Mango (Mangifera indica L., 2n = 40) is a major fruit crop of the tropics and subtropics, and is often called the “king of fruits” (Mukherjee, 1950, 1957; Purseglove, 1972). Mango has been referred to as an allopolyploid (Mukherjee, 1950), but the recent microsatellite marker studies of Duval et al. (2005), Viruel et al. (2005), and Schnell et al. (2005) indicate that M. indica is diploid. Although mango has a high level of heterozygosity, it has a relatively small genome size (approximately 439 Mb; Arumuganathan and Earle, 1991). In recent years, the area under mango plantation has expanded unceasingly worldwide, and mango yields have increased year by year. Currently, more than 100 countries and regions worldwide produce mango, and China is one of the major mango-producing countries. The mango industry has become the mainstay of the local agricultural industry in most of the mango production provinces in China.

Mango breeding has gained increasing attention in recent years, and conventional breeding methods (cross breeding, mutation breeding, and seedling selection) are mainly used for creating new varieties of mango. However, these methods are time-consuming and difficult to improve upon because of some inherent characteristics of mango (Iyer and Schnell, 2009). Many important economic traits of mango are controlled by a number of loci. Quantitative trait locus (QTL) analysis and marker assisted selection (MAS) can shorten selection times and accelerate the breeding process of new varieties of mango. Genetic maps, especially high-density genetic maps, are particularly valuable tools for QTL mapping and MAS. Four genetic maps of mango have been generated to date. Chunwongse et al. (2000) first constructed a genetic map of mango using 197 restriction fragment length polymorphism (RFLP) and 650 amplified fragment length polymorphism (AFLP) markers based on 31 F₁ plants from a cross between Alphonso and Palmer cultivars of mango (M. indica). Kashkush et al. (2001) developed the second genetic map of mango using 34 AFLP markers based on 29 F₁ individuals and covering a length of 161.5 cM with an average marker spacing of 4.75 cM. Fang et al. (2003) developed the third genetic map of mango using 81 AFLP markers based on 60 F₁ individuals and covering a length of 354.1 cM with an average marker spacing of 4.37 cM. Chunwongse et al. (2015) constructed a partial genetic linkage map spanning a distance of 529.9 cM and consisting of 9 microsatellite and 67 RFLP markers using the same plant materials as in their previous report (Chunwongse et al., 2000). However, these maps were developed on the basis of either a small mapping population (60 or fewer) or a limited number of markers, resulting in a relatively low-density genetic map for future QTL analysis.

DNA markers such as AFLP, RFLP, random amplification of polymorphic DNA (RAPD), inter simple sequence repeat (ISSR), sequence-related amplified polymorphism (SRAP), and simple sequence repeat (SSR) have been developed in mango. However, the current number of markers is too small to build a high-density genetic map. Specific-locus amplified fragment sequencing (SLAF-seq) technology is an efficient method of de novo single nucleotide polymorphism (SNP) discovery and large-scale genotyping, which is based on reduced-representation library (RRL) and high-throughput sequencing (Sun et al., 2013). The efficiency of SLAF-seq was tested using rice and soybean data, and has been used to construct the highest-density genetic map of common carp, without a reference genome sequence. To date, SLAF-seq has been used successfully to construct high-density genetic maps and study the genomes of many crops (Chen et al., 2013; Huang et al., 2013; Zhang et al., 2013, 2015; Li et al., 2014; Qi et al., 2014; Wei et al., 2014; Guo et al., 2015; Jiang et al., 2015; Xu et al., 2015; Zhu et al., 2015).

We conducted a hybridization breeding study of mango 10 years ago and established many segregating populations. Following several years' field investigation, we selected an intraspecific cross of M. indica for constructing its genetic linkage map. We employed the recently developed SLAF-seq approach to identify a large number of SNP markers for mango and thereby developed a high-density genetic map of mango. The characteristics and value of this genetic map were analyzed and discussed.

The development of abundant and reliable molecular markers was very important for the construction of a genetic map. In the present study, SNP markers were used for map construction. SLAF-seq technology is an efficient method of de novo SNP detection and large-scale genotyping, which is based on RRL and high-throughput sequencing (Sun et al., 2013). The markers developed using SLAF-seq technology were of relatively higher density, better consistency and effectiveness, and lower cost than those developed using traditional methods. Since the development of SLAF-seq technology, it has been applied in many plant studies and has produced remarkable results. Zhang et al. (2013) employed SLAF-seq technology to detect 71,793 high-quality SLAFs, of which 3673 were polymorphic; they used 1233 of the polymorphic markers to construct the first high-density genetic map for sesame. Qi et al. (2014) applied SLAF-seq to develop 12,577 polymorphic SLAFs, and used 5308 of them to construct a soybean genetic map. Similar studies have been conducted on many other crops, such as wax gourd (22,151 polymorphic SLAFs; Jiang et al., 2015), tea plant (25,014 polymorphic SLAFs; Ma et al., 2015), walnut (49,174 polymorphic SLAFs; Zhu et al., 2015), mei (93,031 polymorphic SLAFs; Zhang et al., 2015), grapevine (42,279 polymorphic SLAFs; Guo et al., 2015), red sage (62,834 polymorphic SLAFs; Liu et al., 2016), and cucumber (15,946 polymorphic SLAFs; Zhu et al., 2016). In this study, 66.02 Gb of raw data containing 330.64 M reads were generated through SLAF-seq. There were a total of 318,414 SLAF markers, of which 156,368 were polymorphic. After a series of screenings, 7394 SLAF markers were found to be effective and were used for linkage analysis; however, only 6594 markers containing 13,844 SNP loci were mapped successfully onto the high-density genetic map. The average sequencing depth of the markers that failed to be included in the map was 535,794, and the average integrity of all samples was 95%, which suggested that this failure was not due to a sequencing problem. The polymorphic markers of mango developed by SLAF-seq technology were more numerous than those of the previously mentioned crops, further demonstrating the potential of SLAF-seq as a low-cost technique to effectively develop numerous and reliable molecular markers for fruit trees.

Genetic maps are the basis of QTL analyses of the agronomic traits, which are important for the improvement of breeding programs. However, the construction of genetic maps is difficult, and there is a shortage of genetic maps for fruit trees, especially the tropical ones. There were a few reports for papaya (Sondur et al., 1996), avocado (Sharon et al., 1997), mango (Chunwongse et al., 2000, 2015; Kashkush et al., 2001; Fang et al., 2003), and guava (Padmakar et al., 2015) with respect to genetic maps. Although four genetic maps have been previously reported for mango, there were several deficiencies in them. Firstly, the maximum number of F₁ individuals used for these studies was 60. If the size of the population used for map construction is too small, it cannot reflect small recombination rates, and is not beneficial for increasing the efficiency of QTL detection. Secondly, the dominant molecular markers, such as AFLP, were mainly employed, whereas the co-dominant markers, such as SSR, were rarely used. Thirdly, the number of mapped markers was relatively small, which resulted in relatively large average distances between the adjacent markers. The progress of the genetic map construction was very slow, and to the best of our knowledge, no high-density map has been reported to date. The total length of the new genetic map, which was constructed using 6594 SLAF markers in this study, was 3148.28 cM spanning 20 LGs, and the average distance between adjacent markers was 0.48 cM. This map could be considered the first high-density genetic map of mango, and its evaluation using haplotype and heat maps suggested that it was of high quality (Supplementary Material Presentations 1, 2).

Some of the seedling trees of the F₁ progenies of the two cultivars of mango began to set fruit two years ago, and the phenotypic data of the two parents and F₁ progenies were collected. The characteristic of disease resistance was segregated. Since the Jin-Hwang mango is resistant to anthracnose and Irwin mango is susceptible to anthracnose, the cross between the two cultivars might be used to breed a new variety with high anthracnose resistance and enable the study of the inheritance of resistance. The high-density genetic map of mango constructed here will be applied to map the QTL/QTLs of this trait.

Additionally, our sequencing results provided a mass of SLAF markers for mango. Because they were explored at the whole-genome level, the sequence and location information of these markers can be used as a reference in the future genome assemblies of mango. Furthermore, a total of 13,844 SNPs, which are co-dominant markers, were found among the SLAF markers, and they can be used for comparative genomic studies and future MAS in mango breeding program. Like other molecular markers, the SLAFs and SNPs developed here also can be applied to other mango cultivars and progenies for identification of the germplasm or hybrid and analysis of the genetic diversity among cultivars.

Conceived and designed the experiments: CL, BS, and QY. Performed the experiments and analyzed the data: CL, BS, and HW. Wrote the manuscript: CL, BS. Read and approved the final manuscript: WX, SW.