Agriculture Reference
In-Depth Information
Figure 2.4. Accuracy and precision of an archer when the objective is to place all arrows in
the central circle: (a) accurate and precise; (b) not accurate, but precise; (c) not accurate
and not precise. (From Campbell and Madden, 1990a).
Although assessment of disease incidence is traditionally based on visual disease
symptoms, the definition can easily accommodate other more modern methods such
as the enzyme-linked immunosorbent assay (ELISA) and polymerase chain reaction
(PCR) (section 2.5.3); disease incidence is a binary variable, that is, a plant unit is
either (visibly) diseased or not (Madden and Hughes, 1999). Disease incidence would
be suitable for assessing systemic infections which may result in total plant loss (e.g.
viruses or cereal smuts) as well as many root diseases, or where a single lesion causes
leaf death (e.g. axil lesions in barley caused by Rhynchosporium secalis) but may also
be useful in the early stages of an epidemic caused by a cereal foliar pathogen when
both incidence (number of tillers affected) and severity (leaf area affected) are related
and increase simultaneously (James, 1983). Hughes and Madden (1995) and Madden
and Hughes (1999) reviewed the methodology for the analysis of disease incidence
data especially where aggregated patterns of disease occur. The authors pointed out
that generalized linear models (GLMs) can be used for binomially distributed (i.e.
random) data and overcome the problems of applying analysis of variance (ANOVA)
to proportions; however if diseased plants or leaves are aggregated or in clusters
(i.e. beta-binomially distributed), difficulties can arise in determining statistical
significance between experimental treatments. Ridout and Xu (2000) explored the
relationships between several quadrat-based statistical methods that have been
applied to spatial aspects of disease incidence data. In contrast to incidence, disease
severity is a continuous variable typically bound by 0 and 1, and a measure of the
quality of plant tissue rather than the number of plant units affected (Madden and
Hughes, 1999). In general, incidence is easier and quicker to assess than severity and
is therefore more convenient to use in disease surveys where many observations are
needed or when non-experts are used to collect data (Madden and Hughes, 1995);
however, severity may be a more important and useful measurement for many
diseases and is sometimes measured as the number of colonies (or lesions) per plant
unit (disease density) (Xu and Madden, 2000).
Relationships between incidence and severity ( I-S relationships) are examples of
data comprising a spatial hierarchy and are an epidemiologically significant concept;
any quantifiable relationship between the two parameters may permit more precise
Search WWH ::




Custom Search