Prunus is one of the largest genera of the family Rosaceae and subfamily Prunoideae (Rehder, 1940). It is economically important as a source of edible and highly appreciated fruits (e.g., peaches, plums, cherries, and almonds) but also as a source of rootstocks for stone fruits production. Modern fruit trees are currently grafted onto a rootstock (Guajardo et al., 2020). The rootstock not only allows scion cultivars to be adapted to unfavorable soil conditions but also determines tree size, vigor, and fruit quality (Font i Forcada et al., 2020). Biotic and abiotic stress tolerance related to soil characteristics is mainly determined by rootstocks (Cummins and Aldwinckle, 1983; Layne, 1987). While new stone fruit scion cultivar development is addressed by many breeding programs around the world, few of them are dedicated to the selection of new Prunus rootstocks (Byrne et al., 2012).

One of the main objectives of these breeding programs for rootstocks is to generate complex hybrids with the maximum number of desirable traits. The new rootstocks should overcome soil and disease problems (Reighard, 2002), showing high-performance and excellent adaption to different regions such as the Mediterranean area. One worldwide abiotic stress presented in these regions is iron chlorosis (Marschner, 1995). This abiotic stress is mainly due to the low iron bioavailability in calcareous soils and the difficulty of iron (Fe) acquisition by the roots (Hell and Stephan, 2003). The selection of rootstocks tolerant to iron chlorosis requires expensive and long-term experiments (Zarrouk et al., 2005; Jiménez et al., 2008). According to Gonzalo et al. (2012), the SPAD (Soil Plant Analysis Development) value seems to be the most reliable measure, and an unbiased and non-destructive method to evaluate the tolerance to iron chlorosis of different genotypes and breeding populations. The selection process can be also accelerated using molecular markers linked to the trait, allowing the early selection of likely seedlings (Etienne et al., 2002; Dirlewanger et al., 2006). Combining SPAD value and low number of molecular markers, Gonzalo et al. (2012) identified different quantitative trait loci (QTL) in linkage groups (LG) 4 and 6 in the almond–peach hybrid “Felinem” (Prunus dulcis × Prunus persica), formerly named “GxN 22” (Felipe, 2009). Although this work allowed localizing the candidate gene PFIT in LG 4, the genetic control of iron chlorosis in Prunus is still unknown. In fact, other important candidates such as PIRT1, PAHA2, and PNramp were not located by the authors (Gonzalo et al., 2012).

At present, new advances in genetic and genomic tools can help to face challenges in both cultivar and rootstocks breeding programs by improving the efficiency and reducing the time necessary to obtain new genotypes. In this sense, high-resolution linkage maps are valuable tools in genetics and breeding because they are the basis for QTLs and the identification of genes associated with traits of interest. For Prunus rootstocks, traditionally linkage maps have been constructed using amplified fragment length polymorphism (AFLPs) markers (Lu et al., 1998), AFLP and single sequence repeat (SSRs) markers (Blenda et al., 2007), and SSRs only (Gonzalo et al., 2012). In the last decade, the use of molecular markers such as single nucleotide polymorphisms (SNPs) has allowed the construction of high-density linkage maps for Prunus cultivars, but their use in linkage maps construction for rootstocks is scarce. For example, mapping populations have been genotyped using SNP chips (Peace et al., 2012; Verde et al., 2012) for the construction of linkage maps in peach (Eduardo et al., 2013; Martínez-García et al., 2013; Pirona et al., 2013; Yang et al., 2013; Frett et al., 2014; Romeu et al., 2014; da Silva et al., 2015, 2018; Donoso et al., 2015; Núñez-Lillo et al., 2015; Fresnedo-Ramírez et al., 2016; Lambert et al., 2016; Zeballos et al., 2016; Serra et al., 2017), sweet cherry (Klagges et al., 2013; Campoy et al., 2015; Calle et al., 2018), and sour cherry (Cai et al., 2018).

More recently, Prunus linkage maps have been constructed using SNPs identified through sequence-based genotyping (SBG, Elshire et al., 2011), a next-generation sequencing methodology that provides a reduced representation of the genome by using single and/or double-digestion of DNA with restriction enzymes to reduce genome complexity and genotype multiple DNA samples (Elshire et al., 2011; Poland et al., 2012). Single digestion using ApeKI restriction enzyme has been used for Prunus linkage maps construction in peach (Bielenberg et al., 2015; Núñez-Lillo et al., 2019), sweet cherry (Guajardo et al., 2015), Japanese plum (Salazar et al., 2017; Carrasco et al., 2018), almond (Goonetilleke et al., 2018), P. mume (Kitamura et al., 2018), and apricot (Pina et al., 2021). Although good genome coverage has been reported using this methodology, it has been described that the double-digest protocol greatly simplifies quantification of the library prior to sequencing and could generate a suitable and uniform complexity reduction of the genome (Poland et al., 2012; Guajardo et al., 2020).

In this study, we used the double-digest SBG for the identification of SNPs and to construct linkage maps for a mapping population derived from the cross “Adafuel” × “Flordaguard,” two Prunus rootstocks. Filtered sequence reads were aligned to the version 2.0 of the Prunus reference genome (Verde et al., 2017). The physical position of each SNP was determined, and the distribution of SNPs along eight peach pseudomolecules was performed to analyze the location of each SNP, identifying SNP-carrying genes and predicting the SNP effect of each SNP. Linkage maps constructed for both peach-based rootstocks were used for the identification of QTLs for iron chlorosis evaluated during two years in the mapping population. The results obtained here will provide a list of genes relevant for Prunus rootstocks breeding.

In this work, a successful strategy covering the discovery of single nucleotide polymorphism (SNP) through double-digest sequence-based genotyping (SBG), genetic linkage maps construction, and QTL identification has been performed in a segregating peach–almond progeny to improve the knowledge about the genetic control of the tolerance to iron chlorosis in Prunus. The physical position and distribution of each SNP along the eight peach pseudomolecules were determined to identify SNP-carrying genes and to predict the effect of each SNP. To our knowledge, this is the first report on linkage map construction and QTL analysis using SNPs obtained from SBG for Prunus rootstocks.

Our results showed excellent performance for double digest-SBG and deep coverage paired-end sequencing for reduced representation of Prunus rootstocks. A high mean depth (1,244) per SNP and high coverage of peach pseudomolecules were reached, which was higher than a coverage of 12.1x obtained by Núñez-Lillo et al. (2019) for map construction in peach using single-digestion SBG. Our results confirm that paired-end sequence reads are more accurately mapped onto the reference genome, showing a significantly lower amount of missing data and a greater number of quality SNPs, as was suggested by Shirasawa et al. (2016) and Guajardo et al. (2020). In total, the SNP density was found to be ~1 SNP per 11.9 Kbp, which can be enough for mapping and QTL identification in an F1 population.

The length of “Adafuel” linkage map, composed by eight linkage groups, is comparable with other maps reported for peach cultivars (Martínez-García et al., 2013; Yang et al., 2013; Núñez-Lillo et al., 2015) and for almond x peach maps (Donoso et al., 2015). A high number of SNPs shared map positions in Ad-LG1, Ad-LG2, and Ad-LG3 (Table 2), probably due to the absence of recombination between some SNPs. On the other hand, and given the low number of SNPs showing paternal segregation (868 SNPs, or 4.6% of the useful SNPs for mapping), it was possible to construct only four linkage groups for “Flordaguard.” It might be due to the origin of this rootstock, which originated from six generations of peach crosses and open-pollinated seedlings (Sherman et al., 1991), generating low levels of heterozygosity in certain regions, or gaps, of the “Flordaguard” genome.

In general, for both “Adafuel” and “Flordaguard” linkage maps, the genetic positions of mapped SNPs are in agreement with the physical position of SNP along the pseudomolecules in the peach genome sequence (Peach v2.0; Verde et al., 2017), and the distribution of SNPs is similar to those observed for the PxF map (Figure 3), which reflect the correction of inter-pseudomolecule misassembles of the Peach v1.0 in the new version of the peach genome sequence.

A high number of skewed markers (p < 0.01) was mapped at the bottom part of Ad-LG5 (Figure 2). Although it has been described that the use of distorted markers could increase the genome coverage of the genetic maps (Zhang et al., 2007, 2013; Cai et al., 2015) and it may be beneficial for QTL mapping (Xu, 2008; Zhang et al., 2010), many authors eliminate the skewed markers and they are not used for linkage maps construction, diminishing the information that it is possible to obtain. It has been proposed that the study of segregation distortion is significant because distorted markers may be linked to genes or traits of interest, and these genes may be beneficial or lethal to organisms (Kirungu et al., 2019). Then, the exclusion of such markers could bias the data and result in the loss of some important genetic information. One example is the reported clustering of loci with distortion near the bottom of LG6 in sweet cherry in semi-compatible crosses, near the Prunus self-incompatibility S locus (Dirlewanger et al., 2004; Klagges et al., 2013; Guajardo et al., 2015). Until our knowledge, this is the first report of clustering of loci with high distortion in the bottom part of the LG5 in Prunus, and additional studies are necessary to determine the biological significance of the skewed markers in this region.

In a preliminary study in Prunus, two different mechanisms associated with tolerance to iron chlorosis were proposed (Gonzalo et al., 2012). The timing to overcome the lack of this essential element is the main difference between both mechanisms. The two linkage groups, LG1 and LG8, were associated with the delayed response observed in the Myrobalan plum P 2175 by Gonzalo et al. (2012). These two linkage groups were also identified in our study. However, QTLs associated with a quick response observed in LG4 and LG6 detected in “Felinem” were not identified in our study. On the other hand, the additional QTLs in LG5 and LG7 identified in “Adafuel” were not previously identified in response to iron chlorosis in this studied genotype.

A total number of 312 distinct genes were affected by significant SNPs associated with iron (Fe) deficiency-induced chlorosis, which has a large impact on yield in tree fruit crops, such as peach (Álvarez-Fernández et al., 2003; Abadía et al., 2011). However, the candidate gene PFIT, related to iron metabolism, and previously identified in Pp04 of Prunus (Gonzalo et al., 2012), was not identified in the candidate genes list obtained here. Peach and almond are considered to be Strategy I plants. The reduction-based strategy (Strategy I) has two main steps, the first step is the acidification of the rhizosphere that leads to an increase in the chelated Fe (III) concentration (Santi and Schmidt, 2009; Ivanov et al., 2012).

Subsequently, the root surface-localized ferric chelate reductase FERRIC-REDUCTION OXIDASE2 (FRO2; Robinson et al., 1999) reduces Fe(III) to soluble Fe(II), which is then taken up into epidermal cells by the Fe-REGULATED TRANSPORTER1 (IRT1; Eide et al., 1996). In our study, although FRO2 was not identified, two Ferric-chelate reductase genes [Prupe.1G587800 (FRO7) and Prupe.5G166400 (FRO08)] were affected by significant SNPs. Interestedly, both genes are located in different peach pseudomolecules. According to the literature, it is possible that once Fe³⁺-citrate complexes pass to the chloroplastic intermembrane space, FRO7 may reduce ferric (Fe³⁺) to ferrous iron (Fe²⁺) via its reductase activity (Jeong et al., 2008; Solti et al., 2016). A possible role of FRO7 in the reduction of Fe (III) to soluble Fe (II) in root Fe uptake responses to Fe-deficiency in peach should be investigated to clearly confirm this hypothesis.

Several studies using commercially available Affymetrix ATH1 GeneChips (Dinneny et al., 2008; Yang et al., 2010; Mai et al., 2016) and RNA-seq data set (Lan et al., 2013) have been carried out to study Fe deficiency and the involvement of ethylene in Strategy I plants. As a result, a total of 71 differentially expressed genes (61 being up-regulated and 10 down-regulated) overlapped between these studies. In the list of up-regulated genes was AT3G12900, which encodes a protein belonging to the 2OG-Fe(II) oxygenase family and is strongly induced by Fe deficiency. Proteins in this family are generally considered to possess oxidoreductase activity catalyzing the 2-oxoglutarate- and Fe(II)-dependent oxidation of an organic substrate using a dioxygen molecule. A previous study showed that the ethylene synthesis protein ACC oxidase also belongs to 2OG-Fe(II) oxygenase superfamily (Aravind and Koonin, 2001). In our study, the gene Prupe.1G468200 is located in Pp01 of peach and it was affected by eight significant SNPs mapped in this pseudomolecule and showing different effects, low, moderate, and modifier. This gene is an ortholog of AT1G52820 and encodes a 2-oxoglutarate (2OG) and Fe(II)-dependent oxygenase superfamily protein. The possible expression of this gene, by Fe deficiency, should be strongly verified in Prunus (peach or almond).

Another candidate gene was Prupe.1G541100 (ppa003848m.g), predicted as glutamyl-tRNA reductase 1, which had a modifier (3'-UTR variant) effect by a significant SNP located in Pp01 (44,202,463 bp) and mapped at 63.65 cM. The study of the transcriptional response of Arabidopsis to Fe deficiency has indicated that glutamyl-tRNA reductase (HEMA1), NYC1, and CGLD27 may represent key players in preventing photooxidative damage in Fe-deficient leaf cells (Rodríguez-Celma et al., 2013). The fast and pronounced down-regulation of HEMA1 expression suggests a rapid and dramatic decline of tetrapyrrole synthesis at an early stage of Fe deficiency. The possible implication of Prupe.1G541100 preventing photooxidative damage in Fe-deficient leaf cells in peach must be clearly confirmed.

Other important genes detected in our study were Prupe.5G039500 and Prupe.1G577000 (ppa011050.m), encoding a ferredoxin protein (FdC2) and a NIFU-like protein 2 (NIFU2), respectively. The interaction of these both genes in peach and their association with iron chlorosis response should be corroborated in further analysis. In this sense, in Arabdopsis, the study of the physiological and molecular changes during Fe deficiency confirmed that at the protein level, ferredoxin, the cytochrome-b6f complex, and Fe-containing enzymes of the plastid sulfur assimilation pathway were major targets of Fe deficiency, whereas other Fe-dependent functions were relatively less affected (Hantzis et al., 2018).

In the present study, we have constructed, to the best of our knowledge, the first Prunus rootstocks linkage maps using SNPs obtained from double-digest SBG. These linkage maps were the basis to identify QTLs and putative genes related to iron chlorosis in a segregating peach–almond progeny. SNPs identified in this study present a valuable set of new SNPs identified in Prunus that would be useful for genetic studies in the future. Candidate genes for iron chlorosis QTL were proposed; however, future fine mapping and transcriptomic analysis are needed to enable the identification of the underlying gene(s).

The dataset generated for this study can be found in the NCBI-SRA database, BioProject number PRJNA773812.

VG, PM-G, and MM conceived and designed the experiments. VG, PM-G, and SS performed the experiments. VG, SS, CS, PM-G, and AC-S analyzed the data. CS and MM contributed with reagents, materials, and analysis tools. All authors contributed to the manuscript preparation, read, and approved the final version.

This research was funded by the Spanish Ministry of Science, Innovation and Universities (MICINN) grants RFP2015-00019 and RTI2018-094176-R-C33, co-funded by ERDF A way of making EUROPE and the Regional Government of Aragon (A44 and T07-17R). It was also funded by CONICYT-REGIONAL/GORE O'HIGGINS/CEAF/R19A10003, FONDECYT 3160316, and CONICYT R16F20006 from Chile.

The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.

All claims expressed in this article are solely those of the authors and do not necessarily represent those of their affiliated organizations, or those of the publisher, the editors and the reviewers. Any product that may be evaluated in this article, or claim that may be made by its manufacturer, is not guaranteed or endorsed by the publisher.