Figure 1.
Ordinal estimated breeding values and standard errors from a multi-threshold mixed-model analysis of BVAHTs of a cohort of Australian German Shepherd Dogs.
Table 1.
Average estimated breeding values (EBV) and average standard errors (SE) for Australian German Shepherds with and without a set of BVA/KC hip phenotypes available.
Figure 2.
Relationship between ordinal EBVs and their standard errors.
Animals with phenotypic scores are shown in blue, animals without phenotypic scores are shown in red, NORB = Norberg Angle, SUBL = Subluxation, CrAE = Cranial Acetabular Edge, DAE = Dorsal Acetabular Edge, CrEAR = Cranial Effective Acetabular Rim, AF = Acetabular Fossa, CaAE = Caudal Acetabular Edge, FHNE = Femoral Head and Neck Exostosis, FHR = Femoral Head Remodelling.
Figure 3.
Boxplots of EBV accuracies for BVAHTs for a cohort of Australian German Shepherd Dogs.
NORB = Norberg Angle, SUBL = Subluxation, CrAE = Cranial Acetabular Edge, DAE = Dorsal Acetabular Edge, CrEAR = Cranial Effective Acetabular Rim, AF = Acetabular Fossa, CaAE = Caudal Acetabular Edge, FHNE = Femoral Head and Neck Exostosis, FHR = Femoral Head Remodelling.
Table 2.
Accuracies of phenotype-only selection compared with average accuracies from EBVs obtained by the ordinal model (See Table 1 for abbreviations).
Figure 4.
Correlation between EBVs for nine BVAHTs calculated by ordinal logistic regression.
NORB = Norberg Angle, SUBL = Subluxation, CrAE = Cranial Acetabular Edge, DAE = Dorsal Acetabular Edge, CrEAR = Cranial Effective Acetabular Rim, AF = Acetabular Fossa, CaAE = Caudal Acetabular Edge, FHNE = Femoral Head and Neck Exostosis, FHR = Femoral Head Remodelling.
Table 3.
Correlation between EBVs calculated by binary logistic regression with cut-points introduced between different scores and EBVs calculated by the ordinal model (See Table 1 for abbreviations).
Figure 5.
A- Linear EBVs (x) vs ordinal EBVs (y) for Femoral Head and Neck Exostosis (FHNE; a Group 2 trait) and B- fitted regression values(x) vs standardised residuals(y).
Note that although the correlation is high (0.96) and the relationship on the left appears approximately linear; it is clear from the graph on right that the assumptions of a simple linear regression are not met.
Table 4.
Pearson correlation between EBVs using a linear mixed model (LMM) and an ordinal model for nine BVAHTs in Australian German Shepherd Dogs (See Table 1 for abbreviations).
Table 5.
Correlation of estimated breeding values calculated by a linear mixed model on summed BVAHTs with EBVs calculated by an ordinal model on individual BVA/KC hip phenotypes in a cohort of Australian German Shepherd Dogs (See Table 1 for abbreviations).
Figure 6.
Trend of EBVs by year of birth expressed in terms of their effects on the observed BVAHT scale for a cohort of Australian German Shepherd Dogs born from 1980–2005.
Proprtions for each score are derived from an ordinal model analysis of observed BVAHT scores. Increasing proportions in lower scores represent a genetic improvement over time.