Racing data on 44,372 Standardbred trotters in Sweden, spanning a period of
nineteen years, were analyzed. Racing speed of any i-th horse was measured
as the best average racing time (k(1) = sec/km) obtained in a trotting rac
e of a length greater than or equal to 1640 m, as a 3- to 5-year-old. The d
istribution of the best average racing time records was found to be asymmet
ric within the population. A scaled logarithmic function of best average ra
cing time of male horses (y(1) = ln(k(i) - 68.2)) was found to be normally
distributed (zero skewness and low kurtosis). This means that best average
racing time records can be expressed as: k(i) = e(y(1)) + x, where the cons
tant x = 68.2 sec/km was interpreted as the asymptotic limit for trotting s
peed in the population of male Standardbred trotters. An equation Kbirth-ye
ar = x(1 + e(-pt)) was fitted to the data for estimating trends in average,
minimum and maximum best average racing time records at time t = (birth-ye
ar - z) in Swedish Standardbred trotters. The constants p and z were estima
ted by least-squares grid search. The trend in the average of best average
racing time in male trotters can be predicted by the following expression:
AverK(birth-year) = 68.2(1 + e((-0.015 (birth-year-1861)))). The correspond
ing prediction of the fastest racing time records is: MinK(birth-year) = 68
.2(1 + e((-0.019 (birth-year - 1853)))). The log linear scale effects may b
e interpreted as successive reductions in marginal substitution effects of
genes and environmental factors affecting racing time as measured on the or
iginal scale. (C) 2001 Elsevier Science B.V. All rights reserved.