually distinctiveacoustic property could potentially serve as a cue for vocally mediatedneighbor recognition. However, we might expect receivers to use only asubset of properties to discriminate among individuals, such as those thatbest identify individuals or propagate through the environment with mini-mal attenuation and degradation or those to which the animals’ sensorysystems are especially sensitive (Bee & Gerhardt, 2001c; Bee, Kozich,Blackwell, & Gerhardt, 2001). Finally, receivers must acquire the internaltemplate to which they match their perception of signals. Recognition ofa territorial neighbor is based on an animal’s previous experience withthat neighbor and thus requires learning. Because the dear enemy effect isdefined by relatively lower levels of aggression toward neighbors,

38 M.A. Bee et al.

ARTICLE IN PRESS

habituation is commonly suggested as a form of learning that could underliethis behavior (Brooks & Falls, 1975; Shettleworth, 2009; Wiley & Wiley,1977). In the remainder of this section, we review studies of anurans thathave investigated rival recognition and the components of rival recognitionsystems.

3.2 Neighbor RecognitionThere have been relatively few studies of neighbor recognition and the dearenemy effect in anurans. In addition, the species in which this behavior hasbeen investigated are taxonomically restricted, coming from only twogenera in each of two families (Ranidae and Dendrobatidae). These studiesdemonstrate conclusively that, in some species, dear enemy relationships doexist between territorial neighbors. However, some of these studies also suf-fer from methodological shortcomings that complicate interpretationsabout acoustically mediated neighbor recognition in territorial anurans(see Box 2).

3.2.1 North American BullfrogsThe first published study, and still one of the most comprehensive studies, ofneighbor recognition in anurans was conducted by Davis (1987) on bullfrogs(R. catesbeiana). As described in Section 2.2.4, bullfrogs are prolongedbreeders. Males aggressively defend aquatic areas that function as breedingterritories because females use them as oviposition sites. Males may alsoopportunistically feed in their territories (Schwartz, Bee, & Tanner, 2000),potentially making them multipurpose territories. It is not uncommon formales of this species to possess the same territory for periods lasting severalweeks during their 2e3 month breeding season. This level of site tenacitywould provide ample time for adjacent territorial neighbors to establishdear enemy relationships.

Davis (1987) first tested the hypothesis that males respond less aggres-sively to hearing the calls of their nearby territorial neighbor comparedwith hearing the calls of a stranger. Territory residents were presentedwith both types of calls from a speaker placed near the territory boundaryshared with the neighbor. Males produced more advertisement calls andmore aggressive calls, and approached closer to the playback speaker, inresponse to hearing the calls of a stranger compared with hearing the callsof their neighbor (Fig. 7A). This result, thus, demonstrated that territorialmale bullfrogs can learn to perceptually discriminate between the calls of

Box 2 Best Practices in Studies of Rival RecognitionFuture discoveries about the function, evolution, and mechanisms of rival recog-nition in anurans will be made by adhering to several best practices that easilyavoid methodological issues that make interpretation of results difficult.1. Know the animal’s natural history. Anurans exhibit remarkable diversity in

their social and reproductive behaviors (Wells, 1977b, 2007). Some speciesdefend fixed, resource-based territories for long periods of time, othersdefend calling sites in leks, while still others defend no territory or callingsite, and may instead defend females directly. In the context of rival recog-nition, it is imperative that researchers know what is being defended. Suchknowledge is not only key to interpreting results in light of theoretical andconceptual treatments of rival recognition (Temeles, 1994), it is also essentialfor drawing inferences from broad-scale comparative studies across species.

2. Measure aggressive behaviors. In studies of the dear enemy effect, the keybehavioral difference of interest is whether or not subjects respond lessaggressively toward neighbors compared with strangers. It is, therefore,absolutely imperative to choose response variables that reflect aggression.This is straightforward when the study species produces either acousticallydistinct or graded aggressive calls that are used in territorial interactions(Bourne et al., 2001; Davis, 1987). Most territorial male anurans also exhibitpositive phonotaxis toward calling males perceived as intruding into theirterritories, sometimes using characteristic displays (Bee, 2003a; Davis,1987). Reliance exclusively on the so-called “evoked vocal response,” inwhich male anurans simply increase their vocal output by producing morecalls (Feng, Arch, et al., 2009) or longer calls (Lesbarr�eres & Lodé, 2002), isproblematic if research with the study species has not already confirmedthat such measures are indicators of aggression. Male frogs often increasetheir vocal output in the context of vocal competition for mates (Reichert& Gerhardt, 2012), and this behavior may have little to do with overtaggression.

3. Test animals in their natural habitat. Key to understanding rival recogni-tion is the resource that animals defend (Temeles, 1994). This is why fieldplayback experiments are so widely used to study this behavior. They allowresearchers to investigate the aggressive behaviors animals exhibit in de-fense of resources while they are actually defending those resources. Proto-cols that involve removing animals from their natural habitat to test them inneutral arenas (Feng, Arch, et al., 2009) are best avoided, at least at the initialstages of empirical investigation of a particular species.

4. Use territorial neighbors as stimulus donors. Compared with the litera-ture on songbirds, some studies of neighborestranger discrimination in an-urans have used much looser criteria for determining which animals areneighbors. Davis (1987) and Bee (2003a), in keeping with the previous song-bird literature, regarded neighbors to be those individuals that shared a

(Continued)


ARTICLE IN PRESS

Box 2 Best Practices in Studies of Rival Recognition (cont'd )territorial boundary with the subject. In contrast and departing from thisstandard, Lesbarr�eres and Lodé (2002) treated any frog calling in the samepond as familiar, and Feng, Arch, et al. (2009) considered any two frogs call-ing within the same 400 m2 area to be familiar. Some playback studies havefailed to report any details about the males that served as neighbor stimulusdonors (Bourne et al., 2001). Such loose (or absent) definitions of what con-stitutes a neighbor are problematic. If the goal of a playback test is to deter-mine whether territory residents exhibit relatively lower levels of aggressiontoward neighbors compared with strangers, then negative results are espe-cially difficult to interpret using any stimulus donor other than an immedi-ately adjacent territorial neighbor. Determining neighbor status couldrequire playbacks or markerecapture methods to map territories prior to ex-periments aimed at testing neighborestranger discrimination.

5. Replicate stimuli. Collectively, studies of rival recognition in anurans haveignored the intense debate concerning important issues of experimentaldesign, pseudoreplication, and external validity that erupted a quarter cen-tury ago in the songbird literature (Catchpole, 1989; Kroodsma, 1989a,1989b, 1990; Kroodsma, Byers, Goodale, Johnson, & Liu, 2001; McGregoret al., 1992; Searcy, 1989). This debate highlighted that problems of interpre-tation can arise when researchers (1) use a single exemplar to represent anentire class of stimuli (eg, using a single stranger to represent all strangers),(2) replicate the experiment by presenting this single exemplar to a numberof subjects, and then (3) conclude that observed differences are due to theclass of stimuli presented. A more appropriate experiment with greaterexternal validity would replicate stimulus exemplars within a particularclass and present each subject with a different exemplar. Sadly, studies ofneighborestranger discrimination in anurans have either used only one ora few stimulus exemplars to represent the classes of “neighbor” versus“stranger” (Bee, 2003a; Feng, Arch, et al., 2009; Lesbarr�eres & Lodé,2002) or failed to describe their methods in sufficient detail to determinewhether multiple class exemplars were used (Bourne et al., 2001; Davis,1987). The use of one or only a few stimulus exemplars to represent the clas-ses of “neighbor” versus “stranger” is problematic because it increases theprobability that factors unrelated to identity could have a disproportionateeffect on the outcome of the experiment. For example, as described in Sec-tion 2, male anurans often respond more (or less) aggressively to stimulidepending on acoustic properties potentially unrelated to identity, such asthose indicating RHP or motivational state. If calls recorded from only oneor a few males are used as exemplars of the stranger class, and these callsare, by chance, particularly effective (or ineffective) at evoking aggression,using only a small number of exemplars could increase (or decrease) the

40 M.A. Bee et al.

ARTICLE IN PRESS

Box 2 Best Practices in Studies of Rival Recognition (cont'd )overall aggression given in response to strangers. This example highlightsone way that putative treatment effects could be biased due to factors un-related to a focal male’s familiarity with the signaler, thereby producingmisleading results that appear consistent (or inconsistent) with the dear en-emy effect. Future studies should properly replicate stimuli representing theclasses of neighbors and strangers. Post hoc acoustic analyses comparingneighbor and stranger stimuli should also be conducted to rule out the pos-sibility that these two classes of stimuli differed systematically in someacoustic property potentially related to the signaler’s RHP or motivation.

6. Control for temperature effects. Because they are ectotherms, anuranvocalizations vary predictably with temperature (Gerhardt & Huber, 2002).Therefore temperature effects must also be accounted for when choosingstimuli for acoustic playbacks to rule out the possibility that individual differ-ences in temperature (instead of individual voice differences) elicit differ-ences in behavioral responses (see Box 3 for additional information).


ARTICLE IN PRESS

different individuals and that they behaviorally discriminate between the callsof neighbors and strangers in ways predicted by the dear enemy effect.

In a second experiment, Davis (1987) tested the hypothesis that reducedlevels of aggression toward familiar neighbors depend upon the location in

Figure 7 Acoustically mediated neighbor recognition in territorial male bullfrogs. Meannumber of aggressive calls and distance moved toward a speaker. (A) When calls werebroadcast from the direction of a neighbor’s territory, males were more aggressive inresponse to the calls of a stranger than those of a neighbor. (B) Males were moreaggressive in response to a neighbor’s calls played from a novel location comparedwith the same calls played from the direction of the neighbor’s territory. Drawn fromdata presented in Davis, M. S. (1987). Acoustically mediated neighbor recognition in theNorth American bullfrog, Rana catesbeiana. Behavioral Ecology and Sociobiology, 21(3),185e190.

42 M.A. Bee et al.

ARTICLE IN PRESS

which the neighbor is encountered. Territory residents were presented withthe calls of their neighbor from two different locations, the “normal” loca-tion corresponding to the neighbor’s usual position and a “new” locationcorresponding to the opposite side of the subject’s territory. Subjectsresponded more aggressively to hearing a familiar sound from the new loca-tion compared to when the familiar sound was presented from its normallocation (Fig. 7B).

Davis’ (1987) results are important because they demonstrated for thefirst time that a territorial frog could learn both to perceptually discriminatebetween familiar and unfamiliar calls based on individual differences in theiracoustic properties, and to associate calls having familiar acoustic propertieswith specific spatial locations. In many ways, the pattern of results reportedby Davis (1987) paralleled earlier results reported for many songbird species(reviewed in Falls, 1982; Lambrechts & Dhondt, 1995; Stoddard, 1996),suggesting some degree of convergence in rival recognition in these twodistantly related groups of animals. This groundbreaking work on bullfrogswould later form the foundation for research investigating the componentsof rival recognition systems in this species. But for nearly 15 years,Davis’ (1987) work would stand as the only demonstration of rival recogni-tion in anurans.

3.2.2 Golden Rocket FrogsThe second demonstration of the dear enemy effect in anurans was a studyby Bourne, Collins, Holder, and McCarthy (2001) of golden rocket frogs,Anomaloglossus beebei (Dendrobatidae). Restricted to a small geographicrange in Guyana, this species breeds exclusively in the water-filled phytotel-mata of large terrestrial bromeliads. Males defend territories in these brome-liads, from which they advertise to, and court females as potential mates.Oviposition occurs in the phytotelmata in the male’s territory, and malescare for the embryos and tadpoles developing in their territories (Bourneet al., 2001; Pettitt, 2012). Thus the territories males defend containbreeding resources. Males can defend the same territory for long periodsof time lasting, in some cases, up to several years ( J. Tumulty andM. A. Bee, unpublished data). Males produce a distinct aggressive call thatis given during territorial interactions with other males (Pettitt, 2012).

In a comprehensive study of the social and reproductive behaviors ofgolden rocket frogs, Bourne et al. (2001) reported that males of this speciesexhibit the dear enemy effect and discriminate between the advertisementcalls of a stranger and their nearby territorial neighbor. Using an experimental


ARTICLE IN PRESS

design modeled after a two-alternative choice test, Bourne et al. (2001) foundthat males were more likely to produce aggressive calls and approach thespeaker broadcasting a stranger’s call than one broadcasting a neighbor’scall. Unfortunately, however, Bourne et al. (2001) provided few details abouttheir methodology and did not present quantitative results, so it is difficult tointerpret their findings and compare them with other studies. More recentwork has confirmed the results of Bourne et al. (2001), showing that territo-rial males have lower thresholds for responding aggressively to the calls ofstrangers compared with neighbors ( J. Tumulty and M. A. Bee, unpublisheddata).

3.2.3 Agile FrogsAgile frogs, Rana dalmatina (Ranidae), are pond-breeding anurans that arefound throughout much of southern Europe and that have a relatively shortbreeding season in early spring (Lodé & Lesbarr�eres, 2004; Schneider,Sofianidou, & Kyriakopoulou-Sklavounou, 1988). There so far appears tobe little evidence from behavioral observations to indicate that male agilefrogs defend breeding or multipurpose territories. Males produce long, pul-satile advertisement calls (Schneider et al., 1988), but distinct aggressive callshave apparently not been described for this species, nor has the advertise-ment call been described as grading into an aggressive signal. Nevertheless,Lesbarr�eres and Lodé (2002) tested the dear enemy effect in this species bycomparing the evoked vocal responses of males to playbacks of “familiar”and “unfamiliar” calls. The authors report that males gave longer calls,with more pulses, in response to unfamiliar calls suggesting to Lesbarr�eresand Lodé (2002) that males discriminated between familiar and unfamiliarmales. However, this study suffers from some critical methodological limi-tations (see Box 2). The most significant of these is that the authors use aless-restrictive definition of “familiar” than that used in nearly all otherstudies of neighborestranger discrimination. In most studies, researchersbroadcast the calls recorded from the nearest territorial neighbor as thefamiliar or neighbor stimulus. In stark contrast, Lesbarr�eres and Lodé(2002) defined familiar calls as those recorded from frogs inhabiting thesame pond. These ponds varied in size from 45 to 212 m2 and contained be-tween 4 and 17 males. The authors did not report how widely spaced themales were, but given the size of the ponds and number of calling males,it is not evident that frogs that happened to be in the same pond wouldnecessarily be familiar with each other. Moreover, there is currently little ev-idence for territoriality in this species nor is there evidence that longer calls

44 M.A. Bee et al.

ARTICLE IN PRESS

signal some aspect of aggression in this species. In light of these concerns, theresults of Lesbarr�eres and Lodé (2002) are problematic and difficult to inter-pret in the context of rival recognition.

3.2.4 Strawberry Poison FrogsThe strawberry poison frog, O. pumilio, is a small, diurnal, and aposemati-cally colored dendrobatid that is a common inhabitant of rain forests inthe southern half of Central America. Similar to bullfrogs and in contrastto golden rocket frogs and agile frogs, the ecology and the social and repro-ductive behaviors of this species have been studied for many years (Brust,1993; Bunnell, 1973; Crothers & Cummings, 2015; Donnelly, 1989a,1989b; Haase & Pr€ohl, 2002; McVey, Zahary, Perry, & MacDougal,1981; Pr€ohl, 1997a, 1997b, 2002, 2003; Pr€ohl & Berke, 2001; Pr€ohl &H€odl, 1999; Stynoski, 2009; Stynoski & Noble, 2011). Males of this specieshave been reported to establish and defend multipurpose territories thatcontain foraging sites, shelter, courtship areas, oviposition sites, and possiblytadpole-rearing sites. Moreover, males can defend the same territory overperiods lasting several weeks or months to perhaps as long as a few years(Donnelly, 1989b; Pr€ohl & Berke, 2001; Pr€ohl & H€odl, 1999; Pr€ohl,1997b). The advertisement calls that males produce from within their terri-tory are also individually distinctive (Pr€ohl, 2003; Section 3.3.4). Given thecombination of individually distinct vocalizations and long-term defense ofmultipurpose territories, the strawberry poison frog was a prime candidatefor a study of rival recognition in anurans.

In two field playback experiments, Bee (2003a) tested the hypothesis thatterritorial males of the strawberry poison frog exhibit the dear enemy effect.In the first experiment, the playback speaker was located in the direction ofthe nearest neighbor’s territory, midway between the central positions fromwhich the subject and neighbor were usually observed calling. Playbacks ofthe calls of the neighbor and a stranger both elicited strong responses fromsubjects in terms of evoked calling, movements, and approaches towardthe speaker. However, there was no evidence of a discriminative behavioralresponse to the calls of neighbors versus strangers: subjects responded equallyaggressively to both types of calls (Fig. 8). One possible explanation for thisoutcome was that, because both neighbors and strangers were heard callingat a location midway between the subject’s territory and that of theneighbor, they were perceived as equivalent threats to territory ownership.To evaluate this possibility, Bee (2003a) repeated the experiment with theplayback speaker located in the approximate center of the neighbor’s

Figure 8 A test of acoustically mediated neighbor recognition in strawberry poisonfrogs. Mean � SE (A) number of call groups, (B) number of movements, (C) distanceof closest approach toward speaker, and (D) maximum approach distance. Results oftwo experiments comparing the aggressive responses of territorial males to the callsof neighbors and strangers are shown: one in which calls of neighbors and strangerswere played from the approximate territory boundary (halfway between estimated ter-ritory centers of two neighbors) and a second in which the calls were played from thecenter of the neighbor’s territory. Data redrawn from Fig. 3 in Bee, M. A. (2003a). A test ofthe “dear enemy effect” in the strawberry dart-poison frog (Dendrobates pumilio).Behavioral Ecology and Sociobiology, 54(6), 601e610.


ARTICLE IN PRESS

territory. This time, aggressive responses were more subdued, but again,there was little evidence for a discriminative behavioral response to the callsof neighbors and strangers (Fig. 8).

Thus despite having individually distinctive calls and reportedly defend-ing long-term, multipurpose territories, males of the strawberry poison frogdid not behaviorally discriminate between their neighbors and strangers inways consistent with the dear enemy effect. At first, this negative result ap-pears to be at odds with the relative threat hypothesis. However, this resultmust be considered in light of more recent work highlighting the ambiguityacross published studies about precisely what breeding resources, if any, aredefended by territorial males of the strawberry poison frog (Pr€ohl, 2005).

46 M.A. Bee et al.

ARTICLE IN PRESS

Wells (2007), for example, even suggests strawberry poison frogs have a mat-ing system closer to that of lek breeders. That a lek-breeding species wouldnot behaviorally discriminate between neighbors and strangers might well beconstrued as consistent with Temeles (1994) relative threat hypothesis.Together, these studies of rival recognition in strawberry poison frogs serveto emphasize the importance of basic natural history data. Even well-researched species can present challenges in interpreting results from play-back experiments.

3.2.5 Concave-Eared Torrent FrogsThe concave-eared torrent frog, Odorrana tormota (Ranidae), breeds alongstreams in Central China. Although little is presently known about the spe-cies’ mating system, reproductive biology, or territorial behavior (Feng,Arch, et al., 2009), its vocal behavior has been the subject of intensiverecent study (Arch, Burmeister, Feng, Shen, & Narins, 2011; Feng &Narins, 2008; Feng et al., 2006; Feng, Narins, & Xu, 2002; Feng, Arch,et al., 2009; Feng, Riede, et al., 2009; Gridi-Papp et al., 2008; Narinset al., 2004; Shen et al., 2008, 2011; Suthers et al., 2006). The species isnotable for its exceptionally complex vocal repertoire with large fluctua-tions in spectral properties within calls and prominent ultrasonic harmonics,which aid communication in noisy stream environments. The function ofdifferent call types within this repertoire has not been investigated, and itis presently unknown whether any of the call types signal aggressiveness.In addition to a large repertoire, there exists substantial within-individualvariation in acoustic properties within a single call type (Feng et al.,2002; Feng, Riede, et al., 2009; Section 3.3.5).

Feng, Arch, et al. (2009) compared the responses of male concave-earedtorrent frogs to the calls of neighbors and strangers. Unlike studies ofneighbor recognition in other anurans, which tested subjects in situ, maleswere collected from the field and tested in a transparent plastic tank(22 � 12 � 15 cm) in a laboratory. Acoustic stimuli recorded from malesin the field were broadcast from directly above the tank, and the authorsquantified the evoked vocal response of males exposed to these calls. Malesgave more calls in response to stranger calls than either neighbor calls or con-trol conditions in which no call was played, suggesting to Feng, Arch, et al.(2009) that males recognized and responded less aggressively to familiar calls.However, there are several methodological problems with this study (seeBox 2). First, similar to the study of Agile frogs by Lesbarr�eres and Lodé(2002), Feng, Arch, et al. (2009) used an atypical definition of territorial


ARTICLE IN PRESS

“neighbor,” which they represented by a call recorded from a male thatoccupied the same general area (up to 400 m2) in the field as the subject.Feng, Arch, et al. (2009) argued that this definition was valid by quantifyingsound attenuation in the field to estimate that the calls of males would besufficiently loud (>60 dB) to be heard and recognized by other frogs occu-pying the same area. Second, males were tested in a neutral arena in a lab-oratory, which eliminates the defended resource, and potentially, thebiological significance of neighbors versus strangers. Finally, variation inthe evoked vocal response of this species has not been empirically linkedto aggression as opposed to mate choice.

3.3 Individual Vocal DistinctivenessMany studies of anuran vocalizations have examined patterns of individualvariation in the context of female mate choice (Bee et al., 2010; Castellano& Giacoma, 1998; Friedl & Klump, 2002; Gerhardt, 1991; Howard &Young, 1998). The studies reviewed in this section have done so explicitlyin the context of rival recognition. Two important trends collectivelyemerge from these studies. First, statistically reliable individual differencesappear to be ubiquitous. This finding is consistent across a wide range of spe-cies with different mating systems and breeding ecologies, suggesting that inall anurans, the interindividual signal variability necessary for recognitionmay be present whether or not recognition is likely to occur. Second, thereis diversity among anuran species in terms of which acoustic propertiescontribute most toward statistically discriminating among individuals. Insome species, spectral properties are the most individually distinctive whilein others this is true of temporal properties. Signal analyses such as thosedescribed in Box 3 and illustrated in the following case studies are usefulfor understanding the discrimination tasks that receivers might face, andfor predicting which signal properties receivers might use to discriminateamong individuals. Readers should bear in mind, however, that whateverthe outcomes of statistical analyses of individual distinctiveness in animal sig-nals, whether or not recognition occurs is always an empirical questionabout receiver behavior, not signal design.

3.3.1 North American BullfrogsRecall from Section 3.2.1 that bullfrogs, R. catesbeiana, exhibit the dear en-emy effect. The bullfrog advertisement call is a slow sequence of long,sonorous notes that males repeat about once every minute or two duringperiods of active calling (Fig. 9A). Each note has a “missing” fundamental

Box 3 Statistical Methods to Describe Individual Distinctivenessin Vocalizations1. Removing temperature effects. As noted in Box 2, anuran vocalizations

depend on temperature. This fact is especially relevant to studies of individ-ual vocal distinctiveness. There are well-established statistical methods forremoving variation in calls due to variation in temperature (Bee & Gerhardt,2001c; Platz & Forester, 1988). Unless calls from different individuals arerecorded over narrow temperature ranges, temperature effects must alwaysbe controlled prior to conducting any of the analyses described in thisarticle. Studies that fail to do so should be regarded with skepticism.

2. Coefficients of variation (CV). Computed as the standard deviation dividedby its corresponding mean, CVs represent a dimensionless measure of vari-ability that can be compared across acoustic properties and across species.More importantly, it is possible to compare CVs both among individuals(CVa) and within individuals (CVw). CVa is computed based on the standarddeviation and mean of the individual means computed separately foreach individual. CVw is computed separately for each individual based onthe mean and standard deviation calculated over the sample of signalsrecorded from that individual. Properties useful for recognition shouldgenerally vary relatively more among individuals (eg, CVa/CVw > 1.0).

3. Model II ANOVA. A commonly used statistical test of individual distinctive-ness is model II ANOVA, which can be used to test the hypothesis that asignal property varies significantly among individuals (Beecher, 1989; Sokal& Rohlf, 1995). This analysis treats individual as a random effect, and it can beused to partition the overall variance in a given property into among-individual variance and within-individual variance. The proportion ofvariance in each signal property that can be attributed to differences amongindividuals versus within individuals can be compared across signal proper-ties by examining variance components (Bee, 2004) or effect sizes (Bee et al.,2010; Pettitt et al., 2013). Beecher (1989) provides some importantcautionary notes about the use of model II ANOVAs to document individualdistinctiveness.

4. Repeatability. As a statistical measure borrowed from population genetics,repeatability represents the proportion of total variability that is due to vari-ability among individuals (Boake, 1989; Falconer, 1981; Lessells & Boag,1987). It is based on making repeated measurements of some phenotypictrait (eg, an acoustic property of a vocalization) from the same individual.It can be computed as the intraclass correlation coefficient from model IIANOVAs. In studies of social recognition, high values of repeatability indicatethat repeated measurements of a signal property within an individual exhibitless variation than measures of the same property across different individ-uals. In studies of individual vocal distinctiveness, measures of repeatabilitycan be used to quantify the temporal stability of acoustic properties within

48 M.A. Bee et al.

ARTICLE IN PRESS

Box 3 Statistical Methods to Describe Individual Distinctivenessin Vocalizations (cont'd )

individuals recorded multiple times on different occasions (Bee & Gerhardt,2001c; Pettitt et al., 2013).

5. Multivariate approaches. Animal signals consist of multiple properties thatare often intercorrelated, but researchers rarely know a priori which proper-ties are relevant to receivers. Therefore signal analyses often use multivariatestatistical techniques to consider simultaneous variation in multipleproperties.a. Discriminant function analysis (DFA). One commonly used multivariate

technique is DFA, which creates canonical discriminant functions thatmaximally separate individuals in multivariate signal space (Nelson &Marler, 1990). Discriminant functions represent linear combinations ofthe original acoustic properties. Interpreting factor loadings revealshow acoustic properties contribute to the statistical discriminationamong individuals. Discriminant functions can also be used to classify in-dividual calls as belonging to a specific individual. Classification successrepresents one measure of overall individual distinctiveness because itquantifies how well calls can be statistically assigned to individuals. How-ever, classification success is highly dependent on sample size, makingcomparative applications of this measure problematic (Bee & Gerhardt,2001c; Bee et al., 2001; Beecher, 1989). Including more individuals in ananalysis decreases classification success due to increased probabilitythat individuals in the sample have similar call properties. Analyses ofthe calls of large samples of individuals potentially overestimate the scaleof the recognition problem faced by territorial residents because territoryholders rarely have more than a few adjacent neighbors. For this reason,the classification success for smaller, more biologically relevant groupsizes should also be considered (Bee & Gerhardt, 2001c; Bee et al.,2001; Pettitt et al., 2013).

b. Principal components analysis (PCA). Signal properties are often inter-correlated. Treating correlated properties as independent can result inoverestimates of the individual distinctiveness of signals (Beecher,1989). One solution to this problem is using PCA to generate orthogonalvariables that can be used as inputs into other analyses such as a DFA(Bee, 2004; Bee et al., 2010; Pettitt et al., 2013). To understand how theoriginal call properties contribute to successful classification in a DFA af-ter transformation using PCA requires interpreting both the factor load-ings of the original variables on principal components and then thefactor loadings of these transformed variables on discriminant functions.

c. Information content. Using information theory (Shannon & Weaver,1949), Beecher (1989) derived a method for quantifying the identity infor-mation capacity of animal signals (Hs). This statistic expresses a signal’s

(Continued)


ARTICLE IN PRESS

Box 3 Statistical Methods to Describe Individual Distinctivenessin Vocalizations (cont'd )

ability to reduce uncertainty about the identity of the signaler, and isparticularly useful for comparative studies because it is both dimension-less and independent of the particular conditions of the sample (eg,sample size, signaling modality). No studies of anuran vocalizationshave reported this statistic, but it is easily derived from the results ofmodel II ANOVAs of PCA-transformed call properties.

Figure 9 Spectrograms and oscillograms of (A) a single advertisement call of the NorthAmerican bullfrog, Rana catesbeiana, recorded in Boone County, Missouri (M. A. Bee,unpublished data), and (B) three consecutive advertisement calls of the golden rocketfrog, Anomaloglossus beebei, recorded in Kaieteur National Park, Guyana (J. Tumulty andM. A. Bee, unpublished data).

50 M.A. Bee et al.

ARTICLE IN PRESS


ARTICLE IN PRESS

frequency near 100 Hz and a harmonically rich, bimodal frequency spec-trum. Each mode is well matched to the tuning of one of the two innerear organs in frogs (Simmons, 2013). The dominant frequency is the secondharmonic and is near 200 Hz. Both the fundamental and the dominant fre-quency are strongly, negatively correlated with male body length. Haas(1976) published an abstract reporting individual differences in bullfrogcalls, but few details were provided. Bee and Gerhardt (2001c) investigatedthe individual distinctiveness of bullfrog advertisement calls by examiningpatterns of among-individual and within-individual variation in 10 acousticproperties analyzed for each of nearly 1100 calls recorded from a sample of27 territorial males. Linear regression analyses were used to remove the ef-fects of temperature on call properties while preserving the influences ofother important sources of individual differences, such as body size (seeBox 3). All 10 properties varied significantly among individuals and werehighly repeatable within individuals across multiple nights. Fundamentalfrequency and the highly correlated property of dominant frequency hadthe lowest coefficients of variation within individuals (CVw ¼ 1.5%), thehighest ratios of among-individual to within-individual coefficients of vari-ation (CVa/CVw ¼ 4.7), and the highest repeatabilities (intraclasscorrelation ¼ 0.93).

In a discriminant function analysis (DFA), greater than 72% of calls wereassigned to the correct individual, which represents a classification successsignificantly higher than expected by chance (3.7%). Inspection of thecanonical structure revealed strong correlations between fundamental fre-quency (or dominant frequency) and the first canonical root, whichaccounted for some 70e80% of the variability among individuals acrossdifferent analyses. Although these analyses were performed on the entiresample of 27 individuals, territorial male bullfrogs rarely have more than 2or 3 neighbors. To evaluate model performance using more biologicallyrelevant sample sizes, Bee and Gerhardt (2001c) determined classificationsuccess for 1000 random samples of 5 individuals drawn from the sampleof 27 males. In these analyses, classification success typically ranged between91% and 95% correct (Fig. 10).

In a follow-up study, Bee (2004) investigated the patterns of individualvariation within the multiple-note advertisement calls males most commonlyproduce. The goal was to assess whether patterns of within-call variationmight either constrain or provide additional cues for recognition. Althoughsix of eight acoustic properties exhibited significant within-call variation, alleight call properties still varied significantly more among individuals. Again,

Figure 10 The properties of advertisement calls of male bullfrogs in two-dimensionalsignal space defined by the first two discriminant functions from a discriminant func-tion analysis. Points represent means and ellipses represent 95% confidence intervalsfor each individual. The top left plot shows an analysis based on a sample size of 27males while the five additional plots show random subsamples of five individualseach. Reprinted from Bee, M. A., & Gerhardt, H. C. (2001c). Neighbour-stranger discrimi-nation by territorial male bullfrogs (Rana catesbeiana): I. Acoustic basis. Animal Behav-iour, 62(6), 1129e1140 with permission from Elsevier.

52 M.A. Bee et al.

ARTICLE IN PRESS

fundamental frequency (and dominant frequency) contributed most towardstatistically discriminating among individuals. In addition, not only were theacoustic properties of notes individually distinct, so too were the differencesamong the notes within the calls an individual produced. Thus individuallydistinctive patterns of within-individual variation might also function inrecognition.

3.3.2 Golden Rocket FrogsLike bullfrogs, territorial males of the golden rocket frog, A. beebei, alsoexhibit the dear enemy effect and discriminate between the calls of neigh-bors and strangers (Section 3.2.2). The advertisement calls of these two spe-cies are quite different (cf. Fig. 9A and B). In contrast to the slow, sonorousnotes of the bullfrog, male golden rocket frogs produce high-pitched adver-tisement calls consisting of three or four short, rapidly repeated pulses havingan average dominant frequency of 5.4 kHz (Fig. 9B; Pettitt, Bourne, & Bee,


ARTICLE IN PRESS

2012, 2013). Pettitt, Bourne, and Bee (2013) analyzed 8 call properties froma sample of 760 advertisement calls recorded from 40 males. Prior to statis-tical analyses, call properties significantly correlated with temperature werecorrected by adjusting their values to a common temperature of 24�C(near the mean temperature of all recordings). All properties varied signifi-cantly among individuals, and the properties of dominant frequency andpulse rate had the highest CVa/CVw ratios (3.2 and 2.0, respectively). Fora subset of 16 males recorded on two different occasions, intraclass correla-tion coefficients ranged between 0.46 and 0.65 and were highest for pulseduration (0.65) and pulse interval (0.60), indicating a moderate to high de-gree of repeatability within individuals. A DFA based on principal compo-nent scores assigned 79% of calls to the correct individual for the full data set.Higher classification success was observed for smaller group sizes chosen tobetter reflect the maximum number of neighbors a territorial male is likelyto have (eg, 97% for random groups of six males). In contrast to bullfrogs, inwhich the fundamental and dominant frequencies were the most individu-ally distinctive call properties (Bee, 2004; Bee & Gerhardt, 2001c), it was thefine temporal properties of pulse rate, pulse interval, and pulse duration thatcontributed the most toward statistically discriminating among males of thegolden rocket frog (Pettitt et al., 2013). Furthermore, the high frequenciesof the calls of golden rocket frogs are almost certainly encoded by the basilarpapilla, one of the two inner ear sensory papillae in anurans. This papillacannot encode differences in frequency independently of differences in level(reviewed in Gerhardt & Huber, 2002), potentially rendering individual dif-ferences in call frequency of limited use in neighborestranger discrimina-tion. Based on these results, Pettitt et al. (2013) predicted that malegolden rocket frogs discriminate between neighbors and strangers basedon individual differences in temporal call properties. So far, this predictionremains to be tested. If this prediction is confirmed in future work, thengolden rocket frogs and bullfrogs would represent an instance of conver-gence in overt behavioral discrimination and divergence in the underlyingperceptual discrimination.

3.3.3 Agile FrogsAs discussed previously (Section 3.2.3), male agile frogs, R. dalmatina,reportedly discriminate between familiar and unfamiliar calls. Lesbarr�eresand Lodé (2002) report results from acoustical and statistical analyses ofadvertisement calls, which consist of a series of approximately 25 pulsegroups, composed of 3e12 pulses each (Geisselmann, Flindt, & Hemmer,

54 M.A. Bee et al.

ARTICLE IN PRESS

1971; Lesbarr�eres & Lodé, 2002; Schneider et al., 1988). Four acousticproperties were measured: call duration, the number of pulse groupsper call, the ratio of call duration/pulse groups per call, and fundamentalfrequency. An undisclosed number of calls were analyzed based onrecording each of 67 males for 1 min. It is unclear whether call propertieswere corrected for variation introduced by variation in temperature at thetime recordings were made. To quantify individual variation, the authorsreport a CV for each call property, which was higher for call duration(54%) and the number of pulse groups (52%) compared with the ratio ofthese two properties (22%) and fundamental frequency (9%). Unfortu-nately, however, the authors did not specify whether these values corre-spond to within-individual (CVw) or among-individual (CVa) measures,making interpretation of their results in the context of individual vocaldistinctiveness impossible. No other measures of individual distinctivenesswere reported.

3.3.4 Strawberry Poison FrogsRecall that males of the strawberry poison frog, O. pumilio, exhibit long-term defense of multipurpose territories and aggressively interact withterritorial intruders but did not behaviorally discriminate between the callsof neighbors and strangers in playback tests (Section 3.2.4). Pr€ohl (2003)investigated patterns of individual variation in advertisement calls in twoCosta Rican populations of this species by recording and analyzing a sampleof 492 call sequences from 40 territorial males. Ten calls from each callsequence were analyzed and one spectral property (dominant frequency)and four temporal properties (call rate, call duration, pulse rate, and dutycycle) were analyzed. Properties that were significantly correlated withtemperature were adjusted to a common temperature of 24�C. All fivecall properties varied more among individuals than within individuals.Consistent with these outcomes, all of the CVa/CVw ratios were greaterthan 1.0, ranging between 1.2 (for call rate) and 1.7 (for pulse rate). Acrossrecording sessions of the same individual, all five call properties alsoexhibited moderate repeatabilities with values ranging between 0.23 (forcall rate) and 0.56 (for pulse rate). This result indicated that individualdifferences in calls were stable over time. Multivariate analyses such asdiscriminant function analyses were not used to estimate classification suc-cess or determine which call properties contributed most toward individualdistinctiveness. Nevertheless, Pr€ohl’s (2003) analyses suggest the temporalproperties of call rate (CVa/CVw ¼ 1.7, r ¼ 0.56) and call duration


ARTICLE IN PRESS

(CVa/CVw ¼ 1.5, r ¼ 0.50) may contribute more to individual distinctive-ness than the spectral property of dominant frequency (CVa/CVw ¼ 1.4,r ¼ 0.39).

3.3.5 Concave-Eared Torrent FrogsThe concave-eared torrent frog, O. tormota, has an exceptionally complexcall repertoire, consisting of eight different call types with unusually highfundamental frequency and substantial fluctuations in frequency within calls(Feng et al., 2002; Feng, Riede, et al., 2009; Narins et al., 2004). Feng,Riede, et al. (2009) examined individual variation in 111 calls recordedfrom a sample of 6 males, focusing on the call type given most frequently(“long calls”). Most acoustic properties varied significantly more among in-dividuals than within individuals. A DFA obtained a mean classification suc-cess of 55%. While this value was above the chance level of correctclassification (16.7%, or 1/6), it is lower than values reported in other studiesof frogs, which have used larger sample sizes and, therefore, should have hadlower, not higher, classification success (Bee et al., 2001; Beecher, 1989).Feng, Riede, et al. (2009) did not present information on which acousticproperties contributed most to the discriminant functions, but maximumfundamental frequency had the highest CVa/CVw ratio at 1.0. These resultssuggest that the calls ofO. tormota are much less-reliable cues of identity thanthe calls of males of other anuran species, probably due to the substantialvariation in acoustic properties that occurs within individuals. Given theseresults and their large repertoire of calls, we might have expected thatneighborestranger discrimination in this species would be too difficult forreceivers (Kroodsma, 1976). Nonetheless, males of this species have been re-ported to discriminate between the calls of familiar and unfamiliar males(Feng, Arch, et al., 2009), although this has not yet been confirmed usingfield playback experiments (Section 3.2.5).

3.3.6 Studies of Individual Vocal Distinctiveness in Other TerritorialSpecies

Green frogs, R. clamitans, are close relatives of bullfrogs, and the two specieshave very similar territorial behavior and mating systems (Section 2.3.2). Beeet al. (2001) investigated individual vocal distinctiveness in the advertisementcalls of male green frogs to examine the potential for these signals to mediateneighbor recognition. All nine of the acoustic properties examinedexhibited significant among-individual variation. Similar to bullfrogs, funda-mental frequency and dominant frequency exhibited low variation within

56 M.A. Bee et al.

ARTICLE IN PRESS

males and had the highest CVa/CVw ratios. A DFA assigned between 52%and 61% of calls to the correct individual for the full data set of 198 callsrecorded from 20 individuals, but classification success was much higherfor smaller group sizes that better reflected the magnitude of the potentialrecognition problem facing territorial males (eg, 82% and 96% for groupsof five and two individuals, respectively). Fundamental and dominant fre-quency contributed most toward statistically discriminating among individ-ual males.

The brilliant-thighed poison frog, Allobates femoralis (Dendrobatidae), isfound throughout Amazonia. Males of this species defend relatively stablebreeding territories, which include terrestrial oviposition sites, for periodsup to several months (Gasser, Amézquita, & H€odl, 2009; Ringler, Ringler,Maga~na Mendoza, & H€odl, 2011; Ringler, Ursprung, & H€odl, 2009;Roithmair, 1992, 1994). Males also aggressively defend these territoriesagainst conspecific males and show rapid aggressive approaches to playbacksof advertisement calls within their territories (H€odl, Amézquita, & Narins,2004; Ringler et al., 2011; Ursprung, Ringler, & H€odl, 2009). Thus malesof this species might benefit from discriminating between familiar neighborsand strangers on the basis of advertisement calls (Temeles, 1994). Gasseret al. (2009) analyzed 285 calls from a sample of 19 males. Nearly all acousticproperties varied significantly more among individuals than within individ-uals. Mean classification success from a DFA was 65%, with a higher successof 88% for smaller groups of five males. Similar to the other dendrobatidspecies described in Sections 3.3.2 and 3.3.4, temporal call features (noterepetition rate and note duration in particular) were the most individuallydistinctive. Given their long-term defense of breeding territories andindividually distinctive vocalizations, tests of the dear enemy effect in thisspecies would be worthwhile.

3.3.7 Studies of Individual Vocal Distinctiveness in Lek-BreedingSpecies

An important question is whether the benefit of being recognized by terri-torial neighbors, in terms of receiving reduced aggression, has resulted in se-lection for greater individual distinctiveness in the vocalizations of territorialanuran species (the “signature adaptation” hypothesis; Beecher, 1989, 1991;Tibbetts & Dale, 2007). If so, we should expect species defending long-termbreeding or multipurpose territories to have more individually distinctivecalls than those that do not defend such territories. To date, no such directcomparison has been made. However, studies of lek-breeding species, which


ARTICLE IN PRESS

typically exhibit short-term defense of calling sites that lack breedingresources, suggest individual vocal distinctiveness is widespread among an-urans. Treefrogs in the family Hylidae often exhibit mating systems bestcharacterized as leks. Nevertheless, studies of two species of hylid treefrogssuggest their calls are individually distinctive. Boreal chorus frogs, Pseudacrismaculata, are a common and widespread species in North America. Bee et al.(2010) investigated the individual distinctiveness of their pulsatile advertise-ment calls by measuring or calculating 26 call properties for 20 advertise-ment calls each from 36 individual males. All call properties variedsignificantly among individuals. In a DFA of call properties following aprincipal components transformation to orthogonal variables, an averageof 91% of calls were assigned to the correct individual. Similar to someterritorial ranid species, the first two discriminant functions loaded mostheavily on fundamental and dominant frequency. In an earlier studydand indeed the first to analyze individual variation in anuran calls in thecontext of individual recognitiondShy (1985) reported that males of theMiddle East treefrog, Hyla savignyi produce individually distinctive calls.This study measured nine acoustic properties, all of which were reportedto vary significantly more among individuals than within individuals.Unfortunately, however, Shy (1985) did not control for temperature,which is known to strongly affect the acoustic properties of anuran calls(see Box 3).

3.4 Perceptual Basis of Neighbor RecognitionCompared with the number of studies documenting individual distinctive-ness in anuran vocalizations and even the number of studies of neighborrecognition in anurans, fewer studies have investigated the perceptual basisof rival recognition in this group. Indeed, how anurans perceptually discrim-inate between neighbors and strangers has been investigated in only onespecies.

3.4.1 North American BullfrogsTerritorial male bullfrogs exhibit the dear enemy effect (Section 3.2.1) andproduce individually distinctive advertisement calls (Section 3.3.1). In a se-ries of field playback experiments, Bee and Gerhardt (2001a, 2001b, 2002)used the habituationediscrimination paradigm (Halpin, 1986) to investigatethe perceptual basis of neighborestranger discrimination in this species.During an initial habituation phase, a synthetic advertisement call was repeat-edly broadcast to a territorial male from a previously unoccupied location

58 M.A. Bee et al.

ARTICLE IN PRESS

adjacent to its territory. The idea was to simulate the arrival of a newneighbor. Subjects initially responded aggressively to these playbacks by giv-ing advertisement and aggressive calls and approaching the speaker usingconspicuous splash displays. As repeated playbacks continued over the courseof the habituation phase, subjects eventually stopped responding aggressivelyand remained in their territory producing only advertisement calls. At thisstage of the experiment, the subject’s aggressive response had habituated(Section 3.5). The interpretation of this behavioral change was that the sub-ject had become familiar with the acoustic stimulus and no longer perceivedit as a threat. Once subjects had either reached a predetermined responsedecrement criterion (Bee & Gerhardt, 2001a, 2001b) or heard a predeter-mined number of stimuli (Bee & Gerhardt, 2002), a single acoustic propertyof the habituating stimulus was changed and presented as a novel stimulusduring a subsequent discrimination phase. The use of synthetic calls was keyto manipulating one property (and physically correlated properties) at atime in order to test hypotheses about specific call properties that mightfunction as acoustic recognition cues.

The primary acoustic cue manipulated in these studies was fundamentalfrequency. This property was chosen for investigation for the three reasonsnoted earlier. It contributed most toward individual distinctiveness (Bee &Gerhardt, 2001c) with approximately 90% of the variation in this propertyattributable to among-individual differences (Bee, 2004). Periodicity cuesrelated to the time-domain expression of fundamental frequency and corre-lated spectral properties like dominant frequency propagate well in both airand water over the distances that typically separate neighboring bullfrogs(Boatright-Horowitz, Cheney, & Simmons, 1999). Lastly, the bullfrog’sauditory nerve robustly encodes fundamental frequency or low-frequencyharmonics (eg, the dominant frequency) in the timing of spike dischargesunder a range of signal-to-noise conditions (Schwartz & Simmons, 1990;Simmons, Reese, & Ferragamo, 1993; Simmons, Schwartz, & Ferragamo,1992).

The distributions of within-individual and among-individual differencesin fundamental frequency were compared to make predictions about thelikely “just-meaningful difference” (JMD; Bee, 2004) required for discrim-ination. In contrast to the more common measure of perceptual discrimina-tion thresholds in psychophysical studies known as the “just-noticeabledifference” (JND), the JMD describes a receiver’s behavioral discriminationthreshold (Gerhardt, 1992; Nelson & Marler, 1990). It represents the min-imum difference between two stimuli required to elicit a difference in


ARTICLE IN PRESS

behavioral response, and thus reflects a receiver’s decision rule aboutwhether or not a perceived difference between stimuli warrants a differencein response. The most commonmagnitudes of among-individual differencesin fundamental frequency were on the order of 5% or larger, whereaswithin-individual differences were typically less than 5% (Fig. 11A). Basedon these patterns of variation, the JMD for fundamental frequency andcorrelated spectral properties was predicted to be about 5%. The reasoningbehind this prediction was that differences smaller than about 5% wouldmore likely reflect the variation that typically occurs within an individual,whereas differences larger than about 5% would more likely represent adifferent individual. Therefore, the predicted outcome of habituationediscrimination playback tests was that changes in the fundamental frequencyof a novel stimulus on the order of 5% or more during the discriminationphase would produce a recovery of habituated aggressive responses, whereasdifferences less than 5% should be treated as the same individual. The datawere largely consistent with these expectations.

During the discrimination phase, when changes in the fundamental fre-quency of the novel stimulus were sufficiently large, males resumed givingaggressive calls and approached the playback speaker. When very large dif-ferences (eg, 24e32%) in fundamental frequency were imposed on novelstimuli, habituated aggressive responses recovered nearly 100% to levelsobserved prior to habituation (Bee & Gerhardt, 2002). These differencesin frequency reflected the most extreme individual differences likely to beencountered between two males in the population. Smaller frequencydifferences on the order of 5e10%, which were also typical of among-individual differences, also elicited recovery of habituated aggression, albeitat lower levels (Fig. 11B and D; Bee & Gerhardt, 2001a, 2001b). When thechange in fundamental frequency was just 2%, a difference typical of within-individual variation, responses were similar to those of subjects in a0%-change control group that heard additional presentations of the habitu-ating stimulus (Bee & Gerhardt, 2001a). These data, thus, indicated that ter-ritorial residents could use individual differences in fundamental frequency asa potential cue for discriminating among neighbors and strangers.

Manipulations of other spectral and temporal call properties in novelstimuli generally failed to elicit recovery of aggressive responses during thediscrimination phase. For example, the responses of subjects that heard10% changes in either note duration, note onset and offset times, or noteduty cycle during the discrimination phase were similar to those in responseto 0%-change controls and exhibited little to no recovery of habituated

Figure 11 (A) Distribution of percentage differences in fundamental frequency fromwithin- and among-individual pairwise comparisons of advertisement calls of malebullfrogs. Most among-individual differences are 5% or greater, while most within-individual differences are smaller than 5%, leading to a predicted just-meaningfuldifference (JMD) of approximately 5%. (Redrawn from Fig. 8 in Bee, M. A. (2004).Within-individual variation in bullfrog vocalizations: Implications for a vocally mediatedsocial recognition system. The Journal of the Acoustical Society of America, 116(6),3770e3781.) (B, C, and D) Aggressive responses of male bullfrogs (number of aggressivecalls, approach distance toward the speaker, number of movements, or the proportionof males responding aggressively) during the discrimination phases of habituationediscrimination experiments. The novel stimuli in discrimination phases (B) differedfrom the habituating stimulus by a 10% change in fundamental frequency, (C) wasidentical to the habituating stimulus but broadcast from the opposite direction relativeto the subject’s territory, or (D) differed by various acoustic properties (a 2%, 5%, or 10%

60 M.A. Bee et al.

ARTICLE IN PRESS


ARTICLE IN PRESS

aggression (Fig. 11D). The 10% change implemented in most of these tem-poral properties was either the same as or smaller than the measuredmaximum CVw and the estimated JMD for the same property (Bee,2004; Bee & Gerhardt, 2001a). Hence changes on the order of 10% failedto exceed the predicted JMD required to elicit a response during thediscrimination phase. Changes to the harmonic fine structure of the call,created by changing the presence/absence and relative amplitudes of har-monics in novel stimuli, also failed to elicit recovery of aggressive responses.There was some indication that a 20% change in note duty cycle, whichexceeded the maximum CVw, elicited recovery of aggression, suggestingthat sufficiently large differences in temporal properties might also functionin discriminating between neighbors and strangers.

Recall that Davis (1987) also showed that in bullfrogs, as in songbirds,lower levels of aggression are only exhibited toward familiar neighborswhen those neighbors are heard vocalizing from the direction of their usualterritory (Section 3.2.1). Familiar sounds broadcast from new locations eli-cited more aggression than familiar sounds broadcast from normal locations.In another series of habituationediscrimination tests, Bee and Gerhardt(2001b, 2002) tested the hypothesis that presenting the same habituatingstimulus from a novel location during the discrimination phase would elicitrecovery of aggression. The data were consistent with this hypothesis.When habituated males heard the same sound from a new location, theyresponded with aggressive calls and approached the speaker in the newposition (Fig. 11C).

3.5 Acquisition of Differential Responses to Neighbors andStrangers

Habituation is commonly suggested as a form of learning that could under-lie, at least partially, the reduced aggressive responses territory holders

=---------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------change in fundamental frequency, harmonic fine structure, note duration, rise and falltimes of individual notes (RT/FT), and a 10% and 20% change in note duty cycle (DC)).The response components in B and C are the median, IQ range, and range of aggressiveresponses during the discrimination phase expressed as the percentage of the male’sinitial aggression during the beginning of the habituation phase. Responses in D are theproportion of males responding aggressively. Redrawn from Figs 2 and 3 in Bee, M. A., &Gerhardt, H. C. (2001b). Neighbor-stranger discrimination by territorial male bullfrogs(Rana catesbeiana): II. Perceptual basis. Animal Behaviour, 62, 1141e1150, and Fig. 6 inBee, M. A., & Gerhardt, H. C. (2001a). Habituation as a mechanism of reduced aggressionbetween neighboring territorial male bullfrogs (Rana catesbeiana). Journal of Compara-tive Psychology, 115(1), 68e82.

62 M.A. Bee et al.

ARTICLE IN PRESS

exhibit toward their neighbors (Brooks & Falls, 1975; Shettleworth, 2009;Wiley &Wiley, 1977). It refers to “a behavioral response decrement that re-sults from repeated stimulation and that does not involve sensory adapta-tion/sensory fatigue or motor fatigue” (Rankin et al., 2009; p. 136). Theextent to which response decrements develop and are maintained overtime is variable, and differences between short-term and long-term formsof habituation are well established (Sharpless & Jasper, 1956). The hypothesisthat habituation learning is involved in the dear enemy effect has receivedconsiderable support from a number of experimental studies in fish(reviewed in Peeke & Peeke, 1973; Peeke, 1984) and songbirds (reviewedin Petrinovich, 1984). In this section, we review studies that have investi-gated habituation of aggression in anurans. Together, evidence from thesestudies of fish, songbirds, and anurans indicates that repeated exposure to aneighbor or its communication signals not only results in long-term decre-ments in aggressive behavior, but also leads to the formation of enduring andstimulus-specific neuronal representations that encode information aboutindividually distinctive signal properties.

3.5.1 North American BullfrogsIn a seminal paper on habituation learning, Thompson and Spencer (1966)operationally defined habituation by outlining nine parametric characteris-tics of changes in response that occur with repeated stimulation. While othercharacteristics have been discussed in the literature (Siddle, 1990; Sokolov,1963; Wagner, 1976; Whitlow & Wagner, 1984), there is generally wide-spread agreement among neuroscientists and behavioral scientists that thecharacteristics outlined by Thompson and Spencer (1966) describe habitua-tion of responses that involve simple reflexes as well as more complex behav-iors of the whole organism (Hinde, 1970; Macphail, 1993; Petrinovich,1984; Rankin et al., 2009; Shettleworth, 2009). A review on habituationlearning has reaffirmed and expanded Thomson and Spencer’s (1966) orig-inal nine characteristics, adding a tenth specifically related to long-termhabituation (Rankin et al., 2009). A series of studies by Bee and Gerhardt(2001a, 2001b, 2002) and Bee (2001, 2003b) investigated the extent towhich territorial aggression in bullfrogs exhibits a subset of 8 of the 10 para-metric characteristics enumerated by Rankin et al. (2009) and quoted herefrom that review.

Repeated application of a stimulus results in a progressive decrease in someparameter of a response to an asymptotic level.


ARTICLE IN PRESS

As noted earlier, male bullfrogs respond aggressively toward acousticallysimulated intrusions into their territories using a combination of advertisementcalls, aggressive calls, splash displays, and approaches toward the speaker (Bee,2002; Davis, 1987; Wiewandt, 1969). With repeated broadcasts of calls simu-lating the arrival of a new neighbor, these aggressive responses exhibitedmarked decrements that generally followed a negative exponential function(Fig. 12; Bee, 2001, 2003b; Bee & Gerhardt, 2001a, 2001b, 2002). Inresponse to repeated stimuli, the numbers of aggressive calls, splash move-ments, and meters moved toward the speaker decreased while the subject’sresponse latency increased. These changes in response were not particularlyrapid. It was not uncommon, for example, for territorial males to continueresponding aggressively to a speaker for 3e6 h, with one particularly obsti-nate male continuing to give aggressive calls to repeated stimuli for 10 consec-utive hours (Bee, 2001). With the exception of advertisement calls, othermeasures of aggression decreased in magnitude at the same rate (Fig. 12A;Bee, 2003b). This finding is noteworthy because it suggests that the changesin neural response that must take place with repeated stimulation occur at alocus in the brain that can feed forward into multiple motor circuits control-ling different behaviors. The situation for advertisement calls was somewhatdifferent from the more overt aggressive responses quantified because adver-tisement call rate increases steadily during a night, reaching its peak between0100 and 0300 h, and then declines steadily until dawn.

If the stimulus is withheld after response decrement, the response recovers at leastpartially over the observation time (“spontaneous recovery”).

After multiple series of stimulus repetitions and spontaneous recoveries, theresponse decrement becomes successively more rapid and/or more pronounced(this phenomenon can be called potentiation of habituation).

Some stimulus repetition protocols may result in properties of the response decre-ment.that last hours, days or weeks.

Bee and Gerhardt (2001a) investigated these three parametric character-istics of habituation by testing males across multiple nights (Fig. 12B). Sub-jects were exposed to 5 h of repeated playbacks of the calls of a newneighbor on each of four consecutive nights. Spontaneous recovery wasexamined by comparing responses from the end of habituation trainingon one night to that at the beginning of habituation training on thefollowing night. There was clear evidence that habituated aggressive re-sponses increased during the 19 intervening hours over which the stimulus

Figure 12 Habituation of aggressive response in male bullfrogs to playbacks of syn-thetic advertisement calls. (A) Mean � SE number of advertisement calls, number ofaggressive calls, approach distance, response latency, and number of aggressive move-ments for each stimulus period. Each stimulus period lasted approximately 7.5 min.(Redrawn from Fig. 4 in Bee, M. A. (2003b). Experience-based plasticity of acousticallyevoked aggression in a territorial frog. Journal of Comparative Physiology A, 189(6),485e496.) (B) Mean and interquartile range of an index of aggression over four nightsof playback. (Redrawn after expressing as an aggressive index data presented in Bee, M. A.,& Gerhardt, H. C. (2001a). Habituation as a mechanism of reduced aggression betweenneighboring territorial male bullfrogs (Rana catesbeiana). Journal of Comparative Psy-chology, 115(1), 68e82.)

64 M.A. Bee et al.

ARTICLE IN PRESS


ARTICLE IN PRESS

was withheld. However, responses did not increase to the same levels atwhich they had started the preceding night, and responses decreased tolow, asymptotic levels more quickly following the first night (Fig. 12B).In other words, habituation became successively more rapid with repeatedseries of habituation training across nights, thus providing evidence forboth spontaneous recovery and the potentiation of habituation. By thefourth night, subjects barely responded at all to playbacks. This outcomewas noteworthy because Haas (1977), based on field observations, hadearlier reported that territorial male bullfrogs reacted aggressively towardnewly established neighbors for a period of 1e4 days, a result quite consis-tent with results from playback experiments. Discrimination tests conductedon the fifth consecutive night showed that the long-term decrementsobserved over the first four nights of habituation training were specific tothe fundamental frequency of the stimulus (Bee & Gerhardt, 2001a).

Other things being equal, more frequent stimulation results in more rapid and/ormore pronounced response decrement, and more rapid spontaneous recovery (ifthe decrement has reached asymptotic levels).

Within a stimulus modality, the less intense the stimulus, the more rapid and/ormore pronounced the behavioral response decrement. Very intense stimuli mayyield no significant observable response decrement.

In another series of playback experiments, Bee (2001) tested the pre-dicted effects of stimulus repetition rate and stimulus intensity on the rateof habituation of aggression. The stimulus was a series of five 5-note callsseparated by 30-s intercall intervals. Stimulus repetition rate and stimulusintensity were manipulated in a 2 � 2 factorial design. Stimulus repetitionrate (fast versus slow) was varied by manipulating the interval betweenconsecutive groups of five calls. Stimulus intensity (high versus low) wasvaried by manipulating the SPL of the playback stimulus. Consistent withexpectations from Thompson and Spencer (1966), response decrementswere more rapid when the stimulus repetition rate was fast and the stimulusintensity was low. The slowest response decrements occurred in the condi-tion in which stimuli were presented at the slow repetition rate and highintensity.

Within the same stimulus modality, the response decrement shows some stimulusspecificity.

Stimulus specificity and stimulus generalization represent ends of a con-tinuum commonly referred to as a stimulus generalization gradient. In their

66 M.A. Bee et al.

ARTICLE IN PRESS

original formulation of this characteristic, Thompson and Spencer (1966)emphasized stimulus generalization in response to novel stimuli. In theirmore recent review, Rankin et al. (2009) shifted the focus of this character-istic to stimulus specificity, thereby highlighting that recovery of habituationin response to a novel stimulus in the same modality can rule out sensoryadaptation and motor fatigue as explanations for the response decrement.As already discussed in Section 3.4.1, playbacks of novel stimuli differingin fundamental frequency or location during a discrimination phasefollowing habituation training resulted in recovery of habituated aggressiveresponses. These results thus demonstrated stimulus specificity and ruled outsensory adaptation and motor fatigue. In contrast, the presentation of othernovel stimuli differing in temporal properties and harmonic fine structureduring the discrimination phase failed to elicit recovery of habituated aggres-sive responses (Bee & Gerhardt, 2001a). Because the magnitude of the dif-ferences imposed on temporal properties and harmonic fine structure almostcertainly exceeded the relevant JNDs (although perhaps not the JMDs), thesedata can be interpreted as evidence for stimulus generalization, in that ani-mals responded to the habituating and novel stimuli similarly followinghabituation training.

Presentation of a different stimulus results in an increase of the decrementedresponse to the original stimulus. This phenomenon is termed “dishabituation.”

Sometimes the recovery of a habituated response elicited as a response topresentation of a novel stimulus is referred to as dishabituation. This is incor-rect. While the response to the novel stimulus demonstrates the stimulusspecificity of the response decrement, it is not to be confused with dishabi-tuation. Instead, dishabituation refers to a reinstatement of a response to sub-sequent presentations of the habituating stimulus following presentation ofsome other, usually strong, stimulus. Thompson and Spencer (1966) referto this strong stimulus as a “dishabituatory stimulus.” According to thedual-process theory of habituation, dishabituation reflects the process ofsensitization (Groves & Thompson, 1970). Bee and Gerhardt (2001a) exam-ined dishabituation of aggression in bullfrogs by examining the extent towhich presentation of a dishabituatory stimulus differing in fundamental fre-quency was able to elicit renewed aggression to subsequent presentations ofthe original habituating stimulus. Compared with subjects in a controlgroup, males that were habituated when they heard a novel stimulusdiffering in fundamental frequency were more likely to give aggressiveresponses during subsequent presentations of the habituating stimulus.


ARTICLE IN PRESS

Thus brief exposures to a stimulus simulating an individual different fromthat simulated by the habituating stimulus caused dishabituation.

3.5.2 Studies of Habituation in Other Territorial SpeciesAs noted in previous sections, green frogs, R. clamitans, are close relatives ofbullfrogs and exhibit similar territorial and aggressive behaviors (Sections2.3.2 and 3.3.6). At present, they represent the only other territorial frogspecies in which habituation of aggression has been investigated. Male greenfrogs defend relatively stable territories in ponds, sometimes for severalweeks (Martof, 1953; Wells, 1977a, 1978), and fundamental and dominantfrequency contribute most toward individual vocal distinctiveness (Beeet al., 2001; Section 3.3.6). Given these similarities to bullfrogs, Owenand Perrill (1998) hypothesized that male green frogs likely also learn torecognize their neighbors’ calls as a result of habituation. They tested this hy-pothesis using the habituationediscrimination paradigm to demonstrate twoof Thompson and Spencer (1966) parametric characteristics of habituation(response decrement and spontaneous recovery). During the habituationphase, males were initially aggressive to broadcasts of synthetic calls butshowed a negative exponential decrement in their aggressive response afterapproximately 1 h. Aggressive responses recovered spontaneously after ashort (15 min) silent period. While this study of green frogs provides clearevidence of habituation of aggression, it has been criticized for attemptingto draw a causal relationship between the short-term response changesobserved and the dear enemy effect, which has not been demonstrated tooccur in this species (Bee & Schachtman, 2000).

3.5.3 Studies of Habituation in Lek-Breeding SpeciesHabituation of aggression has also been investigated in three lek-breedingtreefrog species (Hylidae) as a mechanism that allows males to track changesin the local density of calling males in a chorus. The Pacific treefrog, Pseu-dacris regilla, forms dense choruses, but males maintain nonrandom spacingfrom other calling males in the chorus (Whitney, 1980; Whitney & Krebs,1975). Males produce distinct aggressive calls (Allan, 1973; Whitney, 1980)when the amplitude of the calls of neighboring males exceeds their “aggres-sive thresholds” (Brenowitz, 1989; Rose & Brenowitz, 1991). These thresh-olds are positively correlated with the amplitude of their nearest neighbors’calls (Rose & Brenowitz, 1991). Brenowitz and Rose (1994) tested the hy-pothesis that aggressive thresholds are plastic and fluctuate based on dynamicchanges in a male’s local social environment. By repeatedly broadcasting

68 M.A. Bee et al.

ARTICLE IN PRESS

advertisement calls at an amplitude that exceeded a focal male’s previouslydetermined aggressive threshold, they were able to raise a male’s aggressivethreshold. The aggressive responses elicited in the experiments of Brenowitzand Rose (1994) and Rose and Brenowitz (1997) showed the characteristicsof habituation: focal males were initially aggressive, but responses decreasedexponentially over a relatively short period of time, until after approximately2 min males produced almost exclusively advertisement calls (Fig. 13A). Torule out sensory adaptation and effector fatigue, Brenowitz and Rose (1994)showed that the response decrement was stimulus specific; aggressive

Figure 13 Habituation of aggressive calls in (A) Pacific treefrogs and (B) spring peepers.Mean � SE percentage aggressive calls given by males during each 15-s interval ofplaybacks of conspecific advertisement calls. Redrawn from (A) Fig. 2 in Brenowitz, E.A., & Rose, G. J. (1994). Behavioural plasticity mediates aggression in choruses of the Pacifictreefrog. Animal Behaviour, 47(3), 633e641 and (B) Fig. 4 in Marshall, V. T., Humfeld, S. C.,& Bee, M. A. (2003). Plasticity of aggressive signalling and its evolution in male springpeepers, Pseudacris crucifer. Animal Behaviour, 65(6), 1223e1234.


ARTICLE IN PRESS

thresholds to playbacks of advertisement calls were raised followingprolonged exposure to advertisement calls, but aggressive thresholds toaggressive calls remained unchanged. In a different experiment, Rose andBrenowitz (1997) showed that males also habituate specifically to aggressivecalls without a corresponding increase in aggressive threshold to advertise-ment calls. This series of experiments showed convincingly that short-term habituation is a likely mechanism mediating aggression in chorusesof lek-breeding species in ways that allow males to track the local densityof competitors.

Spring peepers, P. crucifer, are closely related to Pacific treefrogs, and den-sities of calling males in a chorus can vary widely from night to night andeven within a night (Gerhardt, Diekamp, & Ptacek, 1989). Males of thespring peeper also produce aggressive calls in defense of short-term callingsites, and the thresholds for eliciting aggressive calling are positively corre-lated with the amplitude of the call of a male’s nearest neighbor (Marshall,Humfeld, & Bee, 2003). Marshall et al. (2003) and Humfeld, Marshall,and Bee (2009) showed that these aggressive thresholds, like those of Pacifictreefrogs, are plastic and can become elevated after repeated exposure toconspecific advertisement calls at an amplitude above the male’s previouslydetermined aggressive threshold. The changes in a male’s aggressive responsethat occurred with repeated stimulation by a conspecific call were consistentwith the operation of short-term habituation. Males initially responded toplaybacks by producing aggressive calls, but this response quickly decreasedin an exponential fashion (Fig. 13B). Following a brief rest period in whichno stimulus was broadcast, aggressive thresholds returned to their lower,prestimulus levels consistent with a spontaneous recovery of aggression.Humfeld et al. (2009) additionally showed that habituation of aggressionin response to advertisement calls was stimulus specific. A change from astimulus containing only advertisement calls to a novel stimulus containingboth advertisement and aggressive calls elicited a recovery of the aggressiveresponse.

Hourglass treefrogs, D. ebraccatus, are a species of Neotropical treefrogwith a lek mating system similar to Pacific treefrogs and spring peepers, inwhich choruses can be exceptionally dense and males use aggressive callsin agonistic interactions with other males (Wells & Schwartz, 1984b). Addi-tionally, like Pacific treefrogs and spring peepers, males must balance a trade-off between the benefits of repelling competing males and the costs ofaggressive signals, which are less attractive to females (Wells & Bard,1987) and may be more energetically expensive than advertisement calls.

70 M.A. Bee et al.

ARTICLE IN PRESS

Reichert (2010) used a methodology similar to Brenowitz and Rose (1994)to show that the aggressive thresholds of male hourglass treefrogs in responseto conspecific advertisement calls are plastic and that these thresholds areraised following habituation to an advertisement call stimulus. In contrast,aggressive thresholds to playbacks of aggressive calls were lowered followingrepeated presentation of an aggressive-call stimulus. This latter result indi-cates a process of sensitization (Groves & Thompson, 1970), according towhich individuals initially increase response to a particularly strong stimuluswith repeated stimulation.

3.6 Summary of Rival RecognitionThere is now abundant evidence that anuran vocalizations are individuallydistinctive. However, there is still only limited evidence that anuransmake use of this information in the contexts of rival recognition and thedear enemy effect. Studies of bullfrogs and golden rocket frogs demonstratethat territorial male frogs can learn to recognize the calls of their nearbyneighbors and use this information to direct relatively lower levels of aggres-sion toward them. In bullfrogs, reduced aggressive responses to neighbors’vocalizations are also location specific as in songbirds. However, there isso far no evidence to suggest strawberry poison frogs behaviorally discrimi-nate between neighbors and strangers in ways consistent with the dear en-emy effect. This negative result is important because although males ofthis species produce individually distinctive advertisement calls and havebeen reported to defend long-term multipurpose territories, neighborsand strangers are treated similarly. Therefore, evidence from a small numberof studies of territorial anurans suggests potential diversity in rival recogni-tion and the dear enemy effect.

Consistent with the notion that habituation learning contributes to thedear enemy effect, intensive study of territorial bullfrogs has shown thataggressive responses to playbacks of calls simulating a new territorialneighbor exhibit many of the parametric characteristics of habituation. Inaddition, habituation to the calls of a simulated new neighbor is long-term and specific not only to an individually distinctive voice property,but also to the sound’s location of origin. This overall pattern of results sup-ports the hypothesis that habituation learning could play an important role asa mechanism underlying the dear enemy effect in anurans. Current evidencefrom bullfrogs also suggests a potentially tight coupling between patterns ofamong-individual and within-individual variation in calls and the decisionrules receivers use in behaviorally discriminating between neighbors and


ARTICLE IN PRESS

strangers. Studies of lek-breeding species indicate that short-term habitua-tion of aggression allows males to track changes in the local density of nearbycallers, thereby reducing the potential costs of aggressive signaling after call-ing sites have been established. We hypothesize that generally similar neuralmechanisms could mediate habituation of aggression in all anurans, but thatkey products of these mechanisms (eg, rates of response decrement andspontaneous recovery, stimulus generalization versus specificity) can beevolutionarily tailored to reflect differences in social interactions betweenterritorial and lek-breeding species. According to this hypothesis, the samegeneral mechanisms could give rise to the dear enemy effect in territorialspecies and the short-term plasticity of aggressive thresholds observed inseveral lek-breeding species.

4. FUTURE DIRECTIONS

Anurans exhibit remarkable diversity in social, reproductive, andcommunicative behaviors. This diversity provides both significant challengesand unparalleled opportunities to behavioral biologists interested in under-standing the function, evolution, and mechanisms of contest behaviors,assessment strategies, and recognition systems. What advances should westrive to achieve in the next 50 years? Here are some of our suggestions.

4.1 FunctionRival assessment and recognition have closely related hypothesized func-tions. The hypothesized function of rival assessment is that it results in less-costly interactions than outright physical aggression. But this hypothesis hasrarely been tested (Logue et al., 2010; Rillich, Schildberger, & Stevenson,2007). Anurans present a potential challenge for this hypothesis because oftheir relatively noninjurious combat (in most species) and potentially highcosts of aggressive signaling (Halliday & Tejedo, 1995). The dear enemy ef-fect is hypothesized to function in reducing the costs of repeated aggressiveinteractions. Unfortunately, this hypothesis too has rarely been tested(Beletsky & Orians, 1989; Jaeger, 1981). With the exception of reducedattractiveness to females (Marshall et al., 2003), other costs of aggressivebehavior in anurans (eg, energetic costs, time costs, risk of predation, orinjury) are often suggested but rarely quantified. Future efforts to quantifythese contest costs and the hypothesized cost savings arising from assessmentand recognition are critical to advancing our understanding of the functions

72 M.A. Bee et al.

ARTICLE IN PRESS

of both behaviors in anurans. An additional method to study the adaptivevalue of both behaviors would be to measure directly the relationshipbetween individual variation in fitness and variation in assessment or recog-nition behavior without specifying the source of costs. For instance, individ-uals’ aggressive phenotypes could be quantified by measuring averageaggressive signal characteristics, RHP, aggressive thresholds, and propensityto escalate or persist in response to standardized playback trials. Variation inthese phenotypes could then be related to variation in reproductive successin natural choruses.

Theoretical treatments of rival recognition are far less well developedcompared with those for assessment. A current limitation of all existingmodels to explain the adaptive value of the dear enemy effect (Getty,1987, 1989; Ydenberg et al., 1988, 1989) is that they assume a priori thatcontestants can recognize each other. Lack of a dear enemy effect in thesemodels would be interpreted as animals recognizing neighbors but treatingneighbors and strangers similarly. But the absence of a dear enemy effectmight also result from the absence of recognition altogether. The sensoryand cognitive machinery enabling recognition may be so costly as tooutweigh any functional benefits of recognition. Similar arguments havebeen made concerning the evolution of self-assessment versus mutual-assessment (Elwood & Arnott, 2012). Recognition strategies in futuregame theory models should bear these potential costs of recognition.Game theoretic models arising from signal detection theory (Wiley, 2015)could serve as a useful starting point to develop a better theoretical frame-work that explicitly links costs and benefits of recognition to specific out-comes of correct and incorrect decisions about fighting opponents.

Additional considerations relevant to the functions of assessment andrecognition in anurans stem from the fact that anuran choruses representnoisy communication networks (Grafe, 2005). Noise is ubiquitous in animalcommunication and induces errors in decision-making (Brumm, 2013;Wiley, 2015). Previous research has focused on how male anurans modifytheir calls and calling behavior to mitigate the impacts of noise (Schwartz& Bee, 2013), and how receivers, primarily females, are adapted to perceivecalls in noise (Vélez et al., 2013). Future work should investigate how thenoise of a chorus might constrain a male receiver’s behavioral decisions dur-ing assessment and recognition of contest rivals. Noise should induce errors,which in turn might play important roles in shaping the evolution of assess-ment and recognition systems (Wiley, 1994, 2015). In addition, the envi-ronment of a breeding chorus may present some listeners with the


ARTICLE IN PRESS

opportunity to eavesdrop on aggressive interactions between males, asshown in some other animals (Peake, 2005). The extent to which female an-urans might choose mates after eavesdropping on contests, or whether maleanurans might make decisions about engaging in or escalating contests aftereavesdropping on their opponent’s previous contests, remain open questionsfor future studies.

4.2 EvolutionAnurans represent ideal systems in which to discover both the evolutionaryhistories of rival assessment and recognition within a phylogenetic frameworkand the evolved adaptations that make these behaviors possible. A necessaryfirst step will be to learn more about the prevalence of both behaviors in an-urans in relation to the diversity present in their social, reproductive, andcommunicative behaviors. We are very far from understanding what aspectsof a species’ breeding ecology (eg, what is defended and for how long) selectfor rival recognition or for different rival assessment strategies, and how theirevolution might be constrained by physiology or phylogeny. Additional andmore rigorous field studies of rival assessment and recognition in anuran spe-cies having well described breeding ecologies should be a top priority forfuture research. This is especially true for the study of rival recognition in an-urans, which remains in its infancy. To advance our understanding of assess-ment and recognition, it will be imperative to publish both positive andnegative results. Without robust negative results, we will never elucidatethe evolutionary histories of rival assessment and recognition in anurans,which demands knowledge of which species do and do not exhibit these be-haviors. Following the best practices outlined in this review (Boxes 1 and 2)will help ensure that negative results can be trusted.

For studies of rival assessment in anurans, theory has largely outpacedempirical work. Experiments that enable a clear distinction betweenmutual- and self-assessment (Box 1) have only been performed for a fewspecies, and these should be applied over a much broader taxonomic range.However, researchers should keep in mind that mutual- and self-assessmentare endpoints on a continuum of strategies that weight the relative influenceof own and opponent qualities (Elias et al., 2008; Mesterton-Gibbons &Heap, 2014; Prenter, Elwood, & Taylor, 2006). In some cases, animalsmay even switch between different assessment strategies at different pointsin the contest (Hsu, Lee, Chen, Yang, & Cheng, 2008; Yasuda, Takeshita,& Wada, 2012). An especially powerful approach to the comparative studyof rival assessment strategies will be to test the predictions of models that

74 M.A. Bee et al.

ARTICLE IN PRESS

allow for variation in assessment strategies and explicitly examine the condi-tions under which different strategies are expected to evolve (Mesterton-Gibbons & Heap, 2014). These models could also be used to test hypothesesfor the evolution of plasticity in aggressive signaling (ie, frequency alterationand graded aggressive calling). Specifically, feedback between signal contentand assessment strategy is predicted (Maynard Smith & Parker, 1976; Prenteret al., 2006), and this feedback could be examined by comparing the role ofsignaling plasticity in assessment between taxa that differ in assessmentstrategy.

Additional advances will be made in understanding rival recognitionthrough comparative studies that focus on elucidating evolutionary adapta-tions of the components of recognition systems. All anurans might produceindividually distinctive vocalizations, yet certainly not all species exhibit rivalrecognition. This diversity should be exploited to test the signature adapta-tion hypothesis (Beecher, 1989, 1991; Tibbetts & Dale, 2007), according towhich the benefits of being recognized impose selection on signal designthat results in increased individual distinctiveness. Testing this hypothesisin anurans will require quantifying the amount of individual identity infor-mation (sensu Beecher, 1989) present in the signals of carefully selectedspecies such as pairs of closely related species with different breeding ecolo-gies. Support for the signature adaptation hypothesis would be found if spe-cies that exhibit rival recognition also have more individually distinct calls.Comparative studies are also needed to evaluate species-specific perceptualadaptations for rival recognition. Species potentially differ in the suite ofcall properties that contribute most toward individual distinctiveness (eg,spectral properties in ranids and temporal properties in dendrobatids).Thus an important goal for future research should be to uncover species dif-ferences in the perceptual basis of rival recognition. We would predictcoevolutionary matches between the patterns of individual variation presentin a species’ signals and the decision rules by which receivers perceive and actupon individual differences in specific signal properties.

4.3 MechanismsThe role of cognitive processes in the decision to persist, escalate, or with-draw from contests is controversial (Elwood & Arnott, 2012, 2013).Although very little is known about the mechanisms of assessment in an-urans, their utility as subjects of neuroethological studies provides an oppor-tunity to study the cognitive basis of rival assessment. Creative experimentaldesigns manipulating the information available to individual contestants


ARTICLE IN PRESS

(Rillich et al., 2007) could reveal how sensory processes influence theperception of the costliness of contests. The methods used in studies oflearning in the context of neighbor recognition (Section 3.5) could bemodified for testing how recognition of an opponent’s qualities is mediatedby its acoustic signals. In cases of mutual-assessment, it will be important todetermine if individual differences in aggressive signals or in graded signals indifferent contexts are actually meaningful to receivers (ie, is this variationgreater than the JMD?; Section 3.4.1). Learning and memory are also impor-tant considerations for understanding the extended temporal dynamics ofaggression in frog choruses, in particular if there are effects of winning orlosing experiences on an individual’s subsequent assessment of its ownfighting abilities in future contests (Hsu et al., 2006). Finally, very little isknown about the neural and molecular basis of rival assessment in any animalspecies (Stevenson & Rillich, 2015), although more general mechanisms ofaggression are reasonably well understood in humans and other mammals(Nelson & Trainor, 2007) and methods from these studies could be adaptedfor studies of rival assessment in frogs.

Identifying the neural basis of how anurans recognize species differencesin signals has long been a goal of neuroethological research (Gerhardt &Huber, 2002; Narins, Feng, Fay, & Popper, 2007). A new research goalfor the next generation of neuroethologists should be to elucidate the neuralmechanisms underlying the recognition of individual differences. Immediateearly gene expression has been used to identify the functional connectivityof auditory nuclei in the anuran brain in response to conspecific signals(Hoke et al., 2004). These same genomic responses have been used to iden-tify potential brain nuclei involved in individual recognition in songbirds(Mello, Nottebohm, & Clayton, 1995) and might be informative in futurestudies of anurans as well. Single-unit studies have revealed how neurons inthe anuran brain encode spectrotemporal acoustic properties used in speciesrecognition, as identified in behavioral studies (Gerhardt & Huber, 2002;Narins et al., 2007). These well-established methods might also be used toprovide in-depth analyses of how neural responses change with repeated ex-posures to signals (Megela & Capranica, 1983), and how neural circuitsencode the decision rules used by receivers to discriminate between neigh-bors and strangers. Studies of songbirds highlight the role of auditory corticalareas in individual recognition (Gentner, 2004). In contrast, anurans lackany functional or anatomical equivalent of auditory cortex (Wilczynski &Endepols, 2007). Given the presumed functional similarities between song-birds and some territorial anurans in terms of the dear enemy effect,

76 M.A. Bee et al.

ARTICLE IN PRESS

exploring how and where stimulus-specific and location-specific decre-ments in aggression develop in the anuran nervous system will be importantfor understanding the neural basis of convergence in rival recognition inthese two groups.

4.4 ConclusionThe preceding half century of research on rival assessment and recognitionhas uncovered a rich diversity of contest-related behaviors in anurans. Intheir sum, these studies reveal the existence of considerable complexityand variation in male behavior that is frequently and unduly eclipsed bystudies of female mate choice in this group. Nevertheless, previous studiesof maleemale contest behavior in anurans form a solid foundation on whichto frame new and exciting questions about animal contests. Our hope is thatthis review might stimulate future research agendas over the next half cen-tury to explain this behavioral diversity in terms of function, evolution, andmechanisms.

ACKNOWLEDGMENTSWe thank the editors, especially Marc Naguib, for inviting this contribution. We also thankMarc Naguib and one anonymous reviewer for helpful comments on a previous draft. Theauthors collectively thank the National Science Foundation and the National Institutes ofHealth for generously supporting their research over the years.

REFERENCESAlcock, J. (2013). Animal behavior: An evolutionary approach (10th ed.). Sunderland, MA:

Sinauer.Allan, D. M. (1973). Some relationships of vocalization to behavior in the Pacific treefrog,

Hyla regilla. Herpetologica, 29(4), 366e371.AmphibiaWeb. (2015). Information on amphibian biology and conservation. Available http://

amphibiaweb.org/. Berkeley, California.Arak, A. (1983). Sexual selection by maleemale competition in natterjack toad choruses.

Nature, 306(5940), 261e262.Arch, V. S., Burmeister, S. S., Feng, A. S., Shen, J.-X., & Narins, P. M. (2011). Ultrasound-

evoked immediate early gene expression in the brainstem of the Chinese torrent frog,Odorrana tormota. Journal of Comparative Physiology A, 197(6), 667e675.

Archer, J. (1988). The behavioral biology of aggression. Cambridge: Cambridge University Press.Arnott, G., & Elwood, R. W. (2008). Information gathering and decision making about

resource value in animal contests. Animal Behaviour, 76(3), 529e542.Arnott, G., & Elwood, R.W. (2009). Assessment of fighting ability in animal contests.Animal

Behaviour, 77(5), 991e1004.Axelrod, R., & Hamilton, W. D. (1981). The evolution of cooperation. Science, 211(4489),

1390e1396.Bee, M. A. (2001). Habituation and sensitization of aggression in bullfrogs (Rana catesbeiana):

testing the dual-process theory of habituation. Journal of Comparative Psychology, 115(3),307e316.

http://refhub.elsevier.com/S0065-3454(16)30001-8/sref1





http://amphibiaweb.org/

http://amphibiaweb.org/
























ARTICLE IN PRESS

Bee, M. A. (2002). Territorial male bullfrogs (Rana catesbeiana) do not assess fighting abilitybased on size-related variation in acoustic signals. Behavioral Ecology, 13(1), 109e124.

Bee, M. A. (2003a). A test of the “dear enemy effect” in the strawberry dart-poison frog(Dendrobates pumilio). Behavioral Ecology and Sociobiology, 54(6), 601e610.

Bee, M. A. (2003b). Experience-based plasticity of acoustically evoked aggression in a terri-torial frog. Journal of Comparative Physiology A, 189(6), 485e496.

Bee, M. A. (2004). Within-individual variation in bullfrog vocalizations: Implications for avocally mediated social recognition system. The Journal of the Acoustical Society of America,116(6), 3770e3781.

Bee, M. A. (2006). Individual recognition in animal species. In N. Naguib (Ed.), The encyclo-pedia of language and linguistics (Vol. 2, pp. 617e626). London: Elsevier Science.

Bee, M. A., & Bowling, A. C. (2002). Socially mediated pitch alteration by territorial malebullfrogs, Rana catesbeiana. Journal of Herpetology, 36(1), 140e143.

Bee, M. A., Cook, J. M., Love, E. K., O’Bryan, L. R., Pettitt, B. A., Schrode, K., & Vélez, A.(2010). Assessing acoustic signal variability and the potential for sexual selection andsocial recognition in boreal chorus frogs (Pseudacris maculata). Ethology, 116(6), 564e576.

Bee, M. A., & Gerhardt, H. C. (2001a). Habituation as a mechanism of reduced aggressionbetween neighboring territorial male bullfrogs (Rana catesbeiana). Journal of ComparativePsychology, 115(1), 68e82.

Bee, M. A., & Gerhardt, H. C. (2001b). Neighbor-stranger discrimination by territorial malebullfrogs (Rana catesbeiana): II. Perceptual basis. Animal Behaviour, 62(6), 1141e1150.

Bee, M. A., & Gerhardt, H. C. (2001c). Neighbour-stranger discrimination by territorialmale bullfrogs (Rana catesbeiana): I. Acoustic basis. Animal Behaviour, 62(6), 1129e1140.

Bee, M. A., & Gerhardt, H. C. (2002). Individual voice recognition in a territorial frog (Ranacatesbeiana). Proceedings of the Royal Society of London B: Biological Sciences, 269(1499),1443e1448.

Bee, M. A., Kozich, C. E., Blackwell, K. J., & Gerhardt, H. C. (2001). Individual variation inadvertisement calls of territorial male green frogs, Rana clamitans: implications for individ-ual discrimination. Ethology, 107(1), 65e84.

Bee, M. A., & Perrill, S. A. (1996). Responses to conspecific advertisement calls in the greenfrog (Rana clamitans) and their role in maleemale communication. Behaviour, 133(3),283e301.

Bee, M. A., Perrill, S. A., & Owen, P. C. (1999). Size assessment in simulated territorialencounters between male green frogs (Rana clamitans). Behavioral Ecology and Sociobiology,45(3), 177e184.

Bee, M. A., Perrill, S. A., & Owen, P. C. (2000). Male green frogs lower the pitch of acousticsignals in defense of territories: a possible dishonest signal of size? Behavioral Ecology, 11(2),169e177.

Bee, M. A., & Schachtman, T. R. (2000). Is habituation a mechanism for neighbor recogni-tion in green frogs? Behavioral Ecology and Sociobiology, 48(2), 165e168.

Beecher, M. D. (1989). Signalling systems for individual recognition: an information theoryapproach. Animal Behaviour, 38(2), 248e261.

Beecher, M. D. (1991). Successes and failures of parent- offspring recognition in animals.In P. G. Hepper (Ed.), Kin recognition (pp. 94e124). Cambridge: Cambridge Univer-sity Press.

Beecher, M. D., & Stoddard, P. K. (1990). The role of bird song and calls in individual recog-nition: contrasting field and laboratory perspectives. In W. C. Stebbins, & M. A. Berkley(Eds.), Complex signals: Vol. II. Comparative perception (pp. 375e408). New York: JohnWiley & Sons, Inc.

Beletsky, L. D., & Orians, G. H. (1989). Familiar neighbors enhance breeding success inbirds. Proceedings of the National Academy of Sciences of the United States of America,86(20), 7933e7936.










































































78 M.A. Bee et al.

ARTICLE IN PRESS

Boake, C. R. B. (1989). Repeatability: its role in evolutionary studies of mating behavior.Evolutionary Ecology, 3(2), 173e182.

Boatright-Horowitz, S. S., Cheney, C. A., & Simmons, A. M. (1999). Atmospheric andunderwater propagation of bullfrog vocalizations. Bioacoustics, 9(4), 257e280.

Bond, A. B. (1989). Toward a resolution of the paradox of aggressive displays: II. Behavioralefference and the communication of intentions. Ethology, 81(3), 235e249.

Bourne, G. R., Collins, A., Holder, A., & McCarthy, C. (2001). Vocal communication andreproductive behavior of the frog Colostethus beebei in Guyana. Journal of Herpetology,35(2), 272e281.

Bradbury, J. W., & Vehrencamp, S. L. (2011). Principles of animal communication (2nd ed.).Sunderland, MA: Sinauer Associates.

Brenowitz, E. A. (1989). Neighbor call amplitude influences aggressive behavior and inter-male spacing in choruses of the Pacific treefrog (Hyla regilla). Ethology, 83(1), 69e79.

Brenowitz, E. A., & Rose, G. J. (1994). Behavioural plasticity mediates aggression in chorusesof the Pacific treefrog. Animal Behaviour, 47(3), 633e641.

Briffa, M., & Elwood, R. W. (2000). Cumulative or sequential assessment during hermit crabshell fights: effects of oxygen on decision rules. Proceedings of the Royal Society B: BiologicalSciences, 267(1460), 2445e2452.

Briffa, M., & Elwood, R. W. (2009). Difficulties remain in distinguishing between mutualand self-assessment in animal contests. Animal Behaviour, 77(3), 759e762.

Briffa, M., & Sneddon, L. U. (2007). Physiological constraints on contest behaviour. Func-tional Ecology, 21(4), 627e637.

Brooks, R. J., & Falls, J. B. (1975). Individual recognition by song in white-throated spar-rows. I. Discrimination of songs of neighbors and strangers. Canadian Journal of Zoology,53(7), 879e888.

Brumm, H. (Ed.). (2013). Animal communication and noise. New York: Springer.Brust, D. G. (1993). Maternal brood care by Dendrobates pumilio: a frog that feeds its young.

Journal of Herpetology, 27(1), 96e98.Bunnell, P. (1973). Vocalizations in the territorial behavior of the frog Dendrobates pumilio.

Copeia, 1973(2), 277e284.Burmeister, S. S., Konieczka, J., & Wilczynski, W. (1999). Agonistic encounters in a cricket

frog (Acris crepitans) chorus: behavioral outcomes vary with local competition and withinthe breeding season. Ethology, 105(4), 335e347.

Burmeister, S. S., Ophir, A. G., Ryan, M. J., & Wilczynski, W. (2002). Information transferduring cricket frog contests. Animal Behaviour, 64(5), 715e725.

Burmeister, S. S., Wilczynski, W., & Ryan, M. J. (1999). Temporal call changes and priorexperience affect graded signalling in the cricket frog. Animal Behaviour, 57(3), 611e618.

Caldwell, M. S., Johnston, G. R., McDaniel, J. G., & Warkentin, K. M. (2010). Vibrationalsignaling in the agonistic interactions of red-eyed treefrogs. Current Biology, 20(11),1012e1017.

Castellano, S., & Giacoma, C. (1998). Stabilizing and directional female choice for male callsin the European green toad. Animal Behaviour, 56(2), 275e287.

Catchpole, C. K. (1989). Pseudoreplication and external validity: playback experiments inavian bioacoustics. Trends in Ecology and Evolution, 4(10), 286e287.

Clutton-Brock, T., & Albon, S. (1979). The roaring of red deer and the evolution of honestadvertisement. Behaviour, 69(3), 145e170.

Crothers, L. R., & Cummings, M. E. (2015). A multifunctional warning signal behaves as anagonistic status signal in a poison frog. Behavioral Ecology, 26(2), 560e568.

Crothers, L., Gering, E., & Cummings, M. (2011). Aposematic signal variation predictsmaleemale interactions in a polymorphic poison frog. Evolution, 65(2), 599e605.

Crump, M. L. (1988). Aggression in harlequin frogs: maleemale competition and a possibleconflict of interest between the sexes. Animal Behaviour, 36(4), 1064e1077.














































































ARTICLE IN PRESS

Davies, N. B., & Halliday, T. R. (1977). Optimal mate selection in the toad Bufo bufo.Nature,269(5623), 56e58.

Davies, N. B., & Halliday, T. R. (1978). Deep croaks and fighting assessment in toads Bufobufo. Nature, 274(5672), 683e685.

Davies, N. B., & Halliday, T. R. (1979). Competitive mate searching in male common toads,Bufo bufo. Animal Behaviour, 27, 1253e1267.

Davies, N. B., Krebs, J. R., &West, S. A. (2012). An introduction to behavioural ecology (4th ed.).Oxford: Wiley-Blackwell.

Davis, M. S. (1987). Acoustically mediated neighbor recognition in the North Americanbullfrog, Rana catesbeiana. Behavioral Ecology and Sociobiology, 21(3), 185e190.

Dawkins, R., & Krebs, J. R. (1978). Animal signals: Information or manipulation? InJ. R. Krebs, & N. Davies (Eds.), Behavioral ecology: An evolutionary approach (pp.282e309). Oxford: Blackwell.

Donnelly, M. A. (1989a). Demographic effects of reproductive resource supplementation in aterritorial frog, Dendrobates pumilio. Ecological Monographs, 59(3), 207e221.

Donnelly, M. A. (1989b). Effects of reproductive resource supplementation on space-use pat-terns in Dendrobates pumilio. Oecologia, 81(2), 212e218.

Dunn, D., Jander, K., Lamas, A., & Pereira, R. (2015). Mortal combat and competition foroviposition sites in female pollinating fig wasps. Behavioral Ecology, 26(1), 262e268.

Durant, P., & Dole, J. W. (1975). Aggressive behavior in Colostethus (¼Prostherapis) collaris(Anura: Dendrobatidae). Herpetologica, 31(1), 23e26.

Dyson, M. L., Reichert, M. S., & Halliday, T. R. (2013). Contests in amphibians. InI. C.W. Hardy, &M. Briffa (Eds.),Animal contests (pp. 228e257). Cambridge: CambridgeUniversity Press.

Egge, A., Brandt, Y., & Swallow, J. (2011). Sequential analysis of aggressive interactionsin the stalk-eyed fly Teleopsis dalmanni. Behavioral Ecology and Sociobiology, 65(2),369e379.

Elias, D. O., Kasumovic, M. M., Punzalan, D., Andrade, M. C. B., & Mason, A. C. (2008).Assessment during aggressive contests between male jumping spiders. Animal Behaviour,76(3), 901e910.

Elwood, R.W., & Arnott, G. (2012). Understanding how animals fight with Lloyd Morgan’scanon. Animal Behaviour, 84(5), 1095e1102.

Elwood, R. W., & Arnott, G. (2013). Assessments in contests are frequently assumed to becomplex when simple explanations will suffice. Animal Behaviour, 86(5), e8ee12.

Emlen, S. T. (1968). Territoriality in the bullfrog, Rana catesbeiana. Copeia, 1968(2),240e243.

Emlen, S. T. (1976). Lek organization and mating strategies in the bullfrog. Behavioral Ecologyand Sociobiology, 1(3), 283e313.

Enquist, M., & Leimar, O. (1983). Evolution of fighting behavior: decision rules and assess-ment of relative strength. Journal of Theoretical Biology, 102(3), 387e410.

Enquist, M., & Leimar, O. (1987). Evolution of fighting behaviour: the effect of variation inresource value. Journal of Theoretical Biology, 127(2), 187e205.

Enquist, M., Leimar, O., Ljungberg, T., Mallner, Y., & Segerdahl, N. (1990). A test of thesequential assessment game: fighting in the cichlid fish Nannacara anomala. Animal Behav-iour, 40(1), 1e14.

Falconer, D. S. (1981). Introduction to quantitative genetics (2nd ed.). New York: Longman.Falls, J. B. (1982). Individual recognition by sounds in birds. In D. E. Kroodsma, &

E. H. Miller (Eds.), Acoustic communication in birds (Vol. 2, pp. 237e278).Falls, J. B., & Brooks, R. J. (1975). Individual recognition by song in white-throated spar-

rows. II. Effects of location. Canadian Journal of Zoology, 53(10), 1412e1420.Fellers, G. M. (1979). Aggression, territoriality, and mating behaviour in North American

treefrogs. Animal Behaviour, 27(1), 107e119.












































































80 M.A. Bee et al.

ARTICLE IN PRESS

Feng, A. S., Arch, V. S., Yu, Z., Yu, X.-J., Xu, Z.-M., & Shen, J.-X. (2009). Neighbor-stranger discrimination in concave-eared torrent frogs, Odorrana tormota. Ethology,115(9), 851e856.

Feng, A. S., & Narins, P. M. (2008). Ultrasonic communication in concave-eared torrentfrogs (Amolops tormotus). Journal of Comparative Physiology A, 194(2), 159e167.

Feng, A. S., Narins, P. M., & Xu, C.-H. (2002). Vocal acrobatics in a Chinese frog, Amolopstormotus. Naturwissenschaften, 89(8), 352e356.

Feng, A. S., Narins, P. M., Xu, C.-H., Lin, W.-Y., Yu, Z.-L., Qiu, Q.… Shen, J.-X. (2006).Ultrasonic communication in frogs. Nature, 440(7082), 333e336.

Feng, A. S., Riede, T., Arch, V. S., Yu, Z., Xu, Z.-M., Yu, X.-J., & Shen, J.-X. (2009).Diversity of the vocal signals of concave-eared torrent frogs (Odorrana tormota): evidencefor individual signatures. Ethology, 115(11), 1015e1028.

Fisher, J. (1954). Evolution and bird sociality. In J. Huxley, A. C. Hardy, & E. B. Ford (Eds.),Evolution as a process (pp. 71e83). London: Allen & Unwin.

Friedl, T. W. P., & Klump, G. M. (2002). The vocal behaviour of male European treefrogs(Hyla arborea): implications for inter- and intrasexual selection. Behaviour, 139(1),113e136.

Frost, D. R. (2015). Amphibian species of the world: An online reference. New York, USA: Amer-ican Museum of Natural History. Version 6.0. Database accessible at http://research.amnh.org/herpetology/amphibia/index.html.

Gasser, H., Amézquita, A., & H€odl, W. (2009). Who is calling? Intraspecific call variation inthe aromobatid frog Allobates femoralis. Ethology, 115(6), 596e607.

Geisselmann, B., Flindt, R., & Hemmer, H. (1971). Studien zur biologie, €okologie undmerkmalsvariabilitat der beiden braunfroscharten, Rana temporaria L. und Rana dalmatinaBonaparte. Zoologisches Jahrbuch Der Physiologie, 98, 521e568.

Gentner, T. Q. (2004). Neural systems for individual song recognition in adult birds. Annalsof the New York Academy of Sciences, 1016(1), 282e302.

Gerhardt, H. C. (1991). Female mate choice in treefrogs: static and dynamic acoustic criteria.Animal Behaviour, 42(4), 615e635.

Gerhardt, H. C. (1992). Multiple messages in acoustic signals. Seminars in the Neurosciences,4(6), 391e400.

Gerhardt, H. C., Diekamp, B., & Ptacek, M. (1989). Inter-male spacing in choruses of thespring peeper, Pseudacris (Hyla) crucifer. Animal Behaviour, 38(6), 1012e1024.

Gerhardt, H. C., & Huber, F. (2002). Acoustic communication in insects and anurans. Chicago:The University of Chicago Press.

Getty, T. (1987). Dear enemies and the prisoner’s dilemma: why should territorial neighborsform defensive coalitions? American Zoologist, 27(2), 327e336.

Getty, T. (1989). Are dear enemies in a war of attrition? Animal Behaviour, 37, 337e339.Given, M. F. (1988). Territoriality and aggressive interactions of male carpenter frogs, Rana

virgatipes. Copeia, 1988(2), 411e421.Given, M. F. (1993). Male response to female vocalizations in the carpenter frog, Rana

virgatipes. Animal Behaviour, 46(6), 1139e1149.Given, M. F. (1999). Frequency alteration of the advertisement call in the carpenter frog,

Rana virgatipes. Herpetologica, 55(3), 304e317.Grafe, T. U. (1995). Graded aggressive calls in the African painted reed frog Hyperolius

marmoratus (Hyperoliidae). Ethology, 101(1), 67e81.Grafe, T. U. (2005). Anuran choruses as communication networks. In P. McGregor (Ed.),

Animal communication networks (pp. 277e299). Cambridge: Cambridge University Press.Grant, T., Frost, D. R., Caldwell, J. P., Gagliardo, R., Haddad, C. F. B., Kok, P. J. R.…

Wheeler, W. C. (2006). Phylogenetic systematics of dart-poison frogs and their relatives(Amphibia: Athesphatanura: Dendrobatidae). Bulletin of the American Museum of NaturalHistory, 299(299), 1e262.

























http://research.amnh.org/herpetology/amphibia/index.html

http://research.amnh.org/herpetology/amphibia/index.html


















































ARTICLE IN PRESS

Gridi-Papp, M., Feng, A. S., Shen, J. X., Yu, Z. L., Rosowski, J. J., & Narins, P. M. (2008).Active control of ultrasonic hearing in frogs. Proceedings of the National Academy of Sciencesof the United States of America, 105(31), 11014e11019.

Groves, P. M., & Thompson, R. F. (1970). Habituation: a dual-process theory. PsychologicalReview, 77(5), 419e450.

Haas, J. D. (1976). Individual differences in the vocalizations of the Bullfrog, Rana catesbeiana.Herpetological Review, 7(2), 86.

Haas, J. D. (1977). Individual differences in the vocalization of bullfrogs, Rana catesbeiana. Ph.D.thesis. Tufts University.

Haase, A., & Pr€ohl, H. (2002). Female activity patterns and aggressiveness in the strawberry poi-son frog Dendrobates pumilio (Anura: Dendrobatidae). Amphibia-Reptilia, 23(2), 129e140.

Hack, M. A. (1997). Assessment strategies in the contests of male crickets, Acheta domesticus(L). Animal Behaviour, 53(4), 733e747.

Halliday, T., & Tejedo, M. (1995). Intrasexual selection and alternative mating behavior. InH. Heatwole, & B. Sullivan (Eds.), Social behavior: Vol. 2. Amphibian biology (pp. 419e468). Chipping Norton, Australia: Surrey Beatty.

Halpin, Z. T. (1986). Individual odors among mammals: origins and functions. Advances in theStudy of Behavior, 16, 39e70.

Hardy, I. C. W., & Briffa, M. (Eds.). (2013). Animal contests. Cambridge: Cambridge Univer-sity Press.

Hauser, M. D., & Nelson, D. A. (1991). “Intentional” signaling in animal communication.Trends in Ecology and Evolution, 6(6), 186e189.

Hebets, E. A., & Papaj, D. R. (2005). Complex signal function: developing a framework oftestable hypotheses. Behavioral Ecology and Sociobiology, 57(3), 197e214.

Heying, H. E. (2001). Social and reproductive behaviour in the Madagascan poison frog,Mantella laevigata, with comparisons to the dendrobatids. Animal Behaviour, 61(3),567e577.

Higham, J. P., & Hebets, E. A. (2013). An introduction to multimodal communication.Behavioral Ecology and Sociobiology, 67(9), 1381e1388.

Hinde, R. A. (1970). Behavioural habituation. In G. Horn, & R. A. Hinde (Eds.), Short-termchanges in neural activity and behaviour (pp. 3e40). London: Cambridge University Press.

H€odl, W., & Amézquita, A. (2001). Visual signaling in anuran amphibians. InM. J. Ryan (Ed.),Anuran communication (pp. 121e141). Washington, DC: Smithsonian Institution Press.

H€odl, W., Amézquita, A., & Narins, P. M. (2004). The role of call frequency and the audi-tory papillae in phonotactic behavior in male Dart-poison frogs Epipedobates femoralis.Journal of Comparative Physiology A, 190(10), 823e829.

Hoke, K. L., Burmeister, S. S., Fernald, R. D., Rand, A. S., Ryan, M. J., & Wilczynski, W.(2004). Functional mapping of the auditory midbrain during mate call reception. Journalof Neuroscience, 24(50), 11264e11272.

Howard, R. D. (1978a). The evolution of mating strategies in bullfrogs, Rana catesbeiana. Evo-lution, 32(4), 850e871.

Howard, R. D. (1978b). The influence of male-defended oviposition sites on early embryomortality in bullfrogs. Ecology, 59(4), 789e798.

Howard, R. D., & Young, J. R. (1998). Individual variation in male vocal traits and femalemating preferences in Bufo americanus. Animal Behaviour, 55(5), 1165e1179.

Hsu, Y., Earley, R. L., & Wolf, L. L. (2006). Modulation of aggressive behaviour by fightingexperience: mechanisms and contest outcomes. Biological Reviews, 81(1), 33e74.

Hsu, Y., Lee, S.-P., Chen, M.-H., Yang, S.-Y., & Cheng, K.-C. (2008). Switching assess-ment strategy during a contest: fighting in killifish Kryptolebias marmoratus. Animal Behav-iour, 75(5), 1641e1649.

Humfeld, S. C., Marshall, V. T., & Bee, M. A. (2009). Context-dependent plasticity ofaggressive signalling in a dynamic social environment. Animal Behaviour, 78(4), 915e924.











































































82 M.A. Bee et al.

ARTICLE IN PRESS

Huntingford, F. A., & Turner, A. K. (1987). Animal conflict. London: Chapman & Hall.Hurd, P. L. (2006). Resource holding potential, subjective resource value, and game theo-

retical models of aggressiveness signalling. Journal of Theoretical Biology, 241(3), 639e648.Jaeger, R. G. (1981). Dear enemy recognition and the costs of aggression between

salamanders. The American Naturalist, 117(6), 962e974.Jennions, M. D., & Backwell, P. R. Y. (1996). Residency and size affect fight duration and

outcome in the fiddler crab Uca annulipes. Biological Journal of the Linnean Society, 57(4),293e306.

Kemp, D. J., Alcock, J., & Allen, G. R. (2006). Sequential size assessment and multicompo-nent decision rules mediate aerial wasp contests. Animal Behaviour, 71(2), 279e287.

Kime, N. M., Burmeister, S. S., & Ryan, M. J. (2004). Female preferences for socially variablecall characters in the cricket frog, Acris crepitans. Animal Behaviour, 68(6), 1391e1399.

Kluge, A. G. (1981). The life history, social organization, and parental behavior of Hyla rose-nbergi Boulenger, a nest-building gladiator frog. Miscellaneous Publications of the Museum ofZoology of the University of Michigan, 160, 1e170.

Kokko, H. (2013). Dyadic contests: modelling fights between two individuals. InI. C. W. Hardy, & M. Briffa (Eds.), Animal contests (pp. 5e32). Cambridge: CambridgeUniversity Press.

Kroodsma, D. E. (1976). The effect of large song repertoires on neighbor “recognition” inmale song sparrows. The Condor, 78(1), 97e99.

Kroodsma, D. E. (1989a). Inappropriate experimental designs impede progress in bioacousticresearch: a reply. Animal Behaviour, 38(4), 717e719.

Kroodsma, D. E. (1989b). Suggested experimental designs for song playbacks. Animal Behav-iour, 37, 600e609.

Kroodsma, D. E. (1990). Using appropriate experimental designs for intended hypotheses in“song” playbacks, with examples for testing effects of song repertoire sizes. Animal Behav-iour, 40(6), 1138e1150.

Kroodsma, D. E., Byers, B. E., Goodale, E., Johnson, S., & Liu, W.-C. (2001). Pseudore-plication in playback experiments, revisited a decade later. Animal Behaviour, 61(5),1029e1033.

Lambrechts, M. M., & Dhondt, A. (1995). Individual voice discrimination in birds. InD. M. Power (Ed.), Current ornithology (Vol. 12, pp. 115e139). New York: Plenum Press.

Leimar, O., Austad, S., & Enquist, M. (1991). A test of the sequential assessment game efighting in the bowl and doily spider Frontinella pyramitela. Evolution, 45(4), 862e874.

Lesbarr�eres, D., & Lodé, T. (2002). Variations in male calls and responses to an unfamiliaradvertisement call in a territorial breeding anuran, Rana dalmatina: evidence for a “dearenemy” effect. Ethology Ecology and Evolution, 14(4), 287e295.

Lessells, C. M., & Boag, P. T. (1987). Unrepeatable repeatabilities: a common mistake. Auk,104, 116e121.

Lodé, T., & Lesbarr�eres, D. (2004). Multiple paternity in Rana dalmatina, a monogamous ter-ritorial breeding anuran. Naturwissenschaften, 91(1), 44e47.

Logue, D. M., Abiola, I. O., Rains, D., Bailey, N.W., Zuk, M., & Cade,W. H. (2010). Doessignalling mitigate the cost of agonistic interactions? A test in a cricket that has lost itssong. Proceedings of the Royal Society B: Biological Sciences, 277(1693), 2571e2575.

Lopez, P. T., Narins, P. M., Lewis, E. R., & Moore, S. W. (1988). Acoustically induced callmodification in the white-lipped frog, Leptodactylus albilabris. Animal Behaviour, 36(5),1295e1308.

Macphail, E. M. (1993). The neuroscience of animal intelligence: From the sea hare to the seahorse.New York: Columbia University Press.

Marler, C. A., Chu, J., &Wilczynski, W. (1995). Arginine vasotocin injection increases prob-ability of calling in cricket frogs, but causes call changes characteristic of less aggressivemales. Hormones and Behavior, 29(4), 554e570.












































































ARTICLE IN PRESS

Marshall, V. T., Humfeld, S. C., & Bee, M. A. (2003). Plasticity of aggressive signalling and itsevolution in male spring peepers, Pseudacris crucifer. Animal Behaviour, 65(6), 1223e1234.

Martin, W. F. (1971). Mechanics of sound production in toads of the genus Bufo: passiveelements. Journal of Experimental Zoology, 176(3), 273e293.

Martin, W. F. (1972). Evolution of vocalization in the genus Bufo. In W. F. Blair (Ed.), Evo-lution in the genus Bufo (pp. 279e309). Austin: University of Texas Press.

Martof, B. S. (1953). Territoriality in the green frog, Rana clamitans. Ecology, 34(1), 165e174.Maynard Smith, J. (1979). Game theory and the evolution of behaviour. Proceedings of the

Royal Society B: Biological Sciences, 205(1161), 475e488.Maynard Smith, J. (1982a). Do animals convey information about their intentions? Journal of

Theoretical Biology, 97(1), 1e5.Maynard Smith, J. (1982b). Evolution and the theory of games. Cambridge: Cambridge Univer-

sity Press.Maynard Smith, J. (1994). Must reliable signals always be costly? Animal Behaviour, 47(5),

1115e1120.Maynard Smith, J., & Parker, G. A. (1976). The logic of asymmetric contests. Animal Behav-

iour, 24(1), 159e175.Maynard Smith, J., & Price, G. R. (1973). The logic of animal conflict. Nature, 246, 15e18.McGregor, P. K., Catchpole, C. K., Dabelsteen, T., Falls, J. B., Fusani, L., Gerhardt, H. C.…

Weary, D. M. (1992). Design of playback experiments: the Thornbridge Hall NATOARW consensus. In P. K. McGregor (Ed.), Playback and studies of animal communication(pp. 1e9). New York: Plenum Press.

McVey, M. E., Zahary, R. G., Perry, D., & MacDougal, J. (1981). Territoriality and homingbehavior in the poison dart frog (Dendrobates pumilio). Copeia, 1981(1), 1e8.

Megela, A. L., & Capranica, R. R. (1983). A neural and behavioral study of auditory habitu-ation in the bullfrog,Rana catesbeiana. Journal of Comparative Physiology A, 151(4), 423e434.

Mello, C., Nottebohm, F., & Clayton, D. (1995). Repeated exposure to one song leads to arapid and persistent decline in an immediate early gene’s response to that song in zebrafinch telencephalon. The Journal of Neuroscience, 15(10), 6919e6925.

Mesterton-Gibbons, M., & Heap, S. M. (2014). Variation between self- and mutual assess-ment in animal contests. The American Naturalist, 183(2), 199e213.

Mesterton-Gibbons, M., Marden, J. H., & Dugatkin, L. A. (1996). On wars of attritionwithout assessment. Journal of Theoretical Biology, 181(1), 65e83.

Meuche, I., Linsenmair, K., & Pr€ohl, H. (2012). Intrasexual competition, territoriality andacoustic communication in male strawberry poison frogs (Oophaga pumilio). BehavioralEcology and Sociobiology, 66(4), 613e621.

Moore, J. C., Obbard, D. J., Reuter, C., West, S. A., & Cook, J. M. (2008). Fighting stra-tegies in two species of fig wasp. Animal Behaviour, 76(2), 315e322.

Morrell, L. J., Backwell, P. R. Y., & Metcalfe, N. B. (2005). Fighting in fiddler crabs Ucamjoebergi: what determines duration? Animal Behaviour, 70(3), 653e662.

Narins, P. M. (1990). Seismic communication in anuran amphibians. Bioscience, 40(4),268e274.

Narins, P. M., Feng, A. S., Fay, R. R., & Popper, N. (Eds.). (2007).Hearing and sound commu-nication in amphibians. New York: Springer.

Narins, P. M., Feng, A. S., Lin, W., Schnitzler, H.-U., Denzinger, A., Suthers, R. A., &Xu, C. (2004). Old world frog and bird vocalizations contain prominent ultrasonicharmonics. The Journal of the Acoustical Society of America, 115(2), 910e913.

Nelson, D. A., & Marler, P. (1990). The perception of birdsong and an ecological concept ofsignal space. In W. C. Stebbins, & M. A. Berkley (Eds.), Complex signals: Vol. II. Compar-ative perception (pp. 443e478). New York: John Wiley & Sons, Inc.

Nelson, R. J., & Trainor, B. C. (2007). Neural mechanisms of aggression. Nature ReviewsNeuroscience, 8(7), 536e546.












































































84 M.A. Bee et al.

ARTICLE IN PRESS

Ophir, A. G., Burmeister, S. S., & Wilczynski, W. (1997). The effects of stimulus calls on thebehavioral outcome of a simulated agonistic interaction in male cricket frogs. AmericanZoologist, 37, 33A.

Owen, P. C., & Gordon, N. M. (2005). The effect of perceived intruder proximity and resi-dent body size on the aggressive responses of male green frogs, Rana clamitans (Anura:Ranidae). Behavioral Ecology and Sociobiology, 58(5), 446e455.

Owen, P. C., & Perrill, S. A. (1998). Habituation in the green frog, Rana clamitans. BehavioralEcology and Sociobiology, 44(3), 209e213.

Parker, G. A. (1974). Assessment strategy and the evolution of fighting behaviour. Journal ofTheoretical Biology, 47(1), 223e243.

Parker, G. A. (1984). Evolutionarily stable strategies. In J. R. Krebs, & N. B. Davies (Eds.),Behavioral ecology: An evolutionary approach (2nd ed., pp. 30e61). Blackwell Scientific.

Parker, G. A., & Rubenstein, D. I. (1981). Role assessment, reserve strategy, and acquisitionof information in asymmetric animal conflicts. Animal Behaviour, 29(1), 221e240.

Partan, S. R., & Marler, P. (1999). Communication goes multimodal. Science, 283(5406),1272e1273.

Payne, R. J. H. (1998). Gradually escalating fights and displays: the cumulative assessmentmodel. Animal Behaviour, 56(3), 651e662.

Payne, R. J. H., & Pagel, M. (1996). Escalation and time costs in displays of endurance. Journalof Theoretical Biology, 183(2), 185e193.

Payne, R. J. H., & Pagel, M. (1997). Why do animals repeat displays?Animal Behaviour, 54(1),109e119.

Peake, T. (2005). Eavesdropping in communication networks. In P. K. McGregor (Ed.),Animal communication networks (pp. 13e37). Cambridge: Cambridge University Press.

Peeke, H. V. S. (1984). Habituation and the maintenance of territorial boundaries. InH. V. Peeke, & L. Petrinovich (Eds.), Habituation, sensitization, and behavior (pp. 393e421). Orlando: Academic Press.

Peeke, H. V. S., & Peeke, S. C. (1973). Habituation in fish with special reference to intra-specific aggressive behavior. In H. V. S. Peeke, & M. J. Herz (Eds.),Habituation: I. Behav-ioral studies (pp. 59e85). New York: Academic Press.

Petrinovich, L. (1984). A two-factor dual-process theory of habituation and sensitization. InH. V. S. Peeke, & L. Petrinovich (Eds.), Habituation, sensitization, and behaviour. NewYork: Academic Press.

Pettitt, B. A. (2012). Paternal effects in relation to acoustically mediated mate choice in a neotropicalfrog. Ph.D. dissertation. University of Minnesota.

Pettitt, B. A., Bourne, G. R., & Bee, M. A. (2012). Quantitative acoustic analysis of the vocalrepertoire of the golden rocket frog (Anomaloglossus beebei). The Journal of the AcousticalSociety of America, 131(6), 4811e4820.

Pettitt, B. A., Bourne, G. R., & Bee, M. A. (2013). Advertisement call variation in the goldenrocket frog (Anomaloglossus beebei): evidence for individual distinctiveness. Ethology,119(3), 244e256.

Piper, W., Walcott, C., Mager, J., & Spilker, F. (2008). Fatal battles in common loons: a pre-liminary analysis. Animal Behaviour, 75(3), 1109e1115.

Platz, J. E., & Forester, D. C. (1988). Geographic variation in mating call among the foursubspecies of the chorus frog: Pseudacris triseriata (Wied). Copeia, 1988(4), 1062e1066.

Prenter, J., Elwood, R. W., & Taylor, P. W. (2006). Self-assessment by males during ener-getically costly contests over precopula females in amphipods. Animal Behaviour, 72(4),861e868.

Prestwich, K. N. (1994). The energetics of acoustic signaling in anurans and insects. AmericanZoologist, 34(6), 625e643.

Pr€ohl, H. (1997a). Patr�on reproductivo en Dendrobates pumilio (Anura: Dendrobatidae).Revista de Biología Tropical, 45(4), 1671e1676.










































































ARTICLE IN PRESS

Pr€ohl, H. (1997b). Territorial behavior of the strawberry poison-dart frog, Dendrobatespumilio. Amphibia-Reptilia, 18(4), 437e442.

Pr€ohl, H. (2002). Population differences in female resource abundance, adult sex ratio, andmale mating success in Dendrobates pumilio. Behavioral Ecology, 13(2), 175e181.

Pr€ohl, H. (2003). Variation in male calling behaviour and relation to male mating success inthe strawberry poison frog (Dendrobates pumilio). Ethology, 109(4), 273e290.

Pr€ohl, H. (2005). Territorial behavior in dendrobatid frogs. Journal of Herpetology, 39(3),354e365.

Pr€ohl, H., & Berke, O. (2001). Spatial distributions of male and female strawberry poisonfrogs and their relation to female reproductive resources. Oecologia, 129(4), 534e542.

Pr€ohl, H., & H€odl, W. (1999). Parental investment, potential reproductive rates, and matingsystem in the strawberry dart-poison frog, Dendrobates pumilio. Behavioral Ecology andSociobiology, 46(4), 215e220.

Pyron, R. A., & Wiens, J. J. (2011). A large-scale phylogeny of Amphibia including over2800 species, and a revised classification of extant frogs, salamanders, and caecilians.Molecular Phylogenetics and Evolution, 61(2), 543e583.

Ramer, J. D., Jenssen, T. A., & Hurst, C. J. (1983). Size-related variation in the advertisementcall of Rana clamitans (Anura, Ranidae), and its effect on conspecific males. Copeia,1983(1), 141e155.

Rankin, C. H., Abrams, T., Barry, R. J., Bhatnagar, S., Clayton, D. F., Colombo, J.…Thompson, R. F. (2009). Habituation revisited: an updated and revised description ofthe behavioral characteristics of habituation. Neurobiology of Learning and Memory, 92(2),135e138.

Reichert, M. S. (2010). Aggressive thresholds in Dendropsophus ebraccatus: habituation andsensitization to different call types. Behavioral Ecology and Sociobiology, 64(4), 529e539.

Reichert, M. S. (2011a). Aggressive calls improve leading callers’ attractiveness in the treefrogDendropsophus ebraccatus. Behavioral Ecology, 22(5), 951e959.

Reichert, M. S. (2011b). Effects of multiple-speaker playbacks on aggressive callingbehavior in the treefrog Dendropsophus ebraccatus. Behavioral Ecology and Sociobiology,65(9), 1739e1751.

Reichert, M. S. (2012). Call timing is determined by response call type, but not by stimulusproperties, in the treefrog Dendropsophus ebraccatus. Behavioral Ecology and Sociobiology,66(3), 433e444.

Reichert, M. S. (2013a). Patterns of variability are consistent across signal types in the treefrogDendropsophus ebraccatus. Biological Journal of the Linnean Society, 109(1), 131e145.

Reichert, M. S. (2013b). Sources of variability in advertisement and aggressive calling incompetitive interactions in the grey treefrog, Hyla versicolor. Bioacoustics, 22(3), 195e214.

Reichert, M. S. (2014). Playback tests and studies of animal contest dynamics: concepts andan example in the gray tree frog. Behavioral Ecology, 25(3), 591e603.

Reichert, M. S., & Gerhardt, H. C. (2011). The role of body size on the outcome, escalationand duration of contests in the grey treefrog, Hyla versicolor. Animal Behaviour, 82(6),1357e1366.

Reichert, M. S., & Gerhardt, H. C. (2012). Trade-offs and upper limits to signal performanceduring close-range vocal competition in gray tree frogsHyla versicolor. The American Natu-ralist, 180(4), 425e437.

Reichert, M. S., & Gerhardt, H. C. (2013a). Gray tree frogs, Hyla versicolor, give lower-frequency aggressive calls in more escalated contests. Behavioral Ecology and Sociobiology,67(5), 795e804.

Reichert, M. S., & Gerhardt, H. C. (2013b). Socially mediated plasticity in call timing in thegray tree frog, Hyla versicolor. Behavioral Ecology, 24(2), 393e401.

Reichert, M. S., & Gerhardt, H. C. (2014). Behavioral strategies and signaling in interspecificaggressive interactions in gray tree frogs. Behavioral Ecology, 25(3), 520e530.

















































































86 M.A. Bee et al.

ARTICLE IN PRESS

Riechert, S. E. (2013). Maynard Smith & Parker’s (1976) rule book for animal contests,mostly. Animal Behaviour, 86(1), 3e9.

Rillich, J., Schildberger, K., & Stevenson, P. A. (2007). Assessment strategy of fighting cricketsrevealed by manipulating information exchange. Animal Behaviour, 74(4), 823e836.

Ringler, M., Ringler, E., Maga~na Mendoza, D., & H€odl, W. (2011). Intrusion experimentsto measure territory size: development of the method, tests through simulations, andapplication in the frog Allobates femoralis. PloS One, 6(10), e25844.

Ringler, M., Ursprung, E., & H€odl, W. (2009). Site fidelity and patterns of short- and long-term movement in the brilliant-thighed poison frog Allobates femoralis (Aromobatidae).Behavioral Ecology and Sociobiology, 63(9), 1281e1293.

Robertson, J. G. M. (1986). Male territoriality, fighting and assessment of fighting ability inthe Australian frog Uperoleia rugosa. Animal Behaviour, 34(3), 763e772.

Roithmair, M. E. (1992). Territoriality and male mating success in the dart-poison frog,Epipedobates femoralis (Dendrobatidae, Anura). Ethology, 92(4), 331e343.

Roithmair, M. E. (1994). Field studies on reproductive behaviour in two dart-poison frogspecies (Epipedobates femoralis, Epipedobates trivittatus) in Amazonian Peru. HerpetologicalJournal, 4(3), 77e85.

Rose, G. J., & Brenowitz, E. A. (1991). Aggressive thresholds of male Pacific treefrogs foradvertisement calls vary with amplitude of neighbors calls. Ethology, 89(3), 244e252.

Rose, G. J., & Brenowitz, E. A. (1997). Plasticity of aggressive thresholds in Hyla regilla:discrete accommodation to encounter calls. Animal Behaviour, 53(2), 353e361.

Rutte, C., Taborsky, M., & Brinkhof, M. W. G. (2006). What sets the odds of winning andlosing? Trends in Ecology and Evolution, 21(1), 16e21.

Ryan, M. J. (1980). The reproductive behavior of the bullfrog (Rana catesbeiana). Copeia,1980(1), 108e114.

Schneider, H., Sofianidou, T. S., & Kyriakopoulou-Sklavounou, P. (1988). Calling behaviorand calls of Rana dalmatina (Anura, Ranidae) in Greece. Zoologisches Jahrbuch Der Physio-logie, 92(2), 231e243.

Schwartz, J. J. (1987). The importance of spectral and temporal properties in species and callrecognition in a neotropical treefrog with a complex vocal repertoire. Animal Behaviour,35(2), 340e347.

Schwartz, J. J. (1989). Graded aggressive calls of the spring peeper, Pseudacris crucifer. Herpeto-logica, 45(2), 172e181.

Schwartz, J. J. (1994). Male advertisement and female choice in frogs e recent findings andnew approaches to the study of communication in a dynamic acoustic environment.American Zoologist, 34(6), 616e624.

Schwartz, J. J. (2001). Call monitoring and interactive playback systems in the study ofacoustic interactions among male anurans. In M. J. Ryan (Ed.), Anuran communication(pp. 183e204). Washington, DC: Smithsonian Institution Press.

Schwartz, J. J., & Bee, M. A. (2013). Anuran acoustic signal production in noisyenvironments. In H. Brumm (Ed.), Animal communication and noise (pp. 91e132). NewYork: Springer.

Schwartz, J. J., Bee, M. A., & Tanner, S. D. (2000). A behavioral and neurobiological study ofthe responses of gray treefrogs, Hyla versicolor, to the calls of a predator, Rana catesbeiana.Herpetologica, 56(1), 27e37.

Schwartz, J. J., & Simmons, A. M. (1990). Encoding of a spectrally-complex communica-tion sound in the bullfrog’s auditory nerve. Journal of Comparative Physiology A, 166(4),489e499.

Schwartz, J. J., & Wells, K. D. (1983). An experimental study of acoustic interferencebetween two species of neotropical treefrogs. Animal Behaviour, 31(1), 181e190.

Schwartz, J. J., & Wells, K. D. (1984a). Interspecific acoustic interactions of the Neotropicaltreefrog Hyla ebraccata. Behavioral Ecology and Sociobiology, 14(3), 211e224.














































































ARTICLE IN PRESS

Schwartz, J. J., & Wells, K. D. (1984b). Vocal behavior of the neotropical treefrog Hylaphlebodes. Herpetologica, 40(4), 452e463.

Schwartz, J. J., & Wells, K. D. (1985). Intraspecific and interspecific vocal behavior of theneotropical treefrog Hyla microcephala. Copeia, 1985(1), 27e38.

Searcy, W. A. (1989). Pseudoreplication, external validity and the design of playbackexperiments. Animal Behaviour, 38(4), 715e717.

Searcy, W. A., & Beecher, M. D. (2009). Song as an aggressive signal in songbirds. AnimalBehaviour, 78(6), 1281e1292.

Searcy, W. A., & Nowicki, S. (2005). The evolution of animal communication: Reliability anddeception in signaling systems. Princeton, NJ: Princeton University Press.

Seidel, B. (1999). Water-wave communication between territorial male Bombina variegata.Journal of Herpetology, 33(3), 457e462.

Sexton, O. J. (1960). Some aspects of the behavior and of the territory of a dendrobatid frog,Prostherapis trinitatis. Ecology, 41(1), 107e115.

Shannon, C. E., & Weaver, W. (1949). The mathematical theory of communication. Urbana:University of Illinois Press.

Sharpless, S. K., & Jasper, H. (1956). Habituation of the arousal reaction. Brain, 79(4), 655e680.Shen, J.-X., Feng, A. S., Xu, Z.-M., Yu, Z.-L., Arch, V. S., Yu, X.-J., & Narins, P. M.

(2008). Ultrasonic frogs show hyperacute phonotaxis to female courtship calls. Nature,453(7197), 914e916.

Shen, J.-X., Xu, Z.-M., Yu, Z.-L., Wang, S., Zheng, D.-Z., & Fan, S.-C. (2011). Ultrasonicfrogs show extraordinary sex differences in auditory frequency sensitivity.Nature Commu-nications, 2, 342.

Sherman, P. W., Reeve, H. K., & Pfennig, D. W. (1997). Recognition systems. InJ. R. Krebs, & N. B. Davies (Eds.), Behavioral ecology: An evolutionary approach (4th ed.,pp. 69e96). Oxford: Blackwell Science Ltd.

Shettleworth, S. J. (2009). Cognition, evolution, and behavior. New York, NY: Oxford Univer-sity Press.

Shy, E. (1985). Individuality in the mating calls of the treefrogHyla arborea savignyi. Amphibia-Reptilia, 6(4), 343e353.

Siddle, D. A. T. (1990). Orienting, habituation, and resource allocation: an associativeanalysis. Psychophysiology, 28(3), 245e259.

Simmons, A. M. (2013). “To Ear is Human, to Frogive is Divine”: Bob Capranica’s legacy toauditory neuroethology. Journal of Comparative Physiology A, 199(3), 169e182.

Simmons, A. M., Reese, G., & Ferragamo, M. (1993). Periodicity extraction in the anuranauditory nerve. II. Phase and temporal fine-structure. Journal of the Acoustical Society ofAmerica, 93(6), 3374e3389.

Simmons, A. M., Schwartz, J. J., & Ferragamo, M. (1992). Auditory nerve representation of acomplex communication sound in background noise. Journal of the Acoustical Society ofAmerica, 91(5), 2831e2844.

Sokal, R. R., & Rohlf, F. J. (1995). Biometry: The principles and practice of statistics in biologicalresearch (3rd ed.). New York: Freeman.

Sokolov, E. N. (1963). Perception and the conditioned reflex. Oxford: Pergamon.Starnberger, I., Preininger, D., & H€odl, W. (2014). From uni- to multimodality: towards an

integrative view on anuran communication. Journal of Comparative Physiology A, 200(9),777e787.

Stevenson, P. A., & Rillich, J. (2015). Adding up the oddsdNitric oxide signaling underliesthe decision to flee and post-conflict depression of aggression. Scientific Advances, 1(2),e1500060.

Stoddard, P. K. (1996). Vocal recognition of neighbors by territorial passerines. InD. E. Kroodsma, & E. H. Miller (Eds.), Ecology and evolution of acoustic communication inbirds (pp. 356e374). Ithaca, NY: Cornell University Press.







































































88 M.A. Bee et al.

ARTICLE IN PRESS

Stuart-Fox, D. (2006). Testing game theory models: fighting ability and decision rulesin chameleon contests. Proceedings of the Royal Society B: Biological Sciences, 273(1593),1555e1561.

Stynoski, J. L. (2009). Discrimination of offspring by indirect recognition in an egg-feedingdendrobatid frog, Oophaga pumilio. Animal Behaviour, 78(6), 1351e1356.

Stynoski, J. L., & Noble, V. R. (2011). To beg or to freeze: multimodal sensory integrationdirects behavior in a tadpole. Behavioral Ecology and Sociobiology, 66(2), 191e199.

Sullivan, B. K. (1982). Sexual selection in Woodhouse’s toad (Bufo woodhousei) I. Chorusorganization. Animal Behaviour, 30(3), 680e686.

Suthers, R. A., Narins, P. M., Lin, W.-Y., Schnitzler, H.-U., Denzinger, A., Xu, C.-H., &Feng, A. S. (2006). Voices of the dead: complex nonlinear vocal signals from the larynxof an ultrasonic frog. Journal of Experimental Biology, 209(24), 4984e4993.

Taylor, P. W., & Elwood, R. W. (2003). The mismeasure of animal contests. Animal Behav-iour, 65(6), 1195e1202.

Tejedo, M. (1988). Fighting for females in the toad Bufo calamita is affected by the operationalsex ratio. Animal Behaviour, 36(6), 1765e1769.

Temeles, E. (1994). The role of neighbours in territorial systems: when are they “dearenemies”? Animal Behaviour, 47(2), 339e350.

Test, F. H. (1954). Social aggressiveness in an amphibian. Science, 120(3108), 140e141.Thompson, R. F., & Spencer, W. A. (1966). Habituation: a model phenomenon for the

study of neuronal substrates of behavior. Psychological Review, 73(1), 16e43.Tibbetts, E. A., & Dale, J. (2007). Individual recognition: it is good to be different. Trends in

Ecology and Evolution, 22(10), 529e537.Ursprung, E., Ringler, M., & H€odl, W. (2009). Phonotactic approach pattern in the

neotropical frog Allobates femoralis: a spatial and temporal analysis. Behaviour, 146(2),153e170.

Vélez, A., Schwartz, J. J., & Bee, M. A. (2013). Anuran acoustic signal perception in noisyenvironments. In H. Brumm (Ed.), Animal communication and noise (pp. 133e185).New York: Springer.

Wagner, A. R. (1976). Priming in STM: an information-processing mechanism for self-generated or retrieval-generated depression in performance. In T. J. Tighe, &R. N. Leaton (Eds.), Habituation: Perspectives from child development, animal behavior, andneurophysiology (pp. 95e128). Hillsdale, NJ: Erlbaum.

Wagner, W. E. (1989a). Fighting, assessment, and frequency alteration in Blanchard’s cricketfrog. Behavioral Ecology and Sociobiology, 25(6), 429e436.

Wagner, W. E. (1989b). Graded aggressive signals in Blanchard’s cricket frog: vocal responsesto opponent proximity and size. Animal Behaviour, 38(6), 1025e1038.

Wagner, W. E. (1989c). Social correlates of variation in male calling behavior in Blanchard’scricket frog, Acris crepitans blanchardi. Ethology, 82(1), 27e45.

Wagner, W. E. (1992). Deceptive or honest signaling of fighting ability e a test of alternativehypotheses for the function of changes in call dominant frequency by male cricket frogs.Animal Behaviour, 44(3), 449e462.

Walkowiak, W., & M€unz, H. (1985). The significance of water surface-waves in thecommunication of fire-bellied toads. Naturwissenschaften, 72(1), 49e51.

Wells, K. D. (1977a). Territoriality and male mating success in the green frog (Rana clamitans).Ecology, 58(4), 750e762.

Wells, K. D. (1977b). The social behaviour of anuran amphibians. Animal Behaviour, 25,666e693.

Wells, K. D. (1978). Territoriality in the green frog (Rana clamitans): vocalizations andagonistic behaviour. Animal Behaviour, 26, 1051e1063.

Wells, K. D. (1980). Social behavior and communication of a dendrobatid frog (Colostethustrinitatis). Herpetologica, 36(2), 189e199.
















































































ARTICLE IN PRESS

Wells, K. D. (1988). The effect of social interactions on anuran vocal behavior. In B. Fritzsch,M. J. Ryan, W. Wilczynski, T. E. Hetherington, & W. Walkowiak (Eds.), The evolutionof the Amphibian auditory system (pp. 433e454). New York: Wiley.

Wells, K. D. (1989). Vocal communication in a neotropical treefrog, Hyla ebraccata: responsesof males to graded aggressive calls. Copeia, 1989(2), 461e466.

Wells, K. D. (2001). The energetics of calling in frogs. In M. J. Ryan (Ed.), Anurancommunication (pp. 45e60). Washington: Smithsonian Institution Press.

Wells, K. D. (2007). The ecology and behavior of amphibians. Chicago: The University ofChicago Press.

Wells, K. D., & Bard, K. M. (1987). Vocal communication in a neotropical treefrog, Hylaebraccata: responses of females to advertisement and aggressive calls. Behaviour, 101(1),199e210.

Wells, K. D., & Schwartz, J. J. (1984a). Vocal communication in a neotropical treefrog, Hylaebraccata: advertisement calls. Animal Behaviour, 32(2), 405e420.

Wells, K. D., & Schwartz, J. J. (1984b). Vocal communication in a neotropical treefrog, Hylaebraccata: aggressive calls. Behaviour, 91(1), 128e145.

Wells, K. D., & Taigen, T. L. (1986). The effect of social interactions on calling energetics inthe gray treefrog (Hyla versicolor). Behavioral Ecology and Sociobiology, 19(1), 9e18.

Whitlow, J. W. J., & Wagner, A. R. (1984). Memory and habituation. In H. V. S. Peeke, &L. Petrinovich (Eds.), Habituation, sensitization, and behaviour (pp. 103e153). New York:Academic Press.

Whitney, C. L. (1980). The role of the “encounter” call in spacing of Pacific tree frogs, Hylaregilla. Canadian Journal of Zoology, 58(1), 75e78.

Whitney, C. L., & Krebs, J. R. (1975). Spacing and calling in Pacific tree frogs, Hyla regilla.Canadian Journal of Zoology, 53(11), 1519e1527.

Wiewandt, T. A. (1969). Vocalization, aggressive behavior, and territoriality in the bullfrog,Rana catesbeiana. Copeia, 1969(2), 276e285.

Wilczynski, W., & Endepols, H. (2007). Central auditory pathways in anuran amphibians:the anatomical basis of hearing and sound communication. In P. M. Narins,A. S. Feng, R. R. Fay, & N. Popper (Eds.), Hearing and sound communication in amphibians(pp. 221e249). New York: Springer.

Wiley, R. H. (1994). Errors, exaggeration, and deception in animal communication. InL. A. Real (Ed.), Behavioral mechanisms in evolutionary ecology (pp. 157e189). Chicago& London: University of Chicago Press.

Wiley, R. H. (2015). Noise matters: The evolution of communication. Cambridge and London:Harvard University Press.

Wiley, R. H., & Wiley, M. S. (1977). Recognition of neighbors’ duets by stripe-backedwrens Campylorhynchus nuchalis. Behaviour, 62(1), 10e34.

Wilson, E. O. (1975). Sociobiology: The new synthesis. Cambridge, MA: Harvard UniversityPress.

Yasuda, C., Takeshita, F., & Wada, S. (2012). Assessment strategy in maleemale contests ofthe hermit crab Pagurus middendorffii. Animal Behaviour, 84(2), 385e390.

Ydenberg, R. C., Giraldeau, L.-A., & Falls, J. B. (1988). Neighbours, strangers, and the asym-metric war of attrition. Animal Behaviour, 36(2), 343e347.

Ydenberg, R. C., Giraldeau, L.-A., & Falls, J. B. (1989). Remarks on Getty’s “fighting tolearn” hypothesis. Animal Behaviour, 37(2), 336e337.

































































Assessment and Recognition of Rivals in Anuran Contests · 2016-03-09 · Assessment and...

Documents

Transcript of Assessment and Recognition of Rivals in Anuran Contests · 2016-03-09 · Assessment and...