IN THE MAGPIE GOOSE ANSERANAS SEMIPALMATAespace.cdu.edu.au/eserv/cdu:260/Whitehead_260.pdf · IN...

BOOFHEADS WITH DEEPVOICES: SEXUAL DIMORPHISMIN THE MAGPIE GOOSE ANSERANAS SEMIPALMATA

PETER J WHITEHEAD

Parks and Wildlife Commission of the Northern Territory and School of Biological and Environmental Sciences,Northern Territory University, Darwin NT 0909,Australia.

Magpie Geese use a unique polygynous mating system, involving apparently stable triosof one male and two females. Plumage shows little sexual differentiation, but there isconsiderable dimorphism in body size, with males being about 30% heavier. The malesalso develop an extraordinarily elongated and elaborately folded trachea early in life,whereas less than 8.5% of females show even minor tracheal elaboration. Head height, ameasure of the size of a cranial bump of spongy bone increases with age in both sexes,but most markedly in males. Males found in association with nests have larger bumpsand highly elaborated tracheal morphology. The deeper and louder calls associated withgross tracheal elongation, which probably compromises respiratory exchange, may influencefemale choice of mates by providing a reliable signal of male viability. Despite significantoverlap in individual dimensions, especially among younger birds, more than 92% can beaccurately sexed using a discriminant function based on three simple measures (head-billlength. head height, and tarsus length). Discrimination can be improved by checking birdsassigned as females against an index of tracheal morphology. Simulations indicate thatbias in estimates of sex ratios arising from application of the discriminant function, whencombined with tracheal examination, is likely to be less than 2%.

Keywords: Magpie Goose, Sexual Dimorphism, Sexual selection, Morphometries, Growth patterns,Tracheal elaboration, Cranial bump.

Magpie Geese Anseranas semipalmata display amating system that is unique among waterfowl.Most reproductive groups comprise trios ofone male and two females, both of which layapproximately synchronously in a shared nest(Frith & Davies 1961). Equally unusuallyUohnsgard 1978), while there is no obviousdifference between the sexes in plumage, thespecies shows considerable sexual dimorphismin body size, with weights of adult malesaveraging about 30% more than females. Inaddition to greater general bulk, males alsoshow much greater development of adistinctive, skin-covered cranial bump of spongybone and an extraordinarily elongated trachea,lying in folds beneath the pectoral skin (Frith &Davies 1961).

© The Wildfowl & Wetlands Trust

Despite such an array of potentiallyquantifiable dimorphic characters, Frith andDavies (1961) concluded, from examination ofa large sample (761 birds), that there wassufficient overlap in dimensions withinpopulations (including birds of all post-fledgingage classes) to prevent use of any singlemeasurement to reliably determine sex.Perhaps as a result, there has been nosystematic attempt to relate the species'unusually pronounced sexual dimorphism to itsequally enigmatic mating system. This paperseeks to fill part of that gap.

Movever, the species is exploited forsubsistence and recreation (eg Dexter 1988,Whitehead et al. 1988), and faces changes inhabitat quality over parts of its already

WILDFOWL (1998) 49: 72-91

SEXUAL DIMORPHISM IN MAGPIE GEESE 73

Table I. Summary of categories of tracheal loop length and configuration to whichobservations were assigned for analysis. Positions and configurations of loops weredetermined by palapating the skin at the base of the neck and overlying the breast.

<5 left side

<10 left side

<15 left side

<20 left side

>20 left side

>15 2 long loops on left, Ishorter loop on right

>20 2 long loops on left, 2shorter loops on right

Class Number of loops

0 0

2 or 2

3 2

4 2

5 2

6 3

7 4

Length (cm) oflongest loop

Location(relative to sternal crest)

drastically contracted contemporary range(Whitehead et al. 1992). Harvests and habitatdeterioration will affect different elements ofthe life cycle in many interacting ways, and thedynamics of populations are too poorlyunderstood (Frith & Davies 1961) for thecumulative effects of these changes to bepredicted. Robust methods of sexing andassigning birds to age or other classes based onreproductive status will therefore be importantfor developing models of population dynamicsas tools for management.

Accordingly, I also extend this re-examination of morphometric variation toprovide:

(I) a method for sexing birds from a smallnumber of simply- and non-invasivelymeasured dimensions;(2) measures of potential error in estimatingsex ratios with this method; and,(3) evidence of age-related variation in somedimensions that, with further study, might beemployed to assign broad age classes or lifecycle stages.

Methods

Study sites are located in the seasonal tropicsof the Northern Territory, Australia. Theregion's climate is characterised by heavyrainfalls during a hot summer monsoon(November to April) when most of the annualrainfall of more than 1,600 mm occurs. For theremainder of the year (the cooler dry season)there is little or no rain. During the late dryseason (early September to early December)Magpie Geese were captured in baited walk-infunnel traps or by cannon nets at Berrimah onthe outskirts of Darwin, and at two sites on thewestern margin of the Mary River floodplain,about 60 km east of Darwin. During thebreeding seasons (March/April) birds were alsocaught on nests by hand or using remotely-triggered clam traps in or near a breedingcolony at Opium Creek, also on the Mary Riverfloodplain. The Mary River sites are describedin detail inWhitehead et al. I990a.

All birds caught were released fitted with astainless steel leg band and an aluminium(Maclnnes et al. 1969) or laminated plastic neck

74 SEXUAL DIMORPHISM IN MAGPIE GEESE

collar similar to that described by Johnson andSibly (1989). As well, 26 birds of known agewere hatched from artificially incubated eggsand raised in captivity for various periods. Datafrom live birds were supplemented by birds (i)killed by recreational hunters and (ii) taken fora genetic study while they were escortinggoslings (Horn et al. 1996).

Head-bill length, head height and legdimensions (tarsus) were measured withvernier calipers (±O.I mm) as described byWhitehead et al. (1990b), and many birds werealso weighed with a spring balance (±50g). Themeasurement of head height used here, fromthe top of the cranial bump to the bottom ofthe lower mandible, varies from that used byFrith & Davies (1961) - the distal corner of theeye to the top of the skull - to avoid the needto position calipers near the eye of live andstruggling birds.

Live birds were sexed at the time of captureby cloacal examination (n=349) or subjectivelyby an experienced observer (n=428). Sex ofdead birds was determined by dissection andgonadal examination. Only dissected birds orcaptive birds sexed by repeated cloacalexamination were used in quantitative analysesused to discriminate sexes (below).

Many Magpie Geese show tracheal 'loops' ofvarying length and configuration originating atthe base of the neck and overlying the pectoralmuscle and wall of the abdominal cavity. Theloops can be easily detected and their locationdetermined by palpating the skin over thebreast and abdomen (Frith & Davies 1961).Because of the time required to measure lengthof subcutaneous structures precisely, trachealdimensions were estimated relative toconspicuous anatomical features such as thebase of the sternum, and variation summarisedby assigning observations to one of eight broadcategories (Table I).

Birds caught on nests were regarded asreproductively active, although it was notpossible to demonstrate that captured birdshad contributed genetically to the clutch. Forall numerical analyses, only birds known to beless than one year old (ie captive bred) wereassigned to the juvenile category.

Discriminant analysis (SAS procedure

DISCRIM) was used to derive functions toassign individual observations to sex classesfrom a range of linear dimensions, using a groupof 64 birds for which sex was unambiguouslyknown (from dissection for wild birds (n=38)or repeated cloacal examination for birds ofknown age raised in captivity (n=26)). Thesample of 26 captive birds was of known age(75 days) and was included to ensure that thediscriminant function incorporatedobservations from the youngest birds likely tobe encountered in free-flying populations.

A number of wild birds were recaptured andremeasured at intervals ranging from a few daysto more than three years. Post-fledgingvariation in head and leg dimensions wasmodelled from re-measurements at intervalsexceeding 30 d using the von Bertalannfygrowth equation (which takes the form y=a( 1-be"'), where y is the dimension of interest attime t, a is the asymptotic size, band k areconstants. As ages of birds were mostlyunknown, methods summarised by Fabens(1965) were used. Models were derivediteratively using the SAS NUN procedure (SASInstitute Ine. 1989), to fit values for the growthconstant, k, and asymptote, a.

Analysis of covariance employed SASprocedure GLM. Frequency distributions werecompared with the Kolmogorov-Smirnov test(SAS NPAR IWAY procedure). Means are givenwith the standard deviation unless otherwiseindicated.

Results

Sex Differences

Head and Leg Dimensions and Body Weight

Mean dimensions for fledged wild MagpieGeese caught live or measured post-mortemare summarised in Table 2, and some patternsof association among these measures areillustrated in Figures Ia-I c. Males weretypically larger than females in all dimensions.

Body weight was strongly correlated(P<O.OOOI) with all three linear dimensions inboth sexes. The rate of increase in body weightwith head-bill length (males, r=0.43; females,


Table 2. Mean (+SD) and range of measured dimensions of wild Magpie Geese (422males and 420 females). Inter-sex variation was highly significant for all dimensions (P<O.OOO I).All birds were flying but varied widely in age.

Dimension

Mean±SD

Male

Range

Female

Mean±SD Range

Weight (g) 2432±316 1480 to 3200 1858±232 I 150 to 2420

Head-bill length (mm) 130.3±4.0 120.0 to 141.6 I 17.8±3.3 109.2 to 126.0

Head height (mm) 62.2±8.6 44.0 to 85.0 52.1 ±5.2 40.7 to 65.8

Tarsus length (mm) 103.8±4.9 87.7 to 117.7 92.3±4.2 80.6 to 104.3

r=0.43) was consistent between sexes (FigureIa:ANCOYA, sex by head-bill interaction term,F,7IS=O.7,P=0.40) but males were significantlyheavier than females with the same head-billlength (factor sex, F,,7IS=21.3,P<O.OOO I). Fittinga power relationship between head-bill lengthand weight did not produce a significantlybetter fit. The relationship between weight andtarsus length (males, r=0.40; females, r=0.36)was similar to the head-bill/weight relationship.

Body weight was most strongly positivelycorrelated with head height (males, r=0.69;females, r=0.65), a dimension that appears tovary with age in both sexes (Frith & Davies1961; and below). Entry of head height intoregressions already containing head-bill ortarsus length significantly increased explainedvariance in body weight in both sexes(P<O.OOO I).

Tarsus length was correlated with head-billlength in both sexes (Figure Ib). Slopes ofsimple regression lines were similar for eachsex (ANCOYA: F"m=O.4I, P=0.52), butintercepts varied significantly (FI,m=29.6,P<O.OOO I). Males tend to have longer tarsi thanfemales of the same head-bill length.

The relationship between head-bill lengthand head height was relatively weaker, probablydue to post-fledging growth of the cranial bumpin both sexes (Figure Ic and below). Thegreat post-fledging increase in head height isillustrated in Figure I c by inclusion of a

regression line showing the relationship ofhead-bill length to head height in juvenile birds.There was no between-sex variation in thisrelationship for young birds (FI.5I<0.24,P>0.62).In contrast, the association between these headdimensions in the sample of wild birds of allages varies significantly between sexes(FI,774=7.2,P=0.008), with males developingrelatively larger cranial bumps than females atequivalent head-bill lengths. If it is assumed thatthe age structure of the male and femalesamples are broadly similar, this variation inrelationships implies more rapid growth in headheight among males than in females.

Tracheal morphology

There was also marked between-sex variationin the length and configuration of the trachea.Most females (91.5%) lacked tracheal loops(Figure 2), or had small ones extending nomore than a few centimetres from the left sideof the neck.

Among birds with greater development ofthe trachea, there were invariably two loops onthe left side of the sternum (classes 2 to 5 inFigure 2), one lying inside the other so thatfour closely-associated tracheal diameterscould be felt beneath the skin. Many males alsoshow a third, smaller loop on the right side ofthe neck (class 6), and one bird had a fourthright-side loop (class 7). The left-side loops are


3500

3000

2500

2000

1500

(a)

1000

o

o

o

105.0 115.0 125.0 135.0 145.0

Head-bill length (mm)

Figure I a. Relationships between head-bill length and body weight for Magpie Geese.Males are shown with solid symbols and females with open symbols. Separateregression lines are shown for males (solid line) and females (dashed line). Much of thewithin- and between sex variation in body weight can be attributed to differences in general bodysize as reflected in head and leg measurements, although females are significantly lighter than malesin the small region of Dverlap in head-bill length.

120.0 (b)


.~----.--------

•·88

,.-.....E 105.0E •

'--"..r: 0

-+- 0

rncQ) •Cl):JCl) 8L0 90.0I- 8

'b00

0

75.0

105.0 115.0 125.0 135.0 145.0


Figure Ib. Relationship between tarsus length and head-bill length in Magpie Geese.Symbols and lines are as in Figure Ia. The slope of the regession lines is similar for males andfemales, but there is a significant difference in intercepts. Males have longer tarsi at the same head-bill length.


40.0

105.0 115.0 125.0 135.0 145.0


Figure I c. Variation in head height of Magpie Geese with head-bill length. Symbols andlines are similar to Figure I a, with the exception that the relationship is also shownseparately for for recently fledged Magpie Geese raised in captivity, measured beforesubstantial development of the cranial bump (lower dashed line). Much of the variation in headheight is related to growth in the cranial bump, which appears to continue for many years (see text).

Loop configuration class

invariably the longest, often extending from thebase of the neck to the bottom of the abdominalcavity near the cloaca.

Associations of tracheal morphology with body size

Within sexes, loop dimensions andconfigurations were strongly related to generalvariation in body size as indicated by measuresof head and leg length (Figure 2). For example,in females neither head-bill length nor tarsuslength varied significantly among loopconfiguration classes (head-bill; F2.266=1.7, P=0.19:tarsus; F2266=2.0, P=0.14). Tarsus length did notvary significantly with loop class in males(F7301=2.0,P=0.055).

Variation in head-bill length with loop classwas significant among males (F7301=2.5, P=O.O 18),but the proportion of variance explained wassmall (r'=0.054). The only pairwise comparisonsthat showed significant variation were betweenclass 0 and class 6 and class 0 and class 4 (TukeysHSD, P<0.05). The class 0 sample was comprisedentirely of birds regarded as less than one yearold, as determined from similarities of plumageand soft tissue colouration to birds of knownage reared in captivity.

In contrast, variation in head height amongloop classes was considerable (Figure 2) andsignificant for both sexes (females; F2,266=5.3,P=0.006: males (F73ol)=95.1,P<O.OOO I). In bothsexes, birds with loops have larger cranial bumps,and longer or more elaborate looping isassociated with larger bumps. In males, 62.6% ofvariance in head height could be associated withtracheal loop class. Among males, head height inall individual loop classes up to and includingthree differed significantly from all individualclasses at or above four. Within each of thesegroups (0 to 3 and 4 to 7) the only significantpairwise differences were between classes 0 and3 and classes 4 and 6 (Figure 2). In pairwisecomparisons for females, head heights variedbetween loop classes 0 and I (Tukey's HSD,P<0.05), but other contrasts were not significant.

Discriminant Analysis

All of the variables included in the discriminantfunction shown in Table 3 varied significantly


150

E(75)

140(31)

-S.<::0, 130c~:;; 120

I

"0 110" (245) (22)I

100

120

E 110

~

-S~ 1000>C~

90•.E ~0 80I-

70

80

E r3-S 70

:E0> 60';;.<::

"0 50

"I40

Figure 2. Variation in head-bill and tarsuslengths and head heights among trachealloop classes. Dots are means, lines andbars delimit 2 standard deviations.Figures in parentheses are sample sizes. Plotsare offset slightly on the horizontal axis forclarity. There are marked differences betweensexes in the range of tracheal morphologiesand their relative frequency. There is strongassociation between other measures of bodysize and tracheal morphology only in respect ofhead height. Birds of both sexes with longerand more complex tracheal configurations tendalso to have larger cranial bumps.


Ta.ble 3. Details of a discriminant function derived from a sample of unambiguouslysexed birds (n=64). The numbers are coefficients used to calculate a score from the sum ofproducts of the relevant coefficient and measure of individual dimensions. An individual is assignedto the sex for which the discriminant function returns the highest value.

Sex

Variables Female Male

Constant -757.2 -908.4

Head-bill length (mm) 12.573 13.680

Head height (mm) 0.601 0.674

Tarsus (mm) -0.0356 0.0645

between sexes. Head-bill length showed thegreatest between-sex differentiation, and headheight the least. The function correctlyclassified 63 of 64 birds of unambiguouslydetermined sex (Table 4).

The function also produced assignments ofsex in a larger sample of live birds that werereasonably congruent with sex determinedfrom cloacal examination, and a still largercombined sample that included the cloacally-sexed birds plus a number of adult birds sexedsubjectively by an experienced observer. Birdssexed as females in the field were infrequentlymisclassified by the discriminant function.However, the trend for a substantial number of(smaller) males to be apparently misclassified asfemale was maintained in the larger samples(Table 4).

Some apparent misclassifications in the largersamples may result from actual errors of sexingor inaccurate data transcription. However, thisis not a sufficient explanation for highermisclassification rates among males. Fourteen(61%) of 23 putative males classed as female bythe discriminant function showed otherdistinctly male characteristics; in obviouslyolder birds, tracheal lengths never recorded infemales (Frith & Davies 1961), or in very young(first-year) birds with little development of thecranial bump, there was minor but distinct(class I or 2) elaboration of the trachea.

To explore potential bias in field studies,estimates of sex ratios were derived from 100samples of 100 randomly-selected birds sexed

by the discriminant function, and the ratio soderived compared with estimates based onsexes assigned to the same birds by othercriteria. Calculated ratios were re-examinedfollowing application of the additional trachealcriterion. Any bird classed as female by thediscriminant function but possessing trachealelaborations greater than ever measured infemales (class 3 or above: Frith & Davies 1961)was reassigned to the male group. Similarly,birds with small head heights «52 mm)classified as female showing significant trachealdevelopment were reassigned as male. Resultsof applying the tracheal criteria are summarisedin Table 5.

Growth rates - recaptured birds

Head-bill and tarsus lengths

There was no evidence of growth in head-billlength or tarsus length in the small sample ofbirds recaptured and remeasured at nearannual or longer intervals (350 to 1,079 days).Fitting the von Bertalannfy growth model topaired measurements (Fabens 1965) generatedestimates of growth constants k that did notdiffer significantly from zero (95% asymptoticconfidence intervals straddled zero).

Head Height

Similarly, it was not possible to derive a modelfor growth of head heights of females. Given


Table 4. Performance of discriminant function for sexing Magpie Geese. Figures inparentheses are the percentage of the sample correctly classified. The validation sample is based onbirds sexed by cloacal examination in the field.

Sample Sexed by dissection Validation sample(cloacal sexing)

Classified from: Classified to:

Female Male Female Male

Female 32 0 197 4

(100.0) (0.0) (98.0) (2.0)

Male 31 II 137

(3.1 ) (96.9) (7.4) (92.6)

Whole sample(cloacal & subjective)

Female Male

385 6

(98.5) ( 1.5)

23 373

(5.8) (94.2)

the small sample of remeasured birds (n=9),growth increments (the longest interval for afemale was one year) may have been obscuredby the error inherent in measurement on livebirds under field conditions. In the slightlylarger male sample (n= I I) that also covered alonger period (up to 1,079 d), head height didincrease significantly (k=O.I96 ± 0.042(SE); a=73.7 ± 2.3: Figure 3). The data are too few,particularly at larger head heights, and theasymptotic standard errors of parameterestimates too wide for interpretation as acomprehensive summary of age-related changein head height. The model's limitations areillustrated by estimation of an asymptote lowerthan the maximum head height observed in thestudy (Figure 3). An alternative model basedon a fixed asymptote (85 mm) similar to thelargest head height measured is shown forcomparison. These alternative models divergelittle within the range for which repeatmeasures are available (:::75.3 mm), and suggestthat males may achieve head heights above theminimum observed among reproductivelyactive birds about 4-5 years after fledging. Adatum from one male reared in captivity fromhatching for nearly three years is alsoconsistent with this interpretation (Figure 3).

Tracheal Morphology

The manner in which tracheal loops weredescribed inhibits detailed quantitativetreatment of patterns of growth in tracheallength. However, some inferences can bedrawn from patterns of within-sample variationin male tracheal morphology. In particular, thelow relative frequencies (Figure 4d) of classeso to 3 (loops less than 15 cm in length), suggestthat male tracheal elaboration begins early inlife and tracheae rapidly attain substantialdimensions. Males caught during the periodSeptember to December and classedsubjectively as first year birds - on billcolouration (large black areas on the bill andespecially the nail), dull leg colouration (adultsare bright yellow/orange), and generally darkerplumage (Marchant & Higgins 1990)sometimes showed significant trachealelaboration (20 of 52). Exact ages of thesebirds are unknown but, given the typical time ofnesting around the capture sites, is likely torange from 5 to 10 months (Whitehead 1998).Tracheal development ranged from a small'kink' at the base of the neck, to doubled loopsup to 7 cm long. One male caught initially whenless than six months old with no loopsapparent, had class two loops (doubled loops


- ------

90Morphology of Birds of Known Status

extending between 5 and 10 cm from the baseof the neck) when re-caught a year later. Acaptive-reared male accidentally killed at a localwildlife park when aged 22 months (and hencewell into its second post-fledging wet season,the time of breeding) had doubled loopsreaching close to the base of the keel (class 4).

In contrast, patterns of variation in femaletracheal morphology imply a much slowerdevelopment, and the low frequency (8.5%) ofsubstantially developed loops (classes I or 2) inbirds with larger head heights implies thatthey develop predominantly in older birds(below). No evidence of tracheal elaborationwas ever recorded in a first year female.

Figure 3. von Bertalannfy model ofgrowth in male head heights from themean at fledging (46.5 mm) estimatedfrom non-linear regression on re-measurements of II wild birds. Onemodel (solid line) fitted both asymptote andgrowth constant, and the second (dashed line)took the largest observed head height asasymptote. The horizontal line shows theminimum head height observed in a sample ofbirds captured on nests or escorting broods.The dot shows head height of a captive rearedmale of known age (2.8 years). Minimum headheights observed in breeding males appearlikely to be reached in 4 or 5 years, althoughmost of the breeding population appears to besubstantially older (Figure 4).

The total sample includes individuals of bothsexes of known reproductive status. The firstgroup comprises pre-reproductive birdshatched and raised in captivity for periods of upto seven months. The other group of knownstatus comprises birds considered reproductiveactive because they were caught live on nests ortaken while escorting broods. The position ofthese groups within the whole-populationdistribution of head and tracheal dimensions isshown in Figures 4a-d.

There was considerable overlap in head-billand tarsus lengths between juveniles andreproductively active birds of both groups.Indeed the distribution of head-bill lengths ofjuveniles suggests that growth in this dimensionis minor after the first year (Figure 4a). Thesituation is similar in regard to tarsus length(Figure 4b). In neither sex did frequencydistributions of these dimensions varysignificantly between juvenile and nesting birds,or nesting birds and the sample as a whole(Kolmogorov-Smirnov tests, P>0.05)

Head height showed greater divergencebetween juveniles and reproductive adults,especially among males (Figure 4c). There wasno overlap in male head height between knownjuveniles (mean 47.8 ± 2.9, range 44.1 to 52.5,n== 14) and breeding birds (mean 69.6 ± 3.7;range 61.9 to 76.6, n==42). The distribution ofhead heights among breeding males differedsignificantly from the sample population as awhole (P<O.OOO I). Even when comparisons areconfined to that portion of the sample withhead heights exceeding the minimum (61.9 mm)found in birds at nests, the distribution of headheights in the known reproductives wassignificantly skewed (P==0.0007) towards largerhead heights (Figure 4c).

Among females, overlap was minor (juveniles;mean 45.0 ± 2.7, range 40.9 to 49.9, n== 15:adults; mean 56.3 ± 3.3, range 49.9 to 62.7,n==24) and the distibution of head heightsdiffered significantly between these classes.There was some suggestion of a skew towardslarger bumps in the sample of knownreproductives (Figure 4c) but whencomparisons are confined to the birds with head

2016

symptote fitted

128

;.-..-..-..- .

Minimum head height for nesting male

Time elapsed (years)

40o

E.s1:: 70.~Q).c"tJ 60(tIQ)

:I:

--- ---

80


"IQ)

25, 0'01 200-t:Q) 150•....Q)

Cl.. 10

5

0

4035

Q) 30010 25-t:Q) 200•....Q) 15

Cl..

1050

110 120 130 140 150


Figure 4a. Variation in the relative distribution of head-bill lengths in differentcomponents of the sample population of fledged birds. The wild-caught sample population ofunknown reproductive status is shown by open bars, the juvenile sample of captive reared fledgedbirds by hatched bars, and the sample of birds taken in association with nests or broods by filled(black) bars. In both sexes there is considerable overlap in head-bill lengths between the sub-samplesof juvenile (first-year) birds (hatched bars) and birds caught at nests or escorting broods (solid bars).

heights above the minimum found in a birdassociated with a nest (49.9 mm), the variation indistribution was not significant (P=0.16).

Discussion

Sexual dimorphism and sexual selection

From the time of hatching, male Magpie Geesegrow faster to greater asymptotic dimensionsthan females, so that at fledging there is asubstantial difference in body size (Whitehead

et al. 1990b). This study shows thatmorphological divergence between the sexescontinues to widen with post-fledging age inrespect of two conspicuous characters: thelength and configuration of the trachea and thesize of a distinctive cranial 'bump'.

Among females, maximum trachealdevelopment is confined to a single loop of afew centimeters near the base of the neck andis found in a small proportion of the population.In contrast, in all males the trachea growsrapidly to form an elaborately looped structure

80


- -.-------------

.30

25Q)

0> 20ci:Q) 150•...Q)a.. 10

5

090 110

Tarsus length (mm)

---------.--

120

Figure 4b. Distribution oftarsus lengths in the sample population. Different bars relateto sample status in the same way as in Figure 4a. There is little evidence of age-relatedvariation, with the samples of juvenile (first-year: hatched bars) birds and adults (nesting attendingbroods: solid bars) showing a similar distribution.

ultimately exceeding 125 cm in total length. Ofthe dimensions measured, trachealdevelopment was the morphological featuremost closely associated with male reproductiveactivity (Figure 4). All males on nests or withbroods had greatly elongated paired trachealloops, and most (84.2%) possessed a third loop.

Gross elongation of the trachea is likely tocarry some physiological penalty, including thecosts of growth and maintenance of additionaltissue. More importantly, increased 'dead space'in an additional 80 cm or more of tracheallength is likely to compromise respiratory gasexchange (Hinds & Calder 1971). The length ofthe adult male trachea exceeds five times the

length (24.8 cm) calculated from allometricrelationships between tracheal dimensions andbody weight in birds, a ratio much greater thanfor the most elongated trachea described (3.3for the Trumpeter Swan Olar buccinator) in thesummary provided by Hinds & Calder (1971).

Such a feature, in a bird that frequently flieslong distances (Whitehead 1998) and can bepresumed to benefit from efficient respiratoryfunction, implies important non-respiratoryroles for tracheal structure in determining thefitness of male Magpie Geese. Males producelouder, lower pitched calls than females(Marchant & Higgins 1990). The functionalsignificance of pitch and volume of calls has not


300'25

Q)01 200-CQ) 15uL..Q)

CL 10

5

0

30

25Q)01 200-cQ) 15uL..Q)CL 10

5

040 50 60 70 80

Head height (mm)

Figure 4c. Distribution of head height (and hence cranial bump) dimensions in thesample population. In contrast to other dimensions the juvenile (first year) age class is distinct(hatched bars) and contributes to a bimodal distribution in the population as a whole (open bars).Among males there is a large gap between juvenile (hatched bars) and reproductively active birds(solid bars). This trend is less clear in females, in which there is minor overlap.

been directly studied, but may involveenhanced contact among group members.However, Magpie Geese inhabit open swampsand mostly move in flocks. The lower rate ofattenuation of lower-pitched calls withdistance or in structurally complexenvironments often invoked as an importantadvantage of such calls (eg Endler 1992),appears an inadequate compensation for thelikely physiological penalty. Rather, aconjunction of such extraordinaryexaggeration of the trait in males and theequally unusual polygynous mating system

suggest an important role for sexual selectionin its maintenance.

Exaggerated traits maintained by sexualselection are common in birds, includingplumage dimorphism in some wildfowl, despiteadverse impacts on male survival (Promislow etal. 1994, Andersson & Iwasa 1996). Sexualselection has previously been invoked toexplain sex differences in some ecologicallysignificant attributes of Magpie Goose biology.Whitehead et al. (1990b) argued that extendedjuvenile growth periods and consequent largebody size in male Magpie Geese are maintained


100

80, 0-

Cl) IC'Ic

:: I+-r::Cl)0•...Cl)

a...

20

0 -

100

80 I

Cl) IC'Ic 60+-r::Q)0•... 40Q)

a...

20

0

0 2 3 4 5 6 7

Loop configuration class

Figure 4d. Distribution of loop configuration classes among the sample population.Different bars relate to sample status in the same way as in Figure 4a. Most females showno development of loops, whether of reproductive age (filled bar) or not, while all males found inassociation with nests or broods had well-developed tracheal loops.

in populations by sexual selection, despite thepotential for the survival of juvenile males tobe compromised under some environmentalconditions.

Male contests (intra-sexual selection) andmate choice (inter-sexual selection) may bothbe important in driving evolution of larger,louder male Magpie Geese. Large body sizecould play a role in both arenas, and also havea direct functional role in the capacity todefend a nest or family group against predators(Whitehead, in press). A role for the trachea indirect male-to-male competition cannot be

discounted (eg Kodric-Brown & Brown 1984,Zahavi 1987), for example in jamming thesignals of competitors. However, the exerciseof considerable female discrimination in choiceof mates and maintenance of bonds is indicatedby the stable polygynous mating system (Frith& Davies 1961). Magpie Geese are highlymobile and do not defend recognisableterritories: foraging, nesting and brood-rearingsites vary daily, seasonally and from year-to-year (Frith & Davies 1961). It is unlikely thatcohesion of reproductive groups could bemaintained over long periods solely by male

sequestration of females or territory.In many vertebrates, females exercising

choice of mates based on acoustic stimuli selectstronger signals given at higher rates (Ryan &Keddy-Hector 1992). The greater volume andlower pitch of calls from males withexaggerated tracheal development may providefemales with an honest signal of male viability(eg body condition: Genevois & Bretagnolle1994), that is too expensive to be 'faked' by lessrobust birds (Zahavi 1975, Grafen 1991). Inaddition to this auditory signal, patterns ofgrowth in the cranial bump and the greaterfrequency of larger bumps among malesengaged in reproductive activity also suggest arole for mate choice based on this potentialvisual signal. No role unrelated to mateselection has been described or, so far as I amaware, suggested for the bump. Correlation ofhead height (and hence bump size) with bodyweight suggests that it may also signal bodycondition. However, the manner in which largecranial bumps might handicap males and soensure the honesty of the signal is less obvious.It is possible that a bird depending on an abilityto dig with its bill for food at depthsconsiderably greater than the length of its head(Whitehead & Tschirner 1991) may beinconvenienced by a 3-4 cm protrusionperpendicular to the shortest route to thosefood items.

But even if bump size, in isolation, offeredambiguous signals to females regarding malecondition, in combination with vocal attributesthis age-related trait may offer a highly reliablemeasure of quality. Evidence from this study(below) and Johnsgard (1961) suggest that ratesof increase in male tracheal lengths plateauafter about two years, before males arecommonly recorded in association with nests(Figure 3), whereas head height continues toincrease much longer (below). If costs ofassessing additional traits are low, female choicemay select for multiple signals in the malepopulation. Indeed, it has been suggested thatin birds carrying multiple ornaments, one traitwill often be an indicator of viability, whileothers are attractive 'Fisherian' traits,unconnected to wider variation in fitness, butmaintained by self-reinforcing inheritance of the


female preference and of the preferred trait inher male offspring (Iwasa & Pomiankowski1994). Older, big-bumped males with 10ud,low-pitched vocalisations demonstrate their viabilityby surviving long enough to develop largebumps (Buchholz 1991), despite theirprominent tracheal handicap. More detailedstudies of behavioural interactions between thesexes and, in particular, the initiation andmaintenance of bonds among group memberswill be required to improve understanding ofthe separate or joint influence of these traits onfemale choice.

Growth patterns and estimating age

Patterns of variation in head-bill and tarsuslengths and remeasurements of recapturedadult birds suggest limited post-fledging changein these dimensions. Bimodal distributions ofhead heights, with the lesser peak dominated bycohorts of first year birds and the larger byadults, clearly demonstrates that the bumpincreases in size with age in both sexes (Figure4b), although the increase is very much slowerin females (Frith & Davies 1961). While I wasunable to directly demonstrate growth ofcranial bumps by remeasurements ofrecaptured females, this reflects the inadequacyof the sample rather than robustlydemonstrating the absence of growth.

Birds with larger bumps tended to beheavier, even when other measures of body sizewere taken into account. This association couldhave many sources, including othermorphological change directly associated withmaturation, or older experienced birds beingbetter able to exploit a capricious environmentand hence maintain better condition thanyounger animals. I am currently unable toidentify the most plausible link. However, it willbe necessary to explore sources of variation inrates of bump growth and their associationwith other measures of condition if their size isultimately to be employed as a tool for ageingbirds (below).

Development of the male trachea beginsduring the first year and is substantiallyadvanced during the second, with doubledloops reaching close to the distal end of the


sternum (classes 3 or 4). Subsequent rates oftracheal development, or at least theirexpression in terms. of the development of adistinctive third loop on the right side of thesternum, appear to be somewhat variable.Head height was similar in birds with the thirdloop absent and in those with well-developedthird loops, suggesting that this tracheal featuremay not provide a reliable indicator of relativeage. More precise measurements of totaltracheal length rather than a series ofsomewhat arbitrary classes would probably benecessary to explore this question in detail.

Reproductive status, age and size

There is little evidence in these data thatgenerally larger birds of either sex are morelikely to be reproductively active. The span ofhead-bill and tarsus lengths in both males andfemales found in association with nests orbroods was similar to that of the population asa whole (Figures 4a and 4b). Tarsus and headbill lengths of 94.9% and 92.4% respectively ofnon-juvenile males exceeded the minimumdimensions of males found in association withnests or broods.

The range of bump sizes amongreproductively-active males was somewhatnarrower: about 71.0% of the non-juvenilesample had head heights exceeding theminimum measured in a nesting male. Perhapsmore importantly, among nesting males therewas a significant skew towards the upper end ofthe frequency distribution of head heights anda large gap in head heights between known-agejuvenile males and nesting males. If it isassumed that all males caught at nests have agenetic stake in the clutch, the growth andfrequency distribution data are consistent insuggesting that breeding in males is delayed toat least four years (Figure 3), withreproductive activity being significantly biasedtowards even older birds (Figure 4c). Delayedbreeding in males is consistent with intensecompetition for mates and/or places withinbreeding colonies (Zack & Stutchbury 1992).Moreover, over-representation of larger-bumped, older, males in the breeding populationmay be stronger than suggested by these data.

Some nests may be attended by younger'auxiliary' males, which may not have access tolaying females (Marchant & Higgins 1990,Whitehead, in press). One or more of these'auxiliaries' may have been included in oursample, and comprehensive genetic studies willbe required to clarify association between malereproductive activity, bump size and age.

In contrast, head height of juvenile andnesting females overlapped slightly (Figure 4),and the frequency distribution of head heightsof breeding females suggests little skew tolarger bumps. These observation are consistentwith suggestions by Frith & Davies (1961) thata large proportion of the female population isbreeding within two years.

Discriminating sex using morphometrics

Despite marked differences in meandimensions of male and female Magpie Geese,some overlap exists in all individualmeasurements, especially when first-year birdsare included. Nonetheless, a simple lineardiscriminant function derived from a smallsuite of simply and non-invasively measuredvariables can be used to sex wild birds withaccuracies exceeding 92%. The function isapplicable to free-flying birds of all ages, as thesample used in its derivation includes asubstantial number of recently fledged birds ofboth sexes. That this should be possible isunsurprising, given that Whitehead & Tschirner(1990) indicated that goslings as young as 20days could be reliably sexed employingdiscriminant analyses based on a small numberof morphometric characters, similar to thoserecorded by Frith & Davies (1961).

The discriminant function reported herewas derived from samples taken over arelatively small part of the species' total range,and its applicability to distant and hencepotentially distinct populations (eg NewGuinea or Queensland) is uncertain. Studies ofmovement patterns in the Top End of theNorthern Territory (Whitehead 1998) suggestthat there is considerable mixing ofpopulations, so it is likely to be applicablethroughout the core populations along thenorthern coast from Darwin to Murgenella,


Table 5. Estimates of sex ratios derived from 100 samples of 100 birds randomlyselected from the total sample population (n=787). Differences between the mean sexratios calculated from the sex of the randomly-selected birds and sexes assigned by thediscriminant function provide a measure of bias potentially arising from application in field studies.Bias is reduced if additional criteria for tracheal morphology are applied to birds assigned asfemales by the discriminant function.

Method

Nominal

Discriminant

Discriminant functionplus tracheal morphology

Sex ratio

Mean ratio+SD

1.024+0.215

0.941 +0.188

1.008+0.209

Mean bias+SD

-0.083+0.076 (8.1%)

-0.016+0.052 (1.6%)

north of Kakadu National Park.However, there is potential for some

systematic bias in sex ratios estimated from thisdiscriminant function alone. Males show greatervariance in all dimensions and hence are subjectto a greater risk of misclassification. MonteCarlo simulations drawn from a samplepopulation with a nominal sex ratio of about I: I(d': 9) indicate a mean bias of -8.1% in sex ratioestimates based on the discriminant function(Table 5). Incidence of misclassifications can bereduced greatly by checking assignments asfemale against criteria relating to trachealdevelopment. First, birds with two or morewell-developed loops are invariably male(Figure 4d). Second, any bird with no orlimited development of the cranial bump (headheight less than 52 mm) but having distinctdevelopment of tracheal loops (class 2 andabove) is almost certainly male (Frith & Davies1961, Whitehead et al. 1990b, Figure 2)because no female with head height less than55.3 mm was recorded with such loops. Theseattributes of tracheal morphology, used inconjunction with the discriminant function,reduce mean bias calculated from a similarseries of simulations to -1.6%. Given thepurposes for which sex ratio estimates arerequired and the sampling error inherent in anysample of achievable size (see, for example,

Brennan et al. 1991), I consider that thispotential bias can effectively be ignored.

Management implications

Despite some limitations, particularly in regardto numerical models of patterns of growth,these data and the associated analysesdemonstrate the potential to derive importantdemographic and other biological insights frommore comprehensive studies of morphometricvariation in the Magpie Goose. In particular,future studies should include capture-recaptureprograms to provide (i) information for robustmodels of growth in the cranial bump andsubsequent estimates of population agestructures and (ii) a substantial markedpopulation to examine variation in malereproductive success and its morphologicalcorrelates. Such studies of marked birds willalso help provide age-specific measures ofvariation in dispersal patterns, fecundity andsurvival rates, all of which are needed forinformed management.

Taxonomic uniqueness, an extraordinarybiology, and a particular sensitivity to habitatchange, as evidenced by displacement from alarge part of the species' previous Australianrange, all justify special efforts to improveunderstanding of the factors influencing the

Iwasa, Y. 1996. Sexualin Evolution and Ecology


status of this icon of north Australia's tropicalwetlands.

The study was funded by the Parks and WildlifeCommission of the Northern Territory. Manyassisted in the captures and measurement ofbirds. In particular, I thank David Wurst,judithGal/en, Kurt Tschirner,james McKinnon, StevenTurner, Donna jackson, Oswald Tory, and Vicki,Scott,jason, Kim and Adam Whitehead. Themanagement of Marrakai and Opium CreekStations provided unrestricted access to the studysites and assisted ih various ways. Peter Dostine,Richard Kingsford, and John Woinarski madevaluable critical comment on earlier versions ofthe manuscriPt. Amotz Zahavi made a numberof suggestions regarding the likely significance ofmale 'ornamentation' in the Magpie Goose.

References

Andersson, M. &selection. TrendsI 1:53-58.

Brennan, L.A, Buchanan, J.B., Schick, eT. &Herman, S.G. 1991. Estimating sex ratioswith discriminant function analysis: theinfluence of probability cutpoints and samplesize. journal of Field Ornithology 62:357-366.

Buchholz, R. 1991. Older males have biggerknobs: correlates of ornamentation in twospecies of Curassow. Auk 108: 153-160.

Dexter, N. 1988. The effect of experimentalclutch harvest on Magpie Geese Anseranassemipalmata in subcoastal Northern Territory,Australia. Master of Applied Science Thesis,Canberra College of Advanced Education,Canberra.

Endler, J.A 1992. Signals, signal conditions, andthe direction of evolution. AmericanNaturalist 139:S125-S 153.

Fabens, AJ. 1965. Properties and fitting of thevon Bertalanffy growth curve. Growth2:9265-289.

Frith, H.J. & Davies, S.J.J.F.1961. Ecology of theMagpie Goose, Anseranas semipalmataLatham (Anatidae). CSIRO Wildlife Research6:91-141.

Genevois, F.& Bretagnolle, V. 1994. Male BluePetrels reveal their body mass when calling.Ethology Ecology & Evolution 6:377-383.

Grafen, A 1991. Modelling in behaviouralecology. In: Behavioural ecology: an evolutionaryapproach (Krebs, J.R & Davies, N.B. (Eds.).).Blackwell Scientific Publications, Oxford.

Hinds, D.S. & Calder, WA 1971. Tracheal deadspace in the respiration of birds. Evolution25:429-440.

Horn, P., Rafalski, AJ. & Whitehead, P.J. 1996.Reproductive biology of Magpie Geese usinga molecular genetic (RAPD) analysis. Auk113:552-557.

Iwasa,Y.& Pomiankowski,A I994.The evolutionof mate preferences for multiple sexualornaments. Evolution 48:853-867.

Johnsgard, P.A 1961. Breeding biology of theMagpie Goose. Report of the Wildfowl Trust12:92- 103.

Johnsgard, P.A 1978. Ducks, geese and swans ofthe world. University of Nebraska Press,Lincoln.

Johnson, LP.& Sibly,R.M. 1989. Effects of plasticneck collars on the behaviour and breedingperformance of geese and their value fordistant recognition of individuals. Ringing andMigration 10:58-62.

Kodric-Brown, A & Brown, J.H. 1984. Truth inadvertising: the kinds of traits favoured bysexual selection. American Naturalist 124:309-323.

Maclnnes, CD., Prevett, J.P.& Edney, HA 1969.A versatile collar for individual identificationof geese. journal of Wildlife Management33:330-335.

Marchant, S. & Higgins, P.J. 1990. Handbook ofAustralian, New Zealand and Antarctic birds.(I st ed.) Oxford University Press, Melbourne.

Prom islow, D., Montgomerie, R. & Martin, T.E.1994. Sexual selection and survival in northAmerican waterfowl. Evolution 48:2045-2050.

Ryan, M.J.& Keddy-Hector,A 1992. Directionalpatterns of female mate choice and the roleof sensory biases. American Naturalist 139:S4-S35.

SAS Institute Inc. 1989. SAS/STAT User's Guide,Version 6. (Fourth ed.) Cary, NC, SASInstitute Ine.

Whitehead, P.J. 1998. Dynamics of habitat use bythe Magpie Goose Anseranas semipalmata:implications for conservation management.Unpublished PhD thesis, Northern TerritoryUniversity, Darwin.

Whitehead, P.J.(in press). Aspects of the nestingbiology of the Magpie Goose Anseranassemipalmata: incubation period, hatchingsynchrony and patterns of nest attendanceand defence. Emu 99.

Whitehead, P.J.& Tschirner, K. 1990. Eggs andhatchlings of the Magpie Goose Anseranassemipalmata. Emu 90: I54-160.

Whitehead, P.J.& Tschirner, K. 1991. Lead shotingestion and lead poisoning in magpie geeseAnseranas semipalmata foraging in a northernAustralian hunting reserve. BiologicalConservation 58:98-1 I2.

Whitehead, P.j., Bayliss, P.G. & Fox, R.E. 1988.Recreational hunting activity and harvests inthe Northern Territory, Australia. AustralianWildlife Research 15:625-63 I.

Whitehead, P.J.,Wilson, BA & Bowman, D.M.j.S.I990a. Conservation of coastal wetlands of theNorthern Territory ofAustralia: the Mary Riverflood plain.Biological Conservation 52:85- Ill.


Whitehead, P.J., Freeland, WJ. & Tschirner, K.I990b. Early growth in the Magpie GooseAnseranas semipalmata: sex differences andinfluence of egg size. Australian Journal ofZoology 38:249-262.

Whitehead, P.J.,Wilson, BA & Saalfeld, K. 1992.Managing the Magpie Goose in the NorthernTerritory: approaches to conservation ofmobile fauna in a patchy environment. In:Moffatt, I. & Webb, A. (Eds.). Proceedings ofConservation and Development Issues inNorthern Australia. North AustralianResearch Unit, Australian NationalUniversity, Darwin. Pp 90-104.

Zack, S. & Stutchbury, B.J. 1992. Delayedbreeding in avian social systems: the role ofterritory quality and 'floater' tactics.Behaviour 123: 194-219.

Zahavi,A. 1975. Mate selection - a selection fora handicap. Journal of Theoretical Biology53:205-214.

Zahavi, A. 1987. The theory of signal selectionand some of its implications. In: Delfino, VP.(Ed.). Proceedings of InternationalSymposium on Biological Evolution. AdriaticaEditrice, Bari. Pp 305-327.

IN THE MAGPIE GOOSE ANSERANAS SEMIPALMATAespace.cdu.edu.au/eserv/cdu:260/Whitehead_260.pdf · IN...

Documents

Transcript of IN THE MAGPIE GOOSE ANSERANAS SEMIPALMATAespace.cdu.edu.au/eserv/cdu:260/Whitehead_260.pdf · IN...