2 The insect-pest situation in agroforestry

The science of agroforestry is of recent origin, although the practice is age old. There are numerous types of agroforestry system in different parts of the world. There have been few studies of insect pests in agroforestry context, although insect pests of crops that are components of agroforestry systems have been studied. Scanty information is available about the insects associated with the multipurpose trees and shrubs that are gaining greater economic importance as components of agroforestry systems. In any particular location, the insect fauna occurring on certain plant species is more or less the same, whether that plant species is in monoculture or in a polycultural assemblage such as an agroforestry system. However, the activities of these insects are not likely to be identical in any two situations. There are several factors that influence the activities of insects in agroforestry.

The mechanisms governing the insect-pest situation in agroforestry systems are yet to be investigated fully, and there have been no studies comparing the insect-pest situation in monoculture or block plantation and in agroforestry combinations. Most publications that mention pests of agroforestry either contain information on the insect pests of one component (Verma 1986, 1988; Sagwal 1987; Sen-Sharma 1987; Singh and Singh 1987; Khan et al. 1988) or underline the need for research in this field (Epila 1986; Huxley and Greenland 1989). The management of insect pests through agroforestry field design has been the subject of discussion in a few publications (Altieri et al. 1987; Epila 1988; Gold et al. 1989). Liping (1991) suggested directions for research on biological control of pests and diseases in agroforestry systems.

There have been several studies on insect activities in windbreaks, shelterbelts and hedges under temperate conditions, but these were not in an agroforestry context as most of them were restricted to woody plants. Insect dynamics in windbreaks have been investigated by Lewis (1965a, b, 1966a, b), Lewis and Stephenson (1966), Lewis (1967, 1970) Lewis and Dibley (1970), Smith and Lewis (1972), Solomon (1981), Dix and Leatherman (1988), Norton (1988) and Pasek (1988) among others. The activities of insects in shelterbelt plantations have also been subject of study (Galecka and Zeleny 1969; Gorny 1970; Kyrilenko and Pysariev 1976; Slosser and Boring 1980), and there have been a few studies on hedges (Lewis 1969b, c; Hawkes 1973; Bowden and Dean 1977).

The insect pests of an agroforestry system are essentially the pests of its components: the crops and woody perennials. The dynamics of insect pests and their natural enemies are be governed by the complexity and composition of the agroforestry system. The pest situation in these systems will be influenced by the degree of interaction between the components, the type of agroforestry system and the composition of the plant communities in each component.

Interactions among the components of the agroforestry system can be either positive, negative or neutral. They are regarded as negative when pest problems are increased in an agroforestry system when compared with a monoculture block plantation. A reduction in pest activity under agroforestry indicates a positive interaction, while no change in pest intensities between monoculture and agroforestry denotes a neutral interaction from the insect-pest management point of view. Many factors govern insect-pest intensity in agroforestry and each factor may have a different effect on pests at different times and under different situations. The net outcome will be the sum of favourable and unfavourable effects on pests and their natural enemies. Some of the factors that govern the pest situation in agroforestry are described below.

2.1 Vegetational diversity

Uniformity in plant genetic material has been recognized as one of the main causes of an increase in pest problems in monoculture fields. A large number of experiments carried out under different conditions indicated a reduction in pest activity with diverse vegetation as compared with monoculture.

There are several reviews of insect abundance in diversified vegetation (Southwood and Way 1970; van Emden and Williams 1974; Goodman 1975; Cromartie 1981; Altieri and Letourneau 1982, 1984; Pimm 1984; Altieri and Liebman 1986). Studies on the effect of multiple cropping patterns have been carried out by Marcovitch (1935), Dempster and Coaker (1974), Litsinger and Moody (1976), Perrin (1977), van Emden (1977), Altieri et al. (1978), Kroh and Beaver (1978), Risch (1979), Altieri (1980) and Altieri et al. (1990). There have also been some trials with perennial orchard plants (Peterson 1926; Peppers and Driggers 1934; O'Conner 1950; Chumakova 1960; Leius 1967; Syme 1975; Dickler 1978; Altieri and Schmidt 1985; Altieri 1986).

Weeds in crop fields may affect the activity of insects on the crops. A number of experiments have been carried out to assess this effect (Pimentel 1961; Dempster 1969; Tahvanainen and Root 1972; Root 1973; Smith 1976a, b; Speight and Lawton 1976; Altieri et al. 1977; Altieri and Whitcomb 1979; Theunissen and den Ouden 1980; Altieri and Todd 1981; Altieri, Todd  et al. 1981; Horn 1981; Gliessman and Altieri 1982; Altieri and Gliessman 1983; Ahmed et al. 1988).

Various workers have attempted to elucidate the ecological mechanisms underlying differences in the dynamics of insect herbivores and their natural enemies in simple and diverse crop habitats (Tahvanainen and Root 1972; Root 1973; Bach 1980a, b; Risch 1980, 1981; Altieri and Letourneau 1982; Altieri and Gliessman 1983; Kareiva 1983). To explain the general reduction of pest densities in diverse plant combinations, Root (1973) proposed two hypotheses, the resource-concentration hypothesis and the enemies hypothesis. The resource-concentration hypothesis suggests that in monoculture fields where the same plant species is cultivated over large areas the herbivores find a concentrated source of food in one place that supports uninterrupted population build up. The food plants in pure stands are easily detected and colonized. The pests, particularly the specialists, exhibit longer tenure periods and higher feeding and reproductive success.

Agroforestry introduces plant diversity in a land unit, over both time and space. Complex agroforestry systems may be close to though not equivalent to natural plant communities in a stable ecosystem or a system in ecological succession. In the latter, the type and pattern of vegetation is governed by the forces of nature. The plant communities developing through natural selection have a degree of in-built resistance to insect attack. In agroforestry, however, the choice of vegetation is determined by people and depends on the objective of the system being practised. This freedom to introduce selective diversity enables agroforesters to choose plants with the desired attributes for accomplishing their objectives.

Different degrees of insect injury occur when a host plant is raised with different companion plants. A reduction in pest numbers and increase in predators was observed when blackgram (Vigna mungo) was intercropped with sorghum or pigeonpea, while intercropping with greengram (Vigna radiata) provided favourable conditions for an increase in pest numbers (Dhuri et al. 1986). The mite populations on cassava were higher in a eucalyptus–cassava combination than in a banana–cassava combination in experiments conducted by Ghosh et al. (1986). The grasshopper populations in fields of pearlmillet and sorghum with interspersed neem trees were lower than those in fields with Acacia arabica (Amatobi et al. 1988). Six years after eucalyptus trees were introduced in Malnad, India, a survey revealed that the number of insect species was reduced to a quarter compared with that in areas where no eucalyptus were present (Chakravarthy et al. 1986). Interplanting beans or allowing weeds to grow with collards considerably decreased flea beetle densities on the collards and minimized leaf damage (Altieri et al. 1990).

Insect population dynamics are greatly influenced by the type of vegetation in any plant assemblage. Generally, pest levels are not reduced to the same degree in polyculture systems (Risch et al. 1983). Polyphagous pests exhibit varying levels of activity on different plants in an assemblage. Studies on the biology of Diacrisis oblique on different host plants demonstrated significant differences in growth and fecundity of the pest (Shaw et al. 1988). Similar results have been reported in respect of Heliothis armigera (Bilapate 1988). The type of vegetation in a field also affects the activities of the natural enemies of insect pests. In Israel, the cottony cushion scale, Icerya purchasi, was found to be resistant to predation by Rodolia cardinals on Spartum junceum and Erythrina corallodendrum plants (Mendel et al. 1988), whereas the predator is known to be an efficient biocontrol agent of the scale on other plants elsewhere.

Diversity does not always result in reduced pest populations. It appears to be pest specific and also site specific, as well as being affected by other factors. Weed diversity reduced the incidence of fall armyworm, Spodoptera frugiperda, but not of the earworm, Herliothis zea, in a corn field (Altieri and Whitcomb 1980). The intensity of Trachylepida sp attack on Cassia fistula seeds was less in isolated plants as compared with that in mixed stands (Bhatta and Bhatnagar 1986), indicating that diversity does not always result in a reduction in pest attack.

Many multipurpose woody perennials used in agroforestry possess the inherent properties of wild plants, including genetic diversity, as they have not been domesticated for a long time and have not been subjected to rigorous genetic selection, unlike most other plants of economic importance. Before species considered for agroforestry are subjected to breeding for improvement, their insect-resistance characters should be studied in different provenances.

2.2 Taxonomic alliance

Plants belonging to the same or a very close taxonomic group have the tendency to share common pests. In agroforestry systems, aligophagous and polyphagous insect pests are expected to thrive if both components belong to the same or a closely allied taxonomic group. A large number of plants considered for agroforestry are legumes. Combining crops such as pulses or some oilseeds with leguminous woody perennials may result in supporting pest populations common to both components. The most common example is that of the bruchids. Species of caryedon beetles such as Caryedon serratus infest groundnut as well as a number of legume tree seeds. During the early establishment phase of agroforestry systems, tree components are likely to suffer the major injury with attack of the seeds in the limited pods produced. Later, the crops are likely to be more affected as the bruchid populations built up on the tree seeds begin to infest the pulse crops. The damage caused by the pod borer, Etiella zinckelia, in peas has been reported to be accentuated by the presence of acacia trees in the vicinity, as reported by Szeoke and Takacs (1984).

Plant species belonging to the same taxonomic group may contain common or closely related biochemicals that are sought by the insect pests. An insect feeding on a plant with a certain biochemical make-up will adapt more easily to closely related plants with similar biochemical constituents than to species that have entirely different constituents because of taxonomic differences. An agroforestry system comprising plant species belonging to different taxonomic groups is expected to be less affected by insect pests than a system composed of closely related species.

The availability of food over an extended period afforded by closely related plants in the agroforestry field contributes to multiplication of pest populations. The perennial plants in the system may provide a year-round food supply for the pests and thus favour maintenance of the pest population between seasons when the main food—the crop—is absent from the field. It will also result in the maintenance of the population of the pests' natural enemies.

Thus, when considering species for agroforestry, it is advisable to include plants from different taxonomic groups to avoid sharing common pests. The advantages accruing from the maintenance of natural-enemy populations also need to be weighed up.

2.3 Non-taxonomic alliance

Under natural conditions, even insects with a limited host range have been observed to feed on taxonomically diverse species of plants. Thus, taxonomically distant plant species could also be hosts for insect pests. Cocoa grown under leucaena shade suffered more from attacks of defoliating Lepidoptera than when it was grown under some forest trees because the pests were able to use leucaena as an alternative food source (Room and Smith 1975). Eucalyptus grown as shade trees with tea shares attack by Chrysolampra flavipes in Assam, India (Gope 1985). Cross infestation of cassava mites, Tetranychus spp, on banana and of thrips, Retithrips syriacus, on leucaena have been reported by Ghosh et al. (1986). In Hawaii, the thrips, Frankliniella occidentalis, which is a vector of spotted wilt virus on lettuce, tomato, cabbage, etc., occurred on leucaena blossom (Yudin et al. 1986). The bagworm, Oiketicus kirbyi, a pest of coffee, has been demonstrated to develop on eucalyptus leaves in Brazil (Arce et al. 1987). In an agroforestry system, therefore, the plant assembly should consist of species that do not double as host for insect pests of other plants in the system—whether crops or woody perennials.

Some insects utilize different host plants as food in their larval stages from those eaten in the adult stage. Thus an even greater range of plants in an agroforestry system may be attacked by different stages of an insect pest. A plant believed to be a non-host for an insect pest in one stage may turn out to be a host plant for another stage. An example of such a situation occurs in the co-cultivation of crops such as pearlmillet, cowpea, greengram, mothbean and sesame with such tree species as Prosopis cineraria, Azadirachta indica and Zizyphus spp in the rainy season in semi-arid regions of India. The woody species are host plants for adult chafer beetles (Lachnosterna spp, Holotrichia spp, Anomala spp, Adoretus spp and others) whose larvae cause heavy losses to these rainy-season crops in some areas. The presence of such host plants in the crop fields serves to attract the adult beetles. Although the feeding time on the woody plants is very short, the eggs released by the gravid beetles hatch into white grubs, which infest the crops, causing heavy losses.

The insect vectors of plant diseases often survive on a variety of host plants that are not taxonomically close. White flies (aleurodids) and some aphid and jassid species that are vectors of viral, bacterial and other plant diseases often attack a number of different types of plants. The diseases may not affect all these collateral or refuge hosts, but the survival and multiplication of the carrier insect is favoured by them. The presence of refuge host plants in an agroforestry system may provide a source of infection for the crops or human beings if the plant harbours vectors that spread human disease.

The presence of component plants in a system that may serve as host to a vulnerable stage of the insect pest, or that act as a collateral host for plant, animal or human disease, may be utilized to trap and kill these insects while keeping the main crop untreated. Such plants are, however, different from trap plants that are specifically grown to attract insects from the main crop and are not otherwise a component of the agroforestry system.

When establishing an agroforestry system, it is useful to consider whether the component plants serve as host to any vulnerable stage of an important insect pest of the other component plants or as collateral or refuge hosts for the vectors of diseases. Exclusion of such plants from the system could help minimize insect and disease attack on crops or woody perennials.

2.4 The host range of pests

The severity of pest infestation in an agroforestry system will depend on the host range of the attacking pests and their relative abundance in the system. Insects with a wide host range will be able to multiply on a number of host plants, while monophagous insects will be restricted to a limited number of host plants within the system. If all or most plants in a mixed system are palatable to a polyphagous pest, then it is likely that the insect will stay longer and become more numerous, causing greater damage (Speight 1983). Populations of Empoasca krameri were reduced when beans were interplanted with non-host grasses (Altieri et al. 1977; but were not affected by the presence of Amaranthus dubius. The abundance of the leafhopper Scaphytopius acutus increased in peach orchards when the ground cover consisted of host plants but was not affected when a non-host grass was used.

If the insect is a relatively specialized feeder, the population density will be lower in plant communities with a higher diversity. On the other hand, the more general pests are likely to increase in abundance if increasing plant diversity in an agroforestry field increases the number of potential host plants in the system. Even for relatively oligophagous pests, increasing diversity may provide a greater number of suitable hosts and lead to increased abundance. A hypothesis to explain lower numbers of herbivores in diverse plant communities proposes that the rate of emigration, rather than the rate of colonization, is the factor most affected by the presence of non-host plants in the system. Studies of various leaf beetles indicate that they have a shorter residence time in plant patches with non-host plants and move farther after encountering a non-host plant (Bach 1980a, b; Risch 1980, 1981). Saxena and Basit (1982) found that both colonization and residence time of the leafhopper Amrasca devastans on a host plant were influenced by the presence of the non-host plants.

The presence of non-host plants in an agroforestry system is of paramount importance in managing pests. In fact, the major variable determining herbivore abundance is the ratio of host to non-host plants rather than the actual number of plant species in an agroforestry system. Through entomological investigations it is possible to manipulate the proportions of host and non-host plants in any agroforestry system. Intelligent manipulation of crop and woody-perennial combinations can minimize insect damage in an agroforestry system.

Monophagous pests can be controlled altogether by not including their host plants in the system. The host range of oligophagous and polyphagous pests can also be narrowed by eliminating palatable species from the assemblage and replacing them with non-host plants. Monophagous pests are most easily managed in this way. In a review of polyculture systems, Andow (1983a) reported that monophagous herbivores are more likely to decrease in diverse systems than polyphagous pests (61.3% versus 27.1%) and less likely to increase (10% versus 43.8%).

2.5 Biological control potential

Agroforestry systems, particularly the complex ones, have a great potential for controlling pest populations through increasing the efficiency of biological control agents. The natural-enemy hypothesis proposed by Root (1973) to explain reduced herbivory in polyculture systems has been tested in a large number of field experiments and the results were mostly supportive of this hypothesis. Polycultural systems such as agroforestry offer alternate prey, nectar sources and suitable micro-habitats for parasitoids and predators. Being perennial, these systems support the natural enemy population within and between seasons, especially during the off season of the main crop.

Greater colonization and abundance of natural enemies in a mixed culture of plants has been demonstrated in many experiments. Smith (1969) studied the colonization of a brussels sprouts field by Anthocorus nemorum with and without weedy vegetation. The predators were more abundant in samples from weedy areas (representing diverse vegetation). Rapid colonization or higher densities of predator species in dense vegetation have been reported by Sprenkel et al. (1979) and Horn (1981). Letourneau and Altieri (1983) and Letourneau (1990) suggested that predator-colonization rates could be manipulated through vegetational diversification of the crop habitat. Similarly, an increase in the diversity of tree species might increase food sources for adult parasitoids (Mendel 1988).

The presence of alternate prey associated with non-crop vegetation can prevent the local extinction of predator species (Doutt and Nakata 1973) or increase the proximity of colonizer sources (Flaherty 1969). The population of Plutella maculipennis on cabbage raised with a background vegetation of Crataegus sp was regulated by the parasitic wasp Horogenus sp, for which the background vegetation served as an alternate host (van Emden 1965). The predating activities of ground beetles were enhanced when cabbage was undersown with white and red clover, resulting in regulation of populations of Erioischia brassicae, Brevicoryne brassicae and Pieris rapae (Dempster and Coaker 1974). Regulation of populations of Mamestra brassicae, Evergastis forficalis and Brevicoryne brassicae due to predator enhancement as a result of co-cultivation of brussels sprouts with Spergula arvensis has been reported by Theunissen and den Ouden (1980). Increased abundance of predators in collards with a weedy background checked the growth of Myzus persicae (Horn 1981). A greater potential for pest control in complex systems as compared with simple ones was demonstrated by an increased colonization rate by a generalist predator in experiments involving the flower thrip Frankliniella occidentalis and the predator bug Orius tristicolor (Letourneau 1990). The same author reported increased visits of hymenopteran parasitoids in mixed-crop assemblages as compared with pure stands of squash.

Polycultures, especially those containing flowering trees and shrubs, can provide more pollen and nectar sources attractive to and sustaining predators than monocultures. Ageratum conysoides bears flowers all year round, providing pollen for mites and favouring colonization and build up in citrus orchards (Mai et al. 1979). Introduction of flowering perennials or short-lived plants in an agroforestry system will contribute towards biological control of pests.

Some plants produce chemicals that enhance the efficacy of predators. Parasitization of corn earworm eggs by wild Trichogramma sp wasps was promoted by applying extracts of the weed Amaranthus sp (Altieri et al. 1983). The presence of these plants that increase the predatory activities of insects by virtue of producing such chemicals will benefit an agroforestry system. The parasitization of Cotesis kazak was more serious on Heliothis armigera larvae feeding on cotton, okra and tomato, than on those feeding on chickpea, pigeonpea, cowpea and dolichos bean under similar circumstances (Jalali et al. 1988).

A major factor determining natural-enemy abundance in mixed vegetation is the suitability of the microhabitat provided by one or more of the plants in the assemblage. Trees and shrubs often provide better shelter and mating sites than do short-lived annual plants. Hedges provide very favourable environments for parasitic Hymenoptera and Diptera. The Braconidae tend to colonize the leeward side of hedges and other thick vegetation, while the Vespidae and some Diptera accumulate more on the windward side (Pasek 1988). These plants also provide pupation, mating and over-wintering sites for natural enemies. Many predatory spiders prefer to inhabitate woody plants rather than annual plants. The web density of the spider Stegduphus sarasinerum was higher on such plants as Acacia tortilis, Prosopis cineraria, Capparis sp, Tecomella spp and Zizyphus spp (Chandra 1987) than on less woody species.

The presence of different herbivores in an agroforestry system may encourage predators to remain when their main prey is rare. Prey densities that fall below a certain threshold may cause emigration of natural enemies from an area. It is, therefore, important that prey availability is maintained in the system. Predators often have a wider host range than parasites and thus have a better chance of survival in the event of the population of the main prey falling to a low level.

2.6 Microclimate

The interactions among plants in a mixed agroforestry system create a micro-environment that is different from that of the surrounding area. Shade is the most prominent consequence of tree–crop combinations and it may have both a direct and an indirect effect on the activity of pests and natural enemies. Among other effects, shade provides protection from direct sunlight and regulation of light intensity underneath the tree. Reduction in temperature and an increase in humidity are indirect effects. Most diurnal insects, especially soft-bodied larval stages, prefer to feed while avoiding direct exposure to the sun. Under shady conditions their feeding activity could be enhanced, thus increasing their damage potential. Light intensity governs the habitation of fields by some dung beetles (Doube 1983). Most aphid species prefer shady conditions in the warm climates. Some pests prefer sunny conditions. Risch (1981) observed that the density of beetles on beans under the shade of maize plants was less than on beans in monoculture. Infestation by Polyura naraea was higher on stands at the edges of forest canopy facing the sun than on the shady side (Zhou et al. 1985). Shade from trees can interfere with the host-seeking and reproductive behaviour of some insects (Risch 1981; Yang et al. 1988). Many hymenopteran parasites exhibit greater host-searching capacity under bright-light conditions, so their activity may be retarded in the shade. Whether shade has a positive or a negative effect on pest activity depends on the habits of the pests in the agroforestry system. The shade intensity will depend on the growth habits, structure and arrangement of the trees in the system.

The humid conditions in an agroforestry system may be favourable for the development of disease in insect pests. Coupled with the absence of direct sun, the effectiveness of entomopathogenic fungi may be increased by humidity (Jaques 1983). In experiments on the effect of sunlight on the field persistence of Nomuraea rileyi, Fargues et al. (1988) found that the half-life and viability of spores of this fungus were much longer in the shade. The effective infective period of Bacillus thuringiensis, the most widely used entomopathogenic bacterial preparation, is greatly reduced in sunlight. Increased humidity has also been reported to favour parasitization of pest eggs by Trichogramma sp (Pu 1978).

2.7 Masking effect

Many plants emit volatile chemicals or odours into the environment. These odours are perceived by herbivorous insects and utilized to orient themselves to the host fields. Onion flies, Hylemya antigua, have been found to fly directly towards the odour released by an onion bait (Dindonis and Miller 1980). In monocultures, the odours released by the plants spread out in all directions and are perceived by the herbivores without any interruption. But in mixed vegetation, the odours released by some plants may mask the effect of those released by other plants. Under these circumstances, the insects find it difficult to locate the hosts on which to feed and reproduce (Altieri 1986).

Thus, odoriferous plants, when raised with host plants of insect pests, can deter recognition, feeding and reproduction of the pests on their host plants (Dethier et al. 1960; Schoonhoven 1968). Tahvanainen and Root (1972) attributed the lower density of Phyllotreta cruciferae on cabbage raised with tomato, tobacco and ragweed to feeding inhibition by odours from the non-host plants. The lower activity of Plutella xylostella on cabbage intercropped with tomato has been attributed to the repellent effect of the chemical produced by the companion crops. Many plant residues contain allelopathic compounds that could affect the growth of adjoining plants in a mixture, changing the attractiveness of the plants to their pests (Steinsiek et al. 1982). In experiments with living mulches, Altieri et al. (1990) observed lower aphid colonization and population build-up on collards grown with vetch (which they suspected to contain allelopathic compounds) as compared with that on collards in monoculture plots.

The masking effect is not restricted to the insect pests. The performance of natural enemies may also be affected by chemical cues from the associated plants (Altieri, Lewis et al. 1981; Nordlund et al. 1988). The prey-seeking behaviour of insect pests that utilize the odours of the host plants of their prey to locate them is influenced by the presence of other vegetation in the system (Monteith 1960; Shahjahan and Streams 1973). Such a situation will reduce the biological-control potential of the system. However, there could also be cues from companion plants that help establish natural enemies in the system. Identification of the plants that release pest-antagonistic and natural-enemy-attractive odours may be of great importance in developing a pest-free agroforestry system.

2.8 Barrier effects

The tall woody plants in an agroforestry system act as a physical barrier to the movement of insects to, from and within the system. The non-host plants of insect pests raised with their host plants act as biological barriers, restricting their movement towards the host plants. Hedges, boundary plantations and windbreaks affect the colonization and dispersal of both herbivorous and predatory parasitic insects. Both horizontal and vertical movements of insects are affected by tall or thick woody plantations (Pasek 1988). The permeability of the vegetation affects the movement of insects in and out of the system. In- and-out migration patterns of insects in polycultures have been studied by Bach (1984), Risch (1981), and Wetzler and Risch (1984).

Non-host plants mixed in with host plants either act as a mechanical barrier to the dispersal of the insect pest (Kennedy et al. 1959; Root 1973) or physically repel the pests because of unpleasant morphological features such as hairy leaves (Levin 1973). Weakly flying insects such as aphids, thrips, flies and small beetles are carried far afield by the wind. Once in the air, these insects cannot land directly on their host-plant plots because of the high speed of the wind carrying them. They utilize tall trees as obstructions to settle on and later move to the crop fields. Thus, the trees in an agroforestry system may serve as agents for facilitating colonization of insect pests by providing a platform from which they initiate flight to infest the crops.

The woody plants in hedges and boundaries serve as reservoirs for insects (van Emden 1965; Lewis 1969a; Solomon 1981; Onillon 1988). The most abundant taxa within hedges are parasitic Hymenoptera and Diptera. Although hedges contribute to maintaining some populations of insects, most of the increase in insect density results from insects blown in from elsewhere. The Braconidae tend to accumulate on the leeward side, while the Aphididae, Vespidae, large Diptera and Lepidoptera predominate on the windward side (Pasek 1988).

Whether they act as physical or biological barriers, the woody species in an agroforestry system will restrict the movement of insects through the system. This situation is advantageous if the entry of the insect pests is blocked or if outward movement of the natural enemies of the pests is hindered.

2.9 Field configuration and design

Some insect species, while flying high in the air, recognize their host plants by the field configuration. The configuration of a land unit is determined by the type of plants, their colour, structure, height, density in the field and the type and colour of the soil in the background. Any change in these characters may change the configuration of the field, which will affect its recognition by the insects. Dempster and Coaker (1974) observed that even small changes in cropping practices can greatly alter the attractiveness of plots to pests and their natural enemies.

Through the introduction of heterogeneity in a field, it is possible to reduce the number of insect visitors to the field. Heterogeneity can be introduced into the field in several ways, such as through changes in plant density or by mixing different types of plants. The companion crops provide a camouflage for the host crops (Altieri and Liebman 1986), thus preventing the pests from recognizing the host from a distance. Some pests prefer plants of a particular colour or texture (Cromartie 1981). In the Philippines, corn borers were found to avoid a green colour on the ground in corn fields (IRRI 1974). Aphids were found to colonize plants more readily when they stood against a background of bare soil (Kring 1972). The attractiveness of collards to aphids decreased when these were grown against a vetch background (Altieri et al. 1990). Smith (1976a) found that colonization of Pieris rapae on brussels sprouts was reduced when weeds were allowed to grow with the crop. Similar observations have been reported by Dempster (1969). Dempster and Coaker (1974) found that the colonization of cabbages by Erioischia brassicae, Brevicoryne brassicae and Pieris rapae was greatly interfered with when the cabbages were undersown with white and red clover. Colonization of brussels sprouts by Mamestra brassicae, Evergestis forficalis and Brevicoryne brassicae was adversely affected when the crop was grown with Spergula arvensis (Theunissen and den Ouden 1980), probably because of a configurational difference.

Plant-community structures also affect the biology of herbivorous insects and their natural enemies. The number of eggs laid by the pyralid, Cactoblastis cactorum, was affected by plant size, cladode condition, the conspicuousness of the plant, and height above ground (Robertson 1987). Leigh et al. (1974) found plant density to be positively correlated with densities of the predator Orius tristicolor in cotton fields. Carabaeid beetles were more destructive of Pieris rapae caterpillars in weedy plots than in the weeded plots of brussels sprouts (Dempster 1969).

The arrangement of the different plant species in an agroforestry system may have a profound effect on the activities of insect pests and their enemies. The situation of the crops and woody perennials with respect to the direction of wind and sun determines the effect of these agencies on the dynamics of visiting insects. The effect of shading should be minimal in systems where trees have been planted in alignment with the sun's path. Crops raised on the windward side of boundary plantations are less likely to be infested by insect species that are carried by wind.

The design of an agroforestry system will define the extent of the influence of weather on insects. The effect will be the least in systems that have thick vegetation vertically and horizontally. The pest situation in agroforestry may be manipulated through suitable design of systems so that the conditions are favourable for the predatory insects and unfavourable for the pests. The size and density of plants and their relative arrangement in time and space will determine the sphere of activity of the insects.

2.10 Exotic plants and pests

Plants with desirable attributes are often moved from one place to another. In the absence of proper phytosanitary measures and non-compliance with quarantine rules, these plants carry their pests with them to areas where they did not occur previously. In the absence of natural enemies at the new location, these insect species rapidly multiply and establish themselves as serious pests in the new location. The psyllid Ctenartaina eucalypti, the curculionid Gonipterus scutellatus and the cerambycid Phorocantha semipunctata have moved with eucalyptus and established themselves where the tree was introduced (Cadahia 1986). The margarodid Auloicerya acaciae was accidentally introduced into Australia, where it became established on a number of acacia species (Gullan 1986). The Australian sawfly, Phylecteophaga froggati, is causing damage to oak and eucalyptus in New Zealand (Nuttall 1985). Of the 500 species recorded on agricultural crops and forestry plantations in the former USSR, aproximately 80 were introduced insects (Konstantinova and Gura 1986). In Africa, Rastrococcus invadens is becoming established on horticultural and forest plants after its introduction with plant material (Agounke et al. 1988).

When an exotic plant is introduced into a locality, it may affect or be affected by the local insect fauna in a number of ways:

• The introduced plant may become a host for established pests in the area. After it was introduced into Australia, Delonix regia became a host to the longicorn Aridaeus thoracicus, a local pest of pear (Hawkeswood 1985). Acacia tortilis, an introduced plant in India, is attacked by a local pest, Julodis sp, in desert areas. A number of local insects have adapted to eucalyptus in areas where it has been introduced all over the world (Cadahia 1986). In Hawaii, the thrips Frankliniella occidentalis became established on leucaena after the introduction of the latter on the island (Yudin et al. 1986).
• The introduced plant may become a suitable host for insect species that were not considered important pests on local plants. When such insects multiply and establish on the exotic plant, they acquire the status of pests. Before the introduction of eucalyptus into India, Celosterna scabrator was considered a minor pest affecting the acacias. It adapted to eucalyptus and is now considered an important pest on that genus (Chatterjee et al. 1987).
• Exotic plant materials may bring with them some of the pest insects from their native habitat. Or the insects may arrive accidentally in the area where the exotic plant is introduced. In the absence of natural enemies at the new locations, these insects may multiply very fast and become established as major pests. Leucaena leucocephala, introduced in several parts of the world, is heavily infested by the psyllid Heteropsylla cubana , which arrived accidentally at the new sites. The cypress aphid, Cinara cupressi, is causing extensive damage to cypress in Africa.
• An exotic pest coming in with exotic plant material or being accidentally introduced later may establish on other plants in the location in addition to the original host plant. The leucaena psyllid, Heteropsylla cubana, has been found feeding on Samania saman in Asia.

Severe pruning of plants in alley cropping gives rise to new vegetative growth, which is more palatable to the insects than the older growth. Removal of infested plant parts by pruning may help to get rid of sessile sucking pests. Fallowing part or the whole of a unit of land affects insect abundance in the unit, especially of the soil-dwelling stages. Intensification of fallow resulted in increased grasshopper numbers, while cultivation and afforestation affected the grasshoppers adversely in investigations carried out by Amatobi et al. (1988).