Last update: August 18, 2011 02:22:01 PM E-mail Print


Ewe Productivity - (1)

Use of a mixed model approach.


G J Delport, A O de Lange1, G J Erasmus2 and J J Olivier3.

S A Fleece Testing Centre, ADSRI, Middelburg Cape.



A scheme for measuring ewe productivity was implemented in 1983 by the National Performance and Progeny Testing scheme for Wool led Sheep. It was devised to provide a method of selection for higher mutton production from Merino ewes without decreasing wool production. In order to apply this method a multiple trait mixed model was used on total mass of lamb weaned and fleece mass of the ewe. Both traits were measured for two consecutive lambing seasons and therefore resulted in four observations on each ewe. Since computations proved demanding, canonical transformations were used to reduce the multiple trait model to four single trait models. This resulted in the following, general model specification in matrix notation: Yi* = Xbi* + Zui* + ei*.

Where Yi*= a data vector of the i-th canocical trait

bi*= a vector of fixed effects

ui*= a vector of random effects (breeding values)

X,Z= Incidence matrices associated with bi* and ui* respectively. Total mass of lamb weaned is a non-normally distributed trait which causes difficulties. Employing the model above resulted in accurate breeding values for both total mass of lamb weaned and fleece mass of the ewe. Results presented indicate normally distributed breeding values for total mass of lamb weaned. Indications are that these breeding values can be combined into a single measure for ewe productivity which is both practical and feasible.

(1) ARDRI, University of Fort Hare, Alice, Ciskei.

(2) University of the OFS, Bloemfontein.

(3) Karoo Region, Middelburg Cape.



Proceedings 28th SASAP congress