Longitudinal genomic analyses of automatically-recorded vaginal temperature in lactating sows under heat stress conditions based on random regression models

Table 1 Number of model parameters (N), knot position, log likelihood (Log L), Bayesian information criterion (BIC) for random regression models fitted to longitudinal Tv data using B-splines functions or Legendre orthogonal polynomials

Model^a	N	Knot position^b	Log L	BIC
B-spline (Linear)
BSL55	31	5, 363, 721, 1079, 1435	− 419,827	840,079
BSL66	43	5, 291, 577, 863, 1149, 1435	− 397,465	795,520
BSL77	57	5, 243, 481, 719, 957, 1195, 1435	− 387,717	776,218
B-spline (Quadratic)
BSQ66	43	5, 363, 721, 1079, 1435	− 392,435	785,460
BSQ77	57	5, 291, 577, 863, 1149, 1435	− 386,227	773,237
BSQ88	73	5, 243, 481, 719, 957, 1195, 1435	− 384,591	770,186
B-spline (Cubic)
BSC77	57	5, 363, 721, 1079, 1435	− 389,409	779,602
BSC88	73	5, 291, 577, 863, 1149, 1435	− 385,532	772,006
BSC99	91	5, 243, 481, 719, 957, 1195, 1435	NC	NC
Legendre orthogonal polynomials^c
LEG2	7		− 581,803	1,163,702
LEG3	13		− 440,933	882,045
LEG4	21		− 430,771	861,831

NC: the analyses did not converge
^ain italics is indicated the best model based on BIC
^btime points of the day (in minutes) where the knots were placed
^cLegendre orthogonal polynomial of order X (with X = 2, 3, or 4) for the additive genetic and permanent environmental effects

ISSN: 1297-9686