A Field Study of Ornaments, Body Size, and Mating Behavior of the Gambel's Quail

12 Pages·2003·8.3 MB·English

by  Julie C Hagelin

Wilson Bull, 1 15(3), 2003, pp. 246-257
A FIELD STUDY OF ORNAMENTS, BODY SIZE, AND MATING
BEHAVIOR OF THE GAMBEL’S QUAIL
JULIE C. HAGELIN’ 2
—
ABSTRACT. Male Gambel’s Quail {Callipepla garnbelii) have strikingly ornate plumage. Yet, captive ex-
periments indicate that removing multiple ornaments does not necessarily alter patterns of female mate choice
or male-male competition. To test these unexpected results, I observed a wild population of banded quail for
three seasons to determine ornamental and body size traits associated with pairing date and winners of male
contests. I also documented mating behaviors (e.g., pairing date, mate fidelity, brood size). Consistent with
captive studies, male mass, rather than ornate plumage, was the primary feature related to winners of male
contests and early pairing. Heavier males paired earlier, regardless ofage, butdid not exhibit significantly larger
ornaments. Adults of both sexes were heavier and paired earlier than yearlings. Early pairing also correlated
positively with brood size, suggesting that heavy, early nesting birds experienced greater fitness. Mating behav-
iors were flexible across seasons. Social monogamy decreased from 83% in 1996 to 30% in 1998, while polyg-
amy (sequential, long term pairings) increased, particularly among yearlings and adult males. Adult females
were equally likely to exhibit social monogamy or polygamy each season. They also exhibited the highest
frequency ofearly pairing and the greatestkeel scores (ageneral measureofcondition), suggestingtheircapacity
for breeding was high. At least two adult females abandoned their first mate after hatching and re-paired, in an
apparent attempt to double brood. Received31 August 2002, accepted 28 February 2003.
The Gambel’s Quail {Callipepla garnbelii) laboratory findings by studying Gambel’s
is a highly sexually dimorphic game bird of Quail in the field. Wild populations typically
the arid southwestern United States. Males ex- are male biased (Brown and Gutierrez 1980),
hibit multiple plumage ornaments, such as a suggesting that ornaments may signal infor-
long head plume, light and dark belly patches, mation to discriminating females or male
and a rusty head patch (Johnsgard 1973). Con- competitors. I aimed to determine the orna-
trary to prediction, tests of captive Gambel’s mental and body size traits associated with (1)
Quail and other galliform birds indicate that early pairing, which I found to be positively
many extravagant feather traits of males do related to reproductive success, and (2) the
not play a primary role during female choice winners of male contests. I observed a popu-
(Beani and Dessi-Fulgheri 1995, Buchholz lation of banded, free-ranging Gambel’s Quail
1995, Ligon and Zwartjes 1995, Hagelin and for three seasons (1996-1998) and recorded
Ligon 2001, but see Calkins and Burley 2003) when mates paired and the outcome of ag-
or male-male competition (Ligon et al. 1990, gressive interactions between males. I pre-
Buchholz 1997, Hagelin 2()01a). Even the re- dicted that if male ornaments functioned dur-
moval of multiple ornaments in Gambel’s ing sexual selection, ornament size would cor-
Quail and Red Junglefowl {Callus gallus) did relate positively with early pairing or male
not alter patterns of female preference (Ligon status (Senar and Camerino 1998, Wolfenbar-
and Zwartjes 1995, Hagelin and Ligon 2001). ger 1999).
Instead, body size and testosterone-mediated A second goal of the study was to charac-
traits often are favored and may function as terize the mating behavior of Gambel’s Quail.
reliable signals of male quality or condition Though generally considered socially monog-
(Ligon and Zwartjes 1995; Berglund et al. amous (Johnsgard 1973), at least one account
1996; Hagelin 2001a, 2001b; Hagelin and Li- suggests that double brooding may occur
gon 2001). (Gullion 1956). Recent studies have docu-
The principal goal of this study was to test mented multiple broods in other North Amer-
ican quail, and breeding strategies appear to
be more flexible than previously surmised
que'rDqeupet,,NoMf B8i7o1l3o1g,y,USUAni.v. of New Mexico, Albu- (Curtis et al. 1993, Burger et al. 1995, Dele-
-Current address: Dept, of Biology, Swarthmore hanty 1995, Guthery and Kuvlesky 1998).
College, Swarthmore, PA 19081, USA; To assess mating behavior, I determined (1)
e-mail: jhageli @swarthmore.edu the proportion ofsocially monogamous versus
1
246
—
Hagelin • MATING IN THE GAMBEL’S QUAIL 247
polygamous pairs in the population, (2) the 1998 only. We calculated mean trait sizes for
sex and age classes ofbirds that exhibited dif- birds that were trapped and measured more
ferent breeding strategies, and (3) how mating than once. All quail handling was conducted
strategies changed across years. Climatic pa- under Animal Welfare Assurance #A4023-01
rameters, such as temperature and rainfall, are and Univ. of New Mexico Animal Care Pro-
thought to influence the fluctuations of quail tocol #9801-B.
populations (e.g., Raitt and Ohmart 1967, Pairing, brood size and hatching date.
1968; Campbell 1968; Heffelfinger et al. Using binoculars, we monitored quail from
1999; Guthery et al. 2001; Lusk et al. 2001). the BDA observation window from March
However, the underlying breeding patterns of through mid-August. Observations usually oc-
different sex and age classes are not well un- curred on alternate weeks during hours of
derstood. peak activity (dawn until late morning, late
afternoon until dusk). An observation period
STUDY AREA AND METHODS
typically spanned two consecutive days and
A field assistant and I monitored quail from totaled 10-30 h of monitoring. An individual
January 1996 to August 1998 at Bosque del was considered paired when it was seen for
Apache (BDA) National Wildlife Refuge, San >2 weeks (>2 observation periods) in close
Antonio, New Mexico (34° 49' N, 106° 53' proximity to the same individual of the op-
W). Quail commonly are seen foraging, sun- posite sex. The behavior of pairs was discern-
m
bathing, and drinking within 10 of a 3-m able from unpaired birds; pairs typically dis-
X 3-m observation window at the BDA visi- tanced themselves from others, followed each
tor’s center throughout the year. The habitat other while foraging, and the male usually
consists of open sandy soil interspersed with chased off all other quail. We used only those
dense shrubs (four-winged saltbush,Artemesia birds that exhibited obvious changes in pair-
tridentata). Refuge staff provided quail and ing status (e.g., unpaired before but consis-
other wintering birds with approximately 1 kg tently seen with mate after 23 April) in anal-
of wild bird feed (millet, cracked corn, milo, yses of pairing behavior.
sunflower seeds) daily between Decemberand We recorded the number and age of chicks
February. that accompanied banded birds. Chick age
—
Trapping and measurements. We trapped was estimated in 1-week increments. Raitt
quail near the visitor’s center from December (1961) described plumage-based age estimates
to February using funnel traps baited with for California Quail (C. californica), which
seed. Each bird received a unique combination are virtually identical to Gambel’s Quail (Raitt
of plastic color leg bands. We determined sex and Ohmart 1967). We estimated each brood’s
from the presence of ornate plumage and age hatching date by subtracting the age of a
(yearling or adult) from primary wing coverts brood (in weeks) from the date we observed
(Brown et al. 1998). Measures of body size it. To determine the median hatching date of
followed Hagelin and Ligon (2001) and in- the population more accurately, we also in-
cluded mass and length of tarsus, flattened cluded data for unbanded parents with newly
wing, tail, and culmen. We also palpated the hatched broods (<1 week old). We distin-
sternal keel (Gregory and Robins 1998) to ob- guished between new broods during each ob-
tain a general measure of body condition (fat servation period by the number of young. We
and muscle). We categorized a bird’s keel as: avoided problems of duplicate counting, be-
0 (no fat or muscle; sharp keel), 1 (some fat cause observation periods took place every
and muscle; moderately sharp keel), or2 (high other week. Newly hatched chicks of unband-
fat and muscle; dull keel). Ornament mea- ed birds grew markedly between these periods
surements also followed Hagelin and Ligon and were not counted again.
—
(2001). Briefly, we measured the flattened Male contests and mating behavior. Male
plume length of both sexes. We also calculat- aggression was common during each season.
ed the area (in mm^) of ornaments unique to We noted the band combination, the pairing
males (light and dark belly patches, rusty head status of each (whenever possible), and the
patch) by multiplying patch width and height. outcome ofencounters between pairs ofband-
We measured rusty head patch in 1997 and ed males. “Winning” males caused their op-
248 THE WILSON BULLETIN • Vol. 115, No. 3, September 2003
ponents to retreat from the immediate area. after the median pairing date, and age com-
“Losing” males either retreated immediately, pared traits between yearlings and adults.
or fled when an opponent escalated aggres- Stepwise discriminant analysis selected the
sion. specific traits that best described each signif-
We recorded the mating behavior ofbanded icant categorical variable (pairing date, age, or
MANOVA
birds during pairing and breeding (March year) from analysis. These traits
through mid-August). Individuals remaining subsequently were analyzed with ANOVA
with the same mate throughout this period and 2-tailed /-tests to determine how they var-
were considered socially monogamous. In ied relative to categorical variables. Since the
contrast, a bird seen with a partner for <2 rusty head patch of males was measured only
MANOVA
weeks (approximately the time required to lay in 1997 and 1998, an additional
a clutch; Gorsuch 1934, Johnsgard 1973), that model that included the rusty patch was run
next associated with another mate for <2 for males from these two years only. I also
weeks, was considered polygamous. We used Pearson correlations to clarify the rela-
scored polygamy only when we saw both tionship(s) between male ornaments and body
members ofthe original pair, as to exclude the size.
possibility of mate mortality. We also noted I determined the median hatching date an-
extrapair behavior for birds that partnered nually and estimated the mean size of broods
with a new mate for one observation period, that hatched relative to the median date. Mean
bmuattehabdy rtheetunrenxetd pteoritohde.irFionrailgliyn,alw,elornegcotredremd bsirzoeodofsiczheicwkass<ca1lcwuleaetkedoilnd,twaondwabyrso:odbrsoiozde
breeding and parenting behaviors, including pooled across all other chick ages. I used
single parents with broods and multiple par- brood size ofnewly hatched chicks to estimate
ents with communal broods. clutch size and hatching success, whereas data
— pooled across ages was used to estimate mean
andStasteixstriactailosaonfaltyhseisp.opuIlactailocnulaantneudalltyh.e Daigfe- brood size during chick rearing. I tested the
relationship between hatch date and brood
ferences in trait size between the sexes and
age classes (yearling, adult) were assessed size with 1-tailed /-tests, as late hatching
broods typically are smaller (Price et al.
wtriatith s2i-zteasileadcro/-stsestyse,arwshewreeraes adniaffleyrzeendceswitihn 1988). 1 also compared mean brood size of
ANOVA. Data were tested for normality prior adult and yearling pairs. Broods of banded
birds were not observed frequently enough to
to any parametric test. 1 used 2-tailed Fisher’s determine the total number of young fledged.
exact tests to assess patterns related to keel Therefore, seasonal changes in brood size (see
scores.
above) were essential to understanding wheth-
I calculated the median pairing date for the er early pairing birds also experienced greater
population each season. I tallied the number reproductive success. This approach was rea-
of adults and yearlings that paired before, ver- sonable only if the pairing date of a bird cor-
sus after, the median date and analyzed pat- related with the hatching date of its brood.
terns with Fisher’s exact tests. I used chi- That is, birds pairing before the median pair-
square goodness of fit tests to assess age-as- ing date also should hatch broods before the
sortative mating when both members ofa pair median date and vice versa. Significance of
MANOVA
were banded. determined any sig- the association between pairing and hatching
nificant differences in ornaments or body size was evaluated with a chi-square contingency
relative to pairing date, age class, or across table.
MANOVA
the three years studied. Separate I used binomial and chi-square tests to de-
analyses were conducted for males and fe- termine if winning a contest was related to a
males, and each bird was used only once in male’s age class or pairing status (paired ver-
each data set. Trait sizes were the dependent sus unpaired). I subtracted ornamental and
MANOVA
variables in each analysis, whereas body size traits of the losing male from the
pairing date, age and year were independent, winner to understand how traits related to
categorical variables. Pairing date compared winning. Winner-minus-loser (W — L) scores
traits of individuals that paired before versus were analyzed with Wilcoxon signed rank
Hagelin • MATING IN THE GAMBEL’S QUAIL 249
TABLE I. MANOVA and discriminant analysis of Gambel’s Quail breeding at Bosque del Apache, New
Mexico, indicated that the significant effect ofpairing date in males between 1996 and 1998 was best described
by differences in mass and not ornaments. The significant effects ofyear and age were based on differences in
ornaments (males) and body size (males and females).
Discriminant
Sex Effect F df p analysis^ F p
Males Pairing date 2.0 8, 141 0.0506 Mass 9.1 ().()()3()
Year 12.4 8, 142 0.0001 Tarsus, tail, dark patch, white patch 14.9 O.OOOl
Age 4.6 8, 141 0.0001 —Mass, wing, culmen, plume 1—2.7 0.0—001
Eemales Pairing date 1.4 6, 101 0.1996
Year 3.1 6, 102 0.0070 Tarsus, tail 9.0 ().00()2
Age 2.9 6, 101 0.0119 Mass, wing, tail 10.3 O.OOOl
^Variables includedmass (g); length (mm)oftarsus,culmen,tail,flattened wing,and flattenedheadplume;andarea(mm^)ofdarkandwhitebelly
patches.
tests. Each pair of males usWeci in signed rank RESULTS
tests was unique; therefore, - L scores rep- Pairing—We made 618 hours ofbehavioral
resented independent data points. When male .
observations over three seasons and recorded
pairs interacted more than once, the bird that
pairing status for 313 of 450 banded adults
dominated a simple majority of contests was
considered the winner, and only one W — L and yearlings. Median pairing dates occurred
on 14 April (1996), 11 April (1997) and 4
score was included in the data set. Ornaments
April (1998). Contrary to predictions of or-
might operate differently over the course of MANOVA
nament size, analysis identifiedthe
the breeding season. For example, early con-
tests that occur during pair formation may dif- mass of males as the only trait of either sex
that correlated significantly with pairing date
fer from late season contests that involve mate
or chick guarding. Since the pairing status of (Table 1). Regardless of year or age class,
bsmeiafgloneresed rwvaaenrsksutnseosttasftaoelnrwaWtyhse—mkLendodiwaantn,a speIatiscroicnongldluedccattteeed.d hi1ne8ga4v.di5aetrge,m(aafmlteeeras:np1ma7ia8r.se5sdgo,bfefaodru=eltst3.hb0ee4,fmoePrdei=amne0d.ip0aa0in3r:;-
Gsiigvneendtrhaenklatregsetsn,utmheberriskofomfaTlyepetraIitesrruosredwaisn y1e7a4r.l5ingg, m^8a9le=s3b.e8f9o,rPe m=ed0i.a0n0:02)1.85.M5AgN,OafVteAr:
high, which could have caused some variables also identified other traits of both sexes that
to be significant by chance (Rice 1989). varied significantly relative to year and age
Therefore, I analyzed the entire W — L data class (Table 1). These are considered sepa-
set simultaneously in a multiple logistic re- rately, below (see Differences between years,
gression with a stepwise selection procedure. age classes, and sexes). Interaction terms (e.g.,
The analysis identified traits that best de- year X pairing date) were not significant. Re-
scribed winning males. sults did not differ when data included rusty
Finally, I determined the number of birds head patch. Male mass exhibited a weak pos-
that were socially monogamous versus polyg- itive correlation only with dark patch area {P
amous each season. I assessed any significant = 0.025, P = 0.056, n = 162; all other or-
shifts away from social monogamy over the naments P < 0.0009, P ^ 0.70). Weak cor-
three year period with chi-square contingency relations between other body size traits and
< <
tables and two-tailed Fisher’s exact tests. Bi- ornaments also were positive (0.026
nomial tests analyzed patterns of monogamy: 0.063, 108 <« < 170, 0.001 < F < 0.095).
polygamy within each year, as compared to Adults of both sexes were more likely to
the 50:50 null expectation. I also used a con- pair before the median date (66% of 13~6
tingency table and Fisher’s exact test to de- adults versus 43% of 177 yearlings, x^i
termine whether adult or yearling pairs were 15.9, P < 0.0001). This pattern was more
more likely to act monogamously or polyga- marked in adult females (75% of 57 paired
mously. early) than adult males (60% of79, x^i = 3.7,
1
250 THE WILSON BULLETIN • Vol. 115, No. 3, September 2003
TABLE 2. Gambel’s Quail broods hatching before the median hatching date were larger than those hatched
later in the season at Bosque del Apache, New Mexico. Two ages categories are given: newly hatched chicks
(<1 week) andbroodsofall otherage classes (2-6 weeks). Significance isonetailed, asbrood size wasexpected
to decline for late hatching birds.
Meanbroodsize ± SE
Year Broodtype n Earlybroods Latebroods t p
1996 <1 week 28 11 ±0.6 6 ± 0.9 2.5 0.020
All ages 35 10 ± 0.9 5 ± 0.5 3.3 0.002
1997 <1 week 31 12 ± 0.9 8 ± 1.0 1.8 0.080
All ages 78 11 ±0.7 7 ± 0.6 2.8 0.007
1998 <1 week 1 12 ± 1.2 5 ± 0.9 3.2 0.010
All ages 29 9 ± 0.9 4 ± 0.4 3.2 0.003
P = 0.054). Yearling birds showed no such pairing before the median pairing date also
pattern; exactly 43% of females {n = 83) and produced offspring before the median hatch-
43% of males {n — 94) paired before the me- ing date and vice versa. The pattern did not
dian. Of 35 pairs in which both individuals differ significantly between the sexes (P >
were banded, 26 were the same age (18 adult 0.50). Median hatching dates fell between 6
pairs, 8 yearling pairs), while 9 were ofmixed June (1996) and 11 June (1998). Broods that
(X^i ~ 8.1, f* = 0.004). In only four cases, hatched before the median date were almost
pairs from a previous season mated again the twice as large as those hatched later (Table 2).
following year; these birds always paired be- Adults and yearlings produced similar-sized
fore the median pairing date. No birds paired broods (mean size pooled across all ages of
with the same partner during all three years. chicks: adults: n = 53, mean = 8.0 chicks ±
Mortality probably inhibited pair bonds >2 0.7 SE; yearling: n — 42, 8.0 chicks ± 0.8
years, as we noted it only three times. SE). We observed 15 single parents (nine
Since age was related to pairing and some male, six female) tending broods. In five in-
ornaments correlated with male size, I used stances (three female, two male), foster par-
ANCOVA
multivariate to determine if early enting apparently occurred, in which a single
pairing males had ornaments that were larger parent and brood associated with a new mate,
than expected for their body size. Foreach age and remained together during and following
class, I regressed male ornaments against two chick rearing. Broods also merged. In five cas-
predictor variables: (1) PRINl scores from a es, 3—5 parents tended excessively large num-
principle components analysis that described bers of chicks (mean brood size = 30 chicks
variation in male body size (mass, tarsus, ± 7 SE).
—
wing, tail), and (2) pairing date, which denot- Male contests. We observed 110 aggres-
ed whether a male had paired before or after sive encounters between different pairs of
the median pairing date. Body size (PRINl) banded males. Mated males won contests
conelated with significant differences in or- against unpaired opponents 29:4 (binomial
nament size, but pairing date did not (adults: test, P < 0.0001), and adults tended to dom-
PRINl: F433 - 5.08, P = ().0()3, pairing date: inate yearlings 36:22 (y^i = 3.6, P = 0.056).
P = 0.70; yearlings: PRINl: F = 3.45, P When both males were paired but differed in
- 0.013, pairing date: P = 0.304),., Therefore, age, either male was likely to win (18:17).
males pairing before versus after the median Winning males were larger than losers with
W
date did not differ significantly in ornament respect to mass (mean — L score: 4.7 g, w
size, once body size had been taken into ac- = 1210, /z = 103, P < 0.0001). No ornaments
count. — correlated with winning (P > 0.50). I obtained
Brood size and hatching. We recorded the a similar pattWern relative to the median pairing
pairing date and estimated the hatching date date (mean — L score pre-median: 3.8 g,
for 100 birds with broods. Pairing date cor- post-median: 5.1 g; 200 < w < 556, 0.008 <
related positively with hatching date in 83 cas- P < 0.020, 33 < /z < 70; all ornaments P >
(X“i “ 43.6, P < 0.0001). Namely, birds 0.30). Stepwise logistic analysis also selected
Hagelin • MATING IN THE GAMBEL’S QUAIL 251
mass as the primary variable that accounted criminant analysis indicated that the size of
for the differences between winners and losers ornamental patches of males varied signifi-
(—21og Lx^i estimate—: 12.8, P < 0.001). cantly among years, but not between age
Mating behavior. Of 164 birds followed classes (Tables 1, 3). Two body size traits of
during breeding, 56% were socially monoga- both sexes (tarsus, tail length) also varied
mous and 44% were polygamous. Social mo- among years (Table 1). Both decreased slight-
nogamy in the population declined from 83% ly over the study period (mean tarsus of both
in 1996 to 30% in 1998, while polygamy in- sexes, 1996: 32.1 mm, 1997: 31.2 mm, 1998:
cshriefatsseadw(axy\f=ro3m1.m5o,nPog<am0.y00o0c1c)u.rrSeidgniinfibcoatnht 3109.986:mm10;0.P22.4m48m=, 3189.917,:P99<.00.m0m00,1;1m99e8a:n9t7ai.l5,
sexes of yearling birds (males: Fisher’s exact mm; P2.444 ^ 5.0, P = 0.001).
P = 0.004; females: P = 0.001; Fig. lA, B) Discriminant analysis distinguished be-
and adult males (P == 0.001; Fig. 1C). Adult tween age classes of both sexes on the basis
females, however, were equally likely to ex- of body size (mass, wing, tail, culmen; Table
hibit social monogamy or polygamy each year 1). Adults were larger than yearling birds
(h(=iP5g1h=%6e;7r0%F.f;i9rs0eh;FqeiurFg’ei.sgn.cey1xICaDo)c)f,t.sPAcocdoiu=malplt0a.mrm0oea2dnl8;eosgtFoiaegxm.hyyielbaA(ri,ltmieendBag)n.sa 0(11.m77096e0..a387n;ggw,,mia^nys2g4es4,a,r=tlaiaind3lg,u.:l1ct8u,1lm7mPa2e.ln=4e::g01,.19089022.<;4=a/?dgu,<2l.ty92e7f4a,e9rm,lPailn2eg.=::3
Pairs of adults tended to be more monoga- < r < 3.4, 0.0008 < P < 0.007). Length of
omfou3s7;(F5i0sh%ero’fs2e6x)actthaPn p=ai0r.e0d62y)e.arTlihnegspo(l2y7g%- head plumes varied relative to age class in
males only (Table 3). Sui*prisingly, plumes of
amous behavior of adult females (Fig. ID)
adult males were slightly, but consistently,
tapcoetpnudPleadt=tioon0d.ic0f6of1em)rb.ifrnWoeemd ae(lFxlicgol.tuhd1eeArd—mC2e;3mbFbieisrrhdsesr’osffretoxhm-e as3d.hu1ol8rt,t:ePr40t=.h9a0n.m0y0me2,a)r.lyienagrlminagl:es42(.T1ablmem;3; ^2m46ea~n
analyses that appeared to act polygamously,
Males were somewhat larger than females,
but for which we could not exclude the pos-
sibAilitmyajoofrimtayteofmosrotcailailtlyy. monogamous birds rfmeeagmlaaerl:del:e1s11s77.3o6.f0magmg,e,L(4f7meem=aanl3e.m:0a,s1sP1,5=.m4a0l.me0:m0,31;78w.i8ngg=,,
(59%) exhibited at least one short (<2-week) 8.27, P < 0.0001; tarsus, male: 31.7 mm^4,48fe-
cexhtircakpsaiwritbhontdh.eiArllorriegitnuarlnemdattoe.anMdosttendpeoldyagnay- male: 30.8 mm, L48 = 7.46, P < 0.0001; tail,
I qmmuaoelunetsiabblliurepd-aslierfi(tn7gs8(.%B)LF)oreaxehnxidbaimftpeelmdea,lleoanybgealntldeoerwmd-,rpiagsihert-, f6m.ea8ml7ae,l:eP:10<111..002.m0m0mm0,1,;fLce4(um,lalm=ee:n,29.68m.9a1,le:Pmm1=,1.40^4.4m040m4,-;
i (YR), remained together between 4 April and plume: male: 41.7 mm, female: 31.3 mm, t^(,
13 May. By 31 May, BL associated with an- = 46.17, P < 0.0001). We palpated the keel
other female and remained with this mate of 163 quail. Ten of 18 adult females obtained
through the end of the breeding season. YR the highest score, indicating they had the
also paired with another male. Polygamous greatest amount of muscle and fat on their
birds typically switched mates one to two breasts, compared to only 26% of all other
times during a season (e.g., in May and again birds (Fisher’s exact P —= 0.014).
in June or July). Two adult females in 1996 Population changes. The 450 Gambel’s
BDA
abandoned their mates and 5- to 6-week-old Quail trapped at exhibited a male-biased
broods and re-paired with new partners. Three sex ratio, due to a scarcity of adult females
pairs that separated early in the season (May), each year and an over abundance of yearling
and had acquired new mates, reunited again males in 1998 (Table 4). The age distribution
later in July. Two reunited pairs were tending of the population also shifted. In 1996, year-
a brood. lings were approximately 20% more common
Differences between years, age classes, and than adults, but by 1998, they were 50% more
—
I sexes. BothMyAeaNrOaVndAage were significant ef- common (Table 4). The resulting 30% in-
fects in the model (Table 1). Dis- crease in the relative abundance of yearling
i
252 THE WILSON BULLETIN • Vol. 115, No. 3, September 2003
A Yearling males B Yearling females
0.010
Percent
1996 1997 1998 1996 1997
li cn /?= 12 A7 = 18 II c: II CNJ
c Adult males D Adult females
—
P = 0.0001
100 100 n
• n
90 ^ - 90
80
70 .
P=0.50 P^O.50 P"0.39
go
Percent
1996 1997 1998
1996 1997 1998
n = 13 n=22 r?=14 n = 7 a?=13 a?=12
Fig. 1. Significant shifts away from monogamy occurred across years for all groups of Gambel’s Quail at
Bosque del Apache, New Mexico, except adult females: (A) yearling males, Fisher’s exact P = 0.001; (B)
yearling females, P = 0.001; (C) adult males, P = 0.001; and (D) adult females, P = 0.90. Open bars are
monogamous individuals, solid bars are polygamous individuals. Patterns within each year were compared to
the null expectation of50:50 (monogamy polygamy); the cumulative binomial probability P{x ^ k) for each is
:
given above the bars.
Hagelin • MATING IN THE GAMBEL’S QUAIL 253
TABLE 3. Ornament sizes ofmale Gambel’s Quail at Bosque del Apache, New Mexico, varied by yearand
by age class between 1996 and 1998. Contrary to prediction, adult males did not have larger ornaments than
yearlings. Significant differences across years orbetween age classes are indicated in theTrait and Agecolumns,
respectively. Rusty head area was not measured in 1996.
1996 1997 1998
Trait Age (n = 66) (n = 82) (« = l(X))
Dark patch area (mm^)** Yearling 1407.4 1 11 1.0 1356.7
Adult 1431.2 1092.8 1636.8
White patch area (mm^)** Yearling 1990.7 1482.9 1738.3
Adult 19—77.9 1478.1 1881.9
Rusty head area (mm^)* First-year — 489.1 514.8
Adult 487.7 527.6
Plume (mm) Yearling** 42.5 43.0 41.5
Adult 41.5 41.0 39.5
*P= 0.05,**F = 0.001.
birds was significant {n = 450, x^2 ~ 8.3, P dominance hierarchy were successful in ob-
= 0.016). taining a mate (Mastrup 1987).
Unlike mass, male ornaments varied signif-
DISCUSSION icantly among years and between age classes
—
Ornamental versus body size traits. Field (Tables 1, 3). Flexibility ofornament size sug-
observations of Gambel’s Quail were consis- gests that ornaments potentially could serve as
tent with the results of captive experiments. indicator traits during some years. Yet, patch-
Instead of ornaments, male body size, specif- es were not predictably larger (or smaller) in
ically mass (Table 1), was the primary trait adults (Table 3), and all ornaments exhibited
that correlated with early pairing and winners rather weak correlations with mass and body
of male contests. In captive tests, the body size traits. Although early pairing males were
ANCOVA
size traits and behavior rates of males were heavier (Table 1), discriminant and
primarily related to female preferences andthe analyses indicated they did not exhibit signif-
winners ofmale contests, whereas natural var- icantly larger ornaments than other birds in
iation of single ornaments and experimental the population. Early pairing, however, result-
removal of multiple ornaments was not (Ha- ed in larger broods (Table 3), suggesting that
gelin 2001a, 2001b; Hagelin and Ligon 2001). heavier males experienced greater success, re-
Heavier (and older) males at our held site gardless ofornament size. Interaction terms of
also were the best competitors. Mass was the the MANOVA model (e.g., year X pairing
primary trait related to winning early season date) also were not significant, indicating the
contests and the only trait that varied signih- relationship between ornaments and pairing
cantly relative to pairing date (Table 1), indi- date did not change from year to year. The
cating that mass (or male size) may influence failure to detect any significant pattern of or-
mate acquisition. In the related California naments relative to pairing date (Table 1) did
Quail, only those birds thatbelonged to a male not result from a general lack of statistical
TABLE 4. Yearling and adult Gambel’s Quail trapped at Bosque del Apache, New Mexico, exhibited a
male-biased sex ratio (% males), a scarcity of adult females, and an increase in the relative abundance of
yearlings between 1996 and 1998.
Males(n) Females(n) %Males(%) Agedistribution(%)
Age 1996 1997 1998 1996 1997 1998 1996 1997 1998 1996 1997 1998
Yearling 35 53 75 35 52 63 50.0 50.4 54.3 59.8 70.0 75.4
Adult 31 29 25 16 16 20 66.0 64.4 55.6 40.2 30.0 24.6
Total 66 82 100 51 68 83 56.4 54.6 54.6 100.0 100.0 100.0
254 THE WILSON BULLETIN • Vol. 115, No. 3, September 2003
power. Samples were capable of detecting in- male quality, sexual selection may operate on
termediate and large effects of ornament size other, more reliable traits (Kodric-Brown and
relative to pairing date, while retaining 80% Brown 1984, Moller and Pomiankowski 1993,
statistical power (Cohen 1988). Understanding Hill 1994). Body size generally is thought of
smaller effects of ornaments will require larg- as an honest signal, because it is related to an
er samples or perhaps the use of composite underlying quality that cannot be faked (May-
indices to enhance the combined effect of nard Smith and Harper 1988, Guilford and
multiple ornaments (Calkins and Burley Dawkins 1995, Taylor et al. 2000). Body size
2003). of wild quail was clearly linked to age, size,
In previous captive manipulations (Hagelin early pairing in both sexes (see also Mastrup
2001a), only one ornate trait, the head plume 1987), and to male status. Given that Gam-
of Gambel’s males, affected the outcome of bel’s Quail do not usually live beyond their
aggressive contests. Hagelin (2001a) suggest- first breeding season (Sowls 1960), both size
ed that plume position (erect versus flattened) and age appear to honestly indicate an indi-
affected contests, rather than differences in vidual’s overall breeding quality. Experiments
plume length. Field observations confirmed using captive quail were consistent with these
lab findings. Natural variation in plume size results; females typically preferred larger
did not correlate with winning or early pair- males and largermales also won male contests
ing. Adult males, which were larger and usu- (Hagelin 2001a, Hagelin and Ligon 2001).
ally dominated yearling males, actually had High keel scores (greater fat and breast mus-
slightly (but significantly) shorter plumes (5% cle) were also associated with age and large
or 1-2 mm; Table 3). Any social benefit of size of wild adult females, which may reflect
slightly longer plumes appeared to be minimal a greater potential for reproduction (Reynolds
for yearlings. They were unlikely to win con- 1997).
tests or pair early, regardless of plume size. Rather than assessing each ornament sepa-
Several investigations of ornately-feathered rately, Calkins and Burley (2003) considered
species within the order Galliformes have re- “composite traits” by combining multiple or-
ported similar, and equally counter-intuitive naments of California Quail in analyses offe-
results to those of Gambel’s Quail. Specifi- male choice. The exact role, if any, of com-
cally, a strong relationship frequently is lack- posite traits in Gambel’s Quail is presently un-
ing between the ornate plumage of males and clear. Manipulations that removed both single
breeding behavior in species such as the Red and multiple ornaments from captive yearling
Junglefowl (Callus gallus\ Ligon et al. 1990, males did not alter mating preferences of fe-
Ligon and Zwartjes 1995), Wild Turkey (Me- males (Hagelin and Ligon 2001). Ligon and
leagris gallopavo; Buchholz 1995, 1997), and Zwartjes (1995) have reported similar results
Grey Partridge (Perdixperdix: Beani and Des- for Red Junglefowl.
—
si-Fulgheri 1995). Even the simultaneous re- Mating behavior. Gambel’s Quail exhib-
moval of multiple ornaments did not alter pat- ited a flexible breeding strategy which
terns of female preference (Ligon and Zwart- changed markedly from 83% monogamy in
jes 1995, Hagelin and Ligon 2001). Rather 1996 to 70% polygamy in 1998. The resulting
than a direct involvement ofornaments in sex- 53% increase in polygamy within the popu-
ual selection, many traits appear to have lost lation was due to two factors. First, the fre-
their key signaling function and. instead, may quency of polygamy increased in yearling
be maintained through alternative mecha- birds and adult males between 1996 and 1998
nisms, such as weak selection (Moller and (Fig. lA-C). Second, the relative abundance
Pomiankowski 1993, Hagelin and Ligon of yearlings increased by 30% over three
2001). It is unclear, however, how weak sex- years (Table 4), and yearling pairs were more
ual selection must be in order to maintain or- likely to exhibit polygamy than adults.
naments. Our data revealed a poor correlation Although the proximate cues underlying in-
between body size and ornate traits, which dividual breeding behavior are unknown, the
may be a sufficient mechanism for ornament environmental conditions that influence the ir-
maintenance. ruptive nature of semi-arid quail populations
If certain plumage ornaments do not reflect (Forrester et al. 1998, Heffelfinger et al. 1999,
Hagelin • MATING IN THE GAMBEL’S QUAIL 255
Giithery et al. 2001) appear to provide a pairing date than any other birds in the pop-
mechanism. Extreme heat or lack of precipi- ulation, indicating they had a high potential
tation, for example, can influence brood fail- for reproduction (Table 2). A male-biased sex
ure (Heffelfinger et al. 1999, Guthery et al. ratio (Table 4) and high keel scores may have
2001), thereby increasing the frequency ofre- provided adult females with the opportunity
pairing. High levels of polygamy at our study to behave polygamously and obtain additional
site (1998; Fig. lA-C) coincided with a ten- matings. Accordingly, adult females were the
year low in rainfall at the time of hatching only birds that attempted to double brood.
(June), whereas high levels of monogamy Though male incubation in the Northern Bob-
(1996; Fig. lA-C), corresponded to a 25-year white {Colinus virginianus) contributes to fe-
high. male abandonment and double brooding (Cur-
Mean mass of both sexes at Bosque del tis et al. 1993, Suchy and Munkel 1993, Bur-
Apache was within the range reported for oth- ger et al. 1995), it is unknown in Gambel’s
er wild populations (160-200 g; Brown et al. Quail. In the California Quail, a related spe-
1998), suggesting that birds were not abnor- cies that occasionally double broods (Francis
mally fattened by supplemental winter feed. 1965), male brood patches are rare and may
However, feed could have created a high local develop only after the death of a mate (Cal-
nesting density, thereby enhancing re-pairing kins et al. 1999).
opportunities relative to other breeding sites. Gambel’s Quail at Bosque del Apache ex-
Though absolute measures may differ, the rel- hibited both biparental and uniparental care.
ative changes I observed in brood size (Table Birds also showed communal brooding and
2) and in the breeding behavior ofindividuals the potential for foster care. Communal
(Fig. lA-C) occurred independently ofwinter broods in California Quail can provide fitness
feed, which remained constant throughout the benefits (Lott and Mastrup 1999). However,
study. both communal brooding and foster care are
Adult males, the largest, most dominant unstudied in Gambel’s Quail.
birds in the population, exhibited the highest In conclusion, the traits related to success-
frequency of social monogamy (Fig. lA-D), ful breeding in a wild population of Gambel’s
indicating monogamous behavior was a pre- Quail were consistent with captive studies that
ferred breeding strategy. Male California assessed the role of both single and multiple
Quail also exhibited an unwillingness to be- plumage ornaments. Early pairing in males
have polygamously, even during manipula- was related to mass and age rather than to
tions of sex ratio (see Lott and Mastrup un- ornament size. Male mass was also the pri-
publ. data in Calkins et al. 1999). Our data mary trait associated with winners of male
provide several insights into why social mo- contests. The mating behavior of Gambel’s
nogamy may be favorable. First, pairing date Quail was flexible. Most birds exhibited a
and age-assortative mating indicated that adult shift in breeding strategy from monogamy to
males had a good chance of obtaining a qual- polygamy in three years. Adult females, how-
ity mate (a large adult female) early on. ever, exhibited both strategies with equal fre-
Heavier adults that paired early had a high quency. Future investigations should aim to
likelihood of successful breeding (Table 2), understand the mechanisms that cause shifts
and some adult females attempted second in breeding behavior and the costs and bene-
broods. Were a male to abandon its first mate fits of different mating strategies. Studies will
and re-pair, it would be difficult to obtain an- require detailed data from radio-tagged breed-
other mate, particularly an adult female, given ers under known environmental conditions
their scarcity in the population (Table 4). and estimates of brood paternity, in order to
Adult females consistently maintained a compare the annual and lifetime reproductive
mean frequency of 46% polygamy, regardless success of different sex and age classes.
of year (Fig. ID). This pattern was unlike the
ACKNOWLEDGMENTS
more variable breeding behavior of other
members ofthe population (Fig. 1A-C). Adult This study was made possible by the careful, accu-
females also had the highest keel scores and rate, and tireless observations of A. MacDonald, and
were more likely to pair before the median the support of Bosque del Apache National Wildlife