Agriculture Reference
In-Depth Information
virulences. In a bulk isolate, the frequency of resistance can be measured in a similar
way as for that of virulence, as the amount of disease on plants treated with a
differential dose of the fungicide compared to that on untreated plants.
Analysis of continuously distributed resistance generally requires the assumption
that log-tolerances are normally distributed. If this is the case, the mean and standard
deviation of the log-ED
50
s of a random sample of isolates should be the same as the
log-ED
50
and the standard deviation of log-ED
50
in the population. For a bulk
isolate, log-ED values are estimates of the corresponding log-ED values in the
population as a whole (e.g. the log-ED
95
of a bulk isolate is an estimate of the log-
ED
95
of the population), provided that the bulk isolate is indeed a random sample of
genotypes in the population.
Matters are a little trickier with a fungicide to which there several distinct levels
of resistance. If individual isolates are tested, a population may be described by a
table of frequencies of the various levels of resistance, as estimated by a measure
such as the ED
50
or MIC. Changes in resistance, for example in response to selection
by the use or withdrawal of a fungicide, may be summarised as shifts to increased
frequencies of higher or lower levels of resistance. There is no fully satisfactory way
of summarising the response of a bulk isolate when there are two or more discrete
levels of resistance to a fungicide. Probit models are often used but the same ED
50
may be estimated if, on the one hand, there is a high frequency of genotypes with an
intermediate level of resistance, or on the other hand, if there are high frequencies of
genotypes with either high or low levels of resistance.
The ED
50
is not always the most useful measure of resistance from the point of
view of providing guidance on fungicide applications to crops, see section 3.2.2(c).
Higher EDs can be estimated from tests on bulk isolates, provided that an
appropriate model can be fitted to the data; i.e. a probit model will produce an
unbiased estimate of ED
95
if log-tolerances are normally distributed but an
inaccurate estimate if there are several discrete levels of resistance (Brown, 1998).
They can also be estimated from the distribution of ED
50
s in a sample of single-
genotype isolates, even though estimates of higher EDs of the individual isolates
themselves have no value.
MICs are often estimated for bulk isolates and it is well known that in doing so,
one is essentially estimating the MIC of the most resistant genotype in the bulk,
even if that genotype is only a small fraction of the total. This makes data of this
kind difficult to use in describing a population. However, if the aim of an
experiment is to predict the dose of a fungicide needed to control a particular
pathogen population or to judge the risk of a control failure should a particular dose
be applied, estimating the MIC of a bulk isolate may give a cheap, quick, useful
answer to the question.
(b) Baseline data
When one wishes to understand how the use of a fungicide has affected a pathogen's
resistance, one needs to know the level of resistance both before and after the
fungicide was used. In order to obtain baseline data on the responses, pathogen
