A. Diaz  Agriculture, Ecosystems and Environment 78 2000 249–259 255
Table 5 Spearman Rank correlations r between species’ abundance  cover and levels of rabbit incidence number of faecal pellets in 625 m
2
sites in each region
a
Fintray n = 40 sites Oyne n = 40 sites
Dalcross n = 30 sites Species
r P
Species r
P Species
r P
Aphanes arvensis +
0.66 0.001
Myosotis discolor +
0.73 0.001
Poa annua +
0.80 0.001
Bellis perennis +
0.66 0.001
Holcus lanatus +
0.67 0.001
Veronica officinalis +
0.76 0.001
Poa annua +
0.66 0.001
Cerastium fontanum +
0.67 0.001
Myosotis discolor +
0.75 0.001
Agrostis capillaris +
0.54 0.001
Trifolium repens +
0.62 0.001
Erophila verna +
0.70 0.001
Myosotis discolor +
0.54 0.001
Lolium perenne +
0.58 0.001
Aphanes arvensis +
0.69 0.001
Cerastium fontanum +
0.50 0.001
Cirsium arvense +
0.50 0.001
Capsella bursa-pastoris +
0.64 0.001
Trifolium dubium +
0.50 0.001
Veronica serpyllifolia +
0.47 0.002
Myosotis arvensis +
0.64 0.001
Senecio jacobaea +
0.48 0.002
Ranunculus repens +
0.45 0.004
Viola arvensis +
0.58 0.001
Sagina procumbens +
0.48 0.002
Bellis perennis +
0.36 0.024
Agrostis capillaris +
0.53 0.003
Veronica officinalis +
0.48 0.002
Cirsium vulgare +
0.33 0.037
Vicia sativa +
0.52 0.003
Viola tricolor +
0.41 0.009
Papaver rhoeas +
0.50 0.005
Trifolium repens +
0.38 0.015
Trifolium dubium +
0.49 0.005
Cirsium vulgare +
0.35 0.025
Geranium molle +
0.44 0.015
Rumex acetosa +
0.37 0.044
Senecio vulgaris +
0.36 0.049
Dactylis glomerata −
0.54 0.001
Elymus repens −
0.52 0.001
Dacylis glomerata −
0.76 0.001
Poa trivialis −
0.48 0.002
Holcus mollis −
0.48 0.002
Taraxacum officinale group −
0.72 0.001
Arrhenatherum elatius −
0.42 0.007
Poa trivialis −
0.56 0.001
Epilobium obscurum −
0.32 0.048
Chamaenerion angustifolium −
0.53 0.003
Ranunculus repens −
0.47 0.009
Geranium dissectum −
0.46 0.010
a
Only species that occurred in at least 20 of the sites per region are shown. Only correlations of P  0.05 are shown.
correlated to levels of rabbit incidence Table 5. Species that were significantly more abundant
where rabbit incidence was higher tended to be low-growing forbs, e.g., Bellis perennis and Myoso-
tis discolor, or species of disturbed ground, e.g., Papaver rhoeas and P. annua. By contrast, species
whose abundance was negatively related to the inci- dence of rabbits were often taller-growing grasses
e.g., Dactylis glomerata and Poa trivialis.
3.3. The relationship between plant palatability and sward composition
The extent to which differences in sward compo- sition between grazed and ungrazed sites could be
related to differences in plant palatability was inves- tigated for the 22 species in the palatability trials.
Although the sites showed a range of incidence of rabbits and plant abundance Fig. 2, they were for
the purpose of this analysis categorised as ‘grazed’ or ‘ungrazed’ based on whether they had more than 125
faecal pellets per 100 m
2
per week recorded on the site. The mean relative abundance of each plant species on
‘ungrazed’ and ‘grazed’ sites was calculated. A clear relationship emerged when the percent palatability of
each plant species were plotted against the ratio of its mean relative abundance on ungrazed and grazed sites
Fig. 3. Most very palatable species were seen to be far more abundant on ungrazed sites and most un-
palatable species were more abundant on grazed sites.
4. Discussion
This study found a relationship between measured plant palatability to rabbits and the comparative abun-
dance  cover of these plant species in grazed and ungrazed swards. In general, the more palatable the
species, the greater its relative  cover in ungrazed swards compared to that in grazed swards. This rela-
tionship was particularly strong for most of the highly palatable and highly unpalatable species. Given that
strong positive correlations have been found between
256 A. Diaz  Agriculture, Ecosystems and Environment 78 2000 249–259
Fig. 2. The relationship between the incidence of rabbits on each of the 110 sites studied measured as number of faecal pellets found and the abundance measured as  cover of a palatable plant species Poa trivialis; an unpalatable species Myosotis discolor; a species
of intermediate palatability Poa annua. Open triangles represent Poa trivialis. Crosses represent Myosotis discolor. Solid circles represent Poa annua.
competitive ability and palatability for species in the Silwood grassland Crawley, 1990, this finding is con-
sistent with the prediction of competitor release of unpalatable species when exposed to rabbit grazing
Pacala and Crawley, 1992. The quite low levels of palatability found for P. annua and H. lanatus agree
with previous work which suggests that, compared to other grasses, these are avoided by rabbits Fenton
1940 and Gillham 1955 for H. lanatus and Ham- bler et al. 1995 for P. annua. For other species
with intermediate palatabilities, the relationship was weaker, perhaps because the direct effect of grazing
is less important than other direct and indirect effects of rabbits. For example, P. annua is far more abun-
dant in heavily grazed areas despite its intermediate palatability to rabbits. P. annua germinates more read-
ily in light conditions on disturbed ground Grime et al., 1988 and the greater availability of favourable
microsites for seed germination may explain some of its greater abundance in grazed swards. The two
species for which predictions of abundance based on palatability data would have given the poorest fit to
observed results were Rumex acetosella and F. rubra. R. acetosella was far more abundant on grazed sites,
despite its apparently high palatability to rabbits. It is probable that its greater abundance in grazed swards
is due to the indirect effect of competitor release as many of the other plant species on the fertile, ex-arable
land studied can grow far taller than can R. acetosella. The palatability observed for F. rubra was lower than
might be expected from accounts of it increasing cover in the absence of rabbits e.g., Gillham, 1955; Ran-
well, 1960; Thomas, 1963; Crawley, 1990. It may be that the palatability data for F. rubra obtained in this
A. Diaz  Agriculture, Ecosystems and Environment 78 2000 249–259 257
Fig. 3. The relationship between a plant species’ palatability to rabbits and the ratio of its abundance  cover in grazed and ungrazed sites.
study is less reliable than for other species as the vari- ance of these data was relatively high.
The precise interaction of factors which determines feeding preference for different plant species by rab-
bits is unknown and may vary over time Chapuis and Forgeard, 1982. Roles have been suggested for tis-
sue protein content Beaumont et al., 1933; Burton et al., 1964; Myers and Bults, 1977 and for chemical
and physical defences Burton et al., 1964. Such de- fences may become more pronounced as the plant ages
Beaumont et al., 1933; Burton et al., 1964. Also, it is known that tissue damage caused by rabbit grazing
may induce the production of chemical defences, such as glucosinolates, in some species Macfarlane Smith
et al., 1991. In this study, the plants in the field var- ied in age and included flowering adults. By contrast,
those in the palatability trial were at a pre-flowering stage of development and had not been previously sub-
jected to herbivore attack. However, despite all of these possible sources of variation, results from this study
showed that, for most species growing in established set-aside grasslands, their relative plant abundance in
grazed and ungrazed swards could be predicted from their performance in palatability trials. This suggests
that palatability trials can offer a methodology that need not be limited in its validity to providing esti-
mates of relative herbivory of field plants that are at a similar stage of growth as those in the palatability trial.
In this study, the palatability of species was as- sessed by placing plants into heavily grazed swards,
all of which were quite close to warrens. Although the species were found to differ greatly in their palatabil-
ity, this method may produce a conservative estimate of differences in palatability. This is so because areas
close to warrens have been found to be most grazed by rabbits when they first exit the warren Southern,
1940 when they are perhaps most hungry and least selective in their grazing. The relative palatability of
plant species to rabbits was found to be consistent across a number of different rabbit populations and a
258 A. Diaz  Agriculture, Ecosystems and Environment 78 2000 249–259
range of background vegetation types. There is con- siderable anecdotal evidence that suggests that some
species are consistently ignored by rabbits if there is alternative food, e.g., Urtica dioica, Cirsium spp. and
S. jacobaea, whilst others, including many grasses, appear to be consistently palatable reviewed in Tans-
ley, 1949; Thompson and Worden, 1956; Thompson, 1994. Despite this, the degree of consistency found
in this study could be considered as surprising as in- fluences on relative palatability have been detected
in captive rabbits. These include pre-exposure to the plant species either directly or through maternal trans-
mission Bilkó et al., 1994 and effects of specific plant species’ combinations when the palatability of a
number of species is measured by offering only pairs of species at any one time Soane, 1980. It may be
that pre-exposure also affects palatability in field con- ditions but that the range of plants eaten by the wild
rabbits was so large that this effect was not sufficient to be detected in the present field study. Likewise,
it may be that the palatability of a species was not greatly affected by the nature of its adjacent ‘back-
ground’ species in this study because the rabbits were free ranging and had ample food sources.
5. Conclusions