INTRODUCTION
Numerous seabird studies have dealt with the factors affecting breed-
ing success (e.g., Coulson, 1900; Mills, 1973; Davis, 1970; Brooke, 1978;
Ollason and Dunnet, 1978). Complementary to these studies is the in-
vestigation of causes and timing of breeding failures. Unfortunately it
is impossible in practice to examine natural losses of eggs or chicks in
isolation from unnatural ones because the observer's presence is likely
to increase the losses, i.e., cause disturbance. However, the stage in the
breeding cycle (egg or chick) at which losses are more likely to occur
and their timing in relation to date can be determined.
In the Fulmar (Fulmarus glacialis) Dunnet et al. (1903) suggest that
most failures occur at the egg stage: of all egg losses in the first 9 days
after laying, 71% occur within the first 3 days. Mougin (1907) shows two
peaks in egg losses, one shortly after laying, the other around hatching.
Ollason and Dunnet (1978) demonstrate that late-laid eggs are more
likely to fail.
Breeding success in the Fulmar can vary widely from year to year,
e.g., on the island of Eynhallow in Orkney, Dunnet et al. (1979) quote
a range of 16% to 52% over 28 years, although no known change oc-
curred in conditions on the island. This does not account for variations
between years in observer effort, which will now be considered in detail.
The aim of this paper is to determine for the Fulmar the most vul-
nerable stage of the breeding cycle, to describe the pattern of egg losses
in relation to date, and to examine the effects of observers studying
breeding biology on the response of breeding adults and their subse-
quent breeding success. Causes of failure for a generalized seabird will
be described theoretically, and the Fulmar fitted into this framework.
METHODS
In 1950, R. Carrick and G.M.D. began color-banding breeding Ful-
mars caught from their nests on the small uninhabited island of Eyn-
hallow, Orkney. Observations during the breeding season have contin-
ued annually to the present time. General methods have already been
described (Dunnet and Ollason, 1978). The initial aims of the study
were to collect data on breeding and survival. Only recently was it re-
alized that we have useful data on the effects of disturbance.
The data analyzed here were collected over 21 years from 1958 to
1978. Three annual vists were taken to Eynhallow: in May (to count
eggs), in July (to determine hatching success), and in August (to deter-
mine fledging success).
70--
60--
E 30--
z
20--
10--
14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 1
May
I
Date June
FIGURE 1. Average number of eggs laid in 1960, 1961, 1962, and 1978.
Visit in May
In 1960, 1961, 1962, and 1978 the May visit lasted for about three
weeks during laying. Nests were checked as often as possible (usually
daily) to obtain laying dates and the number of eggs laid (apart from
any eggs laid and lost between consecutive checks of nests). The mean
laying dates in these years ranged from 21.6 May in 1961 to 23.5 May
in 1962 (Variance Ratio F = 7.61, df = 3.498, P < 0.001). However, it
was felt that combining the data from all four years should produce a
distribution of laying dates typical of an average year (Fig. 1). In the
remaining 17 years (1958, 1959, and 1963 to 1977), Eynhallow was
visited during the fourth quarter of the laying distribution for 2 to 6
days beginning between 26 May and 8 June. The number of eggs ob-
served in each of these years is therefore less than the total laid (because
some egg losses would have occurred before the first annual visit); it
consists only of the number of eggs present on the first day of the visit,
plus any laid subsequently. A correction therefore needs to be
applied to the observed totals of these 17 years to allow for egg
losses before the first day of the May visit. Using the data from 1960,
1961, 1962, and 1978, the pattern of eggs laid and eggs lost during
laying has been determined for an average year: Figure 2 shows (a) the
number of eggs to be laid after a particular day, expressed as a pro-
1'01
0'8--
0'4--
0'2
0
A
0000
0
0
0
ß
o
B
0000 0
0000
May
14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 1
Date June
FGURE 2. Pattern of egg-laying and egg-loss during the laying period of an average
year: (A) eggs laid after day x as a proportion of eggs present on day x plus eggs
laid after day x; (B) eggs lost by day x as a proportion of eggs present on day x.
Data from 1960, 1961, 1962, and 1978.
portion of those present on that day plus those laid subsequently, and
(b) the number of eggs lost by a particular day, expressed as a propor-
tion of those present on that day. The totals were adjusted as follows:
the observed total was multiplied by the proportion (a) referring to the
first day of the May visit to give an estimate of the number of eggs laid
after the first day of the visit. Subtracting this number from the observed
total gives an estimate of the number of eggs present on the first day,
which is then multiplied by the relevant proportion (b) referring to the
first day of the May visit to give the estimated number of eggs lost by
that day. Adding the three estimates gives an adjusted total for each of
the 17 years which can then be compared with each other and with the
observed totals of 1960, 1961, 1962, and 1978. No significant difference
was found between the distributions of the observed and adjusted totals
(Kolmogorov-Smirnov test X22 = 5.29, NS) and the overall breeding suc-
cess (percentage eggs that fledge) calculated from the observed totals is
highly correlated (% = 0.961, df = 17, P < 0.001) with that calculated
from the adjusted totals. The observed number of eggs present on the
first day is not significantly different from the estimated number of eggs
present on the first day (: = 144.9 and 148.8; SD = 36.28 and 39.39,
n = 16) and they are highly correlated (r = 0.988, P < 0.001). There-
fore in all subsequent analyses in this paper, the adjusted total number
of eggs has been used for 1958, 1959, and 1963 to 1977.
Visit in July
In all years of the study, a second visit to Eynhallow of three or four
days occurred in July, covering part of the hatching period. By the end
of' this visit most eggs had hatched. However, some eggs remained, a few
of which may have hatched but failed before fledging. Hatching success
will therefore tend to be slightly underestimated.
Visit in August
The final annual visit, of one day, occurred during August, by which
time almost all adults had left the area. All nests with fledglings were
recorded, and therefore a good estimate of overall breeding success
could be determined. Since the estimate of hatching success will be less
accurate than that of overall breeding success, all analyses involving
hatching success have also been carried out on overall breeding success.
In 1978, the response of Fulmars on an egg or a chick to the stan-
dardized approach of one particular observer (J.C.O.) was categorized
as follows: "on"--bird remained on the egg or chick all the time while
the observer was visible; "at"--bird left its egg or chick but did not fly
away and usually stayed within a few inches of its nest site; and "off"--
bird left its egg or chick and flew away. For each nest observed more
than five times, an index of remaining "on" the nest was determined by
observing the number of times that the parent(s) remained "on" the
nest and expressing this as a proportion of the total observations of
parents at that nest, while it had an egg or a chick. Nests with fewer
than six observations (55% of the total 215 nests) were eliminated to
avoid the biases of small samples. This procedure will also have removed
nests with birds very sensitive to disturbance, i.e., nests that failed after
five or fewer disturbances. Thus the index of remaining "on" the nest
cannot include the most sensitive parents.
The data were coded and stored on magnetic disc in a data base (using
the Aberdeen University Data Base Management System). Most analyses
were carried out using a Honeywell Level 66 computer.
RESULTS
The Laying Period: Eggs Laid and Eggs Lost
The combined laying distributions for 1960, 1961, 1962, and 1978
are assumed to be typical of an average year ( -- 22.7 May, SD = 3.40,
T,BI, F 1.
Mean percentage failures before hatching for various sets of years.
Mean SD n d P
(a) Years with long May visit = 66.90 3.289 4
1960 to 1962 and 1978
(b) Years with short May visit = 65.78 8.624 17
1958 and 1959, 1963 to 1977
(c) Years with large group in 71.79 7.007 8
July = 1971 to 1978
(d) Years with small group in 62.42 6.115 13
July = 1958 to 1970
(e) Years with large group in 71.79 7.568 7
July excluding years with
long May visit = 1971 to
1977
(f) Years with small group in 61.57 6.804 10
July excluding years with
long May visit = 1958 and
1959, 1963 to 1970
(g) Years with long May visit and 65.27 0.473 3
small group in July = 1960 to
1962
NS
3.119 <0.01
2.854 <0.02
NS
n = 502; Fig. 1). The proportion of eggs laid each day that fail to reach
hatching increased significantly with laying date (rs = 0.759, df = 17,
P < 0.001). No similar change occurred with laying date in the propor-
tion of chicks hatching that eventually fledged.
During the laying periods of 1960, 1961, 1962, and 1978, the day on
which each egg was lost was known. The proportion of eggs lost each
day was not constant (Kolmogorov-Smirnov test X., 2 = 13.52, P 0.01)
but varied in an inconsistent way with date.
Disturbance
The presence of observers on Eynhallow likely caused disturbance to
the breeding Fulmars, possibly resulting in lower hatching success and
overall breeding success. In the years 1960, 1961, 1962, and 1978, the
first annual visit covered about three weeks during laying, whereas in
the remaining years from 1958 onwards, the May visit covered only 2
to 6 days. In all years, 2 to 6 people were present at this time. Since the
egg stage especially is vulnerable to predation, the years with a long May
visit might be expected to have a higher failure rate. This was the case,
but the mean failure rates for the two sets of years were not significantly
different at hatching (Table 1, a, b) or overall (Table 2, a, b).
T^BLE 2.
Mean percentage overall breeding failures for various sets of years.
Mean SD n d P
(a) Years with long May visit = 73.98 4.568 4
1960 to 1962 and 1978
(b) Years with short May visit = 72.54 8.038 17
1958 and 1959, 1963 to 1977
(c) Years with large groups in 77.69 7.366 8
July = 1971 to 1978
(d) Years with small groups in 69.82 5.890 13
July = 1958 to 1970
(e) Years with large groups in 77.27 7.854 7
July, excluding years with
long May visit = 1971 to
1977
(f) Years with small groups in 69.23 6.645 10
July, excluding years with
long May visit -- 1958 and
1959, 1963 to 1970
(g) Years with long May visit and 71.77 1.429 3
small group in July -- 1960 to
1962
NS
2.560 (0.05
2.211 (0.05
NS
In the years 1971 to 1978, groups of from 8 to 12 students plus 3 or
4 staff visited Eynhallow for 3 or 4 days during July, at the beginning
of the hatching period. In the earlier years of the study, the July visit
was similar in date and duration, but the group consisted only of 3 or
4 people. Comparsion of these two sets of years shows that the hatching
success and the overall breeding success were significantly worse in the
years when larger groups visited the island, regardless of whether the
years with long May visits are included (Table 1 and 2, c, d), or not
(Tables 1 and 2, e, f). Since the larger groups apparently had a signif-
icant effect, this effect might be masking the effect of the long May visit
(Tables 1 and 2, a, b). The'efore using only years with no large group
in July, years with a long May visit were compared with years with a
short May visit and found to have lower success (Tables 1 and 2, g, f),
but differences were still not significant.
From the data it is possible to determine at what stage failures oc-
curred and approximately at what date (Table 3). Egg failures occurring
before the July visit, or in and after the July visit have been compared
for (i) the short-May-visit-small-July-group years (1958-1970, excl.
1960-1962) and the short-May-visit-large-July-group years (1971-
TABLE 3.
Timing (in relation to date) of egg and chick failures.
(a) Failures at egg stage
May visit (i) Short (ii) Long
July group size Small Large Small Large
Years 1958-1970 1971-1977 1960-1962 1978
(excl.
1960-1962)
Before July visit 828 724 172 68
In and after July visit 121 236 42 82
x 2 = 43.18 x 2 -- 46.66
P < 0.001 P < 0.001
(b) Failures at chick stage
May visit (i) Short (ii) Long
July group size Small Large Small Large
Years 1958-1970 1971-1977 1960-1962 1978
(excl.
1960-1962)
In July visit 46 23 5 6
After July visit 62 55 16 13
,)(12 = 2.80 X? = 0.04
NS NS
1977) (Table 3a [i]) and (ii) the long-May-visit-small-July-group years
(1960-1962) and the long-May-visit-large-July-group year (1978) (Ta-
ble 3a [ii]). Both tables show significant differences between the sets of
years, there being a much larger proportion of failures in and after July
in the years with a large group present in July. Using the same sets of
years, comparisons of failures at the chick stage in July or after July did
not show significant differences associated with the large group present
in July (Table 3b, [i] and [ii]).
In this long-term study, adults need to be recaptured at regular in-
tervals in order to ensure that color and metal bands remain legible.
Most captures were made in July, and usually birds were caught off
chicks. Perhaps the larger group of people present in July allowed more
birds to be caught and thereby decreased breeding success by increasing
disturbance. However, captures, expressed as percentage captures of the
number of breeding adults for each year, was not sigmficantly correlated
with overall breeding success (rs = -0.058, df = 19) or with hatching
success (rs = -0.111, df = 19). However significant negative correla-
tions were found between hatching success and man-days in July for
each year (rs = -0.630, df = 19, P < 0.01; Fig. 3) and between overall
breeding success and man-days in July (rs -0.538, df = 19, P < 0.02;
50--
170
oe 40 1 600
õ 68
-- 6
o
30--
20-
10--
o I
o
69
63 71
i o
65
58
64 77
72
I 0 0
76
oo _o
75 78
74
20 40 60 80 100
Man-days on Eynhallow in July
F[ct;] 3. Prcentage hatching success plotted against man-days spent on Eynhallow
during July for 1958 to 1978. Closed circles: years with small July group. Open circles:
years with large July group. Underlined symbols: years with long May visit. Captures
(as percentage of breeding adults) were most numerous in 1960, 1959, 1962, and
1958, respectively.
Fig. 4), but neither hatching success nor overall success correlated sig-
nificantly with man-days in May (rs = 0.146, df= 19 and rs = 0.117,
df-- 19, respectively). If the years are arranged into those with a large
group present in July (1971-1978) and those with a small group present
in July (1958-1971), overall breeding success correlates with man-days
in July when the group was large (rs = -0.838, df = 6, P < 0.01) but
not when the group was small (rs -- -0.149, df = 11).
Response of Breeding Adults to Approach of an Observer
The response of a Fulmar on an egg or a chick to the approach of an
observer varied considerably. Some birds allowed an observer to see its
color and metal bands and to remove an egg or chick for weighing
without becoming disturbed or leaving the nest site. At the other ex-
treme, birds left their nest site as soon as the observer appeared, and
did not return while the observer was able to see the site.
40--1 67 ß
__ ß 69
_68ß ß7o
ß60
-- 64 75
-- 62 0
77 0 0
76
20- 5e 0
78
59
-
o-
o I I I I I I I
0 20 40 60
Man-days on Eynhallow in July
63 71
ß o
65
72
74
I I I
80 1 O0
FIGURE 4. Percentage overall breeding success plotted against man-days spent on Eyn-
hallow in July for 1958 to 1978. Closed circles: years with small July group. Open
circles: years with large July groups. Underlined symbols: years with long May visit.
Captures (as percentage of breeding adults) were most numerous in 1960, 1959,
1962, and 1958, respectively.
Nests that failed during the egg stage had a significantly lower mean
proportion of birds remaining "on" (see methods) than nests in which
the egg hatched or the chick fledged (Table 4). The variation in pro-
portion "on" was significantly greater for nests that failed at the egg
stage (Table 4).
Some evidence indicates that the proportion "on" increases as the
breeding experience of the bird increases; e.g., for nests in which the
length of breeding experience of the male was known, those in which
the male's experience was 1 to 3 years had a significantly lower mean
proportion of parents "on" than those in which the male's experience
was over 3 years (Table 5a). A similar, but nonsignificant trend is ap-
parent in relation to the female's experience (Table 5b). If breeding
success is held constant by partial correlation, the proportion of parents
"on" still correlates significantly with length of experience of the male
(r = 0.242, df = 76, P < 0.033).
DISCUSSION
Possible Causes of Failure for a Generalized Seabird
Figure 5 is a schematic representation of a bird breeding season, from
egg production to fledging, showing various factors that may increase
TABLE 4.
Mean proportion of occasions when parent remains "on" the nest at the approach of an
observer, for nests successful to (i) the egg stage, (ii) the chick stage, or (iii) fledging.
Stage which Mean
nest reaches proportion "on" SD n
(i) egg only 0.6375 0.311 40
(ii) chick only 0.7813 0.177 18
(iii) fledgling 0.7612 0.213 36
Comparisons: F ratio P t or d P
(i) with (ii) 3.09 <0.015 2.23 <0.030
(ii) with (iii) 1.45 NS 0.35 NS
(i) with (iii) 2.13 <0.025 2.04 <0.046
the probability of breeding failure. The factors can be classified as those
due to either or both birds and those due to external causes. Their
relative importance will vary depending on the species and the parental
contributions of the male and female.
The past history of both parents (age, breeding experience, and pair-
bond) is known to affect breeding success (e.g., in Kittiwake, Coulson,
1966; Red-billed Gull, Mills, 1973; Arctic Skua, Davis, 1976; Manx
Shearwater, Brooke, 1978; Fulmar, Ollason and Dunnet, 1978). These
factors are presumably effective in terms of the physiology and behavior
of the parents during the breeding season. Thus physiological condition
(Fig. 5a) of both birds is probably important during egg production:
poor quality eggs are less likely to result in fledged chicks; e.g., a small
egg produces a small chick (fowl, Wiley, 1950; Brown-headed Cowbird,
Nolan and Thompson, 1978; Laughing Gull, Ricklefs et al., 1978), with
a lower survival rate (Herring Gull, Parsons, 1970); or the egg may be
TABLE 5.
Mean proportion of occasions that parent remains "on" the nest at the approach of an
observer, for parents with 1 to 3 years breeding experience, or over 3 years breeding
experience.
Length of
breeding Mean
experience, proportion
years "on"
SD n t P
(a) Male
1 to 3 0.6368
over 3 0.7769
(b) Female
1 to 3 0.6570
over 3 0.7335
0.262 13
2.08 <0.042
0.209 56
0.264 19
0.213 47 1.23 NS
infertile; or it may have constituents in the wrong proportions. Possibly
undue stress at this time might result in resorption of the egg, but this
is unlikely to be detected. If physiological condition is not good at the
beginning of breeding, there may be effects later in the season; e.g.,
nutritional state of the female may affect overall breeding success (Red
Grouse, Moss et al., 1975).
If an egg is laid (Fig. 5b), both birds need to be efficient incubators
(Fig. 5c; Baerends, 1959; Drent, 1973) and communicate with each oth-
er to maintain a shared incubation schedule. Poor incubation behavior
may result in eggs being left uncovered: although eggs may have some
chilling resistance (Manx Shearwater, Mathews, 1954) and effects of
weather are not always predictable (Adlie Penguin, Ainley and Le-
Resche, 1973), the probability of killing eggs will be increased by ex-
posure to excessive heat (Herring Gull, Hunt, 1972), cold (Wilson's
Storm Petrel, Beck and Brown, 1972) or wind (Common Tern, Goch-
feld, 1978). Uncovered eggs are also vulnerable to predation (Lesser
Black-backed Gull, Davis and Dunn, 1976). Similarly any disruptive ex-
ternal factors that cause the incubating bird to leave the nest will expose
the eggs to weather and predators (gulls, Harris, 1964; Herring Gull,
Hunt, 1972; Double-crested Cormorant, Kury and Gochfeld, 1975;
Western Gull, Robert and Ralph, 1975; Double-crested Cormorant, El-
lison and Cleary, 1978).
If the egg hatches (Fig. 5d) the parents need sufficient foraging skill
to obtain food for themselves and their young (Fig. 5e; Adlie Penguin,
Ainley and Schlatter, 1972; Brown Pelican, Orians, 1969 and Blus and
Keahey, 1978) and they must be able to feed them efficiently. The chicks
also need protection from predators and weather, and disturbance of
the parents may reduce or remove this protection (Adlie Penguin,
Reid, 1968; Glaucous-winged Gull, Gillet et al., 1975). However, in some
species adults brooding chicks have been shown to be more tolerant of
disturbance than those with eggs (Double-crested Cormorant, Kury and
Gochfeld, 1975), or chicks have become less frightened after repeated
disturbance (Western Gull, Robert and Ralph, 1975). The vulnerability
of the egg and chick stages therefore varies with the species.
Thus the probability of breeding failure is increased if the parents
are in poor condition or if their behavior is inadequate. External dis-
turbance, whether intentional or not, exaggerates the losses by altering
the behavior of the parents. Some species may be less susceptible, since
their success is improved when they nest under cover (gulls, Brown,
1967; Lesser Black-backed Gull, Davis and Dunn, 1976; Common Tern,
Nisbet, 1975). In others it is shown that adults nesting on the periphery
of the colony fled more easily at the approach of an observer than those
nesting in the center (Adlie Penguin, Tenaza, 1971). Also disturbance
in one year can have an effect on breeding in the following year (Adlie
Penguin, Oelke, 1975 and Reid 1968; Fulmar, Ollason and Dunnet,
1978).
(a)
EGG
PRODUCTION
AND
FERTILISATION
I Male in Female in Male in [Female in
poor [ poor ] I good I I gøød
cond,t,on I conit,on I cond,tionl I condit,on
Egg of poor quality Egg of good quality
(b)
LAYING
(c)
INCUBATION
EGG LAID
incubator incubator incubator incubator
Egg poorly incubated Egg well incubated
(d)
HATCHING
(e)
CHICK
BROODING
AND
FEEDING
EGG HATCHES
Male poør I I Female Pøør I I Male gøød I I Female gøød I
feeder feeder feeder feeder
Chick develops slowly Chick develops normally
(f)
FLEDG ING CHICK FLEDGES
FIGURE 5. Schematic diagram of a generalized seabird breeding season from egg pro-
duction to fiedging. Disturbance refers to any factor external to the parents which
alters the behavior of the parents. In general, factors which cause the flow to be down
the left side of the diagram will increase the probability of breeding failure.
Causes of Failure in the Fulmar
In the Fulmar, both parents go on a prelaying exodus (Dunnet et al.,
1963), which is typical of Procellariiformes (Warham, 1964): on the av-
erage the female is away for 20 days; the male is away for 9 days and
returns before the female (Macdonald, 1977). Presumably they go to
distant feeding grounds to build up reserves for the breeding season.
The nutritional state and physiological condition of both birds would
therefore seem to be critical (Fig. 5a). The single egg weighs about 13%
of the female and relaying has never been known to occur. Some un-
published evidence suggests that, within the history of a female, larger
eggs are more likely to be successful than smaller ones; i.e., larger eggs
are of a better quality and presumably use more of the female's reserves.
Possibly disturbance during egg production might cause resorption, but
this would be difficult to demonstrate, and was unlikely in this study
since observers were not normally present at this time and the female
is away from the colony.
After the egg is laid, the male normally relieves the female at the nest
within a few hours, and then incubates for about 10 days. Thereafter
the incubation stints are approximately equal (Dunnet et al., 1963).
Therefore, for incubation to be successful the male and female must be
in the right place at the right time and communicate successfully with
each other (Fig. 5c). The egg is white (i.e., not camouflaged) and no
nest material is present, so that if the egg is left uncovered it is vulner-
able to predators as well as to extremes of weather. Late-laid eggs seem
more at risk than early-laid ones, since fewer of the late-laid eggs
reached hatching. This is consistent with the fact that more experienced
Fulmars tend to lay earlier and are more successful (Ollason and Dun-
net, 1978). The data presented here show some lowering of hatching
success (and overall success) in the four years in which observers were
present during the whole of laying (Tables 1 and 2; a and b, g and f)
although the differences were not significant, probably because the
number of observers was small. A significant increase in breeding failure
was found in years in which a group of at least 12 people was present
for 3 or 4 days during hatching (Tables 1 and 2; c and d, e and f), i.e.,
at a time corresponding to the second peak of egg losses of Mougin
(1967). Thus disturbance (measured in terms of man-days) during in-
cubation (Fig. 5c) does have a significant effect upon hatching success
(and overall success).
After the egg hatches, both parents feed and brood the chick. It is
not known what food is obtained, nor at what distance the parents for-
age. Disturbance at this time (Fig. 5e) leaves the chick exposed, although
Fulmar chicks can protect themselves to some extent by ejecting noxious
oil (Armstrong, 1951). In this study, most captures of adult birds oc-
curred when the chick was present--capture of breeding Fulmars can
affect their subsequent breeding success (Ollason and Dunnet, 1978).
However, no relationship was found here between overall breeding suc-
cess of the population as a whole each year and captures in that year.
This is probably because capture only affects the success of the nest at
which an individual was captured, and captures normally only occurred
at about 10% of nests in any one year. But it is interesting to note that
in two years (1958 and 1959, Figs. 3, 4) in which success was low and
man-days were low, disturbance due to captures was relatively high.
Thus disturbance at the chick stage can have some effect, but at least in
this study, the chick is not as vulnerable as the egg.
Response to the Approach of an Observer
How does an observer cause disturbance? Significant differences be-
tween birds are shown (Table 4) in the response to a particular observer
at nests that subsequently failed (64% remained on site) and those that
were subsequently successful (76% remained on site). The differences
are similar to those described for the Ad61ie Penguin by Tenaza (1971),
and the behavior of birds at unsuccessful nests is more erratic. Thus
putting a bird off its nest an additional once or twice out of 10 occasions
could make the difference between a successful and an unsuccessful
nest.
The evidence indicates that the behavior of birds to the approach of
an observer tended towards staying on the nest as breeding experience
increased (Table 5), and this may be one of the factors that results in
improved success of more experienced birds (Ollason and Dunnet,
1978), rather than being a consequence of it.
Effects of disturbance probably vary with species (e.g., Glaucous-
winged Gull, Gillet et al., 1975; Double-crested Cormorant, Kury and
Gochfeld, 1975), as well as with habitat (gulls, Brown, 1967; Lesser
Black-backed Gull, Davis and Dunn, 1976; Common Tern, Nisbet, 1976)
and frequency of disturbance (Western Gull, Robert and Ralph, 1975).
Perhaps factors such as these should be considered when designing field
projects.
SUMMARY
Data were collected over 21 years as part of a long-term study of breeding Fulmars on the uninhabited island of Eynhallow in Orkney. The pattern of egg losses is described. Eggs laid early in the season were more likely to hatch than those laid late.
The relationship between number of observers each year and subsequent breeding success that year was examined. In four years in which 2 to 6 observers were present for 2 to 3 weeks during laying (May), hatching success and overall success were lowered, but not significantly so. No relationship existed between man-days present in May and breeding success. In eight years in which at least 12 observers were present for 3 to 4 days during hatching (July), hatching success and breeding success were significantly lowered. A significant negative correlation was found between man-days present in July and breeding success.
Capture of breeding adults had no significant effect upon subsequent breeding success of the population as a whole, probably because captures only occurred at about 10% of nests. The responses of sitting birds to the approach of an observer varied: at nests which were subsequently
successful, 76% of birds remained on site: the figure for unsuccessful nests was 64% and their behavior was more erratic.
The causes of failure for a generalized seabird are described, and the Fulmar is fitted into this framework.
ACKNOWLEDGMENTS
We are grateful to the late Miss J. M. Robertson, Kirkwall, for allowing
us to observe Fulmars and to use her house on Eynhallow, Orkney. We
thank Mr. A. Anderson, Aberdeen University, for his long-time assis-
tance, and the numerous others who have helped at some time over the
years. The useful criticisms of Dr. I. J. Patterson, Aberdeen University,
were much appreciated. Aberdeen University Computer Centre provid-
ed facilities for the data analysis.
LITERATURE CITED
AINLEY, D. G., AND R. E. LEREscIE. 1973. The effects of weather and ice conditions on
breeding in the Adlie Penguin. Condor, 75: 235-238.
AINLEY, D. G., AND R. P. SCHLATTER. 1972. Chick raising ability in Adlie Penguins. Auk,
89: 559-566.
ARMSTRONO, E. 1951. Discharge of oily fluid by young Fulmars. Ibis, 93: 245-251.
BAERENDS, G. P. 1959. The ethological analysis of incubation behaviour. Ibis, 101: 357-
368.
BECK, J. R., AND D. W. BROWN. 1972. The biology of Wilson's Storm Petrel Oceanites
oceanicus (Kuhl) at Signy Island, South Orkney Islands. Br. Antarct. Surv. Sci. Rep., 69:
1-54.
BLUS, L. J., AND J. A. KEAHEY. 1978. Variations in reproductivity with age in the Brown
Pelican. Auk, 95: 128-134.
BROOKE, M. DE L. 1978. Some factors affecting the laying date, incubation and breeding
success of the Manx Shearwater, Puffinus puffinus. J. Anim. Ecol., 47: 477-495.
BROWN, R. G. B. 1967. Breeding success and population growth in a colony of Herring
and Lesser Black-backed Gulls, Larus argentatus and L. fuscus. Ibis, 109: 502-515.
COULSON, J. C. 1966. The influence of the pairbond and age on the breeding biology of
the Kittiwake Gull Rissa tridactyla. J. Anim. Ecol., 35: 269-279.
DAVIS, J. W. F. 1976. Breeding success and experience in the Arctic Skua Stercorarius
parasiticus. J. Anim. Ecol., 45: 531-536.
DAVIS, J. W. F., AND E. K. DUNN. 1976. Intraspecific predation and colonial breeding in
Lesser Black-backed Gulls Larus fuscus. Ibis, 118: 65-77.
DRENW, R. 1973. The natural history of incubation. In Breeding Biology of Birds. D. S.
Farher (ed.). Washington, D.C., Nat. Acad. Sci.
DUNNET, G. M., A. ANDERSON, AND R. M. CORMACK. 1963. A study of survival of adult
Fulmars with observations on the pre-laying exodus. Brit. Birds, 56: 2-18.
DUNNEW, G. M., AND J. C. OLLASON. 1978. The estimation of survival rate in the Fulmar
Fulmarus glacialis. J. Anita. Ecol., 47: 437-450.
DUNNET, G. M., j. c. OLLASON, AND A. ANDERSON. 1979. A 28-year study of breeding
Fulmars Fulmarus glacialis in Orkney. Ibis, 121: 293-300.
ELLISON, L. N., AND L. CLEARY. 1978. Effects of human disturbance on breeding of
Double-crested Cormorants. Auk, 95:510-517.
GILLET, W. H., J. L. HAYWARD, JR., AND J. F. STOUT. 1975. Effects of human activity on
egg and chick mortality in a Glaucous-winged Gull colony. Condor, 77: 492-495.
GOCIFELD, M. 1978. Incubating behaviour in Common Terns: Influence of wind speed
and direction on orientation of incubating adults. Anita. Behav., 26: 848-851.
HARRIS, M.P. 1964. Aspects of the breeding biology of the Gulls Larus argentatus, L.
fuscus and L. marinus. Ibis, 106: 432-456.
HUNT, G. L., JR. 1972. Influence of food distribution and human disturbance on the
reproductive success of Herring Gulls. Ecology, 53: 1051-1061.
KUR3/4, C. R., AND M. GOCHFELD. 1975. Human interference and Gull predation in Cor-
morant colonies. Biol. Conserv., 8: 23-34.
MACDONALD, M. A. 1977. The pre-laying exodus of the Fulmar Fulmarus glacialis (L.).
Ornis Scand., 8: 33-37.
MATHEWS, G. V. T. 1954. Some aspects of incubation in the Manx Shearwater, Procellaria
puffinus, with particular reference to chilling resistance in the embryo. Ibis, 96.' 432-
440.
MILLS, J. A. 1973. The influence of age and pair-bond on the breeding biology of the
Red-billed Gull ( Larus novaehollandiae scopulinus ). J. Anita. Ecol., 42: 147-162.
Moss, R., A. WATSON, AND R. PARR. 1975. Maternal nutrition and breeding success in
Red Grouse (Lagopus lagopus scoticus) J. Anita. Ecol., 44.' 233-244.
MoUCN, J. L. 1967. -tudecologique des deux espces de Fulmars--le Fulmar Atlan-
tique (Fulmarus glacialis) et le Fulmar Antarctique (Fulmarus glacialoides). œ'Ois. et la
R.F.O., 37 57-103.
NISBET, I. C. T. 1975. Selective effects of predation in a tern colony. Condor, 77: 221-
226.
NOLAN, V., Jm, AND C. F. ThOmPSON. 1978. Egg volume as a predictor of hatchling
weight in the Brown-headed Cowbird. Wilson Bull., 90: 353-358.
OELE, H. 1975. Breeding behaviour and success in a colony of Adlie Penguins Pygoscelis
adeliae at Cape Crozier, Antarctica. In The Biology of Penguins. B. Stonehouse (ed.).
London and Basingstoke, Macmillan.
OLLASO, J. C., AN G. M. DUNNET. 1978. Age, experience and other factors affecting
the breeding success of the Fulmar Fulmarus glacialis in Orkney. J. Anita. Ecol., 47:
961-976.
ORGANS, G. H. 1969. Age and hunting success in the Brown Pelican (Pelecanus occidentalis).
Anim. Behav., 17: 316-319.
PARSONS, J. 1970. Relationship between egg size and post-hatching chick mortality in the
Herring Gull (iLarus argentatus). Nature, 228.' 1221-1222.
REID, B. 1968. An interpretation of the age structure and breeding status of an Adlie
Penguin population. Notornis, 15: 193-197.
RIngLEts, R.-E., D.C. HAHN, AN W. A. MONTEVECCm. 1978. The relationship between
egg size and chick size in the Laughing Gull and Japanese Quail. Auk, 95: 135-144.
ROBERT, H. C., AND C. J. RaLph. 1975. Effects of human disturbance on the breeding
success of Gulls. Condor, 77 495-499.
TNAZA, R. 1971. Behaviour and nesting success relative to nest location in Adlie Pen-
guins (Pygoscelis adeliae). Condor, 73: 81-92.
WARnA, J. 1964. Breeding behaviour in Procellariiformes. In Biologie Antarctique. R.
Carrick, M. Holdgate, and J. Prevost (eds.). Paris, Hermann.
WILEY, W. H. 1950. The influence of egg weight on the pre-hatching and post-hatching
growth rate in the Fowl. Poultry. Sci., 29: 570-574.
Culterty Field Station, Department of Zoolog?i, University of A berdeen, Aberdeen
AB40AA. Received 3 May 1979, accepted 29 August 1979.