Agronomy of Grassland Systems C. J. Pearson R. L. Ison - Library of

Second Edition

Agronomy of Grassland Systems

C.J.Pearson R.L. IsonTheUniversity of Queensland, Australia TheOpenUniversity,MiltonKeynes, UK

PUBLISHED BY THE PRESS SYNDICATE OF THE UNIVERSITY OF CAMBRIDGE

The Pitt Building, Trumpington Street, Cambridge CBá "RP, UnitedKingdom

CAMBRIDGE UNIVERSITY PRESS

The Edinburgh Building, Cambridge CBá áRU, UnitedKingdomãòWest áòth Street, NewYork, NY "òò""-ãá"", USA

"ò StamfordRoad, Oakleigh,Melbourne â"åå, Australia

#CambridgeUniversity Press "ñðæ, "ññæ

This book is in copyright. Subject to statutory exceptionand to the provisions of relevant collective licensing agreements,

no reproduction of any part may take place withoutthe written permission of CambridgeUniversity Press.

F|rst published "ñðæSecond edition published "ññæ

Printed in theUnitedKingdom at theUniversity Press, Cambridge

Typeset in ñ34/"ápt Baskerville [PN]

A catalogue record for this book is available from theBritish Library

Library of Congress Cataloguing in Publication dataPearson, C. J.

Agronomy of grassland systems / C.J. Pearson andR.L. Ison. ^ ánd ed.p. cm.

Includes bibliographical references (p. ) and index.ISBN ò äá" äåò"ò " (hardback). ^ ISBN ò äá" äåððñ æ (pbk.)

". Pastures. á. Grassses. â. Rangemanagement. I. Ison, R. L. II. T|tle.SB"ññ.Pâæ "ññæ

åââ.á'òá ^ dcá" ñå-ãääæá CIP

ISBN ò äá" äåò"ò " hardbackISBN ò äá" äåððñ æ paperback

Contents

Preface xi

1. Overview:perspectives on grassland systems "1.1. The social construction of grassland systems á1.2. Grassland issues or problems ã1.3. Systems thinking ð1.4. Representing grassland systems ñ1.4.1.Modelling "ò1.5. Purposes of grassland systems "ä1.6. Further reading "æ

2. The emergence of grassland systems "ð2.1. Biological and ecological models that give rise tograssland systems áò2.1.1. Clementsian succession áò2.1.2. New' thinking á"2.1.3.The origins of `range science' áâ2.1.4.Traditional or `mainstream' approaches to grasslandsystem development áâ2.2. Technologies that reveal and conceal â case study áã2.2.1. Increasing productivity áä2.2.2.Thirty to ¢fty years later áæ2.2.3. Stepping outside our traditions áæ2.2.4. Responsibility in technology design áð2.3. Present grassland systems áð2.3.1. Global distribution of grassland systems áð2.3.2. Animal industries âò2.3.3. Grassland plant domestication ââ2.4. The agronomicmanipulation of grasslandsystems âä2.5. Further reading âå

3. Generation âæ3.1. Sources of seed âæ3.2. Sown seed ãá3.2.1.Time, rate and depth of sowing ãâ3.2.2.Method of cultivation and sowing ãã3.2.3.Nutrition and seed treatment ãä3.2.4. Grazing ãå3.2.5.Herbicides ãæ3.2.6. Burning ãæ3.3. Dynamics of the seed bank ãð3.3.1. Seasonality of seed banks ãð3.3.2.Dormant and active seed ãñ3.4. Germination äâ3.5. Vegetative generation ää3.6. Systemsmodelling äå3.7. Further reading äñ

4. Vegetative growth åò4.1. Emergence å"4.1.1. Seedling type åá4.1.2. Seed size and genotype åá4.1.3. Seed bed åâ4.2. Establishment åâ4.3. Forms of development åä4.3.1. Apex position åä4.3.2. Leaves åä4.3.3. Branches or tillers åæ4.3.4. Roots åð4.4. Growth åñ4.4.1. Interception of radiation æò4.4.2.Utilization of radiation æ"4.4.3. Carbon balance æ"4.4.4. E¤ciency of net primary productivity æá4.5. Regrowth æâ4.6. Environmental e¡ects on growth æã4.6.1.Water availability æã4.6.2.Temperature æä4.6.3.Nutrition æä4.6.4. Fire æä4.7. Competition æå4.7.1.Thinning æå4.7.2. Competitiveness and growth rate æå4.7.3.Weeds ææ4.7.4.Trees ææ4.8. Grazing e¡ects on growth and development æð4.9. Long-term changes in species composition ðò4.10.Modelling plant growth ðò4.11. Further reading ðá

5. Flowering and seed production ðâ5.1. Juvenility ðâ5.2.Morphological changes at £owering ðã5.3. Flowering ðä5.3.1. Environmental controls of £owering ðä5.3.2. Autonomous £owering ðð5.3.3.Development of the in£orescence ðð5.4. Fertilization and seed formation ðñ5.4.1. Breeding systems ðñ5.4.2. Anthesis and fertilization ñ"5.4.3. Seed production ñá5.5. Implications for grassland growth andmanagement ñã5.5.1. Flowering and growth rate ñã5.5.2. Flowering and quality ñä5.5.3. Selection of cultivars ñä5.5.4. Sowing time ñå

5.5.5.Management by defoliation ñå5.5.6. Fertilizer application ñæ5.5.7.Diseases and pests associatedwith £owering ñæ5.6. Systemsmodelling ñæ5.7. Further reading ñð

6. Mineral nutrition ññ6.1. The nutrient network ññ6.2. Soil fauna and £ora "òá6.2.1. Rhizosphere organisms "òá6.3. Uptake by plants "òä6.3.1.Uptake of inorganic nutrients "òä6.3.2. Cation anion balance "òä6.3.3. Soil acidi¢cation by legume pastures "òå6.4. Distribution of nutrients within the plant "òæ6.5. Senescence and element release from deadmaterial "òæ6.5.1.Herbage death "òð6.5.2.Dung and urine "òð6.5.3.Decomposition "òð6.6. Animal intake ""ò6.7. Losses from the system ""ò6.8. Element de¢ciency and fertilizer needs """6.9. Implications for grassland growth andmanagement ""á6.9.1. Pesticides and herbicides ""á6.9.2.Types ofN, P,K, S fertilizer ""â6.9.3. E¤ciency of utilization of fertilizer ""ã6.9.4. Amount of fertilizer required ""ã6.9.5.Timing of fertilizing ""å6.9.6.Type andmethod of application of fertilizer ""å6.9.7. Grasslandmaintenance and restoration ""å6.10.Models for nutrient management ""æ6.11. Further reading ""ð

7. Herbage quality and animal intake ""ñ7.1. The basis of herbage quality ""ñ7.1.1. Chemical composition "áò7.1.2. Cell structure "á"7.1.3. Variation among species "áá7.1.4. Ageing "áã7.1.5. Environment "áä7.1.6. Palatibility and edibility "áå7.2. Injurious substances "áå7.3. Sward structure "áñ7.4. Herbage availability: grazing pressure "âò7.5. Animal type and productivity "âá7.5.1. Animal type "âá7.5.2.Utilization of energy "ââ7.5.3. Intake and productivity "ââ

7.6. Implications for management "âã7.6.1. Production systems "âã7.6.2. Research and development towards managingnutritive value "âð7.7. Systems approaches "âñ7.8. Further reading "âñ

8. Grassland ^ animal interactions andmanagement "ãò8.1. Animal e¡ects on grassland "ãò8.1.1. Selective grazing "ãò8.1.2. Pulling "ã"8.1.3.Treading and poaching "ãá8.1.4. Fouling "ãá8.2. Grazingmanagement systems "ãâ8.2.1. Production per animal and per area "ãâ8.2.2.Herd experience, composition and timing ofoperations "ãä8.2.3. Grazing interval: set stocking and block grazing "ãå8.2.4. Regional grazing systems "ãð8.3. Conservation and supplementation "ãñ8.3.1. Carryover feed "ãñ8.3.2.Hay and silage "äò8.3.3. Crop residues and by-products "äá8.3.4. Fodder crops "äâ8.3.5. Supplements "äã8.4. `Feed year' planning "ää8.5. E¤ciency of livestock production "äæ8.5.1. Energy budgeting "äæ8.5.2. Legume versus nitrogen-fertilized grass systems "äñ8.5.3. E¤ciency of use of support energy "äñ8.6. Systems perspectives "äñ8.7. Further reading "åò

9. Grassland systems design "å"9.1. Grasslands in farming systems "å"9.1.1. Farming systems perspective "å"9.1.2. Integration of grasslands and cropping "åá9.1.3. Economic analysis "åå9.1.4. Gender and culture "åñ9.1.5. Farm technology "æò9.2. Grasslands and other forms of humanactivity "æ"9.2.1. Agro-industrial systems for using grassland plants "æá9.2.2. Grasslands in aesthetics and conservation "æã9.3. Design of future grassland systems "æã9.3.1. Participation in designing "æä9.3.2. Future scenarios "ææ9.4. Regional design "ææ9.5. National issues "ðò

Contents

9.5.1.National purposes for grassland systems "ðò9.5.2. Grassland research "ðá9.6. Global design "ðâ9.6.1.Human health and food supply "ðâ9.6.2. Biodiversity "ðä9.6.3. Sustaining organic matter and nutrients "ðä9.6.4. Responding to climate change "ðå9.7. Conclusion "ðå9.8. Further reading "ðæ

References "ðð

Index á"ð

Contents

Chapter 1

Overview: perspectives ongrassland systems

Recent understandings from the `new biology' andother areas of research are challenging the ways inwhich grassland systems have historically been `seen'.When we say `seen' wemean the way we think about,and use language to talk about `grassland systems'.Most texts on agronomy or grasslands will not addressthese issues. Historically they have taken theperspective that certain `real types' of grasslandsystems exist that can be described and classi¢ed andthat these understandings can be used as a basis forintervention and change.We start from a di¡erentperspective. In recent years there has been growingacceptance of several complex, seemingly intransigentproblems, often collectively referred to asenvironmental problems, or problems of sustainabledevelopment. Speci¢c examples, related to grasslands,include the loss of organicmatter from soils, loss ofbiodiversity, rangeland degradation and deserti¢cationand the contribution of grasslands to `greenhousee¡ects'. The perspective of this book is that if we are tomanage these complex issues, there is a need ¢rstly tolook at howwe think about the issues and the ways inwhich historically we have chosen tomanage.Werecognize that many of the issues arise because of ourown historical practices, and thatmore of the samepractices will not necessarily be a good thing.

The concept of perspective is important.Werecognize from daily life that none of us have the samehistory, that we all have di¡erent and uniqueexperiences, and as a consequence we all have our ownperspective on things and events. Historically, genderperspectives have often been ignored in the`construction' of grassland systems; fortunately newapproaches are available (Chapter ñ; CIAT, "ññ").Of course, because we live in families and particularcultures we share similar perspectives aboutmanythings. These cultural perspectives often determinehowwe see and interpret our experiences and lead us tohave particular understandings and to name things in

particular ways. It is thus common to talk about atransport system or a grassland system as if everyonewould agree, followingmore detailed attention, onwhat these were. This is clearly not the case.

Human beings live with the strong wish toexplain.Whenwe, as humans, generate explanations ofphenomenawe often refer to these as theories. Di¡erenttheories, however, carry with them di¡erentperspectives or viewpoints. For example, the theoryof plant succession, accredited to Clements ("ñ"å), hasledmany ecologists, agronomists, and landmanagersand administrators to see grasslands in a particularway. This has led to particular research questions,de¢nitions of what are `good' and `bad' grasslands and`good' and `bad' management, which in turn often ledto particular forms of regulation. In recent years adi¡erent theory, the state and transitionmodel forgrassland dynamics (see Chapter á), has emerged as analternative means to explain phenomena in some typesof grasslands. This has happened because the othertheory did not seem to explain what people wereexperiencing ^ it was no longer useful. Unfortunatelywe often learn about theories as a universal truth ^ the`right way' ^ and as a consequence there have beenmany examples of applyingmanagement prescriptionsthat match the theory but not the context. Examplesare given below.

Theories thus bring certain things into focus,but leave other perspectives blurred or unconsidered.Another way of saying this is to consider any theoryas an example of a social technology, and like anytechnology when it is used certain aspects are revealedand other aspects concealed. This is quite a challengingnotion, but because we see it as important to howfuture grassland agronomists might conductthemselves, we devote some attention to it. A topicalexample is how, in recent concern with geneticengineering, we have come to think of life as beingmade up of sets of genes organized in particular ways.

1

BrianGoodwin ("ññã) draws our attention to how thewhole organism seems to have disappeared from sightfrom this perspective. As a consequence the specialemergent properties (Table ".") of whole organismsare concealed from consideration. Social technologieswhich derive from our theories andmodels ofunderstanding are just as powerful in realizing di¡erentgrassland systems as are `harder' technologies such astillage, ¢re and new plants. In fact we would go furtherand say from our perspective that grassland systemsarise as result of our ways of thinking ^ grasslandsystems do not exist in themselves but as a relational

unity between social factors and what we call ournatural or biophysical environment (Russell & Ison,"ññâ). This is exempli¢ed in Fig. "." which picturesgrassland systems as arising from the relationshipsbetween humans with di¡ering histories andperspectives and what we call a `grassland space'.

1.1 The social construction ofgrassland systemsIn "ñâá the physicistMax Planck said: `Science cannotsolve the ultimatemystery of nature . . . because, in thelast analysis, we ourselves are part of nature, and

Table 1.1 Generalized system concepts employed in systems thinking

Concept De¢nition

Boundary The borders of the system determined by the observer(s) that de¢ne where controlaction can be taken: a particular area of responsibility to achieve system purposes

Communication (i) First-order feedbackmay be regarded as a form of communication, but should not beconfused with human communications, which has a biological basis(ii) Second-order occurs in languages amongst human beings and gives rise to newproperties in the communicating partners who each have di¡erent histories

Connectivity Logical dependence between elements (including sub-systems) within a system

Decision-taking Information collected according tomeasure of performance is used tomodify theinteractions within the system

Emergent properties Properties that are revealed at a particular level of organization and which are notpossessed by constituent sub-systems. Thus these properties emerge from an assembly ofsub-systems

Feedback A form of inter-connection, present in a wide range of systems. Feedbackmay benegative (compensatory or balancing) or positive (exaggerating or reinforcing)

Hierarchy The location of a particular systemwithin the continuum of biological organization(Fig.1.4). This means that any system is at the same time a sub-system of some widersystem and is itself a wider system to its sub-systems

Measure of performance Information collected according tomeasures of performance is used tomodify theinteractions within the system

Monitoring and control Information collected and decisions taken aremonitored and controlled and action istaken through some avenue of management

Purpose Objective, goal or mission; the `raison d'etre' that in terms of themodel is to achieve theparticular transformation that has been de¢ned

Resources Elements that are available within the system boundary

Transformation Changes, modelled as an interconnected set of activities, which convert an input thatmay leave the system (a `product') or become an input to another transformation

Source:Adapted fromWilson ("ñðã).

2 Overview: perspectives on grassland systems

therefore part of themystery we are trying to solve'.This captures the notion that we wish to convey ^grassland systems emerge from the diverse relationshipswe have with `nature'.Wewish to explore howtechnology shapes our relationship with nature and theresponsibility we, as human beings, hold with regard tothis relationship.More speci¢cally we are concernedwith how a future generation of grassland agronomistsmight participate in the design of grassland systemsthat are sustainable and ethically justi¢able. Theperspective we as authors take is that there is value in:(i) thinking of grassland systems as social constructs(Fig. ".á) and (ii) using systems concepts (Table ".")to think about, describe, and inform action in thedesign of future grasslands. These perspectives guideour thinking and the organization of this book.

Historically texts about grasslands have tendedto be based on the classi¢cation of types of grasslands â typology or taxonomy (see Table ".á) ^ based oncommon features such as species, rainfall, soils andother edaphic features. Such classi¢cations haveincluded particular types of human activity, but haverarely taken the perspective that grassland systems aresocial constructs, tending to refer to them as `naturalsystems' or modi¢ed àgro-ecosystems' (see Fisher,"ññâ). From our perspective we would wish to situatethese earlier ways of seeing `grassland systems' in adi¡erent context.We do not wish to lose that which

was valuable from the earlier ways of thinking but seethe need for newways of thinking and acting to dealwith the complex issues that underpin grasslandsustainability.

A `construct' is the particular viewpoint orperspective of `reality' unique to an individual andspeci¢c to time and place (Bannister & Fransella,"ñæ"). A constructivist perspective is one in which theobserver is part of the system rather than independentof, or external to, it. In Fig. ".á we depict the diversityof existing socially constructed grassland systems andhow these change over timewith the changingperspectives of those who are `stakeholders' in thesystem. To see `grassland systems' as social constructsis to take seriously research on the biology of our owncognition. The word cognitionmeans literally `togetherto know'. Thus a group of experts with di¡erentexperiences and training, when distinguishing aparticular grassland system,may see quite di¡erentthings (Fig. ".â). This can lead to di¤culties,particularly if individuals are not aware of it, whenthey try to work together in teams or groups, and whenany collectivemust decide on a course of action whenthere is no clear right way to proceed. Russell ("ñðå)points out important elements of a constructivistperspective when he states: `My real world is di¡erentthan your real world and this must always be so. Thecommon ground, which is the basis of our ability tocommunicate with one another, comes about throughthe common processes of perceiving andconceptualizing. The processes may be the same butthe end products are never the same.What we share iscommunication of the worlds we experience, we do notshare a common experiential world.'

From this perspective any individual only hasaccess to what we call a grassland system throughcommunication in its many and diverse forms ^communication is a process that relates us to each otherand to our environment and enables us to distinguishand recognize di¡erent grassland systems. It would benecessary, for example, for the di¡erent expertsdepicted in Fig. ".â to engage in some form ofcommunication if they were to decide on how thegrassland systemmight be best described, changedor ìmproved'. If they did not, decisions about changewould be likely to be based on a very narrowperspective and for many of the complex issues orproblems to be dealt with in grassland systems thiswould be likely to prove unsustainable.

If students of grassland agronomy recognize thattheymust iterate between di¡erent perspectives andlevels of biological and social organization (Fig. ".ã),

Fig. 1.1 Grassland systems are social constructs unique toindividuals or groups at particular times and in particular places.Peoplewith theories andwith different experiences have differentperspectives (Px,Py). Grassland systems emerge from the dynamicrelationship between people in relationshipwith a grassland `space',inwhich soils, water plants, etc. are recognized. Placing a boundaryaround this system,we are able to recognize different `grasslandsystems'.

The social construction of grassland systems 3

and that together with farmers, researchers orpastoralists, theymust make decisions within a systemthat is socially constructed and often so complex thatthere may be no single, universal best solution, then itis likely that each professional will more fullyunderstand and appreciate the other. From thisrecognition it also follows that prescriptive advice froma textbook is likely to have very limited applicability.In this book we try to avoid prescriptions: we areconcerned with concepts and principles.

1.2 Grassland issues or problemsFuture grassland agronomists will be engaged in theformulation and resolution of grassland `problems'.This is the same as being involved in the design or`construction' of grassland systems (Chapter ñ).When

we use these terms we are not referring to themin the sense commonly associated with building orarchitecture, but in terms of social processes. To dothis it will be necessary for grassland agronomists to beaware of: (i) the problem formulation process; (ii) theneed to work with others whomay have di¡erentperspectives; (iii) particular ways of thinking aboutgrassland problems; (iv) the historical understandingswe have about grassland systems and (v) the languageand concepts that are used to talk about them and toguide action.

From our perspective grassland `problems' are notsomething that exist independently of the processes bywhich they are named and recognized ^ we call thisproblem formulation, recognizing that it is a socialprocess and that problems do not exist òut there' just

Fig. 1.2 Grassland systems differ both spatially and temporally as a result of the different relationships between peoplewith differentperspectives (P1^P8) and a grassland `space'. This gives rise to change and diversity and newperspectives (P9^P16).


Table 1.2 Some typical `grassland systems' inAfrica and Latin America based on typologies of classification

Grassland system Major crops Major animals Main regions Feed source

AfricaPastoral herding(animals veryimportant)

Vegetables(compound)ab

Millet, vegetables

Cattle, goats sheep

Cattle goats, sheep

Savanna (SouthernGuinea)

Savanna (NorthernGuinea and Sahel)

Natural rangelands,tree forage

Natural rangelands,tree forage

Mixed farming(farm size variable,animalsimportant)

Rice/yams/plantains

Rice/vegetables,yams, cocoyams

Sorghum/millet,groundnuts, cotton,tobacco, maize,cowpeas, vegetables

Two ormore species(widely variable)

Some cattle

Cattle, goats, sheep,poultry, horses,donkeys, camels

Humid tropics

Transition forest/savanna

Savanna (Guineaand Sahel)

Fallow, straw,brans, vines

Fallow, vines, straw

Stover, vines, fallow

Latin AmericaPerennial mixtures(large farms)(livestockrelativelyunimportant)

Coconuts, co¡ee,cacao, plantains,bananas, oil palm,sugarcane, rubber

Cattle, swine Allc Natural pastures,by-products, cullmaterial

Commercial livestockExtensivelarge to very large(livestockdominant)

IntensiveMedium to large,livestock dominant

None are important

Improved pasture,some gains

Cattle (beef )

Cattle (dairy),swine, poultry

C, V, Br, Bo, G,CAc

Allc

Natural grasslands

Natural andimproved pasture,feed grains, by-products

Mixed cropping, smallsize in settled areas;medium size in frontierareas; subsistence orcash economy (livestockrelatively important)

Rice, maize,sorghum, beans,wheat, cacao,plantains, co¡ee,tobacco

Cattle, poultry,goats, sheep,donkeys, horses,mules, swine

Allc Natural pastures,crop residues, cutfeed

aEnclosed areas around household or village.b Present or absent, depending on the area.cAll, all countries; Bo, Bolivia; Br, Brazil; C, Colombia; CA, Central America; CI, Caribbean Islands; E, Ecuador;G, Guyana; P, Peru; V, Venezuela.Source:McDowell &Hildebrand ("ñðò).

Grassland issues or problems 5

Fig. 1.3 Perspectives of a grassland system taken by disciplinary experts differ from that taken by an effective interdisciplinary team.(Source: adapted fromU. Scheuermeier, unpub.)

6

waiting to be identi¢ed. Some problemsmay beformulated in a way that is quite speci¢c, e.g. the poorgrowth response of plants to a non-optimal conditionsuch as increasing aluminium toxicity in the soilsolution. Such problems are amenable to analysis usinghypotheses that are testable. The analysis should leadto a de¢nite solution or, at least, to a `best' or anoptimal solution to that speci¢c problem. At the otherextreme, in terms of levels of complexity, are problemsthat arise because of the uncertain interrelationshipsthat are a part of farming systems: there is generally nosingle, static `best' solution to problems ofinterrelationships in a dynamic system involvingenvironment, plants, animals, technology (types ofploughs, etc.), economics and the social values andgoals of the farmer. It is therefore important todistinguish between problems for which there is an

agreed solution and those for which there is no clear`right' answer.Many grassland `problems' are of thesecond type, and for this reason we often refer to themas issues, rather than problems.

The way in which we formulate issues shapes thesort of answers we get; this is because of the ways inwhich we choose to think about issues. Thinking aboutissues in only one way leads us into traps fromwhich itis often di¤cult to escape (OpenUniversity, "ññ").Botswanan and Australian examples demonstrate thisnotion of `traps'. In Botswana, Louise Fortmann's("ñðñ) case study of äò years of rangeland use showedhow o¤cial policy consistently de¢ned themajorproblem of the pastoral regions as overstocking leadingto certain ecological disaster. The problemwas clear,as was the technical solution (destocking). Localexperience, however, de¢ned the problem as too little

Fig. 1.4 Hierarchy of levels of systemorganization. The complexity of such a schema is increasedwhenwe recognize themany perspectivesweas observers bring and that this operates at each `level' of organization.

Grassland issues or problems 7

land. The local solution was also very di¡erent:renting, or using land previously let to a Europeanmining company. The local experience was that thelocal range could and did carry an increased cattlepopulation and that besides localized problems, thedire o¤cial predictions did not eventuate.While thereis general agreement that the quality of theenvironment (as indicated by the quality of thegrazing, the number of trees and the extent of erosion)is deteriorating, there is clearly no agreement on causesor solutions. It is signi¢cant that the story has been toldconsistently from both perspectives for äò years: thisshows how di¡erent and how unconnected traditionscan be (Russell & Ison, "ññâ).

AnAustralian study (Kersten, "ññã) demonstratesa similar contest between researchers and pastoralistsover the nature and causes of a so-called `rangelanddegradation problem' in the semi-arid pastoral zone.Similar experiences have occurred in theUSA, so this isnot a phenomenon peculiar to either the `developing'or `developed' world. These are examples of failure ofthe problem-formulation process. They demonstratehow easy it is to fall into traps when particular ways ofthinking are consistently employed.

1.3 Systems thinkingThe twomost commonways of thinking in our(Western) society have been logical and causalthinking. The former is of the type: Ìf all grasses areplants, and this is a grass, then it is a plant'. As theOpenUniversity team ("ññ") point out, this statement:`starts with a generalisation, a premise which isassumed to be true and then deduces a conclusionabout a particular case.' Three things characterise thistype of thinking: (i) it is objective ^ the conclusion doesnot depend on your point of view, your opinions andvalues about the world; (ii) it is necessary: that is theconclusion always follows from the premise; (iii) thestructure of the thinking is sequential or linear ^ it hasthe form of ìf a, then b', often called a chain ofreasoning. Logical thinking is a way of linking ideastogether.

Causal thinking is a way of linking eventstogether. A grassland agronomist explaining why acow is not growingmay tell you that there is a proteinde¢ciency in the pasture at the end of a dry season.This is of the form à causes b' and super¢cially is notunlike logical thinking; many would suggest it has thesame three characteristics. However causal thinking isalways an explanation by an observer of an event, andan event is dynamic, with participants. So causalthinking is really a statement about a relationship and

the nature of the event (e.g. the protein de¢cientpasture and the cow) is dependent on the properties ofthe participants in the event that is distinguished by anobserver. As observers we tend to put boundariesaround what we are studying ^ thus an animalnutritionist may place a boundary around therelationship between the animal and the plant, whereasa plant nutritionist may place the boundary around theplant and its nitrogen supply. The poor nitrogen supplymay be causing low protein but this in itself is a poorexplanation of the overall problem if it onlyconcentrates on one set of relationships. This is wherethe so-called objectivity of causal thinking breaksdown, because explanations are not o¡ered from avalue-free perspective.We all have perspectives. Ourperspective often determines where we place ourboundaries around problems or issues. Economists referto this as externalities.

The so-called objective forms of reasoning are notvery suitable by themselves for sorting outmany of theissues to do with grassland systems ^ they are not veryhelpful when it comes to preferences about breeds ofanimal, family values, lifestyle questions, enterprisegoals nor deciding what to do about vegetationmanagement in a whole watershed, rangelanddegradation or policies tomitigate contributions byruminants to greenhouse gas emissions. Hence there isa need for a way of thinking about systems that takesthe characteristics of systems into account (Table ".").There are two adjectives derived from `system':`systemic', or thinking in wholes and `systematic', step-by-step thinking or procedures. Checkland ("ñðð)notes that to many people a computerized informationsystem ìs the very paradigm of what theymean by`system''. This is not what wemean when we talk about`system' but it does largely account for the lack ofattention to the development of the ideas of `system' inagriculture and the lack of attention to systems-basedmethodologies that are not computer based.

Systems thinking is a special form of holisticthinking ^ dealing with wholes rather than parts. Oneway of thinking about this is in terms of a hierarchy oflevels of biological organization and of the di¡erentèmergent' properties that are evident in, say, thewhole plant (e.g. wilting) that are not evident at thelevel of the cell (loss of turgor). It is also possible tobring di¡erent perspectives to bear on these di¡erentlevels of organization (Fig. ".ã). Holistic thinking startsby looking at the nature and behaviour of the wholesystem that those participating have agreed to beworthy of study. This involves: (i) takingmultiplepartial views of `reality', as exempli¢ed by the


interdisciplinary team in Fig. ".â; (ii) placingconceptual boundaries around the whole, or system ofinterest and (iii) devising ways of representing systemsof interest.

1.4 Representing grassland systemsWe represent grassland systems in three ways:(i) in the choice of language to describe them; (ii) usingdiagrams and (iii) usingmodels.We do not proposeto spendmuch time talking about language, becausethe book itself is an introduction to the languagecommonly used to talk about grassland systems. It is,however, worth pointing out that some particularfeatures of language can be important in shaping theperspective we use to `construct' a grassland system.Metaphors are a good example: if we choose to see adairy farming system as if it were a factory, then wegenerate certain images and ways of looking at thedesign of a system. If we chose to look at it as if it werean ecological system this might reveal di¡erent featuresand lead to di¡erent designs.

Diagrams have the advantage of showinginterconnections in ways that words alone can not. Thedevelopment of diagramming skills is a useful adjunctto creative problem solving ^ an important ingredientis iteration ^ repeating the process ^ until you aresatis¢ed with the outcome. Figure ".ä is a diagram

showing the interconnections between biologicalcomponents of a grassland system. In our diagram thebiological components of grassland or pastoral systemsare the environment, plants and animals. These areclosely interrelated in a cyclical fashion.We emphasizethe quantitative interrelationships within the biologicalcycle and the feedback nature of the cycle. Thiscontrasts with the perspective of the traditionalagronomist, who usually thinks of the cycle as a simplecatena: environment ^ growth of grass ^ animalproduction ^ product removal. Diagrams such as thatshown in Fig. ".ä can bemade speci¢c by emphasizingparticular developmental aspects, environmentalvariables or loops feeding backmaterial orinformation. An example of this speci¢city is providedby a grassland comprised of amixture of the tropicalannual legumeTownsville stylo (Stylosanthes humilis)and annual grasses in the Australian wet-and-drytropics. Here, in a climate in which rainfall during thewet season dominates the productivity and compositionof the grassland, the agronomist pays particularattention to estimating how the composition andproductivity are a¡ected bywater through rainfall,in¢ltration, run-o¡, soil moisture, soil drying anddrought (Fig. ".å).

It is useful to note, in passing, that the amount ofcontrol which the farmer is able to exercise over the

Fig. 1.5 Themain biological components in the function andmanagement of grassland systems. A grassland system is dynamic: various poolsor state variables (&) are linked by flows ofmaterial, e.g. seed, leaf (arrows) and governed by rate variables ( ).

Representing grassland systems 9

components in the grassland system ranges fromvirtually nil to total (Table ".â).

1.4.1 ModellingModelling is a means of understanding and reducingcomplexity. The word `model' and the act of`modelling' havemanymeanings. Here we use a verybroad interpretation ofWilson's ("ñðã): à model is theimplicit interpretation and representation of one'sunderstanding of a situation, ormerely of one's ideasabout the situation constructed for some purpose. Itcan be expressed inmathematics, symbols or words,but is essentially a description of entities and therelationships between them. Itmay be prescriptive orillustrative, but above all it must be useful.' At the endof each chapter we demonstrate how the use of modelscan enhance analysis and decision-making using the

material that has been discussed. Here we wish togive an overview of the approaches tomodelling.

Modelling is carried out for some purpose.Weneed to be careful not to use models for purposes otherthan which they were designed and to recognize thatthe learning that occurs in the process of developingmodels is qualitatively di¡erent from the use of modelsalready developed by someone else. Purpose re£ectsthe notion that amodel is not a goal in itself but rathera part of some analysis associated with problemresolution that leads to increasing knowledge andsometimes an increasing need for quantitative results.Di¡erent forms of modelling ¢nd greater use indi¡erent problem-formulating strategies and atdi¡erent levels of biological organization (Fig. ".ã).Used in this way, modellingmay serve as a unifyingfunction in the study of grassland systems if the stepsoutlined in Table ".ã are followed within a relevantsystemic or interdisciplinary framework (Rykiel,"ñðã).

Conceptual modellingConceptualization of what the system under study is ormight be is usually a starting point (Rykiel, "ñðã;Wilson, "ñðã). Thus conceptual modelling precedesother forms, as well as being amodelling form in itsown right.

Conceptual or qualitative models may be used:(i) as an aid to clarifying thinking about an area ofconcern; (ii) as an illustration of a concept; (iii) as anaid to de¢ning structure and logic; and (iv) as aprerequisite to design. These uses are not mutuallyexclusive.Wewill use examples that combines aspectsof (ii) and (iii), presenting some of the key concepts andinteractions in plant generation.

Themodelling of human activity systems, asdistinct from biological or natural systems, utilizes aparticular form of conceptual model. Themodellinglanguage developed byCheckland ("ñð") andcolleagues is the use of verbs to model functions oractivities. Plant domestication can be viewed as ahuman activity system (Chapter á) andmodelled inthis way. Conceptual modelling often relies onparticular types of diagrams (Fig. ".æ).

Expert systems, fuzzy thinking and chaosExpert systems are also commonly referred to ascomputer-based decision support systems (DSS). Theyusually combine data (for example, by spreadsheets aslook-up tables) and quantitative relationships withrule-based or qualitative inferences. They are thusmixes of qualitative and quantitative models andmay

Fig. 1.6 Environmental and plant factors that dominate grasslandpattern and species cycling in a grassland comprised of Townsvillestylo and annual grasses in thewet-and-dry tropics. The factors exertquantitative effects on yield and species composition; the plant factorsgermination, establishment, competition and seed productionmay beseen as a series of filters throughwhich individual plants attempt topass. (FromTorssell, 1973.)


contain either as sub-models. Their common purpose isto assist decision-making; their output is usually a set ofoptions for action with likely bene¢ts and perhaps risks,associated with each option (see Stuth&Lyons, "ññâ).Hochman et al. ("ññä) have coupled quantitative,rule-based outputs from an expert systemwitheconomic analyses to assign likely economic scenariosto each option for management of a beef-based,grassland production system.

Fuzzy thinking probably has much to o¡er futuregrassland agonomists; Zhang&Oxley ("ññã) foundfuzzy set ordination (FSO) able to analyse andsynthesize ecological information. This form ofthinking relies on both quantitative and qualitativemodelling. Fuzzy cognitive mapping (FCM), aparticular form of fuzzy thinking, enables èveryone topack their ownwisdom and nonsense into amathpicture of some piece of the world. But once packed in,the FCMpredicts outcomes andwe can compare thesewith data to test them' (Kosko, "ññã). FCMemploysthe power of computing to examine feedback incomplex systems. It thus acts like a neural net. Kosko("ññã) argues that `fuzzy logic will change our worldviews in small ways and deep ways. It will bring us

closer tomachines and bring them closer to us. Andfuzzy logic will poke holes in moral absolutes. It willhelp solve some problems andwill muddy up others.'

Chaos is the sensitive dependence of a system toinitial starting conditions. That is, very smalldi¡erences in a system, say two populations, can overtime lead to vastly di¡erent trajectories and outcomes.Such systems have chaotic dynamics (Hastings et al.,"ññâ). Understandings from studies of chaos andcomplexity will also increasingly inform grasslandagronomists (e.g. Grenfell, "ññá; Godfray, Cook&Hassell, "ññá). As Brian Goodwin ("ññã) notes: `lifeexists at the edge of chaos, moving from chaos intoorder and back again in a perpetual exploration ofemergent order.' He explains further: `For complexnon-linear dynamic systems with rich networks ofinteracting elements, there is an attractor that liesbetween a region of chaotic behaviour and one that is`frozen' in the ordered regimewith little spontaneousactivity. Then any such system, be it a developingorganism, a brain, an insect colony, or an ecosystemwill tend to settle dynamically at the edge of chaos. If itmoves into the chaotic regime it will come out again ofits own accord; and if it strays too far into the ordered

Table 1.3 Level of controlwhich farmers are able to exercise over environmental and biological variables in grassland systems

Environmental factor Level of control Method of control

Grazing pressure Good Stocking rate, stockmovement and herd (population) structure

Plant population Good Sowing, selective herbicides, stocking rate and stockmovement

Defoliation Good Stocking rate and stockmovement

Nutrition Fair Fertilizer application where economic or where species respond

Pests and diseases Variable, usually poor Ranges from good short term control in intensive situations tocontrol relying on plant resistance in extensive systems

Moisture Poor Irrigation: selection between existing wetter and drier sites anduse of cultivars that have development patterns tomatchavailable soil water; farmer can usemechanical aids (contourfurrows, drains, etc.)

Soil structure Poor Tillage and stocking a¡ect particle size distribution, etc. butknowledge is empirical (e.g. don't ploughwhen soil is `too wet'):farmer can select between existing soils, andmake amendments,e.g. add gypsum

Temperature Indirect Selection of sites with di¡erent aspects; selection of species withdi¡erent growth responses to temperature, or alteration of time ofsowing


regime it will tend to `melt' back into dynamic£uidity where there is rich but labile order, one thatis inherently unstable and open to change.'

Quantitative modellingQuanti¢cation of processes within grassland systemsmodels aids both research andmanagement.We referto quantitative models extensively in this text, usuallyas algebraic equations or statistical relationships. Aswith all models these are a simpli¢cation and includeassumptions made by themodeller; they are usedmainly to predict the behaviour of some aspect of thesystem being considered and they require some form ofvalidation. An understanding of this area of modellingmay be gained from the schema proposed byWilson(Fig. ".ð).

Deterministic models include any algebraicrelationship; they are well suited to problemsconcerned with the allocation of some limited resource(e.g. land, money, labour, fertilizer) wheremanyalternatives exist. Linear programming is a well-knownform of a steady state deterministic model. It involvesnumerical optimizing and has been frequently used byagricultural economists to study complex problemssuch as theminimization of costs or themaximizationof pro¢ts, by animal nutritionists for determining theleast cost or pro¢tmaximizing rations (e.g. Sauvant,Chapoutot &Lapierre, "ñðâ) andmore recently withthe development of powerful algorithms for integratingbiological and economic data fromwhole farmingsystems. Linear programming is also an appropriatetool for studying the integration of new grasslandsystems and cattle management options (e.g. Teitzel,

Table 1.4 Modelling: the purposes forwhich amodelmight be used and a four-stepmodelling procedure that enables the development of acommonperspective by different discipline groups involved in problem-solving or research in connectionwith pastoral systems

Purposes1. Exploration Objectives are very often general or intuitive, usually with no speci¢c criteria for meeting

them; themain aims are insight, clari¢cation and understanding of the factors thatcontribute signi¢cantly to system behaviour

2. Explanation The general objective is to understand the structural and functional relationships betweencomponents and sub-systems that explain the pattern of interconnections within the systemand generate system behaviour; speci¢c objectives are related to the level of resolution andthe level of the study; system, sub-system, and component levels

3. Projection The objective is to examine the dynamic behaviour of system variables at any level (i.e.component, sub-system or system) and the e¡ects of changes in the values of parameters orvariables and their variability; variability in the occurrence of events andmaking ofdecisions; the patterns of behaviour represented in the dynamic relationships amongvariables aremore important than the actual values of the variables

4. Prediction The speci¢c objective is to estimate future values of particular system variables and/or thenature and timing of events and decisions; emphasis is on the accuracy and utility of theprediction, and the reasonableness of the explanation of the prediction

Procedures1. Formulation The development of ideas, problem statements, and approaches to solutions by

conceptualization, de¢nition and design

2. Clari¢cation The formal expression of objectives, model structure, functional relationships, etc. throughorganization, documentation and accounting

3. Analysis The interpretation and explanation of model and system behaviour through statistics,simulation and system analysis

4. Application The use of the results of analysis and of models through communication, experimentationand technology transfer

Source:Adapted fromRykiel ("ñðã).


Monypenny&Rogers, "ñðå). The technique oflinear programming is described by, for example,Gass ("ñðä).

Steady-state non-deterministic models applywhere themechanisms governing behaviour are notknown, but where it can be assumed that certainvariables are wholly or partially dependent on others.Seed yield, for instance, may be wholly or partlydependent on plant density (Chapter ã). Regressionrelates variables of interest. In its simplest form, e.g.seed yield ( y) may be related linearly to plant density(x). The constants in linear regression, curvilinearregression, multi-linear regression andmultipleregression are derived tominimize the sum of the errorssquared at the discrete data points.Wilson ("ñðã)includes probabilistic modelling as a variant withinsteady-state quantitative modelling.Models with someelement of randomness, to which wemay assign aprobability, are called stochastic models.

Steady-statemodelling is sometimes described asempirical modelling. Empiricism involves attemptingto ¢t somemodel or equation to data and thenmakingdeductions about themechanisms involved. Theopposite to this is mechanistic modelling, whichattempts to understand the response of biological

systems in terms of mechanisms.Mechanistic modelsare constructed by looking at the structure of thesystem and dividing it into its components in anattempt to understand the behaviour of the wholesystem in terms of the actions and interactions of itscomponents (Thornley, "ñæå).

Dynamic, deterministic models includedi¡erential equations. Di¡erential equations allowthe simulation of situations in which time dependencecan be represented continuously. State variables, i.e.variables that describe quantities or amounts (e.g.pasture drymatter, livestock weight, nitrogen content)are each associated with a rate variable which describeschange with time (t), e.g. pasture growth rate, livestockgrowth rate, rate of nitrogen cycling. The value of astate variable at any point in time (T) can be derivedby integration which, when repeatedmany times,completes the simulation of interest.

RATE (T7 ") = t [STATE (T7 ") . . .] (".")

STATE (T) = [RATE (T7 ")6DELT ]+ STATE (T7 ") (".á)

where RATE and STATE refer to any rate and statevariable respectively,DELT is the time interval over

Fig. 1.7 Aschemadepicting the different forms of diagrams that can be used in conceptualmodelling. (Source: OpenUniversity, 1991.)


which integration occurs, the length of which isdetermined by the type of model.

Dynamic, non-deterministic models are concernedwith simulation of discrete events using di¡erenceequations (as opposed to di¡erential equations).Quantitative models range over all levels of biologicalorganization (Fig. ".ã).Morley&White ("ñðä)describe quantitativemodels relevant to grasslandagronomy ranging from ones concerned with nitrogen¢xation at the cellular level to whole farm and nationalsectoral models. One useful division of modelsaccording to their use is between predictive, andexplanatory or experimental models (P.Martin, "ñðå,personal communication):

(i) Predictive models. These are frequently concernedwith grassland productivity and e¤ciency andthus with input/output relationships, theenvironment and production interactions andeconomics. This form ofmodelling contributes toestimations of the reliability of feed options, theanswering of farmmanagement questions such as`how to use a given foragemost e¤ciently withwhat type of animal?' and/or to estimations of thee¡ect of variability in the weather and prices onenterprise performance. Predictive models also aidfarm development decision-making by simulatingdevelopment under variable conditions. Thegeneral structure of one such predictive model(McKeon&Rickert, "ñðå) begins with a waterbalancemodel, predicts monthly grassland growthand then considers a feed year, simulated on aseasonal time step (DELT, Eqn (".á)).

(ii) Explanatory or experimental models. Predictivemodels can be run and re-run using e.g. historicalclimatic data and by varying the growth rate ofthe forage, the stocking rate, forage allocation andanimal age classes (e.g. O'Leary, Connor&White, "ñðä). This allows experimentation on awhole farm basis andmanipulation of variableswhich would be impracticable and too expensiveto investigate with ¢eld research. From rerunningamodel with di¡ering inputs or input/output

relationships the researcher can predict how theoutput of themodel might respond to varyingfactors. Hypotheses may be formulated, e.g. thatfodder crop yield is dependent on the sowing date,and then tested by re-running themodel.

Concluding remarks on modelling of grassland systemsThemodelling of grassland systems has adopted all ofthe approaches outlined above.Most grasslandmodelshave tackled speci¢c aspects of grass or animalproduction; rarely have there been attempts tomodelthe entire systemmechanistically, even for a speci¢clocality. One attempt at comprehensivemodelling ofgrassland systems was theUnited States GrasslandBiome study, which generated themodel described byInnis ("ñæð).

All modelling is heuristic: it provides a learningexperience which furthers investigation. Thus theresearchers involved inmodelling learn about thesystem or area they are attempting tomodel and wherede¢ciencies in knowledge and understanding exist;these may be further explored bymaking predictionsfrom themodel or formulating hypotheses and testingthem by conducting ¢eld or laboratory experiments orfurther modelling. Once constructed themodels mayalso be used to help others learn, e.g. extension o¤cers,farmers, other researchers. Suchmodels need to beùser-friendly'; one typical form is the `what if?' modelwhich, if formulated to, say, respond to usermanipulation or farmers' questions about fertilizerapplication strategies, helps the farmer understand thedynamic interaction of fertilizer application rate, price,stocking rate, beef output and pro¢t. The processes bywhichmodels are developed and then used has,however, received less attention than is warranted.

Problems relevant to grassland systemmodellinghave been reviewed by Smith ("ñðâ), Freer &Christian ("ñðâ), Emmans&Whittemore ("ñðâ),Bennett &Macpherson ("ñðä), Ison ("ññâa), Seligman("ññâ) and Stuth et al. ("ññâ). Authors conclude thatmodelling has yet to achieve its potential, largelybecause of the lack of appreciation bymodellers,biological researchers and administrators of many ofthe concepts and processes of modelling outlinedabove.We believe that future grassland agronomistswill rely heavily onmodels, but it is worth noting threeviews with regard to the future: (i) that the usefulnessof the output frommodels is the key issue inmodellingbut to date little has been done about checking that theoutputs are useful and that they are worth themodelling e¡ort (Bennett &Macpherson, "ñðä);(ii) DSS assist managers in dealing with complex

Fig. 1.8 Forms of quantitativemodelling that are used in the study ofgrassland systems. (FromWilson, 1984.)


problems, but a fundamental requirement of asuccessful DSS is understanding the human learningprocess and involving users in their development(Stuth et al., "ññâ); and (iii) there is a rich array of non-computer basedmodels andDSSwhich have a role inthe sustainable management of grassland systems(Ison, "ññâa).

1.5 Purposes of grassland systemsThe purpose for any grassland system is determined bythose who participate in the construction of the system.Future grassland agronomists will undoubtedlyparticipate in the design of novel grassland systems thatre£ect changing ways of thinking and cultural values(Chapter ñ). This will require innovative and creativethinking (Carter et al., "ñðã; Buzan&Buzan, "ññâ)that also values, and is able to work with, social andbiological diversity. In this book we focus on sevenmeasures of performance of grassland systems(Fig. ".ñ). Each is a measure related to one aspect ofsystem purpose, which is often a unique combination ofmeasures re£ecting the preferences and trade-o¡smadeby those involved in constructing andmanaging thesystem (Bawden& Ison, "ññá).

Some of the outputs from a grassland system aremeasured in units of meat, milk, wool, hide ormoney.This is called production. In this book we de¢neproductivity (Fig. ".ña) as the output per unit of time,e.g. kg beef per year or kg beef per ha per year.Productivity is a `rate variable', i.e. a measure of thedynamic nature of the system, or how it operates. Ratevariables contrast to state variables such as the amountof standing feed, which tell us only what a particulargrassland system looks like at a particular point intime.

De¢ning the units of biological output from agrassland as kg of beef, etc. does not, however, indicate,even in biological terms, whether the system isoperating successfully or e¤ciently. Success is usuallymeasured in terms of the short-term (seasonal orannual) output relative to inputs. Here we call this ameasure of the e¤ciency of grassland production(Fig. ".ñb). Of equal importance is the stability ofthe system, i.e. its ability to return to an èquilibriumstate' after a temporary disturbance (Fig. ".ñc). Our

Fig. 1.9 The properties of grassland or pastoral systems defininggeneralized situations of high (left-hand column) and low (right-hand column) productivity (a), efficiency (b), stability (c),sustainability (d), responsibility (e), interconnectivity (f) andequity (g). The x-axis in a^f is time, t. (Adapted fromConway, 1993.)

Purposes of grassland systems 15

approach, which does not di¡erentiate between thesocial and biophysical aspects of grassland systems,enables us to consider three further measures of systemperformance. The bene¢ts of grassland system outputsmay £ow disproportionately to those who have, or wishto have an interest in the system. This raises questionsof equity and the need to consider how innovation andtechnological changemay a¡ect the distribution ofbene¢ts (Fig. ".ñg). As participants in the construction,the bringing forth, of grassland systems we also have anethical responsibility. Responsibility, the extent towhich the system enables individuals to participate andto be responsible is thus our next measure ofperformance (Fig. ".ñe). It is possible to speak of theextent to which participants can be responsible andwe propose as ameasure the number of peopleexperiencing an inviation to participate. Equity andresponsibiliy are a¡ected by the extent to which thereis interconnectedness in the grassland system.Interconnectivity is a measure of the relationshipbetween participants and the relationship they havewith animals and land. It encompasses notions of self-identity and stewardship and the satisfaction humansderive from these. As ameasure of interconnectivitywe propose the extent to which diverse perspectives areinvolved in the design of future grassland systems(Fig. ".ñf). Finally, some ecologists would di¡erentiatesustainability, i.e. the ability of a system tomaintainitself or the degree of di¤culty of management requiredtomaintain it (Fig. ".ñd). Others would wish to de¢nesustainabilitymore broadly (Bawden& Ison, "ññá); inthis book we recognize both the ecological perspectiveon sustainability as well as the view that sustainabilityis not an endpoint, but is an emergent property of anongoing process.

Throughout the book we consider these measuresof performance inmore detail and in the light ofdi¡ering historical perspectives on grassland systemsand contemporary environmental issues in whichgrassland systems feature.We examine how particularmodels of understanding have arisen and how theyshape interventions and technological change.Mostof this book, however, is devoted to the principles thatunderly the biological operation of the grasslandsystem, particularly those principles that relate to thedynamics of the biological system. Because farmershave a diversity of forage sources fromwhich theymay draw to sustain animal production, grasslandagronomists will increasingly be concerned with theintegration of forage sources, including crop and agro-industrial residues, into animal production systems.Accordingly, some economic principles and likely

future roles of grasslands in farming and environmentalsystems are discussed. A feature of this book is that wesituate this biological understanding in a broader socialcontext commencing in Chapter á.

Notwithstanding our emphasis on a systems viewof grassland agronomy, we are bound by the fact thatbooks start at the beginning and end at the last page, tostructure this book in a catenary fashion. The ¢rst partof the catena is generation (Fig. ".ä, Chapter â). Thiscomprises the dynamics of the bank of seed in the soil,seed germination and vegetative generation fromstolons and rhizomes, leading to plant emergence.Generation leads to vegetative growth (Chapter ã)and the life cycle of grassland plants ends with seedproduction (Chapter ä). Nutrition (Chapter å) linksthe plant and animal components of the system. Thequality and quantity of feed available from living,dead and conserved pools determines animal intake(Chapter æ). The animal in turn a¡ects theproductivity and composition of the grassland(Chapter ð). Finally, the agronomist, economist,farmer and other relevant stakeholders integratethe principles of pasture development, growth andutilization into the design andmanagement of thegrassland system.Management is associated withfarmers' goals, which vary within socio-culturalsystems and between them; wheremanagementinterventions are made to improve the quantity orquality of herbage (e.g. saving, fertilizing), there is aneed formanagers to estimate the likely annual cycleof production of grasslands and the requirements ofthe farm livestock. These must bematched, makingallowance if necessary for the conservation or purchaseof feed, in a way that ensures some return on the initialinvestment. Such returns are usually economic(Chapter ñ), but in some societies theymay be socialor cultural rewards. Consequently this book, althoughmainly devoted to biological aspects of grasslandagronomy, does conclude with a systems perspectiveof some of the environmental, economic and socialissues that are pertinent to grassland agronomy(Chapter ñ).

In describing the biological aspects of the systemour bias is towards assessing productivity, e¤ciency,stability and sustainability, but particularlyproductivity and e¤ciency. This unequal emphasisusually leads technologists to the conclusion thatgrassland agronomy should becomemore productive:that there are large areas of native grasslands, scrub orforest that could be ìmproved' and that productivitycould, and should, be increased in grassland systemsthat are currently based on sown pastures. For


example, it has been calculated that there are âòòmillion ha of ìmprovable' grasslands in both humidtemperate regions and the wet-and-dry tropics of SouthAmerica and wemight estimate that there are a further"òòmillion ha in Africa. Thus the total area ofpractically improvable grassland, if we include AsiaandAustralia, probably exceeds æòòmillion ha(Norman, Pearson& Searle, "ññä).Much of this æòòmillion ha is in a belt of high potential net primaryproductivity (NPP; Chapter ã). The biologicaladvantages of intensi¢cation of productivity are welldocumented (Henzell, "ñðâ). However, in Chapter ñwe explore whether intensi¢cation is necessarytechnically or appropriate socially.

1.6 Further readingBAWDEN, R. J . & ISON, R. L. ("ññá). The purpose of

¢eld-crop ecosystems: social and economic aspects.In Field-CropEcosystems, ed. C. J. Pearson, pp. ""^âä.Amsterdam: Elsevier.

CONWAY, G. R. ("ññâ). Sustainable agriculture: thetrade-o¡s with productivity, stability andequitability. InEconomics andEcology.NewFrontiersand SustainableDevelopment, ed. E. Barbier, pp. ãæ^åä.London: Chapman&Hall.

ISON, R. L. ("ññâ). Soft systems: a non computer viewof decision support. InDecision Support Systems For theManagement of Grazing Lands:Emerging Issues, eds J.WStuth&B. G. Lyons, pp. ðá^""ä. UNESCO-Manand the Biosphere Book Series. Vol "". Carnforth,U.K.: Parthenon Publishing.

PRETTY, J . N. ("ññã). Alternative systems of inquiry fora sustainable agriculture. Institute ofDevelopmentStudies Bulletin, áä, âæ^ãð.

RUSSELL, D. B. & ISON, R. L. ("ññâ). The research-development relationship in rangelands: anopportunity for contextual science. Proceedings of theIVth International Rangeland Congress,Montpellier,"ññ", Vol. â, pp. "òãæ^äã. Association Francaise dePastoralisme.

Further reading 17

Documents

Agronomy of Grassland Systems C. J. Pearson R. L. Ison - Library of