Plumage Mimicry in Avian Mixed-Species Flocks

0 downloads 0 Views 306KB Size Report
entre le modèle putatif et l'imitateur (s'il n'y avait aucun modèle hypothétique, nous avons considéré que les deux espèces pouvaient être des modèles l'une ...
3OXPDJH0LPLFU\LQ$YLDQ0L[HG6SHFLHV)ORFNV0RUHRU/HVVWKDQ0HHWVWKH(\H" 0LP«WLVPHGXSOXPDJHGDQVOHVYRO«HVPL[WHVG RLVHDX[SOXVRXPRLQVTXHOHVDSSDUHQFHV" $XWKRU V *X\%HDXFKDPSDQG(EHQ*RRGDOH 6RXUFH7KH$XN9RO1R -XO\ SS 3XEOLVKHGE\University of California PressRQEHKDOIRIWKHAmerican Ornithologists' Union 6WDEOH85/http://www.jstor.org/stable/10.1525/auk.2011.11016 . $FFHVVHG Your use of the JSTOR archive indicates your acceptance of the Terms & Conditions of Use, available at . http://www.jstor.org/page/info/about/policies/terms.jsp JSTOR is a not-for-profit service that helps scholars, researchers, and students discover, use, and build upon a wide range of content in a trusted digital archive. We use information technology and tools to increase productivity and facilitate new forms of scholarship. For more information about JSTOR, please contact [email protected].

University of California Press and American Ornithologists' Union are collaborating with JSTOR to digitize, preserve and extend access to The Auk.

http://www.jstor.org

The Auk 128(3):487 496, 2011 ‘ The American Ornithologists’ Union, 2011. Printed in USA.

PLUMAGE MIMICRY IN AVIAN MIXED-SPECIES FLOCKS: MORE OR LESS THAN MEETS THE EYE? G UY B EAUCHAMP1,3 2

AND

E BEN G OODALE 2

1 Faculty of Veterinary Medicine, University of Montréal, P.O. Box 5000, St.-Hyacinthe, Québec J2S 7C6, Canada; and Section of Ecology, Behavior and Evolution, Division of Biological Sciences, University of California, San Diego, 9500 Gilman Drive, La Jolla, California 92093, USA

A.—Social mimicry was postulated as a possible evolutionary mechanism that would produce convergence in the characteristics of species that interact extensively and, thus, act as an opposing force to competition and character displacement. The examples for this hypothesis were drawn mostly from visual resemblances among bird species that flock together. We evaluated plumage mimicry among groups of birds, asking  raters to score the resemblances among  sets of putative mimicry postulated in the past four decades. The resemblance between the putative model and the mimic (if there was no one model hypothesized, we considered both species to be possible models for each other) was compared with that between the model and () a species closely related to the mimic and () another species that only occasionally associated with the mixed-species groups but lived in the same habitat. We found significant support for  of the cases. Our results emphasize the importance of assessing similarity by using multiple raters and that several mechanisms may drive phenotypic resemblance, some reflecting phylogenetic inertia or habitat constraints. Nevertheless, these mechanisms may be accentuated in groups of birds that interact extensively. Received  January , accepted  May . Key words: avian plumage, mimicry, mixed-species flocks, phylogenetic independence, ratings.

Mimétisme du plumage dans les volées mixtes d’oiseaux : plus ou moins que les apparences? R.—Le mimétisme social a été postulé comme un mécanisme évolutif possible pouvant causer une convergence des caractéristiques chez les espèces qui interagissent considérablement et, ainsi, agir comme une force opposée à la compétition et au glissement de niche. Les exemples pour cette hypothèse proviennent surtout de ressemblances visuelles entre les espèces d’oiseaux qui forment des volées mixtes. Nous avons évalué le mimétisme du plumage entre des groupes d’oiseaux, en demandant à  évaluateurs de marquer les ressemblances entre  ensembles de mimétisme putatif postulés au cours des quatre dernières décennies. La ressemblance entre le modèle putatif et l’imitateur (s’il n’y avait aucun modèle hypothétique, nous avons considéré que les deux espèces pouvaient être des modèles l’une pour l’autre) a été comparée à celle entre le modèle et () une espèce apparentée à l’imitateur et () une autre espèce qui n’est qu’occasionnellement associée aux groupes multi-spécifiques mais qui vit dans le même habitat. Nous avons trouvé des appuis significatifs à l’hypothèse pour  des cas. Nous résultats soulignent l’importance d’évaluer la similarité en utilisant plusieurs évaluateurs et que plusieurs mécanismes peuvent guider la ressemblance phénotypique, certains reflétant une inertie phylogénétique ou des contraintes de l’habitat. Néanmoins, ces mécanismes peuvent être accentués dans les groupes d’oiseaux qui interagissent considérablement. M   convergence of phenotypic traits over evolutionary time between a model and a mimic that imitates the model to gain some fitness benefits (Ruxton et al. ). Such convergence may involve acoustic, chemical, or visual traits in animals and can be displayed statically (i.e., the mimicked signal is always displayed) or dynamically (i.e., the mimicked signal is part of an ephemeral behavioral display; Norman et al. ). Mimicry has evolved for many reasons; the best studied may be avoidance of attacks by predators that have learned to associate particular traits with unpalatability in another species (Wickler ).

Although the classic cases of mimicry involve species that occur together in the same area (Pfennig et al. ), convergence among species that not only live in the same area but forage together in groups has also been hypothesized in many animal species. In birds, for instance, convergence in dynamic acoustic traits (e.g., Ficken , Chu , Goodale and Kotagama ) and dynamic (e.g., wing-flicking patterns; Moynihan ) or static visual traits (e.g., Moynihan , ; Willis ; Cody ; Diamond ) has been reported in mixed-species flocks. We will focus here on static

3

E-mail: [email protected]

The Auk, Vol. , Number , pages  . ISSN -, electronic ISSN -. ‘  by The American Ornithologists’ Union. All rights reserved. Please direct all requests for permission to photocopy or reproduce article content through the University of California Press’s Rights and Permissions website, http://www.ucpressjournals. com/reprintInfo.asp. DOI: ./auk..

— 487 —

488

— BEAUCHAMP

visual traits, which are both the most widely reported and the most straightforward to test. Although mimicry in some avian cases may involve signals of unpalatability (Dumbacher et al. ), other factors appear to play a more prominent role in birds. Potential factors include greater cohesion among flock members due to economy of signals that have to be learned to coordinate different species in a moving flock (Moynihan ) and the “oddity effect,” in which species that look different from others in the flock may be targeted more often by a predator (Barnard , Landeau and Terborgh ). Mimicry in flocks has also been linked to dominance, because a subordinate species may benefit from resembling a dominant species to reduce interspecific aggression while they are feeding together (Diamond ). To be convincing, cases of putative mimicry should eliminate two alternative hypotheses. First, because cases of mimicry deal with species living in the same habitat, it is important to consider independent evolution of similar traits in each species in response to shared habitat requirements (Willis b, Burtt and Gatz ). For instance, many unrelated species that forage at sea have a mainly white plumage (Tickell ). Similarity in plumage in such species probably reflects adaptations to a shared habitat. Second, species that are closely related may have inherited similar traits from a common ancestor. Shared ancestry may thus lead to similarity among closely related species even when they occur in different habitats. A further methodological challenge that also arises when investigating putative cases of mimicry is the simple identification of convergence among species. As far as we can tell, almost all cases of visual mimicry have been identified by human eyes (for an exception, see Tobias and Seddon b) even though the targets of mimicry are most often avian predators or companion species in the same flock (Cuthill and Bennett ). It is now clear that avian vision is distinctive from human vision in that birds have different color receptors in their eyes and, thus, a broader range of perception (Cuthill ). Nevertheless, there is broad overlap in avian and human vision such that researchers have argued that although humans may miss certain subtleties in avian coloration, our perception of color is a good approximation of avian vision (Seddon et al. ). This approximation may be sufficient to identify putative mimicry cases, given that visual mimicry is not expected to produce a perfect match between mimic and model species (Sherratt ). This may be particularly relevant in mixed-species flocks in which predators can only catch glimpses of their fast-moving prey during an attack. In addition, convergence to some key plumage traits, rather than a close match to all traits, may be all that is needed to coordinate flock movements. Perhaps a more insidious consequence of using human eyes to evaluate convergence is that perception of resemblance may be specific to the samples evaluated and, more importantly, may vary from one person to another. To our knowledge, such biases have not been evaluated in the mimicry literature. Here, we examine putative cases of static visual mimicry in mixed-species avian flocks in light of the aforementioned issues. We asked several raters to evaluate resemblance between the plumages of different species in a set. In addition to a model and a putative mimic species, each set also included a control species to test for similarity caused by habitat sharing and a control

AND

GOODALE —

AUK, VOL. 128

species to test for common ancestry. The control species for habitat sharing, which we refer to as “habitat control species,” was a species that occurred in the same flock but at a lesser frequency than the model and putative mimic species. Therefore, this species essentially shared the same habitat and thus experienced similar pressures on plumage patterns. Nevertheless, because it occurred less frequently in the flock with the model and mimic species, the evolutionary pressures to converge in plumage, as a result of the oddity effect or economy of signals for instance, should be reduced. The control species for shared ancestry, which we refer to as the “ancestry control species,” was a species closely related to the putative mimic species, but which occurred in a different geographic area than the model species. Therefore, the ancestry control species was unlikely to have been exposed to the factors that promoted mimicry in the putative mimic species. Putative mimicry would be more convincing if the putative mimic was a closer match to the model than both the habitat and ancestry control species. We asked not one but several raters to evaluate resemblance between species. Pooling results across many raters should reduce the effect of idiosyncratic biases. Raters were blind to the nature of the project, which we believe is essential to provide a reliable test of our hypotheses. In addition, we replicated the samples to avoid results specific to a particular sample. With this procedure, we believe that we provide a rigorous test of mimicry in birds. M ETHODS Search strategy and definition.—We searched the literature to find published accounts of putative mimicry in mixed-species avian flocks or aggregations at resources or nesting sites. All cases of mimicry had to involve at least two species not only living in the same habitat but foraging together. A species set included a model species, a mimic species, an ancestry control species, and, when possible, a habitat control species. Some species sets involved only two species, and in such cases no habitat control species from the same flock could be assigned. For species sets with many potential habitat control species, we selected one species randomly. We used recent molecular phylogenetic trees along with distribution maps to select a closely related species that did not have an overlapping distribution (Table ). We did not look at photographs or drawings of any of the species in each set until all species were assigned a status. Data sets.—We could not always determine the status of a species as model or mimic, because authors sometimes remarked only on the similarity between the two species without discussing the probable evolution of mimicry in this system. Therefore, we created two data sets using the same species but switched the labels “model” and “mimic” between the two species involved in mimicry. These two data sets were replicated once. In replicate , we used photographs of each species in the field, except once when a drawing was used instead because we could not find a photograph. Photographs were taken from the Internet and had to be closeups. Replicate  consisted of the same species sets, but this time we used scanned drawings of the same species from field guides. Photographs or drawings for each species set were arranged on a page with the model at the top and the remaining species, namely the mimic, the habitat control species, and the ancestry control

J ULY 2011

— R EASSESSMENT

OF

M IMICRY

IN

B IRDS —

489

TABLE 1. List of the species in each set used to evaluate plumage mimicry in birds. The status of each bird in a set is presented along with the references describing mimicry and those used to determine the phylogenetic (ancestry) control (see text for the different types of set). Species role Set

Model

Mimic

1a: Ambiguous model

Sporophila corvina Volatinia jacarina

Volatinia jacarina

Myrmotherula fulviventris Cathartes aura Calidris pusilla

Tachyphonus luctuosus Buteo albonotatus Calidris mauri

Agelaius phoeniceus Tangara icterocephala Chlorospingus ophthalmicus Diglossopis cyanea Tangara vassorii

Quiscalus quiscula

Conirostrum sitticolor Delathraupis castaneoventris Sphecotheres viridis Phrygilus unicolor Diuca diuca Carduelis barbata Phrygilus gayi

Delothraupis castaneoventris Conirostrum sitticolor Oriolus sagittatus

Muscisaxicola flavinucha Muscisaxicola albilora Fulica leucoptera Fulica rufifrons Oriolus bouroensis forsteni

15: Presumed model

Muscisaxicola albilora Muscisaxicola flavinucha Fulica rufifrons Fulica leucoptera Philemon molucenssis subcorniculatus Oriolus szalayi

16: Presumed model

Philydor rufum

17: Presumed model

Anabazenops fuscus Poecile atricapillus

1b: Ambiguous model 2: Presumed model 3: Presumed model 4: Probe 5: Probe 6a: Ambiguous model 6b: Ambiguous model 7a: Ambiguous model 7b: Ambiguous model 8a: Ambiguous model 8b: Ambiguous model 9: Presumed model 10a: Ambiguous model 10b: Ambiguous model 11a: Ambiguous model 11b: Ambiguous model 12a: Ambiguous model 12b: Ambiguous model 13a: Ambiguous model 13b: Ambiguous model 14: Presumed model

18: Presumed model 19: Presumed model 20: Presumed model 21a: Ambiguous model 21b: Ambiguous model

Pitohui ferrugineus Sporophila luctuosa Picoides pubescens Picoides villosus

Sporophila corvina

Chlorospingus ophthalmicus Tangara icterocephala Tangara vassorii Diglossopis cyanea

Diuca diuca Phrygilus unicolor Phrygilus gayi Carduelis barbata

Pycnopygius stictocephalus Orchesticus abeillei Biatas nigropectus Vermivora chrysoptera Pomatostomus isidorei Conothraupis speculigera Picoides villosus Picoides pubescens

References

Habitat control species

Ancestry control species

Sporophila nigricollis Sporophila nigricollis Eucometis penicillata Charadrius semipalmatus Molothrus ater Basileuterus melanogenys Basileuterus melanogenys Thraupis cyanocephala Thraupis cyanocephala Margarornis squamiger Margarornis squamiger Philemon citreogularis Phrygilus gayi Phrygilus gayi

Phylogeny

Mimicry

Tiaris olivacea

1, 2, 3

4

Sporophila leucoptera

1, 2, 3

4

Tachyphonus rufiventer Buteo swainsoni Calidris minuta

5

4, 6, 7

8 11

9, 10 12, 13, 14

Agelaius xanthomus

15

16, 17, 18

Chlorospingus semifuscus Tangara schrankii

19, 20

7, 21, 22

19, 20

7, 21, 22

Tangara dowii

19, 23

7, 21, 24

Diglossa baritula

19, 23

7, 21, 24

Dubusia taeniata

19, 25, 26

7, 21, 24

Conirostrum bicolor

19, 25, 26

7, 21, 24

Oriolus flavocinctus

2, 27

28

Lophospingus pusillus Haplospiza unicolor Phrygilus punensis Carduelis xanthogastra Muscisaxicola juninensis Muscisaxicola alpina

26 26 26, 39 26, 39

29 29 29 29

31

29

31

29

32 32 33

29 29 34

Creurgops verticalis

35

7, 36, 37, 38

Sakesphorus canadensis Limnothlypis swainsonii Pomatostomus superciliosus Pyrrhocoma ruficeps

39

38

40

41

42

43

44

7, 45

Picoides stricklandi

46

47

Picoides scalaris

46

47

Fulica gigantea Fulica gigantea Oriolus chinensis

Lichenostomus flavescens Orthogonys chloricterus Attila rufus

Lichenostomus flavescens

Sphyrapicus varius Sphyrapicus varius

Certhionyx variegatus

(continued)

490

— BEAUCHAMP

AND

GOODALE —

AUK, VOL. 128

TABLE 1. Continued. Species role Set

Model

Mimic

22: Presumed model

Egretta thula

23a: Ambiguous model

Arremon brunneinucha Pipilo ocai

Egretta caerulea (white immature) Pipilo ocai

23b: Ambiguous model 24: Probe

Philomachus pugnax (female)

25: Presumed model

Pomatostomus isidorei Icterus galbula (female)

26: Probe

Habitat control species

Ancestry control species

Egretta tricolor

Egretta novaehollandiae Pipilo erythrophthalmus Arremon castaneiceps Philomachus pugnax (non female-like male) Conopophila rufogularis Icterus galbula (adult male)

Arremon brunneinucha Philomachus pugnax (female-like male) Timeliopsis griseigula Icterus galbula (subadult male)

References

Pitohui cristatus

Phylogeny

Mimicry

48

49, 50

51, 52

29, 53

51, 52

29, 53

54

55

56

57

References: (1) Lijtmaer et al. 2004; (2) Jønsson and Fjeldså 2006; (3) Klicka et al. 2007; (4) Moynihan 1960; (5) Burns and Racicot 2009; (6) Gradwohl and Greenberg 1980; (7) Isler and Isler 1999; (8) Lerner et al. 2008; (9) Willis 1963; (10) Mueller 1972; (11) Thomas et al. 2004; (12) Ashmole 1970; (13) Stawarczyk 1984; (14) Tripp and Collazo 1997; (15) Johnson and Lanyon 1999; (16) Williamson and Gray 1975; (17) Brugger et al. 1992; (18) Werner et al. 2005; (19) Burns and Naoki 2004; (20) Weir et l. 2008; (21) Moynihan 1968; (22) Buskirk et al. 1972; (23) Mauck and Burns 2009; (24) Remsen 1985; (25) Burns et al. 2003; (26) Klicka et al. 2007; (27) Norman et al. 2009; (28) Beland 1977; (29) Cody 1970; (30) Cardoso and Mota 2008; (31) Chesser 2000; (32) Livezey 1998; (33) Driskell and Christidis 2004; (34) Diamond 1982; (35) Fjeldså and Rahbek 2006; (36) Davis 1946; (37) Willis 1976a; (38) Willis 1989; (39) Tobias and Seddon 2009a; (40) Lovette and Hochachka 2006; (41) Ficken and Ficken 1974; (42) Dumbacher et al. 2008; (43) Diamond 1987; (44) Burns 1997 (45) Witt 2005; (46) Weibel and Moore 2005; (47) Morse 1970; (48) McCracken and Sheldon 1998; (49) Caldwell 1981; (50) Caldwell 1986; (51) Cadena et al. 2007; (52) DaCosta et al. 2009; (53) Short 1961; (54) Jukema and Piersma 2006; (55) Driskell and Christidis 2004; (56) Bell 1983; (57) Flood 1984.

species, at the bottom. We took care to change the order of the non-model species from one species set to another. Rating procedure.—The whole data set consisted of four files, each containing the same species sets. There were two files for each replicate because we duplicated cases with ambiguous models, as explained above. These files were sent to colleagues in four countries—the United States, Australia, China, and Sri Lanka— ensuring that each country received at least one full set of four files. Some countries received more files than others. Each person rated photographs or drawings from one file only. To ensure that ratings were blind, all raters were unaware of the purpose of the study. For the purpose of rating, the species at the bottom in each species set were called “templates” and those at the top “samples” so as to avoid giving away the purpose of the study. The task of each rater was to rate each template species, independently of the others, according to its resemblance to the sample species. Resemblance scores took on the following values: , , , and , with  being a very close match and  a very poor match. To facilitate scoring, we suggested to raters that they mentally break down each species into  parts. If resemblance was very good in nine or more of these parts, the score should be ; if resemblance was good between five and eight parts, the score should be ; if resemblance was good in two to four parts, the score should be , and if resemblance was good in fewer than two parts, the score should be . We specified that two criteria for matching should be used: plumage color and plumage pattern (spots, bars, etc.). We asked raters to ignore bird size and shape because these could not be scaled reliably for each species. We asked that raters complete the task alone in about  min.

Probes.—Within each file, we also inserted four probes, each including a full species set. One set consisted of sandpipers in the genus Calidris (Scolopacidae). Calidris species from around the world are mostly streaked brown, with a white belly (Prater et al. ). The model and putative mimic that we selected, which are very often found in the same flocks in the non-breeding season (Ashmole ), are very similar and many birders cannot distinguish these different species at a distance (Cartar ). This probe represents a case in which phylogenetic constraints can explain the resemblance between two species, because even species that do not currently live in the same geographic area still look very similar. We expected that both the putative mimic and the ancestry control species should be judged a good match to the model species. Another set consisted of blackbirds (Icteridae), which often have blackish plumage (Jaramillo and Burke ). These blackbirds of open country flock in large numbers after breeding (White et al. ). The resemblance in this case between the putative mimic and the model can be explained by habitat requirements, shared ancestry, or both. We expected that both the mimic and the two control species would be judged a good match to the model species. The third probe consisted of breeding male and female Ruffs (Philomachus pugnax; Scolopacidae). In Ruffs, breeding males come in several morphs, one of which is colored like a female, and typical males cannot distinguish this female-like morph from females (Jukema and Piersma ). In the fourth probe with Baltimore Orioles (Icterus galbula), dull subadult males are similar to adult females and adult males cannot distinguish between the two (Flood ). These two last probes thus represent cases in which

J ULY 2011

— R EASSESSMENT

OF

bird eyes are deceived by mimicry. Therefore, we expected that human eyes should be similarly deceived. Statistical analysis.—We compared the resemblance score of the putative mimic to the model with the resemblance score of the habitat or the ancestry control species to the same model. The habitat contrast consisted of the difference between the resemblance score for the putative mimic and that for the habitat control species. The phylogeny contrast consisted of the difference between the resemblance score for the putative mimic and that for the ancestry control species. As noted earlier, the habitat contrast could not be calculated in flocks that included only two species. The Wilcoxon signed-ranks tests (signed rank statistic S) was used to determine whether the median contrast value differed statistically from zero, in which case a statistically significant shift in the distribution of resemblance scores between the two species involved in the contrast took place. Negative contrasts indicate that the mimic was judged to be more similar to the model than either the habitat control species or the ancestry control species, depending on the type of contrast used. For species sets in which the putative mimic was determined a priori, we used the full data set (i.e., all four files) to test our hypotheses. For species in which such a distinction could not be made, we performed the statistical analysis separately for each model–mimic combination. R ESULTS We obtained data for  species sets in which mimicry was suspected and included four probes among these species sets for a total of  species sets per file. We did not obtain data from one rater among the  that we contacted. The list of species is provided in Table  along with the references that we consulted to identify putative cases of mimicry and to determine the identity of the ancestry control species in each set. Blackbird probe.—When assessed against the model blackbird, the median difference between the putative mimic resemblance score (median [min–max] =  [–]) and the resemblance score of the habitat control species (median [min–max] =  [–]), which occurs in the same habitat but is not very likely to flock with the model and the mimic species, was not significantly different from zero (S = –., P = ., n = ). The putative mimic and habitat control species were thus considered an equal match to the model species. The ancestry control species, however, was judged quite differently from the model (median [min–max] =  [–]; S = , P < ., n = ). Sandpiper probe.—When assessed against a model Calidris, the median difference between the putative mimic resemblance score (median [min–max] =  [–]) and the ancestry control species resemblance score (median [min–max] =  [–]) was not significantly different from zero (S = –, P = ., n = ). Here, both the putative mimic and the ancestry control species were considered close matches to the model sandpiper. Ruff and oriole probes.—When assessed against a female Ruff, the median difference between the female-like breeding-male resemblance score (median [min–max] =  [–]) and a typical breeding non-female-like male resemblance score (median [min– max] =  [–]) was significantly smaller than zero (S = –, P =

M IMICRY

IN

B IRDS —

491

., n = ). When assessed against an adult female Baltimore Oriole, the dull-subadult-male resemblance score (median [min– max] =  [–]) was significantly smaller than the resemblance score of an adult male (median [min–max] =  [–]) (S = –, P < ., n = ). In both cases, the putative mimic was judged a close match to the model. Cases with presumed models.—If the presumed model was identified using information from the literature, the resemblance score of the mimic should be smaller than those of the habitat control species and the ancestry control species. In the  sets of species with presumed models, the null hypothesis was rejected for both contrasts in the predicted direction in nine sets (Table ). Cases with ambiguous models.—Here, again, we expected that the resemblance score of these mimic species should be smaller than the score for the two control species. These hypotheses were tested in the two sets of species, each assuming a different model, given that we could not determine a priori which species was the model. Combining results from the two sets, the null hypothesis was rejected for both contrasts in the predicted direction in five sets (Table ) of  cases, including  cases (Tangara–Diglossopis, Picoides spp., and Arremon–Pipilo) in which evidence of convergence in plumage occurred in both species involved in mimicry. D ISCUSSION Our rating procedure reduces two common types of errors associated with subjective evaluation: () the rater is not blind as to the research hypothesis and can thus show biases unconsciously and () the rating is done by only one person. By using raters who were blind to our predictions, we obtained a more objective assessment of resemblance among species. As witnessed by the large range of resemblance scores for most species across raters, we conclude that raters do not necessarily assess the evidence similarly, and therefore several and possibly many raters are necessary to obtain an overall picture. Results from the probes support the case for a consideration of alternative hypotheses for the evolution of resemblance between species. We found that the putative mimic was as closely matched to the model as the habitat control species in one probe and as the habitat and ancestry control species in the other. It is therefore important to consider the effect of habitat sharing and phylogeny when evaluating mimicry between species. Results from the other two probes indicate that human raters perceived a great resemblance between a putative mimic species and the model in cases in which avian subjects in the field also cannot make the same distinction. Although we are dealing with only two cases, the results are consistent with the hypothesis that our vision is a good approximation of avian vision for cases of mimicry in birds. This probe procedure could be used to investigate other suspected cases of mimicry in birds. In their classic articles, Moynihan () and Cody () postulated that “social mimicry” could provide a force in opposition to competition, driving the characteristics and niches of animals closer together, rather than farther apart. Here, we show that many purported cases of mimicry may need reevaluation: in  of our  cases, the resemblance between the two species was not significantly greater than resemblances to closely related species or to other species in the same habitat that do not interact extensively

492

— BEAUCHAMP

AND

GOODALE —

AUK, VOL. 128

TABLE 2. Evaluation of mimicry through resemblance scores (RS) in sets of species including a putative mimic, a model, and controls for ancestry and habitat (n = 31).

Species set (model– putative mimic) Myrmotherula fulviventris– Tachyphonus luctuosus Cathartes aura–Buteo albonotatus Sphecotheres viridis– Oriolus sagittatus Philemon moluccensis– Oriolus bouroensis Oriolus szalayi– Pycnopygius stictocephalus Philydor rufum– Orchesticus abeillei Anabazenops fuscus– Biatas nigropectus Poecile atricapillus– Vermivora chrysoptera Pitohui ferrugineus– Pomatostomus isidorei Sporophila luctuosa– Conothraupis speculigera Egretta thula–Egretta caerulea (white immature) Pomatostomus isidorei– Timeliopsis griseigula

Median Median Is RS smaller in the Median resemblance score resemblance score Is RS smaller in the putative mimic than resemblance score for habitat control for ancestry control putative mimic than in the ancestry for putative mimic species species in the habitat control control species (min–max) (min–max) (min–max) species (S (P))? (S (P))? Conclusion 4 (1–4)

4 (2–4)

3 (1–4)

25 (0.29)

Not likely

2 (1–4)



3 (2–4)



–177.5 (