Abstract
SPSS model syntax was defined and used to evaluate the individual performance of 49 linear and non-linear models to fit the lactation curve of 159 Murciano-Granadina goats selected for genotyping analyses. Lactation curve shape, peak and persistence were evaluated for each model using 3107 milk yield controls with an average of 3.78 ± 2.05 lactations per goat. Best fit (Adjusted R2) values (0.47) were reached by the five-parameter logarithmic model of Ali and Schaeffer. Three main possibilities were detected: non-fitting (did not converge), standard (Adjusted R2 over 75%) and atypical curves (Adjusted R2 below 75%). All the goats fitted for 38 models. The ability to fit different possible functional forms for each goat, which progressively increased with the number of parameters comprised in each model, translated into a higher sensitivity to explaining curve shape individual variability. However, for models for which all goats fitted, only moderate increases in explanatory and predictive potential (AIC, AICc or BIC) were found. The Ali and Schaeffer model reported the best fitting results to study the genetic variability behind goat milk yield and perhaps enhance the evaluation of curve parameters as trustable future selection criteria to face the future challenges offered by the goat dairy industry.
Highlights
When research involves excessively high costs, researchers may be compulsorily forced to perform sample selection procedures [1]
R2 value of 46.90% in the context of a value of 42.31% for variation coefficient for milk yield, we could infer that ALISCH model may capture all the variability that may be attributed to the evolution of milk yield in time, plus around 4% of linear dependence with other factors
ALISCH model could be considered a parametric complex model, due to the number of regressors that it comprises, our results suggest that the inclusion of logarithmic forms in the formula may somehow promote the adaptation of lactation curves described by each goat individually to the properties of the model, which may result in the improvement in the variability capturing ability of these models when compared to the rest
Summary
When research involves excessively high costs, researchers may be compulsorily forced to perform sample selection procedures [1] These procedures seek to achieve the highest representativity of the population under study in the minimum possible number of effective individuals. The effects of sample size limitation become determinant when traits are obtained after the application of functions to model the trends that such traits describe (as happens in milk yield, composition or growth, among others) In such circumstances, sample size limitations may compromise population representativeness, and may reduce the buffer effect derived from the dilution of the loss of information when larger numbers of individuals are considered. It may not be able to represent the reality of the populations under study
Sign-in/Register to view highlights & summary Sign-in/Register to access full text options 
Talk to us
Join us for a 30 min session where you can share your feedback and ask us any queries you have
Schedule a callDisclaimer: All third-party content on this website/platform is and will remain the property of their respective owners and is provided on "as is" basis without any warranties, express or implied. Use of third-party content does not indicate any affiliation, sponsorship with or endorsement by them. Any references to third-party content is to identify the corresponding services and shall be considered fair use under The CopyrightLaw.
