Pasture Ecology Meeting Plant-Herbivore Relationships · ty, biogeochemical cycles and landscapes. Thus, there is a considerable need to improve scientific knowledge by including

Pasture Ecology Meeting

Plant-Herbivore Relationships

Clermont-Ferrand / Theix, 31 May and 1 June 2002

Conference organized by

P. Carrère and J.F. Soussana

(INRA Agronomy Unit)

R. Baumont

(INRA Herbivore Research Unit)

Published in:

ANIMAL RESEARCH - Volume 52 - Number 2 - 2003 - pp. 141 - 190http://www.edpsciences.org/animres/

Foreword

The Pasture Ecology meeting was organized in Clermont-Ferrand/Theix on 31May and 1st June as a satellite meeting of the 19th congress of the EuropeanGrassland Federation held in La Rochelle (France) in May 2002. Eighty-eightscientists from seventeen countries joined this meeting and had deep and stimu-lating scientific exchanges on vegetation dynamics, plant-herbivore interactionsand management in grassland ecosystems. We are pleased to publish here thethree main papers that were given in the session focussed on plant-herbivorerelationships.

Grazing is one of the major processes that affect grassland ecosystem dynamics.Grasslands are no longer seen only as a feeding resource for domestic herbivo-res, but also as an environmental resource through its contribution to biodiversi-ty, biogeochemical cycles and landscapes. Thus, there is a considerable need toimprove scientific knowledge by including a comprehensive view of how gra-zing can contribute to the multi-functionality of grasslands. Grazing results fromcomplex interactions between animals and vegetation. These interactions occurat different spatio-temporal scales from the smallest that is the prehension ofbites to the largest at the plot level. On all of these scales, animals create andrespond to heterogeneity of the vegetation. The creation and maintenance ofheterogeneity at different scales can be analysed in terms of constraints or bene-fits for grazing animals' diets and for grassland management for environmentalpurposes, biodiversity for example. The three papers gathered here bring anoverview of recent scientific advances in the comprehension of the grazing pro-cess from small to large scales and give some prospects in the development ofgrazing management for a multi-purpose use of grasslands.

R. BaumontGuest Editor

Review article

Sward structural resistance and biting effort

in grazing ruminants

Wendy M. GRIFFITHS*, Iain J. GORDON

Macaulay Institute, Craigiebuckler, Aberdeen AB15 8QH, Scotland, UK

(Received 5 September 2002; accepted 25 February 2003)

Abstract — Grazing ruminants face complex decisions in searching for and harvesting adequate for-

age to meet their requirements in the face of heterogeneity in the abundance, nutritive value and distri-

bution of resources. Some of the major decisions which affect the bite mass and therefore, forage

intake adjustments in bite depth, bite area and exerted bite force, are made in relation to heterogeneity

in sward structure. A number of relationships have been established linking the adjustment of bite

mechanics in response to variations in sward height and bulk density on temperate forages for animals

of a range of body sizes. However, there is less consistency in the response of the bite mechanics to the

greater variations in structural resistance that arise from the vertical and horizontal arrangements of

the plant morphological organs of leaf, pseudostem and stem. Furthermore, only limited progress has

been made in quantifying the biting forces involved in grazing, the linkages between the plant mor-

phological organs and their effect on bite force and, the interaction with body mass. In this paper, we

discuss the different hypotheses that have been proposed to explain the regulation of bite depth and

provide evidence for their acceptance or rejection. We comment on the knowledge gap in understand-

ing biting force mechanisms across animal species of contrasting body mass, and stress the need for

differentiation between the concepts of biting force and biting effort.

bite / effort / force / ruminants / structure

Résumé — Résistance de structure du couvert végétal et effort de prélèvement des bouchées

chez les ruminants au pâturage. Face à l’hétérogénéité de la ressource en abondance, en valeur nu-

tritive et en répartition, les ruminants au pâturage doivent prendre des décisions complexes pour re-

chercher et récolter les fourrages nécessaires à leurs besoins. Certaines de ces décisions affectent la

masse de la bouchée et donc, l’ajustement de l’ingestion en terme de profondeur et de surface de la

bouchée, ainsi que de force de prélèvement. Ces décisions sont prises en relation avec l’hétérogénéité

de la structure du couvert végétal. Nombre de relations ont été mises en évidence liant l’adaptation de

la mécanique de la bouchée aux variations de la hauteur et de la densité du couvert végétal de fourra-

ges tempérés pour des animaux de différentes tailles corporelles. En revanche, les grandes variations

145

Anim. Res. 52 (2003) 145–160

INRA, EDP Sciences, 2003

DOI: 10.1051/animres:2003012

* Correspondence and reprints

Tel.: + 44 (0) 1224 498200; fax: + 44 (0) 1224 311566; e-mail: [email protected]

de résistance des végétaux, qui résultent des arrangements verticaux et horizontaux des organes mor-

phologiques de la plante (feuille, gaine et tige), ont des effets variables sur la mécanique de la

bouchée. De plus, seuls des progrès limités ont été accomplis pour quantifier les forces de prélève-

ment des bouchées impliquées au pâturage, les liens entre les organes morphologiques de la plante et

leurs effets sur les forces de prélèvement, ainsi que l’interaction avec le format des animaux. Dans cet

article, nous discutons les différentes hypothèses qui ont été proposées pour expliquer la régulation de

la profondeur des bouchées, et nous proposons des éléments pour leur acceptation ou leur rejet. Nous

exposons les lacunes qui subsistent pour comprendre les mécanismes des forces de prélèvement à tra-

vers différentes espèces animales de formats distincts et nous soulignons le besoin de différencier

deux concepts : la force de prélèvement et l’effort de prélèvement.

bouchée / effort / force / ruminants / structure

1. INTRODUCTION

The harvesting of forage is a particularly

time-consuming exercise for ruminants; a

large number of bites, of a relatively small

bite mass, are required to meet the nutrient

intake requirements for maintenance,

growth and reproduction. A number of re-

views have discussed the mechanisms of

foraging behaviour [7, 9, 45, 50, 59, 60], in-

cluding the considerable advances that

have been made in understanding the influ-

ence of the independent effects of sward

height and sward bulk density on the com-

ponents (e.g. bite mass and bite rate) of

daily herbage intake [19, 30, 38, 59]. How-

ever, despite the comprehensive informa-

tion available, researchers are still unable to

predict the intake responses of grazing ani-

mals to changes in the vegetation structure

and morphology of more complex swards.

Illius and Hodgson [41] suggested that this

lack of predictive power emphasised the

abundance of descriptive studies, and

called for studies to actively seek a mecha-

nistic explanation of sward-bite interac-

tions. Descriptive and mechanistic approaches

are, however, essentially intertwined due to the

difficulty of clarifying the many confounding,

and indeed complex sward-bite interac-

tions.

Our aim is to focus attention on the im-

portance of the interaction between the

morphological plant organs encompassed

within the bite, and biting effort, an area

that has received considerably less atten-

tion, even though the effects of the distribu-

tion of plant morphological organs on the

feeding strategy in ruminant species has re-

cently been stressed [59]. The principles

should have direct relevance to ruminants

foraging in both temperate and tropical eco-

systems.

In this paper we define sward structure

and the terminology associated with the

study of bite mechanics, describe the anat-

omy of the grass leaf with respect to sever-

ance, and highlight the potential contrasts

in severance action across animals of differ-

ent body mass. The inconsistency in termi-

nology, the methodologies and their

suitability for assessing bite force are iden-

tified. Evidence of the hypothesised link-

age between bite depth, bite area and bite

force is discussed, and lastly the idea of as-

sessing biting effort is introduced.

2. SWARD STRUCTURE AND

BITING FORCE TERMINOLOGY

Studies investigating the linkages be-

tween the structural organs and food

comminution in the mouth have been re-

viewed [56], but there is comparatively less

information on the linkages between sward

structure and biting effort. Further progress

146 W.M. Griffiths, I.J. Gordon

within this field calls for a review of the def-

initions of the terminology used in this area.

Morphological changes that occur with

plant maturation lead to different propor-

tions, and relative age, of leaf, pseudostem,

stem, and inflorescence in the vertical di-

mension of the sward [13]. The architecture

of these organs within the sward canopy is

termed sward structure. Sward bulk den-

sity; the weight of herbage per unit volume,

increases with increasing sward depth. This

pattern is more pronounced for tropical pas-

tures than for temperate pastures [28, 67],

and the slope of the relationship can also

vary markedly across swards of different

structure within temperate pastures [see

25]. The volume of a bite is the major con-

tributor of intake rate, and comprises bite

depth, which is the vertical distance that an-

imals insert their muzzle into the sward,

and measured as the difference between

sward surface height and the residual height

post-grazing, and bite area, which is the

horizontal area of herbage encompassed

within a single bite [38, 48]. Evidence from

temperate swards shows that bite area ex-

hibits less sensitivity to sward variation

than does bite depth [12, 48, 53], although

Wade and Carvalho [74] suggest that bite

area is the primary parameter that animals

adjust to increase bite mass.

In linking the disciplines of plant me-

chanics with foraging behaviour we define

five plant-based terms: (i) fracture force in

tension, which is a measure of herbage

strength and is estimated from the maxi-

mum force that produces fracture of the

plant organ/s [80]; (ii) tensile strength,

which is the fracture force in tension per

unit cross-sectional area of the plant speci-

men [80, 81]; (iii) maximum force in shear,

also a measure of herbage strength and is

the maximum force required to fracture the

plant organ/s, and is determined from the

height of the highest peak on the force dis-

placement curve (N.B.: it does not repre-

sent the force to fracture the whole plant

organ) [80]; (iv) specific work to fracture,

equally known as toughness as it is mea-

sured by the energy required to shear the

test specimen per unit cross-sectional area

of the plant specimen [18, 80] and (v) resis-

tance, which is the theoretical, accumu-

lated force required for severing all of the

plant organs encompassed within a bite.

Two important animal-based terms are: (i)

bite force, which represents the three-dimen-

sional force applied in severing a bite (N.B.:

this term does not represent the force exerted

during chewing as given by Pérez-Barbería

and Gordon [57]) and (ii) biting effort, which

represents the power applied by the animal

to sever a bite and includes components of

resistance, bite force and head acceleration

(N.B.: the importance of the individual com-

ponents may interact with species body

mass and species differences in severance

action, see below SEVERANCE ACTION).

3. GRASS LEAF ANATOMY

Plants are the primary producers in most

food chains, and animal subsistence and

production is dependent upon them. For-

ages, like all plants, however, have evolved

structures that resist ingestion by grazing

animals (grazing resistance; [11]), as a

means of ensuring their own continued ex-

istence within ecosystems.

The grass leaf anatomy responsible for

protection against herbivory can be mod-

elled as a three component system [71],

consisting of sclerenchyma tissue, vascular

bundles and epidermal cells covered by a

protective cuticle. These components com-

prise the ‘fibre’ fraction, along with a ma-

trix consisting of thin-walled parenchyma

cells found in the leaf mesophyll. However,

these thick-walled cells that are heavily lig-

nified, and hence confer mechanical

strength to the plant by way of attachment

to chained bundles of vascular tissue only

account for a small proportion of leaf

cross-sectional area. Despite this, the fibre

Sward structure and biting effort 147

component accounts for 90–95% of the

longitudinal stiffness of grass leaves [71].

The critical disparity between the leaf

anatomy of grasses from the C3 and C4

photosynthetic pathways lies in the greater

number of veins, and smaller interveinal

distance [62, 78], and a higher proportion of

sclerenchyma [1, 78] in C4 species. These

structural characteristics strengthen plant

tissues [72]. In tropical ecosystems, envi-

ronmental temperature increases the rate of

plant growth, which increases plant

strength [72], and also reduces the moisture

content between the sclerenchyma fibres

[72], which leads to increased herbage

strength [34, 36]. Nevertheless, despite

possible categorical groupings of plants,

plant anatomy will vary between plant

groups, genera, species and even lines

within a species [77], but while some plants

may be stronger in tension, fracture proper-

ties will be more dependent on the organi-

sation of the bundles of sclerenchyma

which determines brittleness [73].

Wright and Vincent [81], in their review

of the mechanics of fracture in plants, dis-

cussed the three modes of crack propaga-

tion in a leaf; shear, tension and torsion, and

it is the modes of shear and tension that

have attracted interest from plant and ani-

mal scientists alike. Assessment of the me-

chanical properties of grass leaves is far

from new. It has provided information for

the design and improvement of the cutting

mechanisms in forage harvesting machines

[34, 51], and for plant breeding programmes

where the primary objective is to increase

forage intake [20, 36, 52], through increas-

ing the forage’s feeding value [70]. These

studies have focussed primarily on the

properties of fracture in shear. The interest

in relating the tensile properties of plants to

aspects of foraging behaviour, particularly

prehension, is of more recent origin [42, 68,

80], and has led to a clearer definition and

use of tensile strength to quantify fracture

properties in tension and the specific

work to fracture (i.e. toughness) in shear

for determining properties such as those as-

sociated with food comminution in the

mouth. Henry et al. [35] contended, how-

ever, that there was a lack of evidence to

support the partitioning of the fracture me-

chanics between prehension and chewing.

The sclerenchyma tissue content is often

reported in biomechanical studies where

the structural and fracture properties of

plants have been extensively evaluated [73,

80]. Fracture force in tension is well correlated

with the volume-fraction of sclerenchyma tis-

sue [80] when the sclerenchyma tissue is be-

low 15% of the volume-fraction of the leaf,

and the fibres are laterally separated [71,

73]. Above 15%, the fibres can be continu-

ous across the leaf, rendering the leaf brittle

[81], and thus more susceptible to fracture

since the leaf will lose a large proportion of

its strength with only a very small applied

force [72]. The plant properties that in-

crease mechanical resistance, chiefly the fi-

bre component, are essentially the same as

those that reduce plant digestibility [66,

80]. However, to rely on the nutritional in-

dices of Neutral Detergent Fibre (NDF) or

Acid Detergent Fibre (ADF) to provide

measures of herbage strength and resis-

tance would be inappropriate since fracture

force in tension (the maximum force that

produces fracture) is poorly correlated with

both NDF and ADF [80], whereas, for ex-

ample, the work to fracture in shear (the

area under the displacement curve) is well

correlated with both NDF and ADF, as well

as the lignin content of the leaf [80].

4. SEVERANCE ACTION

All ruminant species have a common

dental structure that consists of four paired

incisors set on the lower jaw, with a thick

pad of connective tissue, also known as the

dental pad, positioned on the upper jaw.

The act of severance entails the incisors

closing against the dental pad to grip each

mouthful of herbage before fracture of


plant tissue occurs, although cattle have

been observed to grip herbage against the

incisors using the tongue [16].

The structural arrangement of parallel

venation in grass leaves, ensures that leaves

exhibit little fracture sensitivity, thus the

damage at the point of incisor insertion

does little to act as a concentrator of stress

[72]. Wright and Vincent [81] commented

that unlike small herbivores which can

avoid the problems of indigestible fibres in

plant tissues, by eating around them (e.g.

larvae of the gum-leaf skeletoniser moth,

Uraba lugens) or cutting the lamina with

their mandibles or incisors (e.g. rabbits

have upper and lower incisors), ruminants

must use their strength. If strength scales

with body mass [42], then how do small-

bodied ruminants cope with the same con-

straints as large-bodied ruminants? Vincent

[72] suggested that sheep (Ovis aries), like

geese, probably use a strategy that involves

creating a crease on the lamina, through

pulling the lamina at an angle across the in-

cisors, and thereby reducing the force re-

quired to create fracture. There is no known

quantitative evidence for this severance ac-

tion in sheep, and there is the puzzling con-

cern, as Vincent noted, that it is difficult to

create creases across multiple lamina. Cat-

tle (Bos taurus), sheep and goats (Capra

hircus) foraging on a range of swards have

been observed to insert and/or move their

mouth sideways into swards to gather herb-

age [21, 28, 61]. In these examples it is

likely that animals were making use of the

shearing cusps on the molars to grip and

sever plant tissue [see also 63], thereby ob-

taining a bite mass above what would other-

wise be possible.

5. QUANTIFYING BITE FORCE

The idea of measuring the mechanical

resistance of plants is far from new, but the

motivation, approach and reporting have

varied considerably [5, 18, 22, 32, 35, 51,

52, 73, 80]. One consequence has been the

number of different descriptive terms that

have surfaced and these are often used in-

terchangeably. ‘Strength’ and ‘toughness’

are both biological terms that have been

adopted in quantifying plant mechanical re-

sistance [e.g. 17, 22], possibly since they

represent a physical attribute that is easily

understood. There is, however, some confu-

sion since toughness commonly implies

strength. Correctly defined, toughness is a

measure of the energy required to propa-

gate a crack, and not a measure of the force

involved in fracturing material [81]. This

definition is crucial since it implies that

toughness is an important mechanical pa-

rameter for the chewing dynamics, but not

for the process of severance of plant mate-

rial.

Other inconsistencies have emerged in

the definition of tensile strength and the use

of the term shear strength, which has no

clear definition in its own right, and, there-

fore, varies in terms of its units of expres-

sion [e.g. 23, 69, 80 for tensile and 20,

43 for shear]. According to Wright and

Vincent [81] tensile strength should be re-

ported as the force for crack propagation

per unit cross-sectional area of plant mate-

rial, and consistent expression of tensile

strength would greatly facilitate across-

study comparisons. However, it is not al-

ways a feasible use of resources to conduct

laborious estimates of tensile strength

within the context of animal-based studies.

Fracture force is a measure of the maximum

force required to fracture a test specimen

[80]. It takes no account of cross-sectional

area of the plant material [80], and so has

less acceptability for comparison across

different plant species with contrasting

tiller size-density relationships, and thus

potentially contrasting lamina cross-sec-

tional area. Fracture force measurements,

however, can be carried out more quickly

since they omit the additional laborious cross-

sectional area measurement, and thus can

provide objective information on herbage


strength, as long as the above constraints

are borne in mind.

When answering questions on the causal

relationship between bite force and bite me-

chanics, an area of concern lies with the es-

timates on single, whole or separated

fractions of plant organs and their relation

with clumped plant material encompassed

within a bite. For single tests to have direct

relevance, one question of particular impor-

tance, and that does not appear to have been

answered, is the potential error associated

with estimates using a constant multiplier

[e.g. 42]. In practical terms, it is unlikely

that the bite force required for the severance

of 10 leaves together will be the summation

of the strength of 10 single leaves, unless a

correction for the packing effect of leaves

encompassed within the bite is applied. The

absence of a constant multiplier would have

implications for the scaling of bite force

with modulations in bite area [e.g. 42, 68].

Other factors worthy of consideration are

that the number of leaves able to be creased

i.e. running at an angle over the incisors,

would be of a magnitude lower than the to-

tal number grasped. This is of particular im-

portance for the severance action in small

ruminants [see 72]. Furthermore, it is diffi-

cult to conceive that integrating the esti-

mates of fracture force in tension from

single, separate tests of leaf, pseudostem

and stem can model the predicted force for

a bite encompassing two or more morpho-

logical components, unless measurements

are made at a very fine level of precision

(i.e. 1 cm strata), forming serious con-

straints for swards more applicable for test-

ing with large-bodied animals.

Indirect assessments of bite force [42,

68] are derived from estimates of the force

required to fracture plant organs in the ver-

tical dimension. Although the contribution

from Illius et al. [42] predicted bite forces

indirectly from regressions based on a sepa-

rate set of swards, they amassed their data

from three studies, which may not be possi-

ble where time is a major constraint. Small-

scale apparatus’ that provide direct, quick

and reliable estimates from field conditions

would aid progress. There has been some

recognition of the need for the development

of portable field operated devices that mea-

sure fracture force in tension [68] and ten-

sile strength [76] in situ, providing an

estimate of the magnitude of the contrast in

resistance, for a given bite volume. Prog-

ress has nonetheless been slow and fraught

with difficulties; Westfall et al. [76] in a

short communication described a portable

tensilmeter with hand-operated modified

pliers and an electronic recording system,

and Tharmaraj [68] discussed a portable le-

ver operated clamping apparatus mounted

on a tripod frame, which expressed the peak

fracture force reading (Fig. 1). However,

for neither apparatus was it clear how the

rate of acceleration during fracture was

controlled.

Furthermore, an indirect estimate of bite

force from fracturing plant organs in ten-

sion does not fully mimic the mode of the

characteristic jerk action in the horizontal

plane during severance. By comparison, di-

rect quantification of the forces involved in

severing herbage in grazing animals yields

potentially more valuable information as

biomechanical force plates [75], such as

those used by Hughes et al. [40] and Laca

et al. [47] record three dimensional forces

(Figs. 2 and 3). Consequently, some dispar-

ity between values obtained from indirect

estimates as compared with direct method-

ology must be expected (Griffiths and

Gordon, in preparation). In the foreseeable

future, such technical difficulties will be-

come inconsequential if effort is made to

partition and present the force according to

the Cartesian coordinates. This is also a

critical point of evaluation as animal spe-

cies of contrasting body mass may allocate

the force between the Cartesian coordinates

differently. It is important to recognise that

bite force predicted indirectly from esti-

mates of fracture force and tensile strength,

provide only a theoretical framework for


comparative purposes. These predictions

also leave many unanswered questions as to

the effort exerted, particularly as evidence

does not suggest a 1:1 response in bite ge-

ometry with plant resistance (see below

SUMMIT FORCE HYPOTHESIS).

6. BITE DEPTH AND THE LINKAGE

WITH BITE AREA AND BITE

FORCE

An increasing number of studies have

focussed on the responses of the bite me-

chanics to the structural complexity and the

rigidity of plant organs, but can we draw

any consistency between studies and quan-

titatively relate the response patterns to bite

force? Three hypotheses have been pro-

posed for the regulation of bite depth pene-

tration:

– Summit Force (6.1).

– Balancing reward against cost of bite

procurement (6.2).

– Marginal Revenue (6.3).

In discussing the available evidence to

support or reject the above three hypotheses

we pull together studies which have pro-

vided supporting data on fracture force in

tension, tensile strength and leaf toughness

and/or bite force when examining the pa-

rameters of bite volume. There are rela-

tively few studies that meet this criterion,

and all of these come from the vegetative

temperate forages. Only a few studies [e.g.

54, 69] have assessed tensile strength on

forages in more extensive grassland envi-

ronments, but these studies have not pro-

vided parallel data on bite volume. There is a

need to evaluate the hypotheses for swards

displaying greater structural complexity


Figure 1. Tensile bite fracture force apparatus (from Tharmaraj [68]).

including those present in extensive grass-

land environments.

6.1. Summit force

The Summit Force hypothesis [37] was

originally offered as an explanation for the

association between bite depth and

pseudostem height [8, see also 2, 3] with the

rational response being one of animals being

constrained by the effort required to sever

pseudostem within the lower stratum. The

estimated fracture force of pseudostem was

later quantified [80] as approximately three

times that measured for a single leaf, and

arose from the complex arrangement of

concentric sheaths around the immature

leaves and growing points. In order to main-

tain consistency in bite force, the Summit

Force hypothesis implied that once the limit

in force had been reached, that animals


Base frame bolted to thefloor

Force plate transducers

Sward

Threaded holes for securingtop plate and sward to forceplate

Boards protecting theforce plate

Force plate

Figure 2. Exploded view of a force plate apparatus (adapted from Webb and Clark [75] and Griffiths

and Gordon, unpublished).


-30

-20

-10

0

10

20

30

40

Tugging the bite to the left

Tugging the bite to the right

Late

ralf

orce

(N)

5 10 15 20 25 30

-200

-150

-100

-50

0

50

Tugging the bite vertically

Pushing the herbage around during the grasping of the bite

Ver

tical

forc

e(N

)

Time (sec)

-60

-40

-20

0

20

40

60

Exclusive chewing jaw movement

Tugging the bite backwards

Tugging the bite forwards

Long

itudi

nalf

orce

(N)

Figure 3. An example of the output from a force plate apparatus for a series of 15 bites (from Griffiths

and Gordon, unpublished).

would adjust bite area to maintain a con-

stant bite force in the face of increasing

strength or tiller density.

There are comparatively few reports

documenting tiller, or leaf and stem

strength, and tiller density on bite area. De-

spite this, there appears to be general con-

sistency across studies in that the rate of

reduction in bite area is much lower than the

rate of increase in tiller density or tiller

strength [39, 48, 53, 68]. This suggests that

there is only a partial adjustment in bite area

in relation to sward structure, and further-

more, the adjustment has little effect in re-

ducing the energy required to sever a bite

[79]. This partial compensation in bite area

also explains, at least in part, the absence of

significant patterns in sweeping tongue

movements in cattle in response to decreased

lamina length across swards of different

structure [25]. Mitchell [53] demonstrated,

using pooled data from sheep and red deer

(Cervus elaphus), that bite area was only

35% greater on wheat (Triticum aestivum)

swards than ryegrass (Lolium perenne)

swards of identical height, despite the

ryegrass swards exhibiting 110% stronger

tissues (measured by shear). The reduction

of bite area in an attempt to maintain the bit-

ing force momentum can thus be consid-

ered to be a small effect. This inference is

supported by the few direct evaluations of

biting force. Hughes and colleagues [39,

40], using boxed swards anchored to a force

plate, found no supporting evidence for a

maximum force that goats or sheep exerted

across a range of sward arrangements

within immature swards and across sward

species. Similarly, cattle grazing from

hand-constructed swards mounted on a

force plate were not observed to exhibit a

maximum force per bite [46]. Furthermore,

Illius et al. [42], in a study with goats, found

appreciable variation in the indirect esti-

mates of bite force, derived by substituting

the grazed heights into polynomial regres-

sion equations summarising force-canopy

structure relationships, across broad- and

fine-leaved plant species. More recently,

Tharmaraj [68] indirectly estimated bite

forces using a field-operated bite force me-

ter and found no indication of a maximum

force that cattle exerted across a range of

plant species and tiller densities. The ab-

sence of a maximum bite force that animals

exert, irrespective of how the bite dimen-

sions are balanced in the face of tiller struc-

tural rigidity and bulk density constraints,

does not, therefore, lend support to the

Summit Force hypothesis. This is also

consistent with the fact that not all leaves

within the bite are severed simulta-

neously but rather in close succession

(W.M. Griffiths and I.J. Gordon, unpub-

lished data). In practical terms obtaining a

measure of the peak bite fracture force is of

little importance, and it is the average force

applied for a given bite that needs to be as-

sessed. Furthermore, rejection of the Sum-

mit Force hypothesis is supported by the

consistent findings that, across a wide

range of body sizes, animals do not always

fully exploit the available depth of regrowth

[sheep and guanacos (Lama guanicoe) [6],

bison (Bison bison) [10], cattle [27, 33]].

6.2. Balancing reward against cost

of bite procurement

A second hypothesis utilising economic

principles described an efficient forager as

one that forages to maximise, or at the very

least, balance the reward against the cost of

bite procurement i.e. force per bite [40].

However, as with the Summit Force hy-

pothesis there does not appear to be any

constancy in the force per bite. Hughes [39]

found that sheep were more willing to in-

crease the force exerted if the force was

compensated by a greater reward, while

Illius et al. [42] found that the expendi-

ture:reward ratio for goats was not consistent

across a range of plant species. More re-

cently, Tharmaraj [68] found a fluctuating


expenditure: reward ratio across short and

tall swards, but a constant ratio when

swards varied in tiller density only. The cal-

culation that the energy gain from increased

penetration will always exceed the energy

cost [42], implies that the greatest gain will

be achieved from severance of material en-

compassed within the buccal cavity at the

base of the sward, in the absence of uproot-

ing. Why then, do cattle and other species,

restrict their grazing depth to some fraction

of sward height, even when the sward is

composed solely of lamina [48]? Further-

more, goats have been found to be insensi-

tive to the differential costs in penetration

depth for broad-leaved and fine-leaved spe-

cies, with the ratio between pseudostem

height and grazed height changing little

across the sward species [42]. It, therefore,

seems likely that the avoidance of a lower

stratum may be unrelated to the greater ef-

fort required by the animal to detach plant

organs relative to bite reward. Rather, it

may reflect the fact that encompassing

pseudostem within the bite slows down bite

formation, and would require additional

jaw movements to chew the more fibrous

material. Therefore, the animal strikes a

balance between severance and chewing

constraints, as discussed in the model of

Spalinger and Hobbs [65] and further eval-

uated by Shipley and Spalinger [63]. The

relationship between bite mass and chew-

ing behaviour has also been ascribed to the

need to strike a balance in the maintenance

of a grazing momentum [49]. However, the

importance attached to striking a balance

between severance and processing may be

over-emphasised since cattle are able to

prehend, and manipulate the contents of ex-

isting bites, within the same jaw movement

[49], effects that were assumed to be mutu-

ally exclusive in Spalinger and Hobbs’model

[65]. The overlap between prehension and

chewing can be, however, incorporated into

models, as shown by Farnsworth and Illius

[24].

6.3. Marginal revenue

Delving deeper into the application of

economic principles, a study by Illius et al.

[42] marked a further attempt to understand

the conceptual basis of bite depth. These

authors demonstrated, using goats and a

group of fine- and broad-leaved grasses,

that depth of penetration into a sward could

be predicted when account was taken of the

marginal revenue i.e. the differential in-

crease in bite force relative to the differen-

tial increase in energy intake rate. Despite

the variation in the predicted bite forces (by

indirect measures) and contrasting grazing

depths, the goats grazed to a common mar-

ginal revenue, although the value of mar-

ginal revenue differed between the fine-

and broad-leaved species. This hypothesis

is both puzzling and plausible. Among the

three fine-leaved species, there was a clear

linkage between the mass of material re-

moved and the force involved in severing

those bites, but the reason for differing

common marginal revenue between the

fine- and broad-leaved species remains un-

clear. It is worth noting that had these au-

thors examined only one plant species from

each of the fine- and broad-leaved catego-

ries, an approach that would have been rea-

sonable, their conclusion would not have

held, emphasising that the marginal reve-

nue hypothesis may not explain penetration

responses across swards of greater structural

complexity. Critical evaluation of the hy-

pothesis across swards of contrasting struc-

ture and animals of different body size

awaits.

To summarise, the evidence to support

the three current hypotheses is weak and

new initiatives are required to shed light on

the determinants of bite mechanics.

7. EFFORT VS. FORCE

Increasingly it has become obvious that

there is no conclusive evidence to support


any of the three key hypotheses that have

been put forward as mechanisms for ob-

served bite depth penetration responses. It

may not be that we need to put forward new

hypotheses, rather we need to discover new

approaches of thinking the problem. Cen-

tral to our understanding of biting force has

been the assumption that the force applied

in severance will be proportional to the

strength of individual plant components, al-

though Illius et al. [42] argued that the

structural arrangement of the plant compo-

nents has the greater bearing on the force

exerted. Nevertheless, the implication is

that severance occurs as a result of muscle

moving against a fixed anchor of body

mass, and the costs of severance for a given

force are then lower for species of greater

body mass, owing to the allometric scaling

of force with body mass (W0.67 [58]; W0.69

[42] using the assumption that the force

generated by muscle is proportional to

muscular cross-sectional area [14]). Leaf

tensile strength has been found to be of

greater importance in determining plant ac-

ceptability for sheep than for cattle [54], a

finding that more likely reflects the con-

straints imposed on the smaller body mass

and thus smaller potential muscular effort,

rather than the suggested shorter retention

time and lower digestive ability of small ru-

minants [54]. An intra-specific allometric

relationship between body mass and graz-

ing depth (M–0.15) has been reported [31,

42], although the relationship was only

present on the tall swards in the study by

Gordon et al. [31]. By contrast Canigano

et al. [15] did not find evidence for an intra-

specific relationship between body mass

and grazing depth in cattle, despite a wide

contrast in body mass (256–608 kg). Fur-

thermore, there has been consistency in bite

depth for inter-species variation in body

mass; cattle vs. sheep [55], sheep vs.

guanacos (Lama guanicoe) [6]. Given these

inconsistencies in the literature on intra-

and inter-specific species depth of penetra-

tion responses we are left wondering about

the real costs involved in grazing [42]. It

might then be helpful to begin by thinking

of the act of severing plant material as a

two-step process; (i) ‘head’ resistance,

which is the force required to accelerate the

mass of the head in a steady momentum and

(ii) ‘herbage’ resistance, the strength and

architectural arrangement of the plant or-

gans in 3D-space, with the summation of

these two resistances shaping the biting ef-

fort ([53], J. Hodgson personal communi-

cation). This provides a clear distinction

between biting effort and biting force.

Further, the laws of motion state that for

a given force, the rate of acceleration will

be greater for a small object than for a large

object. Chambers et al. [16] reported the

rate of head acceleration as some 65%

greater for sheep than for cattle grazing the

same sward. There is little quantification of

the effort and associated costs incurred in

upholding the tension in the muscles of the

neck to maintain the head momentum, al-

though Illius et al. [42] comment that it

must be considerable. The contribution of

the ‘head’ resistance to maintain a cyclic

momentum for the applied effort can be

predicted to be of some significance. This is

despite the relatively small energy cost as-

sociated with bite severance as a result of

the small displacement, although the dis-

placement will increase with increasing

number, and height, of tillers captured

within the bite. If tiller length is viewed as a

moment arm, the distance and angle per-

pendicular to the tiller, from the tiller’s

rooted position to that at severance will in-

crease with sward height and/or tensile

strength, implying that displacement and

acceleration are important components of

biting effort. In support, Chambers et al.

[16] found that the rate of head acceleration

of sheep decreased with increasing sward

height. However, these authors did not ob-

serve a similar pattern for cattle.

The allometric relationship between

force and body mass would provide the

generality that researchers seek, but as biol-

ogists and ecologists we need to recognise


the existence of many exceptions. In the

only known evaluation of biting force

across species body mass, Hughes [39]

showed that goats (23.5 kg W) did not fully

utilise their body mass (W0.37, estimated

from Tab. 5.12 and text from p. 82 [39]) to

sever bites from 15 cm perennial ryegrass

swards, with bite depth substantially

smaller than that recorded for sheep (64 kg W)

(W0.64, estimated from Tab. 5.12 and text

from p. 82 [39]). Other studies have also ob-

served goats to exhibit a shallow penetra-

tion depth on vegetative swards [26],

suggesting that goats may not exploit the

force stored in their muscles for

prehension. This suggests that constraints

imposed by mandibular length may be a

further critical parameter that needs to be

corrected for in assessing biting effort.

There is ample evidence that incisor

breadth scales with body mass [e.g. 29, 31],

and that total masseter weight, which re-

flects masseter tissue size, is correlated

with body mass [4]. However, whilst man-

dibular length does scale with body mass,

there appears to be no compensation be-

tween mandibular length and masseter

muscle mass in ruminants, unlike in some

carnivores [44]. By inference a lack of cor-

relation between mandibular length and

masseter muscle mass implies a weaker

gripping force at the incisors (see also [64])

leading to increased tiller slippage during

initial clamping of herbage as well as dur-

ing the jerking motion of the head. This im-

plies that, irrespective of body mass,

animals with longer mandibles are likely to

incur substantially reduced bite masses for

the applied bite effort with increasing

canopy depth. Placing this into perspec-

tive, goats have a higher mandibular

length:body mass ratio than sheep (F.J.

Pérez-Barbería and I.J. Gordon, unpub-

lished data), and this may account for the

observed responses by goats in the study by

Hughes [39]. These additional animal-

based constraints strengthen the argument

that it is biting effort rather than biting force

that needs to be evaluated, and therefore

providing a ‘real’ measure of the con-

straints in handling forages rather than just

a measure of the force required to fracture a

given plant organ.

8. CONCLUSION

Given the importance of bite depth as a

contributor to bite volume and intake rate,

there remains considerable scope for un-

derstanding the mechanistic basis for bite

depth regulation. Increasing evidence

points towards responses in bite depth to

the vertical positioning, and maturation, of

the morphological organs within the sward

canopy. Assessment of biting effort would

represent a more holistic approach to un-

derstanding the animal’s response to sward

structural complexity, and might explain

why the adjustments in bite geometry have

been of a smaller magnitude than the

change in herbage resistance. More explicit

arguments have been hampered by the lim-

ited available bite force data, much of

which has been collected from swards ex-

hibiting relatively limited variation in

structural strength, and from comparisons

across plant species where other physical

and chemical characteristics may also vary

[40, 42, 46, 68]. Despite the obvious tech-

nological difficulties in quantifying the var-

ious components of biting effort that have

been described in this paper, progress on

this front will remain essential to aid the de-

velopment of predictive foraging models.

ACKNOWLEDGMENTS

We thank two anonymous referees for

comments on the manuscript. This contribu-

tion was funded by a Foundation for Research

Science and Technology New Zealand, Grant

no: MLUR001 to WMG, and funding from the

Macaulay Development Trust and the Scottish

Executive Environment and Rural Affairs De-

partment.


REFERENCES

[1] Akin D.E., Burdick D., Percentage of tissue

types in tropical and temperate grass leaf blades

and degradation of tissues by rumen microor-

ganisms, Crop Sci. 15 (1975) 661–668.

[2] Arias J.E., Dougherty C.T., Bradley N.W.,

Cornelius P.L., Lauriault L.M., Structure of Tall

Fescue swards and intake of grazing cattle,

Agron. J. 82 (1990) 545–548.

[3] Astigarraga L., Peyraud J.L., Effects of sward

structure upon herbage intake by grazing dairy

cows, Ann. Zootech. 44 (1995) 126.

[4] Axmacher H., Hofmann R.R., Morphological

characteristics of the masseter muscle of 22 ru-

minant species, J. Zool. (Lond.) 215 (1988)

463–473.

[5] Baker S., Klein L., de Boer E.S., Purser D.B.,

Genotypes of dry, mature subterranean clover

differ in shear energy, Proceedings of XVII In-

ternational Grassland Congress, New Zealand,

1993, pp. 592–593.

[6] Bakker M.L., Gordon I.J., Milne J.A., Effects of

sward structure on the diet selected by guanacos

(Lama guanicoe) and sheep (Ovis aries) grazing

a perennial ryegrass-dominated sward, Grass

For. Sci. 53 (1998) 19-30.

[7] Bailey D.W., Dumont B., WallisDeVries M.F.,

Utilization of heterogeneous grasslands by do-

mestic herbivores: Theory to management, Ann.

Zootech. 47 (1998) 321–333.

[8] Barthram G.T., Grant S.A., Defoliation of

ryegrass-dominated swards by sheep, Grass For.

Sci. 39 (1984) 211–219.

[9] Baumont R., Prache S., Meuret M., Morand-

Fehr P., How forage characteristics influence be-

haviour and intake in small ruminants: a review,

Livest. Prod. Sci. 64 (2000) 15–28.

[10] Bergman C.M., Fryxell J.M., Gates C.C., The ef-

fect of tissue complexity and sward height on the

functional response of Wood Bison, Funct. Ecol.

14 (2000) 61–69.

[11] Briske D.D., Strategies of plant survival in

grazed systems: a functional interpretation, in:

Hodgson J., Illius A.W. (Eds.), The Ecology and

Management of Grazing Systems, CAB Interna-

tional, Wallingford, UK, 1996, pp. 37–67.

[12] Burlison A.J., Hodgson J., Illius A.W., Sward

canopy structure and the bite dimensions and

bite weight of grazing sheep, Grass For. Sci. 46

(1991) 29–38.

[13] Burns J.C., Pond K.R., Fisher D.S., Measure-

ment of Forage Intake, in: Fahey G.C. Jr., Moser

L.E., Mertens D.R., Collins M. (Eds.), Proceed-

ings of the National Conference on Forage Qual-

ity, Evaluation, and Utilisation, University of

Nebraska, 1994, pp. 494–532.

[14] Cachel S., Growth and allometry in primate

masticatory muscles, Arch. Oral Biol. 29 (1984)

287–293.

[15] Cangiano C.A., Galli J.R., Pece M.A., Dichio L.,

Rozsypalek S.H., Effect of liveweight and pas-

ture height on cattle bite dimensions during pro-

gressive defoliation, Aust. J. Agric. Res. 53

(2002) 541–549.

[16] Chambers A.R.M., Hodgson J., Milne J.A., The

development and use of equipment for the auto-

matic recording of ingestive behaviour in sheep

and cattle, Grass For. Sci. 36 (1981) 97–105.

[17] Choong M.F., What makes a leaf tough and how

this effects the pattern of Castanopsis fissa leaf

consumption by caterpillars, Funct. Ecol. 10

(1996) 668–674.

[18] Choong M.F., Lucas P.W., Ong J.S.Y., Pereira B.,

Tan H.T.W., Turner I.M., Leaf fracture tough-

ness and sclerophyll: their correlations and eco-

logical implications, New Phytol. 121 (1992)

597–610.

[19] Demment M.W., Peyraud J.L., Laca E.A., Herb-

age intake at grazing: a modelling approach. In

Recent developments in the nutrition of herbi-

vores, in: Journet M., Grenet E., Farce M.H.,

Theriez M., Demarquilly C. (Eds.), Proceedings

of the IVth International Symposium on the Nu-

trition of Herbivores, INRA Editions, Paris,

1995, pp. 121–141.

[20] Easton H.S., Variability of leaf shear strength in

perennial ryegrass, N.Z. J. Agric. Res. 32 (1989)

1–6.

[21] Edwards G.R., Parsons A.J., Penning P.D.,

Newman J.A., Relationship between vegetation

state and bite dimensions of sheep grazing con-

trasting plant species and its implications for in-

take rate and diet selection, Grass For. Sci. 50

(1995) 378–388.

[22] Evans P.S., A study of leaf strength in four

ryegrass varieties, N.Z. J. Agric. Res. 7 (1964)

508–513.

[23] Evans P.S., Leaf strength studies of pasture

grasses: I. Apparatus, techniques and some fac-

tors affecting leaf strength, J. Agric. Sci. (Camb.)

69 (1967) 171–174.

[24] Farnsworth K.D., Illius A.W., Large grazers

back in the fold: generalizing the prey model to

incorporate mammalian herbivores, Funct. Ecol.

10 (1996) 678–680.

[25] Flores E.R., Laca E.A., Griggs T.C., Demment

M.W., Sward height and vertical morphological

differentiation determine cattle bite dimensions,

Agron. J. 85 (1993) 527–532.

[26] Fraser M.D., Gordon I.J., The diet of goats, red

deer and South American camelids feeding on

three contrasting Scottish upland vegetation

communities, J. Appl. Ecol. 34 (1997) 668-686.

[27] Ginnett T.F., Dankosky J.A., Deo G., Demment

M.W., Patch depression in grazers: the roles of


biomass distribution and residual stems, Funct.

Ecol. 13 (1999) 37-44.

[28] Gong Y., Hodgson J., Lambert M.G., Gordon

I.L., Short-term ingestive behaviour of sheep

and goats grazing grasses and legumes. 1. Com-

parison of bite weight, bite rate, and bite dimen-

sions for forages at two stages of maturity, N.Z.

J. Agric. Res. 39 (1996) 63–73.

[29] Gordon I.J., Illius A.W., Incisor arcade structure

and diet selection in ruminants, Funct. Ecol. 2

(1988) 15–22.

[30] Gordon I.J., Lascano C., Foraging strategies of

ruminant livestock on intensively managed

grasslands: potential and constraints, in: Pro-

ceedings of XVII International Grassland Con-

gress, New Zealand, 1993, pp. 681–687.

[31] Gordon I.J., Illius A.W., Milne J.D., Sources of

variation in the foraging efficiency of grazing ru-

minants, Funct. Ecol. 10 (1996) 219–226.

[32] Greenberg A.R., Mehling A., Lee M., Bock J.H.,

Tensile behaviour of grass, J. Mater. Sci. 24

(1989) 2549–2554.

[33] Griffiths W.M., Hodgson J., Arnold G.C., The in-

fluence of sward canopy structure on foraging

decisions by grazing cattle. II. Regulation of bite

depth penetration, Grass For. Sci., submitted.

[34] Halyk R.M., Hurlbut L.W., Tensile and shear

strength characteristics of Alfalfa stems, Trans.

ASAE 11 (1968) 256–257.

[35] Henry D.A., Macmillan R.H., Simpson R.J.,

Measurement of the shear and tensile fracture

properties of leaves of pasture grasses, Aust. J.

Agric. Res. 47 (1996) 587–603.

[36] Henry D.A., Simpson R.J., Macmillan R.H.,

Seasonal changes and the effect of temperature

and leaf moisture content on intrinsic shear

strength of leaves of pasture grasses, Aust. J.

Agric. Res. 51 (2000) 823–831.

[37] Hodgson J., The control of herbage intake in the

grazing ruminant, Proc. Nutr. Soc. 44 (1985)

339–346.

[38] Hodgson J., Clark D.A., Mitchell R.J., Foraging

behaviour in grazing animals and its impact on

plant communities, in: Fahey G.C. Jr., Moser

L.E., Mertens D.R., Collins M. (Eds.), Proceed-

ings of the National Conference on Forage Qual-

ity, Evaluation, and Utilisation, University of

Nebraska, 1994, pp.796–827.

[39] Hughes T.P., Sward structure and intake of rumi-

nants, Ph.D. thesis, Lincoln University, New

Zealand, 1990.

[40] Hughes T.P., Sykes A.R., Poppi D.P., Hodgson

J., The influence of sward structure on peak bite

force and bite weight in sheep, Proc. N.Z. Soc.

Anim. Prod. 51 (1991) 153–158.

[41] Illius A.W., Hodgson J., Progress in understand-

ing the ecology and management of grazing sys-

tems, in: Hodgson J., Illius A.W. (Eds.), The

Ecology and Management of Grazing Systems,

CAB International, Wallingford, UK, 1996,

pp. 429–457.

[42] Illius A.W., Gordon I.J., Milne J.D., Wright W.,

Costs and benefits of foraging on grasses varying

in canopy structure and resistance to defoliation,

Funct. Ecol. 9 (1995) 894–903.

[43] Inoue T., Brookes I.M., John A., Hunt W.F.,

Barry T.N., Effects of leaf shear breaking load on

the feeding value of perennial ryegrass (Lolium

perenne) for sheep. I. Effects on leaf anatomy

and morphology, J. Agric. Sci. (Camb.) 123

(1994) 129–136.

[44] Jaslow C.R., Morphology and digestive effi-

ciency of red foxes (Vulpes vulpes) and grey

foxes (Urocyon cinereoargenteus) in relation to

diet, Can. J. Zool. 65 (1987) 72–79.

[45] Laca E.A., Demment M.W., Foraging strategies

of grazing animals, in: Hodgson J., Illius A.W.

(Eds.), The Ecology and Management of

Grazing Systems, CAB International,

Wallingford, UK, 1996, pp. 137–158.

[46] Laca E.A., Demment M.W., Distel R.A., Griggs

T.C., A conceptual model to explain variation in

ingestive behaviour within a feeding patch, Pro-

ceedings of XVII International Grassland Con-

gress, New Zealand, 1993, pp. 710–712.

[47] Laca E.A., Ungar E.D., Seligman N., Ramey

M.R., Demment M.W., An integrated methodol-

ogy for studying short-term grazing behaviour of

cattle, Grass For. Sci. 47 (1992) 81–90.

[48] Laca E.A., Ungar E.D., Seligman N., Demment

M.W., Effect of sward height and bulk density on

bite dimensions of cattle grazing homogeneous

swards, Grass For. Sci. 47 (1992) 91–102.

[49] Laca E.A., Ungar E.D., Demment M.W., Mecha-

nisms of handling time and intake rate of a large

mammalian grazer, Appl. Anim. Behav. Sci. 39

(1994) 3–19.

[50] Launchbaugh K.L., Biochemical aspects of

grazing behaviour, in: Hodgson J., Illius A.W.

(Eds.), The Ecology and Management of

Grazing Systems, CAB International,


[51] McRandal D.M., McNulty P.B., Mechanical and

physical properties of grasses, Trans. ASAE 23

(1980) 816–821.

[52] Mackinnon B.W., Easton H.S., Barry T.N.,

Sedcole J.R., The effect of reduced leaf shear

strength on the nutritive value of perennial

ryegrass, J. Agric. Sci. (Camb.) 111 (1988)

469–474.

[53] Mitchell R.J., The effects of sward height, bulk

density and tiller structure on the ingestive be-

haviour of red deer and romney sheep, Ph.D. the-

sis, Massey University, New Zealand, 1995.

[54] O’Reagain P.J., Plant structure and the accept-

ability of different grasses to sheep, J. Range

Manage. 46 (1993) 232–246.

[55] Orr R.J., Harvey A., Kelly C.L., Penning P.D.,

Bite dimensions and grazing severity for cattle


and sheep, Proceedings of the Vth Research

Conference, BGS, Seale Hayne Faculty of Agri-

culture, Food and land Use, University of Plym-

outh, 8-10 September 1997, pp.185–186.

[56] Pérez-Barbería F.J., Gordon I.J., Factors affect-

ing food comminution during chewing in rumi-

nants: a review, Biol. J. Linn. Soc. 63 (1998)

233–256.

[57] Pérez-Barbería F.J., Gordon I.J., The functional

relationship between feeding type and jaw and

cranial morphology in ungulates, Oecologia 118

(1999) 157-165.

[58] Peters R.H., The Ecological Implications of

Body Size, Cambridge University Press, Cam-

bridge, 1983.

[59] Prache S., Peyraud J.J., Foraging behaviour and

intake in temperate cultivated grasslands, Pro-

ceedings of XIX International Grassland Con-

gress, Brazil, 2001, pp. 309–319.

[60] Provenza F.D., Villalva J.J., Cheney C.D.,

Werner S.J., Self-organization of foraging be-

haviour: from simplicity to complexity without

goals, Nutr. Res. Rev. 11 (1998) 199–222.

[61] Ruyle G.B., Hasson O., Rice R.W., The influ-

ence of residual stems on biting rates of cattle

grazing Eragrostis lehmanniana Nees, Appl.

Anim. Behav. Sci. 19 (1987) 11–17.

[62] Scheirs J., De Bruyn L., Verhagen R., A test of

the C3-C4 hypothesis with two grass miners,

Ecology 82 (2001) 419–421.

[63] Shipley L.A., Spalinger D.E., Mechanics of

browsing in dense food patches: effects of plant

and animal morphology on intake rate, Can. J.

Zool. (1992) 1743–1752.

[64] Shipley L.A., Gross J.E., Spalinger D.E., Hobbs

N.T., Wunder B.A., The scaling of intake rate in

mammalian herbivores, Am. Nat. 143 (1994)

1055–1082.

[65] Spalinger D.E., Hobbs N.T., Mechanisms of for-

aging in mammalian herbivores: new models of

functional response, Am. Nat. 140 (1992)

325–348.

[66] Spalinger D.E., Hanley T.A., Robbins C.T.,

Analysis of the functional response in foraging

in the Sitka Black-Tailed deer, Ecology 69

(1988) 1166–1175.

[67] Stobbs J.H., The effect of plant structure on the

intake of tropical pastures. II. Differences in

sward structure, nutritive value, and bite size of

animals grazing Setaria anceps and Chloris

gayana at various stages of growth, Aust. J.

Agric. Res. 24 (1973) 821–829.

[68] Tharmaraj J., Effects of resistance to prehension

and structure of pastures on grazing behaviour

and intake of dairy cows, Ph.D. thesis, Univer-

sity of Melbourne, Australia, 2000.

[69] Theron E.P., Booysen P. de V., The tensile

strength properties of ten indigenous grasses,

Proc. Grassl. Soc. South Afr. 3 (1968) 57–61.

[70] Ulyatt M.J., The feeding value of herbage, in:

Butler G.W., Bailey R.W. (Eds.), Chemistry and

Biochemistry of Herbage, Volume 3, Academic

Press, London, UK, 1973, pp. 131–178.

[71] Vincent J.F.V., The mechanical design of grass,

J. Mater. Sci. 17 (1982) 856–860.

[72] Vincent J.F.V., The influence of water content on

the stiffness and fracture properties of grass

leaves, Grass For. Sci. 38 (1983) 107–114.

[73] Vincent J.F.V., Strength and fracture of grasses,

J. Mater. Sci. 26 (1991) 1947–1950.

[74] Wade M.H., Carvalho P.C. de F., Defoliation pat-

terns and herbage intake on pastures, in: Lemaire

G., Hodgson J., de Moraes A., Carvalho P.C. de F.,

Nabinger C. (Eds.), Grassland Ecophysiology

and Grazing Ecology, CAB International,


[75] Webb N.G., Clark M., Livestock footfloor inter-

actions measured by force and pressure plate,

Farm Building Prog. 66 (1981) 23–36.

[76] Westfall R.H., van Staden J.M., Britz P.J.,

Pretorius, C., The development of a leaf

tensilmeter for in situ measurement of leaf ten-

sile strength, J. Grassl. Soc. South Afr. 9 (1992)

50–51.

[77] Wilson J.R., An interdisciplinary approach for

increasing yield and improving quality of for-

ages, Proceedings of XVth International Grass-

land Congress, Kyoto, Japan, 1985, pp. 49–55.

[78] Wilson J.R., Akin D.E., McLeod M.N., Minson

D.J., Particle size reduction of the leaves of a

tropical and a temperate grass by cattle. II. Rela-

tion of anatomical structure to the process of leaf

breakdown through chewing and digestion,

Grass For. Sci. 44 (1989) 65–75.

[79] Woodward S.J.R., Bite mechanics of cattle and

sheep grazing grass-dominant swards, Appl.

Anim. Behav. Sci. 56 (1998) 203–222.

[80] Wright W., Illius A.W., A comparative study of

the fracture properties of five grasses, Funct.

Ecol. 9 (1995) 269–278.

[81] Wright W., Vincent J.F.V., Herbivory and the me-

chanics of fracture in plants, Biol. Rev. 71 (1996)

401–413.


Review article

Spatial heterogeneity and grazing processes

Anthony J. PARSONSa*, Bertrand DUMONTb

aAgResearch Grasslands, Private Bag 11008, Palmerston North, New ZealandbINRA, Unité de Recherches sur les Herbivores, Theix, France

(Received 19 September 2002; accepted 25 February 2003)

Abstract — Large-scale spatial patterns, whether in the height, density, or species composition of

vegetation, are one of the most demonstrable and widely recognised features of heterogeneity in large

herbivore grazing systems. But to understand how their existence relates to grazing processes, and

what the implications of the patterns are for plants, animals, and for land users, requires adding spatial

concepts, and dynamics to our knowledge of the interactions between plants and animals. Neither ap-

proach has been traditional in agricultural research. In this paper, we provide an overview of what we

propose are some of the key topics and questions that arise in attempts to understand spatial aspects of

the interaction between plant growth (food resources) and animals’behaviour. Rather than review ad-

vances in any one area in detail, we look at some basic principles of the fundamentally different ways

in which animals eating from vegetation (with or without selectivity) affect the components of plant

regrowth; the variance about these; the way this ‘seeds’ the creation and maintenance of heterogene-

ity, and most important the outcome (intake) for the animals. Likewise we outline some basic features

of animals’ behaviour, given heterogeneous and so spatially distributed food, which includes the ex-

pected rates of encounter; learning and memory; and both the benefits and costs of social interactions

when foraging as a group. In this way we combine knowledge from several disciplines (plant physiol-

ogy; animal science; behavioural ecology and not least from practical agriculture) with a goal of pro-

viding a basis for the development of simple pragmatic means for manipulating a grazed but multi-

purpose landscape to balance diversity, heterogeneity and agricultural performance.

cognitive abilities / foraging costs / optimal grazing / rate of encounter / social behaviour

Résumé — Pâturage et hétérogénéité spatiale. Les motifs d’hétérogénéité à large échelle, en terme

de variabilité de hauteur, de densité ou de composition botanique du couvert, sont l’un des aspects les

plus reconnus de l’hétérogénéité spatiale dans les systèmes pâturés. Afin de comprendre en quoi ils

résultent du processus de pâturage, et ce qu’ils induisent pour l’évolution des couverts, les perfor-

mances animales et les décisions des gestionnaires, il est indispensable de prendre en compte la com-

posante spatiale des dynamiques de végétation et du comportement animal, ce qui représente en soi

une approche innovante. Dans cet article, nous donnons une vue d’ensemble de ce que nous pensons

161

Anim. Res. 52 (2003) 161–179


DOI: 10.1051/animres:2003013


E-mail: [email protected]

être les éléments clés des interactions spatiales entre les herbivores et les couverts pâturés. Nous dé-

clinons comment différents modes de prélèvement de la végétation (sélectifs ou non) affectent la

croissance des repousses, leur variabilité spatiale, les motifs d’hétérogénéité à large échelle, et les

performances des animaux qui exploitent ces couverts. Ce faisant, nous soulignons les processus mis

en jeu lorsque des herbivores exploitent l’hétérogénéité spatiale des couverts, en particulier le rôle

des probabilités de rencontre avec les items alimentaires préférés, les mécanismes comportementaux

d’apprentissage et de mémorisation, ainsi que les coûts et les bénéfices de la sociabilité sur les choix

individuels. Ceci nous amène à faire appel à des compétences disciplinaires variées, relevant de l’éco-

physiologie végétale, de la nutrition et du comportement animal, et de l’écologie comportementale,

ainsi qu’à des compétences appliquées en agronomie et élevage, afin de proposer des règles de ges-

tion des couverts prairiaux dans un contexte de multifonctionnalité de leurs usages.

aptitudes cognitives / comportement social / coût de sélection / pâturage optimal / probabilité de

rencontre

1. INTRODUCTION

There are numerous sources of hetero-

geneity in vegetation, to the extent it is

probably more difficult to explain how veg-

etation can ever be homogeneous, than it is

to accept its spatial complexity [27]. There

may, for example, be an intrinsic heteroge-

neity in the soil or other resources and this

may or may not interact with the distribu-

tion and propagation of plant species, and

the re-distribution of resources by animals

[32]. Each of these topics deserves ongoing

review. But what we address here are some

specific features of the creation and mainte-

nance of heterogeneity – those induced by

decisions made by herbivores about where

and when to place their bites.

Even within these confines, a study of

grazing is unavoidably a study of a complex

system. It is spatially heterogeneous as ani-

mals both create, and respond to, variations

in vegetation state. Most important, it is a

dynamic system, as defoliation impacts on

the rate of replacement of the vegetation. At

least one of the major processes, biting, is

discrete (rather than continuous) and sto-

chastic (as there is uncertainty as to where

and when animals may place their bites)

[44]. There are numerous non-linearities in

the biological responses. It is recognized

that such features generate spatial and tem-

poral complexity at a number of scales, and

that such a system would be expected to ex-

hibit a potentially bewildering array of phe-

nomena, even without the vagaries of

animal behaviour [33]. In physical sci-

ences, one approach to understanding such

a system would be to develop a body of the-

ory – a framework – for how we would

expect the system to behave under a suc-

cession of carefully defined circumstances,

against which we could overlay the poten-

tially complex phenomena observed in

experiments and in practice, in which nu-

merous factors may interact, and so allude

to interpretations of how these phenomena

arise. This is what we attempt here. In all

cases we work with grazing seen as a spatial

and dynamic system, but we start with some

conceptually simple examples and system-

atically add in factors, notably those for

which information from controlled experi-

ments is available.

Complex phenomena can arise in part

from simple rules operating locally and at a

fine scale. We look first at what is known

about the processes operating at the fine

(bite) scale and consider how far we can go

in explaining the origin of phenomena at

larger scales, without at this stage evoking

any larger scale behaviours.

162 A.J. Parsons, B. Dumont

Next, we review evidence for larger

scale (e.g. social, flock/herd) foraging be-

haviours, and then discuss how these may

add three major aspects to the overall un-

derstanding. First we consider information

gathering (sensory knowledge and spatial

memory) and discuss its adaptive value ac-

cording to environment complexity. Sec-

ond we consider social behaviours and how

these may both help and hinder foraging

success in individuals. Here we explore the

probability that the very ‘rules’for foraging

(notably preference) which are the driving

processes for the fine patch scale impacts,

may change at different scales as animals

are seen to respond to the allocation of

space, and/or their proximity to group

members. Third, we consider the way ani-

mals moving around the resources as an ag-

gregated group, might alter their success in

approaching the maximum marginal value

for the rate of supply of resources by modi-

fying where and when they distribute their

defoliations. We parallel this with accounts

derived from different rotational grazing

scenarios in agriculture.

While we recognise the continuum of

these impacts across the scales, we proceed

systematically in an attempt to unravel the

complexity of the phenomena that emerge

in grazing systems. Our aim is to focus at-

tention on those aspects of heterogeneity

that might be manipulated or controlled

with a view to increasing the sustainability

of grasslands for alternative goals.

2. HOW FAR CAN FINE SCALE

INTERACTIONS EXPLAIN

THE ORIGIN OF PHENOMENA

AT LARGER SCALES?

2.1. Sequential (deterministic)

vs. random biting

First we use a previously published

model [39, 44] to consider the impact of some

fundamentally different ways in which animals

might place their bites in space and time. In

the first case we consider what would be the

outcome if animals were to take bites in a

strict sequence such that no bite sized

‘patch’ was revisited before all other

patches had been eaten from. Note this

would give rise to the situation where ani-

mals were at all times eating from the next

largest patch in the vegetation. We contrast

this with the case where we imagine ani-

mals may take bites totally at random. In all

cases, we perform a full mechanistic analy-

sis of the constraints to searching for and

handling food [37, 46], such that the state of

each vegetation patch determines bite

mass, and the animals face the time costs of

prehending and masticating each bite,

which in turn affects the rate at which ani-

mals progress through the array of patches.

But in these first ‘simplest’ cases, we can

consider foraging with the minimal in-

volvement of any costs of searching, as if

the next bite were adjacent to the last.

The outcome of even these simple forag-

ing examples (Fig. 1a) is revealing, as it

emphasises the importance of the dynamics

of vegetation (resource replacement) and

the impact of foraging on the components

of this. At low stocking rates, intake would

be seen to be unaffected by grazing method,

as intake is not constrained by vegetation

state and each animal eats its ‘fill’. But at

high stocking rates, animals (each and all)

would appear to do better grazing spatially

and temporally at random. The explanation

is that stochasticity gives the vegetation a

chance. Where the animals graze patches in

strict sequence, each patch suffers the same

defoliation (which becomes severe at high

stocking rates as animals progressively re-

duce the overall vegetation state), and each

patch is defoliated at the same (frequent)

deterministic interval [38]. There is spatial

heterogeneity in vegetation state, simply

because animals can only bite from a small

proportion of the total available patches in

any one day, but there is no variance about

the determinants of regrowth – the residual

Spatial heterogeneity and grazing processes 163

patch state or regrowth duration. Even

though animals would always be eating the

next largest patch in the sward, a situation

some have proposed as being optimal for

intake [51], intake is low as there is little

vegetation, so patches and bites are small.

By contrast, under random grazing, some

patches would escape grazing by sheer

chance (there is now not only spatial heter-

ogeneity but also variance in residual patch

states and in regrowth duration). This gives

the opportunity for more growth in those

patches. Intake is greater purely because

the system comes to equilibrium with a

greater mean vegetation state (and so bites

are larger).

Both these simple systems generate spa-

tial heterogeneity in the sense that there

would be a frequency distribution of patch

states at the fine (bite) scale [38]. Spatially

explicit accounts of these methods for plac-

ing bites in space and time reveal that larger

scale patterns can arise purely by chance

under even spatially random grazing,

whether it is portrayed as a Poisson process

or if the animals are constrained to walking

a random path. Such larger scale patterns

are only transient however. But as we shall

see later, these can become the focus for

subsequent larger scale animal behaviours,

and so could contribute to sustained vegeta-

tion patterns.

2.2. Adding preference at the fine scale

We now add to this framework what

would be the effects of grazing with prefer-

ence. Here we still consider the case where

animals might encounter potential bites

spatially and temporally at random, but

now make local, state-dependent decisions

whether to eat from the bite sized patches

they encounter. There are numerous criteria

by which animals might prefer one vegeta-

tion patch over another, but here we focus

on size (density, mass or height). This high-

lights how, although preference is simple to

conceive in a static system, it is more difficult


cons

umpt

ion

(kg/

ha/d

)

0

10

20

30

40

50

60

70

2 4 6 8 10 12 140

stocking rate

cons

umpt

ion

(kg/

ha/d

)

0

10

20

30

40

50

60

70

2 4 6 8 10 12 140 2 4 6 8 10 12 140 2 4 6 8 10 12 140

stocking rate

2 4 6 8 10 12 140 2 4 6 8 10 12 140

(a) (b)

optimal solution

Figure 1. The effect of some fundamentally contrasting ways of placing bites in space and time on

the yield (intake per ha) of the grazed system across a wide range of stocking rates. In (a) bites were

grazed at random (solid line); in strict sequence (dashed line), or encountered at random but grazed

preferentially with an 80% or 95% (lower line) partial rejection of tall patches (dotted lines). In (b)

animals grazed as an enforced group (a 30 day regrowth: 1 day grazing rotation) which suppressed

preference at high stocking rates (solid line), as compared with continuous grazing (dotted line).

According to [44].

to conceive dynamically. We have seen how

grazing animals create a frequency distri-

bution of patch states. All patches in this

size-structured population will endeavour

to regrow (move to the right in Fig. 2a). But

preference rules filter this flow. Patches that

are eaten will in effect be ‘thrown back’

down the size structure, to contribute to the

small categories of the frequency distribu-

tion (Fig. 2b). Of the many simple state-

based preference rules we tested [38], one

in particular was seen to have a very delete-

rious impact – a partial rejection of ‘tall’

(relative preference for short). It was shown

[44] with a simple dynamic model of this

mechanism, that where there is a high prob-

ability of tall patches being eaten, a skewed

but unimodal frequency distribution of

patch states results (Fig. 2c). However,

where there is a low probability of tall

patches being eaten (a stronger stochastic

rejection of tall) a bimodal frequency distri-

bution of patch states emerges (Fig. 2d).

This form of patch selection can lead dy-

namically to considerable reductions in in-

take, and so foraging success (dotted lines

Fig. 1a). When animals partially reject tall,

the mean state of the vegetation (even at

high stocking rates) is larger, but intake by

animals is reduced as they eat preferentially

from the population of smaller patches. In-

take is therefore reduced even at low stock-

ing rates, where it would otherwise not have

been constrained by vegetation. Bi-modal

frequency distributions have been widely

reported in field experiments [23].


increasing patch size (height)

(d) low probability of defoliation of tall patches

increasing patch size (height)

(c) high probability of defoliation of tall patches

increasing patch size (height) increasing patch size (height)

(b) but preference ‘rules’ filter this flow(a) all patches attempt to grow

Figure 2. Diagram to convey the impact in a dynamic context of stochastic preference ‘rules’(here a

partial rejection of tall) on the frequency distribution of patch states, and the possibility of the emer-

gence of sustained markedly bi-modal distributions that could ‘seed’larger scale patterns in heteroge-

neity. According to [38], using the same spatial model [44] as in Figure 1.

A bimodal frequency distribution of bite

scale patch states will clearly be a powerful

force to ‘seed’ spatial patterns at larger

scales should there be other state-depend-

ent processes that cause the subsets (e.g. tall

and short) of the frequency distribution to

become aggregated in space [5, 32].

2.3. Moving up a scale: adding foraging

(search) costs

So far, we have considered the dynamic

consequences of how animals place their

bites in space and time, without considering

the additional time costs of searching (to be

precise, where search costs are less than

handling costs and these are deemed to

overlap, so there are no additional search

costs for foraging) [37, 44].

Grazing selectively can, however, sub-

stantially increase the costs of foraging ([48]

and Fig. 3). Costs increase as animals pass

by less desirable items (‘lost opportunity’),

in effect travelling further per unit preferred

food. Grazing selectively is shown in Fig-

ure 3 to increase foraging costs in two ways.

First, costs increase substantially with the

degree of selectivity, and second, costs in-

crease substantially if the preferred food is

less abundant in the vegetation. Both are in-

tuitive, though the scale of the increase is

perhaps surprising. Grazing selectively

may clearly create spatial heterogeneity,

both at a fine scale and as a larger scale pat-

tern, but it is important to consider next how

the potential size of foraging costs might

limit animals desire to graze selectively in

the first place.

We can consider the expected feedback

of search costs on animals desire to forage

selectively by modelling the optimal solu-

tions for trading off the benefits of eating a

given diet component, against the costs of

selecting for it, under different circum-

stances (for details see [48]). Optimising

the cost-benefits of foraging, and a desire


Rel

ativ

ein

crea

sein

spec

ific

cost

s

Grass acceptance coeff., Clover acceptance coeff.,

Clover preference coefficient,

Grass rejected Clover rejected

10

5

0

10

5

0

10:90

20:80

50:50

100:0

90:10

80:20

50:50

0:100

0 0.5 1

0.5

0

10 5 2 1 0.8

0.5

0.2 0

�c : g

ac = 1 ag = 1

ag ac

ac / ag

c : g� � �

Figure 3. The calculated relative increase in specific foraging costs (i.e. per unit mass eaten) in a

2 species mixture as affected by the acceptance coefficients (ac, ag) and the relative abundance (λc: λg)

of the two species (assumed here to be clover and grass). From [48].

for a mixed diet [36, 41] predict complex

selective foraging phenomena. Figure 4a

describes how the optimal proportion of

preferred food in the diet should vary with

the proportion (abundance by cover) of the

preferred food in the vegetation. The model

suggests that, if the searching costs are neg-

ligible, the animal would be expected to ex-

tract always its preferred diet (in this case

presumed to be 70% of the preferred food).

If search costs were very high, the animal

would be expected to eat whatever was in

front of it. With intermediate costs, seem-

ingly complex behaviours would be antici-

pated, with animals sustaining their

preferred diet while the composition of the

vegetation is close to that mixture, but in sit-

uations where the abundance of the pre-

ferred food in the vegetation is lower,

animals would be expected to progressively

forego preference. In some cases, e.g. when

the cost structure is very high, then at low

abundance of preferred food, animals

should resort again to eating whatever was

in front of them. Experimental studies of

the effects of the relative abundance of al-

ternative foods, on selectivity, confirm

these predictions in grassland [17].

3. INFORMATION GATHERING:

ANIMAL RESPONSES

TO POTENTIALLY PROHIBITIVE

COSTS

Foraging (search) costs are modified by

the way food is distributed (dispersed or ag-

gregated) and the extent to which animals

are able to exploit this opportunity spa-

tially. It is well established that animals (in-

cluding domestic herbivores) can use

sensory cues (sight or smell) to detect food

items at a distance; can form flexible, even

abstract, associations between food appear-

ance and its value [6, 19] and so learn, and

use spatial memory [15, 18] to aid foraging.

This, clearly and intuitively, can help them

reduce foraging costs, though it is not a

simple task to analyse to what extent, and

under what circumstances, the benefits pre-

vail.

3.1. ‘Theory’ for how spatial

distribution per se affects rate

of encounter

We can argue from the basis of probabil-

ities alone that, if the distribution of pre-

ferred food is approaching random at a fine

scale (or the distribution is for any reason

cryptic), then for any level of abundance,

the potential search costs for grazing selec-

tively would be greatest. There would be

little animals could do other than to re-

spond, as in the examples above, by making

very local decisions, trading off the benefits

of being selective against the potentially

high costs.

However, food may be distributed in

patches (aggregated). To consider how this

would intrinsically affect animals’expecta-

tion of finding the preferred food, we can

start by imagining a situation where the

same amount of food is distributed in one

case in a large number of small patches,

whereas in a second case, it is distributed in

a small number of correspondingly larger

ones. Let us assume the density of food

within the patches is the same and that there

is therefore the same abundance (by cover)

in the area to be searched. We can consider

what problems the animals face, and their

expected encounter with preferred food,

from the well-established theory behind

vegetation sampling techniques [29].

Totally random point sampling would

clearly give an identical level of success in

finding preferred food in both cases. But

this is not the way ground-based animals

can sample as they are constrained to walk-

ing a path. If animals were to sample only

locally along a path (e.g. as with random

line transects, or a random walk), then

when food is distributed patchily, the ex-

pected rate of encounter with the patches is

far greater where there are more smaller

patches than with fewer larger ones. However,


this would be compensated for in that more

food is found in each larger patch and so the

expected rate of encounter with preferred

food (as opposed to with patches) could be

the same (though patch shape has some im-

pact). There would however be greater vari-

ance about the rate of encounter with food

when it is in fewer larger patches, with lon-

ger periods with no preferred food followed

by longer periods with preferred food.

If animals were to search along a path

but with a wider field of view, and were

drawn toward any preferred food item

within that, they would be sampling in ef-

fect, as if using contiguous ‘quadrats’ and

their success would relate more to the ‘fre-

quency’ and not the ‘cover’ of preferred

items [29]. This sampling approach implies

that for any given level of abundance (by

cover), the probability of encountering a

patch will depend on the size of the field of

view relative to the ‘grain’ of the pattern of

aggregation. There would be an optimal

size field of view (cf. ‘quadrat’ size) which

maximizes the rate of encounter with

patches for each level of aggregation [29].

The probability of encountering patches

would be far greater (in some cases ap-

proaching certainty) where the same

amount of food is distributed in many small

patches.

3.2. Learning and spatial memory

Where the distribution of food items is

not detectable at a distance (e.g. when no

visual or olfactory cues are available for lo-

cating them), learning and spatial memory

become potentially more valuable. To study

the use of learning and spatial memory by

herbivores, and so the impact on animals of

aggregated patterns of food distribution,

experiments and models have been used in

which preferred food is for example offered


prop

ortio

nin

diet

0

0.2

0.4

0.6

0.8

1

0.2

proportion on offer

0

00.2 0.4 0.6 0.8 10.40.6 0.8 1

(a) (b)

low

medium

preferred diet

v.high

t

Figure 4. (a) The complex effects of partial preference, and of low, medium, and high (specific) for-

aging costs on the proportion of preferred food in the diet, in relation to its abundance in the vegeta-

tion, as predicted by a model that seeks the optimal trade-off between the benefits and costs of

foraging selectively [48]. (b) Data from a range of sources show the effect of aggregated (solid sym-

bol) and dispersed (open symbol) spatial distributions of food are consistent with differences in for-

aging costs [17]. It is assumed that the preferred diet is a mixture in (a) of 70% and in (b) of 76% of the

preferred species.

in bowls in an arena [18] or hidden within a

pasture [15 ], or herbage species are planted

in patches mown to look like surrounding

vegetation [16]. The layout of one such ex-

periment is shown in Figure 5a. When naïve

animals are released into this arena, the re-

sults show a marked increase over time in

the success of the animals in finding the

preferred (pellet) food (Fig. 5b), typically

demonstrating the capacity for learning and

spatial memory, as the animals subse-

quently move more directly to and between

the aggregated preferred areas [15]. It is no-

table animals did better, both initially, and

even after learning, when the area to be

exploited was smaller (as in this design, the

overall density of the preferred food was si-

multaneously greater).

Independently manipulating the many

aspects of the complexity of the environ-

ment in which animals forage, in experi-

ments, would be prohibitively exhaustive.

But following this experiment, Dumont and

Hill [14] constructed an individual-based

model to explore the adaptive value of spa-

tial memory in relation to environmental

complexity (e.g. plot size, and consequent

overall density of the preferred food) using

experimental data to calibrate the parameters

of sheep searching behaviour. Comparison of


0

20

40

60

80

100

M-1

00%

M-8

0%

M-6

0%

M-4

0%

M-2

0%M

odel

M+2

0%

M+4

0%

M+6

0%

M+8

0%

M+1

00%

Nb

bow

lsvi

site

dby

each

ewe

in30

'on

day

12

40 x 40m

80 x 80m

120 x 120m

160 x 160m

200 x 200m

240 x 240m

0

10

20

30

40

50

60

70

80

90

1 2 3 4 5 6 7 8 9 10 11 12

Day

Nb.

bow

lsvi

site

dby

each

ewe

in30

'

80 x 80 m

160 x 160 m

large plot small plot

80m

(a) (b)

(c)

Figure 5. (a) Layout of experimental plots with individual bowls shown as dots [15], (b) the increase

in the number of bowls visited by each ewe in 30 min as animals learn the location of bowls, shown

for different plot sizes [15], (c) simulation of the number of bowls visited by each ewe in 30 min after

12 tests according to the animals’memory capacity and plot size [14]. Model outputs were similar to

that of the ewes in (b) (‘Model’), and the predicted effects of increasing (or decreasing) memory size

and memory persistence by 20%, 40%, 60%, 80% and 100% are shown.

the real system behaviour and of model pre-

dictions was successful for both the visual

features of animal movement paths and for

the main model outputs, i.e. the simulated

data were within the 95% confidence inter-

val of real data. Some examples are shown

in Figure 5c, which portrays the effects of

manipulating the memory capacity of the

animals (reducing or increasing by 20%,

40%, 60%, 80% and 100% the memory size

and memory persistence parameters in the

model) within a range of plot sizes.

This confirms, first, that when the area to

be searched is small, animals should be

more successful (both initially and after

several days learning) in finding the pre-

ferred food, but adds that variations in

memory capacity will have relatively small

effect (the number of bowls visited in

30 minutes, after several days learning, dif-

fered little with major changes in memory

capacity). Second, the benefits of spatial

memory were greatest in intermediate-

sized plots, foraging success increasing by

22–25 bowls with memory capacity. In the

largest plot size considered, increasing

memory capacity had less effect (only

11 extra bowls found) suggesting there is an

upper limit to the benefits of using spatial

memory in herbivores.

3.3. Area-concentrated searching

within patches

Whatever means animals use to increase

their chances of finding a food patch, or

even if they encounter one by pure chance,

it is recognized that animals can concen-

trate foraging within a preferred food patch,

once any part of that patch has been found.

This is achieved by increasing the rate of

turning, to remain within the locality [16,

50, 53]. It can be shown easily (e.g. by tak-

ing simple transects across patches drawn

on graph paper), that area-concentrated for-

aging should increase foraging success

many-fold, compared to where the foraging

path within a patch is not altered. More

elegantly, Benhamou [8] determined the

theoretical efficiency of area-concentrated

searching using computer simulation. The

model simulated searching with high sinu-

osity and low speed within high resource

density areas, but low sinuosity and high

speed between these areas. Tested in habi-

tats having the same mean overall density,

the efficiency of this movement control was

higher in coarse-grained (a few large

patches) than in fine-grained habitats (more

smaller patches) and increased also with

intra-patch resource density. Depending on

the habitat, an animal exhibiting optimal

spatial memory-based area-concentrated

searching behaviour was able to harvest

three to five times more food items than if it

did not exhibit any area-concentrated

searching behaviour but moved in a straight

line with an optimal constant speed [9].

3.4. How well do animals exploit

these abilities?

We can now return to the issue of hetero-

geneity and the prospects offered by these

sensory behaviours to reduce foraging

costs. Data from a range of studies, in

which food was distributed in different

ways, and at different abundances, are plot-

ted together in Figure 4b, which shows ani-

mals (large herbivores grazing a range of

vegetations) are more successful in select-

ing preferred food where it is more aggre-

gated [17]. Relating this to the theoretical

responses to different overall costs of forag-

ing (Fig. 4a), the data suggest this is consis-

tent with the aggregated food leading to

lower foraging costs. But the benefits of

foraging in patchy environments seen in

such studies are not always as great, per-

haps, as would be expected. In this context,

we feel it would be of value if more studies

of the foraging success of herbivores in dif-

ferent (patchy) environments, compared

the observed results with some expectation

of success (e.g. the expected encounter

with patches if animals foraged without


knowledge; if they foraged close to ran-

dom; or assessed how the outcome would

be expected to increase from area concen-

trated foraging). Below, we discuss some

possible reasons why, despite proven abili-

ties in learning and spatial memory, ani-

mals may yet, in some cases, exhibit a

foraging success closer to that expected by

random.

Any increase in knowledge about food

distribution offers animals the opportunity

to devise travelling strategies to minimize

the searching time costs of moving between

patches of preferred food. But, even with

total knowledge of the area to be grazed and

of the location of all preferred food patches,

it becomes progressively more difficult to

exploit this knowledge as the number of

these increases. Seeking minimum distance

travel paths between patches poses com-

plex ‘travelling salesman’ problems. De-

tailed analyses of the search paths observed

in experiments are necessary to confirm if

animals are responding to heterogeneity in

an intelligent fashion, or opting to graze

closer to at random.

Another reason for the distribution of

animals eating not matching the distribu-

tion of food at any instant is that in situa-

tions where animals revisit areas of

vegetation frequently, they may instead be

responding more to the spatial patterns in

the rate of replacement of resources. Spa-

tially this alludes to there being an ‘ideal

free distribution’ [22]. Some notable forag-

ing experiments and models test ‘against’

some of the theories for the expected distri-

bution of animals in a dynamic resource en-

vironment [20, 21]. Several of these

emphasise how the expected outcome of

foraging, in a dynamic system, is distinctly

different from that in a static one, and how

foraging may match, or both increase and

decrease, heterogeneity in the vegetation

[1]. But, a third explanation is that social in-

teractions between animals can create moti-

vation conflicts that can reduce (as well as

enhance) foraging success.

4. FORAGING AS A GROUP

AND ITS EFFECTS ON THE

FORAGING OF INDIVIDUALS

Many large herbivore species forage in

groups (flocks or herds), itself a recogni-

tion, in part, of the impacts of aggregating

the distribution of food (themselves, as

prey) on the expectations and foraging

costs of their predators [28]. Herbivores

have been domesticated more to enhance,

rather than remove, the evolved flocking

and associated anti-predatory (e.g. vigi-

lance) behaviours, many being managed

using wolf-like predators. When sheep are

made to graze in only small groups (e.g. be-

low five) grazing time per day decreases

[40] and vigilance postures increase [13].

4.1. Costs and benefits of foraging

within a group

Social behaviours can be both beneficial

and detrimental to individual (and arguably

even group) foraging success. Animals for-

aging within a group benefit from the feed-

ing sites (e.g. preferred patches) discovered

by other members of the group and from

shared vigilance, but they can face the nega-

tive effects of intra-specific competition for

food within the food patches. Even with

perfect knowledge about food distribution,

herbivores always run the risk of returning

to a patch that has been largely depleted by

other animals of the group. Sociability thus

adds uncertainty in animal’s expectations

of patch value and potentially increases for-

aging costs [25]. The way animals would be

expected to optimise the trade-offs between

‘cooperation’ and competition has been

considered for a number of species, using

game theory [34], and social foraging the-

ory has been established around this [24].

Several models have looked at the impact of

these interactions on the expected distribu-

tion of grazing herbivores [7, 14].

Hence, group behaviour can impose mo-

tivation conflicts, as animals from socially


stable groups are very reluctant to graze

away from their peers. This has been shown

in intensive grasslands. Sheep were al-

lowed to forage in a long narrow grass field

which contained an area of preferred food

(taller patch) at a distance of either 15 or

50 m from where, at the end of the field,

there was a small sealed paddock contain-

ing a group of their social peers [13]. At

15 m, the tendency to eat from the preferred

patch was unaffected whether animals

grazed alone or in small groups. But their

results suggest that a sheep is very reluctant

to graze the preferred patch when it is lo-

cated further away (50 m in their experi-

ment, this critical distance varying with

sheep breed) unless it is accompanied by

several other peers, and even then (in

groups as great as 7) time spent grazing the

preferred patch was depressed.

Dumont and Hill [14] modelled that in

an unfamiliar and complex environment,

where the cost of finding preferred food

items is high, the foraging success of an ani-

mal decreased together with the increase in

conspecific attraction within the group, and

this was for two reasons. First, ewes were

frequently attracted by peers feeding on

previously discovered sites and therefore

missed the opportunity to discover new

feeding sites. Second, animals faced the ef-

fects of feeding competition for a preferred

and rare resource on sites. Their efficiency

(in g of pellets consumed per minute of

searching) therefore decreased with in-

creasing social attraction index, while in

the same time the number of bowls visited

per minute increased.

Conversely, when animals are aware of

food location and when food is not limiting,

social bonds within a group can favour

patch and habitat selection. Boissy and

Dumont [11] observed the behaviour of

ewes in their motivation conflict procedure

with animals being familiar (reared to-

gether from the young age) or non familiar to

them. The ewes with familiar companions

more easily split from the paddock contain-

ing a group of peers to graze the preferred

patch located away, vocalised less and were

less vigilant than those with unfamiliar ani-

mals. Differences in the strength of social

bonds within a flock are thus likely to affect

the formation of subgroups and the way

herbivores forage in patchy grasslands.

Similarly, under more extensive conditions,

cattle from the same herd share the use of a

common home range, which is very similar

to that of their mothers [26].

A third way in which social interactions

may reduce foraging success is that compe-

tition with peers can encourage animals to

leave a current patch prematurely. Like-

wise, a remaining animal may leave a re-

warding patch to follow its group mates. It

is well recognized that herbivores should

leave patches before exploiting all the food

these patches contain (this is a central tenet

of the Marginal Value Theorem [12] dis-

cussed later) and there is a theoretical opti-

mal time to leave a patch, which is modified

by the time spent reaching the patch [47].

Social factors also alter patch residence

times [24]. It has been proposed that, ac-

cording to the departure criterion (intake

rate falling below a threshold value or dura-

tion without finding a food item) and to the

number of foragers that are likely to take

this patch leaving decision (anyone or only

a leader decides), living in a group can be

either beneficial, neutral or detrimental to

individual foraging success in this regard

[42]. If grazing herbivores use a departure

criterion based on an intake rate threshold,

it has been proposed that foraging for

patchily distributed food, as a group, would

cause a reduction in food intake rate com-

pared to as a solitary forager [52]. Any ten-

dency for group foragers to leave rewarding

patches prematurely would readily explain

why the foraging success of herbivores

in patchy environments falls short of

what would be expected if they fully ex-

ploited their abilities in area-concentrated

foraging.


4.2. Scaling issues: preferences may

change with scale due to social

interaction

One problem in interpreting the behav-

iour of animals foraging as a group, as a

means to understanding how animal behav-

iour interacts with the vegetation, is that the

phenomena may be dependent on scale

(Fig. 6). At small scales (relative to natural

range and herd size), animals may be com-

fortable and sense their preferred inter-animal

distances are satisfied, while distributing

themselves even randomly across the entire

vegetation area, with consequently widely

distributed impacts on the vegetation pat-

tern. That is, at small scales, the desire for

grouping may have no different impact on

foraging success and vegetation dynamics

than if animals forage independently. At

larger scales, animals satisfying a desire for

the same inter-animal distances would ap-

pear to be aggregated in their distribution

(Fig. 6a). On a homogeneous sward,

Sibbald et al. [45] measured the effects of

space allowance on the grazing behaviour

and spacing of groups of ten Scottish

Blackface ewes. At space allowance from

50 to 133 m2 per head, there were no signifi-

cant differences between mean observed

inter-animal distances and those expected

by chance (i.e. there was no grouping pat-

tern), but observed values were lower than

expected values at 200 m2 per head. In very

heterogeneous environments, herds [30]

and flocks [4] may split into subgroups, ac-

cording to the size and distribution of vege-

tation patches.

What matters more, for the interaction

of animals and vegetation, is whether the

act of forming a group, or questions of

space allocation, alter the very rules, at the

fine scale, by which animals graze. The fact

that animals might move around an area,

foraging as a group, should not be seen as

synonymous with them having an obvious


effic

ienc

y:%

pref

erre

dite

ms

eate

n

‘space allocation’ prompting social behaviours

enforced grouping

social grouping

. .

.

... ..

..

..

.

.

.

.

..

..

..

..

enforced grouping

..

.

.social grouping

..

.

.

....

..

..social grouping

(a) (b)

Figure 6. A speculative graph (b) on how fine scale foraging rules might be anticipated to change

with scale by evoking social foraging behaviours. Preference may be constrained when animals

graze as an enforced group (dotted line), or when grazing as a social group (fall-off at large scales), as

shown in (a).

effect on the large scale spatial heterogene-

ity in the vegetation, nor on the fine scale

heterogeneity. If the group of say 5 moves

at random, and revisits areas on a fairly

short time scale (e.g. where stocking rate is

high), it is unlikely their impacts would be

distinguishable from that of 5 individuals

eachmoving at random. The group may act

like a ‘meta-animal’. If the group grazed

with the same fine scale rules as individu-

als, the outcome would be the same, and as

seen in Figure 1.

But fine scale foraging rules might be

anticipated to change with scale as indi-

cated in Figure 6b. Such a graph is, we feel,

an essential component for a foraging

model (either as an input or preferably an

emergent property) that aims to analyse

heterogeneity and its impacts across a range

of scales. But such an analysis is rarely pre-

sented, and is largely speculative here. The

axis of ‘space allocation’ should be seen

perhaps only as a surrogate for the extent to

which various social behaviours are

evoked. We propose that foraging effi-

ciency (the apparent selection of preferred

items) might decline at large spatial scales

as a combination of environmental com-

plexity and the constraints of group forag-

ing limit the foraging success of individuals

(as we have discussed earlier). What we add

is that individuals’ foraging success might

also be expected to decline when animals

are forced into groups at very small spatial

allocations. Consistently, increasing the in-

stantaneous stocking rate from 50 to

150 ewes per ha and per day in an oak cop-

pice where the availability of the preferred

herbaceous layer was low resulted (for a

same overall high stocking rate) in an in-

crease of browse consumption by the ewes

[31]. What is critical here is that not all ani-

mals are equal. Social groups impose con-

straints on subordinate individuals. In

grazing red deer, for example, the subordi-

nates do not have access to preferred

patches [2] are less synchronised with the

dominants [10] and have a lower biting rate

when near the dominants [49]. Similar

hierarchical social relationships may, for

example, be expected to reduce the forag-

ing efficiency of subordinate members of

groups of domestic herbivores at the high

stocking densities in each successive pad-

dock in a rotational grazing system.

5. TO WHAT EXTENT DO ANIMALS

COME CLOSE TO ACHIEVING

THE OPTIMAL SOLUTION?

One final consideration about the conse-

quences of how animals place their bites in

space and time, is to ask to what extent ani-

mals moving as individuals, or as a group,

come close to achieving the optimal solu-

tion for repeatedly harvesting the vegeta-

tion.

5.1. Optimal foraging modelled

at the bite scale

Although the Marginal Value Theorem

[12] elegantly describes the optimal means

to exploit a succession of patches of vegeta-

tion, it is very difficult to conceive and ap-

ply in grazing systems where the vegetation

is spatially more continuous (albeit hetero-

geneous), where grazing and regrowth are

at a field scale simultaneous, and where

patches may be frequently revisited, so that

the rate of replacement of resources be-

comes paramount [3]. However, the opti-

mal solutions can be perceived readily, even

under continuous grazing, again by work-

ing at the fine (bite) scale [35]. Using the

same, previously published models of bite

scale foraging [39, 44], we can recognise

that individual bites are discrete, assumed

instantaneous, events and so generate a

regrowth pattern specific to each and every

bite taken. Just two of numerous possible

examples are shown in Figure 7. The shape

of the regrowth curve, and the amount that

may be harvested subsequently, depends

uniquely on the initial patch state (residual


after the previous bite), and the duration of

regrowth in each case [35]. These are then

two fundamental determinants of the rate of

replacement of resources. For any initial

patch state, (and any ensuing growth curve)

we can identify an optimum timing for har-

vest, which will achieve the maximum av-

erage rate of yield. This (the maximal

marginal value) is shown as the tangents

from Wi to the growth curves in Figure 7. In

Figure 8 (solid lines) we plot the optimal

solutions for all possible initial (residual)

patch states (x axis), in terms of the optimal

defoliation interval that is required and the

maximum sustainable yield that each com-

bination of residual patch states, and defoli-

ation intervals would achieve (for full

details see [35, 39]).

To understand the impact of the funda-

mentally different ways in which animals

may place their bites in space and time, we

can now translate the contrasting animal

foraging behaviours into the components of

regrowth (the residual patch states and de-

foliation intervals) that these would gives

rise to (the emergent properties), and we

can relate these to what would be the opti-

mal solutions. Most important, we consider

not just the mean values but the variance

about these because achieving the optimal

solutions would require that all patches are

defoliated identically – a deterministic so-

lution which is achievable only by uniform

cutting [38].

5.2. Model outputs in the different

grazing scenarios

The modelling predicts that grazing se-

quentially or at random, whether as individ-

uals or a group, would lead to combinations

of residual patch states and defoliation in-

tervals that are clearly suboptimal (Fig. 8a).

At low stocking rates, patches are grazed

too leniently, and defoliation intervals are

too long to approach optimal regrowth


optimal defoliation interval time

biom

ass

-W

(kg/

ha)

max marginal value

Wi

Wmax

Figure 7. The theoretical basis for seeking optimal solutions for a sustainable harvesting of a hetero-

geneous and regrowing resource is simple only at a bite-sized patch scale where defoliation is instan-

taneous. Unique regrowth curves (only two shown here, one from a low residual state, one from a

greater residual state) depend on the initial state of each patch, Wi , and there is an optimal time of har-

vest for that patch [35]. At even greater residual patch states, regrowth is monotonic, implying the

patch should be regrazed immediately. Foraging rules affect both these components and generate

unique growth curves, but with simple to derive optimal solutions.

rates. At higher stocking rates, although re-

sidual patch states decline, so too do defoli-

ation intervals. The sustainable yields fall

short of the optimum possible, not only be-

cause the combinations of the components

of regrowth are inappropriate, but also be-

cause under random grazing, there is vari-

ance about these components. Note that in

the examples so far, the animals are at all

times free to move around the whole area at

will, motivated by attempting to meet in-

take demand.

Next, we consider the case where ani-

mals move around the area as a group, but

where their rate of movement is regulated

(e.g. by management with subdivision or

fencing). In this way we impishly relate the

expectations for social foraging behaviour

to what is more widely observed in agricul-

ture – rotational grazing. Grazing as a rota-

tion, the models propose, comes considerably

closer to the optimal solutions for exploiting

resource replacement (Fig. 8b). This is be-

cause constraining the movement of the

animals allows control over at least one of

the two major components of regrowth. This

means that it is now possible, e.g. at high

stocking rates, to combine low residual


20

40

60

0

20

30

40

50

60

70

80

residual(initial) patch state (m)

increase instock rate


sequentialrandom

20

40

60

0

10

20

30

40

50

60

70

80

residual(initial) patch state (m)



(a) (b)op

timal

def

.int

erva

l (d)

0 0.05 0.1 0.15 0.2 0.25

10

0 0.05 0.1 0.15 0.2 0.25

0 0max

.sus

tain

able

yie

ld (

kg h

a d)

Figure 8. The effect of residual patch state on the maximum sustainable yield that may be achieved

and the defoliation interval required to achieve this (solid lines) as predicted by a model that seeks op-

timal solution for all combination of residual patch state and the timing of harvest. In (a) we show the

combinations of residual patch state and defoliation intervals that emerge (separate point for each of a

range of stocking rates) when animals are assumed to graze patches either at random (solid dots), or

in strict sequence (open circles) and in (b) the same for rotational grazing (open symbols, 30 days

with animals not present: 1 day grazing) compared to continuous grazing (solid symbols), all as in

Figure 1. Arrows show the direction of increasing stocking rate. According to [35].

patch states (severe defoliation) with long

intervals between defoliation. Note, many

defoliation intervals in a rotation are less

than one day. What matters more to plant

regrowth, however, is there is one interval

(e.g. here 30 days) that has been imposed,

which allows close to optimal resource re-

placement.

Finally, we combine all the impacts,

some probable social changes in foraging

rules with scale, and the concepts of ex-

ploiting the rate of replacement of re-

sources, in a dynamic context, using the

same models, in Figure 1b. Where animals

graze as an enforced group, and where so-

cial behaviours modify their capability to

graze selectively (e.g. at high stock densi-

ties in a rotation, selectivity is depressed in

some if not all animals), foraging success

per ha (and clearly per animal) is greatest

and close to the maximum sustainable. The

model runs, here, clearly demonstrate there

is an optimal time to leave a patch (this is a

simple extension of the MVT which in-

cludes resource replacement). However an

animal moving through the vegetation in-

tent on meeting its daily demand for intake,

may fail to realise the maximum sustain-

able rate of yield. In this analysis, each and

every animal would do better, in the longer

term, to constrain its movements to opti-

mise resource replacement. Presumably in-

dividual animals do not do this for fear their

longer-term vision would be exploited by

those who are more opportunists [24].

Clearly animals are not motivated by feed-

ing alone and the social interactions neces-

sary to ensure fitness and survival of

individuals might mitigate against achiev-

ing the optimal solution for the whole

group.

6. CONCLUSION

Although one of the most notable fea-

tures of heterogeneity is what we regard (as

humans) as large scale pattern, the challenge

for us is to understand how such pattern

could arise in the first place, and in any

case, what heterogeneity means to foraging

success and vegetation dynamics. The most

satisfying explanations, we propose, are

those that can generate pattern from an ini-

tially homogeneous state. The grazing pro-

cesses described here will achieve this,

though some spatially localising behav-

iours are necessary to aggregate the (e.g.

tall or short, or species) components of the

frequency distributions into large patches

in space. As a final note, we cannot overem-

phasise how many sources of heterogeneity

may yet be pre-determined e.g. by varia-

tions in soil quality; dung or urine return, or

due to the nature of the mechanism of dis-

persal of plant species and their biotic inter-

action with soil [32]. Complex spatial and

temporal phenomena can certainly arise

from such vegetation interactions alone

[43].

Given this complexity it is probable that

only by combining models with carefully

focussed experimentation, will we satisfy

the desire to understand the role of hetero-

geneity and grazing processes in ecosystem

function. Such work would need foremost

to address issues of spatial scale. Analyses

of the growth of vegetation resources in

grassland are soundly based in the physiol-

ogy and morphology of individual plants

but have tended to be modelled as a contin-

uous, deterministic process, as if homoge-

neous, at the field scale. This is in marked

contrast to advances in animal foraging sci-

ence and behavioural ecology which are

predominantly individual based, where for-

aging is perceived at the bite (prey or patch)

scale and considers the (stochastic) expec-

tation of success of finding food. This dis-

parity in scales can lead to substantial

imbalance in models and discrete, stochas-

tic, spatial accounts can give critically dif-

ferent predictions, e.g. of carrying capacity

and stability, from conventional continuous,

deterministic, homogeneous ones [39].

But we propose it is not only important to


address plant animal interactions at the

same scale, but to seek ‘rules’ across a

range of scales (e.g. in Fig. 6). Emphasis on

individual based concepts must not, of

course, overlook group foraging theory

[24].

Major advances in combining vegeta-

tion and animal behaviour have been made

recently, but one component of the system

is substantially overlooked – that of human

intervention in response to risk and uncer-

tainty, and a confusion of regulatory, emo-

tive and socio-economic goals. The models

above are essential for understanding the

biophysical system, but the challenge is

then to capture these insights and rational-

ise them to a scale and level of detail that is

more appropriate for tools for managing the

complexity of grassland ecosystems, and

sufficiently balanced to be able to seek opti-

mal solutions for achieving the human, as

much as the animals, multiple goals.

REFERENCES

[1] Adler P.B., Raff D.A., Lauenroth W.K., The ef-

fectofgrazingonthespatialheterogeneityofveg-

etation, Oecologia 128 (2001) 465–479.

[2] Appleby M.C., Social rank and food access in red

deer stags, Behaviour 74 (1980) 294–309.

[3] Arditi R., Dacorogna B., Optimal foraging on ar-

bitraryfooddistributionandthedefinitionofhab-

itat patches, Am. Nat. 131 (1988) 837–846.

[4] Arnold G.W., Dudzinski M.L., Ethology of free-

ranging domestic animals, Elsevier, Amsterdam,

1978.

[5] Bakker J.P., The impact of grazing on plant com-

munities, in: WallisDeVries M.F., Bakker J.P.,

Van Wieren S.E. (Eds.), Grazing and Conserva-

tion Management, Kluwer Academic Publishers,

Dordrecht, 1998, pp. 137–184.

[6] BazelyD.R.,EnsorC.V.,Discriminationlearning

in sheep with cues varying in brightness and hue,

Appl. Anim. Behav. Sci. 23 (1989) 293–299.

[7] Beecham J.A., Farnsworth K.D., Animal group

forces resulting from predator avoidance and

competition minimisation, J. Theor. Biol. 198

(1999) 533–548.

[8] Benhamou S., Efficiency of area-concentrated

searching behaviour in a continuous patchy envi-

ronment, J. Theor. Biol. 159 (1992) 67–81.

[9] Benhamou S., Spatial memory and searching ef-

ficiency, Anim. Behav. 47 (1994) 1423–1433.

[10] Blanc F., Thériez M., Effects of stocking density

on the behaviour and growth of farmed red deer

hinds, Appl. Anim. Behav. Sci. 56 (1998)

297–307.

[11] Boissy A., Dumont B., Interactions between so-

cial and feeding motivations on the grazing be-

haviourofherbivores:sheepmoreeasilysplit into

subgroups with familiar peers, Appl. Anim.

Behav. Sci. 79 (2002) 233–245.

[12] Charnov E.L., Optimal foraging, the marginal

value theorem, Theor. Pop. Biol. 9 (1976)

129–136.

[13] Dumont B., Boissy A., Grazing behaviour of

sheep in a situation of conflict between feeding

and social motivations, Behav. Proc. 49 (2000)

131–138.

[14] Dumont B., Hill D.R.C., Multi-agent simulation

of group foraging in sheep: effects of spatial

memory, conspecific attraction and plot size,

Ecol. Model. 141 (2001) 201–215.

[15] Dumont B., Petit M., Spatial memory of sheep at

pasture, Appl. Anim. Behav. Sci. 60 (1998)

43–53.

[16] Dumont B., Maillard J.F., Petit M., The effect of

the spatial distribution of plant species within the

sward on the searching success of sheep when

grazing, Grass For. Sci. 55 (2000) 138–145.

[17] Dumont B., Carrère P., D’hour P., Foraging in

patchy grasslands: diet selection by sheep and

cattle isaffectedby theabundanceandspatialdis-

tribution of preferred species, Anim. Res. 51

(2002) 367–381.

[18] EdwardsG.R.,NewmanJ.A.,ParsonsA.J.,Krebs

J.R., The use of spatial memory by grazing ani-

mals to locate food patches in spatially heteroge-

neous environments: an example with sheep,

Appl. Anim. Behav. Sci. 50 (1996) 147–160.

[19] EdwardsG.R.,NewmanJ.A.,ParsonsA.J.,Krebs

J.R.,Useofcuesbygrazinganimals tolocatefood

patches: an example with sheep, Appl. Anim.

Behav. Sci. 51 (1997) 59–68.

[20] Farnsworth K.D., Beecham J.A., Beyond the

ideal free distribution: more general models of

predator distribution, J. Theor. Biol. 187 (1997)

389–396.

[21] Farnsworth K.D., Beecham J.A., How do grazers

achieve their distribution? A continuum of mod-

els from random diffusion to the ideal free distri-

bution using biased random walks, Am. Nat. 153

(1999) 509–526.

[22] Fretwell S.D., Lucas H.L., On territorial behav-

iour and other factors influencing habitat distri-

bution in birds, Acta Biotheor. 19 (1970) 16–36.

[23] Gibb M.J., Huckle C.A., Nuthall R., Rook A.J.,

Effect of sward surface height on intake and graz-

ing behaviour by lactating Holstein Friesian

cows, Grass For. Sci. 52 (1997) 309–321.

[24] Giraldeau L.-A., Caraco T., Social Foraging The-

ory, Princetown University Press, Princetown,

New Jersey, 2000.


[25] Hewitson L., The foraging behaviour of sheep in

response to environmental uncertainty, Ph.D.,

University of Edinburgh, 2002.

[26] Howery L.D., Provenza F.D., Banner R.E., Scott

C.B., Social and environmental factors influence

cattle distribution on rangeland, Appl. Anim.

Behav. Sci. 55 (1998) 231–244.

[27] Hubbell S.P., The Unified Neutral Theory of

Biodiversity and Biogeography, Monographs in

Population Biology, 32, Princetown University

Press, Oxford, 2001.

[28] JarmanP.J.,Thesocialorganisationofantelope in

relation to their ecology, Behaviour 48 (1974)

215–267.

[29] Kershaw K.A., Quantitative and Dynamic Plant

Ecology,2nded.,EdwardArnold,London,1973.

[30] LazoA.,Facteursdéterminantsducomportement

grégairedebovinsretournésàl’étatsauvage,Rev.

Ecol. (Terre & Vie) 47 (1992) 51–66.

[31] Lecrivain E., Leclerc B., Hauwuy A.,

Consommation de ressources ligneuses dans un

taillis de chênes par des brebis en estive, Reprod.

Nutr. Dev. 30 (1990) 207s–208s.

[32] MarriottC.A.,CarrèreP.,Structureanddynamics

of grazed vegetation, Ann. Zootech. 47 (1998)

359–369.

[33] May R.M., Stability and complexity in model

ecosystems, 2nd ed., Princetown University

Press, Princetown, New Jersey, 1974.

[34] NewmanJ.A.,CaracoT.,Cooperativeandnonco-

operative bases for food-calling, J. Theor. Biol.

141 (1989) 197–209.

[35] Parsons A.J., Chapman D.F., The principles of

pasture growth and utilisation, in: Hopkins A.

(Ed.), Grass: its production and utilisation,

3rd ed., Blackwell Science Ltd, Oxford, UK,

2000, pp. 31–89.

[36] Parsons A.J., Newman J.A., Penning P.D.,

Harvey A., Orr R.J., Diet preference of sheep: ef-

fects of recent diet, physiological state and spe-

cies abundance, J. Anim. Ecol. 63 (1994)

465–478.

[37] Parsons A.J., Thornley J.H.M., Newman J.A.,

Penning P.D., A mechanistic model of some

physical determinants of intake rate and diet se-

lection in a two-species temperate grassland

sward, Funct. Ecol. 8 (1994) 187–204.

[38] Parsons A.J., Carrère P., Schwinning S., Dynam-

ics of heterogeneity in a grazed sward, in:

Lemaire G., Hodgson J., de Moraes A., Nabinger

C., de F. Carvalho P.C. (Eds.), Grassland

Ecophysiology and Grazing Ecology, CAB

International, Wallingford, UK, 2000,

pp. 289–315.

[39] Parsons A.J., Schwinning S., Carrère P., Plant

growth functions and possible spatial and tempo-

ral scaling errors in models of herbivory, Grass

For. Sci. 56 (2001) 21–34.

[40] Penning P.D., Parsons A.J., Newman J.A., Orr

R.J.,HarveyA.,Theeffectsofgroupsizeongraz-

ing time in sheep, Appl. Anim. Behav. Sci. 37

(1993) 101–109.

[41] Prache S., Damasceno J., Béchet G., Preferences

of sheep grazing down conterminal monoc ultures

of Lolium perenne – Festuca arundinacea: test of

an energy intake-rate maximization hypothesis,

in: Durand J.L., Emile J.C., Huyghe C., Lemaire

G. (Eds.), Multi-Function Grasslands – Quality

Forages, Animal Products and Landscapes, Brit-

ish Grassland Society, Reading, UK, 2002,

pp. 258–259.

[42] Ruxton G.D., Foraging on patches: are groups

disadvantaged?, Oikos 72 (1995) 148–150.

[43] Schwinning S., Parsons A.J., A spatially explicit

population model of stoloniferous N-fixing le-

gumes in mixed pasture with grass, J. Ecol. 84

(1996) 815–826.

[44] Schwinning S., Parsons A.J., The stability of

grazing systems revisited: spatial models and the

role of heterogeneity, Funct. Ecol. 13 (1999)

737–747.

[45] Sibbald A.M., Shellard L.J.F., Smart T.S., Effects

of space allowance on the grazing behaviour and

spacing of sheep, Appl. Anim. Behav. Sci. 70

(2000) 49–62.

[46] Spalinger D.E., Hobbs N.T., Mechanisms of for-

aging in mammalian herbivores: new models of

functional response, Am. Nat. 140 (1992)

325–348.

[47] Stephens D.W., Krebs J.R., Foraging Theory,

Princetown University Press, Princetown, New

Jersey, 1986.

[48] Thornley J.H.M., Parsons A.J., Newman J.A.,

Penning P.D., A cost-benefit model of grazing in-

take and diet selection in a two-species temperate

grassland sward, Funct. Ecol. 8 (1994) 5–16.

[49] Thouless C.R., Feeding competition between

grazing red deer hinds, Anim. Behav. 40 (1990)

105–111.

[50] Turchin P., Odendaal F.J., Rausher M.D., Quan-

tifying insect movement in the field, Environ.

Entomol. 20 (1991) 955–963.

[51] Ungar E.D. Ingestive behaviour, in: Hodgson J.,

IlliusA.W.(Eds.),TheEcologyandManagement

of Grazing Systems, CAB International,


[52] Valone T.J., Patch information and estimation: a

costofgroupforaging,Oikos68(1993)258–266.

[53] Ward D., Saltz D., Foraging at different spatial

scales: Dorcas gazelles foraging for lilies in the

Negev desert, Ecology 75 (1994) 48–58.


Review article

Grazing and pasture management

for biodiversity benefit

Andrew J. ROOK*, Jeremy R.B. TALLOWIN

Soils, Environmental and Ecological Science Department, Institute of Grassland

and Environmental Research, North Wyke, Okehampton, Devon EX20 2SB, UK

(Received 19 August 2002; accepted 25 February 2003)

Abstract — The primary role of grazing animals in grassland biodiversity management is mainte-

nance and enhancement of sward structural heterogeneity, and thus botanical and faunal diversity, by

selective defoliation due to dietary choices, treading, nutrient cycling and propagule dispersal. Most

research on dietary choices uses model systems that require considerable extrapolation to more com-

plex communities. Grazing animals’ diets are constrained by temporal and spatial changes in sward

structure, plant defence mechanisms, herbage availability, plant phenology and animal physiological

state. Potentially, these could be exploited to manipulate choice in diverse communities. Dietary

choice differs between animal species, driven by factors such as body size, digestive physiology and

dental anatomy. There is anecdotal evidence for breed differences but little experimentation, with ge-

netic effects often confounded with background experience. There is information about landscape-

scale breed and background effects but little about parameters such as bite and feeding station areas

that allow reconstruction of the development of small-scale sward patchiness. An experiment at five

European sites is examining breed effects on grazing behaviour, structural, floral and faunal diversity,

animal production and economic impacts. In another project, calves are being reared by their own

mothers or by cows of another breed allowing genetic effects on grazing behaviour to be separated

from effects of early experience. ‘Designer animals’may be needed to deliver desired grazing behav-

iour and biodiversity outcomes, either by breeding or by the use of training and previous experience to

manipulate choices. Application of research results requires consideration of conservation goals,

whether at landscape, habitat, plant community or plant species level. There is a need to replace

stocking rate prescriptions with sward-based methods and to integrate biodiversity goals into inten-

sive systems. Major gaps in our knowledge of grazing behaviour and its impact on biodiversity re-

main, necessitating greater integration of plant ecophysiology, plant community ecology and animal

behavioural ecology research.

grazing / biodiversity / pasture management / dietary choices / animal type

181

Anim. Res. 52 (2003) 181–189


DOI: 10.1051/animres:2003014


Tel.: +44 1837 883548; fax: +44 1837 82139; e-mail: [email protected]

Résumé — Pâturage et gestion des prairies pour davantage de biodiversité. Le rôle majeur des

animaux au pâturage dans la conservation de la biodiversité des prairies permanentes est de maintenir

et de développer l’hétérogénéité du couvert végétal par la défoliation sélective, selon les choix ali-

mentaires, par le piétinement, le recyclage des éléments nutritifs et la dispersion des graines. Ainsi le

pâturage favorise la diversité floristique de la prairie et la diversité faunistique associée. La plupart

des recherches sur les choix alimentaires utilisent des situations modèles qui exigent des extrapola-

tions considérables pour des communautés végétales plus complexes. En effet, le régime alimentaire

des animaux au pâturage dépend des changements spatio-temporels de la structure du couvert végé-

tal, des mécanismes de défense des plantes, de la disponibilité en herbe, de la phénologie des plantes

et de l’état physiologique des animaux. Potentiellement, ces facteurs peuvent être exploités pour

orienter les choix alimentaires dans diverses communautés végétales. Les choix alimentaires diffè-

rent entre les espèces animales, selon des facteurs comme la taille de l’animal, la physiologie de la di-

gestion et l’anatomie dentaire. Des différences entre races ont été mises en évidence sur la base d’une

démarche plus empirique qu’expérimentale, les effets génétiques étant souvent confondus avec les

effets de l’environnement et de l’expérience alimentaire des animaux. Il existe également des don-

nées sur l’effet de la race et du milieu dans lequel évolue l’animal à l’échelle large du paysage, mais

peu au sujet de paramètres tels que la surface des bouchées et des stations alimentaires qui permettent

la restauration de l’hétérogénéité du pâturage à une échelle plus petite. Un essai conduit actuellement

sur cinq sites européens a pour but d’étudier l’influence de la race sur le comportement au pâturage, la

diversité de la structure et de la flore de la prairie ainsi que celle de la faune associée, la production

animale et l’impact économique. Dans une autre étude, des veaux sont élevés soit par leur mère soit

par des vaches d’une autre race afin de séparer sur le comportement au pâturage les effets liés aux fac-

teurs génétiques de ceux liés à l’apprentissage dans le jeune âge. En orientant les choix alimentaires

des animaux, par la sélection génétique ou bien par l’expérience dans le jeune âge et l’apprentissage,

on pourrait utiliser des animaux ‘modèles’ pour favoriser des comportements de pâturage désirés, et

par voie de conséquence la biodiversité de la prairie. La mise en application des résultats de la re-

cherche doit prendre en compte les objectifs de conservation des espaces pastoraux aussi bien à

l’échelle du paysage, que de l’habitat, de la communauté ou de l’espèce végétale. Il devient alors fon-

damental de remplacer les méthodes traditionnelles de conduite du pâturage, faisant appel à la notion

de chargement, par des méthodes basées sur l’état du couvert végétal, et aussi d’intégrer des objectifs

de biodiversité dans les systèmes d’élevage intensif. De nombreuses lacunes dans notre connaissance

du comportement de pâturage et de son impact sur la biodiversité demeurent. Il est nécessaire de

mieux intégrer les recherches menées en écophysiologie des plantes, en écologie des communautés

végétales et écologie comportementale des animaux.

pâturage / biodiversité / conduite de la prairie / choix alimentaires / type d’animal

1. INTRODUCTION

In this paper, we review the role of the

grazing animal in the management of grass-

lands for biodiversity, and the mechanisms

by which this role occurs. We identify some

of the gaps in our knowledge and describe

experiments at the Institute of Grassland

and Environmental Research, North Wyke

(UK) that are attempting to fill some of

these gaps. We consider the goals of

biodiversity management and some of the

important current issues. In the light of

these goals, we consider some of the poten-

tial management tools that may assist graz-

ing managers to enhance biodiversity.

If we are to exploit current knowledge

and to take sensible directions in applied re-

search, it is necessary to consider the goals

of our conservation management. To some

extent, this is an issue of scale. The goal

might be to manage for a cultural land-

scape, that is not just the physical features

of the landscape but also the human aspects.

182 A.J. Rook, J.R.B. Tallowin

For example, the species and breed of ani-

mal employed might be chosen to reflect

traditional local practices. The possibility

that this might compromise biodiversity

outcomes per se need to be considered. An-

other goal might be to manage at the habitat

level over several different plant communi-

ties or at the community level. Alterna-

tively, interest may centre on rare or

emblematic species or on the fauna rather

than the flora. These scenarios may require

very different grazing management prac-

tices.

Having established the goals, there is

also a need to consider some of the current

issues in the management of grasslands for

biodiversity. One of these is the need to

move away from crude, though easily im-

plemented, management methods such as

stocking rates prescriptions on which many

current agri-environment schemes are

based, to sward based management guide-

lines such as sward height or sward height

distribution [5]. Another pressing issue is to

obtain reliable evidence about animal breed

and background effects as many current

management prescriptions rely heavily on

anecdotal evidence. There is also a need to

integrate biodiversity goals into intensive

systems. Currently, many of our agri-envi-

ronment schemes are aimed at farmers in

marginal areas rather than at, for example

the intensive dairy farmer.

2. THE ROLE OF THE GRAZING

ANIMAL

Most temperate grasslands require peri-

odic defoliation to control succession, if

they are not to succeed to scrub and ulti-

mately woodland. Except on very steep or

uneven ground, it is usually possible to

achieve this defoliation by mechanical har-

vesting of the herbage. Indeed some com-

munities such as hay meadows have

evolved in response to such management.

However, the grazing animal has a unique

role to play. This is to maintain and enhance

structural heterogeneity of the sward can-

opy, which in turn has a vital influence on

floral and faunal diversity.

3. MECHANISMS FOR CREATING

HETEROGENEITY IN GRAZED

SWARDS

Probably the most important mecha-

nism by which grazing animals create

sward heterogeneity is selective defoliation

as a result of dietary choices both between

species and between plant parts within spe-

cies. This alters the competitive advantage

between plant species both by direct re-

moval of phytomass and by altering the

light environment and competition for soil

nutrients [3]. A second mechanism is tread-

ing which opens up regeneration niches for

gap-colonising species. A third mechanism

is nutrient cycling. This has the effect of

concentrating nutrients into ‘hot spots’ at

dung and urine patches and again may alter

the competitive advantage between species,

both directly and by feedback effects on di-

etary choice and thus on heterogeneity, as

cattle in particular will not graze near dung

patches. Grazing animals also have a role in

propagule dispersal. This may be either

endozoochorous (i.e. by seeds passing

through the animal’s digestive system) or

exozoochorous (i.e. by seeds attaching to

the animal’s coat) dispersal but we particu-

larly stress the role of the endozoochorous

route as the mower can also effectively dis-

tribute many exozoochorous species. For a

more comprehensive review of plant re-

sponses to grazing see [2].

The direct effects of grazing on sward

canopy structure and the plant community

lead to secondary effects on faunal diversity

both by changing the abundance of food

plants and by providing breeding sites. The

direct effects on invertebrate diversity feed

through to vertebrate diversity (e.g. [20]).

Another secondary effect of the changes in

Grazing management for biodiversity 183

structure and community brought about by

grazing is the feedback on the grazing be-

haviour of the animals by changing the

choices available to them.

4. DIETARY CHOICES

Since much of the system is driven by

the animal’s dietary choices (both between

species and between potential feeding sta-

tions), it is important to understand the

mechanisms driving these choices. It

should be stressed that most of our knowl-

edge is derived from simple model systems,

such as perennial ryegrass-white clover and

that there has been relatively little detailed

work in more complex communities, at

least in temperate lowland environments.

Generally, behavioural ecologists have

assumed that the animal is striving to opti-

mise its evolutionary fitness. In the context

of foraging, rate of energy intake has usu-

ally been taken as a surrogate measure for

evolutionary fitness [18]. However, in

many situations animals appear to behave

sub-optimally. For example, grazed grass

has a carbon/nitrogen ratio too low to be op-

timal for the animal’s requirements but both

cattle and sheep offered a free choice with

minimal physical constraints consistently

chose a diet containing around 70% clover,

with an even lower C/N ratio [17]. Further-

more, the mixed diet is not due to intake rate

maximisation since in this case animals

would choose 100% clover as this species

can be eaten faster [17]. This suggests that

rate of energy intake is not the currency that

the animal is optimising, and that the true

currency remains to be identified. To opti-

mise fitness the animal has to trade-off

many currencies, for example nutrient in-

take with predation risk and these trade-offs

are not fully understood [17]. These limita-

tions to our knowledge make it difficult to

extrapolate from our simple model systems

to the more complex swards of interest to

biodiversity managers.

In discussing dietary choices, many peo-

ple, particularly those working in conserva-

tion management use the term ‘palatability’

as a plant descriptor. Unfortunately, the

term has been so misused as to have become

almost meaningless. Palatability refers to

the acceptability to an animal of a food

based purely upon organoleptic properties

independent of post-ingestive conse-

quences. Animals can learn to associate

post-ingestive consequences (for example

toxicity) with the taste of a food and subse-

quently use taste as a cue to avoid this food

but there are relatively few examples of

choices being made solely on organoleptic

grounds. Palatability is primarily an animal

not a food characteristic as there are many

situations in which an animal’s food choice

is altered even though the foods themselves

remain unchanged. We therefore recom-

mend that the term is not used as a food

descriptor and suggest that it is of little use

in understanding the basis of dietary

choices (for a fuller discussion see [14]).

It is more useful when describing food

choices to regard the animal as having a po-

tential intake but being constrained by vari-

ous factors, including those inherent to the

food. When grazing there are important

physical constraints on intake and therefore

on the choices between plants with differ-

ent levels of these constraints. These con-

straints include sward structure (sward

height, leafiness, tiller density and horizon-

tal patchiness) and plant physical defence

mechanisms (for review see [15]). This

latter may be a driving force in patch

formation in some situations, for example,

the animals may avoid an area around

Cirsium sp.

Dietary choice changes over time at all

scales. This is due to the availability of

herbage, phenology of the plant and the

physiological state of the animal. An exam-

ple of a relatively short-term temporal ef-

fect is the change in preference between

grass and clover that has been observed over

the day. Both dairy cows and sheep include


more clover in the diet in the morning and

more grass in the evening [16]. It has been

speculated that this might be either due to

higher sugar levels in the grass at this time

[11] and hence higher digestibility or, alter-

natively, it may be because the animal fills

its rumen with relatively slowly digesting

material (compared to clover) in order to

maintain rumen microflora populations

during the overnight fast. At present, it is

not possible to offer a definitive answer. If a

similar circadian effect was to be seen in

choices between elements of plant commu-

nities of interest for biodiversity, it might be

possible to exploit the effect to manipulate

overall choice and hence effect on sward

structure and diversity.

There are also spatial effects at many

scales. Most animals in lowland systems in

Europe have no opportunity to make

choices at the landscape scale and hence

much of our research in these systems re-

lates to choice at the bite or feeding station

(i.e. without moving the legs) scale. In hill

and upland systems (and range systems in

other countries), we know that animals es-

tablish home ranges within which they

move on daily and longer time scales.

Choice of location may be driven by other

factors than food, such as water, shelter and

social cohesion (itself an anti-predation

strategy) (e.g. [6]).

5. ANIMAL TYPE

Animal type has a major effect on di-

etary choice. The most fundamental effect

is that of body size. Because larger animals

have relatively large gut capacity in relation

to their metabolic requirements, they can

retain digesta in the tract for longer and thus

digest it more thoroughly. This means that

they can deal with a lower digestibility diet

and hence can forage less selectively than

smaller animals which must of necessity

select higher quality items [10]. The ani-

mal’s physiological state will also affect its

selection. For example, hungry animals are

less selective [12].

Species effects are of great importance.

Some of these are driven by body size, for

example, sheep are more selective than cat-

tle. Digestive physiology is also important,

for example, ruminants such as cattle have

more efficient digestion than hind-gut

fermenters such as horses [10]. The latter

therefore rely on high throughput and this

can necessitate long grazing times of up to

19 hours per day (e.g. [4]). Dental anatomy

is also important; horses, with both top and

bottom incisors can graze much closer to

the ground than cattle and appear to con-

centrate their grazing in short areas that rep-

resent only a small proportion of the

available area and thus produce a quite dif-

ferent sward structure [7]. The extent to

which grazing by horses results in a differ-

ent plant community to grazing by cattle is

still the subject of some debate. Many horse

grazed pastures are overstocked, leading to

poor structure and loss of diversity [1]. This

has probably resulted in an unjust, negative

perception of grazing by horses as a tool for

conservation management.

There is much anecdotal evidence (e.g.

[19]) for breed differences in diet selection

and hence in impact on sward structure and

composition but little experimental evi-

dence. In these anecdotal reports, true ge-

netic differences between breeds are often

confounded with the environmental effects,

particularly prior experience of biodiverse

pastures during early life that may affect

subsequent selection.

Secondary evidence on breed effects is

also patchy. There is some good informa-

tion about breed and background effects on

animal movements at a landscape scale. For

example in an experiment in which Scottish

Blackface or Suffolk ewes raised either

lambs of their own breed or of the other

breed [8], the distances between Blackface

ewes was greater than between Suffolks but

Blackfaces kept their lambs much closer to

them, whatever the breed of the lamb. The


ewes had a choice of using upland or low-

land pasture; the Blackface ewes made

much more use of upland and this persisted

in the lambs that they had reared whatever

the lamb breed, although some effect of

lamb breed was also evident (Tab. I).

While we have reasonable information

on breed effects on movement at a landscape

scale in heterogeneous environments, we

know very little about differences at the

scale of the grazing bout or feeding patch.

There is information from single breeds

grazing homogeneous pastures (e.g. [9,

13]); these provide information on parame-

ters such those shown in Table II that allow

the development of patchiness in the grazed

sward to be reconstructed [2]. However, we

have no idea if and how breeds differ in

these parameters, how any such effects

would be modified in heterogeneous pas-

tures or how they would interact with the

background of the animals, either immedi-

ately prior to moving to the target area or

during early life. It is also possible that

small-scale selection and the heterogeneity

this creates will differ depending on the

scale of enclosure in which the animal is al-

lowed to make its choices. Some of the

main gaps in our knowledge are summa-

rised in Table III. Because of these gaps, we

are currently ill-placed to predict effects of

different grazing managements and breed

on biodiversity.

6. CURRENT EXPERIMENTS

In this section, we describe two recently

initiated experiments which illustrate re-

search approaches to some current issues in

biodiversity management. In an EU pro-

ject (FORBIOBEN), we are comparing

North Devon (a traditional breed) with

Charolais × Holstein-Frieisan (a commer-

cial breed) yearling steers. The aim of this

project is to test if there are any breed differ-

ences in grazing behaviour and impact on

biodiversity. With the commercial breed,

we will also look at grazing intensity. The

animals are grazing agriculturally semi-im-

proved grassland and rush pasture contain-

ing 5–10 species per m2. We have chosen

the North Devon as it was originally devel-

oped on this type of grassland, particularly

rush pastures. We are monitoring botanical

structure and flora and faunal diversity,

herbage and animal production and eco-

nomic outputs. Similar trials are taking

place at 4 other sites across Europe

(Tab. III).

In a second project (BEFORBIO),

funded by the UK Department for the Envi-

ronment, Food and Rural Affairs, we are at-

tempting to separate true breed (genetic)

effects from the effects of early experience.

North Devon or Hereford-Friesian suckler

cows have been mated to North Devon and

Charolais bulls, respectively. At birth we


Table I. Inter-animal distance and percentage of time spent on upland pasture by Scottish Blackface

and Suffolk ewes rearing lambs either of their own, or of the other breed (according to Dwyer and

Lawrence [8]).

Ewe

Lamb

Blackface

Blackface

Suffolk

Blackface

Blackface

Suffolk

Suffolk

Suffolk

Ewe-ewe distance (m) 12.77 5.30 12.01 3.84

Ewe-lamb distance (m) 6.03 11.64 5.45 10.88

% upland use (ewes) 78 10 82 2

% upland use (lambs) 82 28 55 4

will cross-foster half the calves onto cows

of the other breed. Devon cows with their

own or fostered calves will graze a fen

meadow/rush pasture while Hereford-

Friesian cows + calves will graze a ferti-

lised ryegrass sward. In their second year,

all the calves will graze on fen meadow/

rush pasture and we will compare behav-

iour and impact on the sward of the differ-

ent breeds and backgrounds.

7. MANAGEMENT OPTIONS

In this section, we consider some of op-

tions for biodiversity management that we

believe current research findings in grazing

behaviour make possible. We believe there

is a need for ‘designer animals’that are cho-

sen to deliver desired grazing behaviour

and hence biodiversity goals. This might be

achieved by exploiting the animal’s genet-

ics. We do not believe that genetically mod-

ified animals per se will be acceptable to

European consumers in the foreseeable fu-

ture, particularly in a conservation context.

However, the new insights provided by

genomic technologies open up opportuni-

ties for greater understanding of the genetic

basis of behaviour and the use of marker


Table II. Examples of small scale (within head

down grazing bout) movement parameters for

heifers and ewes. According to Harvey et al. [9].

Heifers Ewes

Bout duration (s) 180 51

Distance moved during

bout (m)

7.3 1.8

Bites per m 32.3 37.4

Bites per bout 244 67

Bites per min 80 79

Bite area (cm3) 36.4 16.7

Table III. Gaps in current knowledge on the ef-

fects of grazing on biodiversity.

Dietary choices of animals in temperate

multi-species swards

Small scale animal movements in

heterogeneous swards

Relative importance of genetics

and environment

Effects of spatial scale at which grazing

management is applied

40

20

0

20

40

7 9 11 13 15 17 19 21 23 1 3 5Hour

clover

grass

Min.

Figure 1. Minutes grazing clover or grass by dairy cows in each hour of the day (according to Rutter

et al. [18]).

assisted selection to obtain animals with

desirable traits. In the short term, we be-

lieve there is opportunity to better exploit

the background of animals and to train them

to produce the biodiversity outcomes that

we desire. There is also much scope for ex-

ploiting temporal behaviour patterns to ma-

nipulate dietary choices so as to ensure that,

whilst animals are productive and provide

the farmer with an acceptable economic re-

turn, they also make the desired choices

when grazing biodiverse pastures. An ex-

treme example of this might be the use of

folding systems (such as practised in the

past in many chalk downland systems) in

which animals are removed to fallow arable

land for part of the day as a means of export-

ing nutrients.

8. CONCLUSIONS

In conclusion, grazing animals have a

vital role to play in the management of

biodiverse pastures. However, major gaps

in our knowledge of grazing behaviour in

such pastures and its impact on biodiversity

remain. We believe that there is a need for

stronger integration between research on

plant ecophysiology and plant community

ecology and the behavioural ecology of for-

aging herbivores in order to address these

knowledge gaps.

ACKNOWLEDGEMENTS

We thank the organisers of the Pasture Ecol-

ogy Group Meeting for inviting us to prepare

this paper and INRA for funding attendance at

the Meeting. We acknowledge funding from the

EU Commission Framework 5 programme

(FORBIOBEN – QLK5-2001-00130) and from

the UK Department for the Environment, Food

and Rural Affairs. We also acknowledge the

contribution of our partners to the setting up of

the FORBIOBEN project – M. Petit, B. Dumont,

J. Isselstein, M. Hofmann, K. Osoro, M.F. Wallis

de Vries, G. Parente, S. Venerus, M. Scimone, P.

Gaskell and J. Mills. Animal Research at IGER

is carried out in accordance with the welfare

standards approved by IGER’s Ethical Review

Procedure.

REFERENCES

[1] Bullock D.J., Armstrong H.M., Grazing for envi-

ronmental benefits, in: Rook A.J., Penning P.D.

(Eds.), Grazing management, The principles and

practice of grazing, for profit and environmental

gain, within temperate grassland systems, Brit-


pp. 191–200.

[2] Baumont R., Dumont B., Carrère P., Pérochon

L., Mazel C., Design of a multi-agent model of a

herd of ruminants grazing a perennial grassland:

the animal sub-model, in: Durand J.J., Emile J.

C., Huyghe C., Lemaire G. (Eds.), Multi-func-

tion grasslands: quality forages, animal products

and landscapes, Association Française pour la

Production Fourragère, Versailles, 2002,

pp. 236–237.

[3] Bullock J.M., Marriott C.A., Plant responses to

grazing and opportunities for manipulation, in:

Rook A.J., Penning P.D. (Eds.), Grazing man-

agement, The principles and practice of grazing,

for profit and environmental gain, within tem-

perate grassland systems, British Grassland So-

ciety, Reading, UK, 2000, pp. 17–26.

[4] Crowell-Davis S.L., Houpt K.A., Carnevale J.,

Feeding and Drinking behavior of mares and

foals with free access to pasture and water, J.

Anim. Sci. 60 (1985) 883–889.

[5] Diack I.A., Burke M., Peel S., Grassland man-

agement for nature conservation towards a con-

sistent approach to sward measurement and

description, in: Rook A.J., Penning P.D. (Eds.),

Grazing management, The principles and prac-

tice of grazing, for profit and environmental



pp. 155–156.

[6] Dumont B, Boissy A., Grazing behaviour of

sheep in a situation of conflict between feeding

and social motivations, Behav. Processes 49

(2000) 131–138.

[7] Dwyer C.M., Lawrence A.B., Effects of mater-

nal genotype and behaviour on the behavioural

development of their offspring in sheep, Behav-

iour 137 (2000) 1629–1654.

[8] Fleurance G., Duncan P., Mallevaud B., Daily in-

take and the selection of feeding sites by horses

in heterogeneous wet grasslands, Anim. Res. 50

(2001) 149–156.

[9] Harvey A., Parsons A.J., Orr R.J., Rutter S.M.,

Rook A.J., Movement patterns of sheep and cattle

within and between continuous short-term graz-

ing bouts, Proceedings of the 32nd International


Congress of the ISAE, Clermont Ferrand, 21-25

July 1998, p. 215.

[10] Illius A.W., Gordon I.J., Diet selection in mam-

malian herbivores: constraints and tactics, in:

Hughes R.N. (Ed.), Diet selection: an interdisci-

plinary approach to foraging behaviour,

Blackwell Scientific, Oxford, 1993, pp. 157–

181.

[11] Orr R.J., Rutter S.M., Penning P.D., Rook A.J.,

Matching grass supply to grazing patterns for

dairy cows, Grass For. Sci. 56 (2001) 352–361.

[12] Newman J.A., Penning P.D., Parsons A.J.,

Harvey A., Orr R.J., Fasting affects intake be-

haviour and diet preference of grazing sheep,

Anim. Behav. 47 (1994) 185–193.

[13] Roguet C., Prache S., Influence of forage avail-

ability on feeding station behaviour of ewes,

Ann. Zootech. 44 (Suppl.) (1995) 106.

[14] Rook A.J., Penning P.D., Rook A.J., Limitations

to palatability as a concept in food intake and diet

selection studies, in: Forbes J.M., Lawrence

T.L.J., Rodway R.G., Varley M.A. (Eds.), Ani-

mal choices, British Society of Animal Science

Occasional Symposium 20, British Society of

Animal Science, Edinburgh, 1997.

[15] Rook A.J., Principles of foraging and grazing be-

haviour, in: Hopkins A. (Ed.), Grass, its produc-

tion and utilization, 3rd ed., Blackwell Science,

Oxford, 2000, pp. 229–241.

[16] Rutter S.M., Orr R.J., Rook A.J., Dietary prefer-

ence for grass and white clover in sheep and cat-

tle: an overview, in: Rook A.J., Penning P.D.

(Eds.), Grazing management, The principles and

practice of grazing, for profit and environmental



pp. 73–78.

[17] Rutter S.M., Orr R.J., Penning P.D., Yarrow N.H.,

Champion R.A., Atkinson L.D., Dietary prefer-

ence of dairy cows grazing grass and clover. Pro-

ceedings of the Winter Meeting of the British

Society of Animal Science, Scarborough, UK,

23-25 March 1998, p. 51.

[18] Stephens D.W., Krebs J.R., Foraging Theory,

Princeton University Press, Princeton, 1986.

[19] Tolhurst S., Oates M., The breed profiles hand-

book, English Nature, Peterborough, UK, 2001.

[20] Vickery J.A., Tallowin J.R., Feber R.E., Asteraki

E.J., Atkinson P.W., Fuller R.J., Brown, V.K.,

The management of lowland neutral grasslands

in Britain: effects of agricultural practices on

birds and their food resources, J. Appl. Ecol. 38

(2001) 647–664.

To access this journal online:www.edpsciences.org


Documents

Pasture Ecology Meeting Plant-Herbivore Relationships · ty, biogeochemical cycles and landscapes. Thus, there is a considerable need to improve scientific knowledge by including