Fig 1.
Murciano-Granadina does on field and herd and in the milking room.
Fig 2.
Murciano-Granadina goat breed distribution across the Spanish territories and Location of Caprigen (Goat Biotechnology Center) in Fuente Vaqueros, Granada (Spain) (Navas González, 2026).
The map is an original illustration created by the author for this study, edited in Microsoft PowerPoint (Microsoft 365) and rendered using Paint.
Fig 3.
Frequency distributions of fertility percentages by semen type (fresh/chilled vs. frozen/thawed), day of insemination, and buck batch combined with insemination day.
Histograms depict empirical fertility frequencies, with overlaid kernel density estimates and fitted Gaussian curves illustrating distributional trends. The unimodal shape and approximate symmetry of the distributions indicate fertility values approaching normality, supporting the definition of fertility categories based on empirical distribution tails and central ranges.
Fig 4.
Analytical workflow of the study.
Schematic overview of the analytical framework, from data acquisition and preprocessing through confirmatory multivariate analysis (rCCA), supportive analyses (CDA, CHAID), validation, and biological interpretation of fertility–lactation relationships.
Table 1.
Descriptive statistics (mean, standard deviation (SD), Standard error of the mean (SEM), maximum, minimum and variance (α2) fertility related parameters across semen type of the doses with which inseminations were performed and unstandardized and standardized milk yield and composition parameters at 150, 210, 240 and 305 days.
Fig 5.
Scree plots of canonical functions.
The top panel shows results from regularized canonical correlation analysis (rCCA), where fertility rates were analyzed as continuous variables together with milk yield and composition. The bottom panels show canonical discriminant analyses (CDA), where fertility rates were treated as categorical clustering variables (very low to very high) for three contexts: fertility per day of insemination, fertility per buck batch and day of insemination, and fertility per day by semen type. In all cases, the first canonical function (F1) explained the majority of variance, with higher-order functions contributing minimally.
Table 2.
Canonical discriminant analysis efficiency parameters showing the significance of each canonical discriminant function based on Bartlett’s test.
Fig 6.
Vector plot for discriminant loadings for fertility traits for regurlarized Canonical Correlation analyses (rCCA) and for Canonical Discriminant Analyses (CDA).
Fig 7.
Hierarchical clustering dendrograms based on Mahalanobis distances for three fertility rate measures:
(A) Fertility per Day of Insemination, (B) Fertility per Buck Batch and Day of Insemination, and (C) Fertility per Day by Semen Type.In all cases, two distinct clusters emerge, separating Very Low/Low fertility from Medium/High/Very High fertility. The degree of separation varies by measure, with the widest divergence observed for semen type (final merge at ~0.60) and the most compact structure for buck batch and insemination day (final merge at ~0.14).
Table 3.
Pearson’s correlations between unstandardized and standardized milk yield and composition trait pairs after VIF variable dimensionality reduction. Color scale ranges from green (maximum positive value) to red (maximum negative value).
Table 4.
Pearson’s correlations between fertility and unstandardized and 150 days standardized milk yield and composition parameter pairs after VIF variable dimensionality reduction. Color scale ranges from green (maximum positive value) to red (maximum negative value).
Table 5.
Canonical correlations and Redundancy coefficients.
Fig 8.
Optimal cross-validation scores for the values of the parameter of regularization (λ1 and λ2).