The timing of egg laying, clutch size, and egg size of the American Oyster- ...... White-rumped Sandpiper. 6 ..... mere Island· N.W.T. Ibis 115: 202-217. NISBET ...
CLUTCH
INITIATION
EGG SIZE OF THE
DATES, CLUTCH
AMERICAN IN
SIZE, AND
OYSTERCATCHER
VIRGINIA
ERICA NOL, • ALLAN J. BAKER, 2 AND MICHAEL D. CADMAN3 •Department of Zoology,Universityof Toronto,Toronto,OntarioM5S 1A1, Canada;and 2Department of Ornithology, RoyalOntarioMuseum,100 Queen'sParkCrescent, Toronto, Ontario M5S 2C6, Canada
ABSTRACT.--The timing of egg laying, clutch size, and egg size of the American Oystercatcher(Haematopus palliatus)were studiedover six consecutivebreedingseasonsin Virginia. Synchrony of laying datesoccurredin each of five localitiesof the study area in at least one year. Mean clutch size was 2.8 eggs (mode = 3) in first clutchesand 2.4 eggs(mode = 2) in replacementclutches.Individual females laid replacementclutchesof the samesize and laid eggsof similar averagevolume in all years.A change in mate had little effect on the date on which femalesinitiated their first clutchesin successive years.The averageegg size in a clutch was correlatedwith the size of the laying female. Egg-sizeordering occurredwithin clutches,the first-laid egg being smaller than the secondegg and about equal in volume to the third. We proposethat the secondegg is largest becauseit has the highest probability of hatching, and the resulting sibling hierarchy reducesthe frequency of sibling competition. Received19 October1983,accepted19 April 1984. SYNCHRONY of clutch-initiation
dates, uni-
Clutch size in this group is usually less than four eggs,and this may be related to the extensive parental care (Maclean 1972). Egg-size ordering within a clutch is undocumented in oystercatchers, but it might be expectedto oc-
form egg size, and four-egg clutches are assumed to be adaptations of many shorebirds (Charadrii and Scolopaci)for living within the constraints of high-latitude breeding seasons. Clutch-initiation datesoccur over a very brief period (Holmes 1971), and replacementclutch-
cur in them
es are uncommon (Pitelka et al. 1974), because the period of abundant food resourcesis brief
and the season is telescoped (Holmes 1972, Nettleship 1973, Pitelka et al. 1974). A definitive clutch of four eggsof similar size and shape apparently forms the optimal configuration for minimizing the rate of heat losswhen the clutch is uncovered (Norton 1970).
Uniform egg sizewithin a clutchalsoimplies no differential allocation of parental care to the chicks (Miller 1979). Uniform chick size may be as important a consequenceof four similarly sized eggs as that provided by the energetic advantageduring incubation, particularly because a definitive
clutch
of four
rather
than
in shorebirds
that
do
not feed their young. This is because,through preferentialfeeding,oystercatcher parentshave a greater potential for "control" over which of their chickssurvive (Alexander 1974). This potential control might be extended to the eggs, particularlyif food is limiting or variable(Howe ! 976), if preferential feeding is impractical,and if the probability of raising all chicksis dependent on the food supply. The purpose of this paper is to document variations in clutch size and egg size in American Oystercatchers(Haematopus palliatus)and to elucidate seasonaland annual patterns in the timing of egg laying. By comparingindividually known females over several breeding seasons,we have attempted to gauge repeat-
is also found
in many shorebirds nesting in the temperate
ability measures(Falconer 1981) of these char-
zone (Maclean 1972). Oystercatchers(Haematopodidae) are one of
acteristics, as well as the effect of female size.
the few familiesof shorebirdswhoseyoungde-
METHODS
pend almost exclusively on the adults for food.
Study area.--Study sites were located on Wallops and Assawomanislands,Virginia (37ø50'N,75ø35'W) in
3Present address:364 Waterloo Avenue, Guelph, Ontario N1M 3K2, Canada.
1978-1983
and
on the
salt marshes
and
dunes
aroundthe ChincoteagueChannel, Virginia (37ø55'N, 75ø23'W) in 1981-1983. For purposes of comparing
855
The Auk 101: 855-867.
October
1984
856
IXToL, BAKER, ANDCADMAN
WALLOPS MAIN
[Auk,Vol. 101
STATION BASE
I SLA'"'N D
O&/4,0 0
ATLANTIC
OCEAN
KM
o
1
2
3
i
I
I
I
Fig. 1. Map of study area including five breeding sites:(1) ChincoteagueIsland (mean of 5 nests per year), (2) ChincoteaguePoint (œ= 13 nests),(3) northern Wallops Island (œ= 7 nests),(4) the sand spit of southern Wallops Island (œ= 8 nests),and (5) AssawomanIsland (œ= 6 nests).
the synchronyof clutch-initiationdates among lo- the period ! March to 31 July for six successive calities,we divided the studyarea into five localities: breeding seasonsfrom 1.978to 1983.Observationsin (1) ChincoteagueIsland, (2) ChincoteaguePoint, (3) all seasonsbeganby at least 6 April. northern Wallops Island, (4) the sand spit of south-
ern WallopsIsland, and (5) AssawomanIsland (Fig. 1). Fieldwork was conductedat the study siteswithin
Data collection.--Overthe study period, 60 adults (30 females,30 males) were trapped at the nestswith
drop trapslined with fish net (Mills and Ryder 1979).
October1984]
EggLaying in American Oystercatchers
The adults were banded
with
aluminum
bands and
unique color-band combinationsand were weighed and measured. From 1978 to 1980, all territories were
searchedperiodically,although we were not always certain of clutch sequence(i.e. whether first or replacementclutch)or the order of egg laying. To obviate this problem, we searchedterritories throughout the prelaying and laying periods in the 19811983breeding seasons. Eggs were marked, weighed, and measured.All weights used in the analyses are of eggs weighed within 48 h of being laid. Weight can be difficult to determine in the field, particularly on windy days, whereaslength and breadth measuresare more precise. We therefore wanted a method for predicting volumes. Sixty-eight eggs from museum collections were
filled
with
water
to determine
their
volumes.
Regressioncoefficientswere calculatedwith volume as the dependent variable and the product of egg length (EL) times breadth2 (EB2) as the independent variable (Vaisanen 1969). The resultant formula was used to calculatethe volume of all eggs measured: Volume (cm3)=0.47736 x EL (mm) x EB2 (mm •)1.318(mm3);r2 = 0.96. Fresh egg weight (EW) was a good predictor of the calculated volume: Volume (cm3)= 0.757 EW (g) + 5.96 (cm3);r• = 0.87, n = 170. In the subsequenttext, egg size and volume are used synonymously. Statistical analysis.--Theclearestmethod for detecting the effects of the female, clutch sequence,and year on egg size is to look at the effectsof each factor in two two-way designs (Sokal and Rohlf 1981). We
857
acter (Falconer 1981). Heritability, the resemblance between relatives, may be much less than repeatability, but it cannot be greater (Falconer1981). The within-individual component arises from environmental
fluctuations
between
successive measures and
the systematiceffectsof age.The between-individual componentcomprisesboth permanentenvironmental effectsand a genetic component(Findlay and Cooke 1982).We calculatedrepeatabilitymeasuresby using the resultsof a one-way analysisof variance applied to (1) egg size among femalesacrossyears, (2) egg size among femalesacrossclutcheswithin a year,and (3) date of first clutchinitiation acrossyears. A three-way contingency table was constructedto observethe pattern of egg-sizeordering in successive years. The table was analyzed using log-linear models after the example of Fienberg (1970) and Bishop et al. (1975). In addition to testing the assumptionof mutual independence,which characterizes classicalcontingencyanalyses,the techniquesof Fienbergcan be used to calculatethe frequenciesof events that would be expected under all possible conditionsof dependence(interaction)betweentwo or moreclassifyingvariables.Goodness-of-fit testsare then used to determine
which
of a series of hierar-
chical models that are basedupon various assumptions of dependencebest explain the observedfrequencies of the event (Harder 1980). If social stimulation tends to promote synchrony
in the initiation of first clutchesin neighbors,then femalesin a locality should tend to lay at more similar dates than
females
in different
localities.
One
testedfor the effectsby using a mixed-model,twoway analysisof variance (ANOVA) with femalesas
way to test for synchronizationis to comparethe
a random factor, and assumed no interaction be-
in one area (Sl2) with the variance of observed clutch-
tween femalesand the fixed factor(e.g. year or clutch sequence).We then estimatedthe variation attributable to femalesand to the fixed factor and the significancelevelsof the two factorsby usingthe inter-
initiation datesof all other birds (s2•) and to compute
action
term
as the error
term.
We used mean
values
to avoid the confoundingeffectsof different cell sizes produced by differently sized clutches.When using year as the main effect, we examined only first clutches to avoid any confounding of clutch sequence in the design. In the comparisonof means, we used t-testson every pair of means,while controlling the experiment-wiseerror rate by lowering the value of a to a/n where n is the number of paired comparisonsmade. To test for differencesin egg sizeacrossyears,across clutches within a year, and within differently sized clutcheswe useda one-way analysisof variance.We also tested for differencesin egg size as a result of egg order by means of a one-way analysisof variThe ratio of the between-individual component of variance
to the total
variance
measures
the correla-
tion between repeatedmeasurementsof the sameindividual and is known as the repeatability of a char-
variance
of observed
clutch-initiation
dates of birds
the ratio of variances (s2•/s?). We obtained 500 ran-
dom partitionsof the data into samplesof size n• (the number of nestsin the locality in question)and n2 (the number of total clutches initiated in a year - n•), computed the variance ratio for each,and then looked to see if our observed
value fell in the
upper 5% tail of the distribution of 500 empirically obtained ratios (Sokal and Rohlf 1981). RESULTS
Clutch initiation.--American Oystercatchers
breeding in Virginia initiated clutchesover a short period. Initiation dates of first clutches rangedfrom asearly as 6 April (1981)to as late as 13 May (1982); the mean spreadof datesfor 1981-1983 was only 25 days,however. The averagedateof initiationof firstclutchesfor 19811983 was progressivelylater in the successive breeding seasons(Fig. 2). Frequencydistributions of clutch-initiation dates before 28 April are similarly shaped in the 1981 and 1982
858
NOL,BAKER, ANDCADMAN
[Auk,Vol. 101
14
M
•
1983
I0
6
2
14
•M
•
1982
I0
6
.-I
2
o
14
•M 1981
I.l.I
I0
m
6
2
• IOO o
*"'•'"*' .... *"'"'.....• 2eggs
5o
3eggs
• •
o m
•
o
o
e-------
14
22
30
APRIL
8
16 MAY
24
I
9
17
JUNE
Fig. 2. Frequencyhistogramsfor 1981-1983breedingseasonsillustratingnumber of first clutches(open bar),secondclutches(graybar),third clutches(crosshatched bar),and clutchesof unknownsequence(bold stippledbar). Locationof mean(Z) and median(M) layingdatesare indicatedby arrows.The bottompanel showsthe proportion of two- and three-egg clutches(n = 156).
breeding seasons.In 1983, the curve is flattened from an increasein the spreadof clutch-
26 April 1981 causedmost nests to be lost, and
most likely causedthe shift in dutch-initiation
lost to both spring tides and predation at about the sametime. This pattern of nest lossresulted in a small secondpeak of laying from 5 to 12 May. The pairs representedby this peak lost their nestsa secondtime to predation,all with-
a peakof renestingoccurredfrom 12 to 18 May. initiation dates, and the mean date of first clutch In 1982, the spring tides occurred before the initiation is slightly later. The 1983 field season majority of females had laid and after eggs in was characterizedby a cold, wet April, and this the first clutch had hatched. In 1983, nests were dates.
The time of initiation of replacementclutches was related
to the date on which
the first
clutch was lost. High spring tides during 24-
October1984]
EggLaying inAmerican Oystercatchers
859
TABLE1. Clutch size of Haematopus palliatuson Chincoteague,Wallops, and Assawomanislands,Virginia in six breeding seasons,1978-1983.
Number of eggsin
Number of eggsin
first clutch Year
Number of eggsin
second clutch
third 2
clutch
n
!
2
3
4
ua
1
2
3
ua
1
1978 1979 1980 1981 1982 1983
33 43 18 83 54 63
0 1 0 1 0 0
1 14 0 2 4 4
2 18 0 31 31 19
1
0 0 0 6 6 13
0 1 0 1 0 1
0 3 0 17 9 8
0 4 0 11 1 6
5 2 1
0 0 0 0 0 0
0 1 0 4 1 6
0 1 0 1 0 1
3
Total
294
2
25
101
1
25
3
37
22
8
0
12
3
Number of eggs in clutchesof unknown
ua
3
3
1
2
number 3
ua
2 0 2 0 0 1
9 0 6 1 0 2
18 0 10 0 0 0
1
5
18
28
1
clutcheswith unknown number of eggs.
in 4 days, resulting in a third peak of laying in late May and early June (Fig. 2). Approximately three quarters of the total
Synchrony of clutch-initiation dates of first clutches(as defined by a significant s22/s• 2ratio,
variance
least one of the three years of intensive study (1981-1983). Females nesting on the sand spit
in
clutch-initiation
dates
was
ex-
plained by the effectsof female and year. About 59% of the variation
in clutch-initiation
dates
was explained by variation between females (P < 0.001) and about 16% by variation between years (P < 0.001). The repeatability (ra) measure
for date of clutch
initiation
for those
femalesthat laid in three consecutiveyearswas high (total phenotypicvariance= 97.8, ra= 0.73, n = 22) and indicates that females tend to lay at about the same time each year. An examination of the year-to-yearcorrelationsbetween clutch-initiation
dates, however, makes clear the
factsthat in 1983femaleswere lesslikely to lay at the same dates as in 1981 and 1982 (19811982, r = 0.740, P < 0.0001, n = 22; 1982-1983, r = 0.485, P < 0.02, n = 23) and were also less
likely to lay on the same datesrelative to their neighbors (Spearman rank correlation coefficients: 1981-1982, r = 0.816, P < 0.0001; 1982-
see methods) occurred in all five localities in at
of southern Wallops Island in 1982 were more
synchronousthan the general population (six nests, P < 0.05). There was evidence of synchrony in all three yearsamong the femaleson Chincoteague Point (nine nests: 1981, 0.05 < P < 0.1; 1982, P < 0.01; 1983, P < 0.01).
At the north end of Wallops Island, there was significant synchrony in 1983 (four nests,P < 0.01) but not in either 1981 or 1982. Synchrony was also detected in 1983 on Chincoteague Island (5 nests, P < 0.01) and in 1981 on Assawoman Island (5 nests, P < 0.01).
Clutchsize.--The averageclutch size declined as the seasonprogressed.Most first clutchesin 1981-1983containedthree eggs(Table 1). About two-thirds of first replacement clutches (second clutch) contained two eggs. Second replacementclutches(third clutch) containedtwo eggs in all but three nests. In 1979, 42.4% of first clutches contained two eggs, and in 1978
1983, r = 0.463, P < 0.02). Again, a cold wet April and persistenthigh tides in somenesting areascausedsome females to lay at a later date one clutch (3.0% of all clutches) contained four in 1983 than in previous years, and this fact eggs. alone would lower the year-to-year correlation. The proportion of three-egg clutches deWhen known females on a territory attained clined through the breedingseason(Fig. 3), but new mates (n = 4) in successivebreeding sea- females laid three-egg clutchesas late as 1 June sons,they still tended to lay at about the same in 1981 and 2 and 13 June in 1983 (Fig. 2). Fedate as in the previous year. In the two in- malesnormally lay a maximum of two replacestances when males retained their territories ment clutcheswhen precedingclutchesare lost, but acquired different matesin the subsequent but one female laid three replacement clutches breeding season,however, these new females in 1979,althoughone and possiblytwo of these laid 16 and 26 days later than the original res- were incomplete. Females that laid replaceidents. ment clutches in 1981 and 1983 tended to lay
860
NOL,BAKER, ANDCADMAN
[Auk,Vol. 101
TABLE2. Egg dimensions(mean + SE) of first, second,and third laid eggsof H. palliatusin Virginia.
Egg sequence • Volume (cc) Length (ram) Breadth (ram) Weight (g)
Eggs of
1
2
3
unknown
(n = 116)
(n = 95)
(n = 69)
(n = 241)
number
41.68 + 0.27a 56.05 _+0.21a 39.46 +_0.11a 47.69 + 0.42a
43.22 + 0.27b 57.09 + 0.19b 39.82 + 0.11a 49.65 _+0.40b
42.28 + 0.32a 56.95 + 0.20b 39.43 + 0.14• 47.72 q- 0.53a
42.34 + 56.31 + 39.68 + 49.30 +
0.18 0.13 0.07 0.50
• Homogeneousgroupsindicatedwith superscripts. Differencestestedusing Multiple t-tests(P < 0.015) following one-way ANOVA (Sokaland Rohlf 1981).
the same sized clutches (X2 = 3.19, df = 1, n = 13, 0.05 < P < 0.10).
Eggs/ze.--Egg size in the combined data set (including data for all eggs of known laying sequence,regardlessof clutchnumberor clutch size) varied systematicallywith laying order (Table 2). On average,the secondegg laid was the largest(mostvoluminous)and the heaviest. The first and third eggs were about equal in volume. When we categorizedthe dataaccording to clutch sequence,we obtained a similar result. In first clutches (a sample of mostly three-egg clutches),the secondegg was larger and heavier than the first (P < 0.015) and
heavier than the third (P < 0.015). The first and
third eggswere of equal size,but the third egg was longer (P < 0.015). In second clutches (a
the clutch. In three-egg clutches, the second egg was largest and heaviestand the first and third eggswere equal in size and weight (P < 0.015).The secondegg was longer than the first (P < 0.004) and wider than the third (P
_ 2
1 < 2
1 (-0.1) 3 (0.8) 2 (-1.1) 3 (-1.7) 4 (0.4) 12 (2.6) 2 (-0.1) 3 (-0.4)
3 (-0.2) 5 (-0.3) 18 (1.3) 25 (0.5) 4 (-1.9) 16 (-0.3) 10 (1.3) 9 (-0.7)
30
90
selected models of interaction
between
clutch size
(C), presenceof egg-sizeordering (E), and year (Y) in the data from Table 3. For a given model, variablesthat appear together within commaswere assumed to be jointly dependent, but the effect of that interaction is independent of the other specified interactions or single factor effects?
Total
4 8 20 28 8 28 12 12 120
• Standardizeddeviates= (Obs.Exp)/EX/-• for modelof completeindependence(Bishopet al. 1975).
Model
G2
df
E, C, Y
19.99
10
0.05
EY C, EY
15.05 6.41
8 7
0.05 < P < 0.1 n.s.
E, CY Y, EC EC, EY EC, EY, CY
14.16 19.21 5.63 0.53
7 9 6 3
0.05 0.05
• Probabilities
P
n.s. n.s.
less than 0.05 that are associated with
modelsof equal or lesscomplexityindicate their inability to provide adequateexplanationsof the data (Sokal and Rohlf 1981).
October1984]
861
EggLayingin AmericanOystercatchers
TABLE 5. Effectof clutchsize on variation in egg-sizeparameters(mean ñ SE). Clutch
size
Egg
2-egg (n = 68)
3-egg (n = 286)
Volume (cc)
42.4 + 0.29
42.8 ñ 0.15
Length (mm)
56.9 + 0.25
56.6 + 0.12
Breadth (mm)
39.5 + 0.14
39.8 + 0.05
Weight (g)
48.7 + 0.44•
48.7 -+ 0.27
pa
0.1
