Discrimination between grass species and nitrogen-fertilized
vegetation by young Barnacle Geese
M YRFYN
O W E N , M A R T IN
NUGENT
Introduction
M any studies have indicated that wild geese
are s e le c t iv e g r a z e r s a n d t h a t th e ir
preferences are related to th e nutritional
characteristics o f their foods (e.g. O wen
1973, 1975a; H arw ood 1975). In a series o f
te sts w ith c a p tiv e S n o w G e e se A n s e r
caerulescens and C anada G eese B ra n ta
can aden sis, Lieff et al. (1 9 7 0 ) found highly
significant differences in the selection o f six
plant species. T hey suggested that selection
becam e m ore efficient as th e experim ent
proceeded, i.e. the geese cou ld ultim ately
recognize their preferred food s. Their tests
included grasses, sedges and spike-rush,
w hich varied in height from 1 -2 cm to
3 0 - 6 0 cm and were o f varying degrees o f
c o a r se n e ss. T here w ere, h o w ev er, no
relationships between the preferences show n
by the geese and the height, nutritional
characteristics o f the plants, or soil pH . W ild
geese grazing winter pastures face a situation
where there is a very lim ited height range
and where all the grass sp ecies could be
described as ‘fine’. Even in this situation,
O w en (1 9 7 6 ) d e m o n stra te d c o n siste n t
preferences in W hite-fronted G eese A n se r
albifron s for certain plant sp ecies in m ixed
sw a rd s, alth o u g h n o rela tio n sh ip w ith
nutritional com position w as dem onstrated.
In an ex p erim en t on th e sa m e sw ard ,
how ever, geese were show n to select nitrogen
fertilized vegetation (O w en 1975a).
This paper describes an experim ent to test
w h eth er y o u n g B arn acle G e e se B ra n ta
leu copsis reared in isolation from their
parents discrim inate between four grass
species clipped to a standard height and
between trays o f one o f the sp ecies treated
with different levels o f nitrogen fertilizer.
M aterials and m ethods
Ten goslings, hatched in an incubator, on
14th and 15th June 1976, were reared for the
first tw o w eeks in wire-floored brooders and
fed on poultry pellets and greenfood (lettuce
and duckw eed L em n a sp.). T h ey w ere then
transferred to a wire enclosure w here poultry
pellets, later m ixed with grain, and flint grit
were alw ays available. The goslin gs were
marked at a day old with m onel foot webW ildfowl 28 (1977): 2 1 -2 6
and N IC O L A
D A V IE S
tags and at four w eek s with coloured plastic
leg rings for individual recognition. T hey
w ere w e ig h e d p e r io d ic a lly and v isited
frequently so that they w ere familiar with the
experimenters.
The four grass species, perennial ryegrass
L oliu m p eren n e, bent A g ro s tis ten uis, red
fescue F estuca rubra and m eadow grass P oa
p ra te n sis, w ere grown from com m ercial seed
in a 3:1 mixture o f soil and peat in 36 x 21
cm seed trays. The soil w as thoroughly
m ixed and 16 trays (4 for each species) filled
from the sam e sam ple. A further 12 trays
were sow n with L oliu m and four fertiliser
treatm ents applied ju st after em ergence
o f the sh oots: (0) control, (1) 1-5 gm /tray,
(2) 3 gm , and (3) 4-5 gm o f fertilizer. The fer­
tilizer used as N itrochalk (25% N ) and the
levels corresponded approxim ately to 0, 2 00,
4 0 0 and 6 0 0 kg/h a. The trays were sow n
3 - 5 w eeks before they w ere presented to the
geese and were clipped to a standard 5 cm
above soil level prior to the tests.
T he original intention w as to present
grasses in pairs to geese in pairs in a sm all
holding pen, but because the geese becam e
nervous and did not attem pt to peck at the
vegetation under these circum stances, this
proved unworkable. T w o gro u p s o f four
geese (A and B) w ere eventually separated,
and kept in contiguous pens, into w hich test
trays were introduced. T rays w ere presented
in c om bin ation s o f tw o sp ecies (or nitrogen
treatm ents) and tw o com binations presented
consecutively constituted a rou nd, which
thus included each o f the four species or
treatm ents. A replicate consisted o f three
ro u n d s, and in clu d ed th e six p o ssib le
com binations o f four treatm ents, e.g. for
species R eplicate 1: round 1— A g ro s tis/
F estuca, L o liu m /P o a ; round 2— A /L , F /P ;
round 3— A /P , L /F . Three replicates were
carried out on sp ecies and four on fertilizer
treatm ents.
In the even in g b e fo r e th e te st little
supplem entary food w as given (w hich w as
consum ed very quickly), and the tests were
carried out between 0 8 .0 0 and 10.00. There
w as little grass available in the pen due to a
severe drought. The trays were placed 20 cm
apart in the centre o f the pen (Figure 1), and
positions with respect to the geese, w hich
21
each species determ ined. After the replicate
trials had been com pleted, trays o f single
species and treatm ents were presented to the
geese and droppings collected so that the
breakdown index (O wen 1975b) could be
determined. O ne hundred faecal fragments
were exam ined for each treatment.
R esults and discussion
S pecies selection
Figure I. Two goslings of Group A approaching
trays o f two different species. Before the actual tests
the grass was clipped to the same height in all trays.
retreated to one side o f the pen, continually
changed, so that on e species w as not alw ays
on the left or the right.
Recordings w ere started when the first
goose pecked at the grass (Figure 2), and
continued for five m inutes. T w o observers
recorded on tape (a) the number o f pecks,
and (b) the tim e spent pecking on each tray
by each o f the four birds. A sam ple o f grass
w as clipped from each tray on tw o dates in
early Septem ber, the water content assessed
and the nitrogen content o f pooled trays o f
The total num ber o f pecks, time spent
feeding (in seconds), and the feeding rate
(pecks per m inute o f feeding tim e, calculated
from total pecks and time) were subjected to
analyses o f variance. A s an exam ple, the full
analysis for total pecks in the grass species
tests, is given in T able 1. In order to reduce
com putation tim e the full bird by bird
analysis w as not carried out on all tests, in
som e cases the group totals were used. A
sum m ary o f the species preference is given in
Figure 3, w here significant differences are
show n by all three m easures o f feeding. A lso
show n is the number o f tests in which a
species w as first selected (by any goose).
This ‘first ch o ice’ is not related to subsequent
p referen ce, and w a s n ot different from
chance selection (X 2 test against 1 :1 :1 :1
Figure 2. Three of the goslings feeding on the preferred Festuca tray.
F o o d d iscrim in ation by B a rn a c le G eese
23
T a b l e 1. O v e ra ll a n a l y s is o f v a r i a n c e fo r th e n u m b e r o f p e c k s in th e s pec ie s te st s.
Source
Degrees of
fredom
SS2
Variance
F
P
287
95
3
2
7
6
21
14
42
2
190
1493705-3
806692-7
27253-5
33931-1
144181-2
98185-1
106381-6
40743-2
110747-1
963-2
686049-5
8491-5
90841-2
16965-6
20597-3
16364-2
5065-8
2910-2
2636-8
481-6
3610-8
2-35
25-16
4-70
5-70
4-53
1-4
<1
<1
<1
<0-001
<0-001
<0-05
<0-001
<0-01
NS
NS
NS
NS
Total
Treatm ents
Species
rounds
birds
Sx R
Sx B
R x B
Sx R x B
Replicates
Error
N S = not significant.
TOTAL
PECKS
FEEDING
P E C K IN G
TIM E
LSD=
26-2
LS D =
0
90-7
S
100
FIRST
RATE
CHOICE
L SD=
365
Figure 3. The selection of species by Barnacle Geese as assessed by four measures. L = Lolium
perenne, F = Festuca rubra, P = Poa pratensis, A = Agrostis tenuis. L.S.D. = Least Significant
Difference (at the 5% level).
ratio: X 2 = 0 -6 6 , P > 0 -7 5 0 ). A lthough the
grasses were clipped to a standard height
they were very different in colou r and tiller
density, and easily distinguishable to the
human eye. The absence o f co n sisten cy in
‘first ch o ice’ even at the end o f the experi­
m ent indicates that the geese did not learn to
r e c o g n iz e at a d ista n c e their preferred
species. There were significant differences
between the com binations presented, the
low est, as expected, being A g ro s tis /P o a .
Thus the geese not only preferred the other
tw o sp ecies but fed less heavily on these tw o
w hen presented together.
S election o f f e r tiliz e d vegetation
The selection o f fertilized L o liu m is sum ­
marized in Figure 4. A lthou gh each feeding
m e a su r e g iv e s a s ig n ific a n t d iffe r e n c e
between treatm ents this is alw ays between
the unfertilized and one or m ore o f the fer­
tilizer treatm ents. There w ere highly signifi­
cant differences between the activity o f
different birds (see T able 1), and there w as a
significant interaction betw een individual
birds and nitrogen level, i.e. individuals were
selecting different treatm ents. It becam e o b ­
vious, especially tow ards the end o f the
experim ent, that the m ore aggressive in­
d iv id u a ls w ere p r even tin g o th e r s from
feeding on the preferred trays. The analysis
w as carried out using on ly three and only
tw o birds from each group and the variance
ratios were thereby increased. There were
still, h o w ev e r, no sig n ifica n t d ifferen ces
between any fertilizer treatm ents.
The ‘first ch o ice’ test w as again not related
M yrfyn O wen, M a rtin N u g en t a n d N ico la D a vie s
24
to preferences and there w as no significant
differences from the expected 1 :1 :1 :1 ratio
(X 2 = 4 -0 8 , P > 0 -2 5 ), despite the fact that
colour differences betw een the control and
fertilized trays w ere obviou s to the hum an
eye.
protein content o f F estu ca w as 46% higher
than that o f L o liu m but the geese had m ore
pecks at L oliu m . Experim ental variability
w as high, how ever, and nitrogen w as deter­
mined on p ooled sam ples, so statistical
analysis o f these relationships can n ot be
carried out. The fragm entation index (Figure
5 b) relates clo sely to water content and if the
in d e x g iv e s a g o o d in d ic a t io n o f th e
availability o f nutrients (O w en 1976), this
m ay indicate the reason for the apparently
closer relationship betw een water content
a n d p r e fe r e n c e . F ig u r e 6 s h o w s th e
relationship betw een preference (total pecks)
C h a ra cte ristic s o f the te st vegetation
The selection o f sp ecies and fertilized vegeta­
tion is related to its water and protein co n ­
tent in Figure 5. In both the sp ecies and the
nitrogen tests the preference o f the geese
more closely relates to w ater content than to
protein content, in term s o f fresh weight. The
TOTAL
PECKS
FEE D IN G T IM E
P E C K IN G
RATE
FIRST
C H O IC E
LSD=
49-6
■
LSD=
204
LSD=
40_
_
35-8
-
0
1
2
3
Figure 4. The selection o f Lolium perenne treated with different levels o f Nitrogen fertilizer. 0 = con­
trol, 1 = 1 - 5 gm/tray, 2 = 3-0 gm/tray, 3 = 4-5 gm/tray.
a)
300
A
P
L
F
A
P
L
F
A
P
L
F
1
2
3
b)
60
d
No. o f
0
Pecks
°/0 W a te r
%
Protein (w e t w t.)
0
1
2
3
F ragm en tation Index
Figure 5. The relation between preference and the water and protein content (%N x 6-25) of
vegetation; (a) species tests, (b) fertilizer tests. The fragmentation index is the percentage of fragments that
appear in droppings with only a single surface intact (Owen 1975b).
25
Food discrim ination by B arnacle Geese
and the ‘available protein ind ex’, w hich is
calculated from % protein (w et w eight) x
fragm entation in d e x /100. A lthou gh this cor­
relation is very clo se (r = 0 -9 9 8 ) a large
am ount o f experim ental error is not taken
into accou n t and it can n ot be regarded as a
precise relationship.
H arw ood (1 9 7 5 ) found that the applica­
tion o f nitrogenous fertilizer to a grass sward
increased the water content as well as the
protein content o f the vegetation. M oreover,
th e w a te r a n d p r o te in c o n te n t w ere
Available
Protein
Index
Figure 6. The relationship between preference (No. of pecks) and the ‘available protein index’ for the
fertilizer tests. ‘Available protein index’ = % protein (wet wt.) x breakdown index/100.
S pecies ■
N1 •
N2
80-
604
*
—T —
10
20
°/0 Protein
(d ry
30
m a tte r)
Figure 7. The relationship between protein and water content of test vegetation. Squares — species,
circles = Nj (fertilizer tests sam pled 6.9.76), and triangles = N 2 (fertilizer tests sam pled 14.9.76).
26
M yrfyn Owen, M artin Nugent an d N icola D avies
significantly and positively correlated. H e
suggested that the birds were responding to
changes in w ater content but the correlation
m ade it im possible for the tw o factors to be
separated. Figure 7 sh ow s the relationship
between protein and water content o f the
present test vegetation. T he overall relation­
ship is significant (r = 0 -6 2 , P < 0 -0 5 ) but
within each set o f sam ples the relationship is
m uch closer (r for Species = 0 -9 3 2 , for N , =
0 -9 4 7 , for N 2 = 0 -8 7 3 ). T he protein content
o f vegetation in the fertilizer tests decreased
from 23—30% ( o f dry weight) at the first
sam pling to 1 0 -1 9 % at the second, but
w ater content rem ained the sam e. This w as
because fertilizer w as leached out o f the
shallow trays by watering. T ow ards the end
o f the experim ental period fertilized trays
were ob viou sly a brighter green than c o n ­
trols. U nfortunately, because com parisons
cannot be m ade betw een groups o f tests, the
effect o f differences in protein and water con ­
tent on selection cannot be separated in the
present experim ent. The closer concord ance
betw een p referen ce and w ater c o n te n t
(Figure 5) than with protein content m ay,
how ever, su ggest that the birds are really
responding to changes in water content.
T he results o f the present experim ent
suggest that geese w ould fare better if they
selected vegetation on the basis o f its water
content than its protein content. This is
because water content is correlated with, (a)
protein content, and (b) the fragm entation in­
dex which affects the availability o f that pro­
tein. It seem s likely that the brittleness o f
leaves w ould be related to their water c o n ­
tent, and ch an ges in such m echanical proper­
ties m ay be detectable by geese. It has been
suggested (O w en 1976) that geese select
vegetation by exerting a certain pressure or
‘puli’ so that m ore brittle leaves break o ff and
are ingested w hile the tougher on es slip
through. The fact that geese in this experi­
m ent did not appear to select trays visually,
and that there is an apparent relationship
betw een w ater content and selection , is c o n ­
sistent with this hypothesis. In order to test
this further, m ore experim ents are needed
where the nitrogen and water content o f
vegetation are varied independently (the
change in protein content betw een N , and N 2
(Figure 7) su ggests that this is possible), and
where the m echanical properties o f leaves
(e.g. tensile strength) are related to their
water content.
Acknowledgements
We are grateful to the C urator, M. R. Lubbock,
for the loan of the birds and the use of the
experimental area, and to S. Lewis for assistance
during the experiments.
Summary
An experiment is described which tested the abili­
ty of young Barnacle Geese Branta leucopsis to
discriminate between grass species and between
trays o f a single species treated with different
levels o f Nitrogen fertilizer.
The geese showed significant preferences for
perennial ryegrass Lolium perenne and red fescue
Festuca rubra against bent Agrostis tenuis and
meadow grass Poa pratensis, and they significant­
ly preferred fertilized to unfertilized Lolium,
although they did not select either species or fer­
tilized trays visually. The preferences were cor­
related with both the water and nitrogen content
of the vegetation, and these were themselves
positively correlated. W ater content seemed to
relate better to the preferences shown than did
protein content.
It is suggested that geese would ingest more
protein by selecting leaves on the basis of their
water rather than their protein content. This was
because w ater content was not only correlated
with protein content but with the ease with which
vegetation was broken up in the gizzard, which in
itself controls the availability o f nutrients to the
bird.
Although not conclusive the results o f this
experiment were consistent with the hypothesis
put forward earlier that geese select grass leaves
and species on the basis of their mechanical not
their visual properties.
R eferences
Harwood, J. 1975. Grazing strategies of Blue Geese A nser caerulescens. Ph.D. thesis, University of
W estern Ontario.
Lieff, B. C., M aclnnes, C. D. & Misra, R. K. 1970. Food selection experiments with young geese.
J. Wildl. M gmt. 34: 321-7.
Owen, M. 1973. The m anagem ent of grassland areas for wintering geese. W ildfowl 24: 123—30.
Owen, M. 1975a. Cutting and fertilizing grassland for winter goose management. J. Wildl. M gm t. 39:
163-7.
Owen, M. 1975b. An assessment of faecal analysis technique in waterfowl feeding studies. J. Wildl.
M gmt. 39: 271-9.
Owen, M. 1976. The selection of winter food by W hite-fronted Geese. J. Appi. Ecol. 13: 715-29.
D r M. Owen, M. J. Nugent & N. Davies, Wildfowl Trust, Slimbridge, Gloucester, G L2 7BT.