Image Analysis: Basic Procedures for Description of Plant Structures - Plant Cell Morphogenesis: Methods and Protocols - page 69

3 Methods

3.1 Sampling

To get unbiased estimation of quantitative parameters of studied
objects, it is necessary to sample the specimens in a SUR way, so
each item has the same probability to be sampled. The more het-
erogeneous the structure is, the more segments, samples, and sam-
pling windows are needed [ 11 ]. The aim of good experimental
design is to quantify natural variability of data due to biological
reasons (treatment, or biological process). To characterize natural
variability of data, the other, artifi cial sources of the error have to
be minimized—such as error due to wrong sampling and design
experimental error (observer, devices, etc.) or sample processing
( see Note 1 ). SUR sampling principles applied in design lead to
unbiased estimations of followed structural parameters.
Answer the following questions and modify the design based
on your answers:

1. What and where you have to sample? Are there known gradi-
ents of the parameter you are going to measure? Sample the
same/comparable parts of plants to exclude natural variation
in anatomical gradients.
2. Do you need the average value for the whole plant/organ/
tissues? In that case you have to apply the SUR principle for
sampling, e.g., segments of an organ to be further processed
to get sections, to which any measurement procedure is
applied. For example, for study on leaves you have to sample
several segments in SUR way.
3. Are you interested just in comparison of some specifi c parts of
the organ/plant? For example, for more extensive study on
leaves you can compromise to comparison of well-defi ned part
within the leaf blade, e.g., middle of the coniferous needle.
4. Principle of SUR sampling for elongated plant structures, e.g.,
coniferous needles: Sample the positions of segments along lon-
gitudinal organ axis. Determine the interval (distance T )
between the segments, in which segments will be sampled along
the organ. The value of T can be derived from the average length
of the axis, divided by 5-10 (you want to sample between 5 and
10 segments). Get a random number from the set {0, 1, 2,…,
T −1} from a table of random numbers or from some generator
(e.g., http://www.random.org ) to determine the position of
fi rst segment— z . Cut the initial segment at distance z , next seg-
ment at z + T , then at z + 2 T, etc. (for details see refs. [ 7 , 8 ]).

5. Principle of SUR sampling for planar structures, such as blade
of broadleaves of many herbs. Sample the positions in a fl at
(2D) organ in a SUR way. Get random numbers for coordi-
nates x , y from a table of random numbers or from some gen-
erator (see above)—for the horizontal coordinate x from the
set {0, 1, 2, …, a −1} and for vertical coordinate y from the set