S. Savary et al., Rice pest constraints in tropical Asia: Quantification of yield losses dueto rice pests in a range of production situations, PLANT DIS, 84(3), 2000, pp. 357-369
A series of experiments was conducted where a range of injuries due to rice
pests (pathogens, insects, and weeds) was manipulated simultaneously with
a range of production factors (fertilizer input, water supply, crop establi
shment method, variety) in different seasons and years. These factors were
chosen to represent lowland rice production situations characterized in sur
veys conducted in tropical Asia and their corresponding range of attainable
yield. Experiments complemented one another in exploring the response surf
ace of rice yields to yield-limiting and yield-reducing factors. The result
ing experimental data base consisted of 445 individual plots and involved 1
1 manipulated injuries in a range of attainable yields of 2 to 11 t ha(-1).
A first, nonparametric, multivariate analysis led to a hierarchy of potent
ial injuries, from marginally (e.g., bacterial leaf blight) to extremely ha
rmful (e.g., rice tungro disease). A second, parametric, multivariate appro
ach resulted in a multiple regression model involving factors generated by
principal component analysis on injuries that adequately described the vari
ation in actual yield. One major finding was that some (attainable yield x
injury factors) interactions significantly contributed to the description o
f variation in actual yield, indicating that some injuries (or their combin
ations) had a stronger or weaker yield-reducing effect, depending on the le
vel of attainable yield. For instance, yield losses due to sheath blight, w
eed infestation, and rice tungro disease tend to increase, remain stable, a
nd decrease, respectively, with increasing attainable yields. Back-computat
ions using the principal component regression model estimated yield losses
caused by individual injuries, using the mean injury levels in a population
of farmers' fields surveyed across tropical Asia. The results indicate tha
t sheath blight, brown spot, and leaf blast are diseases that cause importa
nt losses (between 1 and 10%) regionally. Among the insect injuries, only w
hite heads caused by stem borers appear of relevance (2.3% yield losses). T
hese injuries, however, do not match in importance those caused by weeds, w
hether outgrowing the rice crop canopy (WA) or not (WB), both types of inju
ries causing about 20% yield losses when considered individually. When all
mean injuries were combined into one mean injury profile occurring at a reg
ional attainable yield of 5.5 t ha(-1), a mean yield loss of 37.2% was esti
mated, indicating that injuries were less than additive in their yield-redu
cing effects. Scenario analyses were conducted in a set of (production situ
ations x injury profiles) combinations characterized from surveys in farmer
s' fields in tropical Asia. Depending on the scenario chosen, losses rangin
g from 24 to 41% were found.