Table 3.

Sources of variation in the number of sperm ejaculated by a male in the sequential experiment. ((a) The overall dataset (n_males=27) and (b) Swedish population only (n_males=19). AIC values were used to select the best model, with smaller AIC indicating a better-fit model. The difference between the two AIC values was considered insignificant when equal to or less than 1. Values in bold represent p-values <0.05.)

model	AIC	d.f.	F	p-value
(a) the overall dataset (nmales=27)
total MHC alleles shareda	2983.92	1, 11	5.82	0.035
MHC haplotypes sharedb	2982.87	1, 11	6.97	0.023
comb size	2957.91	1, 11	0.77	0.4
female mass	2942.52	1, 11	2.77	0.124
(b) Swedish population only (nmales=19)
relatedness	2075	1, 4	1.19	0.336
total MHC alleles sharedc	2079.68	1, 4	1.24	0.328
MHC haplotypes shared	2078.68	1, 4	2.36	0.199
comb size	2093.01	1, 4	0.04	0.843
female mass	2077.82	1, 4	1.21	0.333

^aWhen comparing AIC values between the models no other alternative MHC similarity measures, MHC class I alleles shared, MHC class II alleles shared, MHC major alleles shared, MHC major I alleles shared and MHC major II alleles shared were a significantly better fit in explaining variation in number of sperm ejaculated when using the overall dataset.

^bMHC haplotypes shared explained significantly more of the variance in the number of sperm ejaculated than all other MHC similarity measures, MHC class I alleles shared, MHC class II alleles shared, MHC major alleles shared, MHC major I alleles shared and MHC major II alleles shared, when comparing AIC values and using the overall dataset.

^cAlternative MHC similarity measures, MHC class I alleles shared, MHC class II alleles shared, MHC major alleles shared, MHC major I alleles shared and MHC major II alleles shared did not significantly predict variation in the number of sperm when restricting the analysis to the Swedish population.