Heterozygosity-fitness correlations within inbreeding classes: local or genome-wide effects?

Marker-based studies of inbreeding may lead to an enhanced understanding of inbreeding depression in natural populations, which is a major concern in conservation genetics. Correlations between marker heterozygosity and variation in fitness-associated traits-'heterozygosity-fitness correlations' (HFCs)-are of particular importance and have been widely applied in natural populations. In partially inbred populations, HFCs can be driven by selection against inbred individuals and thus reflect inbreeding depression. However, other explanations for HFCs also exist, such as functional effects of the markers per se or that the markers reveal selection on linked fitness genes due to extended linkage disequilibrium (LD) in the population. Accordingly, HFCs do not only arise in partially inbred populations, they may also occur within inbreeding classes such as families, i.e. in situations when there is no variation in the inbreeding coefficient. In this study we focus on the importance of LD for within-family HFCs, thereby aiming at enhancing our general understanding of HFCs. For non-coding markers, within-family HFCs have been proposed to be caused in two ways: either by 'local effects' at linked fitness genes in LD with the markers, or by 'general effects' due to a correlation between proportion of heterozygous markers (H-M ) and heterozygosity at genome-wide distributed fitness genes (H-GW ). To evaluate these contrasting hypotheses for within-family HFCs, we analysed simulated data sets of sexually reproducing populations with varying levels of LD. The results confirmed that segregation induces variation in both H-M and H-GW at a fixed level of inbreeding; as expected, the variation in H-M declined with increasing number of markers, whereas the variation in H-GW declined with decreasing LD. However, less intuitively, there was no positive correlation between the variation in H-M and H-GW within inbreeding classes when the local component of H-GW was accounted for (i.e. when the part of the chromosome in LD with the markers was excluded). This strongly suggests that within-family HFCs are not caused by general effects. Instead, our results support the idea that HFCs at a known level of inbreeding can be driven by local effects in populations with high to moderate LD. Note however that we define the local component of H-GW as the part of the chromosomes in LD with the markers. This implies that when LD is high, the local component will consist of a substantial part of the genome and thus provides a rather 'genome-wide' view. We caution against routinely interpreting positive HFCs as evidence of inbreeding depression and non-significant HFCs as lack thereof, especially when few markers are used.