Recording of frost events and damage
Because of the differences in growth stages among individual plants and between main spike and tillers, spring radiation frost impact can vary significantly between heads. Therefore, precise frost tolerance phenotyping poses a daunting task. In this study, six DH populations were grown in three distinct environments. In total, two locations comprising around 4,000 plots (about 4 ha) suffered from frost damage during the most critical developmental stages. We adopted a visual scoring method to estimate frost damage, commonly used by other researchers (Tóth et al., 2003; Knox et al., 2008; Liu et al., 2014; Sieber et al., 2016; Chen et al., 2019).The entire scoring was carried out by a single person to minimise human biases. As a result, robust QTL detection was achieved, as demonstrated by the same QTL regions showing up consistently in multiple environments and across populations. To further validate our frost scoring, five parental and control lines covering three anthesis windows were used to assess the frost damage using a different scoring method based on floret sterility (Reinheimer et al., 2004; Biddulph et al., 2013). Between 70 to 140 spikes from 7 to 14 randomly selected plots from each line were assessed at the Muresk site, with the results showing the same damage pattern as in our visual scoring.