A review of belowground interactions in agroforestry,focussing on mechanisms and management options

G. SCHROTH

University of Hamburg, Institute of Applied Botany, P.B. 30 27 62, D-20355 Hamburg, Germany(Present address: c/o EMBRAPA Amaz nia Ocidental, C.P. 319, 69011-970 Manaus-AM, Brazil;E-mail: schroth@internext.com.br)

Key words: competition, complementarity, facilitation, root activity, root distribution, rootmanagement

Abstract. This review summarises current knowledge on root interactions in agroforestrysystems, discussing cases from temperate and tropical ecosystems and drawing on experiencesfrom natural plant communities where data from agroforestry systems are lacking. There is aninherent conflict in agroforestry between expected favourable effects of tree root systems, e.g.on soil fertility and nutrient cycling, and competition between tree and crop roots. Root man-agement attempts to optimise root functions and to stimulate facilitative and complementaryinteractions. It makes use of the plasticity of root systems to respond to environmental factors,including other root systems, with altered growth and physiology. Root management tools includespecies selection, spacing, nutrient distribution, and shoot pruning, among others. Root distri-bution determines potential zones of root interactions in the soil, but are also a result of suchinteractions. Plants tend to avoid excessive root competition both at the root system level andat the single-root level by spatial segregation. As a consequence, associated plant species developvertically stratified root systems under certain conditions, leading to complementarity inthe use of soil resources. Parameters of root competitiveness, such as root length density,mycorrhization and flexibility in response to water and nutrient patches in the soil, have to beconsidered for predicting the outcome of interspecific root interactions. The patterns of rootactivity around individual plants differ between species; knowing these may help to avoidexcessive competition and unproductive nutrient losses in agroforestry systems through suitablespacing and fertiliser placement. The possibility of alleviating root competition by supplyinglimiting growth factors is critically assessed. A wide range of physical, chemical and biolog-ical interactions occurs not only in spatial agroforestry, but also in rotational systems. In a finalpart, the reviewed information is applied to different types of agroforestry systems: associa-tions of trees with annual crops; associations of trees with grasses or perennial fodder and covercrops; associations of different tree and shrub species; and improved fallows.

Introduction: The conflicting role of roots and the concept of root management

Tree root systems are involved in some of the major favourable effects onsoils and crops expected from the application of agroforestry techniques. Theseinclude soil carbon enrichment through root turnover, the interception ofleached nutrients, or the physical improvement of compact soil layers. Inaddition to such facilitative effects of trees on associated crops, deep-rootedtrees may use soil resources which are not accessible to the crops, leading toa certain complementarity of the use of soil resources and consequently an

Agroforestry Systems 43: 5–34, 1999. 1999 Kluwer Academic Publishers. Printed in the Netherlands.

increase of the resource use by the association as a whole. On the other hand,trees can compete with associated crops through their root systems, and thismay lead to yield depressions and may contribute to the economic failure ofland-use systems (Schroth et al., 1995b).

Because of this inherent conflict between facilitative, complementary andcompetitive effects of root systems in agroforestry, the definition of desir-able root characteristics for agroforestry species is a complex task whichrequires a detailed understanding of root-soil and root-root interactions. Onthis basis, it may be possible to identify suitable root criteria for the selec-tion of the woody and herbaceous components of different agroforestrysystems at different sites and to include them in breeding and selectionprogrammes, together with criteria of productivity and product quality, shadingand shade tolerance, site adaptation and site improvement etc. However, thequestion would remain whether species which fulfil these root-related criteriacan actually be found (or be produced by breeding), and if so, whether theyare also sufficiently interesting in an economic sense to be planted by farmers.

There may however be other possibilities to improve root functions inagroforestry systems. Root systems, i.e. their distribution, activity andturnover, are not only genetically determined, but they are also the productof environmental factors which may be manipulated by the land-user. Thispoint is of critical importance, because it may increase considerably the rangeof species from which he can choose those most appropriate according toany not root-related criteria, such as productivity etc. By applying combina-tions of, e.g. associations with other plant species, fertiliser distribution andshoot pruning, it may be possible to manipulate the root systems of tree speciesin a desirable way, even if these species do not possess an ideal root systemaccording to certain criteria (e.g. vertical distribution, lateral extension) in theabsence of these measures. The application of techniques which aim tooptimise root functions and root interactions in agroforestry or other land-use systems may be called ‘root management’ (Bowen, 1984; Schroth, 1995).

Aim of the paper

The objective of this review is to explore how the root systems of associatedplant species interact under different environmental conditions and whichoptions exist to influence these interactions through the selection of speciesand the design and management of agroforestry systems.

Not only research results from man-made land-use systems will be dis-cussed, but also data and observations from natural vegetation. The study ofnatural plant communities may provide insights into the structures andmechanisms through which excessive competition, leading to mutual exclu-sion, between associated species is avoided. These could then eventually beemployed in a more or less modified way in agroforestry systems. One ofthe basic ideas of agroforestry is in fact to design land-use systems whichare closer in their structure and function to natural plant communities than

6

conventional agriculture and plantation forestry, and which may therefore alsoexhibit some degree of their site adaptation and sustainability, especially underadverse pedoclimatic conditions and at a low level of external inputs (Ewel,1986).

Mechanisms of root interactions and perspectives of their management

Root distribution and root interactions at the root systems level

A fundamental hypothesis of agroforestry is that different plant life forms suchas trees and herbaceous crops or pastures occupy to some extent differentsoil strata with their root systems when grown in association, leading to adegree of complementarity in their use of soil resources. For this reason,rooting depth and the vertical distribution of root systems are of particularinterest for agroforestry. Rooting depth determines to which extent plants (e.g.trees) can use subsoil water and nutrients which make them less dependentfrom the supply in the topsoil and may also be made available to associatedplants (e.g. crops) with shallower root systems through nutrient pumping andhydraulic lift (Emerman and Dawson, 1996). Deep-rooted plants may also beless affected by root competition or by mechanical disturbance, e.g. throughtillage, of their root systems in the topsoil than shallow-rooted species.

Regarding rooting depth, it is important to distinguish between the totaldepth explored by a root system and the distribution of the roots in the soilprofile. Sometimes, plants are characterised as shallow-rooted when theyexhibit a rapid decrease of root mass or length density with increasing soildepth in the first decimetres of soil (Schroth and Zech, 1995a). Such plantsare likely to take up a smaller proportion of their water and nutrients fromsubsoil horizons than plants which do not show such steep gradients of verticalroot distribution. Nevertheless, plants may have a high concentration of rootsin the topsoil and still have some very deep roots which may reduce theirsusceptibility to drought. For example, several fruit tree species of thetemperate zone may root to more than 4 m depth, but most of their roots arenear the soil surface (Buwalda, 1993). Some tree species in Amazonian rain-forests have roots to 6 m (Chauvel et al., 1993) or even 18 m depth (Nepstadet al., 1994), although very dense root systems are typically found in thisvegetation type near the soil surface and in the litter layer (Sanford, Jr. andCuevas, 1996).

The ability to form relatively deep root systems is not restricted to woodyperennials. Many annual crops including maize (Zea mays), groundnut(Arachis hypogaea) and cowpea (Vigna unguiculata) also develop root systemsof more than 1 m depth under favourable soil conditions (Taylor, 1980). Inan ultisol in humid Puerto Rico, four perennial grass species (Panicummaximum, Cynodon nlemfuensis, Erichloa polystachya and Brachiariabrizantha) reached 1.8 m soil depth with some roots, and at 150–180 cm depth,

7

P. maximum (guinea grass) had a higher root length density than the shrubspigeon pea (Cajanus cajan) and coffee (Coffea arabica). Both guinea grassand the also very deep-rooted pigeon pea were known to be particularlydrought resistant in the region (Rivera et al., 1983).

Such differences in root distribution between plant species, growing at thesame site and under similar environmental conditions, illustrate the geneticfactors of root development. These interact with environmental factors indetermining the shape and functioning of root systems, including soil strength,soil temperature, soil aeration, water and nutrient availability, and microbialand faunal interactions (Smucker, 1993). If dry, infertile or compact soil zonesrestrict the growth of a portion of a root system, increased root branching mayoccur in the less restricted zones in the soil, a phenomenon called ‘compen-satory root growth’ (Miller, 1986). Through its water and nutrient uptake andpossibly adverse chemical (allelopathic) interactions, one root system can alsocause restrictions to the development of another root system, thereby influ-encing its distribution in the soil. In an experiment in Britain, the closer thespacing of apple trees (Malus sylvestris) and thus the intra-specific rootcompetition in the topsoil, the more roots grew vertically into the soil insteadof spreading horizontally near the soil surface (Atkinson et al., 1976). Thismay negatively affect the individual plant, but may be positive for the systemas a whole: Increasing the tree density increased the subsoil water use in asilvopastoral system in Australia (Eastham et al., 1990) and may increase thenutrient uptake from the subsoil in other cases.

Vertical stratification of root systems

In certain cases, associated plant species form vertically stratified root systems,i.e. their root systems occupy more or less separate soil layers (Lyford andWilson, 1964). This situation is desirable in many agroforestry situationsbecause it is likely to reduce overall root competition and to lead to a morecomplete utilisation of soil resources. The ‘safety net’ of tree roots below theroot systems of associated crops (van Noordwijk et al., 1991a) refers to asituation of vertically stratified root systems, in which the tree roots absorbnutrients which have not been taken up by the shallower-rooted crops andhave therefore been leached out of the topsoil. Such a root distribution mayor may not develop in zones of overlapping root systems, as is illustrated bythe following three examples.

1) In the coastal vegetation of California, root competition from an invadingplant species with a very dense root system, Carpobrotus edulis, provoked thedownward displacement of the root systems of two native shrubs, Haplopappusericoides and H. venetus. The maximum number of lateral roots arising fromthe main taproot of H. ericoides was displaced from 0–30 cm without com-petition from the invader to 20–50 cm with competition. Removal of theinvader improved both the water relations and the growth of the shrubs(D’Antonio and Mahall, 1991).

8

2) In a South African savannah with 630 mm rainfall and very sandy soils,the maximum root length density of the dominating tree species Burkeaafricana, Ochna pulchra and Terminalia sericea was between 10 and 40 cmsoil depth, whereas that of the grasses Eragrostis pallens (in the open) andDigitaria eriantha (under the trees) was at 0–10 cm depth. Tree growth wasnot improved by experimental grass removal, suggesting that in this case, themaximum of the tree roots was in the subsoil not because of grass competi-tion but because of periodically dry topsoil (Knoop and Walker, 1985). Thereaction of the tree roots to grass removal was not investigated.

3) In the relatively humid Lamto savannah of Côte d’Ivoire, shrubs as wellas grasses had their maximum fine root mass in the upper few decimetres ofsoil, although some roots of both groups extended to 1.8 m depth (Le Rouxet al., 1995). The water uptake by one shrub species, Crossopteryx febrifuga,was largely restricted to the same topsoil layers (0–0.6 m) as that of the grassHyparrhenia diplandra, whereas another shrub species, Cussonia barteri,explored the water reserves of deeper soil layers during the dry season (LeRoux and Bariac, 1998). This illustrates a genetic difference between the twoshrub species, which precluded a vertically stratified resource use in theassociation with the shallow-rooted grasses for one of the shrub species(Crossopteryx febrifuga), but not for the other species (Cussonia barteri) undersimilar environmental conditions.

These examples show that vertical stratification of root systems can, butdoes not necessarily develop in zones of overlapping root systems, eitherbecause one root system forces the other one into greater soil depth (i.e.through root competition), or because of abiotic factors to which two plantspecies respond differently with their root distribution.

Among cultivated tree species, apple trees possess highly plastic rootsystems which readily respond to changes in their growth conditions. Incomparison to growth under clean-weeded conditions, intercropping withmaize confined the root systems of apple trees laterally and provoked a morepronounced vertical root development in an experiment in Nebraska. Beneaththe root systems of the associated maize at 2.4 m depth, the tree roots con-tinued to spread laterally and may have acted there to some extent as a ‘safetynet’, although these roots were relatively few (Figure 1) (Yocum, 1937). Thestudy shows that competitive, deep-rooted grass species may be used to forcetree root systems into considerable soil depths. However, at this depth, theremay be little lateral root development of the trees. Such deep biologicalbarriers for lateral tree roots may increase nutrient recycling from the subsoilby the trees or protect less competitive species further away from tree rootcompetition. They may fail to induce the expected formation of verticallystratified root systems if they are too deeply rooted. Vertical stratification maydevelop more easily in associations with shallower-rooted species.

In summary, it is clear that plant species differ in their ability to formstratified root systems, and if different species are mixed together with theaim to create this root distribution through competitive downward displace-

9

ment of one root system by another root system, it seems that the followingconditions must be fulfilled: 1) The root system of one species must be suf-ficiently competitive to displace the root system of the other species, if not,the root systems will intermingle (this can be desirable in some situations,e.g. associations of trees with leguminous cover crops, see below); 2) theroot system of the second (displaced) species must be sufficiently flexible torespond to the restriction with compensatory root growth in depth; 3) the

10

The markings on the horizontal and vertical axes indicate intervals of 61 cm (2 feet). Note howthe apple tree roots turned downward as they came into competition with the maize and howthey continued to spread laterally below the maize root system (from Yocum, 1937).

Figure 1. Root system of a three-year old apple tree (Malus sylvestris) developed in competi-tion with maize at 1.05 m (3.5 feet) distance during three seasons in eastern Nebraska, USA.

root system of the first species must be sufficiently shallow so that the secondspecies still exhibits relevant lateral root spread below the root system of thefirst species; 4) the soil conditions must permit root growth in depth, i.e. thesubsoil must not be too compact, dry, infertile etc. in relation to the topsoil.To predict the root distribution of such species associations and select thespecies accordingly, quantitative measures for the relevant root and soilproperties need to be developed.

Further complications arise because the conditions under which a rootsystem develops nearly always change with time. Sinclair et al. (1994)describe an agroforestry technique in Sumatra, where the tree speciesPeltophorum pterocarpum is planted into Imperata cylindrica grassland,develops roots mainly in the subsoil as a response to the competition fromthe grasses, then shades the grasses out and remains as a tree with a favourableroot distribution for intercropping. They suggest that suitable root-relatedselection criteria for trees in this system would be a) deep rooting and adownward displacement response to grass root competition; and b) a reducedsurface root proliferation response during the subsequent intercropping phase(Sinclair et al., 1994). The combination of these properties may however bedifficult to achieve, because criterion a) requires a flexible tree root systemwhich responds to growth restrictions in the topsoil through compensatoryroot growth in depth, whereas criterion b) requires an inflexible root systemof the trees which does not respond to the removal of the grass root compe-tition and the creation of favourable topsoil conditions for the annual crops(fertilisation etc.) through a readjustment of its root distribution, but keepsits root system in the subsoil. Nevertheless, the example illustrates well thatspecies selection in a root management context does not simply mean to selecttrees or other plants with fixed, desirable root characteristics, but rather toselect species which will develop a root system with certain favourableproperties under the conditions of a specific agroforestry situation and whichreact to root management measures in a predictable and desirable way.

Interactions at the individual root level

The restrictions of lateral root development and the formation of verticallystratified root systems in the contact zone of competing root systems may beseen as mechanisms which plants have developed to avoid excessive intra-and interspecific root competition. Similar phenomena have been observedat the scale of the individual root, and an interesting question is whether suchmechanisms can also be exploited to reduce competition effects in agroforestrysystems.

At the mm- to cm-scale, roots of competing plant species have been shownto segregate in the soil with respect to species (Caldwell et al., 1996), and thisprobably reduces interspecific (and increases intraspecific) interactions inzones of overlapping root systems. In the Harvard Forest, Lyford and Wilson(1964) observed that the roots of associated tree species seldom came into

11

direct contact and suggested that ‘there is some mechanism that keeps rootsfrom getting close to each other . . . presumably some rhizospheric effect’.Such a segregation may in part be explained with branching patterns incombination with unspecific reactions to a zone of nutrient and/or water deple-tion around individual roots which reduces the growth of other roots into thiszone (and possibly provokes compensatory root growth in another zone).However, phenomena of highly specific root recognition have been observedbetween cold-desert grasses, which depend on the involved species and evenon the populations from which the plants come. The actual mechanism ofthis recognition is unclear (Krannitz and Caldwell, 1995). It would be unsafeto speculate about the possible importance of such specific recognition phe-nomena in agroforestry systems, yet it might be worthwhile to investigate theiroccurrence also for plant species (and cultivars!) of potential value for agro-forestry.

The spatial separation of the roots of associated plants may reduce com-petition and negative rhizospheric interactions between plants (see below),but also the possibility of positive interactions. These may occur whenshallow-rooted crops take up water which has been hydraulically lifted intothe topsoil by deeper-reaching tree root systems (Emerman and Dawson,1996), or when one species takes up nutrients which have been made avail-able by rhizospheric processes of another species (Jones and Darrah, 1996).So, strategies of spatial segregation or approximation between associated rootsystems imply certain trade-offs for the plants, and the net outcome certainlydepends on specific conditions of the site and the species involved.

Factors of root activity and competitiveness

The description of root distribution and its changes in time give importanthints about zones and periods of possible root-root and root-soil interactions,but the prediction of the outcome of such interactions requires a more detailedunderstanding of the relationships between root structure and function. Onceroot systems of associated plants have come into contact, which parametersdecide about their relative competitiveness? Unfortunately, our knowledge inthis respect is still very incomplete.

Fine root length density is an important parameter for the acquisition ofwater and nutrients when the transport to the root, and not the uptake intothe root, is the limiting step in resource acquisition. It is therefore especiallyrelevant in dry soil and for nutrients of low to intermediate mobility such asP and K (Eissenstat, 1992). Accordingly, a comparison of different familiesof Pinus radiata seedlings revealed that those trees with high numbers offirst and second order laterals were particularly efficient in N and P uptakefrom the soil and showed fastest growth (Theodorou and Bowen, 1993). So,plants with high fine root length densities are likely to be more competitivethan plants with lower root length density. This, however, is only a firstapproximation, because fine root length density is not directly proportional

12

to root function. The uptake efficiency per unit root length for soil P can differby a factor of up to 10 between competing cold-desert grass and shrub species(Caldwell, 1994), and there is no reason to believe that such differences donot also exist between agroforestry species. Other parameters of root effi-ciency and competitiveness include root age, root diameter, presence of roothairs, physiological uptake characteristics, root exudates and root symbioses(Eissenstat, 1992; Goss et al., 1993; Caldwell, 1994; Grayston et al., 1996).

Mycorrhizae increase the effective root length density and thus the uptakeefficiency of a root system for immobile nutrients, especially P. This may beparticularly important for plants with relatively thick and poorly branchedroots and consequently low root length per unit carbon invested (Eissenstat,1992), and under soil conditions which restrict root growth more than thegrowth of mycorrhizae (Bowen, 1985). The increase depends on the per-centage infected root length and the length of the mycorrhizal hyphaeextending from the root surface into the soil. The effective root length densitymay increase by a factor of about 2.5 when 15% of the root length are infected(van Noordwijk et al., 1996). In addition, infection with arbuscular mycor-rhizae may markedly increase root branching (Atkinson and Last, 1994).Mycorrhizae may infect different plant species and may thus interconnectthe roots of associated plants, but the nutrient exchange between the plantsthrough these connections seems to be too small to affect their growth andnutrition significantly (Bethlenfalvay, 1992). Possible effects of mycorrhiza-tion on the competitive balance between associated plant species are illus-trated by an experiment, in which beans (Phaseolus vulgaris) inoculated witharbuscular mycorrhizae fixed more atmospheric N than uninoculated plants,but as the inoculation also increased their competitiveness for soil-N, the Nuptake of associated maize was reduced. This was in contrast to an expectedincrease of the N-transfer between the two species through mycorrhizal links(Reeves, 1992). The author suggested that fungi-host combinations shouldbe selected such that both plant species in an association perform well orthat the economically more important species is favoured by the mycorrhizalsymbiosis. This reflection is relevant for agroforestry associations of legu-minous trees with crops, in which the crops are normally the economicallymore valuable component. Other examples where mycorrhization affected thebalance between competing plant species are given by Newman (1978). InAmazonian rainforest, mycorrhizae have been shown to intervene in thecontact between roots and decomposing litter (Herrera et al., 1978), whereasin a shaded coffee plantation in Venezuela, the contact is made by root hairsbut not by mycorrhizae (Cuenca et al., 1983). Such structures presumablyincrease the competitiveness of root systems with respect to the uptake ofnutrients released during litter decomposition.

Like mycorrhizae, rhizospheric micro-organisms, which may include N-fixing and P-solubilising groups, certainly have the potential to affect the effi-ciency of root systems and thereby also the interactions between neighbouringroots of different species (Davet, 1996; Grayston et al., 1996). For example,

13

a strain of the rhizosphere bacterium Pseudomonas putida increased rootgrowth and P uptake of rapeseed (Brassica campestris) (Lifshitz et al., 1988),and the synergistic effect of an ectomycorrhiza (Pisolithus tinctorius) and arhizospheric bacterium (Bacillus sp.) of Pinus caribaea increased significantlythe P uptake from an organic soil P compound (phytate) (Chakly and Berthelin,1982). The complex interactions between mycorrhizae and the rhizospheric(or ‘mycorrhizospheric’) microflora have been reviewed by Linderman (1992).The influence of above- and belowground interactions between associatedplant species on the composition of their rhizospheric microflora has also beendemonstrated (Nair and Subba Rao, 1977; Newman, 1978). These includethe inhibition of the formation of root symbioses such as mycorrhizae(Robinson, 1972) and nodulation in legumes (Rice, 1968) by root exudatespresent in the rhizosphere of competing species. Skipper et al. (1996) reviewfirst results with the use of rhizospheric micro-organisms in biologicalweed control. However, a full discussion of this vast topic and its possibleimportance and management implications in agroforestry is beyond the scopeof this review.

A possibly very relevant component of root competitiveness is the abilityof a root system to utilise soil gaps (i.e. soil patches without roots) or to pro-liferate in fertile soil patches rapidly and take up nutrients there before theroots of other plants arrive. From two tussock grasses, the competitiveAgropyron desertorum invaded soil gaps created by removing a neighbouringplant more rapidly with its roots than the less competitive Pseudoroegneriaspicata (Eissenstat and Caldwell, 1989). In fertile soil patches created byinjecting nutrient solution, Agropyron roots showed increased proliferationwithin 24 hours, whereas Pseudoroegneria roots did not respond within twoweeks (Caldwell, 1994). Preferential root proliferation in humid micrositeshas also been observed (Smucker, 1993). In cropping systems, soil gaps arecreated at tillage, weeding or harvesting of annual crops. Fertile soil patchesare created by fertilisation, litterfall and death of plants in a stand.

The plasticity to respond to favourable microsites in the soil seems to behigher for roots of small diameter than for thicker roots (Eissenstat, 1992),and it would be interesting to compare different tree species, crops and foddergrasses in this respect as these typically differ in their root diameter. Whereassociated plants with intermingling root systems are known to differ in theirability to respond to fertile soil patches, it may be possible to use fertiliserdistribution to increase the share of the applied nutrients that either of thespecies receives: the species with the more flexible root system should profitmore from fertiliser which is applied in small, localised doses as comparedto uniform fertiliser distribution.

The ability of roots to respond to nutrient patches through increasedbranching and nutrient uptake depends on the carbon and nutrient status ofthe respective plant, and it may be possible to manipulate these with the aimof altering the competitiveness of associated plant species for nutrients witha heterogeneous distribution in the soil. N-deficient conifer seedlings exhib-

14

ited greater root proliferation in an N-rich soil patch than did N-sufficientplants (Friend et al., 1990), and for coconut trees (Cocos nucifera) it haseven been suggested to use the amount of branching in solutions with dif-ferent nutrient composition as a test for nutrient deficiencies (Wiersum, cit.in van Noordwijk et al., 1996). So, a tree which is expected to take up patchyfertiliser residues after the harvest of an annual crop before these are leachedmay do this more efficiently if it is moderately nutrient-deficient. On the otherhand, where shade trees are grown in association with tree crops and fertilisersare applied as placements near the individual crop plants, competition for thefertiliser may be less severe if the shade trees are well-supplied with therespective nutrients than if they are nutrient-deficient. Excessive shading ofthe crops should be avoided in this situation, because reduced light andconsequently carbon supply of plants has been found to reduce the ability oftheir root systems to respond to fertile soil microsites with increased nutrientuptake which is an energy-dependent process (Jackson and Caldwell, 1992).Shoot pruning of a plant may have similar effects to shading in this respectbecause it also reduces the energy available to the roots (Kandiah et al., 1984).If this is so, then shoot pruning may be used to fine-tune the relative abilityof trees and associated crops to respond to water and nutrient patches byinfluencing both the light climate of the crops and the energy available to thetree roots. This hypothesis certainly requires further study.

Patterns of root activity around individual plants

The overall form of a root system, the distribution of young fine roots withinthe root system and differences in mycorrhization and physiological proper-ties between these may be the most important factors which explain observedpatterns of root activity around individual plants. Knowing these patternsshould help to avoid excessive competition between associated species inagroforestry systems through suitable planting designs and to better targetnutrients through localised application of fertiliser into zones of high rootactivity.

In a comparative study of P uptake patterns around different tree crops(IAEA, 1975), the highest root activity was usually found close to the soilsurface, with the exceptions of orange (Citrus sinensis) in Spain (30 cm) andcoffee (Coffea arabica) in Kenya, where it moved to 45–75 cm soil depthduring the dry season. More variable results were obtained for the lateraldistribution of root activity: Banana (Musa paradisiaca), coffee and coconuttrees had the highest P uptake close to the trunk (at 30–100 cm distance),whereas citrus trees, oil palm (Elaeis guineensis, in Malaysia) and cacao(Theobroma cacao) showed a higher P-uptake from 100–300 cm distance thanfrom the soil closer to the plant. For orange trees, the highest activity movedaway from the plants with increasing tree age. Oil palm in the Côte d’Ivoireshowed no dependence of root activity from the tree distance between oneand four metres. It is to be expected that a tree species is particularly com-

15

petitive (with respect to the investigated nutrient) at a distance where its rootactivity is highest.

Differences in the rooting strategies between tree species which are poten-tially relevant for soil occupation and competition with neighbouring plantsinclude the ‘reiterative patterns’, or patterns of root branching, as defined byAtger and Edelin (1994). According to these authors, the ‘immediate reiter-ation’ (or immediate root branching) leads to a front of young feeder rootswhich is progressively pushed away from the tree base, leaving the zonearound the stem and between the coarse main roots increasingly unexploitedduring later growth stages. These spaces are then occupied by roots arisingfrom dormant or secondary meristems in the coarse roots (‘retarded reitera-tion’). The Mediterranean tree species Platanus hybrida occupies its rootingspace rather completely due to intensive reiteration of both types (Figure 2),

16

The dotted areas indicate the finest absorbing roots (from Atger and Edelin, 1994).

Figure 2. Root architecture of Platanus hybrida in southern France, illustrating the comple-mentarity of immediate (white) and retarded (black) reiteration in the colonisation of the soilby the root system.

whereas in other species (e.g. the tropical pioneer tree Cecropia obtusa) theoccupation of the soil near the stem is more restricted (Atger and Edelin,1994). For agroforestry, this difference may be relevant, as species with inten-sive retarded reiteration should make better use of their own litter, can beeasily supplied with fertiliser nutrients close to the stem, and should respondless to the removal of their peripheral roots, e.g. by tillage or root pruning(Korwar and Radder, 1994), than species which rely more heavily on theyounger, peripheral parts of their root systems.

Atkinson et al. (1978) demonstrated that the spatial distribution of tree rootactivity is also influenced by the spacing between trees and associated plants:When apple trees were grown in rows in 1.5 m wide herbicide strips betweenalleys with grass cover, 75% of the tree roots were in the herbicide strips andabsorbed most of their N there. When the soil was either uniformly treatedwith herbicide or covered with grass, the tree root distribution was more homo-geneous, and the N uptake from the soil in the tree row was less than in thefirst treatment. This indicates that when the aim is to confine tree root activityin a certain area around the trees with the help of a root-intensive groundcover, it may be most efficient to leave a certain space with reduced compe-tition near the trees, e.g. by spot weeding around the stem. If the ground coveralready starts close to the trees, more intermingling of the root systems, andconsequently more competition, may occur.

Managing root interactions by supplying limiting growth factors

When two species are known to compete for a certain growth factor, e.g. anutrient, it may be expected that competition can be reduced by raising thegeneral availability of this nutrient in the soil, e.g. by broadcast applicationof fertiliser or by selecting a more fertile site. Increasing the size of the avail-able pool of nutrients (or water) in the soil should mean that more plants,and consequently more roots, can be supplied from this pool without becominglimited by the respective growth factor. In practice, however, this expecta-tion is not always confirmed. Woods et al. (1992) measured additive effectsof the intensity of weed control and N-fertilisation on the growth and N-contents of Pinus radiata on a sandy soil where weeds and trees competedfor soil N, which means that the effect of weed control (and thus competitionfor N) was independent of the level of N fertilisation (Figure 3). Wilson (1988)reviewed twenty-three competition experiments, all involving agriculturalplant species, and concluded that increase and reduction of competition afteraddition of the resource for which competition occurred were equally frequent.So, there is a need for a more detailed understanding of the factors whichmake the addition of nutrients efficient or not in the alleviation of rootcompetition.

Competition between neighbouring roots occurs when the nutrient deple-tion zones around the roots, which are caused by the uptake of this nutrientinto the root, overlap. If roots compete for relatively immobile nutrients, such

17

as P and K, the addition of the respective nutrient increases its mobility inthe soil, and consequently the extension of the depletion zones around theroots (Marschner, 1995). This, in turn, increases the probability that thedepletion zones of neighbouring roots overlap. If the availability of the nutrienthas been raised to a level where it becomes non-limiting to plant growth, com-petition for this resource will be removed. If, however, the nutrient remainsgrowth-limiting, the increased mobility of the added nutrient can result inincreased (or unchanged) competition as compared to an equally fertilised,but competition-free control plant (Wilson and Newman, 1987; Wilson, 1988),even if the growth or nutrient status of the competitors is increased by thefertilisation.

If the growth of associated plants is limited by an immobile nutrient, suchas P, the application of this nutrient may also increase the probability of com-petition by bringing more mobile nutrients such as N into limiting supply,for which the depletion zones around neighbouring roots overlap more easily(Wilson and Newman, 1987). Furthermore, competition may increase whenthe addition of a growth factor stimulates root growth. In a silvopastoral exper-iment with wild cherry (Prunus avium) and ryegrass (Lolium perenne) inScotland, N fertilisation increased grass root growth at a time when soil water

18

After Woods et al., 1992.

Figure 3. Nitrogen concentrations in Pinus radiata needles at four levels of weed control(0 to 3 m width of the weed control strip around trees) and two levels of N fertilisation (0 and150 kg N ha–1 year–1) on a N-deficient soil in southern Australia. The interaction between weedcontrol and N-fertilisation was not significant.

was limiting, thereby increasing the potential for competition between thegrass and the trees (Campbell et al., 1994). Where N is not limiting rootgrowth, however, increased N availability may reduce rather than increase theroot mass of grasses (Kandeler et al., 1994) and trees (Hendricks et al., 1993),presumably by reducing root longevity (Fiala and Studeny, 1988; Hendrickset al., 1993), and this may result in reduced root competition in plant associ-ations. On the other hand, surface application of fertilisers and mulching mayresult in increased root concentration near the soil surface through increasedroot branching in fertile soil (see above), and this may result in increasedroot competition in this zone once the added nutrients have been taken up orbeen leached; thus, it is necessary to repeat such measures at appropriateintervals.

Interactions between root and shoot competition have been investigated inoldfield vegetation in the US, where the total level of competition betweenassociated plant species was independent of the fertility of a site, but the rela-tive contribution of root competition decreased with increasing fertility, andthus increasing shoot growth and mutual shading (Wilson and Tilman, 1991;Wilson and Tilman, 1993). In managed ecosystems such as agroforestry, it ismuch easier to control shading, e.g. through tree pruning, than to control rootinteractions, so that a shift from mainly belowground to mainly abovegroundcompetition may be a desirable consequence of fertilisation. Controlling theshade while increasing the pool size of nutrients in the soil through the appli-cation of nutrient-rich biomass is a working principle of both hedgerow inter-cropping and systems with perennial crops and pruned leguminous shade trees.

Temporal segregation of root activity of associated plants

Just as competing root systems may segregate in space at different levels,there may also be different degrees of the segregation of root activity in time.These may range from a complete separation, when one plant species growsafter the other species has died or been harvested, to differences in the timingof root activity between species growing in permanent association. Thefollowing example from a permanent plant association shows how the spatialand temporal components of root processes can interact (Veresoglou and Fitter,1984). In a grassland in Britain, competing grass species show temporal sep-aration of peaks of nutrient (P, K) uptake, even when the peaks of dry matterproduction coincide. Agrostis capillaris has the deepest root system (spatialsegregation!), and its consequently higher drought resistance allows thespecies to have its peak nutrient uptake later during the growing season, whenthe soil is normally drier, than the dominating Holcus lanatus. Poa pratensisis resistant to low temperature and can take up nutrients earlier in the seasonthan H. lanatus. So, the different morphological and physiological propertiesof the three associated grass species facilitate spatio-temporal resource sharingwithin this plant community. Similar processes may occur in well-designedand -managed agroforestry systems.

19

If root growth patterns are predictable, plant species can be combined inagroforestry associations which do not have high root growth rates at the sametime and not when soil resources are limiting. A relatively simple case is theassociation of perennial species with annual crops (including annual covercrops) in climates with a wet and a dry season, where the annuals completetheir life cycle when soil water is relatively abundant. In the seasonally dryclimate of South Australia, annual lupines (Lupinus angustifolius) can be usedto improve the growth of pines and substitute for mineral fertiliser, whereasa perennial ground cover would compete with the trees for water during thedry season (Nambiar and Nethercott, 1987). This principle only works if thesoil water reservoir is refilled by rains after the harvesting (or death) of theannual plant and before the onset of the dry season; otherwise the trees willsuffer during the dry season from the depletion of the soil water store (Duprazet al., 1998).

The problem becomes more complex when several perennial species aregrown in association. In the aforementioned silvopastoral experiment inScotland, the root growth of the grass (perennial ryegrass) peaked in May, andthe root growth of the trees (wild cherry) peaked in June. So, the probabilityof root competition between trees and grasses was reduced, especially as soilwater was not limiting during these months. In August, the soil was drier,and tree and grass root growth occurred simultaneously. During this time,herbicide application increased growth and leaf N content of the trees, indi-cating that there was competition for soil resources. As in temperate regions,nutrients taken up by trees in autumn are important for growth in the followingspring, the control of competition during this phase may be particularly impor-tant (Campbell et al., 1994).

In some situations it may be possible to influence the temporal pattern ofthe belowground activity of a plant species through the management of itsaerial parts. Shoot pruning can have dramatic effects on root growth dynamics.In an alley-cropping system with Gliricidia sepium in the Côte d’Ivoire,repeated shoot pruning of the trees during the cropping season was the likelyreason for an observed ‘anti-cyclic’ root development with maximum tree rootgrowth during the dry season when the trees were not pruned (Schroth andZech, 1995b). However, not all tree species may react to shoot pruning in sucha way. In fact, Gliricidia sepium is a relatively incompetitive tree species(Schroth and Lehmann, 1995; Schroth and Zech, 1995a), and althoughunpruned Gliricidia is also relatively tolerant to herbaceous understorey plants(Schroth et al., 1996), one reason for the low competitiveness of Gliricidiain agroforestry associations may be the pronounced reaction of its root systemto shoot pruning (another reason is probably the high quality of the Gliricidiamulch.) The root weight of G. sepium seedlings was reduced much more byincreased pruning frequency than that of Leucaena leucocephala seedlings(Ezenwa and Atta-Krah, 1992). Fownes and Anderson (1991) suggested thatthe allocation of a large biomass fraction to belowground organs and espe-cially to coarse storage roots makes tree species like L. leucocephala more

20

tolerant to shoot pruning. This, however, may only be desirable to a certainextent, because the larger this storage pool, the less the root systems may reactto shoot pruning, and the less the timing of tree root activity can be fine-tunedby the aboveground management of the plant. In an alley-cropping experi-ment in the Peruvian Amazon, shoot pruning of Inga edulis hedgerows led totemporary root death, but fine roots grew again already after 8–16 days(Fernandes et al., 1993). These authors suggested that with increasing treeage, the efficiency of shoot pruning in controlling belowground competitionmay decrease because the increasing carbohydrate and nutrient reserves in thestems allow rapid regrowth of roots and shoots.

In the temperate zone, where vegetation processes are slower and moreseasonal than in the humid tropics, it has been shown that at a longer timescale, the speed of shoot and certainly also of root regrowth can be influencedconsiderably through the timing of shoot pruning (Kays and Canham, 1991):Root starch reserves at the end of the growing season and sprout productionin the following growing season of several tree species in New York statewere highest when the trees were cut either at the very beginning or at thevery end of the growing season. Cutting well after the initiation of shootgrowth in the spring, but before the cessation of shoot growth in the summerresulted in minimal sprout growth, because the starch reserves had then beenused up for current season’s growth and had not yet been replenished. So,the timing of shoot pruning of trees is a suitable method of managing shootand root interactions, e.g. in systems with agricultural fields bordered bywoody hedgerows which are common in some temperate regions.

Root interactions in sequential agroforestry systems

The negative effects of root competition in agroforestry systems can beminimised by growing the components in rotations instead of associations inspace, e.g. in improved fallow systems, provided the relatively long tree phasewithout crop production can be justified in economic terms. Highly compet-itive tree root systems can be an advantage in this situation if they lead toefficient weed suppression during the fallow phase (Schroth et al., 1996).However, even when trees and crops or pastures are completely separated intime, there can still be significant interactions between their root systems. Theprocesses described below are not specific to fallow systems but may occuras well in simultaneous agroforestry.

Physical interactions between successive plant covers (e.g. a field cropfollowing fallow vegetation) may develop because the former vegetationinfluences the soil structure in a way which affects the root development ofthe plants which come after it. Plant roots preferentially grow through struc-tural weaknesses within the soil and can frequently be observed within existingmacropores, e.g. from older roots (van Noordwijk et al., 1991b; Schroth andZech, 1995b). So, if a root-intensive vegetation increases the macroporosityand reduces the bulk density of a soil, the root development of the following

21

vegetation may be facilitated. Miller (1986) cites a case where ‘Bahia grass(Paspalum notatum) penetrated soil layers that excluded cotton roots andcotton following the Bahia grass had increased rooting to at least 60 cmcompared to where the grass had not been grown’. Hulugalle and Lal (1986)found that maize growing after pigeon pea (Cajanus cajan) had more rootsthan maize growing after maize on an Alfisol in Nigeria. The rotation alsohad higher earthworm activity than the continuous maize, and it might beasked to which extent the improved root development of the maize in therotation had been caused by the pigeon pea roots and to which extent by theearthworms. Interactions between roots and soil fauna, which may both beinfluenced by agroforestry measures, as well as their relative importance inthe improvement of soil physical conditions and repercussions on plant growthcertainly merit study (Derouard et al., 1997; Springett and Gray, 1997). In anexperiment in the Côte d’Ivoire, root-intensive tree species were more effi-cient in reducing the bulk density of compacted soil horizons than sponta-neous fallow, although the effect on the root development of subsequent cropswas not investigated (Schroth et al., 1996).

The facilitation of root development of one species by a preceding vege-tation is partly a physical phenomenon, but certainly in part also a chemicalone. In acid, nutrient-poor tropical rainforest soils, roots can often be seen toproliferate in decomposing organic matter, including stems and dead roots.In an agroforestry association with Gliricidia sepium in the Côte d’Ivoire, fineroots were frequently found attached to the surface or even growing withindecomposing coarser roots (Schroth and Zech, 1995b). Without doubt, theroots which invade such ‘belowground litter’ take up nutrients that are releasedduring its decomposition. In addition, the organic matter within and aroundsuch old root channels may protect the young roots from toxic Al whengrowing through acid soil horizons (van Noordwijk et al., 1991b). There isalso evidence that the mineralisation of soil N, and thus the N availability tosubsequent crops, can increase with increasing root mass of a leguminousfallow vegetation, although the mechanisms of this phenomenon are not fullyunderstood (Schroth et al., 1995a).

However, chemical interactions between root systems may also be negative.Allelopathic effects of litter (including root) extracts and rhizospheric soil ofagroforestry trees on the root growth of crops and other tree species have oftenbeen detected under controlled conditions (Chou and Kuo, 1986; Hauser, 1993;Lisanework and Michelsen, 1993; Ramamoorthy and Paliwal, 1993), althoughtheir importance in the field, either in associations or in rotational agroforestrysystems, is still rather obscure as they are notoriously difficult to distinguishfrom other soil-root and root-root interactions. It could be argued that in thesoil, allelopathic substances may be rapidly deactivated by decomposition orsorption to soil particles so that their importance is likely to be small and ofshort duration. However, even if the main effect of allelopathic substances inthe roots of a fallow species were to prevent the roots of the subsequent cropfrom growing into its decomposing root mass, taking up the nutrients released

22

and profiting from the created root channels, this could already reduce theefficiency of the respective fallow species in the improvement of crop yieldson some soils.

Biological interactions between the root systems of plants growing inrotation may occur through changes in the abundance and species-composi-tion of mycorrhizal inocula as has been observed in agricultural rotations(Johnson and Pfleger, 1992) and through soil-borne diseases, especiallynematodes. Several woody species which are commonly used in agroforestryare hosts for plant nematodes (Meloidogyne spp., Pratylenchus spp.), such asCajanus cajan, Leucaena leucocephala, Sesbania grandiflora, Sesbaniasesban and Tephrosia vogelii (Page and Bridge, 1993; Mchowa and Ngugi,1994). How important this is for rotations (and associations) of these treeswith susceptible crop species is not yet clear. The fact that ‘there are no reportsin the literature that these nematodes are a constraint to the use of alley-cropping systems’ (Page and Bridge, 1993), or any other agroforestry tech-niques, is not sufficient proof that they are not important. Interactions ofagroforestry measures with plant pests and diseases are clearly a neglectedfield in agroforestry research. In fact, part of the phenomena at the tree-cropinterface which are commonly interpreted as root and shoot competition couldbe due to undetected pest and disease problems (Schroth et al., 1995b).Smucker (1993) suggested that the occupation of the same root channels bysuccessive susceptible plant species could lead to increased incidence of rootdiseases. This would not only apply to monocultures, but also to rotations ofspecies sharing the same root diseases, including some improved fallowsystems.

Application to some types of agroforestry systems

Associations of trees with annual crops

The root ecology of this type of agroforestry association has been treated insome detail in another paper (Schroth, 1995), and only additional remarks willbe made here.

When defining the desirable root characteristics of a tree for this systemtype, it is important to take the intended role of the tree into consideration.As for any type of agroforestry system, the net effect of the association oftrees with annual crops needs to be positive in an economic sense to be inter-esting for farmers. Even an incompetitive and well-managed (e.g. adequatelypruned) tree reduces the space that is available for the crops. So, for achievinga positive net effect of the association, the trees may either yield a commer-cial product (e.g. fodder, fruits, wood), or increase the crop yields in theremaining part of the field, at least in the medium term (e.g. through increasedsoil fertility), or in the ideal case they may do both.

Much agroforestry research has concentrated on associations of crops with

23

mostly leguminous tree species whose main or only function is to improvenutrient cycling and soil fertility. In such systems, the tree does not yield acommercial product, and competition between the trees and the crops forspace, light, soil resources or manpower (e.g. for planting and pruning thetrees) is only acceptable if it is compensated through clear advantages in termsof yield increases and/or yield security of the associated crops (e.g. throughefficient soil protection on slopes). Presumably, most farmers would prefertree species in this situation which are relatively incompetitive, abovegroundas belowground. A potential limitation to the usefulness of such ‘service trees’is that incompetitive species may not be as efficient as competitive speciesin capturing leached nutrients and improving the soil structure through rootactivity (Schroth, 1995). The questions ‘What makes a tree species com-petitive?’ and ‘What makes a tree efficient in soil improvement?’ are offundamental importance here.

More competition between trees and crops may be tolerated if the treesyield a commercial product, such as fruit trees, fodder trees or fast-growingtimber trees. A relevant finding in this context is that there seems to be nodirect relationship between the growth of a tree species and its competitive-ness, meaning that fast-growing trees are not necessarily intolerant to asso-ciated herbaceous plants (Jones and Sinclair, 1996; Schroth et al., 1996).

In agroforestry systems with annual crops, the crops have to develop theirroot system when those of the trees are already established. This should givethe trees at least temporarily an advantage in the competition for limitingsoil resources, e.g. water during drought phases at the onset of the rainyseason. On the other hand, associations of trees with annual crops offer thepossibility to influence root competition through soil tillage, in contrast toall following types of simultaneous agroforestry systems. In hedgerow inter-cropping with Leucaena leucocephala, zero-tillage led to yield depressions,presumably due to root competition (Ssekabembe, 1985), whereas 30 cm deepploughing along the hedgerow (‘root pruning’) increased soil water contentsunder the crops and crop yields (Korwar and Radder, 1994). This indicatesthat by destroying the superficial tree roots through soil tillage, the farmershifts the competitive balance between trees and crops in favour of the crops.The tree roots cut off during tillage enrich the soil organic matter pool andrelease nutrients into the soil which may be taken up by the crops. So, deepand frequent tillage probably has advantages in tree-crop associations whichneed to be balanced against possible increases of soil erosion and minerali-sation of soil organic matter (Schroth et al., 1995b).

Associations of trees with grasses or perennial fodder and cover crops

When trees are planted into a field with perennial fodder crops, the situationis to some extent the opposite of that described for the development of annualcrops in the presence of tree root systems. Initially, the herbaceous specieshave much deeper and denser root systems than the trees and may severely

24

compete with the developing tree seedlings for water and nutrients (Duprazet al., 1995; Dupraz et al., 1998). Especially grasses often have higher rootlength densities than dicotyledonous plants (Taylor, 1980; Bowen, 1985), andso they tend to be rather competitive, either as weeds or as crop plants (Yocum,1937; McDonald, 1986). Besides the direct competition for soil resources,chemical interactions (allelopathy) are suspected to contribute to the com-petitiveness of some grass species, such as Cynodon dactylon, Sorghum sudanense and S. halepense (McDonald, 1986). However, there are consid-erable differences in competitiveness between grass species and varieties: Ina high-rainfall area in Sri Lanka, the fodder grasses Brachiaria brizantha, B.decumbens and Panicum maximum cv. Guinea B were much more detrimentalto rubber trees (Hevea brasiliensis) with which they were associated thanPaspalum plicatulum, Setaria anceps and Panicum maximum cv. green panic.The forage production of the grasses was not clearly related to the growthdepression they caused in the rubber trees, indicating that competitiveness isnot a simple function of growth (Dissanayake and Waidyanatha, 1987). Thisparallels similar findings for trees (see above).

In some situations, the competitiveness of grasses may be used to suppressother, still more problematic weeds: In dry areas such as California and south-west Oregon, grasses need to be controlled in conifer plantations on shallowsoils (e.g. < 0.9 m deep) where their roots occupy the same soil as the treeroots, but on deeper soils, they reduce the growth of shrubs which, after sometime, would become more severe competitors for the trees because they aredeeper-rooted than the grasses and would compete with the trees even in thesubsoil. Controlled grazing is a way to favour grasses which prevent theestablishment of shrubs and reduce the growth and seeding of established ones(McDonald, 1986). In tree crop plantations in the tropics, herbaceous legu-minous species such as Pueraria phaseoloides are often planted as groundcover, and this is partly to reduce the competition from grasses and aggres-sive woody plants.

The desirable root characteristics of the trees are probably different inassociations with either grasses or leguminous cover crops. If tree crops aregrown together with pasture or fodder grasses, the trees should respond tothe competition from the grass roots through downward displacement of theirroot systems, thereby minimising competition and increasing the overall useof the soil resources. In addition, temporal differences between the rootactivity of trees and grasses would be desirable.

In associations of trees with leguminous cover crops, on the other hand,an intensive exploitation by the tree roots of the soil directly beneath theground cover is desirable, because the trees are expected to profit from soilimprovements from the cover crop and especially to take up biologically fixednitrogen which is released from the shoot and root litter of the cover crop atits decomposition and through root exudates. Direct nutrient transfer fromthe cover crop to the trees via mycorrhizal links seems to be of less impor-tance (Ikram et al., 1994). Competition between trees and cover crops for

25

nitrogen, in turn, may stimulate the N-fixation activity of the legumes. So,tree root systems which are relatively aggressive in relation to those of thecover crop and which do not respond to their presence through downwarddisplacement or reduced lateral extension are desirable. The management ofthe tree root systems should favour their lateral growth and intensive exploita-tion of the topsoil horizons, e.g. through the distribution of fertiliser over asufficiently large area around the trees so that lateral root growth is not dis-couraged by steep gradients of nutrient availability with increasing distancefrom the stem.

Associations of different tree or shrub species

The most intensively studied agroforestry systems of this type are planta-tions of perennial crops, especially coffee, cacao and tea, with shade trees.Other examples include the highly diversified tropical homegardens or simplerassociations of different tree crops, such as cacao with coconut palms (Nairand Subba Rao, 1977). A major difference between these systems and the onesdiscussed before is that plantations even of shade-demanding perennial cropsmay need the soil protecting and improving effects of associated ‘service’trees much less than fields with annual crops because of the presence of theperennial root systems of the tree crops themselves. So, an aim of systemsdesign and management may be to avoid root interactions between the asso-ciated tree species as much as possible.

One way to control root competition between shade trees and perennialcrops is spacing. For example, in the climate of Malawi in East Africa with6 months of dry season, coffee suffers from competition of the shade treesAlbizia lebbek and Grevillea robusta for soil water. As the coffee needs someclimatic protection under these relatively dry conditions, it has been recom-mended to plant the trees in shelterbelts and not regularly spaced betweenthe coffee bushes (Foster and Wood, 1963). Also, trees should possess arestricted lateral root development in this situation. In the same region, theroot systems of Albizia gummifera, Gliricidia sepium and Grevillea robustawere compared with respect to potential competition problems when plantedas shade trees in tea (Camellia sinensis) plantations (Laycock and Wood,1963). Albizia gummifera had a prolific, laterally spreading root system whichcertainly intermingled with the roots of the tea, whereas Grevillea robusta hadmost of its roots directly beneath the bole, with few superficial roots spreadinghorizontally. Gliricidia sepium was intermediate. Albizia and Gliricidiareduced tea yields in comparison to Grevillea or no shade, but the authorsconcluded that this was due to aboveground rather than belowground effects.Grevillea has remained a very popular agroforestry species, certainly in partbecause of its incompetitive root system.

Shelterbelts around plots of perennial crops as an alternative or supplementto overhead shade are especially interesting when the protection againstmechanical damage by wind is an important component of the benefit of the

26

shade trees for the crops. This seems to be the case for cacao (Alvim andAlvim, 1980). In Sabah, box-shaped plots have been developed in whichcacao, once established, is only protected laterally by the trees which areplanted on the boundary of plots of variable size (Lim, 1980). In the citedexperiment, root competition for water and nutrients did not seem to be impor-tant, increased radiation was apparently the main reason for cacao yieldincreases in box-shaped plots as compared with plots with overhead shade,but the principle can equally be applied in situations where root competitionis more relevant. Tree species used in the boundary plantings around the cacaoplots included leguminous species such as Gliricidia sepium, but also thetimber trees Grevillea robusta and Swietenia spp. (Lim, 1980).

Associations of perennial crops may present special management problemswhich are related to the intermingling of the root systems. In the top 15 cmof mixed rubber-cacao plantations in Bahia/Brazil, the fine root mass of therubber was twice as high as that of the cacao directly under a cacao tree whichgrew at 2–3 m distance from a rubber tree (Kummerow and Lages Ribeiro,1982). The authors concluded that it would be difficult under these conditionsto fertilise the cacao without loosing much of the fertiliser to the rubber trees.In associations where one root system is displaced into the subsoil by anotherspecies, fertilisation of the species with the displaced root system withimmobile nutrients could become virtually impossible.

Improved fallows

The root systems of a natural or improved fallow vegetation certainly con-tribute considerably to the regeneration of soil conditions during a fallowphase, and their relative importance is further increased when a part of theaboveground biomass is removed from the plot for use as fodder, firewoodetc. (Schroth et al., 1995a). The more the land comes under pressure from anincreasing population, the more intensively fallows will be used, and thecritical question is until which intensity this use is compatible with thefunction of the fallow in soil regeneration. The degradation, instead ofregeneration, of the soil under fallow when population pressure causes over-grazing of the fallow vegetation has been described, for example, from theSudan savannah of the northern Côte d’Ivoire. Protection from grazingincreased the aboveground biomass, but even more so the root mass of thefallow vegetation (César and Coulibaly, 1991).

The tree species which are typically recommended for planted tree fallows,such as Sesbania sesban (Kwesiga and Coe, 1994), Gliricidia sepium(Adesina, 1991) and Leucaena leucocephala (MacDicken, 1991), are alsopotential fodder trees, and the possibility of fodder harvesting during thefallow phase may be an important incentive for farmers to adopt fallowimprovement techniques. However, the intensity of fodder harvesting mayhave to be strictly controlled to maintain the soil-improving role of the fallow.The aforementioned effects of aboveground cutting on root mass and distri-

27

bution are particularly relevant in this respect, as weakened and superficialtree root systems caused by frequent aboveground pruning, or intensivegrazing, would be inefficient in the recycling of nutrients from subsoilhorizons and in their physical improvement.

Conclusions and outlook

Plant species have persisted and developed during evolution because withintheir respective niche, they competed successfully for space, light and soilresources with the species associated with them. So, we should not expectplants, neither herbaceous nor trees, to be altruistic. Nevertheless, it has beenpointed out in this review that the root systems of associated plants interactin many ways which range from severe competition through complementarityto facilitation. Even where competition occurs, this is not necessarily negativefor the system as a whole, as total nutrient and water use may be increased.Defining and obtaining the right amount of competition within tree-cropassociations is a very difficult, but necessary task for agroforestry design andmanagement.

It has also been pointed out that various types of root interactions existand can be of major importance in rotational agroforestry systems. These arenot principally different from interactions in spatial associations, but maybecome more evident when the interacting species are separated in time. Asin spatial agroforestry, these interactions need to be predicted and optimisedthrough appropriate species selection and management.

The majority of management measures in agroforestry systems probablyhave some influence on the form and functioning of the root systems of theplant species present, and consequently on their interactions. In many cases,we can predict from the available information in which manner and directionthe development, distribution and activity of roots in agroforestry systems willchange as a response to a well-defined management measure. Only in a fewcases, however, we can predict these changes quantitatively. It is unlikely thatthis situation will change much in the near future, because interacting rootsystems are influenced by too many (often still unknown) factors, and quan-titative root studies in the field are notoriously difficult and cumbersome, espe-cially when they include a time component. However, in the short term, thequalitative information summarised in this review may already be of helpwhen deciding about plant species and management plans for agroforestrysystems. In the medium term, it is important that root researchers concernedwith agroforestry concentrate on those root processes which are likely to beof practical importance. Thus, we must identify and agree first what is prac-tical in this field. There is a need to test the effectiveness of root manage-ment measures in different plant species and ecological situations, with theaim of identifying those measures which work and which should be refinedthrough further research and be applied in practice. Once this has been

28

achieved, the response to some effective, standardised root managementmeasures should be accepted as part of tree and crop improvement pro-grammes for agroforestry.

Acknowledgements

Some of the ideas presented in this review have ripened during inspiringdiscussions with John Beer and Fergus Sinclair. Petra Marschner, ChristianDupraz and two anonymous reviewers improved the manuscript greatly withtheir critiques. Bernard Mallet corrected too simplistic ideas about the rootecology of some rainforest tree species. The University of Hamburg providedpartial funding for travel to present this paper.

References

Adesina FA (1991) Soil management with cultivated fallows in humid and subhumid Africa.Can J Soil Sci 71: 147–154

Alvim PT and Alvim R (1980) Environmental requirements of cocoa with emphasis on responsesto shade and moisture stress. In: Proceedings of the International Conference on Cocoa andCoconuts, Kuala Lumpur, 1978, pp 93–111. The Incorporated Society of Planters, KualaLumpur

Atger C and Edelin C (1994) Stratégies d’occupation du milieu souterrain par les systèmesracinaires des arbres. Rev Ecol (Terre Vie) 49: 343–356

Atkinson D, Naylor D and Coldrick GA (1976) The effect of tree spacing on the apple rootsystem. Hortic Res 16: 89–105

Atkinson D, Johnson MG, Mattam D and Mercer ER (1978) The effect of orchard soil man-agement on the uptake of nitrogen by established apple trees. J Sci Food Agric 30: 129–135

Atkinson D and Last F (1994) Growth, form and function of roots and root systems. ScottishForestry 48: 153–159

Bethlenfalvay GJ (1992) Mycorrhizae and crop productivity. In: Bethlenfalvay GJ and LindermanRG (eds) Mycorrhizae in sustainable agriculture, pp 1–27. American Society of Agronomy,Madison

Bowen GD (1984) Tree roots and the use of soil nutrients. In: Bowen GD and Nambiar EKS(eds) Nutrition of plantation forests, pp 147–179. Academic Press, London

Bowen GD (1985) Roots as a component of tree productivity. In: Cannell MGR and JacksonJE (eds) Attributes of trees as crop plants, pp 303–315. Institute of Terrestrial Ecology,Huntingdon

Buwalda JG (1993) The carbon costs of root systems of perennial fruit crops. Environ Exp Bot33: 131–140

Caldwell MM (1994) Exploiting nutrients in fertile soil microsites. In: Caldwell MM and PearcyRW (eds) Exploitation of environmental heterogeneity by plants, pp 325–347. AcademicPress, San Diego

Caldwell MM, Manwaring JH and Durham SL (1996) Species interactions at the level of fineroots in the field: influence of soil nutrient heterogeneity and plant size. Oecologia 106:440–447

Campbell CD, Mackie-Dawson LA, Reid EJ, Pratt S M, Duff EI and Buckland ST (1994) Manualrecording of minirhizotron data and its application to study the effect of herbicide andnitrogen fertiliser on tree and pasture root growth in a silvopastoral system. Agrofor Syst26: 75–87

29

César J and Coulibaly Z (1991) Le rôle des jachères et des cultures fourragères dans le maintiende la fertilité des terres. In: Savanes d’Afrique – terres fertiles? Actes des RencontresInternationales, Montpellier, 10–14 Dec. 1990, pp 271–287. Ministère de la Coopération etdu Développement, Paris

Chakly M and Berthelin J (1982) Rôle d’une ectomycorhize ‘Pisolithus tinctorius – Pinuscaribea’ et d’une bactérie rhizosphérique sur la mobilisation du phosphore de phosphatesminéraux et organiques insolubles. In: Gianinazzi SE (ed) Les Mycorhizes, partie intégrantede la plante: biologie et perspectives d’utilisation, pp 215–220. Institut National de laRecherche Agronomique, Paris

Chauvel A, Vital ART, Lucas Y, Desjardins T, Franken WK, Luizao FJ, Araguas LA, RozanskiK and Bedmar AP (1993) O papel das raizes no ciclo hidrologico da floresta Amazonica.Congresso Brasileiro de Meteorologia, 28 Sept–2 Oct 1992, vol. 1, pp 298–302. Sao Paulo

Chou C-H and Kuo Y-L (1986) Allelopathic research of subtropical vegetation in Taiwan III.Allelopathic exclusion of understory by Leucaena leucocephala (Lam.) de Wit. J ChemEcol 12: 1431–1442

Cuenca G, Aranguren J and Herrera R (1983) Root growth and litter decomposition in a coffeeplantation under shade trees. Plant Soil 71: 477–486

D’Antonio CM and Mahall BE (1991) Root profiles and competition between the invasive, exoticperennial, Carpobrotus edulis, and two native shrub species in California coastal scrub. AmJ Bot 78: 885–894

Davet P (1996) Vie microbienne du sol et production végétale. Paris, Institut National de laRecherche Agronomique, 383 pp

Derouard L, Tondoh J, Vilcosqui L and Lavelle P (1997) Effects of earthworm introduction onsoil processes and plant growth. Soil Biol Biochem 29: 541–545

Dissanayake SN and Waidyanatha UPD (1987) The performance of some tropical forage grassesinterplanted with young Hevea trees and their effect on growth of the rubber. Trop Agric(Trinidad) 64: 119–121

Dupraz C, Dauzat M, Suard B, Girardin N and Olivier A (1995) Root extension of young wide-spaced Prunus avium trees in an agroforest as deduced from the water budget. In: EhrenreichJH, Ehrenreich DL and Lee HW (eds) Growing a Sustainable Future, Proceedings of theFourth North American Agroforestry Conference, July 23–28, 1995, pp 46–50. Boise, Idaho

Dupraz C, Simorte V, Dauzat M, Bertoni G, Bernadac A and Masson P (1998) Growth andnitrogen status of young walnuts as affected by intercropped legumes in a Mediterraneanclimate. Agrofor Syst this issue

Eastham J, Rose CW, Cameron DM, Rance SJ, Talsma T and Charles Edwards DA (1990)Tree/pasture interactions at a range of tree densities in an agroforestry experiment. II. Wateruptake in relation to rooting patterns. Aust J agric Res 41: 697–707

Eissenstat D M (1992) Costs and benefits of constructing roots of small diameter. J Plant Nutr15: 763-782

Eissenstat DM and Caldwell MM (1989) Invasive root growth into disturbed soil of two tussockgrasses that differ in competitive effectiveness. Funct Ecol 3: 345–353

Emerman SH and Dawson TE (1996) Hydraulic lift and its influence on the water content ofthe rhizosphere: an example from sugar maple, Acer saccharum. Oecologia 108: 273–278

Ewel JJ (1986) Designing agricultural ecosystems for the humid tropics. Ann Rev Ecol Syst17: 245-271

Ezenwa IV and Atta-Krah AN (1992) Early growth and nodulation in Leucaena and Gliricidiaand the effects of pruning on biomass productivity. In: Mulongoy K, Gueye M and SpencerDSC (eds) Biological nitrogen fixation and sustainability of tropical agriculture, pp 171–178.International Institute of Tropical Agriculture, Ibadan

Fernandes ECM, Davey CB and Nelson LA (1993) Alley cropping on an acid soil in the upperAmazon: mulch, fertilizer, and hedgerow root pruning effects. In: Technologies for sustain-able agriculture in the tropics. ASA Special Publication 56, pp 77–96. American Society ofAgronomy, Madison

30

Fiala K and Studeny V (1988) Cutting and fertilization effect on the root system in severalgrassland stands II. Vertical distribution of root biomass and changes in the carbohydratecontent. Ekológia (CSSR) 7: 27–42

Foster LJ and Wood RA (1963) Observations on the effects of shade and irrigation on soil-moisture utilization under coffee in Nyasaland. Emp J Exp Agric 31: 108–115

Fownes JH and Anderson DG (1991) Changes in nodule and root biomass of Sesbania sesbanand Leucaena leucocephala following coppicing. Plant Soil 138: 9–16

Friend AL, Eide MR and Hinckley TM (1990) Nitrogen stress alters root proliferation in Douglas-fir seedlings. Can J For Res 20: 1524–1529

Goss MJ, Miller MH, Bailey LD and Grant CA (1993) Root growth and distribution in relationto nutrient availability and uptake. Eur J Agron 2: 57–67

Grayston SJ, Vaughan D and Jones D (1996) Rhizosphere carbon flow in trees, in comparisonwith annual plants: the importance of root exudation and its impact on microbial activityand nutrient availability. Appl Soil Ecol 5: 29–56

Hauser S (1993) Effect of Acioa barteri, Cassia siamea, Flemingia macrophylla and Gmelinaarborea leaves on germination and early development of maize and cassava. Agric EcosysEnviron 45: 263–273

Hendricks JJ, Nadelhoffer KJ and Aber JD (1993) Assessing the role of fine roots in carbonand nutrient cycling. Trends Ecol Evol 8: 174–178

Herrera R, Mérida T, Stark N and Jordan CF (1978) Direct phosphorus transfer from leaf litterto roots. Naturwissenschaften 65: 208–209

Hulugalle NR and Lal R (1986) Root growth of maize in a compacted gravelly tropical alfisolas affected by rotation with a woody perennial. Field Crops Res 13: 33–44

IAEA (1975) Root activity patterns of some tree crops. Technical Report Series 170. Vienna,International Atomic Energy Agency, 154 pp

Ikram A, Jensen ES and Jakobsen I (1994) No significant transfer of N and P from Puerariaphaseoloides to Hevea brasiliensis via hyphal links of arbuscular mycorrhiza. Soil BiolBiochem 26: 1541–1547

Jackson RB and Caldwell MM (1992) Shading and the capture of localized soil nutrients: nutrientcontents, carbohydrates, and root uptake kinetics of a perennial tussock grass. Oecologia91: 457–462

Johnson NC and Pfleger FL (1992) Vesicular-arbuscular mycorrhizae and cultural stresses. In:Bethlenfalvay GJ and Linderman RG (eds) Mycorrhizae in sustainable agriculture, pp 71–99.American Society of Agronomy, Madison

Jones DL and Darrah PR (1996) Role of root exudates in plant nutrient acquisition in low inputagricultural systems. Agroforestry Forum 7: 7–9

Jones M and Sinclair FL (1996) Differences in root system responses of two semi-arid treespecies to crown pruning. Agroforestry Forum 7: 24–27

Kandeler E, Eder G and Sobotik M (1994) Microbial biomass, N mineralization, and theactivities of various enzymes in relation to nitrate leaching and root distribution in aslurry-amended grassland. Biol Fertil Soils 18: 7–12

Kandiah S, Wettasinghe DT and Wadasinghe G (1984) Root influence on shoot developmentin tea (Camellia sinensis (L.) O. Kuntze) following shoot pruning. J Hortic Sci 59: 581–587

Kays JS and Canham CD (1991) Effects of time and frequency of cutting on hardwood rootreserves and sprout growth. For Sci 37: 524–539

Knoop WT and Walker BH (1985) Interactions of woody and herbaceous vegetation in a southernAfrican savanna. J Ecol 73: 235–253

Korwar GR and Radder GD (1994) Influence of root pruning and cutting interval of Leucaenahedgerows on performance of alley cropped rabi sorghum. Agrofor Syst 25: 95–109

Krannitz PG and Caldwell MM (1995) Root growth responses of three Great Basin perennialsto intra- and interspecific contact with other roots. Flora 190: 161–167

Kummerow J and Lages Ribeiro S (1982) Fine roots in mixed plantations of Hevea (Heveabrasiliensis H.B.K.Müll.Arg.) and cacao (Theobroma cacao L.). Rev Theobr 12: 101–105

31

Kwesiga F and Coe R (1994) The effect of short rotation Sesbania sesban planted fallows onmaize yield. For Ecol Manage 64: 199–208

Laycock DH and Wood RA (1963) Some observations on soil moisture use under tea inNyasaland. Trop Agric (Trinidad) 40: 35–48

Le Roux X, Bariac T and Mariotti A (1995) Spatial partitioning of the soil water resourcebetween grass and shrub components in a West African humid savanna. Oecologia 104:147–155

Le Roux X and Bariac T (1998) Seasonal variations in soil, grass and shrub water status in aWest African humid savannah. Oecologia 113: 456–466

Lifshitz R, Guilmette H and Kozlowski M (1988) Tn5-mediated cloning of a genetic region fromPseudomonas putida involved in the stimulation of plant root elongation. Appl EnvironMicrobiol 54: 3169–3172

Lim DHK (1980) New developments in shade for hybrid cocoa in Sabah. In: Proceedings ofthe International Conference on Cocoa and Coconuts, Kuala Lumpur, 1978, pp 122–142. TheIncorporated Society of Planters, Kuala Lumpur

Linderman RG (1992) Vesicular-arbuscular mycorrhizae and soil microbial interactions. In:Bethlenfalvay GJ and Linderman RG (eds) Mycorrhizae in sustainable agriculture, pp 45–70.American Society of Agronomy, Macison

Lisanework N and Michelsen A (1993) Allelopathy in agroforestry systems: the effects of leafextracts of Cupressus lusitanica and three Eucalyptus spp. on four Ethiopian crops. AgroforSyst 21: 63–74

Lyford WH and Wilson BF (1964) Development of the root system of Acer rubrum L. HarvardForestry Paper 10: 1–17

MacDicken KG (1991) Impacts of Leucaena leucocephala as a fallow improvement crop inshifting cultivation on the island of Mindoro, Philippines. For Ecol Manage 45: 185–192

Marschner H (1995) Mineral nutrition of higher plants, 2nd edition. London, Academic Press,889 pp

McDonald PM (1986) Grasses in young conifer plantations – hindrance and help. NorthwestSci 60: 271–278

Mchowa JW and Ngugi DN (1994) Pest complex in agroforestry systems: the Malawi experi-ence. For Ecol Manage 64: 277–284

Miller DE (1986) Root systems in relation to stress tolerance. HortScience 21: 963–970 Nair SK and Subba Rao NS (1977) Microbiology of the root region of coconut and cacao under

mixed cropping. Plant Soil 46: 511–519 Nambiar EKS and Nethercott KH (1987) Nutrient and water availability to and growth of young

radiata pine plantations intercropped with lupins. New Forests 1: 117–134 Nepstad DC, de Carvalho CR, Davidson EA, Jipp PH, Lefebvre PA, Negreiros GH, da Silva

ED, Stone TA, Trumbore SE and Viera S (1994) The role of deep roots in the hydrologicaland carbon cycles of Amazonian forests and pastures. Nature 372: 666–669

Newman EI (1978) Root microorganisms: their significance in the ecosystem. Biol Rev 53:511–554

Page SLJ and Bridge J (1993) Plant nematodes and sustainability in tropical agriculture. ExplAgric 29: 139–154

Ramamoorthy M and Paliwal K (1993) Allelopathic compounds in leaves of Gliricidia sepium(Jacq.) Kunth ex Walp. and its effect on Sorghum vulgare L. J Chem Ecol 19: 1691–1701

Reeves M (1992) The role of VAM fungi in nitrogen dynamics in maize-bean intercrops. PlantSoil 144: 85–92

Rice EL (1968) Inhibition of nodulation of inoculated legumes by pioneer plant species fromabandoned fields. Bull Torrey bot Club 95: 346–358

Rivera E, Silva S and Vicente-Chandler J (1983) Distribution of pigeon peas, cassava, coffeeand grass roots in an ultisol. J Agric Univ Puerto Rico 67: 278–285

Robinson R K (1972) The production by roots of Calluna vulgaris of a factor inhibitory togrowth of some mycorrhizal fungi. J Ecol 60: 219–224

32

Sanford RL, Jr. and Cuevas E (1996) Root growth and rhizosphere interactions in tropical forests.In: Mulkey SS, Chazdon RL and Smith AP (eds) Tropical Forest Plant Ecophysiology, pp268–300. Chapman and Hall, New York

Schroth G (1995) Tree root characteristics as criteria for species selection and systems designin agroforestry. In: Sinclair FL (ed) Agroforestry: Science, Policy and Practice, pp 125–143.Kluwer Academic Publishers, Dordrecht

Schroth G, Kolbe D, Balle P and Zech W (1995a) Searching for criteria for the selection ofefficient tree species for fallow improvement, with special reference to carbon and nitrogen.Fertilizer Res 42: 297–314

Schroth G, Poidy N, Morshäuser T and Zech W (1995b) Effects of different methods of soiltillage and biomass application on crop yields and soil properties in agroforestry with hightree competition. Agric Ecosys Environ 52: 129–140

Schroth G, Kolbe D, Balle P and Zech W (1996) Root system characteristics with agroforestryrelevance of nine leguminous tree species and a spontaneous fallow in a semi-deciduousrainforest area of West Africa. For Ecol Manage 84: 199–208

Schroth G and Lehmann J (1995) Contrasting effects of roots and mulch from three agroforestrytree species on yields of alley cropped maize. Agric Ecosys Environ 54: 89–101

Schroth G and Zech W (1995a) Root length dynamics in agroforestry with Gliricidia sepium ascompared to sole cropping in the semi-deciduous rainforest zone of West Africa. Plant Soil170: 297–306

Schroth G and Zech W (1995b) Above- and below-ground biomass dynamics in a sole croppingand an alley cropping system with Gliricidia sepium in the semi-deciduous rainforest zoneof West Africa. Agrofor Syst 31: 181–198

Sinclair FL, Verinumbe I and Hall JB (1994) The role of tree domestication in agroforestry. In:Leakey RRB and Newton A (eds) Tropical trees: potential for domestication and therebuilding of forest resources, pp 124–136. HMSO, London

Skipper HD, Ogg AG, Jr. and Kennedy AC (1996) Root biology of grasses and ecology ofrhizobacteria for biological control. Weed Techn 10: 610–620

Smucker AJM (1993) Soil environmental modifications of root dynamics and measurement. AnnRev Phytopathol 31: 191–216

Springett J and Gray R (1997) The interaction between plant roots and earthworm burrows inpasture. Soil Biol Biochem 29: 621–625

Ssekabembe CK (1985) Perspectives on hedgerow intercropping. Agrofor Syst 3: 339–356 Taylor HM (1980) Modifying root systems of cotton and soybean to increase water absorption.

In: Turner NC (ed) Adaptations of plants to water and high temperature stress, pp 75–84.John Wiley and Sons, New York

Theodorou C and Bowen GD (1993) Root morphology, growth and uptake of phosphorus andnitrogen of Pinus radiata families in different soils. For Ecol Manage 56: 43–56

van Noordwijk M, Hairiah K, Ms S and Flach EN (1991a) Peltophorum pterocarpa (DC) Back(Caesalpiniaceae), a tree with a root distribution suitable for alley cropping on acid soils inthe humid tropics. In: McMichael BL and Persson H (eds) Plant roots and their environ-ment, pp 526–532. Elsevier, Amsterdam

van Noordwijk M, Widianto, Heinen M and Hairiah K (1991b) Old tree root channels in acidsoils in the humid tropics: Important for crop root penetration, water infiltration and nitrogenmanagement. Plant Soil 134: 37–44

van Noordwijk M, Lawson G, Soumaré A, Groot JJR and Hairiah K (1996) Root distributionof trees and crops: competition and/or complementarity. In: Ong CK and Huxley P (eds)Tree-Crop Interactions, pp 319–364. CAB International, Wallingford

Veresoglou DS and Fitter AH (1984) Spatial and temporal patterns of growth and nutrient uptakeof five co-existing grasses. J Ecol 72: 259–272

Wilson JB (1988) Shoot competition and root competition. J Appl Ecol 25: 279–296 Wilson JB and Newman EI (1987) Competition between upland grasses: root and shoot

competition between Deschampsia flexuosa and Festuca ovina. Acta Oecol Oecol Gener 8:501–509

33

Wilson SD and Tilman D (1991) Components of plant competition along an experimentalgradient of nitrogen availability. Ecology 72: 1050–1065

Wilson SD and Tilman D (1993) Plant competition and resource availability in response to dis-turbance and fertilization. Ecology 74: 599–611

Woods PV, Nambiar EKS and Smethurst PJ (1992) Effect of annual weeds on water and nitrogenavailability to Pinus radiata trees in a young plantation. For Ecol Manage 48: 145–163

Yocum W W (1937) Root development of young delicious apple trees as affected by soils andby cultural treatments. Univ Nebraska Agric Exp Stat Res Bull 95: 1–55

34