The main goal of morphometrics is to study how shapes vary and their covariance with other variables.
Even though morphometrics can be used to describe the form of any object it is mostly used in biology to
describe organisms. Morphometrics is very important in biology because it allows quantitative descriptions of
organisms. Quantitative approach allowed scientists to compare shapes of dierent organisms much better
and they no longer had to rely on word descriptions that usually had the problem of being interpreted
dierently by each scientist. This shift to quantitative descriptions was caused by advances in statistical
analysis methods that allowed to interpret collected data.
First method of morphometrics called Traditional morphometrics was done by measuring linear distances
(such as length, width, and height) and multivariate statistical tools were used to describe patterns
of shape variation within and among groups. This approach also sometimes used counts, ratios, areas and
angles measures. The biggest advantage of this method was that it was very simple, however it had several
diculties. The biggest problem was that linear distance measurements are usually highly correlated with
size and this makes shape analysis dicult. Another problem was that measurements taken from two dierent
shapes could produce equal results because the data did not include the location of where the measurements
were taken relative to each other. And it was also not possible to reconstruct graphical representation of
the shape from taken measurements. Figure 1 illustrates the problems of Traditional morphometrics. To
overcome these problems a more sophisticated method called Geometric morphometrics was created.
Landmark-based geometric morphometrics uses a set of landmarks to describe shape. Landmark is a twoor
three-dimensional point described by a tightly dened set of rules. The results that are generated by this
method directly depend on the quality of landmarks. A lot of eorts have to be put to choose landmarks that
would have high evolutionary signicance. Each landmark also has to be present on every studied organism.
If a landmark is not present on at least one of studied organisms it either has to be marked approximately
or it can not be used at all. The number of landmarks selected should not exceed the number of specimen
samples because extra landmarks will be redundant. Usually number of landmarks is approximately equal
to the number of specimen samples. There are three types of landmarks that can be used. True landmarks
that have some biological signicance. Pseudo-landmarks are dened by relative locations e.g. "the point
of highest curvature of this bone". Semi-landmarks are dened by a location relative to other landmarks
e.g. "midway between landmarks X and Y". Landmarks can sometimes have weighted value in analysis
according to their importance.
Extracted landmark data has a lot of variations in position, orientation and scale between specimens.
These non-shape variations have to be removed before further analysis. There are several methods used to
superimpose landmarks each of them having dierent optimization criteria. The most simple one is two-point
registration. This method translates, scales and rotates all landmarks such that two named landmarks are in
the same place in all specimens. The biggest disadvantage of this method is that it removes all the data from
those two landmarks. Another popular method is Generalized Procrustes analysis (GPA, also sometimes
called Generalized least squares). This method rst calculates the centroid of landmark congurations and
translates it to the origin. This is done by taking k points in two dimensional space
The main goal of morphometrics is to study how shapes vary and their covariance with other variables.
Even though morphometrics can be used to describe the form of any object it is mostly used in biology to
describe organisms. Morphometrics is very important in biology because it allows quantitative descriptions of
organisms. Quantitative approach allowed scientists to compare shapes of dierent organisms much better
and they no longer had to rely on word descriptions that usually had the problem of being interpreted
dierently by each scientist. This shift to quantitative descriptions was caused by advances in statistical
analysis methods that allowed to interpret collected data.
First method of morphometrics called Traditional morphometrics was done by measuring linear distances
(such as length, width, and height) and multivariate statistical tools were used to describe patterns
of shape variation within and among groups. This approach also sometimes used counts, ratios, areas and
angles measures. The biggest advantage of this method was that it was very simple, however it had several
diculties. The biggest problem was that linear distance measurements are usually highly correlated with
size and this makes shape analysis dicult. Another problem was that measurements taken from two dierent
shapes could produce equal results because the data did not include the location of where the measurements
were taken relative to each other. And it was also not possible to reconstruct graphical representation of
the shape from taken measurements. Figure 1 illustrates the problems of Traditional morphometrics. To
overcome these problems a more sophisticated method called Geometric morphometrics was created.
Landmark-based geometric morphometrics uses a set of landmarks to describe shape. Landmark is a twoor
three-dimensional point described by a tightly dened set of rules. The results that are generated by this
method directly depend on the quality of landmarks. A lot of eorts have to be put to choose landmarks that
would have high evolutionary signicance. Each landmark also has to be present on every studied organism.
If a landmark is not present on at least one of studied organisms it either has to be marked approximately
or it can not be used at all. The number of landmarks selected should not exceed the number of specimen
samples because extra landmarks will be redundant. Usually number of landmarks is approximately equal
to the number of specimen samples. There are three types of landmarks that can be used. True landmarks
that have some biological signicance. Pseudo-landmarks are dened by relative locations e.g. "the point
of highest curvature of this bone". Semi-landmarks are dened by a location relative to other landmarks
e.g. "midway between landmarks X and Y". Landmarks can sometimes have weighted value in analysis
according to their importance.
Extracted landmark data has a lot of variations in position, orientation and scale between specimens.
These non-shape variations have to be removed before further analysis. There are several methods used to
superimpose landmarks each of them having dierent optimization criteria. The most simple one is two-point
registration. This method translates, scales and rotates all landmarks such that two named landmarks are in
the same place in all specimens. The biggest disadvantage of this method is that it removes all the data from
those two landmarks. Another popular method is Generalized Procrustes analysis (GPA, also sometimes
called Generalized least squares). This method rst calculates the centroid of landmark congurations and
translates it to the origin. This is done by taking k points in two dimensional space
การแปล กรุณารอสักครู่..
