Simulation Overview
The simulation can be split into one program for AI bull selection, and one program for a single herd simulation, in which the herd is replicated with different initial herd compositions. Results are analyzed separately.
Both programs have a 20 year 'burn-in' phase, followed by a 25 year observation phase. The results from the last 5 years are used in the final analyses.
AI Stud
The AI stud simulation follows a traditional progeny testing scheme, with selection entirely on the Lifetime Profit Index. Four hundred bulls are entered per year and 35% of the cow population is mated to young sires. Fifty proven bulls are kept at all times for breeding the remaining 65% of the cow population in a random manner. The cow population size is 1 million animals. Individual cows are simulated as part of one national herd. Every year the variances and covariances of true genetic values of females born is calculated. Cows are also selected on the basis of LPI. Selection, and the Bulmer effect are known to change genetic correlations.
Bulls available for breeding in each year and month are written to a file with their true breeding values (TBV). Another file of new genetic covariance matrices is created. These files are read during the herd simulation program, as needed.
Events occur on a monthly basis. Herd policies are totally ignored as well as costs and incomes of cows or bulls. The actual amount of genetic change in the bull and cow populations can be documented.
Genomic selection on bulls and cows can be incorporated at a later date.
Herd Program
Parameters for herd size, management practices and decision criteria, and other herd related environmental variables are input to the program. Practices such as hoof trimming, milking parlour design, within herd groups, diets, and bull selection strategy can be varied from run to run.
Policies are kept constant during the 45 year simulation period, and also actual costs and relative costs. Relative costs can be varied between runs. Only direct costs and incomes caused by the cow are recorded (e.g. feed, vet costs, breeding costs, calf sales, milk sales, salvage value). All other costs are assumed to be for the benefit of all cows in the herd and are not due to the cow's genetics. Cows are credited with the potential value of a female calf that may become a herd replacement, surviving for three lactations, proportional to the probability of that event actually happening.
Cows calving in the last five years are written to a file with their TBVs and their accumulated income and expenses in different periods of time.
Activities are monitored on a day-to-day basis within the herd, for each animal. Weather patterns over the last ten years in Ontario are replicated in the simulations, and adjustments for heat and cold stress are included in the simulations. Cows have a genetic ability for resisting stress. Diseases have higher risks in certain weather.