Agriculture Reference
In-Depth Information
14.3.2 Relationship between disease dynamics and yield components
Disease-yield loss relationships (see also Chapter 2) must be determined to assess
the agronomic efficacy and economic benefits of control methods. Few studies have
focused on the relationships between soil-borne disease progress curves and crop
response, for three main reasons. First, studies on the development of soil-borne
disease epidemics must include a large number of representative samplings, because
of the patchy distribution of most of these diseases. Second, disease assessment for
soil-borne pathogens involves destructive sampling, which leads to discontinuities in
the dynamic representations of both disease and crop growth. Third, it is not always
possible to compare, in equivalent conditions, a healthy situation with various levels
of disease in the crop.
The question of yield losses due to take-all has been addressed by comparing
situations involving different crop rotations (Slope and Etheridge, 1971), sowing
dates (Bateman
et al
., 1990), or artificial inoculations with different amounts
of fungal inoculum (Rothrock, 1988) or with the same amount of inoculum
incorporated at different depths (Hornby and Bateman, 1990), in order to generate
differences in epidemic patterns. These approaches have generally focused on total
yield at harvest, but rarely, the various yield components formed successively during
the wheat cropping season (Meynard and Sebillotte, 1994) have been investigated:
ear number per square meter (sowing to mid-stem elongation), grain number per ear
(floral initiation to flowering), grain number per square meter (sowing to flowering),
and 1,000 grain weight (flowering to maturity). The impact of the disease is likely
to depend on when disease occurs and, consequently, on the nature of the yield
components affected. However, most studies have simply established correlations
between damage (i.e. yield reduction) and disease level at flowering (Bateman
et al.
,
1990), grain filling (Slope and Etheridge, 1971; Hornby and Bateman, 1990), or
harvest (McNish and Dodman, 1973), and have taken no account of the link between
disease dynamics and crop growth dynamics.
Schoeny and Lucas (1999) carried out a series of experiments in which a
fungicidal seed treatment was used to generate different disease incidence and
severity progress curves at a single location, with identical cultural practices and
climatic conditions. Schoeny
et al.
(2001) then investigated the effects of various
take-all epidemics on yield formation as a function of disease progression. Simple
linear regression models involving various disease variables were compared and
their ability to account for and predict the losses of yield components was assessed.
Yield losses at harvest were strongly linked to the area under the disease progress
curve (AUDPC) for disease incidence calculated between sowing and flowering
(Fig. 14.3). The observed losses were larger for plots to which low rates of fertiliser
were applied than for plots to which high rates of fertiliser were applied. Losses in
terms of ear number per square meter, grain number per ear, and grain number per
square meter were mainly related to cumulative disease incidence, calculated as
AUDPC, during periods corresponding to yield component formation (sowing to
mid-stem elongation, floral initiation to flowering, and sowing to flowering,
respectively). In contrast, 1,000 grain weight losses were linked to disease incidence
at mid-stem elongation (i.e. at a growth stage before the formation of this yield
