Abstract:
Multivariate morphometry has been used to investigate the discreteness
.and interrelationships of stocks with a species. However, there are several biases
and weaknesses inherent to traditional use of morphometric characters for the
purpose. As an alternative a new system of morphomelric measurement called the
Truss network system has been used to differentiate fish species. In the present
study, landmark-based (truss measurements) multivariate morphomelric analysis of
Oreochromis mossambieus (n:100) from Negombo Lagoon and Oreochromis
nifotieus (n : 100) from Seira lake is presented. Twenty morphometrie characters of
these two species were measured. Measurements were then standardized using
. two different methods to remove the size effects. The first was to divide each truss
measurement by standard length. In the second approach, truss measurements
were standardized for fish size using the follOWing equation. Standard measurement of truss length LTs(;)= log,o LT;,) ( )
log,o TLI,)
where TL is the total length, LT(i) is the truss length of ilh fish , TLm is the overall
mean total length and b is the slope, within areas of the geometric mean regression
on the logarithms of LT and TL
Correlation coefficients between each pair of characters were calculated.
According to the analysis, low correlation coefficients were resulted. after removal of
size effect. Multivariate techniques i.e., Principal Component Analysis (PCA) and
Cluster Analysis were performed to analyze transformed and untransfarmed data of
the two species. Two Oreochromis species separate into two groups in the PCA of
transformed data. In cluster analysis, both transformed methods separated
Oreochromis species into two clusters. Nevertheless, the second transformation
method showed greater differences among groups than the first approach.
