How do Black-headed Gulls distinguish
between eggs and egg-shells?
By N. Tinbergen, H. Kruuk, M. Paillette, and
R. Stamm*
(Plates 20-2i)
INTRODUCTION
{Lams ridibundus) remove the empty eggshell shortly after the hatching of the chick. An investigation into
the survival value of this response and of the stimuli eliciting it (Tinbergen et al. 1962a) yielded the following results. An empty egg-shell
at a distance of four inches from a single egg laid out in the dunes
rendered such an egg more vulnerable to predation by Carrion Crows
(Corvus coroni) and Herring Gulls (L. argentatus); this suggests that
egg-shell semoval may help to reduce predation. The response can
be elicited throughout the incubation period by a variety of objects
in the nest or onlts rim, in fact by "any object which does not resemble
an egg, a chick, or nest material", although the egg-shell itself is
optimal. In systematically conducted tests with dummies, it was
found that colour, shape, size and distance between nest and shell
affected the response.
These results suggested, therefore, that this seemingly trivial response, which normally takes no more than twenty seconds of a bird's
time each year, contributes to the survival of the brood; and they
showed that it is controlled by a complicated and well-adapted
mechanism.
This first study, while not allowing us to list all the characteristics
of the situation which elicits egg-shell removal, showed that the birds
distinguish between egg-shell and nest material by at least four
characteristics: in fact, the egg-shell is distinguished by being threedimensional, rounded, less oblong than nest material, and partly white.
The present paper deals with the question of how gulls distinguish
between egg-shells and eggs. Failure to do so would naturally
BLACK-HEADED GULLS
*Dr. Tinbergen is from the Department of Zoology, Oxford University; Mr,
Kruuk from the University of Utrecht, Netherlands; Miss Paillette from the
Laboratoire de Physiologic Acoustique, Jouy-en-Josas, France; and Dr. Stamm
from the Zoologisches Institut der Universitat, Basel, Switzerland,
continued ...
120
GULLS DISTINGUISHING EGGS AND EGG-SHELLS
F I G . I . The first stages of egg-shell removal (Jeff) and egg retrieving {right)
in the Black-headed Gull {Larus ridibmdiis) (see below) {sketches: N. Tinbergen)
endanger the brood, and we have actually never observed a gull
removing its eggs or chicks.
Although an intact egg and an egg-shell as left after the chick has
hatched have much in common, they elicit two entirely different sets
of responses. An egg in the nest is brooded and occasionally shifted.
If an egg happens to He on the nest's rim or even a little outside the
nest, the bird often retrieves it. It may either sit down first or remain
standing in the nest, then may gently touch the egg's upper or distal
surface with the ventral side of its bill. This is then often followed by
"rolling in", which is done by bending over the egg and balancing it
against the under surface of the lower mandible, then rolling it back
towards the breast, i.e. into the nest cup (plate 20b).
A shell in or near the nest is taken in the bill (plate 20a). The bird
gets bold of the thin edge and either walks or flies away with it at once,
or nibbles it for a while before removing it. Nibbling may be interrupted when the bird drops the shell, but it ultimately leads to "carrying". Sometimes a gull pecks at the material which is left behind in
the shell and it may eat some of it.
121
B R I T I S H BIRDS
The situations eliciting the two responses in our experiments were
the same except for the objects (egg and egg-shell) themselves.
The differences we see between an egg and an egg-shell could be
described in various ways, and the following descriptions guided us
in our attempts to find out to which properties the gulls reacted;
(i)
(2)
(3)
(4)
(5)
(6)
Egg
Smooth, oval outline
Closed; no opening
No thin edge or rim
Heavy (c. 37 gm.)
Dotted khaki; no-white
Total egg-coloured surface
Egg-shell
Interrupted, partly serrated outline
With opening; hollow
A thin edge
Light (c. 2 gm.)
White on rim and inside
Small egg-coloured surface
Of these, all except (4) could be perceived visually. Since both
responses are initiated by visual stimuli (the birds focusing the
objects from a distance, irrespective of the direction of the wind, and
visually similar dummies being effective), weight could not play a
part in the release of the response, but it might have an effect once the
bird has made contact.
METHOD
This study was carried out at the Black-headed Gull colony at Ravenglass, Cumberland, in the summer of 1961. Various models, to be
described later, were presented, one at a time, on the rims of nests
containing two or three eggs, and the birds' responses were observed
from hides approximately eight feet away. Each bird was given each
model only once; and each was tested with all models of a series.
The sequence of presentation was varied at the different nests, as much
as possible according to a latin square arrangement. Each individual
test lasted either 15 minutes or—if a bird completely performed one of
the two end-acts (rolling-in or carrying) before the 15 minutes were
over—until either end-act had been completed. After each test, the
observer signalled to an outside helper without disturbing the bird.
The helper then approached, inevitably forcing the bird to fly up, and
as quickly as possible prepared the next test situation. With each nest
the models compared were always put out in the same place.
We could usually distinguish the two partners of a pair and were
able, therefore, to test individual birds. However, because we could
not control the moment of nest-relief, some series had to be broken off
before the relieved bird had been tested with the entire series of
models. The number of presentations varies slightly in each experiment, therefore.
The following terms were used to characterise the birds' behaviour:
Billing: touching the model's surface with the bill, which is usually closed (this
generally precedes rolling)
122
GULLS D I S T I N G U I S H I N G
ECGS AND
EGG-SHELLS
Intention-rolling: bending the neck over the model and touching its distal surface
without actually moving it
Rolling: pushing the model in the direction of the nest cup, irrespective of success
Lifting: picking up the model, then dropping it without carrying
Nibbling: mandibulating the rim of the model
Intention-carrying: getting hold of the model with a stronger grip, and often dragging
it a little way
Carrying: picking up the model, walking or flying away and dropping it at some
distance from the nest
Poking: pecking into the hollow model, sometimes eating material left inside
Ignoring: making no response to the model, apart from looking at it
(For the statistical tests employed, which are mentioned separately
below, see Siegel 1956.)
Experiment A,
This was a pilot experiment in which no more than the two end-acts
(carrying and rolling) were recorded; it was designed to see whether
clear-cut results were at all possible with small numbers of birds.
Empty eggs with holes of different sizes were used in this test. For
reasons not to be discussed here, the holes were made at the sharp ends
of the eggs (the hatching chick invariably cuts off a "lid" at the blunt
end) and the edge of each hole was smooth, not serrated. We used
eggs of approximately the same size; the size of each hole was expressed indirectly in the projected distance between its edge and the blunt
pole of the egg.
The following models were used: "Small hole" (empty egg with
edge of hole 51 mm. from end); "Medium hole" (similar but 45 mm.);
"Large hole" (similar again but 36 mm.); and "Real shell" (egg-shell
from which a chick had actually hatched the previous year). The
models are all illustrated on plate 21 (b, c, d and h). The diameter of
"Large hole" was approximately the same as that of the hole of a real
shell. Each model was presented to twelve birds. Table 1 summarises the results.
TABLE I—RESULTS OF EXPERIMENT A WITH BLACK-HEADED GULLS
{Icarus ridibundm), C U M B E R L A N D , 1961
This experiment was confined to recording the end acts of rolling and carrying
(see above and plate 20). The total column gives the number of presentations of
each model egg. The names of the model eggs and the small letters with them
correspond to those on plate 21
Total
Rolling
Carrying
Neither
presentations
(b)
(c)
(d)
(h)
Small hole
Medium hole
Large hole
Real shell
9
9
4
0
1
2
12
1
2
12
6
9
2
12
3
12
123
B R I T I S H BIRDS
For computing the significance of these results, the Fisher exactprobability test was used. The scores for "Small hole" and "Medium
hole" are significantly different from those for "Large hole" (p<.oo5).
Thus we reached three conclusions. First, a real shell is not rolled in
(though in other tests it sometimes was, as can be seen from, for
example, Table 2); when the bird responds to it at all, it removes it.
Second, the hole must be fairly large to make the bird switch over to
carrying in part of the tests. Third, a real shell elicits more carrying
and less rolling than even "Large hole".
These results were, however, difficult to interpret. We did not
know whether the birds reacted to the size of the hole, the hollowness,
the extent of the rim, the surface area of khaki-colour or what; nor
did we know whether the difference between "Real shell" and "Large
hole" was due to the serrated edge or to the former's showing more
white.
Experiment B
We next presented the following four models to new birds, and
observed their responses in more detail: "Empty egg" (a blown-out
egg, with hardly visible holes at both ends); "Filled shell" (a real
egg-shell filled to the rim with plaster of Paris, so that it was not
hollow and did not show a thin edge, but did have an interrupted and
partly serrated contour); "Shell-with-lead" (a real egg-shell with,
inside and at the obtuse end, a piece of white-painted lead equal in
weight to a real egg); and "Real shell" (as in Experiment A). "Shellwith-lead" was heavier than "Filled shell". These models are also all
illustrated on plate 21 (a, e, g and h). Table 2 summarises the results.
TABLE 2—RESULTS OF EXPERIMENT B WITH BLACK-HEADED GULLS
Lifting*
Billing
Nibbling
Poking
Ignoring
Total
presentations
Empty egg
Filled shell
Shell with lead
Real shell
Carrying
(a)
(e)
(g)
(h)
Rolling*
(Larus ridibundus), C U M B E R L A N D , I 96 I
The responses of the birds presented with the model eggs were divided into the
categories listed (pages 122-123). Several birds showed more than one response
and so the sum of the individual columns exceeds the total number of presentations
in each case. The names of the model eggs and the small letters with them correspond to those on plate 21
24
0
0
10
0
0
0
26
20
0
0
13
7
4t
3
5
13
7
1
21
2
4
16
27
*5
27
4
5
12
3
1
3
*These categories include intention-rolling and intention-carrying respectively
t Doubtful responses
124
GULLS D I S T I N G U I S H I N G EGGS AND E G G - S H E L L S
" E m p t y eggs" were always rolled in the same way as normal eggs.
In Table 2 there is no significant difference in the treatment between
"Empty egg" and "Filled shell", at least as far as the frequency of the
responses is concerned. However, the latency of the response (i.e.
the time between presentation and the start of the response) is significantly larger for "Filled shell" than for " E m p t y egg" (p=.oo4*).
For other conclusions we must bear in mind that some birds showed
responses of both systems to one and the same model. As a basis of
our statistical treatment, therefore, we took the numbers of birds which
showed a certain response to some models but not to others, and
compared these figures for each response with respect to different
models.
The following conclusions can be drawn about rolling and intentionrolling. In all, 15 birds showed either rolling or intention-rolling to
"Empty egg" but not to the other models, not a single bird did the
opposite, and eight treated the models in roughly the same way.
These figures are significant ( p < . o o i ) .
Further, 13 birds showed rolling or intention-rolling to "Filled
shell" but to neither "Shell-with-lead" nor "Real shell", one did the
opposite, and twelve responded in one of these two ways to all three
models. These figures are also significant ( p < . o o i ) .
As soon as a model had a thin edge or was hollow, therefore, it
was less likely to be rolled in: this was independent of its weight.
The interrupted outline did not affect the scores, though, as we have
seen, the latency of the response is larger for "Filled shell" than for
"Empty egg".
Ten birds showed rolling or intention-rolling to both " E m p t y
egg" and "Filled shell", but not to the two other models; no birds
did the opposite; seven did not treat these two categories differently,
and three behaved similarly to part of the two categories. These
differences, too, are significant ( p = . o o i ) .
Billing gave very similar results. Eleven birds billed "Empty
egg" but not "Shell-with-lead" nor "Real shell", four did the opposite,
and ten treated the models similarly (p^.059). The number of birds
which showed billing to "Filled shell" but to neither "Shell-with-lead"
nor "Real shell" was 13; four did the opposite, and eight did not
show this difference ( p = . o 2 j ) . Finally, the number pf birds which
showed billing to either "Filled shell" or " E m p t y egg" but to neither
"Shell-with-lead" nor "Real shell" was 2 1 ; four did the opposite, and
three did not distinguish clearly either way ( p < . o o i ) .
Table 3 summarises further information that can be extracted about
the difference between "Real shell" and "Shell-with-lead". The first
*A11 p values in the remainder of this paper were computed according to the sign
test.
125
B R I T I S H BIRDS
T A B L E $•—COMPARISON
OF RESPONSES OF B L A C K - H E A D E D G U L L S
(Larus ridibundus) T o " R E A L S H E L L " A N D " S H E L L - W I T H - L E A D " ,
C U M B E R L A N D , 1961
This shows the numbers of birds which responded with each of the reactions listed
in the first column to "Real shell" but not to "Shell-with-lead" ( + ) , which did the
opposite (—), or which did not differentiate (o) (p values based on sign test)
Carrying
Lifting*
Rolling
Nibbling
Poking
+
—
0
7
0
0
2
1
0
5
0
5
2
15
10
3
4
2
Significance
p=
p=
,008
.031
not significant
p=<
p=
.001
.001
*This category includes intention-carrying
line allows us t o say that "Real shell" elicited much more carrying,
and the second shows that "Shell-with-lead" was more often merely
lifted and then dropped. The fourth line shows that "Shell-withlead" very often gave rise to mere nibbling, while the fifth says the
same about poking.
These data demonstrate that neither model elicited rolling, but
that both elicited some or all of the removal sequence. This obviously
consists of two parts, nibbling and carrying, and it is also obvious that
the "Shell-with-lead", while eliciting nibbling, stopped the real
carrying which normally follows nibbling. In other words, the
presence of the lead inhibited the action chain half-way. That this
was due to the weight of the lead will become clear in the next experiment.
'Experiment C
This experiment was designed to test four hypotheses which we had
developed as the results of Experiment B took shape.
First, we had gained the impression that the "Real shell" elicited
removal mainly through having a thin edge, rather than through being
hollow. We therefore offered an empty egg with a flange of shell
glued on it at right angles to the surface. The flange measured
2 cm. x 1.5 cm. and could be considered flat. This model, which we
called "Flanged egg" (plate 21I), was presented with the flange turned
up and the longitudinal axis pointing radially. It thus offered a thin
edge, but no hollow.
Second, to see whether the fact of a shell's being hollow had some
effect, and also to have a control model in connection with "Shellwith-lead", we filled a real shell with cottonwool so that it had very
126
GULLS D I S T I N G U I S H I N G
EGGS AND
EGG-SHELLS
much the same appearance as the plaster-filled egg or "Filled shell".
However, whereas the plaster neatly fitted the edge, the cottonwool
(though concealing the hollowness) allowed the bird to see the thin
edge as separate from it. This model we named "Cottonwool shell"
(plate 2if).
Third, to see whether a serrated edge was more stimulating than a
smooth one, we introduced two model eggs with similar sized holes
40 mm. from the blunt end. One had a smooth edge and the other
was artificially serrated. These we called "Smooth r i m " (plate 21J)
arid "Notched r i m " respectively (the latter is not illustrated here, but
the notches were of the same size as the white patches on "Painted
rim", plate 21k).
Fourth, we thought that a real egg-shell might owe part of its effectiveness to the amount of white it showed. A smooth-rim model
(with a hole 40 mm. from the blunt end) was therefore given a series
of white triangles along the outside of the rim, so as to fake a serrated
edge to the khaki exterior. This was known as "Painted r i m " (plate
2lk).
With the "Real shell" as a control, there were thus six models in this
experiment. Each of these was presented 18, 19 or 20 times, the slight
difference being due, as in the previous tests, to some of the series
being prematurely ended by nest-relief. The results are given in
Table 4.
Although the scores for rolling and intention-rolling in Table 4
are higher for "Flanged e g g " and "Cottonwool shell" than for "Real
12
l
i
1
14
15
2
9
9
8
12
12
12
0
1
!9
'3
0
1
20
11
5
6
4
5
2
20
2
1
18
1
0
J
5
0
21
* These categories include intention-rolling and intention-carrying
127
Total
presentations
Billing
4
5
6
5
7
Ignoring
6
6
Poking
Flanged egg
Cottonwool shell
Real shell
Smooth rim
Notched rim
(k) Painted rim
Nibbling
(1)
(f)
(h)
(j)
Carrying*
Rolling*
TABLE 4—RESULTS OF EXPERIMENT C WITH BLACK-HEADED GULLS
(Larus ridibundus), C U M B E R L A N D , 1961
This shows the numbers of birds which responded to the various model eggs with
each of the reactions listed. The first two columns give the ultimate responses,
irrespective of the incomplete movements which may have preceded them. The
names of the model eggs and the small letters with them correspond to those on
plate 21
9
respectively
B R I T I S H BIRDS
shell", the differences are not significant. "Flanged egg" and "Cottonwool shell" have not been compared in one series with "Empty egg"
and "Filled shell" (Table 2), but neither of the latter models were
carried even once; every single bird which responded to them did so
by rolling. We must conclude, therefore, that "Empty egg" and
"Filled shell" were treated like eggs (i.e. retrieved) and that "Flanged
egg" and "Cottonwool shell" were treated like a real shell (i.e. removed), although we must leave open the question whether the birds
distinguish to a slight extent between the different models within these
groups. The main character to which the gulls respond by carrying
must be one that is found in both "Flanged egg" and "Cottonwool
shell" but absent from the other models. It follows that the thin rim
is the important character and that neither hollowness nor broken
outline, not their combination (as found in "Filled shell"), are sufficient
to make the bird carry, although, as we have seen, "Filled shell"
elicited rolling with a greater delay than an egg does.
The high carrying score for "Cottonwool shell" also shows that the
low carrying response to "Shell-.with-lead" (Table 2) was due to its
weight and not to the sight of something inside the shell.
Turning now to the characteristics of the shell's rim (the figures in
the three lower rows of Table 4), we find that these scores are not
significant either. However, because the differences between these
models are relatively simple and straightforward, it is worth extracting
fuller information from our protocols and counting all responses shown
in all tests instead of merely listing (as in Table 4) whether or not a
response occurred in any one test. Many birds responded more than
once in a test; moreover, many birds showed in one and the same
test elements of both rolling and carrying. By counting all these
responses in every case, we can determine for each test an index which
gives the ratio between the number of egg-rolling movements and the
number of carrying responses—a "retrieving-over-carrying index".
By comparing these indices for the various models, we find that in
ten cases the index was higher for "Smooth rim" than for "Notched
rim", in six the indices were equal for these two models, and one bird
had a higher index for "Notched rim" than for "Smooth rim" (p=
.006). Thus "Notched rim" elicited relatively more carrying than
"Smooth rim", which in turn elicited relatively more rolling.
This same index was higher for "Smooth rim" than for "Painted
rim" in eight cases, it was equal for both models in eight more, and
one bird showed the reversed response. These differences are also
significant (p—.oi).
We can therefore conclude that both serrating the edge, as in
"Notched rim", and adding the white pattern of "Painted shell"
increased carrying and reduced retrieving.
128
GULLS D I S T I N G U I S H I N G EGGS AND E G G - S H E L L S
SUMMARY AND C O N C L U S I O N S
Experiments were carried out at Ravenglass, Cumberland, in the
summer of 1961, to determine h o w Black-headed Gulls (Lams ridibundus) distinguish between eggs and egg-shells. These elicit entirely
different sets of responses: eggs are rolled into the nest and brooded,
while shells are picked u p in the bill and removed. The following
paragraphs summarise the conclusions reached.
The egg-shell elicits removal because it differs from the intact egg
in the following characteristics: it shows a thin edge; this edge is
serrated; and it shows white. The "decision" to remove it is already
taken before the bird can have checked its weight. Neither the interruption of its outline nor its hollowness could be shown to contribute
to removal, though both seem to exert an inhibiting influence on rollingIn addition, it was shown that egg-shell removal is a chain of acts:
nibbling is elicited by visual stimuli; during nibbling the weight is
checked and, if it 'does not grossly exceed that of a real shell, the object
is carried; if it equals the weight of an egg or a chick, the chain is
broken off. This prevents chicks from being carried away when they
have hatched but not yet left the shell; whether other safeguards
exist as well we cannot say.
Most of the population in which these responses were studied must
have had previous breeding experience. Very little can be said,
therefore, about the extent to which the responses are innate or could
be the consequence of conditioning. Three one-year-old birds,
which were tested with eggs and shells when they had not yet laid at
all, rolled in an egg and removed a shell in a way indistinguishable
from experienced birds (Tinbergen, Kruuk and Paillette 1962).
ACKNOWLEDGEMENTS
Our thanks are due to Sir William Pennington Ramsden, Bt, and the
Cumberland County Council for permission t o work in the Ravenglass
gullery; and to the Nuffield Foundation and the Nature Conservancy
for support.
M. Paillette wishes to express her gratitude for financial
assistance received from the British Council and the Zoological Society
of L o n d o n ; R. A. Stamm was helped by a grant from the JanggenPohn Stiftung.
REFERENCES
SIEGEL, S. (1956): Non-parametric Statistics for the Behavioral Sciences. New York.
TINBERGEN, N., BROEKHUYSEN, G. J., FEEKES, F., HOUGHTON, J. C. W., KRUUK, H„
and SZULC, E. (1962a): "Egg shell removal by the Black-headed Gull, harus
ridibundus L.: a behaviour component of camouflage". Behaviour (in press).
TINBERGEN, N., KRUUK, H., and PAILLETTE, M. (1962b): "Egg shell removal by
the Black-headed Gull, Laris ridibundus L. II. The effects of experience on the
response to colour". Bird Study (in press).
129
P L A T E 20A. Black-
headed Gull {Larus
ridibundus)
lifting
egg-shell to remove
it. It holds the
thin edge in its bill
and either takes it
away
immediately
or nibbles it for a
while first. It has
been shown that such
shells near the nest
increase the risk of
predation (page 120)
P L A T E 20B.
Start-
ing to roll an egg
into the nest. This
is an outsize model
that has no connection with the experiments described on
pages 120-129, but
the action is typical.
Bending over the
egg, it rolls it back
with the under-side
of its bill (page 121)
{photos: N. Tinbergen)
(a) Empty egg
(e) Filled shell
(i) Real shell
(b ", Small hole
(f) Cottonwool shell
(j) Smooth rim
(c ) Medium hole
(g) Shell-with-lead
(k) Painted rim
(d) Large hole
(h) Real shell
(1) Flanged egg
P L A T E 2 1 . Model eggs and egg-shells used in experiments with Black-headed
Gulls (Larus ridibundus), Cumberland, 1961 (pages 120-129) (photos: John Haywood)
("Notched rim" is not shown)
© Copyright 2025 Paperzz