Case Study: The effect of artificial defoliation on the yield of pole
1. The time at which the beans are most susceptible to defoliation; and
2. The relationship between pole bean yield and manual defoliation of beans, when beans were defoliated once, repeatedly and at different levels within the upper and lower halves of the plant.
A split-plot design was used to evaluate the effects of defoliation time (main plots) and various defoliation levels (sub-plots). Treatments were replicated four times. Each sub-plot was defoliated only once, at one of the defoliation levels of 100, 80, 60, 40 and 20% with defoliation starting from either the top of the plant (‘upper’ treatment) or from ground level (‘lower’ treatment). Sub-plots which were not defoliated were the controls. The times (main plots) at which defolia- tion was carried out were 1, 3, 5 and 7 weeks after plant emergence.
For the repeated defoliation experiment, defoliation levels of 0, 10, 20, 30 and 50% were assigned to plots in a randomized complete block design. Foliage was removed weekly so that each plot was maintained at the required defolia- tion level when compared with the undefoliated control.
The regression analyses of the relationships between yield and levels of defo- liation were all significant (P < 0.005; Fig. 3.6) except for week 1. Week 5 corre- sponds to the blooming (flowering) period, as it was at this time that the plants appeared most sensitive to defoliation. As expected in the continuous defoliation experiment, yield decreased with increasing defoliation (Fig. 3.7) so that a con- tinuous 50% reduction in foliage resulted in yield losses of approximately 40%.
Van Waddill et al. (1984) concluded from this study that a large proportion of pole bean yield loss could be accounted for by defoliation. Plants were most sen- sitive to defoliation during the blooming and pod set period.
Fig. 3.7. The relationship between pole bean yields and weekly defoliation levels (after van Waddill et al., 1984).
3.6.3 Control of initial levels of infestation Levels of natural infestation can be con- trolled to provide differences of pest inten- sity at the start of a yield loss trial. The most common form of control measure used is insecticides but plant varieties have also been used (Walker, 1987).
Insecticides can be used to create dif- ferent levels of pest infestation through use of different doses applied at specific times (Egwuatu and Ita, 1982), predeter- mined application times (Dina, 1976;
Kirby and Slosser, 1984) or different insecticides can be used to create and maintain different levels of infestation (Yencho et al., 1986). With natural infest- ations the absolute level of attack and hence the degree of yield loss between treatment plots cannot be predetermined.
In seasons where population levels are low yield loss may be negligible and treat- ment differences insignificant. In addi- tion, in situations where a number of different pests attack a given crop differ- ent pesticides may have to be used to con-
trol the different pests in order to separate their effects.
The natural level of pest infestation can also be controlled by the use of plant vari- eties which exhibit different levels of sus- ceptibility to the pest insect. The range of possible intensities that can be produced depends on the number of varieties having variable but distinct levels of resistance.
The biggest drawback with this technique is that ideally in the absence of pest attack or at the same level of attack the different varieties should produce similar yields (Walker, 1981). In practice this is often dif- ficult to achieve.
3.7 Paired Treatment Experiments
The most commonly used and the simplest method of evaluating losses due to insect pests is the paired-treatment experiment (e.g. Fisher and Wright, 1981; Cole et al., 1984; Barnard, 1985). This technique is equally applicable to field crops, orchards,
Fig. 3.8. The layout of a randomized complete block strip trial in a paired-treatment experiment where: P = protected plot; NP = unprotected plot and G = guard rows (after LeClerg, 1971).
forests, stored products and livestock sys- tems and in its simplest form involves a comparison of the yield in control plots where there is no infestation (control by insecticides may be necessary) with the yield from plots having natural or con- trolled levels of infestation. A suitable number of replicates is used. Treatment and control plots should be arranged in pairs, but the allocation of treatment or control to each of the plots should be made at random (Fig. 3.8). In many situations however, it is not possible to follow this simple model, although in principle a comparison is still made between uninfested and infested hosts, but the way this is achieved is dependent on the pest/host system.
In forest systems, a common approach is to compare the growth of trees that are attacked by heavy infestations of defolia- tors with that of less heavily attacked trees (Thomas and Miller, 1994) or to compare growth of the same trees during periods of high and low pest attack (Seaby and Mowat, 1993; Day and Leather, 1997). Such studies have the problem of having only
‘pseudo-controls’ but measuring sub-lethal effects of infestation in forest stands is dif- ficult if not impossible to manipulate in any other way. Aerial photography has also provided a means of estimating yield losses from insect pests in forest plantations (Chiang and Wallen, 1971).