Agriculture Reference
In-Depth Information
2.6 ASSESSMENT OF YIELD LOSS
2.6.1 Confounding factors
Relating the yield of crops to varying levels of plant disease has remained a complex
task both in theory and practice. The difficulty of interpreting disease-yield loss
relationships was recognized by James (1983), one of the pioneers in this field of
research. Madden and Nutter (1995)
argued that the failure to measure disease
intensity (severity and incidence) accurately was a major factor contributing to this
difficulty; they concluded, however, that recent research in crop loss assessment
would lead to a better understanding and to predictions of crop losses through
sophisticated modelling, progress in sampling theory and application, and advances
in instrumentation.
There may be several confounding factors that weaken the statistical relationship
between the independent variable (assessment) and the corresponding dependent
variable (yield loss). Such factors have been identified as: interactions between
diseases and between the pathogen and an environmental factor; the self-limiting
effect of local lesions; overcapacity of the host plant (within-plant compensation);
and between-plant compensation and lesion position effects. In relation to the latter,
Zadoks and Schein (1979) cautioned against the conclusion that lesions of equal size
always had equal effects on yield and crop loss, and cited the example of
Phytophthora infestans
(cause of potato late blight) in which a stem lesion can kill a
haulm with 10 leaves and 50 leaflets whereas the same-sized lesion on a leaflet kills
only one leaflet at most. The same caution would apply to the effects of axil and leaf
blade lesions of equal size caused by
Rhynchosporium secalis
on barley.
Another identifiable confounding factor is the often poor correlation between
visible symptoms and amount of tissue colonization. Precise techniques can now
measure fungal biomass using chitin or ergosterol as biomarkers (section 2.5.3). In
Fusarium
ear blight of wheat, kernels in asymptomatic spikelets may be infected
and mycotoxin content may not be correlated with visible symptoms. Mycotoxin
produced in grain can have serious consequences for the food chain; assays for
mycotoxins may be more important for the milling and baking industries than
estimates of disease symptom incidence and severity (Shaner, 2003). Madden and
Nutter (1995) reviewed the approaches for modelling crop losses in relation to
disease intensity and identified additional factors that might change our
understanding of the disease-yield loss relationship, such as the relevance of healthy
leaf area duration, radiation interception, spatial pattern of disease intensity and time
of infection.
Bryson
et al.
(1995) suggested that a simple model relating loss of green area
within a winter wheat crop canopy to changes in light interception might be useful
in predicting disease-induced yield losses by yellow rust (caused by
Puccinia
striiformis)
;
such a yield loss model is thus based on crop function rather than
measurements of disease severity and related area under the disease progress curve
(AUDPC) values which have no mechanistic link to the productivity of the host
plant. Bryson
et al
. (1995) obtained a significantly better relationship between area
under leaf green area index progress curve (AULGAIPC-synonymous with healthy
