Table 1.

Parameter estimates under the Gaussian and binomial-like ATOMM models for the A. thaliana–X. arboricola pathosystem

Parameter estimate under the null	Gaussian ATOMM estimate (SE)	Binomial-like ATOMM estimate (SE)
Intercept, ${\beta }_0$	0.19 (0.011)	-0.75 (0.013)
Person 1, ${\beta }_1$	0.15 (0.015)	0.16 (0.015)
Person 2, ${\beta }_2$	0.20 (0.015)	0.19 (0.015)
Total residual variance, ${\sigma }_t^2$	1.23	1.45
Proportion of residual variance due to
A. thaliana, ${\xi }_h$	0.021	0.011
X. arboricola, ${\xi }_p$	0.441	0.581
A. thaliana–X. arboricola interaction, ${\xi }_{hp}$	0.048	0.011
Plant/block effect, ${\xi }_J$	0.093	0.069

We included in ATOMM the person effects (i.e., effects due to which lab researcher scored the QDR) as covariates. There were three persons measuring the trait scores. One person effect is subsumed into the intercept, leaving us to estimate effects of person 1 and person 2 only. In addition to the four genetic VC_s described in Eq. 3, we also included the plant effect as arandom effect (see Materials and Methods). The parameters were estimated under the global null hypothesis ${\mathcal{H}}_0:{\gamma }_h={\gamma }_p={\gamma }_{hp}=0$.