For efficient artificial insemination (AI) it is essential to understand the relation between semen quality, quantity characteristics and fertility. For an AI company it is important to monitor its results in practice. Knowing the performance of semen in the field and keeping track of semen processing results at individual AI stations can result in a tool to control efficient semen production and processing.
Routine AI laboratory practice of AIM Varkens KI Nederland (Vught, the Netherlands) created a database of 821,928 processed ejaculates (January 2007 to January 2019), which is continuously updated. The data originated from 9 AI locations within the Netherlands, analyzing semen results from 12,147 individual boars (on average 70 ejaculates per boar), originating from 6 dam lines and 5 sire lines. Semen traits like concentration, volume, motility and morphology were routinely collected. Furthermore, the data also included information related to the boar (e.g., the amount of days a boar rested from the previous ejaculation [collection interval], the age and body temperature of the boar at collection time, and the pedigree of the boar), and information related to the AI station (e.g. barn, laboratory and technicians).
A linear mixed model was implemented in ASReml software. Univariate analyses were performed to estimate the variance components of selected traits, e.g. number of cells per ejaculate (NC) and fresh sperm motility (MOT, measured by computer assisted semen analysis [CASA]). Repeatability estimates were 0.25 and 0.53 for NC and MOT, respectively. The phenotypic variance explained by the permanent environmental effect (13% for NC and 37% for MOT) was higher than the phenotypic variance explained by the additive genetic effect (12% for NC and 17% for MOT). The variance explained by the collection interval, age of the boar at the collection time, line of the boar, year and week of collection, AI station, barn and lab technicians, and the interactions among them, all together, were 28% and 10% of the phenotypic variance. The scrutiny of the latter effects revealed large differences in certain classes.
Availability of a large and growing database allows the estimation of non-animal and of animal effects. The first can be used to learn and optimize between and within AI stations. The latter can be used to improve the development phase of boars prior to AI, since permanent environmental effects are high, especially for motility. Such a data management tool improves the efficiency and reliability in the production of AI doses, which is beneficial for the AI companies and their customers aiming to use AI boars with high genetic merit most efficiently.