Animals seen as mediators of the vegetation value : new advances in animal-vegetation interactions modelling

(1)

..

•

Cirad-Emvt Programme « Productions Animales »

Institut National Recherche Agronomique

Unités de Recherches sur les Herbivores

DYNAMIQUES DE VEGETATION ET RELATIONS

HERBE/ ANIMAL

Organisation et édition scientifique :

Michel DURU (INRA-Toulouse), Philippe LECOMTE (Cirad-Emvt)

et Hubert GUERIN (Cirad-Emvt)

Compte rendu du Séminaire INRA-CIRAD à Montpellier

Les 31 janvier et 1er

février 2001

Rapport n°2001-37

rHF,.JLJ Il, 1 t

(2)

Dynamiques de végétation en système herbager et relations herbeanimal -Séminaire INRA-CIRAD - Montpellier Baillarguet - 31 janvier-!" février 2001

Animais seen as media tors of the vegetation value:

new advances in animal-vegetation interactions modelling

Michel MEURET1 & Bertrand DUMONT2

1 _{INRA, Systèmes}_{Agraires et Développement, Ecodéveloppement, Domaine Saint-Paul,}

Site Agroparc, 84914 Avignon Cedex 9, France 2

/NRA, Elevage et Nutrition des Animaux, Unité de Recherches sur les Herbivores, Theix, 63122 Saint-Genès Champanelle, France

This paper has already been published as :

Advances in Modelling Animal-Vegetation Interactions and Their Use in Guiding Grazing Management From: EEAP Publication N° 97 - Wageningen Pers: 57-72.-5th International Livestock Farrning Systems Symposium

lntegrating Animal Science Advances in the Search of Sustainability- 19-20 August 1999 -Posieux - Switzerland

Summary

Grazing managers and scientists are faced with new social concems, such as those relating to the environmental 'value' of grazinglands, which are no longer viewed as 'natural forages' only. Scientists need to design new models that include a set of decision rules about how to graze animals under multiple objectives. To study how an animal perceives and moves through its environment, grazing has to be considered as an ecological process. Hence, it is crucial to re-scale grazing mechanisms within spatial and temporal levels. Our paper then presents recent advances in the knowledge of animal social behaviour and spatial memory, which help to understand how herds and flocks exploit heterogeneous environments. An analysis of three recent and still developing grazing models finally enables us to discuss how this knowledge can be incorporated into research models. We refute the classic 'carrying capacity' paradigm that ignores one of the major aspects of herbivore at pasture : mobility. W e encourage farmers to become 'designers' of grazing areas, using appropriate paddocking or shepherding rules to impact the animals' spatial distribution and change their expectations about the relative value of feeding sites, in order to graze them in less preferred areas. Hopefully, recent advances in the knowledge of the cognitive abilities of herbivores open the way to a new generation of models that help to better understand how grazing managers can cope with pasture heterogeneity and variability.

Keywords : grazing, herbivores, behaviour, land management, biodiversity, ecological scales Both grasslands and rangelands are being attributed new social values and uses. These areas are perceived today as multiple-use landscapes and reservoirs for biodiversity. This vision is miles away from that which, for over 30 years, has considered these areas as 'natural forages' only (Hubert et al., 1995). Grazing researchers are thus pressed to build new grazing models which include a set of decision rules about how to manage grazinglands under multiple objectives. This implies that the concept of grazingland 'value' be reconsidered and primarily informed through better understanding of animal response to vegetation structure and dynamics. This paper focuses first on the elements of debate about temporal and spatial scaling, which are of primary importance when grazing is seen as an ecological process. Sorne behavioural aspects are then discussed, which are relevant to understand how herds and flocks exploit heterogeneous environments and how a former can upgrade vegetation use by correctly designing paddocks. The greater part of this paper is then taken up by a discussion

(3)

of three recent and still developing grazing models and the management implications that can be proposed at this point.

Guiding animais at pasture

Studies on the grazing process are founded on assumptions about how an animal moves through and perceives its environment. Hodgson & Illius (1996) have recently reviewed the broad field of foraging theory and modelling. They conclude that integrating grazing mechanisms across spatial and temporal scales is a crucial challenge, and that conceptual tools from ecology could help to guide grazing management. Scale is the resolution within the range of a measured quantity (Auger et al., 1992). The basic quantitative item considered in the grazing process is the plant material consumed by herbivores at each point of the area. The finest level of resolution is the bite and the largest is the land area available to the herd over several years. As emphasised by Laca & Ortega (1995), most grazing studies ignore scales and implicitly assume that small-scale measurements can be extrapolated to larger scales just by changing units. Nevertheless, ecological studies have clearly pointed out that the leap from small to larger scales implies a temporal and spatial re-scaling of the phenomenon under study (Allen & Starr, 1982; Dale et al., 1989). Its validity depends on the degree of spatial heterogeneity in the vegetation, on the seasonal variation in environmental conditions, and on the physiological and behavioural responses of the animais.

Despite the dependence of measurements on scale, the study of pasture dynamics in domestic grazing animais has usually focused on a single scale: the daily intake by an individual at plant community level. Scaling variables such as pasture size and number of animals within a group have often been ignored. The reason could be the classical analytical procedure in science, which first tends to build knowledge about elementary units before considering complicated biological processes. However, it is also well known that for many decades, modemising agriculture has been the main concern of animal nutritionists and that they have endeavoured to apply, in ail circumstances, the leading indoor paradigm: optimising the long-term daily individual intake and associated performance by formulating accurate diets (Mertens, 1996). Today, advances in hierarchical theory applied to grazing animais mainly refer to the systems approach proposed by Senft et al. (1987). This approach, inspired by iandscape ecoiogy, was initially concemed with range management at the regional scale (West, 1995), but it is now also applied to swards at finer scales (Prache et al., 1998).

Functional scales and foraging hierarchies

Pasture heterogeneity needs to be defined by identifying the environmental variance inducing changes in the behaviour of grazing animais (e.g., intake rate, movement rate, etc.). Six levels within a hierarchy of behaviours can be defined (Table 1). This functional heterogeneity can also be used to discriminate between spatial elements at the same level, by a change in the rate of the fonction.

Functional heterogeneity differs from statistical heterogeneity primarily because it is scaled to the species and process considered. We may distinguish between statistical and functional heterogeneity by considering the still debated definition of a grazing 'patch' (Bailey et al., 1996). For an herbivore, a patch could be functionally defined as a spatial aggregation of bites in which instantaneous intake and movement rates remain relatively constant over a short period of time. Thus, a patch can be an homogeneous area of grass, one shrub, a group of shrubs in close proximity to each other, or a section of pasture with a

(4)

mixture of grasses and/or shrubs. This definition differs from that frequently used where a patch is a priori defined before the study from soil types and plant cover evaluation, and is assumed to be a specific aggregation of forages. The concept of spatial 'heterogeneity' is scale-dependent. What may be perceived at one scale as heterogeneous may be homogeneous at other scales - the difference between the beetle's and the buffalo's view of the prairie-. This requires space to be defined from the herbivore's point of view (Illius & Hodgson, 1996).

Table 1. The functional scales used to describe the foraging process in domestic herbivores. Each level includes units that herbivores can select from adapted from

Laca & Ortega, 1995 and Bailey et al., 1996).

SQatial level Tem2oral level Defining behaviour Motivation criteria Home range 3-12 months Dispersal and migration Reproduction cycles,

competition with wild herbivores,

thermoregulation,

transhumance, farm

territory organisation

Camp 1-4 weeks Central areas where Plant abundance and

animais drink and rest phenology, paddocking between grazing bouts rules, paddock abiotic

structure, herder seasonal land use rules

Feeding site 1-4 hours Grazing bout Group daily activity pattern, eating and

digestion rates, topography, predation, distance to focal sites (water, sait), design of herded circuit

Patch 1-30 minutes A break in the bout Olfactory and visual eues, sequence with instantaneous plant reorientation to a new palatability, group

location cohesiveness, topography,

phase within a herded grazing circuit

Feeding station 5-100 seconds Front feet placement and Forage type and abundance, change of vegetation intake rate, social

strata interactions

Eating bite 0,5-5 seconds Jaw, tongue and neck Prehensibility of plant movements portions, nutrient and toxin

concentration Italics are mainly related to man-made decisions

(5)

Dynamiques de végétation en système herbager et relations herbe-animal -Séminaire INRA-CIRAD - Montpellier Baillarguet - 31 janvier-!" février 2001

How to design the paddock?

Grazing management impacts mainly the four larger scales, from the home range to the patch. When designing a new paddock, the livestock farmer will first take his land tenure situation into account. Then, he will endeavour to select and fence an area that offers most of the characteristics of a 'camp', in other words a place where a group of herbivores can live for a couple of weeks. Such a place should comprise sufficient feed, water and resting places, correctly distributed in space. From the animals' point of view, 'correctly' means easy to find and remember, without visual breaks or long and costly walks between preferred areas. If the vegetative regrowth, watering and shady places are all located nearby on a flat area, this makes up a nice paddock, especially for cattle and horses. This kind of paddock is typically undergrazed on most of the area and overgrazed in some parts. As livestock farmers do not care to frequently redefine paddock limits, their standpoint is quite different. For them, a correctly designed paddock is one with a spatial structure that induces the animals to be curious and mobile, to prospect everywhere, to compare and finally graze most of the offered edible matter (Meuret et al., 1995).

We will see later how a livestock farrner can change the animals' expectations in forage site value, making them less 'greedy' (as some French shepherds say), using the effect of abiotic factors (distance to water, shade and shelter, salt and supplementary feeds) to make an impact on grazing sites at camp level (Bailey et al., 1996; 1998). A grazing manager must offer the animals a space where they can graze on successive patches. Each sequence making up a feeding site, must have a watering point and resting place nearby, especially when they can be predicted to be less preferred sites. Such paddocks, inciting the animais to compare sites and upgrading their motivation to move, could be designed by livestock farmers who would accept taking time to observe the behaviour of their animais: where are the preferred areas located, is there a trail effect, does the paddock comprise sufficient shady places for resting? Encouraging farmers to become 'designers' of grazing places leads to totally discarding, at the camp level, the concept of 'carrying capacity' if it does not include, locally and momentarily, their own animal's point of view about the relative desirability of grazing sites. What they need to learn is a new professional behaviour, which may reinforce their pasture management skills.

Grazing together

Domestic herbivores are gregarious and the livestock farmer (or land manager) conceives and carries out the grazing management at the group Jevel rather than at the individual level. More than a sum of individuals, a group is an organised entity, in which the dominance status of individuals and social bonds influence the way they forage (Arnold, 1985). Animals recognise, and prefer to associate with, their dam, peers or other members of the groups in which they were bred, the attractiveness of their different partners changing over time (Veissier et al., 1998). Patterns of grazing are shaped by the times of sunrise and sunset (Leclerc & Lécrivain, 1979), or by daytime temperatures, but partners also influence the start and to a lesser extent the end of grazing bouts (Rook & Penning, 1991 ). Spacing patterns are related to the availability and distribution of the resource. Spatial dispersion of animals increases as food availability decreases. If edible vegetation becomes very patchy, flocks and herds tend to subgroup, the spatial location and size of these sub-units then matching food distribution (Lazo, 1992). The initiative to start moving away from one location during grazing is taken by a limited number of usually old or high-ranking animals called 'leaders'. Managers may be able to take advantage of the consistency of leading behaviour within a

(6)

Dynamiques de végétation en système herbager et relations herbea ni mal -Séminaire INRA-CIRAD - Montpellier Baillarguet - 31 janvier-!" février 2001

herd by selecting leaders with desirable grazing patterns to influence the overall herd movements (Bailey et al., 1998).

Social relationships within a group influence individual strategies for dietary choices, exploitation of feeding sites and spatial distribution on the home range (Dumont & Boissy, 1999). Young animals leam to eat navel foods and to avoid harrnful ones by grazing with their dam and peers. This social facilitation is more efficient than individual learning by trial and error. Socially induced diet aversions have been used to prevent animals from grazing toxic plants (Provenza & Burritt, 1991), but these extinguish when averted animals graze with naive partners which eat the toxic plants (Ralphs et al., 1994). Group living is also thought to

be advantageous for individual animals because they can use the feeding sites discovered by the other members of the group (Clark & Mange!, 1984). Increased feeding competition is inherent in group living as soon as vegetation availability is limited. The ingestive behaviour (Thouless, 1990), dietary choices (Appleby, 1980) and growth (Blanc & Thériez, 1998) of low-ranking animais can then be affected by the dominants through direct feeding interference. In heterogeneous pastures, social bonds and food preferences internet to deterrnine the choice of grazing location. Lambs that were reared separately expressed their own dietary preferences and flocks split into subgroups, whereas social influences override individual preferences when animals have always grazed together (Scott et al., 1995). Recent

results suggest that a sheep will move whether alone or with a few peers to a preferred feeding site located close to the core of its social group. Conversely, a sheep will not leave its group to reach a preferred site located further away unless it is followed by several other peers (Dumont & Boissy, 2000). At a larger scale, the high degree of home range fidelity between years in free-ranging herbivores is also partly due to social factors. For example, cattle have been shown to return to the initial locations and associated habitats that they were exposed to by their dams early in life (Howery et al., 1998).

Remembering the pasture

Herbivores have a spatial memory. How they leam and remember the location of patches on a pasture will determine feeding site selection and consequently the impact on plant species and plant community changes. This ability becomes essential when resources are distant or when no visual or olfactory eues are available for locating them (Bailey et al.,

1996). Farmers know from experience that their animals frequently return to some areas where they have previously found preferred forages, but how this is affected by animal, vegetation caver and paddock characteristics is still unclear (Dumont & Petit, 1998).

Spatial memory has been represented in two parts, 'reference' and 'working' memory. Reference memory is the map-like representation of the foraging environment. The spatial configuration and the relative availability of food when animals enter a new paddock are stored in the reference memory. W orking memory is used to remember which parts of the paddock have already been visited so that the animals avoid retuming to sites where food has just been consumed. Working and reference spatial memories are useful at different scales. Working memory can be important at the scale of a feeding station, patch and even feeding site. The use of working memory is limited if a long time elapses between decisions. Reference memory can be used at larger scales, from the patch to the home range. The use of reference memory at smaller scales is limited by the large number of alternatives to be remembered (Bailey et al., 1996).

(7)

Dynamiques de végétation en système herbager et relations herbe-animal

-Séminaire INRA-CIRAD - Montpellier Baillarguet -31 janvier-!" février 2001

Memory testing trials, although conclusive, have often been carried out mainly under

very artificial conditions (Bailey et al., 1989; Edwards et al., 1996). In order to approach

natural conditions, Dumont & Petit (1998) and Dumont et al. (2000) tested at pasture how

groups of sheep leam, without the help of any proximal eue, the distribution of sites with a

preferred food, i.e. bowls containing pellets or small Lolium perenne patches aggregated

within a Dactylis glomerata sward. In these two experiments, ewes quickly memorised not

only the distribution but also the 'value' (dense vs. sparse sites) of these preferred feeding sites, whereas they did not leam the precise location of man y isolated preferred patches. This

can have practical implications, as a preferred plant or mixture of plants will be Jess exploited

(at least at the scale of a few weeks) when evenly dispersed than when aggregated (Dumont et

al., 2000). This spatial memory effect raises new questions in modelling vegetation value for

grazing, especially for the people concemed by grazing interferences with biodiversity structure and dynamics (Blondel, 1995; Valentin-Smith, 1998).

Three grazing models

U.K. bills seen as a huge cafeteria

Background

They are many examples in the literature of how different animal species or grazing

intensities affect the use of vegetation, along with the effect of grazing on plant species dynamics (Gordon & Illius, 1992; Gordon & Lascano, 1994). Armstrong et al. (1997b) used

such knowledge to develop a mode! which aims to predict intake and diet selection by sheep in U.K. hills and montane areas. Managed for extensive livestock production, these areas are

characterised by semi-natural vegetation with low dry matter (DM) production but a high

nature conservation, landscape and recreation value. In the U.K., shepherding has become

increasingly uncommon and sheep are generally left to forage freely almost ail year round.

The authors are concemed with rural development strategies and policies. They contribute to

technology transfer, through the development of foraging sub-models with a strong theoretical basis to be incorporated into Decision Support Tools.

Building the mode/

The monthly growth, senescence, litterfall and standing biomass in 12 major

vegetation communities (7 Calluna-dominated and 5 grass-dominated communities) were

estimated and adjusted to take account of altitude, climatic factors and soil fertility

(Armstrong et al., 1997a). This initial vegetation mode! is run to calculate daily DM

production estimates: sward height, live and dead standing biomass in grass vegetation types,

shoot biomass per unit arca in Calluna-dominated communities. Potential Jaily DM intake from each vegetation type is limited either by the digestibility of biomass or by its availability. For the grass vegetation types, diet digestibility is predicted from an estimate of bite depth, the distribution of live and dead material in the sward, and digestibility values for live and dead biomass. The digestibility of Calluna intake is taken to be that of the current

year's growth of new shoots. Daily DM intake is classically defined as the product of grazing

time, bite weight (g DM/min) and bite rate (bites/min). Maximum grazing time is set at 13

hours. Bite weights on Calluna are calculated from the biomass of the year's new shoots and

the incisor breadth of the sheep. Bite weights on grass vegetation types are calculated from bite depth, distribution of live and dead biomass with height and incisor breadth of the sheep.

(8)

assumed to decrease with sward height in some grass communities, and to be constant when no relationship was available. Bite rates on each Calluna type have been set to obtain the desired degree of selection between Calluna types.

Running the mode[

If the areas of all the vegetation types are the same, the distribution of intake among

• plant communities is driven by the potential digestible DM intake per day in each community.

The model thus explicitly scales up from the level of the bite to that of the patch. Relative preference between vegetation types is directly determined by the mode!, which gives the Decision Support Tools a mechanistic basis. The mode! was partly validated by observations of the seasonal changes in sheep grazing location within an area of 100 ha with six different vegetation types in south-east Scotland (Hunter, 1962). lt would allow the prediction of the

average stocking rates and associated impacts on Calluna, which will satisfy sheep requirements and ensure a sustainable utilisation of the vegetation, for various temperature zones and Grass:Calluna ratios (Table 2).

Table 2. Predictions made by the model of stocking rates and Calluna utilisation rates

which will give sheep minimum sustainable intakes of digestible DM over the year.

Result of simulations for northern Scotland or northern England and western Wales,

with three ratios of land area covered by Agrostis-Festuca to that covered by pioneer

Calluna (Armstrong et al., 1997b).

Grass:Calluna Temperature zone Stocking rate Calluna utilisation(%)

ratio (sheep ha-1)

90:10 N. Scotland 2 3.5 90:10 N. England -W. Wales 9 29.0 50:50 N. Scotland 1 2.5 50:50 N. England - W. Wales 5 19.0 10:90 N. Scotland 0.01 0.1 10:90 N. England -W. Wales 1 3.9

Sorne additional spatial components could, however, be included in the model. In its first version, the distribution of the different vegetation types was assumed to have no effect on foraging behaviour. However, Clarke et al. (1995a) showed that the simple ratio of

Calluna/Grass in a moorland was a very poor predictor of heather utilisation by sheep and of its local dynamics. In the mode!, animals behaved as under free-choice conditions and were assumed to be always able to find the best vegetation types. The results of recent modelling (Turner et al., 1993; Wallis de Vries, 1996) and experimental works (Dumont et al., 1998)

show that, when the rate of encounter of forages is altered, walking costs can be an important factor in foraging decisions. Furthermore, limits in the use of their spatial memory can have consequences on how sheep exploit a mixture of plant species (Dumont et al., 2000). This pleads for the inclusion of recent advances in our understanding of the spatial behaviour of herbivores in research mode!, and for the incorporation of these spatial models into Decision Support Tools.

(9)

Management implications

The aim of the authors was to build a general simulation model, which does not require any widespread, long and expensive field monitoring. The output of this model is the prediction of average stocking rates and associated impacl on Calluna according to its initial abundance, and to climatic and soil conditions (Table 2). However, the use of some specific areas with Calluna is largely underestimated throughout the year. This poses a problem, given the UK's social concerns about conservation of heather landscapes. Specific observations of sheep grazing grass/heather mosaics showed that the utilisation rate of Calluna tended to be higher at the grass-Calluna interface. This can lead to fragmentation of Calluna stands or to a shrinking from the edge of the grass-Calluna interface (Clarke et al., 1995b ). This is an important effect that needs taking into account in landscape and biodiversity conservation. Since the initial version of the model predicts the average utilisation of all the Calluna of one type, it cannot predict the decline of Calluna in this manner. To overcome this problem, the authors propose to consider each Calluna type in terms of likely zones of differential grazing pressure.

Animais seen as site samplers

Background

A landscape grazing distribution model was developed by Bailey et al. (1996) to demonstrate how recent knowledge about cognitive foraging mechanisms in large herbivores,

such as spatial memory, can be integrated with vegetation and abiotic factors to predict grazing patterns. This conceptual individual animal-based mode! focuses on patch and feeding site processes because higher-scale management problems (e.g., overgrazing, undergrazing,

habitat deterioration ... ) and farming practices are related to these levels in the hierarchy. As a better understanding about the spatial and temporal dynamics of landscape use by free-grazing herbivores is critical for ecosystems management (Senft et al., 1987), the mode! is designed to evaluate site specific management practices a priori: e.g., how livestock grazing in an area may force indigenous animais to use marginal habitat, how management tools designed to modify livestock grazing distribution may successfully limit access to critical areas.

The mode! simulates feeding site selection at the beginning of a bout from previous encounters within the camp. The effects of abiotic factors (slope, distance to water, microclimate, predators ... ) are integrated with site forage value, i.e. forage quality, forage quantity and information on secondary compounds obtained while foraging at each site, to define the 'perceived site value'. Abiotic factors, such as slope gradient, are specific of each herbivore species. Site forage value is a function of the relative abundance of nutrients (g N/m2) and of the percentage of secondary compounds. The perceived site value is not useful

in absolute terms. It must be compared to a standard, i.e. the 'reference value'. The reference value is calculated as the moving average of perceived site values encountered during the last four days. Using only recently visited sites as the reference value incorporates the effects of recent foraging experiences. A 4-day moving average is consistent with the time taken by cattle to respond to changes in forage availability (Bailey et al., 1989), but this time duration is not adjusted for other herbivore species, which may have different cognitive abilities. The difference between perceived site value and reference value is termed 'deviation'. Memory of

(10)

Dynamiques de végétation en système herbager et relations herbeanimal -Séminaire INRA-CIRAD - Montpellier Baillarguet - 31 janvier-!°' février 2001

foraging experience at a feeding site diminishes over time, and the cumulative effect of each

encounter is reduced. Memory decay reflects that, when foraging in a variable environment,

animais give a stronger weight to the most recent information because it is more certain. Just

after visiting a feeding site, an herbivore's expectation, or rating of a feeding site, will be more

representative than after several weeks because forage regrowth or grazing by other herbivores can change forage conditions. Memory of the poor sites (negative deviations) decays more slowly than that of the rich sites, because animais are usually risk-adverse (Real,

• 1991). 'Expectation' is the final index, which combines for each feeding site information

decay over time and the difference between perceived site value and the reference value. Expectation is used as the selection criterion.

Running the mode[

Before each bout, the expectation for each feeding site is compared to the current

reference value. Sites with expectations above the reference value are selected. Sites with

expectations near the reference value are either selected or avoided, according to the location

of animais at the beginning of the bout. Selected sites can be abandoned if grazing reduces an

initially high expectation to well below the reference value. Although poor sites are typically

avoided, this model predicts periodic samplings of all patches within a pasture, because

memory of the encounter with a poor site can decay to a level near the current reference

value. Figure 1 is the result from an actual simulation. It gives expectations from an

environment with a nutrient-rich and a nutrient poor site. Initially, the animal had no

expectation of site values and selected the poor site. Then, as the animal only selected the

• nutrient-rich site, the reference value increased to a level equivalent to that of this best site.

Within 20 days, memory of the encounter with the poor site had decayed to a level where the

expectation for this poor site was approximately equal to the reference value, and the poor site

was selected again, beginning a new cycle.

Figure 1. Simulation of expectations from a nutrient-rich and a nutrient-poor site by a Landscape Distribution mode) (Bailey et al., 1996).

1 1 1

,

, , , ,

..

.. ..

-

..

0 2 4 6 8 10 12 14 16 18 20 22 Days - - - Reference value Nutrient-poor site - Nutrient-rich site (multiple visits)

(11)

Dynamiques de végétation en système herbager et relations herbeanimal -Séminaire INRA-CIRAD - Montpellier Baillarguet -31 janvier-!" février 2001

This model predicts day-to-day movement patterns by integrating memory decay and travel costs with forage attributes of feeding sites. In an homogeneous area, such as a sward or a dense mosaic of slightly contrasted feeding sites, perceived site values are near the reference

value and animais will frequently return to the same sites. Conversely, in heterogeneous and

variable environments with distant and dispersed contrasted feeding sites, animais will

periodically visit a great number of feeding sites because they have forgotten their value, and •

hence must update their knowledge. This conceptual model not only predicts grazing distribution, but it is also useful to address various management actions on livestock grazing patterns (Bailey et al., 1998). Abiotic factors such as distance to water, to shelter, to sait or supplementary feed blacks can be strong determinants of herbivore grazing distribution, and act as attractants in which mechanisms based on forage characteristics operate. A grazing

manager could exert a direct impact on these focus sites at camp level, by choosing

appropriate locations. By moving a focus site on a camp, travel distance, or access difficulty to certain feeding sites can be reduced, which may increase their utilisation. Under

free-ranging conditions, strategically placing sait away from water was shown to modify camp

location because animais periodically use the salting area as a resting site (Cassini & Hermitte, 1992). In paddocks where the flocks return daily to the shed, strategically placing the gate of the paddock near the under-utilised feeding sites may upgrade their use (Leclerc & Lécrivain, 1994). This model will enable us to test how such attractants could be used to make under-utilised feeding sites more attractive.

These operating rules may not be really considered as 'advances' in grazing •

management. Sorne of them where pointed out many years ago (e.g., Valentine, 1947), but

they were mostly forgotten in years when improvement of forage quality was the main

objective of range scientists. They were nevertheless cited as central range management skills by authors from the social sciences dealing with farming practices in developing countries (Milleville et al., 1982; Claude et al., 1991 ). Today, such management mies may gain interest

in developed countries as a result of the new environmental stakes of animal-vegetation

studies at a large scale. Sorne of these mies, although qui te unknown to scientists, are already used by experienced shepherds when acting on rangelands (Deffontaines et al., 1989).

The shepherd seen as a Restaurant Chef

Background

A model of a shepherd's circuit organisation was developed by Meuret (1993a,b) to

evaluate how shepherding may interfere with the foraging process in order to upgrade flock

appetite on less desirable patches within a camp. The general objective was to better

understand how to manage landscapes in southern France that have been intensively exploited for centuries and abandoned a few decades ago (Hubert et al., 1995; Deverre et al., 1995). In

that area, shepherding remains relatively common, both with flocks that are transhumed to the

mountains and those that are kept ail year round on the farm territory. The model was

developed so that the technical skills of shepherds could be made understandable and

teachable in shepherding schools, and to assess the result of these skills in grazing

(12)

The model is called MENU and was developed by asking several experienced shepherds while working to record daily in writing and by drawing on a map, information on the design of each of their grazing circuits and to explain precisely the reasons for each intervention (Miellet & Meuret, 1993; Savini et al., 1993). These recordings were combined with direct observation of the activity pattern and spatial behaviour of the flock, together with the continuous recording of the intake activity of an animal within the flock, to estimate its rate of intake (Meuret, 1989; Dumont et al., 1995).

A shepherd is a permanent observer of the animais' appetite, through checking their attitudes, rates of biting and activity patterns. His aim is to use their foreseeable behavioural response in order to impact the flock-vegetation interaction at camp level. 'During a grazing circuit (hich corresponds to a 3 to 4 hours continuous bout), there is a need ta ajfer cantrasted areas, ardered sa as ta frequently revive the animal's appetite' the shepherds say. The design of a grazing circuit may therefore be viewed as an ordered sequence of foraging patches (Laca & Ortega, 1995; Bailey et al., 1996). The model is based on two rules: i) create feeding synergetic sequences within the bout; ii) follow spontaneous behavioural trends while developing flock confidence in the shepherd's interventions.

Using the mode/

This model is an organisational model that helps a shepherd make the appropriate decisions at the level of a half-day grazing route, corresponding to a grazing bout. When intake must be stimulated on a particular patch ('target-zone') with rough and less palatable material (e.g., a patch to be cleared of scrub), the shepherd must detect and use complementary, sometimes contiguous, patches within the feeding site. The different patches could play six distinct roles during a circuit (Fig. 2 left), and the instantaneous value of each patch for a specific phase during the foraging bout is assessed according to two criteria: the relative abundance of edible material and the 'instantaneous palatability' (Sauvant et al.,

1996) of the patch for that flock.

A 'moderation-patch', with abundant but not highly palatable plants, could be used during 20 to 30 min at the beginning of the circuit in order to reduce an excessive appetite of the flock. In contrast, an 'appetite promoting' patch, offering a highly palatable but rare resource, could be used to stimulate a flock having a low initial appetite. The target-patch, with medium plant abundance and palatability, is then used as 'main-course'. When the animais are losing interest in this patch, a 'booster-patch', or a booster-action, has to be introduced during 15 to 20 min to add diversity and revive the appetite. There are four ways of boosting the animais' appetite: 1. By offering a patch with low abundance but high instantaneous palatability; 2. Conversely, by offering a patch with medium abundance and very low instantaneous palatability; 3. By watering the flock; and 4. By offering sait licks. If

the booster-action is successful, another target-patch, with slightly better instantaneous palatability than the main course-patch, is used for a 'second course'. At the end of the circuit, a 'dessert-patch', with high plant abundance and palatability, can be used during 20 to 30 min on condition that this phase cannot be predicted by the animais.

Figure 2 (right) gives an example of a bout kinetic pattern for a lactating goat grazing with the goat herd in a Quercus pubescens coppice and herded by an experienced shepherd using the model.

(13)

Dynamiques de végétation en système herbager et relations herbe-animal -Séminaire INRA-CIRAD - Montpellier Baillarguet -31 janvier-!" février 2001

HrTder's interventions on the h rTd

MC B SC D

50 100 150

lnstant111eous Palatooility + time (rrin)

200

Figure 2. Conceptual flow diagram of the MENU model (left) (Meuret, 1993a,b).

M: Moderation patch - A: Appetite promotion - MC: Main course - B: Booster - SC: Second course - D: Dessert. Plain arrows indicate how a shepherd may detect and mobilise distinct contiguous patches at feeding site level to create synergetic effects during a half-day grazing circuit (dotted arrows are alternatives). The objective is to constantly revive the animais' appetite on a less desirable patch, called 'target-zone'. An example of intake rate recording (right) shows how an individual bout kinetic may be

structured by interventions made by an experienced shepherd using this organisational

model to upgrade intake (Meuret, 1996).

With such herded lactating goats, 60% of dry matter (DM) intake during a bout occurs during strong accelerations of the intake rate, i.e. an increase of more than 4 g DM intake/min in less than 20 min (Meuret et al., 1994). In 52% of cases, these within-bout accelerations were directly impelled by the shepherd while moving the herd during the circuit from one grazing patch to another. In addition, the heterogeneity of the pasture provoked 62% of the accelerations as the animais moved from one vegetation caver to another, with or without direct intervention by the shepherd, and resulted in very high intake :around 1,3 kg DM/bout.

An experiment was conducted in a shepherding school on the relevance of using MENU

as a teaching tool for inexperienced shepherds (Agreil & Meuret, unpublished data). During 48 successive days in the summer, three persons herded each in tum the same herd of 40 lactating goats on successive grazing routes ovcr a half-day period. One persan was an experienced shepherd (S 1) while the other two were not (S2 & S3). The herd' s milk yield was

recorded twice daily, at 7 a.m. and 5 p.m. In summer, milk yield is strongly correlated to the amount and quality of feeding during the previous 12 hours. This means that the herding skills of one shepherd during one grazing route could be assessed from the amount of milk produced at the next milking.

The first three days were considered as the habituation period and the analysis was carried out on the kinetic of moming milk production over the last 45 days. For each of the three shepherds, a statistical NAs was produced by counting the number of times the skill of Shepherd SA was lower than that of S8 who herded on the previous moming. The absence of

(14)

any difference between the shepherds' performances is taken as the null hypothesis. The statistical procedure consists of computing these statistics for a large number of random permutations of the data (1000 times). For each shepherd, the probability of obtaining the observed value is calculated from the statistical distribution on the null hypothesis (Manly,

1997).

During an initial 21-day period, the skill of S3 was significantly lower than that of S 1

• _{who herded the day before and of S2 who herded on the following day}(N31=6 for 7 transitions S1/S3 and N23=1 for 7 transitions S3/S2 , p<0.05). In contrast, over the same period, no

significant differences were observed between the 'skill' of S1 and that of S2 who herded on

the previous day (N 12=3 for 7 transitions S2/S 1, p>0.05). During that period S2 managed to exploit the full sequence of the MENU zones while S₃failed to use both the 'moderation' and the 'booster' patches. In the last 24 days, there were no significant differences between the skills of the three shepherds (N31=5 for 8 transitions S1/S3, N23=2 and N 12=4, p>0.05), since the leaming process of S3 was completed.

A shepherd's grazing management could take full advantage of a patchy environment and could emphasise the concept of instantaneous palatability of feed during a bout. The model confirms that the feed value of such an environment results largely from proper feeding management, at feeding site and patch levels, and that the appetite during a bout can be upgraded. Planning for synergetic phases within bouts may be an efficient way of considering how a farm territory (consisting of several camps) should be shaped over several years. For example, the strategic implantation of small sainfoin (Onobrychis sativa) plots near Jess • desirable patches, can tum these into 'boosters' or 'appetite promoters' in the circuit. This can upgrade the intake of coarse resources by as much as 20 % during the patch utilisation period (Meuret, 1997).

Experienced shepherds observe their flock's behaviour permanently. They tend to evaluate initial hunger, intermediate disaffection for food and signs of satiety. Those who succeed in designing a menu would be able to generate a 'built-up' pastoral value from an heterogeneous and variable vegetation (Meuret, 1996), a perspective which is not confined to European conditions. In the US, herding cattle to under-utilised areas has been recommended for over 40 years (Sklovin, 1957), but the common cow-boy way of herding did not develop herd confidence in the herder, and cattle often retumed to preferred sites just after being herded away (Bailey et al., 1998). Recently, herding was noticed to efficiently keep cattle away from riparian areas in Idaho, using a Peruvian 'sheep-guy' (Butler, 1998).

Three points of view on grazing management

The previous models are all concemed with grazing management and focus on what is usually called 'decision making'. However, modelling approaches, model outputs, as well as the spatial and temporal scales of processes studied are different. Such differences are primarily linked to the scientific background and local social stakes the authors are dealing ,. with. Armstrong et al. ( 1997b) are mostly ecologists, concerned with land planning at a very large scale, in a country where the pressure is now on the role of agriculture regarding the new environmental and urban recreational issues. The authors have extensive knowledge of vegetation dynamics in relation to climate and soil conditions, and on herbivore selectivity and intake. They incorporate this knowledge into a predictive model for land planners at the national/country scale. The output of their model is the prediction of average stocking rates

(15)

Dynamiques de végétation en système herbager et relations herbeanimal

-Séminaire INRA-CIRAD - Montpellier Baillarguet - 31 janvier-!" février 2001

environmental value in the UK hills. Bailey et al. (1996) are ecologists and range scientists,

dealing with range managers in charge of both livestock and wildlife. They face the classical question, which is not particular to the US, of competition for land and resources among herbivores (and between farmers and hunters). Their main objective is to take advantage of the most recent advances in the cognition of herbivores to show how abiotic factors,

especially those resulting from specific management at the farm Jevel, could be used as a

'tool' to impact on grazing spatial patterns of herbivores. Meuret (1993a,b) is an animal nutritionist working within a multidisciplinary group that includes social scientists and ecologists. The group approaches rangelands and environment issues through a better understanding of farmer practices and rural policies. In France, as in most European countries,

new environrnentally-related uses of the land has revived interest in shepherding. This is the reason why this author developed an operational tool to help make understandable and teachable the technical skill of experienced shepherds in designing their grazing circuits.

The common objective of these authors, i.e. decision making in guiding animais at pasture, leads them to deal with the question of re-scaling in assessing the foraging process (Table 1 ). Biological processes are scaled-up from the bite to the patch in Armstrong et al. (1997b ), from the patch to the feeding site in Meuret (1993a,b) and from the feeding site to the camp in Bailey et al. (1996). Time and space scaling is a common interest, although they are not incorporated to the same extent in the three models. The only spatial component in Armstrong et al. ( 1997b) is the distribution of offtake/intake across vegetation communities, but the authors are looking forward to the day when spatial models will be available that they can incorporate into their Decision Support Tool. Conversely, spatial and temporal components (via leaming processes and spatial memory) are of major importance in both the building and the outputs of Bailey et al.' s (1996) model. Space is also of main concem as well as the temporal man-made organisation of a feeding bout in the Meuret ( 1993a,b) model, which aims to provide locally a shepherd with a functional picture of his camp. When considering grazing as an ecological process, the analysis of Jast two models shows that research into the temporal dimension of processes leads to also consider the spatial aspects. Finally, it is to be regretted that none of these models integrate the social aspects of animal grazing. This could be a future objective of grazing studies, given that farmers focus their grazing management on groups of animais, or even on a whole herd or flock.

Conclusion

New, multiple, and sometimes conflicting management objectives for grazinglands have caused for around a decade some managers and researchers to focus on a better understanding of the grazing process. Yet the world-wide, dominant, 'carrying capacity' paradigm still prevails. ln seeking to optimise seasonal adjustment between herbivore requirements and forage abundance and quality, this paradigm implicitly upholds a 'ranching model' that promotes the sedentarisation of livestock farmers (Niamir-Fuller, 1995). lt ignores one of the major aspects of herbivore behaviour at pasture, i.e. mobility. Although based on small scale studies, its recommendations target whole landscapes or regional scales,. lt considers weather, terrain conditions and local pastoral practices as grazing 'constraints' that need to be minimised to ensure the precise implementation of land use recommendations. This is a kind of grazing industrialisation, which may be relevant in some parts of the world where livestock producers want to get rid of the trial-and-error process (Stuth et al., 1999).

(16)

-Séminaire INRA-CIRAD - Montpellier Baillarguet -31 janvier-!" février 2001

Given the new and multiple objectives for grazing, we argue that Jivestock farming will certainly remain a trial-and-error process for a long time. Different social concems surface about the animal-vegetation interaction in the local grazing of herds. The resulting local pasture 'value' may take into consideration biodiversity conservation, wild fire prevention, meat or cheese quality, water quality preservation, etc.. Since these concerns are still

diversifying, modelling efforts must focus first on herbivore behavioural responses when

pasturing conditions change. In the medium term, such as the 5 years of an

agri-environrnental contract, pastoral systems will not be viewed as stable systems. Small

adjustments and sometimes strong re-orientations of farmer practices will be the input

variables in man-animal-vegetation interaction models. Animais could be then considered as

'mediators' of the vegetation value. Hopefully, new advances in the knowledge of the

cognitive abilities of herbivores, which help them to cope with heterogeneity and variability, open the way to a new generation of decision models sensu lato.

References

Allen, T.F.H., Starr, T.B. (dir.), 1982. Hierarchy : perspectives for ecological complexity. The University of Chicago Press, Chicago & London, 310 pp.

Appleby, M.C., 1980. Social rank and food access in red deer stags. Behaviour 74: 294-309 Armstrong, H.M., I.J. Gordon, S.A. Grant, N.J. Hutchings, J.A. Milne & A.R. Sibbald, 1997a.

A model of the grazing of hill vegetation by sheep in the UK. I. The prediction of

vegetation biomass. J. Appl. Ecol. 34: 166-185

Armstrong, H.M., I.J. Gordon, N.J. Hutchings, A.W. Illius, J.A. Milne & A.R. Sibbald,

1997b. A model of the grazing of hill vegetation by sheep in the UK. II. The prediction

of offtake by sheep. J. Appl. Ecol. 34: 186-207

Arnold, G.W., 1985. Associations and social behaviour. In: Fraser, A.F. (Ed.), Ethology of farm animais, Elsevier Scientific Publishing Company, Amsterdam, pp 233-248

Auger, P., J. Baudry & F. Fournier (dir.), 1992. Hiérarchies et échelles en écologie. Naturalia pubs., 300 pp.

Bailey, D.W., L.R. Rittenhouse, R.H. Hart, D.M. Swift & R.W. Richard, 1989. Association of

relative food availabilities and locations by cattle. J. Range Manage. 42: 480-482

Bailey, D.W., J.E. Gross, E.A. Laca, L.R. Rittenhouse, M.B. Coughenour, D.M. Swift & P.L.

Sims, 1996. Mechanisms that result in large herbivore grazing patterns. J. Range

Manage.49: 386-400

Bailey, D.W., B. Dumont & M.F. WallisDeVries, 1998. Utilization of heterogeneous grasslands by domestic herbivores: Theory to management. Ann. Zootech. 47: 321-333 Blanc, F. & M. Thériez, 1998. Effects of stocking density on the behaviour and growth of

farmed red deer hinds. Appl. Anim. Behav. Sei. 56:297-307

Blondel, J., 1995. Du théorique au concret : la biologie de la conservation. Nature Science,

Sociétés, Hors-série ' Recréer la Nature ': 10-18.

Butler, P., 1998. Herding cows like sheep on the National Forest using a Peruvian herder (abstr.) 51st Ann. Mtg. Soc. Range Manage.: 23

Cassini, M.H. & G. Hermitte, 1992. Patterns of environrnental use by cattle and consumption of food blocks. Appl. Anim. Behav. Sei. 32: 297-312

Clark, C.W. & M. Mangel, 1984. Foraging and flocking strategies: information in an

(17)

Clarke, J.L., D. Welch & I.J. Gordon, 1995a. The influence of vegetation pattern on the

grazing of heather moorland by red deer and sheep. I. The location of animais on

grass/heather mosaics. J. Appl. Eco!. 32: 166-176

Clarke, J.L., D. Welch & I.J. Gordon, 1995b. The influence of vegetation pattern on the

grazing of heather moorland by red deer and sheep. Il. The impact on heather. J. Appl.

Eco!. 32: 177-186

Claude, J., Grouzis, M., Milleville, P., 1991. Un espace sahélien : la Mare d'Oursi (Bukina

Faso). Paris, ORS TOM, 241 pp.

Dale, V.H., Gardner, R.H., Turner, M.G., (eds.), 1989. Predicting across scales: theory development and testing. Landscape Ecology, 3: 147-252.

Deffontaines, J.P., Landais, E., Savini, I., 1989. L'espace d'un berger: pratiques pastorales dans les Ecrins., Garabédian, D. (ed), Vidéofilm, Paris, Inra-SAD/ENS St Cloud (prod.): 60 min.

Deverre, C., B. Hubert & M. Meuret, 1995. The know-how of livestock farmers challenged by new objectives for European farming: I. Rangelands usages, greatness and decay. In : West N.E. (editor), Rangelands in a sustainable biosphere. Proc. Vth Int. Range. Cong., Salt Lake City, Utah, July 23-28, 1995, I: 115-116

Dumont, B., M. Meuret & M. Prud'hon, 1995. Direct observation of biting for studying

grazing behavior of goats and llamas on garrigue rangelands. Small. Rumin. Res. 16 :

27-35

Dumont, B. & M. Petit, 1998. Spatial memory of sheep at pasture. Appl. Anim. Behav. Sei. 60: 43-53

Dumont, B., A. Dutronc & M. Petit, 1998. How readily will sheep walk for a preferred forage? J. Anim. Sei. 76: 965-971

Dumont, B. & A. Boissy, 1999. Relations sociales et comportement alimentaire au pâturage. INRA Prod. Anim. 12: 3-10

Dumont, B., J.F. Maillard & M. Petit, 2000. The effect of the spatial distribution of plant species within the sward on the searching success of sheep when grazing. Grass For. Sei. 55 (in press)

Dumont, B. & A. Boissy, 2000. Grazing behaviour of sheep in a situation of conflict between feeding and social motivations. Behav. Proc. 49 (in press)

Edwards, G.R., J.A. Newman, A.J. Parsons & J.R. Krebs, 1996. The use of spatial memory by

grazing animais to locate food patches in spatially heterogeneous environments: an example with sheep. Appl. Anim. Behav. Sei. 50: 147-160

Gordon, I.J. & A.W. Illius, 1992. Foraging strategy: from monoculture to mosaic. In: Speedy, A.W. (Ed.) Progress in sheep and goat research, CAB International, Wallingford, pp

153-177

Gordon, I.J. & C. Lascano, 1994. Foraging strategies of ruminant livestock on intensively managed grasslands: potential and constraints. ln: Baker, M.J. (Ed.) Grasslands for our World, SIR Publishing, Wellington, pp 208-216

Hodgson, J. & A.W. Illius (editors), 1996. The ecology and management of grazing systems. CAB International, Wallingford, New-York, 466 pp.

Howery, L.D., F.D. Provenza, R.E. Banner & C.B. Scott, 1998. Social arid environmental factors influence cattle distribution on rangeland. Appl. Anim. Behav. Sei. 55: 231-244

(18)

Hubert B., Deverre C., Meuret M., 1995. The know-how of livestock farmers challenged by

new objectives for European farming: II. Reassigning rangelands to new,

environmental-related usages. In : West N.E. (editor), Rangelands in a sustainable

biosphere. Proc. Vth Int. Range. Cong., Salt Lake City, Utah, July 23-28, 1995, I: 251-252

Hunter, R.F., 1962. Hill sheep and their pasture: a study of sheep grazing in south-east

Scotland. J. Ecol. 50: 651-680

Illius, A.W. & J. Hodgson, 1996. Progress in understanding the ecology and management of grazing systems. In: The ecology and management of grazing systems, Hodgson, J. &

A.W. Illius (editors), CAB International: 429-458

Laca, E.A. & I.M. Ortega, 1995. Integrating foraging mechanisms across spatial and temporal scales. In: West N.E. (editor), Rangelands in a sustainable biosphere. Proc. Vth Int.

Range. Cong., Salt Lake City, Utah, July 23-28, 1995, II: 129-132

Lazo, A., 1992. Facteurs déterminants du comportement grégaire de bovins retournés à l'état

sauvage. Rev. Ecol. (Terre Vie) 47: 51-66

Leclerc, B. & E. Lécrivain, 1979. Etude du comportement d'ovins domestiques en élevage

extensif sur le Causse du Larzac. Thèse 3e cycle, Univ. Rennes, 165 pp.

Leclerc, B. & E. Lécrivain, 1994. Incidence du retour quotidien en chèvrerie sur le

comportement alimentaire et spatial de caprins dans un taillis. Ann. Zootech. 43: 295

Manly, B.F.J., 1997. Randomization, bootstrap and Monte Carlo methods in biology.

Chapman & Hall, London, 399pp.

Mertens, D.R., 1996. Methods in modelling feeding behavior and intake in herbivores. Ann.

Zootech. 45 (suppl): 153-164

Meuret, M., 1989. Feuillages, fromages et flux ingéré. PhD Diss., Gembloux Agronomy Fac.,

249 pp.

Meuret, M., 1993a. Piloter l'ingestion au pâturage. Et. Rech. Syst. Agr. Dev. 27: 161-198 Meuret, M., 1993b. Les règles de l'Art : garder des troupeaux au pâturage. Et. Rech. Syst.

Agr. Dev. 27: 199-216

Meuret, M., C. Viaux & J. Chadoeuf, 1994. Land heterogeneity stimulates intake rate during grazing trips. Ann. Zootech. 43: 296

Meuret, M., S. Bellon, G. Guerin & G. Hanus, 1995. Faire pâturer sur parcours. Renc. Rech. Rumin. 2: 27-36

Meuret, M., 1996. Organizing a grazing route to motivate intake on coarse resources. Ann. Zootech. 45 (suppl): 87-88

Meuret, M., 1997. How do I cope with that bush : optimizing less palatable feeds at pasture using the MENU model. In : Lindberg J.E. et al. (editors), Recent advances in small ruminant nutrition. Option Medit. A43: 53-57

Miellet, P. & M. Meuret, 1993. Savoir faire pâturer en SIG. Mappemonde. 2/93: 12-17 Milleville, P., J. Combes & J Marchal, 1982. Systèmes d'élevage sahéliens de l'Oudalan:

étude de cas. ORSTOM Ouagadougou, Rapp. multigr. 129 pp.

Niamir-Fuller, M., 1995. A new paradigm in African range management policy and Pratice.

In: West N.E. ( editor), Rangelands in a sustainable biosphere. Proc. Vth Int. Range.

Cong., Salt Lake City, Utah, July 23-28, 1995, II: 79-83

Prache, S., I.J. Gordon & A.J. Rook, 1998. Foraging behaviour and diet selection in domestic

(19)

-Séminaire INRA-CIRAD -Montpellier Baillarguet - 31 janvier-!" février 2001

Provenza, F.D. & E.A. Burritt, 1991. Socially induced diet preference ameliorates conditioned

food aversion in lambs. Appl. Anim. Behav. Sei. 31: 229-236

Ralphs, M.H., D. Graham & L.F. James, 1994. Social facilitation influences cattle to graze

locoweed. J. Range Manage. 47: 123-126

Real, L.A., 1991. Animal choice behavior and the evolution of cognitive architecture. Sei.

253: 980-986

Rook, A.J. & P.D. Penning, 1991. Synchronisation of eating, ruminating and idling activity

by grazing sheep. Appl. Anim. Behav. Sei. 32: 157-166

Sauvant, D., R. Baumont & P. Faverdin, 1996. Development of a mechanistic mode! of intake

and chewing activities of sheep. J. Anim. Sei. 74: 2785-2802

Savani, 1., E. Landais, P. Thinon & J.P. Deffontaines, 1993. L'organisation de l'espace

pastoral : des concepts et des représentations construites à dire d'expert dans une perspective de modélisation. In : Pratiques d'élevage extensif: identifier, modéliser,

évaluer, Landais E. (editor), Et. Rech. Syst. Agraires Dév., 27: 137-160

Scott, C.B., F.D. Provenza & R.E. Banner, 1995. Dietary habits and social interactions affect choice of feeding location by sheep. Appl. Anim. Behav. Sei. 45: 225-237

Senft, R.L., M.B. Coughenour, D.W. Bailey, L.R. Rittenhouse, O.E. Sala & D.M. Swift,

1987. Large herbivore foraging and ecological hierarchies. BioScience 37: 789-799

Sklovin, J.M., 1957. Range riding - the key to range management. J. Range Manage. 10:

269-271

Stuth, J. W., M. Freer, H. Dove & R.K. Lyons, 1999. Nutritional management for

free-ranging livestock. In: Nutritional ecology of hervivores, Proc. Vth Inter. Symp. Nutr. Herbiv., Jung H-J. G. & J.C. Fahey, jr (editors), The ASAS pub., Savoy, Illinois, 696-751

Thouless, C.R., 1990. Feeding competition between grazing red deer hinds. Anim. Behav. 40:

105-111

Turner, M.G., Y. Wu, W.H. Romme & L.L. Wallace, 1993. A landscape simulation mode! of

winter foraging by large ungulates. Eco!. Mode!. 69: 163-184

Valentine, K.A., 1947. Distance from water as a factor in grazing capacity of rangelands. J.

Forest. 45: 749-754

Valentin-Smith, G. (coord), 1998. Guide méthodologique des documents d'objectifs Natura

-2000. Réserves Naturelles de France / Atelier technique des Espaces Naturels,

Quétigny. 144 pp

Veissier, I., A. Boissy, R. Nowak, P. Orgeur & P. Poindron, 1998. Ontogeny of social

awareness in domestic herbivores. Appl. Anim. Behav. Sei. 57: 233-245

Wallis de Vries, M.F., 1996. Effects of resource distribution patterns on ungulate foraging

behaviour: a modelling approach. For. Ecol. Manage. 88: 167-177.

West, N.E. (editor), 1995. Rangelands in a sustainable biosphere. Proc. Vth Int. Range. Cong.,

Salt Lake City, Utah, July 23-28, Soc. Range Manage. Pub, Denver, Colorado, II, 202

pp.

Our thanks to Laurence de Bonneval for checking the text and enjoying our 'appetizers' in the