Patterns of Dominance and Aggressive Behavior in Blue Jays at a Feeder ’

The Condor 99~4346444 0 The Cooper Ornithological Society 1997

PATTERNS OF DOMINANCE AND AGGRESSIVE BEHAVIOR IN BLUE JAYS AT A FEEDER ’

KEITH A. TARVIN~ AND GLEN E. WOOLFENDEN Department of Biology, University of South Florida, Tampa, FL 33620 and Archbold Biological Station, P.O. Box 2057, Lake Placid, FL 33862

Abstract. We studiedinteractions among Blue Jays(Cyanocitta cristuta) visitinga feeder in south-centralFlorida over a 4-year period to examine the influence of sex, time of year, and body size on dominance and aggression,describe changesin dominance among individuals over time, and test for the presence of linear dominance hierarchies. Males dominated females throughout the annual cycle, and in all 24 of the male-female significant dyads. We cannot reject the hypothesisthat male dominance over females results from the larger body size of males. We infer that males also were more aggressive than females becausethey were involved in more interactionsthan expected by chance. Females became more, and males became less, aggressive immediately prior to the breeding season, but fluctuationsin aggressiondid not lead to shifts in intersexual dominance. Dominance relationships among a few high-ranking males were intransitive and changed over time. Dom- inance hierarchies, characterizedby reversals, circular triads, and unknown relationships, were not linear. Whereas linear hierarchies have been shown to exist in New World jays that live in small, stable social groups, we suspectthe variable constituencyand instability of Rocksprecludes the emergence of strictly linear hierarchiesin the genus Cyanocittu. Key words: aggression, Blue Jay, Cyanocittacristata, dominance hierarchies, New World jays, seasonal@, sex-biased dominance.

INTRODUCTION or engage in social interactions (Hardy 1961). Although the Blue Jay (Cyanocittu cristatu) is a However, the flocks vary in membership and are common and familiar eastern North American spatially and temporally ephemeral (Cohen species,remarkably little is known about its ba- 1977). These social patterns hold for the non- sic ecology and social behavior, including dom- migratory population in south-central Florida, inance relations. Only Racine and Thompson which we have studied since 1990. Our present (1983) have reported on dominance in Blue study of dominance and aggression at a feeder Jays, and although they found that rank order of is relevant because members of different pairs jays visiting a feeder was constant over a period often have the opportunity to interact and do so of a few weeks, they did not provide information as they encounter each other on more or less on other features of dominance relationships in “neutral” ground when foraging or harvesting their study population. Here we report on Blue acorns. Because Blue Jays do not defend or Jay dominance and aggression as it relates to maintain exclusive territories, individuals en- sex, time of year, and body size, describe counter many other individuals from within and changes in dominance relationships among in- beyond the local neighborhood, and perhaps en- dividuals over time, and test for the presence of gage with them in competition for food, nest linear dominance hierarchies. Analysis of the in- sites, etc. Thus, although attracting jays to an fluence of site on dominance was outside the artificial food sourceto observe interactions may scope of our study. quantitatively enhance levels of aggression or The basic social unit of Blue Jays is the frequency of interactions, it probably does not breeding pair. These pairs often share spacewith change the quality of the interactions because it other pairs and individuals, and these pairs and mimics the social and spatial mileu within which individuals often coalesce into flocks to forage Blue Jays normally exist. For the congeneric Steller’s Jay Cyanocittu stelleri, Brown (1963) reported dominance as ’ ’ Received 3 July 1996. Accepted 12 December site-dependent, rather than dependent on fixed 1996. ZCorresponding author: Keith A. Tarvin, Archbold social relationships (sex excluded). In contrast, Biological Station, PO. Box 2057, Lake Placid, FL many other New World jays live in small stable 33862, e-mail: [email protected] groups apparently organized around families.

[4341 DOMINANCE IN BLUE JAYS 435

Linear dominance hierarchies have been report- cealed position within a nearby house. By man- ed in three group-living jays for which domi- ually operating the trap, which was otherwise nance is related to sex, age, and breeding status locked open when baited, we were able to catch of group members (Florida Scrub-Jay Aphelo- focal jays, yet allow other jays to enter and leave coma coerulescens,Woolfenden and Fitzpatrick the trap. Because we rarely attempted to capture 1977, Mexican Jay A. ultramarina, Barkan et al. jays (especially marked ones), jays rarely re- 1986, Pinyon Jay Gymnorhinus cyanocephalus, sponded to the trap with apprehension. Further- Marzluff and Balda 1992). However, an analysis more, those jays that were captured usually fed of the degree of linearity of hierarchies has yet again from the feeder within a day or so of their to be performed on Cyanocitta using recently capture. proposed methods (Appleby 1983, deVries 1995). A color-marking program where numer- MORPHOLOGY AND SEX DETERMINATION ous individual Blue Jays were sexed and mea- For most captured jays, we measured bill depth sured provided the opportunity to examine cor- and width at the anterior margin of the nares, relates of dominance and aggression in a popu- bill length from nares to tip, culmen, head length lation of resident Blue Jays in south-centralFlor- from occiput to tip of upper mandible, tarsus, ida, and to compare patterns of dominance in primary 7, central rectrix, and mass. Feather this species with those found in other New measurements were estimated to the nearest World jays. mm; remaining linear measurementswere to the nearest 0.1 mm. Mass was estimated to the near- METHODS est 0.1 g. Multiple measurements of the same STUDY SITE individual taken after the jay had reachedits second calendar year were averaged. We studied dominance interactions among indi- We determined sex of many Blue Jays by vidually color-marked Blue Jays at a feeder at presence of a brood patch (females only), lapa- Archbold Biological Station, Highlands County, roscopy, or, in one case, by behavior at nests. Florida (27”lO’ N, 81”21’ W), from June 1991 Other jays were sexedusing a logistic regression through December 1994. The feeder was located analysis. The logistic regression procedure cor- in human-modified parkland with scatteredlarge rectly classified 88% of 124 Blue Jays of known slashpines (Pinus elliottii), surroundedby native sex, and calculated the probability that each bird scrubby flatwoods and sand-pine scrub habitats of unknown sex was a female based on length (Abrahamson et al. 1984). We determined that of the head and central rectrix. The procedure adult Blue Jays at Archbold are year-round res- made no classification errors within the 15% idents, and that birds in their first year tend to tails of the probability distribution when it was wander during autumn (unpubl. data). Virtually applied to the sample of Blue Jays of known sex. all the jays we observed at the feeder were band- When the probability of being female was 2 ed, and unbanded immigrants that became reg- 0.85 or I 0.15, we classified individuals of un- ular visitors were captured and marked within a known sex as female or male, respectively. In- few days of their arrival. Although many jays dividuals with intermediate probability scores marked at the feeder were not seen again, a core were categorized as unknown sex. Twelve males set of marked jays was seen in most months dur- and one female were sexed using this procedure. ing each seasonof each year, and many of these We were unable to assign sex to 13 jays. core individuals were present for more than one year. Most observations were of these frequent ASSESSMENTOF DOMINANCE visitors, and our use of significant dyads (see From within the house we observed Blue Jays below) ensured that we did not unduly weight as they interacted at the feeding platform placed interactions of drifters and stragglers.The num- about 15 m in front of a window. A Potter trap ber of jays visiting the feeder during an obser- centered on the platform and locked open con- vation sessionranged from 1 to 5 in autumn to tained about 30 g of commercial bird seed or a 10 to 20 in spring and summer. slice of bread that could be obtained by only a We capturedjays in a 4-cell Potter trap placed few jays simultaneously. Blue Jays quickly on the feeding platform by using a string to drop learned to associatethe presence of food with a the door from a remote and essentially con- characteristicwhistle that we emitted as we bait- 436 KEITH A. TARVIN AND GLEN E. WOOLFENDEN ed the feeder, and food (and concomitant obser- verts and presence or absence of bars on the vations of interactions) usually lasted no longer greater upper secondary coverts following Dater than 15 min. Limited accessto a small amount [ 19701 and Bancroft and Woolfenden of food and the whistle alerting jays to its pres- [1982]).We considered one bird dominant over ence combined to concentratejay activity within another only when the win:loss ratio within the a short time span, and thus promoted agonistic dyad (the summary of interactions between interactions used to assessdominance. members of a pair) differed significantly from Observation periods usually were conducted random as determined with a two-tailed bino- l-3 times per day, typically in the early morning mial test. A binomial probability I 0.05 reflect- and around noon. Observation periods were ini- ed a significant dominance relationship (hence- tiated on most days during the study period, ex- forth “significant dyads;” see Barkan et al. cept when we were away. During August 1986). through November, the responsivenessof jays to the whistle and to food provided by us sharply INFLUENCE OF SEX ON DOMINANCE AND declined as a result of the jays’ attention to har- AGGRESSION vesting and eating acorns. During that period, To test the influence of sex on dominance, we we usually did not attempt to initiate new ob- considered only significant intersexual dyads, servation periods if food remained on the feeder and used the two-tailed binomial test to deter- from a previous observation period. However, mine whether the proportion of those dyads we were attentive to the activities of the jays in dominated by jays of one sex differed from that the area, and when we saw them around, we expected by chance. To determine how the pro- attempted to entice them to the feeder with the pensity to interact differed between males and whistle and new food. Thus, our effort toward females, we calculated for each year the proba- observing interactions was generally constant bility that a given interaction would involve ei- throughout the year, although it was modified ther a male or a female, considering that each somewhatduring the acorn seasonin accordance interaction involved two participants, and con- with the responsivenessof the jays. Because we sidering the mean number of males and females observed no aggressive interactions of jays at present during months for which interactions the feeder during September and October, those were recorded. We calculated the total number months were deleted from our analysis of sea- of observed participations (= 2 times the total sonal patterns in aggression. number of interactions) and tallied the number We considered a dominance interaction to of times members of a given sex participated in have occurred when one jay either supplanted interactions, regardless of the sex of the birds another that was feeding or standing on the feed- with whom they interacted or whether jays of ing platform, or successfully resisted a supplan- the focal sex won or lost. We used the two-tailed tation “attempt” from another jay. In the first binomial test to determine whether the observed case, the supplanter was considered the winner proportion of sex-specific participations differed of the interaction; in the secondcase, the resistor from that expected by chance. We likewise ex- was the winner. During rare extended interac- amined the propensity of each sex to be involved tions, distinct supplantationswere consideredin- in significant dyads. Our method of calculating dependent. We made no attempt to assessthe expected frequencies of sex-specific interactions intensity of interactions, and the rare interactions would be improved by considering the mean with ambiguous outcomes or apparently reflect- number of males and females present during ing tolerance among individuals were not con- each observation period. However, because we sidered in our analyses. Because our view of the recorded number of individuals visiting the feed- area surrounding the feeding platform was re- er during each month, rather than during each stricted by the window frame, we did not record observation period, we are unable to generate the presenceof individuals that perched near the more precise expected values. feeder but did not visit it. We included in our We determined how the propensity of males analyses only interactions that occurred between and females to interact varied throughout the an- jays that had reached or surpassedtheir second nual cycle by comparing the monthly distribu- calendar year of age (AHY; ages were inferred tion of sex-specific participations with that of from the color of the greater upper primary co- total interactions, each summed over all 4 years. DOMINANCE IN BLUE JAYS 431

We used a x2 goodness-of-fit test to determine or greater than the mean size differential ob- whether the distributions of sex-specific partici- served in male-female dyads. pations matched that of total interactions. We tested whether birds of a given sex were DOMINANCE HIERARCHIES more likely to be involved in significant dyads We analyzed annual dominance matrices com- with birds of the same or opposite sex. When a posed of jays that were involved in at least one female is involved in a significant dyad, the significant dyad. This criterion served to cull probability of her being in a dyad with a male from the matrix transients and other individuals is that interacted only rarely, thereby reducing the number of uninformative unknown relationships (i.e., empty cells) within the matrix and increas- ing the likelihood that we detected linearity if it is the total number of males that whereKales existed. Within dominance matrices, we consid- may potentially interact, and (Nremales- 1) is the ered one bird to dominate another only when the total number of females that may interact minus dyad was significant (Barkan et al. 1986). We 1, because the focal female cannot interact with followed deVries’ (1995) procedure and evalu- herself. The two-tailed binomial test was used to ated linearity of hierarchies within each year by assess whether the proportion of female-male calculating d, the number of circular triads with- dyads was greater or lesser than expected by in a dominance matrix, and h, Landau’s (1951) chance. We used the same procedure to deter- index of linearity, as described by Appleby mine whether the proportion of male-male dy- (1983). Appleby’s d and h were then trans- ads was greater or lesser than expected by formed as describedby deVries (1995) to derive chance. the unbiased estimates d ’ and h’, respectively. INFLUENCE OF SIZE ON DOMINANCE The parameters d ’ and h ’ are preferred over d and h because Appleby’s technique is overly Within a sex, body size may influence domi- conservative when matrices contain unknown nance relationships (e.g., Seamy 1979). We used relationships (devries 1995). The exact signifi- the Wilcoxon matched-pairs signed-ranks test to cance of d ’ (i.e., the significance of linearity) is examine the influence of body mass and tarsus generally determined via randomization tests. length, features commonly used as indices of However, based on a series of randomization relative body size within bird populations (Free- tests, deVries (1995) concluded that when the man and Jackson 1990), on the outcome of probability of Appleby’s d is less than 0.01, the male-male significant dyads. Male Blue Jays are P-value associatedwith deVries’ d ’ is less than slightly, but significantly, larger than females at 0.05, and the hierarchy is significantly linear at Archbold (unpubl. data), and therefore domi- (Y = 0.05. Likewise, if the probability of Apple- nance of males over females, or vice versa, by’s d is greater than 0.15, deVries’ d’ is greater could be an effect of size. If size per se is an than 0.05 and the hierarchy is not significantly important influence of intersexual dyadic out- linear. come, then it should similarly influence the outcome of intrasexual dyads. However, the conclu- RESULTS sion that size of a given character is not an im- From 1991 through 1994,48 Blue Jays were in- portant influence of dominance within intersex- volved in 1,263 interactions. Over the 4-year pe- ual dyads is valid only if it is unimportant for riod, 27 individuals were involved in at least one intrasexual dyads in which the size differential of 63 significant dyads. Of these 27 jays, 11 between members is at least as great as the mean were males, 8 were females, and 8 were of un- size differential between members of male-fe- known sex. male dyads. Therefore, we tested for effects of body mass, tarsus length, head length, and INFLUENCE OF SEX ON DOMINANCE AND length of the central rectrix (the latter two char- AGGRESSION acters were most useful in distinguishing males Table 1 presents significant dyads grouped by and females using logistic regression analysis) sex of dominant and subordinate. Males consis- on dominance in male-male dyads when the size tently dominated females. Of 316 interactions differential for selected characterswas equal to involving both a male and a female, fewer thar

\ A_ 438 KEITH A. TARVIN AND GLEN E. WOOLFENDEN

TABLE 1. Blue Jay dominance interactionsand significant dyads tallied over 4 years, groupedby sex of both members.

Number of individuals involved in slgniiicant dyads”

TOtal Stgnificant Dominant Subordinate TOtal Direction of dominance interactionr dyads” individuals individuals individuals’ Male over male 493 14 5 10 11 Male over female 313 24 I 8 15 Female over female 99 4 2 3 4 Female over male 3 0 0 0 0

*Numberof sigmficant dyads tallied over all 4 years does not correspond with analyses presented elsewhere for which dyads are tallied wthin single years. h 17 males and I5 females were observed in dommance interactions, but only 11 males and 8 females were involved in the 42 Ggniticant dyads for which sex of both members was known. ‘ Total indiwduals mav be less than the sum of dominant and wbordinate mchwduals because an indwidual may be dominant to some birds, while subordmate to others

1% were won by the female, and the male was 7 male-male and 5 male-female dyads, P = dominant in all significant intersexual dyads 1.00). (two-tailed binomial test, IZ = 24, P < 0.001). Dominance interactions were not distributed Males exhibited a greater propensity to inter- evenly throughout the year, and the propensity act than females. When controlling for the num- of either sex to interact throughout the annual ber of males and females present, males were cycle could not be predicted from the distribu- involved in far more, and females in far fewer, tion of total interactions (Fig. 1). Males inter- interactions than expected by chance in each of acted more frequently than expectedduring May the four years (Table 2). However, the pattern and June, and slightly less frequently from De- held in only two years when considered at the cember through March (Fig. la). Females inter- level of the significant dyad (Table 3). In 1992, acted more frequently than expected in March the observed proportion of female-female ver- and less frequently in April, May, and July (Fig. sus female-male significant dyads was unex- lb). Feeder visits, and therefore interactions, by pected by chance, with females more likely to females were especially uncommon during April be involved in dyads with males than with other and May when they were incubating and brood- females (two-tailed binomial test, it = 4 female- ing. To account for the absence of females during the breeding season, we removed April and female and 16 female-male dyads, P = 0.02). May interactions from the data set. Males still Sample sizes were insufficient for this test in interacted significantly more frequently in sum- other years. Males were involved in significant mer and less frequently in winter and spring. dyads with other males more often than expect- The pattern for females was reversed (Fig. lc ed by chance in one of three years for which and d). sample sizes were sufficient (two-tailed binomial test: 1992, IZ = 8 male-male and 16 male-fe- INFLUENCE OF SIZE ON DOMINANCE male dyads, P = 0.31; 1993, n = 6 male-male Neither body mass, tarsus length, head length, and 0 male-female dyads, P = 0.04; 1994, n = nor length of central rectrices significantly influ-

TABLE 2. Propensity of male and female Blue Jays to participate in dominance interactions.

f number of indiwduals Expected proportion Number of actual present per month* of participatmns” participations’

Yeat Male Female Male Female Male Female ” P 1991 6.5 8.3 0.44 0.56 140 84 5.51

*Based on months during which mteractmns were recorded. h Calculated under the null hypothesis that each sex shares an equal propensity to participate. ‘ Regardless of wins or losses. d Two-tailed binomial test. The normal approximation, corrected for continuity, was used because sample sizes were >25 (Siegel 1956) DOMINANCE IN BLUE JAYS 439

TABLE 3. Propensityof male and femaleBlue Jaysto be involvedin significantdyads.

Probability male is Number of dyads for Involved in dyad when which sex of both Observed number sex of both members members is known of positions Year is known* (number of positions)” occupied by males P 1991 0.44 3 (6) 6 0.01 1992 0.49 28 (56) 32 0.28d 1993 0.59 6 (12) 12 co.01 1994 0.62 12 (24) 19 0.09

‘Based on ratio of mean number of males and females presentaf the feeder per month. See Table 2. h Each dyad is composed of two positlons (i.e., a dominant and a subordinate or 2 jays of equal status) / Two-tailed bmomial test. dz = 1.09.

enced dominance within male-male dyads (Ta- DOMINANCE HIERARCHIES ble 4). For body mass and head length, size dif- Dominance hierarchies (Table 5) among the ferentials were as great or greater than the mean Blue Jays we studied were characterized by re- size difference of intrasexual dyads in only two versals (a subordinate wins an interaction male-male dyads. In these cases, the dominant against a typically dominant individual; Brown had the larger feature in one dyad and the small- 1975), circular triads (e.g., A is dominant to B, er feature in the other. No male-male dyads ex- B is dominant to C, yet C is dominant to A; isted in which the tail length differential was as Appleby 1983), and unknown relationships (lack great as the mean differential among male-fe- of a clear dominance relationship between in- male dyads. The difference in tarsus length be- dividuals arising from a lack, or insufficient tween members of nine male-male dyads was as number, of observed interactions between them; great or greater than the mean difference in tar- deVries 1995). Consequently, none of the hier- sus length calculated from male-female dyads. archies we constructed was linear (Table 6). The male with the longer tarsus was dominant Throughout our study, four or five males seemed in six of those nine dyads (Wilcoxon matched- to be of substantially higher rank than the re- pairs signed-ranks test: z = - 1.16, P = 0.25). maining jays, but dominance relationships

500 400 300 200 100 0

250

200

150

100

0 Jan Feb Mar Apr May Jun Jul Aug NW Dee Jan Feb Mar Jun Jul Aug Nov DBC

Month m observed Month m expected FIGURE 1. Distributionof interactionsof male and femaleBlue Jaysthroughout the annualcycle (a and b). In (c) and (d), datafrom the majorportion of thebreeding season (April and May) aredeleted. Male interactions are representedin graphsa and c; femalesin graphs b and d. Note the difference in the scale of the Y axis for males and females. We never observed interactions in the months of September or October; consequently those months were not considered when calculating goodness-of-fit tests. 440 KEITH A. TARVIN AND GLEN E. WOOLPENDEN

TABLE 4. Relative size of selected charactersof dominant and subordinatemembers of significant male-male dyads.

Mean (SD)

Character Dominant Subordinate n Z” P” Body massb 74.5 (1.0) 74.3 (5.2) 12 -0.24 ns Tarsusc 35.1 (0.5) 34.6 (1.3) 12 -1.10 ns Head lengthc 56.7 (0.5) 56.3 (1.0) 12 -0.94 IlS Tail lengthd 122.3 (3.4) 120.5 (3.2) 11 -1.16 ns

* Wilcoxon matched-pairs slgned-ranks test h Measured to nearest 0.1 g. c Measured to nearest 0.1 mm. *Measured to nearest 1.0 mm. among these high-ranking males changed hierarchy obtained (Brown 1963), it should not through time (Table 7). affect the influence of sex or season on dominance or aggression, nor the degree to which DISCUSSION hierarchies approach linearity within a single Although our study design encouraged jays to year interact within a restricted space for a brief pe- The most conspicuous pattern in our study riod, the social context within which the ob- was the unwavering dominance of males over served interactions occurred probably mirrors females. Such a pattern is consistent with those that of other encounters among the marked jays found in other studies of New World jays at our study site. The fundamental social unit of (Brown 1963, Woolfenden and Fitzpatrick 1977, the Blue Jay is the breeding pair, yet a pair Barkan et al. 1986, Marzluff and Balda 1992), shares most of its home range with other jays as well as many other passerine species (e.g., (Hardy 1961, Cox 1984). Therefore, most Blue Ketterson 1979, Piper and Wiley 1989; but see Jays when foraging or harvesting acorns poten- Samson 1977, and Belthoff and Gauthreaux tially encounter all the other individuals residing 1991 for converse patterns in the genus Curpo- within the neighborhood, as well as transients. ducus). Contrary to reports from a few studies Although the particular site at which observa- (Smith 1980 and references therein), dominance tions are recorded may influence the order of the relationships between male and female Blue

TABLE 5. Blue Jay dominance matrix from 1992. Hierarchy constructedfollowing Brown (1975), except only birds involved in at least one significant dyad were included. Wins are in rows, losses in columns. Numerals above each column correspondwith those adjacenteach individual jay in the far left column. Sex of individuals identified at the top of each column: m = male, f = female, u = unknown sex.

m m m m m f f f f f f Jay 1 2 3 4 : 6 7 : 9 l”o I1 12 1”3 14 15 1”6 1 P-PS 14b 21 44 5 6 6 14 5 2 2 12 4 5 13 3 2 P-BS - 9 1 2 5 2 5 14 - 1 3’2 10 1 - 3 G-RS - 12 1 1 2 6 4 2 3 14 4 G-GS - 16 2 9 - 9 11 14 2 5 23 5 19 - 2 5 G-OS - 6 - - 12 12-4 ; 2 - - 6 P-GS - - 2 - 1 2 16-- 1 3 3 10 7 R_YS ______4 14 1 1 23 3 2 1 - 8 G_PS ------13 1 6- 4-- 9 L_BS ______6 9 16 - 3 - - 10 R_GS ______3 _ 1 l--_- 11 BBS ______1 _ _ _ 1 _ _ 12 GG_S ______1 _ _ _ _ 1 _ _ 13 P-AS ______1 - - 14 G_WS - - - - _ _ - - - - _ 1 _ _ _ 15 _RGS ______16 GAS______

“Dominance matrices from other years of the study are available from the authors h Boldface values indicate number of wins within sigmficant dyads. DOMINANCE IN BLUE JAYS 441

TABLE 6. Characteristicsof dominance hierarchies among Blue Jays visiting a feeding station at Archbold Biological Station, 1991-1994. One bird was considereddominant over another only when the win:loss ratio within the dyad differed significantly from random. Hierarchies were constructedusing only birds involved in at least one significant dyad.

Total number Number of birds Number of Linearity statistics of interacting involved m significant Yeat birds significant dyads dyads d”’ h” pd 1991 29 5” 4 - 1992 25 16 38 114.9 0.325 ns 1993 23 10 14 24.5 0.406 ns 1994 15 9 13 17.1 0.434 ns

a Lmearity statistics for the 1991 matrix are omitted because a hierarchy containing fewer than six individuals cannot be shown to be significantly lmear (Appleby 1983). h d ’ is a.n unbiased estimate representing the number of circular triads present withm a dominance matrix, and it is significant when fewer circular triads occur than expected by chance based on the number of indlwduals in the dominance matrix. A dominance hierarchy cannot be linear when d IS not significant (Appleby 1983). d ’ = d - 0.25~. where d is the number of circular triads derived via Appleby’s (1983) method, and u IS the number of unknown relationships in the dommance matrix (deVries 1995). c h ’ is an unbiased index of linearity that ranges from 0 to 1, with 1 indicating perfect linearity. h ’ = h + (6/[N’ - N])u, where h = Landau’s Index of linearity (Landau 1951, Appleby 1983). N is the number of indwlduals in the dominance matrix. and II is the number of unknown relationships in the matrix (deVries 1995). li P-values derived for Appleby‘s d were > 0.15 for each dominance matrix. When P values derived ford are greater than 0.15, d’ is not significant at OL= 0.05 (deVries 1995).

Jays did not shift during the breeding season. fenden and Fitzpatrick 1977, Barkan et al. 1986, Because our data included few examples of Marzluff and Balda 1992, for similar patterns in male-male dyads exhibiting size differentials other New World jay species), as indexed by the equal to or greater than the average differential relatively high propensity of males to interact within intersexual dyads, we cannot reject the overall, and by the low frequency of observed hypothesis that male dominance over females is female-female interactions and significant dy- an effect of body size. However, male and fe- ads. Interestingly, even during the period of high male Blue Jays in our study population differ by female and low male aggression shortly before less than 5% for the several characterswe mea- the breeding season, females were consistently sured, and such slight sexual dimorphism is gen- subordinate to males. The overall rarity of fe- erally not considered to be an important influ- male-female interactions, and the fact that fe- ence on dominance (Smith 1976). males seem more likely to interact with males In addition to being dominant, male Blue Jays than with other females, suggestsmales may ini- were more aggressive than females (see Wool- tiate interactions as is the case in Florida Scrub-

TABLE 7. Stability of relationshipsamong 5 high-ranking male Blue Jays present in at least 3 of the 4 years of the study. Arrows indicate a significant direction of dominanceunless otherwise noted. “-” denotesneither bird won more than 3 interactions. “=” denotes at least 8 interactions were observed, but the win:loss ratio was not significant. “A” denotesBird B not observed in interactionsin that year.

Number of Ye= Stabibty of possible OVerall relationship interyear Changes Jay A Jay B 91 92 93 94 relationship through time” comparisons observed

G-GS G-RS - < - < < NC 1 0 P-PS > < > F=?z C 3 2 P-GS - - > ; > NC 0 P-BS > iii A > C? 1 l? G-RS P-PS < > z C 2 1 P-GS - - >b ; z ? 0 P-BS - < < A < NC 1 0 P-PS P-GS - > - > NC 0 P-BS > ~~C 0 - P-GS P-BS < - A = ? 0 Totals: 8 4~~

*NC = no change; C = change; ? = evaluation of change equivocal. h P i 0.063, therefore not included in the summary of possible interyear comparisons and changes observed. 442 KEITH A. TARVIN AND GLEN E. WOOLFENDEN

Jays (Woolfenden and Fitzpatrick 1977) and found that shifts in dominance of male Steller’s Mexican Jays (Barkan et al. 1986). Jays at particular sites correspondedwith relo- We suspectthe decreasein male, and the con- cation of nest sites during renesting attempts. comitant increase in female, aggressionprior to The significant changes in dominance over time the breeding season is a function of the nesting among high-ranking male Blue Jays in our study phenology and consequent energy demands of may have reflected similar shifts in nest sites in the jays, as suggested by Smith (1980). Most different years. Blue Jays at Archbold are nesting by early April, The social ranks of Blue Jays visiting our and offspring from first nests fledge by early to feeder were not organized linearly. In a species middle May. However, as nesting successis rel- with a social organization characterized by atively low and replacement nests consequently ephemeral flocks of varying constituency (Hardy are common, many jays continue to nest through 1961), such as the Blue Jays at Archbold, the May and June (pers. oberv.). Heightened ag- numerous circular triads, reversals, and shifts in gression of females at food sources in March dyadic dominance of high-ranking males over may reflect their attempts to obtain the extra cal- time would be expected (Brown 1975). This ories and nutrients needed to produce eggs. Sim- conclusion is supported by the findings of Ra- ilarly, heightened male aggression in May and tine and Thompson (1983), who reported the June may reflect the energy demands of nest- rank order within a flock of wintering Blue Jays lings, which are fed by males until about a week visiting an ad libitum feeder in Massachusetts before fledging. However, we did not directly was relatively stable over short (several weeks), test these ideas. but not long (several months), periods of time. Although sex strongly influenced the outcome Racine and Thompson (1983) also found both of intersexual interactions among Blue Jays vis- reversals and circular triads, a pattern consistent iting our feeder, factors influencing dominance with our observations. Our criterion that domi- relationships among males are complex. Body nance is ascribed only when the outcome of dy- size, or some correlate of body size, may influ- ads is significant may be considered overly con- ence the outcome of interactions to some degree, servative. However, had we simply considered but body size alone cannot fully explain the re- an uneven win:loss ratio as indicating domi- lationships we observed. This is best evidenced nance, we would have been forced to include far by the intransitivity of dominance among males more individuals in our matrices than just those over time, a pattern that precluded a strictly lin- involved in at least one significant dyad. The ear relationship between size and rank. The im- result would have been far more unknown re- portance of the relationship of body size to dom- lationships in the matrices, and consequently, inance varies among bird species, and is prob- even lower degrees of linearity than were ob- ably strongly influenced by other features of the served. biology of those species such as mating system The dominance matrices compiled by Brown and social organization. Other factors not mea- (1963) for the congeneric Steller’s Jay are strik- sured by us probably are better predictors of ingly similar to those we compiled for Blue Jays, Blue Jay intrasexual dominance than is size. in that they include reversals, circular triads, and Such factors may include location of nests many unknown relationships. Brown (1963) as- (Brown 1963, Wechsler 1988), familiarity with cribed inconsistencies within dominance matri- the study site (Glase 1973, Dhindsa et al. 1989), ces to first encounters, mistaken identities, and health (Weatherhead et al. 19954 variation in true changes in dominance among individuals, plumage signals (Holberton et al. 1989), age and we agree that such occurrencesprobably are (Woolfenden and Fitzpatrick 1977), and degree common in species that do not live in stable of relatedness of participants. Unfortunately, groups. Neither Brown (1963) nor Racine and most of the high ranking males in our study Thompson (1983) explicitly tested whether the were banded as unknown age birds in 1990; con- dominance hierarchies they observed were lin- sequently, we do not know whether age or site- ear. Although Racine and Thompson (1983) did familiarity differences existed among them. not provide actual dominance matrices, we an- Likewise, we knew where few individuals nest- alyzed those in Brown (1963) using the same ed, and were thus unable to examine effects of technique we employed for Blue Jays. The three nest proximity on dominance. Brown (1963) matrices in Brown (1963) that contained more DOMINANCE IN BLUE JAYS 443

than 5 individuals did not exhibit significant lin- Biological Station, Florida: an example of the earity (Brown’s Table 1: n = 22 jays, d ’ = southernLake Wales Ridge. Fla. Sci. 47:209-250. APPLEBY,M. C. 1983. The probability of linearity in 371.8, h’ = 0.160, P > 0.05; Table 5: n = 21, hierarchies.Anim. Behav. 3 1:600-608. d ’ = 330.8, h’ = 0.141, P > 0.05; Table 6: n = BANCROFT.G. ‘I, ANDG. E. WOOLFENDEN.1982. The 21. d ’ = 315.1. h’ = 0.182, P > 0.05). molt of Scrub Jays and Blue Jays in Florida. Or- In contrast to the unstable flocks df Cyano- nithol. Monogr. 29:1-51. _ BARKAN,C. I? L., J. L. CRAIG,S. D. STRAHL,A. M. citta, linear dominance hierarchies occur in sta- STEWART,AND J. L. BROWN.1986. Social domi- ble groups of Florida Scrub-Jays (Woolfenden nancein communalMexican JaysAphelocoma ul- and Fitzpatrick 1977) and Mexican Jays (Barkan trumarina. Anim. Behav. 34:175-187. et al. 1986). Analysis of dominance hierarchies BELTHOFF,J. R., AND S. A. GAUTHREAUX,JR. 1991. has been published for only one other New Aggression and dominance in House Finches. Condor 93:1010-1013. World jay, the Pinyon Jay (Marzluff and Balda BROWN.J. L. 1963. Aggressiveness,dominance and 1992), a species that lives in relatively stable, social organization in the Steller Jay. Condor 65: but huge groups (> 50 individuals). Using Ap- 460-484. pleby’s method, Marzluff and Balda (1992) BROWN, J. L. 1975. The evolution of behavior. W. W. found that a dominance hierarchy involving 14 Norton, New York. COHEN, S. M. 1977. Blue Jay vocal behavior. Ph.D. of the highest ranking males in the flock was diss., Univ. Michigan, Ann Arbor, MI. significantly linear. However, they did not em- Cox, J. A. 1984. Distribution, habitat, and social or- ploy significant dyads as indicators of domi- ganization of the Florida Scrub Jay, with a dis- nance. When we reanalyzed their dominance cussion of the evolution of cooperative breeding matrix (Marzluff and Balda 1992, Table 10) us- in New World jays. Ph.D. diss., Univ. Florida, Gainesville, FL. ing the more conservative significant dyad cri- DATER,E. 1970. Dorsal wing covertsof Blue Jay (Cy- terion and devries’ method to account for un- unocitta c&tutu): guide to age. EBBA News 33: known relationships, we found no significant 125-127. linearity (n = 14 jays, d ’ = 81, h’ = 0.29, P > DEVRIES,H. 1995. An improved test of linearity in 0.05). Based on the available analyses, linear dominancehierarchies containing unknown or-tied relationshios.Anim. Behav. 50:1375-1389. dominance hierarchies appear more likely to DHINDSA,M. S:, P E. KOMERS,AND D. A. BOAG. 1989. arise in New World jays when group member- The effect of familiarity with an environment on ship is stable and group size is relatively small. the dominancerelationships between juvenile and Together, these group characteristics provide a adult Black-billed Magpies. Ornis Stand. 20:187- 192. milieu within which dyadic relationships may FREEMAN,S., AND W. M. JACKSON.1990. Univariate solidify over time (Brown 1975). metrics are not adequateto measure avian body size. Auk 107:69-74. ACKNOWLEDGMENTS GLASE,J. C. 1973. Ecology of social organization in We extend deep appreciationto the directors and staff the Black-cappedChickadee. Living Bird 12:235- of Archbold Biological Station for providing us with 267. a study site and financial and logistical support. We HARDY,J. W. 1961. Studies in behavior and phylog- also thank Jan Woolfenden for allowing us to band eny of certain New World jays (Garrulinae).Univ. jays and conduct observationsfrom her-dining room, Kans. Sci. Bull. 42:13-149. Marv Garvin and Bert Remlev for helu with related HOLBERTON,R. L., K. I? ABLE, AND J. C. WINGFIELD. field-work, and Reed Bowman and Dirk Burhans for 1989. Statussignalling in Dark-eyed Juncos,Jun- helpful discussion.Jack Hailman and anonymousre- co hyemulis: plumage manipulationsand hormon- viewers critically read earlier drafts of the manuscript al correlatesof dominance.Anim. Behav. 37:681- and provided suggestionsthat greatly improved its 689. quality. We thank one anonymousreviewer for sug- KETTERSON, E. D. 1979. Aggressive behavior in win- gesting a way to estimate the frequency of year to year tering Dark-eyed Juncos: determinants of domi- reversals of dominance within dyads. The binomial nance and their possible relation to geographic tests were performed using Binomial Test version 1.0, variation in sex ratio. Wilson Bull. 91:371-383. 0 1987 by Bill Engels, Department of Genetics, Uni- LANDAU,H. G. 1951. On dominancerelations and the versity of Wisconsin. KAT was supportedin part by structureof animal societies. I: effect of inherent the Frank M. Chapman Memorial Fund of the Amer- characteristics.Bull. Math. Biophys. 13:1-19. ican Museum of Natural History and Archbold Bio- MARZLUFF,J. M., AND R. l? BALDA. 1992. The Pinyon logical Station. Jay: behavioral ecology of a colonial and cooperative corvid. T & A. D. Poyser, London. LITERATURE CITED PIPER,W. H., AND R. H. WILEY. 1989. Correlates of ABRAHAMSON,W. G., A. E JOHNSON,J. N. LAYNE,AND dominance in wintering White-throated Sparrows: l? A. PERONI.1984. Vegetation of the Archbold age, sex and location. Anim. Behav. 37:298-310. 444 KEITH A. TARVIN AND GLEN E. WOOLFENDEN

RACINE,R. N., AND N. S. THOMPSON.1983. Social SMITH,S. M. 1980. Henpeckedmales: the general pat- organization of wintering Blue Jays. Behaviour tern in monogamy?J. Field Omithol. 51:55-64. 87:237-255. WEATHERHEAD,l? J., K. J. METZ, D. SHUTLER,K. E. SAMSON, E B. 1977. Social dominance in winter MUMA, AND G. E BENNE~. 1995. Blood parasites flocks of Cassin’s Finch. Wilson Bull. 8957-66. and dominance in captive blackbirds. J. Avian SEARCY,W. A. 1979. Morphological correlates of Biol. 26:121-123. dominance in captive male Red-winged Black- WECHSLER,B. 1988. Dominance relationshipsin Jack- birds. Condor. 8 1:417-420. SIEGEL,S. 1956. Nonparametric statisticsfor the be- daws (Corvus monedula). Behaviour 106:252- havioral sciences.McGraw-Hill, New York. 264. SMITH,S. M. 1976. Ecological aspectsof dominance WOOLFENDEN,G. E., AND J. W. FITZPATRICK.1977. hierarchiesin Black-capped Chickadees.Auk 93: Dominance in the Florida Scrnb Jay. Condor 79: 95-107. 1-12.