Introduction
Frost can cause significant grain yield and quality losses in wheat crops (Crimp et al., 2016). In spring, when plants sense the gradual increase of temperature and their development proceeds beyond the jointing stage, both winter and spring types show considerable sensitivity to low temperatures (0-12°C) and frost (<0°C), particularly to short chilling and frost events at night (Powell et al., 2012). The early-flowering phenotype in modern wheat and barley cultivars has resulted in significant grain yield and quality losses from frost damage (Reinheimer et al., 2004). Frost damage during reproductive stage can lead to a multiplicity of symptoms, including dead stems, floret and spikelet abortion, and empty shells along the spikes, thus significantly reducing the seed number per spike.
Many past wheat frost tolerance studies focused on the vegetative development stages (Hayes et al., 1993; Galiba et al., 1995; Snape et al., 2001; Sutka, 2001; Limin & Fowler, 2002; Francia et al., 2004). Genetic segregation for vegetative frost tolerance (or susceptibility) have been reported on chromosomes 5A, 5B, 5D, and 7B in wheat, and 5H in barley. The frost tolerance QTL Fr-A1 and Fr-B1 on chromosomes 5A and 5B were closely linked with the vernalisation genesVrnA1a and VrnB1a (Snape et al., 2001; Tóth et al., 2003).Frost resistance 2 (FR2 ) genes are in control of delayed heading. In combination with VRN1 showed reproductive frost tolerance (Eagles et al., 2016), it is not clear yet whether these frost tolerance loci are identical with VrnA1a and VrnB1a(Galiba et al., 2009). Similarly, it remains unclear whether the frost locus on chromosome 7B is influenced by VRN3 (VrnB4 ) (Hayes et al., 1993). In barley, three Doubled haploid (DH) populations were used to identify reproductive stage frost tolerance QTL (Reinheimer et al., 2004). Two major QTL were identified on chromosomes 2H and 5H. The QTL for frost-induced floret sterility and grain damage overlapped with the growth QTL on the Vrn-H1 locus in all three DH populations, whereas in two of the populations the floret sterility QTL on 2H was not close to the growth QTL or Ppd-H1 , but close to theearliness per se gene (Eps 2 ), the cold-regulated gene (Cor14b ) and the barley low-temperature gene (Blt14 ) loci. It seems that the frost tolerance QTL are closely associated with the vernalisation genes in both vegetative and reproductive stages, although the effects kick in at different growth stages.
Numerous eps and the related flowering-time QTL in wheat have been mapped to chromosomes 1DL, 2B, 3A, 4A, 4B and 6B (Scarth & Law, 1983; Hoogendoorn, 1985; Zikhali et al., 2014). An ortholog to theArabidopsis thaliana LUX ARRHYTHMO/PHYTOCLOCK1 (LUX/PCL1 ) gene was identified as Eps-3Am in einkorn wheat (Triticum monococcum L.). Lines containingEps-3Am showed a distorted circadian clock, spikelet number variation and temperature sensitivity (Gawroński et al., 2014).
Apart from the well-known anthesis-related genes mentioned above, there is a set of T . aestivum GIGANTEA (TaGI ) encoding genes, whose products interact with FLAVIN-BINDING, KELCH REPEAT, and F-BOX 1 (FKF1) domains to form a complex regulating photoperiod-dependent flowering by regulating CONSTANS (CO) expression (Zhao et al., 2005).
A SOC1 (Suppressor of Overexpression of CO 1 )-like gene on chromosome 4DL, WSOC1 , was reported to influence flowering time in wheat (Shitsukawa et al., 2007). A gene for wheat vegetative to reproductive transition on the chromosome 7 group, TaVRT-2 , interacts with VRN1 and VRN2 and regulates the floral transition (Kane et al., 2005). In wheat, homologs to ArabidopsisEarly flowering 3 (ELF3 ) gene have been identified on chromosome group 1 (Wang et al., 2016). Three short-day flowering-time genes on 1B, including flowering locus T3 (TaFT3-B1 ), WUSCHEL-like (TaWUSCHELL-B1 ) and TARGET OF EAT1(TaTOE1-B1 ) have been cloned (Zikhali et al., 2017), with the early-flowering function for TaFT3-B1 having been validated. A set of heading-date genes (TaHD1 ) identified on group 6, are regulated by long-day conditions and the circadian clock, directly affecting vernalisation genes under long-day conditions. Its mutants showed a delayed flowering response in a long-day environment (Nemoto et al., 2003; Shi et al., 2019).
It is difficult to screen frost tolerance in the field at the reproductive stage, since trials need to be hit by a natural frost event at the right developmental stage, which is purely a matter of chance. To increase the chance, trials consisting of a wide range of genotypes usually need to be planted on multiple sowing dates (Reinheimer et al., 2004; Biddulph et al., 2013). The type of damage to the affected plants will also be influenced by the weather on days leading to the frost event and after. This goes a long way in explaining why untangling the complex genetic basis of frost tolerance under controlled conditions in glasshouses and cold chambers has been of limited utility to breeders.
Our 2018 large-scale field trials encountered such a chance event exactly at the right stage, which provides a valuable resource for QTL detection for frost damage. Six DH populations representing genetically divergent origins were impacted by frost in two distinct environments. All populations suffered considerable frost damage. Genes and markers contributing to the frost tolerance or susceptibility phenotype were identified. Frost tolerance segregation patterns in the current study provide valuable genetic information that can be used in wheat frost tolerance breeding.