2 
PACIFIC SCIENCE, Vol. XXII, January 1968 
But on the same islands other species have 
annual breeding regimes, indicating that for 
them seasonal influences are sufficiently strong 
to outweigh the selective advantages of breed- 
ing at shorter intervals (Ascension Island: pa- 
pers in Ibis 103b, 1962-63; Christmas Island: 
Gallagher, I960; Ashmole, 1965; Ashmole and 
Ashmole, 1967; Ashmole, in press; Galapagos 
Islands: Leveque, 1964; Snow, 1965). Some ad- 
vances are now being made in understanding 
the differences in feeding ecology which lead 
to dependence of different birds on different 
groups of prey species, but lack of knowledge of 
the importance of seasonal effects in populations 
of the various prey species prevents further 
progress. The reproductive cycles of oceanic fish 
and cephalopods in relatively seasonless areas 
are even less well understood, but are of both 
theoretical and practical importance. 
The purpose of the present paper is to show 
that a very economical investigation of seasonal 
(and year-to-year) variation in the surface fauna 
of a tropical oceanic area could be carried out 
by making use of the sampling of the oceanic 
environment which is done by sea birds, and 
at the same time by collecting certain other data. 
The method which we propose depends on 
the fact that tropical sea birds of many species 
are easy to catch and, when caught, will often 
regurgitate the whole or part of their latest 
meal. Regurgitations provide, on average, far 
more items in good condition than do stomachs 
of shot or netted birds and their collection does 
not appreciably affect the bird population. Re- 
gurgitations are stored in formalin and the 
individual food items identified and measured 
at leisure. Since the feeding methods of each 
species are reasonably constant, regular sampling 
on an adequate scale would permit detection of 
changes in the relative availability of different 
prey species. Furthermore, each of the bird spe- 
cies takes prey covering a considerable range in 
size, so that the data should reflect changes in 
the size-frequency distributions of the various 
prey species. Deductions could doubtless also 
be made about the timing of reproduction in 
some of the fish and squid species. The method 
could be used on almost any tropical island on 
which appropriate species of sea birds are avail- 
able for a large part of the year, but the decision 
as to which bird species should be used for 
sampling depends on considerable knowledge 
of the ecology of the available species. 
In the course of our study of the food of sea 
birds on Christmas Island during the period 
March 1963 to June 1964, we examined 800 
samples (mainly regurgitations) from eight spe- 
cies of sea birds. We were mainly concerned 
with comparing the food of different species, 
and we did not have the opportunity to collect 
very large numbers of samples from any one 
species, so that our data are not suitable for 
detailed seasonal analysis. However, they do 
demonstrate the potentialities of the method. 
A complete account of the study is available in 
Ashmole and Ashmole (1967), and here we 
shall discuss only those aspects which are perti- 
nent to the planning of a more general inves- 
tigation, and in particular to the choice of bird 
species from which samples should be obtained. 
The samples which we collected on Christmas 
Island consisted of entire or partially digested 
fish and squid, while two bird species also pro- 
vided various invertebrates other than squid, 
mainly water striders ( Halobates ) and crusta- 
ceans. Nearly all the squid belonged to a single 
genus ( Sym plectoteuthis ) of the family Om- 
mastrephidae, but the fish were distributed 
among 33 families. The state of the samples 
varied considerably, but there were fairly con- 
sistent differences among the bird species (Table 
1). These differences resulted partly from the 
fact that in some species it was easiest to obtain 
samples from young birds which had previously 
been fed by adults, while in others it was easy 
to catch adults. The best samples were those ob- 
tained as the birds arrived to feed their chicks, 
while those obtained from roosting adults were 
generally more digested and thus the items were 
more often unidentifiable. It would be impor- 
tant, therefore, to plan future work in such a 
way as to obtain the food items as soon as pos- 
sible after they were caught by the birds. 
The usefulness of a particular species would 
depend largely on the ease with which large 
numbers of identifiable fish could be obtained 
from it, so we have calculated (Table 1) the 
number of fish which we were able to identify 
(to family level) per 100 samples collected. In 
the following short accounts of the most promis- 
ing bird species we have also commented on the 
abundance of each species, the ways in which it 
