I am currently a graduate student working on the stock discrimination of
the Australian pilchard. I am using genetic criteria as well as
morphometric data. However, I am unsure about the type of morphometric
data to collect. I initially thought that it would be worthwhile to collect
landmark data for truss network analyses as well as do some conventional
measurements then analyse that using discriminant and principal component
analyses. However, Bookstein (1990: Morphometric Tools for Landmark Data)
(one of the founders of the truss network) disputes the validity of using
discriminant analysis and discounts the use of truss analysis and
recommends using "Centroid Size and shape coordinates". As yet, I have not
been able to source any papers using these methods on fish. (I am still
waiting for the "Proceedings of the Michigan Morphometrics workshop").
I personally have misgivings about using the truss network because
pilchard shape varies with gonad development and spawning seasons differ
throughout Australian waters, so I thought that I would be bound to find
both temporal and spatial variation. Hence, I am confused about which
method is appropriate to use. Does anyone out there have any
experience/opinions/advice on these issues or know of any software
available to analyse data using the centroid size method? I would
appreciate any replies to be sent to my e-mail address viz:
L.Worland at unsw.EDU.AU
Thanks,
Linda W.
L.Worland at unsw.EDU.AU
