Fig. 5From: A Bayesian generalized random regression model for estimating heritability using overdispersed count dataVariance components estimated from log transformed faecal egg counts using WOMBAT. (a) Additive genetic (AG, brown), permanent environmental (PE, green) and maternal (MAT, pink) variance components for a full rank analysis with two Legendre polynomials and two eigenfunctions (K=M=2, solid curves) and for a reduced rank analysis with one eigenfunction (K=2, M=1 dotted curves). (b) Additive genetic (AG, brown), permanent environmental (PE, green) and maternal (MAT, pink) variance components for a full rank analysis with three Legendre polynomials and three eigenfunctions (K=M=3 solid curves) and for a reduced rank analysis with two eigenfunctions (K=3, M=2 dotted curves). (c) Additive genetic (AG, brown), permanent environmental (PE, green) and maternal (MAT, pink) variance components for a full rank analysis with four Legendre polynomials and four eigenfunctions (K=M=4 solid curves) and a reduced rank analysis with two eigenfunctions (K=4, M=2 dotted curves). The transformation used was log(FEC+1). The model with four Legendre polynomials and two eigenfunctions (C, dotted lines) provided the best fit to the data under model selection based on AIC c Back to article page