Agriculture in the tropics is complex anddiverse, encompassing plants and soils,climate and people, trade and survival. It canbe very diverse biologically, as in mixedgardens; elsewhere, one species can domi-nate vast areas of land. In dry, marginal areas,crops may grow for only a few weeks eachyear when there is sufficient water and, often,their yields are barely adequate to sustainlife. In humid, fertile regions, plantsyield more-or-less continuously, givingproducts for exports that dominate the econ-omy.

Despite this diversity, common ele-ments allow the terms ‘tropical agriculture’and ‘tropical agronomy’. The geographicallocation between the Tropics of Cancer andCapricorn is an approximation, but thewarmth and the bright sunlight between andaround these limits are what most peopleassociate with the word ‘tropical’. To theagriculturalist and ecologist, the soils andplants provide further similarities betweencontinents and countries. Many tropicalsoils are potentially unstable, low in fertili-ty, easily eroded and shallow. The cereals,legumes, roots, fibres and fruits that are nor-mally recognized as tropical crops are oftengrown by similar forms of husbandry inAfrica, Asia, the Pacific and the Americas.Of all factors, temperature most stronglylimits the extent of these species. Mostdevelop and grow between about 10 and40°C and yield economically at meanmonthly temperatures between 15 and 30°C.Very few have any chance of surviving to

yield in the cold or frosty weather of tem-perate lands in winter.

However, there are important excep-tions to these generalities. Several wide-spread subsistence crops such as potato, andcertain cereals including wheat and barley,are common throughout temperate agricul-ture and also occur widely in tropical lati-tudes, usually at high altitude where it iscooler. Indeed, staples such as potato andsome cereals originate from high altitudetropical regions (Chapter 4) and, althoughmuch land is cultivated using hoes andploughs, the management of these crops inmany tropical locations is little different tothat in temperate regions. For example, cere-al production on moist tropical plateauxdeploys machinery, fertilizer and pesticidewith an intensity similar to that in Europeor North America. Nevertheless, the princi-ples advanced in this book should be rele-vant to a wide range of plants, husbandryand regions that fit the loose term ‘tropical’.The examples described in the book are pri-marily of tropical species grown under lowinput husbandry, or in plantations, in lowlatitude regions of Africa, Asia and SouthAmerica.

1.1 The boundaries, scale and scopeof agronomy

Crop management, and its scientific studyagronomy, are part of a system that com-prises the physical elements of the climate,

1

1Contexts

© CAB International 2002. Principles of Tropical AgronomyS.N Azam-Ali and G.R. Squire

soil and land, the biological constituents ofthe vegetation and soil, the economic oppor-tunities and constraints of markets, salesand profit, and the social circumstances andpreferences of those who work the land(Fig. 1.1). In all this, management actsdirectly on a part of a plant, a whole plantor a small group of plants in a stand, or elsean amount of soil that can be lifted or turnedby a person, animal or machine. Each act ofmanagement influences the physiologicalprocesses of the plants, which in turn mod-ify or regulate the flow of environmentalresources – sunlight, water, nutrients – to

economic or useful products. Though largemachines are used, especially in plantationagriculture and with intensely managedcereals, the animal-drawn plough is oftenthe main means by which the soil is turnedand shaped, while the hand-held imple-ment, or the hand itself, sows, weeds, culti-vates and harvests.

Despite being defined at a particularscale in this way, management and its effectsinfluence and are influenced by processesand events at much smaller and largerscales. Any human influence on a plantorgan affects the metabolic pathways that

2 Chapter 1

Fig. 1.1. Flow diagram of physical, biological, economic and social dimensions of agronomy.

give the organ its nutritional and biochemi-cal character; and any disturbance to the soilultimately alters its minute architecture andthe organisms that inhabit it. Conversely, theclimate, the topography of the land, and theeconomic and social factors such as marketsand roads, determine the type of manage-ment that is possible and feasible. Manyindividual acts of management togetherinfluence the pattern of land usage, the fer-tility of the land and its local microclimateand the wealth and health of people andnations.

These interactions at larger and smallerscales than management are largely beyondthe scope of this book. It is neverthelessessential that agronomists are aware ofthem. Historically, the physical, biological,economic and social disciplines each havehad their own methods and units (or ‘cur-rencies’) for which there seem no universal-ly accepted exchange rates. We return to thistheme in the final two chapters. The rest ofthis introductory chapter aims to summarizegeneral aspects of tropical climates, soilsand plant growth.

1.2 The broad dominance of climate

Climate is the long-term nature of the weath-er. Climatic variables that affect the growthof plants include solar radiation, tempera-ture, precipitation, windspeed and atmos-pheric humidity. Other physical variables,such as the concentration of carbon dioxideand pollutants in the air, are also importantto plants, but are not usually regarded as cli-matic, though, of course, they do influencechanges in the climate.

The agronomist or farmer is not con-cerned with whether the field or regionbeing studied is classified as being in thetropics. Attempts to classify what is tropical(e.g. Koppen, 1931, 1936; Paffen, 1967;Nieuwolt, 1977) can even mislead if takentoo strictly. Koppen (1936), for example,demarcated the tropics as places having amean monthly temperature of 18°C in thecoolest month of the year. This definition isunhelpful as it is well above the limit atwhich most tropical plants start developing,

while large yields of tropical crops such asmaize and tea have been won at tempera-tures well below it. Indeed, plants maketheir own distinctions, while farmers andpractising agronomists do not have to be toldwhether they are in the tropics or not. In thisbook we are not concerned with preciselydefining tropical climates, but with how dif-ferent combinations of weather variablesinfluence the growth and yield of crops.

Any simple definition of tropical cli-mates is complicated by the effects that lat-itude, altitude, topography and atmosphericcirculation have on a general pattern. As aconsequence of this variation, local climateswithin the boundaries of the tropics can varyby as much as anywhere in the world.Strictly, the lines of the Tropics of Cancerand Capricorn, respectively 23.5° N and23.5° S of the equator, indicate the outer lim-its where the sun can ever be directly over-head. Between these limits and some waybeyond them, the solar radiation is veryintense. Here, leafy crop plants may have totranspire water quickly to stay near to thesurrounding air temperature or else theywill scorch. However, at these latitudes rain-fall may be sporadic and amounts may besmall. The water held in many tropical soilsmust therefore enable stands to remain alivefor several weeks and sometimes severalmonths without significant rainfall. Atlower latitudes, i.e. closer to the equator,rainfall amounts often substantially exceedthose at the margins of the tropics and rarelyconstrain the growth of vegetation. Betweenthese extremes, topography and particular-ly altitude can influence the pattern andavailability of rainfall over extremely shortdistances.

1.2.1 Climate and vegetation

The broad type of vegetation in a region isstrongly determined by its climate. Evenwhere human activity has greatly altered thecomposition and extent of plant cover, thelocal climate dominates all other factors indetermining where, when and for how longparticular crops can be grown. For a givenspecies, the climate sets the limit to total

Contexts 3

productivity and yield. Even in highlydeveloped agriculture, climatic factors stillalter the annual yields of the main speciessubstantially between one year and the next.In addition to direct effects on the primaryproduction of vegetation, climatic factorsalso affect the performance of plant systemsby changing the populations of pests, dis-

eases and competitors and making soil moreor less difficult to till or work on.

Because other climatic factors changeless dramatically over short distances, rain-fall is usually the most dominant factor con-trolling the type of vegetation and agricul-ture and is the basis of many classificationsof climatic zones (Thornthwaite, 1948; Troll,

4 Chapter 1

Fig. 1.2. Budyko’s diagram of geobotanic zonality. (After Budyko, 1974.) The parameter R/Lr defines therelative components of the heat and water balance (i.e. a dryness index) and R is the heat balance. Theprincipal geobotanic zones are then demarcated along this axis as straight lines. Essentially, we canconsider the vertical axis as representing heat and the horizontal axis as a measure of wetness.

Fig. 1.3. A simplified matrix of the major crop types in relation to climate.

1964). Budyko’s classification (1974) also, ineffect, includes evaporative demandthrough an index of dryness, which dependson the difference between incoming andoutgoing energy and on the rainfall. Frommeasurements at 1600 sites throughout theworld, he was able to assign a range of theindex to a broad type of vegetation – forest,steppe, semi-desert and desert (Fig. 1.2). Wecan take the basic characteristics ofBudyko’s classification of natural vegetationand create a simplified matrix of major agri-cultural vegetation, i.e. crop types that typ-ically fit within the constraints set by radi-ation and evaporative demand (Fig. 1.3).The distribution of most rainfed crops iswithin the comparable zones of forest andsteppe. Outside these zones, farmers with-out access to irrigation must select crops thatare highly resilient to the stresses imposedby shortages of soil moisture, often accom-panied by high temperatures during much,if not all, of the growing season. The physi-ology of the species and skill of managementdetermine the actual ranges of crop plantswithin these zones. (Examples are given inChapters 3, 4 and 5.)

1.2.2 Climate and wealth

From the above it can be seen that the localclimate can determine the affluence of

human communities and ultimately thewealth of nations. For example, Fig. 1.4shows the distribution of the least devel-oped countries of the world as defined bythe United Nations (1985). At the time ofthat report, the majority of the world’s poor-est countries were in a contiguous sequenceacross the Sahelian region of Africa, extend-ing south to Tanzania, Malawi andBotswana. Each country is within a regiondefined as arid or semi-arid which roughlycorresponds to the desert and semi-desertzones defined by Budyko (see Fig. 1.2).Here, although there is little diversity in thenatural ecosystem, agricultural productsstill account for the bulk of national income.In recent years, this group of countries hasmostly shown stagnant or declining agricul-tural productivity with only modest increas-es in the value of their export crops. Most ofthe other poorest countries in the world arein regions of high altitude where poor, thinsoils and topography, in addition to con-straints of climate, make agriculture diffi-cult or impossible.

At the other extreme, much ofBangladesh can receive more than 3750 mmof annual rainfall, often in torrential down-pours accompanied by strong winds. Muchof the country is barely above sea level andstorm waves and sea surges raise the waterlevel along the coast so that widespreadflooding of low-lying areas occurs, adding to

Contexts 5

Fig. 1.4. Geographical distribution of the least developed countries defined by the United Nations (1985).

the devastation caused by the strong winds.A major element of agriculture in theseregions is the uncertainty associated withthe timing, amount, intensity and durationof rainfall. Thus, irrespective of the manysocial and political constraints on them, theeconomies of the least developed nationsultimately depend on a vulnerable, uncer-tain and increasingly impoverished naturalresource base. It is the declining productiv-ity of this base that is the underlying causeof their poverty.

The dominance of a single climatic fac-tor, in this case rainfall, on the economy ofan entire country is demonstrated in Fig. 1.5.Here, despite the inevitable economic com-plexities and social pressures imposed on anewly independent nation, the annual grossdomestic product of Zimbabwe between1986 and 1996 was closely correlated withthe previous year’s rainfall.

1.2.3 Climatic variables and cropproduction

The climate at any one place is the productof year-to-year variations in daily and sea-sonal weather patterns. These patterns are aconsequence of changes in solar radiationand the atmospheric systems that controlcloud cover and rainfall. The exact effect ofeach climatic variable is often difficult todefine because weather variables are highlycorrelated and steady state conditions arerare in the field. However, seasonal trendsin the principal climatic factors can be quan-tified in terms of the latitude, altitude andgeography of a particular site. For example,Table 1.1 describes the mean winter andsummer values of irradiance, daylength andtemperature at sea level for three latitudesin the northern hemisphere.

In terms of their influence on cropprocesses, the major climatic factors inter-act between two atmospheric systems.Directly or indirectly, the solar spectrumcontrols the solar radiation, daylength andtemperature at a particular site. The terres-trial hydrological cycle determines precipi-

6 Chapter 1

Fig. 1.5. Growth in the gross domestic product (GDP, %) of Zimbabwe against previous year’s rainfall(mm).

Year

tation and evaporation. The interaction oftemperature and evaporation controls theatmospheric saturation deficit above a crop.Saturation deficit is a measure of the mois-ture content of the air and it is this factorwhich determines the atmospheric demandfor evaporation at any particular location.

Figures 1.6–1.8 illustrate how the solarspectrum and hydrological cycles influencethe annual totals and seasonal distributions

of the major climatic resources at each lati-tude. These totals and distributions set lim-its on the type of vegetation that can begrown at any location and at which time ofyear it can successfully grow. Figure 1.6shows the annual trend in daily solar radia-tion received at four locations between 1and 64° N of the equator. Interestingly, thepeak value of radiation received during theyear at Reykjavik is actually greater than that

Contexts 7

Table 1.1. The total solar radiation, daylength and mean temperature at sea level for three latitudes inthe northern hemisphere. (Adapted from Woodward and Sheehy, 1983.)

Latitude (°N)

0 20 50

WinterMean irradiance (W m–2) 411 296 87Daylength (h) 12.1 10.9 8.1Mean temperature (°C) 29 19 –1

SummerMean irradiance (W m–2) 417 451 470Daylength (h) 12.1 13.4 16.4Mean temperature (°C) 27 32 18

Rad

iatio

n (

)

Fig. 1.6. Extraterrestrial radiation received at four locations (latitude) on the globe.

at the equator. This is largely because at highlatitudes longer daylengths in the summercompensate for the lower intensities of radi-ation. However, because the daylength at theequator varies little between summer andwinter, the radiation received remains sim-ilar throughout the year and the annualreceipt of radiation therefore greatly exceedsthat at higher latitudes.

The seasonal distribution of solar radi-ation determines the temperature and rain-fall patterns at any location. Figure 1.7shows the variation in mean temperature inJanuary and July with latitude. At any timeof year, there is convergence towards theequator. However, as with solar radiation,the annual range of temperature increaseswith distance from the equator.

All atmospheric moisture originatesfrom the earth’s surface, where water in liq-uid or solid phase is transformed into watervapour via the process of evaporation. Thereare two components of evaporation. The first

is that which occurs directly from water sur-faces or from land areas which have soil orvegetation cover that has recently been wet-ted by precipitation. The second source istranspiration which is physically the sameprocess as direct evaporation but occurs viaplants which draw water through their rootsystems. Direct evaporation and transpira-tion are often combined into the collectiveterm ‘evapotranspiration’, Et, which is ameasure of the total flow of water vapour tothe atmosphere. The forcing effects of solarradiation and temperature, demonstrated inFigs 1.6 and 1.7, mean that the latitudinalvariation in annual evaporation increases atlower latitudes (Fig. 1.8). However, maxi-mum evaporation is not centred at the equa-tor but reaches two peaks located at about25° N and 15° S of the equator.

The annual evaporation at any locationis a measure of the demand of the atmos-phere for water vapour. In most cases, theprinciple means of satisfying this demand is

8 Chapter 1

Fig. 1.7. Mean temperature in January and July in relation to latitude. (Adapted from Budyko, 1974.)

through precipitation, usually in the form ofrainfall. Figure 1.8 also shows the latitudi-nal variation in precipitation across theplanet. Although the shape of the precipita-tion distribution shows a crude similarity tothat of evaporation, the two curves are notcongruent. In fact, the previously mentioneddip in evaporation at the equator coincideswith the increase in cloud cover which isitself a consequence of the maximum valuesof precipitation. These complex interactionsmean that there can be large differences inthe balance between the supply (precipita-tion) and demand (evaporation) for moistureat different locations on the planet. Forexample, whilst locations at about 55° N and25° N each receive approximately 500 mmof annual precipitation, there is an approx-imate doubling of evaporative demandacross the same latitudinal range. As a con-sequence, the supply and demand for wateris broadly similar at 55° N (e.g. temperatenorth-west Europe) whilst at 25° N (e.g.semi-arid Africa), the annual demand isroughly double the supply.

Of course, the general links identifiedabove between major climatic factors, suchas radiation, rainfall and evaporation, andlatitude are a gross oversimplification of thereal world. Other factors such as topographyand, in particular, the atmospheric circula-tion play a crucial role in determining theexact climate at any specific location.

Atmospheric circulation

The local climate and weather that cropmanagement has to contend with in any yearor field are determined by global factors.The poleward flows of air from the tropics,caused by the uneven heating of the earth’ssurface, are the main driving force for thelongitudinal air currents of the tropical cir-culation (Lockwood, 1974). Combined withthe rotational force of the earth, they causeair to move and produce large convectioncurrents which determine the global patternof winds. At the equator, the circumferenceof the earth is about 40,000 km, and anypoint on the equator is moving eastwards at

Contexts 9

Fig. 1.8. Latitudinal variation in mean annual evapotranspiration and mean precipitation. (Adapted fromWust, 1922; Sellers, 1965.)

just over 1600 km h–1. In contrast, at theNorth Pole, there is no such movement. Air,which is not itself moving, has a true east-ward velocity and momentum that is simi-lar to that at the surface. When this air massmoves to a different latitude, it will retainsome of its original eastward momentum,though much is lost with friction at theearth’s surface. As the air moves away fromthe equator, its excess eastward momentumrelative to the earth’s surface makes itappear as a west wind to an observer on theground.

The latent and sensible heat of the mov-ing air masses that originate in the tropicsare transported by the winds and graduallyreleased to colder air. The poleward movingair currents (anti-trade winds) generallydecay at 20–30° latitude. Here, the massesof air they contain fall back to the earth’s sur-face, resulting in areas of relatively highatmospheric pressure (sub-tropical highs).This circulation is completed by a massiveair stream drawn back towards the equator(trade winds). The two trade wind systems,each side of the equator, converge at an areacalled the Inter Tropical Convergence Zone(ITCZ) which is characterized by relatively

low pressure and rising air movements. Thisgeneralized global circulation is describedin Fig. 1.9.

The simplified global pattern shown inFig. 1.9 is complicated by the seasonal dif-ferences in radiation received in tropicaland sub-tropical regions some distance fromthe equator. Latitudes between 15 and 30degrees receive very large amounts of radi-ation during the summer months. The mainconsequence of the resultant heating of theland surfaces at the beginning of summer isthat the subtropical high pressures are trans-formed into centres of low pressure over thetropical continents. As a result, a new areaof intertropical convergence is establishedsome distance from the equator. The tropi-cal continents and adjacent oceans thereforeexperience a temporary reversal in winddirection, and the winds in these systemsare known as ‘monsoons’. The layer of air inwhich most of this circulation occurs – the‘troposphere’ – is shallow, with a maximumdepth near the equator of about 20 km. Incontrast, its horizontal dimensions aremeasured in thousands of kilometres.Because of this shallow and wide character,horizontal air movements (i.e. ‘winds’) pre-

10 Chapter 1

Fig. 1.9. Mean meridional circulation of the northern hemisphere. (Adapted from various sourcesincluding Budyko, 1974; Nieuwolt, 1977.)

vail. However, vertical movements are alsoextremely important because they transportheat and moisture to and from the earth’ssurface and change the stability conditionsof the air masses involved.

The monsoonal systems dominate theclimate in much of tropical Asia, Africa andSouth America and deliver most of the rain-fall at latitudes between 15 and 30°. Theconsequences of monsoons for vegetationand people are enormous. Climates con-trolled by monsoons include more than halfthe world’s people, most of whom derivetheir income almost entirely from agricul-ture. Yet within all three continents, thereare large variations in rainfall and its asso-ciated climatic effects. In Asia, parts ofMalaysia, Indonesia and New Guineareceive rainfall throughout most of the year;the Indian sub-continent, Burma, Thailandand across to the Philippines have a dry-and-wet monsoon climate, with wet sum-mers and cooler dry winters; and there areparts of the area to which monsoons fail tobring rainfall, as in the north-west of theIndian sub-continent. In tropical Africa,there is one monsoonal system in the westand another in the east, but large areas in thecentre and south have no monsoon butexperience winds of the general circulation.Much of the interior is highland plateaux,

generally at elevations above 1000 m, whichgives the interior a highland continental cli-mate. Seasonal rainfall shows a distinctbimodal distribution, with two maxima nearthe equator and a single maximum at lati-tudes greater than about 10 degrees. In trop-ical America, the variations in climate arealso very great, but more systematic, withgenerally wet climates to the east of the cen-tral mountain range, and dry ones to thewest. There is also a very great change fromthe north of South America, where annualrainfall is over 5000 mm, to the deserts ofthe south.

The annual distribution of rain by thesecirculation patterns is the main factor thatsets the start and duration of agriculturalactivities. At any location, it is not only theamount of rain but its variability within andbetween years that affects agriculturaldevelopment. When the rain does fall, it isoften in intense storms that erode the soil orcompact its surface or in such smallamounts as to be inaccessible to plant roots.The variability of mean annual rainfall alsoincreases as the climate becomes more arid,such that rain is least reliable where it ismost needed, increasing the risks associat-ed with crop production.

Sometimes tropical regions suffer asequence of bad years in which rainfall is

Contexts 11

Fig. 1.10. Rainfall index for 20 sub-Saharan stations in West Africa west of 10° E between 11° N and19° N. (Developed by Lamb, 1985.)

below the long-term average. An example ofthis is the drought in the Sahelian region ofWest Africa which lasted for 17 years from1968. Figure 1.10 shows data obtained from20 sites in West Africa, west of 20° E andbetween 11° N and 19° N (Lamb, 1985), anddemonstrates how rainfall can decline dras-tically over a sequence of years. The changeshad devastating consequences for the inhab-itants of the region. There are also recordedinstances of changes in seasonal weatherthat have lasted much longer than this. Ineastern Kenya, for example, a new 3-monthlong dry season appeared in 1913 which haslasted ever since (Musembi and Griffiths,1986), and to which agriculture must havebeen forced to adapt.

1.3 The influence of soil andtopography

At scales below tens or hundreds of kilo-metres, the shape of the land, the state of thesoils and agricultural history have an influ-ence at least as great as that of climate. Steepslopes, marshy ground and other such fea-tures severely limit the types of managementpossible, unless overcome, for example byterracing or drainage. Between extreme fea-tures, though, agriculture grades from oneform to another. Gradients of temperature,dryness, soil depth and fertility might runtogether or in opposition in different partsof the world to create zones where one formof agriculture or another predominates. Atypical transition in Africa, for example, isfrom a dry plane or valley bottom support-ing extensive grazing, through a range ofmixed animal and crop systems on wetterslopes, to vegetative crops such as timberand tea in cool wet uplands.

More local features can blur the bound-aries between zones or create extremes ofzonation over short distances. This is soeven in the humid tropics. In south-east Asiafor example, the common transition in thenatural vegetation from coastal high foreston deep wet clays and peats to low montanescrub on thin lateritic soils is reflected in theyields of crops that replace the natural veg-etation. In extreme instances, a change from

mangroves and swamp forest to seasonalgrassland with xerophytic and insectivo-rous plants can occur over two or three kilo-metres, largely as a result of change in soildepth from several metres to a few centi-metres. Such differences in vegetation aremaintained under one climate.

Whether the transitions occur rapidlyor slowly over distance, the range of cropspecies that can be grown is still determinedby physiological interactions of plants withsoil and air. Vegetative crops predominate athigh, cool regions of the tropics because lowtemperature stops reproduction in mosttropical plants before it stops their foliageexpanding. Similarly, determinate cerealsare rare on the dry plains because more oftenthan not they scorch before grain is set andripens. Selection, breeding and agronomycan all extend the range of a species, for sev-eral years at least. Over a longer time, anagricultural species can be continuallygrown in an environment only if the nutri-ents removed from the soil by the crops arereplenished. Further information on physi-ological differences between species isgiven in Chapter 4 and on sustainable man-agement in Chapter 8.

1.3.1 Heterogeneity of the soil

Zones of specified climate, altitude and ter-rain are themselves not uniform and canonly guide possible agronomic practice.Patterns on a smaller scale of hundreds ofmetres occur in both the natural vegetationand as a result of earlier agricultural activi-ty. In the Amazonian forests of Bolivia, forexample, log jams in rivers start a sequenceof events that alter the natural flora.Flooding and deposition of river-borne soilcaused by the log jams leads to death of for-est trees and temporary clearings that arethen colonized by trees such as broad-leavedmahogany (Swietenia macrophylla), whichnormally are unable to grow in denser for-est. Distributed throughout a catchment,many such events create a heterogeneousvegetation and a shifting system of tributar-ies. Removal of the natural vegetation inparts of the Amazon has also revealed evi-

12 Chapter 1

dence of past agricultural activity nowobscured from aerial view by forest. Thespecies composition of the forest will beinfluenced by this previous agriculture andany future agriculture in this and similar for-est will undoubtedly be influenced by pre-vious ecological change in the distributionof species. In such ways, agriculture and thenatural environment are commonly inter-dependent and interchangeable. It is logicalto consider agricultural management withinthe context of an agroecosystem in whichsimilar biological and physical principlescan be applied to both natural and managedspecies. A wider discussion of these princi-ples is presented in Altieri (1995) andCollins and Qualset (1999) and we return tothis subject in Chapter 8.

Localized changes in soil physicalproperties that result from weathering of theunderlying rock and from soil erosion canbe enhanced by biological activity.Systematic change over metres and tens ofmetres in the physical and chemical natureof the soil commonly occurs from the move-ment of soil particles and nutrients downslopes. Even on flattish land, complexmosaics have developed from the underly-ing chemical nature of the soil. In theKenyan savannah, for example, the patternsof grassland communities were mainly relat-ed to sub-surface sodium concentration andmound-building termites (Belsky, 1988).

Arable agriculture inherits this hetero-geneity, is forced to adapt to it and tries tomodify it. Some of the effects of spatial vari-ability can be erased by using machines toplough and smooth the land, by applyingfertilizer and irrigation and by growingcrops that are genetically uniform or homo-geneous. McBratney (1992) presents resultswhich show that the great spatial variabili-ty in pH at 0.2 m below the surface is muchreduced in an adjacent paddock. However,even when subjected to intense agriculturefor several decades, some spatial patternsare very resilient and difficult to eliminate.Often spatial variability over scales of10–100 m also persists in the populations ofnematodes and other invertebrates in manyfields.

To the farmer or researcher, natural and

imposed variations in the structure, consis-tency and biology of soils and resultant vari-ations in the growth of plant roots are oftendisguised. This is because, unlike the above-ground environment, they are hard to see.Whereas the aerial environment of a cropstand can be relatively well described bymeasuring the fluxes of radiation, tempera-ture, vapour concentrations and windspeed,etc., the vertical and horizontal fluxes ofphysical and biological elements within thesoil are latent and are, therefore, more diffi-cult to measure. In many research environ-ments, efforts have been made to amelioratenatural variations within the soil to try toachieve uniformity for experimental pur-poses. In some instances, attempts havebeen made to create ‘precision’ fields byusing machinery to redistribute soil andprovide a uniform and flat surface. It is ourexperience that, even when such drastic andexpensive procedures are attempted and doindeed produce a visually uniform surface,they are doomed to failure because the phys-ical redistribution of soil rarely achieves ormaintains a uniform composition of struc-tural or biological elements with depth.With soils, at least, what you see is rarelywhat you get.

Whereas highly mechanized manage-ment attempts to achieve homogeneity ofthe soil, farming on infertile or otherwisedifficult soils with few inputs often gains ayield by enhancing spatial variations with-in the soil. Slash-and-burn agriculture,water-harvesting, raised beds all attempt toconcentrate water and nutrients in particu-lar places in relation to crop plants. Thisimposed heterogeneity can remain for con-siderable periods, as evidenced by the per-sistence of field structures from ancienttimes. However, not all of this heterogeneityis persistent. Agronomists in Malaysia haddifficulty establishing oil palm on peats,because once drained, the organic matterwithin them oxidized, the soil level droppedand the palms fell over. Planting the palmsin a hole to an extent solved the problembecause after several years the soil surfacemoved down to be level with the base of theplant.

This type of imposed heterogeneity

Contexts 13

commonly occurs over centimetres to sev-eral metres and is usually visible in thenature of the soil surface or among the plantsthemselves. It is on smaller scales, however,that most tropical management interactsdirectly with the heterogeneity of the soil.The spade and plough more generally affectcrucial processes at scales of millimetresand decimetres: those at which channelsform through which water and gases move,roots explore and exploit, and bacteria,fungi, protozoa, nematodes and other inver-tebrates graze and predate. Some typicalscales at which different processes work inthe soil are shown in Fig. 1.11.

The fine structure of soil

Particularly important for plants are theactions of microorganisms. Some of theseare closely associated with the plants them-selves. The nitrogen-fixing symbionts thatlive in the roots of legumes and certain otherplants contribute nitrogen, while the fungithat combine with roots to form mycorrhizacan enhance the uptake of other elements.Besides these, the fertility of soil dependson many free-living bacteria and fungi,

which break down organic matter and takeup, store, transform and release nutrients,and whose secretions and dead bodies givecohesion and structure to the soil. Structureis important to enable a soil to hold a diverseassemblage of organisms and to resist thepower of rain and wind to erode soil parti-cles and wash out nutrients.

Many soils are able to support a com-munity of microorganisms having a muchgreater mass and containing more nutrientsthan the crops themselves. The microorgan-isms compete with plants for the nutrients,which they lock up for a time, but thenrelease back to the soil or to the atmosphere.One of agronomy’s main aims is to under-stand how to optimize rates of amassingmicroorganisms in stable communities andof releasing nutrients when plants can bestabsorb them. One of husbandry’s greatestskills is to do this blindly, without anyimmediate information about microorgan-isms or nutrients.

Implications for agronomy

Agronomy and husbandry work blindly atthe fine scales of the soil pores and chan-

14 Chapter 1

G

T

C

W

Cl

M M M

B

S

W

N

R F

A

H

Hy

H

B

R

S S

Fig. 1.11. Typical scales at which different processes work in the soil.

nels. To the experienced, the success of apractice can be judged by the feel and work-ability of a soil and the yield of crops overseveral years. To a large extent, science isblind here also. There are enormous diffi-culties in understanding links between themicroscopic nature of soil, tillage and yield.Moreover, the synthesis of knowledge aboveand below ground has been hampered bytraditional divisions in research and teach-ing that still persist, and in some cases arebeing reinvented. At present, science knowsperilously little about the effect of soil struc-ture and function on the way plant standsestablish, absorb resources, grow and yield.Accordingly, quantitative descriptions ormathematical models of soil and plant tendto simplify, and often misrepresent, theprocesses concerned. We know, for exam-ple, that soil is not a homogeneous mass. Itoften has cracks and channels through itcaused by worms, termites and moles orcontraction during drought. One of theauthors remembers pouring several bucketsof water down a 2 cm wide crack in adroughted lateritic soil: the crack showed noindication of being filled with water.Following a dry season, rain washes downsuch cracks for a time, possibly leavingmuch of the bulk soil dry. The reality, how-ever, is more as represented in Fig. 1.12,where many channels of different size pen-etrate a matrix of fine and tortuous poresthrough which water, gases, nutrients,microorganisms, roots and invertebrates areunevenly distributed.

Because the fine structure of soils is socomplex, many models and equations still

treat soil as if it were homogeneous. As afirst step, this provides the student with themain soil factors that can influence crops.However, an important result of soil hetero-geneity is that the value of a soil factor usu-ally changes with the scale of measurement.The rate at which water infiltrates soil, forexample, will often increase if a measuringdevice of larger diameter is used as largerand larger pores and holes become encom-passed by the device. Similarly, at an invis-ible scale, bacterial processes might have agood supply of oxygen if they occur on theoutside of soil aggregates but be starved ofoxygen if they occur deep inside fine pores,only a few millimetres away. Interpretingsoil function through an average or bulkvalue can therefore be misleading. Forexample, a soil at one bulk measure of oxy-gen concentration will likely support bothanaerobic as well as aerobic processes.

There is growing interest in whetherheterogeneity and its effects at all theseregional and microscopic scales can beinterpreted through a common set of princi-ples. We return to this theme in the finalchapter. In Chapters 3 and 7, we presentsome simplified guides to the processeslinking soils to plant growth. Farmers andagronomists know that the reality is morecomplex, but it is hoped the simple conceptswill help to initiate more intensive anddirected agronomic experiments, lessdemanding of time and land, as well asattempts to understand heterogeneity froma common starting point.

Contexts 15

Fig. 1.12. Simple representations of soil structure (diagram provided by Dr B. Marshall).

1.4 Climates, crops, soils, people

One of the biggest challenges facing the trop-ical agronomist is to link the various physi-cal and biological processes that set thepotential and determine the actual perform-ance of a diverse range of crops and crop-ping systems with the human dimensioninvolved in crop management, preferenceand choice. A logical starting point is todefine the environmental conditions withinwhich plants and humans have to operate.

1.4.1 Agroclimatic classification

One response to the unruly diversity of agri-culture has been to classify it. The system ofagroecological zones in Kenya (Jaetzold andKutsch, 1980; Jaetzold and Schmidt, 1982)is based on soil, topography, plants and ani-mals as well as climate. Their classificationis a matrix of rows and columns similar tothe simplified scheme depicted in Fig. 1.3.

Land classifications were originallydone by assimilating knowledge from sur-veys and estimates provided by regionalagricultural departments. Ground surveyscan now be augmented by analysing photo-graphs from aircraft and satellites. The bal-ance of different wavelengths in the spec-trum of visible and infrared radiation issometimes used to assess the developmen-tal stage and health of vegetation. Caution isneeded, however, over received classifica-tions of agricultural land and soil. Some soilscientists and agronomists recognize ‘cottonsoils’, ‘cocoa soils’, ‘rubber soils’ and othersoils that they recommend specific crops tobe grown in. Some national organizationshave even produced maps showing wherespecies should be planted. Land use typesand categories might be useful for legislat-ing against malpractice or encouraging hus-bandry that stabilizes slopes, for instance,but often they do not correspond with whatis growing or will be grown in the ground.

1.4.2 The reality of infrastructure, inputsand profit

While climate, soil and topography togeth-er determine the potential range of a plantspecies, and the range of species that can begrown at any site, other factors contribute to,and sometimes dominate, what is grown ata site. Chief amongst these are the availabil-ity of plants, the desire to grow rich off theland and the need to appease hunger.

Early trade, human migrations andcolonialism moved plants from one conti-nent to another. South-east Asia, for exam-ple, has an ideal climate for perennial treecrops, but massive and profitable plantationagriculture developed there only after rub-ber was brought from South America and oilpalm from West Africa. Potatoes from theAndes became a staple diet in parts ofEurope (hence the ‘Irish’ potato) and arenow found in cool tropical regions through-out the world. Maize from central Americafeeds most families in Africa. Given that fewbarriers now exist to the movement of cropspecies, the range of crops and products incash agriculture, and the resulting pattern ofland usage, often depend as much on policy,economics and preference as on the physi-cal environment (see Chapters 4 and 8).

Tea, for example, is cultivated in broad-ly similar cool-to-warm moist climates at arange of sites throughout Africa and Asia. Itis also grown at hot, moist, previouslyforested lowland sites in Assam and SriLanka, but only in small quantities at simi-lar sites in Malaysia and hardly at all inmany other similar climates. Tea could begrown at most of the hot, moist sites in theworld, but is often simply not as profitablethere as rubber, cocoa or oil palm. Duringthe 1980s, oil palm was so profitable as acrop in south-east Asia that it was grown inpreference to all other crops on all types ofsoil. Similarly, a single species – sugar cane– dominates the landscape of Mauritiusregardless of terrain and local climate. Insubsistence cropping, also, factors otherthan those of the climate and land contributeto profitability and success. Particularlyimportant are the availability of inputs suchas fertilizer, whether farmers can afford to

16 Chapter 1

buy them, their cost in relation to the returnthey bring, the costs of transporting produceto the market, the selling prices and the asso-ciated subsidies, if any. Most importantly,despite any analysis or prediction based onphysical and economic factors, many cropsare grown in preference to others simplybecause farmers and their families knowhow to grow them and like to eat their prod-ucts. The lesson is always to look fartherthan the local physical environment to coverall possible reasons why a crop is where itis and how many human and physicalresources are expended on it.

1.5 Agronomic experiment andanalysis

Agronomy, as any other biological field ofstudy, is faced with understanding complexprocesses. It also has to synthesize knowl-edge from many sources and disciplines tosolve immediate and practical problems.Agronomy can rarely achieve understand-ing at scales of the farmer’s stand and fieldbecause of the many couplings and feed-backs that exist between weather, soil andplants and management. If one factor ischanged, several others change also.Agronomic research therefore developed astrong experimental base.

The essential feature of biological andagronomic experiments is to break some ofthe couplings between variables. Forinstance, plants might be grown in con-trolled environments where temperature isvaried but water held constant, or standsmight be grown in small plots over a rangeof planting density, each receiving the sameamount of fertilizer. Such experimentstighten the ‘boundaries’ of the system: ifthey make temperature vary, for example,they will attempt to keep within definedbounds for the degree of variation in satura-tion deficit. The problem is then examinedwithin the new boundaries. The ultimateaim is to move back to the original problemin the field, and to do this, the true effectsof the broken couplings have to be inferredfrom what was learned about the processes(Squire and Gibson, 1997).

In reality, it is very difficult to make thismove back to the field. Problems tend to betackled at different scales by people fromdifferent scientific disciplines and codessuch as soil science, physiology and genet-ics. When the disciplines have remainedisolated, problems at the field scale haverarely been solved. Even when disciplinescombine, there are still fundamental scien-tific challenges of scaling, in which somemental or mathematical procedure is usedto predict from measurements in experi-ments what is likely to happen at the scalesof crops in soils subject to variable weatherand heterogeneous soil (Marshall et al.,1997). A further complication is that thetools of mathematical interpretation are notalways familiar to the biological practition-er or researcher and provide another barrierto understanding.

In some areas of science, so little isknown about couplings at the scales of thepopulation, field and catchment that theinferences can only be guesses. Predictingeffects of global environmental change onvegetation, for example, is very uncertainbecause experiments are not feasible atscales anywhere near to those of the ques-tion. The fate of introduced genes and cul-tivars also presents particular problems ofscaling. The initial change occurs at thescale of the individual genome or small sec-tion of DNA. The ramifications might makemillionaires, destroy a country’s soil or con-tribute to a stable equitable agriculture.

Agronomy itself is fortunate in thatexperimentation at near-realistic scales ispossible. Indeed, agronomy has generated arange of experimental and theoreticalmethodologies, among which are empiricalfield trials, physiology and environmentalphysics.

1.5.1 Empirical field trials

A multi-factorial field trial is put in place toget a response curve at the field scale to avariable such as fertilizer or planting densi-ty. The purpose of such trials is to draw outeffects of treatment from those of localtopography, nutrient status and other soil

Contexts 17

factors. The treatments are repeated in anumber of blocks sited in different parts ofthe experimental field. A branch of statisticshas been developed to analyse the results ofthese experiments (see, for example, Lane etal., 1987) and assess their validity.

An end-point, such as yield or the qual-ity of yield, has often been the only responsequantified. Nevertheless, trials coordinatedthroughout a country or region can providea description of the broad effects on yield ofsoil type, climatic variation and nutrition.The coordination of disparate fertilizer tri-als in Malaysia by the national oil palmplantation body is a good example of thisapproach (Foster and Goh, 1977).

Scaling-down field trials has been triedto obtain, in a much smaller space, aresponse curve of yield to nutrients or plant-ing density. Examples are fan designs,developed for vegetables (Nelder, 1962;Bleasdale, 1967) but applied to crop plantsas large as oil palm (Corley et al., 1971a), andline-source irrigation experiments (Hanks etal., 1976). The aim of all miniature trials isto establish a gradient of the variable beingstudied. Plants at any single point on thegradient might extract additional resourcesat higher or lower points, so scaling to thelevel of the commercial field is not straight-forward. Nevertheless, the small, response-curve trial can give a good first indicationand breadth of effect generally and, more-over, is at a scale quite appropriate for mixedgardens and uneven terrain where steep gra-dients in fertility and soil depth are com-monplace.

Empirical field trials have remained thestock-in-trade of tropical agronomy fordecades. The working doses of fertilizer andirrigation in many crops have been definedfrom the results of such trials. They havelimitations, however. They are repetitiveand need much land, money and time. Theyrarely identify links between the physicalenvironment and yield, so are site-specific:the results at one site can seldom be appliedelsewhere. Moreover, the scale of the plot ofplants whose yield is measured and that ofthe guard area around them have to be rep-resentative of the real conditions underwhich the crop will be grown. Plantation

crops such as oil palm, rubber and tea cov-ering vast stretches of land in a more or lessunbroken blanket, have very different edgesfrom the single rows and clumps of plantsin subsistence gardens. The purpose of theguard area is to minimize the effect of theedge of the experiment or the edge of anoth-er treatment. Plants at an open edge get morelight and nutrients than those nearer thecentre of a plot, but they also lose morewater. Therefore, they might grow faster orslower than the centre ones according to thecircumstances. As a general guide: the larg-er the extent of a crop, the more importantit is to remove the edge effect in any trial. Sosmall plots of wheat or maize in breeding tri-als, for example, that are grown withoutguard areas will yield much more than theywould in an intensely managed field, sim-ply because individual plants in the plotreceive more light. Scaling these resultsupwards would lead to an overestimate ofthe yield from a large field. Scaling resultsdown from large trials would likely under-estimate the yield of single rows or singleplants.

Replicated field trials will remain animportant feature of agricultural research formany years, especially where they are prac-ticable and affordable. The branches ofagronomy that seek mechanisms that linkweather, soil and crop have the importantfunctions of extending the results from tri-als, finding alternative short-cuts and even-tually removing the reliance on these trialsby many tropical research stations. Theyalso contribute to an understanding ofcause-and-effect which, as well as explain-ing what we already can see, allow us to pre-dict the future performance of crops andcropping systems in uncharted territory. Inthis context, modern techniques, such as theuse of computer simulation models of cropgrowth and yield have an important role toplay in furthering our understanding ofwhat we know and, perhaps more impor-tantly, what we don’t know about particularcrops and cropping systems. We return tothis theme in Chapter 8.

18 Chapter 1

1.5.2 Physiology in controlledenvironments

To understand mechanisms linking plantswith environment, a branch of agronomyretreated to controlled environments wherethe heterogeneity and couplings of the out-doors could be eliminated. In the laborato-ry, agronomy joined with basic physiologi-cal science to investigate the independenteffects of light, water, nutrients, tempera-ture, photoperiod and other variables.Accordingly, several crop plants such asmaize, wheat and barley became used wide-ly in laboratories as objects of basic physio-logical research.

Experimentally restricting the cou-plings among processes and environmentoften brings about untypical responses inplants (Raper and Downs, 1976). Low radi-ation in chambers has so often limited thecarbohydrate for plant growth that theresponse to other variables has simply notoccurred. Altering temperature while allow-ing humidity to vary unchecked is anothercommon limitation, as is maintaining agiven relative humidity over a range of tem-peratures rather than a saturation deficit (seeChapter 3). Some experimenters thereforemoved to systems that did not rigidly con-trol all variables. Open-topped chambersand controlled-environment glasshouses,for example, allow sunlight to vary more orless naturally while still regulating factorssuch as temperature and carbon dioxideconcentration. Others went further to main-tain that, since plants rarely experience aconstant environment, daily variation and‘noise’ in factors such as sunlight, tempera-ture and saturation deficit should be includ-ed within the experimental treatment.Moreover, they insisted that plants’ rootsand canopies should be allowed to formwithout being severely restricted by theircontainers (Monteith et al., 1983).

Physiological studies in controlledenvironments have made important contri-butions to understanding crop growth in thefield, but only for specific processes. As ageneral guide, the responses to environ-mental factors that exert local effects, and donot have to operate through a whole-plant

growth rate, have been reasonably wellquantified by physiological experiment.Many effects of temperature and photoperi-od on rates of leaf initiation and floral devel-opment are of this type (Chapters 2 and 5give examples). In contrast, those effects ofsunlight, water or nutrients, for example,that operate through a whole plant or wholestand have often been misrepresented incontrolled environments. A common error –as already indicated – has been to observeno response to varying one factor, such asnutrient concentration in the medium,while another factor such as light is limitingthe rate of growth. Agronomy should notassume that the variable factor exerts littleor no effect in the field.

1.5.3 The reality and uncertainty of thefield

A parallel approach of plant science was tomeasure and attempt to interpret yieldthrough processes of growth and develop-ment. Early fieldwork yielded growthcurves in different environments, or popu-lation structures of species in competition(e.g. Harper, 1977). In agriculture, the direc-tion of this approach became concerned lesswith the variability within populations thanwith the dimensions of structures that makeup a stand. A central analysis developed inwhich the performance of a stand wasexpressed in relation to some measure of itssize or surface area (Rees, 1963). The rela-tive growth rate, for example, is the dry mat-ter increment over a period divided by thedry matter at the start of the period; whilenet assimilation rate is the dry matter incre-ment divided by the leaf area index.

This form of expression provided acommon means of defining growth, butoften led to misunderstandings. Expressingthe rate at which new dry matter is producedin relation to plant mass, for example, meansthat even if the absolute rate were stable –or even maximum – the relative rate wouldtend to decline as plants become larger. Adeclining rate during the main phase of cropgrowth might suggest that something iswrong. By contrast, a declining relative

Contexts 19

growth rate and stable absolute growth rateare common in healthy crops once canopiesintercept most of the available light (Squireand Corley, 1987).

Growth analysis in the field sufferedsome of the limitations of empiricaltrialling. In particular, the results from anysite, season and crop were not transferableto other environments. A stand of a givenleaf area index (i.e. the (dimensionless) leafarea per unit field area) might grow at aslower rate or for a shorter time at one placecompared with another, but the methodolo-gy was able to give few insights as to whyfactors such as leaf area index were not gov-erning absolute growth rate between envi-ronments. These insights came with themerging of physical and plant sciences overthe last few decades.

1.5.4 The contribution of environmentalphysics

Dimensions such as leaf area index becamemore directly linked to the energy in theenvironment following the concept ofpotential evaporation, which was an esti-mate based on physical theory of the maxi-mum rate that water could be lost from veg-etation (Penman, 1948). Hutchinson et al.(1958), for example, used the concept ofpotential evaporation to estimate water useby cotton in Sudan. They derived, experi-mentally, a relation between leaf area indexand evaporation as a percentage of potentialevaporation. They also brought to the fieldthe concept that the onset of water strainduring the season could be detected bymethods of relative turgidity developed byphysiologists in the laboratory. Their analy-sis accounted for much of almost a tenfoldvariation in yield among planting dates andyears (see Chapter 5).

However, it was not until the physiolo-gy of gas exchange was applied to growth(e.g. de Wit, 1958) and dry matter changesin whole stands were examined in stands,not inside laboratories through flows ofenergy, carbon dioxide and water (Monteith,1965, 1972) that substantive progress with

understanding the basis of growth and yieldwas made. These ideas led to expressing theprocess of primary production as a sequenceof factors linking the resource (value = 1) toyield (a very small fraction). Several of thefactors were entirely of the physical envi-ronment, such as the fraction of sunlight thatcloud or dust reflect or absorb. Others, suchas the energy allocated to a particular struc-ture, were largely physiological, though stillresponded to environment. The crucial fac-tors linked the physical resources to thephysiology: those, for example, that deter-mine how much light is absorbed for a unitleaf area, or water for a unit root length, orhow much plant material is made from aunit of energy absorbed. We return to thistheme in Chapters 2 and 3.

The principles of this approach spreadrapidly, and were applied in many temper-ate and tropical countries. Gradually, bycomparison and synthesis of many studies,it became apparent that several of the factorswere stable. They were not constant, butthey changed only a little between seasonsand fields: they were conservative and – asdemonstrated later – they allow back-of-envelope calculations, as well as complexcomputer simulations, of production andyield.

Some standardization has now arisenfrom applying the principles of environ-mental physics and physiology to agricul-ture. Although nomenclature, symbols andunits of measurement still vary, certainphysical quantities have become, in effect,currencies by which to compare crops in dif-ferent seasons and places. The analysisfavoured in this book as a practice for trop-ical agronomists interprets growth and yieldof different plants and crops using ‘curren-cies’ of solar radiation and water. Ourapproach therefore is founded on the prin-ciple that management seeks to manipulatethe flows of solar radiation and waterthrough individual plants and plant com-munities to a range of desired products. Thefactors controlling fluxes of, specifically,radiation and water are presented inChapters 2 and 3.

20 Chapter 1

1.5.5 Agronomic decisions and sequenceof chapters

Following Chapters 2 and 3, the rest of thisbook examines the way knowledge of fluxcurrencies can be used to inform agronom-ic practice. The sequence of chapters arguesthat choices and operations in managementcan be separated into two broad groups. Inthe first are those that govern what, whenand how to sow a crop: factors that all farm-ers must decide around the time of sowing.They require decisions about the genotypesto be grown, the time of sowing and of otheroperations, and the intended configurationof the stand as influenced by the populationdensity and species composition. Genotype,timing and configuration are examined inChapters 4, 5 and 6, respectively. The sec-

ond group of factors includes operationsthat are not always done, such as applyingfertilizer, water-harvesting techniques andirrigation, and reducing competition fromweeds. These are considered together inChapter 7. Inevitably, several subjects arecovered only briefly or have been omitted.These include the effects of pests and dis-eases and localized soil problems such assalinity, waterlogging and the effects of acidsulphates and peats. In practice, knowledgeof the flux-currencies could help to eluci-date any of these problems. A final chapterexplores the wider implications of theapproach and indicates where progress anddeeper understanding are needed. It alsolooks to the exciting methodologies thatmight be absorbed by agronomy in thefuture.

Contexts 21