Excellent AFLPs, EST-SSRs, and scIvana_1-based marker banding profiles were obtained [see Additional file 1], despite the size and complexity of the sugarcane genome. The 72 enzyme-selective primer combinations tested provided a range of AFLP band numbers per gel (44 to 174), and polymorphic loci (four to 33). Subsequently, 22 combinations were selected as optimal for genotyping the F₁ population [Additional file 2]. The combinations generated 102 to 172 AFLP bands per gel, and 19 to 48 polymorphic loci, which revealed 22.1% (685/3,094) segregating loci. Among the segregating loci, 71.2% (488/685) segregated in only one parent, and 28.8% (197/685) segregated in both parents (Table 1). The ‘IAC66-6’ clone and ‘TUC71-7’ variety contributed a respective 52% (254/488) and 48% (234/488) of the loci that segregated in only one parent. The average number of amplified bands and segregating loci per enzyme-primer combination were 140.6 (3,094/22) and 31.1 (685/22), respectively.

^a AFLPs generated from 22 enzyme-selective primer combinations, EST-SSR alleles generated from 41 loci and scIvana_1-based markers generated from 6 restriction enzyme-primer combinations.

^b Data obtained after Bonferroni’s correction.

^c Markers present in only one parent with a 1:1 segregation ratio in the mapping population.

^d Markers present in both parents with a 3:1 segregation ratio in the mapping population.

From the 184 EST-SSR loci initially investigated, 22.3% (41) were selected for genotyping [Additional file 2]. These loci revealed 273 alleles with an average of 6.7 alleles per locus; 80.6% (220) segregated in the F₁ population, 68.6% (151/220) segregating in only one parent, and 31.4% (69/220) in both parents (Table 1). The ‘IAC66-6’ clone and ‘TUC71-7’ variety contributed a respective 43% (65/151) and 57% (86/151) of the alleles that segregated in only one parent.

Among the 16 restriction enzyme-primer combinations used to amplify the retrotransposon scIvana_1-based markers, six were selected for genotyping the F₁ population. These combinations revealed 357 loci; 24.4% (87) behaved as segregating loci. Among them, 85.1% (74/87) and 14.9% (13/87) segregated in only one parent and in both parents, respectively (Table 1). The average number of amplified bands and segregating loci per enzyme-primer combination were 59.5 (357/6) and 14.5 (87/6), respectively [Additional file 2]. The male and female parents contributed a respective 44.6% (33/74) and 55.4% (41/74) of loci that segregated in only one parent. Sixteen retrotransposon scAle_1 combinations revealed gel profiles; however, amplicon absence, or amplifications associated with non-specific polymorphism prevented profile use in genotyping.

AFLP and scIvana_1-based loci exhibited similar levels of segregating alleles in the mapping population (~25%). Notably, both techniques reveal polymorphisms at restriction enzyme cleavage sites. On the other hand, EST-SSR loci showed high levels of segregating alleles in the mapping population (~80%). The polymorphisms observed at EST-SSR loci are due to differences in the size of multiple alleles; we cannot predict if any allele will be fixed in a sugarcane cultivar, which is expected to be highly heterozygous. Both parents contributed equally to polymorphisms i.e. 49.4% ± 3.7% and 50.6% ± 3.8% of the amplicons derived from the male (‘IAC66-6’) and female (‘TUC71-7’) parent, respectively, independent of marker type.

It is important to clarify that we organized our segregation data assuming that homo(eo)logous chromosomes paired faithfully during meiosis, leading to regular bivalent formation as well as normal gametes. It is imperative to emphasize that sugarcane is an artificial genome, highly polyploid, aneuploid, and has interspecific origins, which impedes our capacity to designate co-dominant markers at any locus. Consequently, loci were divided into heterozygous in one parent and null in the other (simplex x nuliplex), and heterozygous in both parents (simplex x simplex). Based on this model, all markers were scored as dominant (or binary), and were assigned to the expected segregation ratios i.e. 1:1 and 3:1 (Table 1).

A total of 992 segregating loci were genotyped in the mapping population; 685 AFLPs (generated from 22 enzyme-selective primer combinations), 220 EST-SSRs (derived from 41 loci), and 87 scIvana_1-based loci (obtained from six enzyme-primer combinations). The expected segregation ratios at each locus (992) were checked using Chi-square tests, adjusting for multiple tests using Bonferroni correction. Then, 730 (73.6%) loci were safely used to build the map, being 535 AFLPs, 130 EST-SSRs, and 65 scIvana_1-based loci. The global level of significance used to determine the validity of the segregation ratios of 1:1 and 3:1 was 5.04e-05 (alpha = 0.05/992).

The marker number used to perform linkage analyses was 730; 261 were derived from ‘IAC66-6’ (here indicated as D1), 284 from ‘TUC71-7’ (or D2), and 185 were present in both parents (or C). The final sugarcane map was comprised of 546 markers assembled into 92 co-segregation groups (CGs), and 184 markers not assigned to any CG. Coincidently, D1 (208) and D2 (240) markers were distributed into nearly the same number of CGs (51 and 55, respectively). By using the loci that segregated in a 3:1 fashion (98) as bridges, we provided the integration of 19 CGs (I-1, I-2, I-3, I-4, I-6, I-8, I-9, I-12, II-1, II-3, III-1, IV-1, VII-1, U-1, U-2, U-3, U-8, U-13, and U-33). The marker number per CG ranged from two to 39 (Figure 1). The total map length was 4,843.19 cM, with a marker density of 8.87 cM. The CG length covered a range from 1.14 to 404.72 cM, with an estimated average of 52.64 cM. The greatest distance between two adjacent markers was 48.25 cM (CG I-3).

Forty-one CGs (44.6%) were assembled into seven putative independent homo(eo)logous groups (HGs), which were recognized based on 82 alleles generated by 24 EST-SSR loci. The remaining loci (ESTB189, ESTB191, CV11, and CV86) did not contribute to CG assembly; only one allele was positioned on the respective HG. The CG number assembled into HGs ranged from two to 15, the largest (HGI) contained 210 markers, and the smallest (HGVII) six markers, which was clearly a very non-uniform distribution (Figure 1). Eight previously mapped EST-SSR loci 32 were placed here on HGI (ESTA53, ESTB94, ESTB100, and ESTB118), HGIII (ESTA48), HGIV (ESTC80 and ESTB99), and HGV (ESTB60).

The mapped proportion of scIvana_1-based loci was 86.2% (56/65), greater than AFLPs (73.8%, 395/535) and EST-SSRs (73.1%, 95/130). The scIvana_1-based loci (56) were positioned on 21 CGs, which were represented in four HGs. However, there is some evidence that these markers may not be regularly distributed (Figure 1); only one marker was placed in groups I-6, I-13, III-3, III-4, U-1, U-2, U-4, U-7 and U-8, and more than five markers were placed in groups I-1, I-3, and IV-1. The evidence for clustering of scIvana_1-based markers was verified by the Chi-square goodness-of-fit test that results in a p-value 4.64e-04. This model has been used for testing if marker distribution deviates significantly from a random distribution in genetic maps i.e. segregating as expected in a Poisson distribution 47. The number of regions with two (6) and three (1) adjacent markers was higher than expected only by chance (2 and 0.09, respectively). Interestingly, distances between some of the adjacent scIvana_1-based markers were lower than 5 cM, notably in groups, I-1, I-2, III-2, IV-1, and U-51 (Table 2). This value was the same used by Rossi et al.36 to define NBS/LRR RGA clusters.

^a Defined by the distance between the flanking markers.

^b HG, homo(eo)logous groups; CG, co-segregation groups.

We subsequently obtained the total scIvana_1-based marker numbers (357), divided by the number of enzyme-primer combinations used to obtain the amplicons (6), and the result was ~60. We propose this as the number of retrotransposon scIvana_1 copies in the sugarcane genome. Similarly, by obtaining the total number of scIvana_1-based markers each parent contributed separately, the following results correspond to the number of scIvana_1 copies in the respective ‘IAC66-6’ and ‘TUC71-7’ genomes: 53 (316/6) and 54 (324/6).

However, the total number of scAle_1-based fragments was 1008. Nine restriction enzyme-primer combinations were used to amplify the scAle_1-based fragments, an average of 112 per combination. The results did not facilitate the identification of any clear polymorphisms in the gels. Consequently, the data were not used for mapping studies.

Twenty-five scIvana_1-based markers selected for genotyping were sequenced; and 64% (16/25) showed homology with known nucleotide sequences deposited in GenBank (Table 3). Most (13 sequences) showed homology with scIvana_1 sequences, and others showed homology to Zea mays (DIsIvSI337), Oryza minuta (DIsIvSI390, DIsIvSI412), and Sorghum bicolor (SIsIvSI180) sequences. The DIsIvLI228 fragment also exhibited homology with a Saccharum chloroplast sequence. Six of the sequences were mapped, DIsIvSI163, DIsIvSI208, DIsIvSI160, SIsIvLI240, SIsIvLI412, and DIsIvLI415, and two were tightly linked with EST-SSRs. The DIsIvSI337 fragment was submitted for tblastx search, and revealed similarity with a hypothetical, highly conserved protein of unknown function in Arabidopsis thaliana, and other species. Similarly, DIsIvSI390 and DIsIvSI412 fragments were aligned with an Oryza minuta sequence; tblastx indicated similarity with a hypothetical protein of unknown function. The SIsIvSI180 fragment exhibited similarity to two Sorghum sequences, one to a retrotransposon of S. bicolor48. Finally, the DIsIvLI228 fragment exhibited a certain identity with a ribosomal protein chloroplast sequence of Saccharum ‘SP80-3280’ (Table 3).

^a All are standard nucleotide-nucleotide BLASTn scores.

The mapping population was composed of 188 individuals derived from a single cross between ‘IAC66-6’ and ‘TUC71-7’. The male parent ‘IAC66-6’ is a clone with low sucrose content, large diameter stems, and is susceptible to sugarcane smut, a fungal disease caused by Sporisorium scitamineum; the female parent, the Argentinean variety ‘TUC71-7’, exhibits a higher sucrose content, lower diameter stems, and resistance to smut disease. This disease limits the use of recent high-yielding sugarcane varieties developed in Brazil. The cross was made under field conditions at the CanaVialis/Monsanto Company experimental station, located in the northeastern state of Alagoas, Brazil (S 09°39'57''; W 35°44'07''). Sugarcane successfully flowers and sets seed at the locality due to light period duration (photoperiod), and plants were therefore cultivated at this site. Seeds were harvested, germinated in plastic boxes, and transported to the southeastern state of São Paulo (S 22°19'49''; W 47°10'21'') for field cultivation.

DNA was isolated from young leaves of F₁-progeny and parental plants using the CTAB-based extraction procedure 67, with minor modifications. DNA concentrations were carefully estimated following electrophoresis on ethidium bromide-stained agarose gels using molecular weight standards; aliquots of 50 ng/μl were prepared following quantification.

AFLPs were amplified based on the protocol described by Vos et al.68 and applied to sugarcane. Briefly, 250 ng of genomic DNA was double digested with 6U of EcoRI (Promega) and MseI (NE Biolabs) in a 25-μl reaction mixture (10 mM Tris–acetate pH 7.5, 10 mM Mg-acetate, 50 mM K-acetate, 5 mM DTT, 1 X BSA) for 4 h at 37°C. Restrictions were terminated by heat inactivation for 20 min at 65°C. The resulting fragments were ligated to adapter sequences by addition of an equal volume of ligation solution comprised of 0.25 μM EcoRI adapter, 2.5 μM MseI adapter, 1 X enzyme reaction buffer, and 67 U of T4 DNA ligase (400 units/μl, NE Biolabs). Incubations were performed for 16 h at 16°C, and reactions were terminated by heat inactivation. The adapter-ligated DNA (3 μl) was used for pre-selective amplification with primers based on the adapter sequences with one selective nucleotide at the 3’ end (EcoRI + A and MseI + C). Pre-selective amplification was performed in a 20 μl reaction mixture containing 1.5 mM MgCl₂, 0.5 mM each dNTP, 250 nM each primer, 1 X enzyme reaction buffer, and 3 U Taq DNA polymerase (Promega). Amplifications were conducted under the following conditions: 94°C for 2 min; 26 cycles of 94°C for 60 s, 56°C for 60 s, 72°C for 60 s; and a final elongation at 72°C for 5 min. For the selective step, 1.5 μl of a 5-fold water diluted pre-selected PCR product was used as DNA template. The 20 μl reaction mixture contained 1.5 mM MgCl₂, 0.2 mM each dNTP, 250 nM EcoRI + ANN, 300 nM oligo MseI + CNN, 1.6 U Taq DNA polymerase (Fermentas), and 1 X enzyme reaction buffer. Selective amplification was conducted under the following conditions: 94°C for 2 min; 12 cycles of 94°C for 30 s, 65°C for 30 s, 72°C for 60 s; the final 23 cycles had similar conditions with the exception of a 56°C primer annealing temperature, and a final elongation at 72°C for 2 min. Following PCR, the amplified products were mixed with an equal volume of denaturing buffer containing 95% formamide, 10 mM EDTA (pH 8.0), 0.2% bromophenol blue, and 0.2% xylene cyanol. Samples (3 μL) were loaded into 5% (w/v) polyacrylamide gels (acrylamide/bis-acrylamide, 19:1). Electrophoresis was performed at a constant power of 70 W for 4 h, using a Sequi-Gen® GT (Bio Rad) apparatus. Gels were silver-stained according to the protocol described by Creste et al.69.

Seventy-two different restriction enzyme and selective primer combinations were examined using DNA from both parents in duplicate reactions. Combinations that exhibited good profiles, and revealed a large number of loci and polymorphism rates (≥ 20%) were selected for genotyping the F₁ population. The polymorphism rate between parents was calculated by assessing the number of bands present in one parent and absent in the other, in relationship to the total number of amplified bands.

In analyzing the SUCEST database, Pinto et al.70 identified 2005 clusters containing SSRs. Primer sets were subsequently developed from these data, and used in polymorphism analyses 70717273. In addition, Maccheroni et al.74 analyzed 352 and 122 sugarcane ESTs available in both public 75 and private 76 databases to establish sugarcane SSRs. Primer sets were developed from these sequences. In the present study, we used published 727374, and non-published primer sets developed by CanaVialis/Monsanto.

EST-SSRs amplification was performed in a final volume of 10 μl in 96-well thermocycler plates. Approximately 20 ng of template DNA was mixed in a solution of 0.25 μM of each forward and reverse primer, 0.2 mM each dNTP, 2.0 mM MgCl₂, 1X Colorless Go Taq buffer, and 1.0 U Go Taq Flexi DNA Polymerase (Promega). Amplifications were performed using two thermal cycling programs. The first program was conducted under the following conditions: 94°C for 3 min; 31 cycles of 94°C for 60 s; primer annealing at varied temperatures for 60 s; extension at 72°C for 60 s; and a final elongation at 72°C for 2 min. The second was conducted under the following cycle parameters: an initial denaturation step at 94°C for 5 min; followed by 35 cycles of 94°C for 30 s; primer annealing at varied temperatures for 30 s; extension at 72°C for 30 s; and a final elongation at 72°C for 60 min. PCR products were analyzed by two methods, denoted S and F. Amplicons were resolved by 5% (w/v) denaturing polyacrylamide gel electrophoresis, and silver stained (S) as above described; or capillary electrophoresis using a MegaBACE 1000® genotyping system (GE Healthcare Life Sciences). For capillary electrophoresis, forward primers were labeled with fluorescent dyes (F) (fluorophore NED or 6-FAM, Applied Biosystems), and fragments were verified with the Fragment Profiler version 1.2®.

Polymorphisms between parental genotypes were assessed by amplifying 184 EST-SSRs using DNA from ‘IAC66-6', ‘TUC71-7', and a sample from the mapping population (F₁). The data included 33 EST-SSRs developed by Oliveira et al.72, two from Marconi et al.73, and three from Maccheroni et al.74. In addition, 146 EST-SSRs were available from CanaVialis/Monsanto (unpublished data). Results showed 41 polymorphic loci, which did segregate in the progeny sample; therefore, these loci were applied to genotype the mapping population. Details on these sugarcane EST-SSRs are presented in Table 4.

^{a, b, c, d} Develop by Oliveira et al.72, Marconi et al.73, CanaVialis (unpublished); Maccheroni et al.74, respectively.

^e PCR program as described in Material and Methods.

^f Annealing temperature in the amplification reaction.

^g D: Silver-stained polyacrylamide gel electrophoresis (S) or fluorescence-based automated capillary electrophoresis (F) for the detection of EST-SSR alleles.

^h Observed number of alleles per locus and their size ranges in bp.

The principle NBS-profiling technique 77 was applied according to Hanai et al.78 to generate markers based on two retrotransposons named scIvana_1 (GenBank Accession Number JN800016) and scAle_1 (GenBank Accession Number JN800006). Approximately 500 ng of genomic DNA were digested with AluI, DraI, SspI, or RsaI (NE BioLabs). Digestions were performed in a final volume of 30 μl using 7.5 U of enzyme for 7 h at 37°C, according to the manufacturer's recommendations. Reactions were terminated by heat inactivation (20 min at 65°C). Adapters were prepared by incubating equimolar amounts of LA (long arm) and SA (short arm) oligonucleotides at 65°C for 10 min, and respectively cooled to 37°C and 25°C (10 min each). The SA oligonucleotide 3’ end was blocked for Taq DNA polymerase extension by the addition of an amino group, but phosphorylated at the 5' end, which results in an adapter primer-annealing site only following the first PCR cycle. Subsequently, the digested material and a solution containing a 1.6 μM adapter (when restricted with AluI or RsaI), or a 0.2 μM adapter (when DraI or SspI was used), 1 X ligation buffer (NE BioLabs), and 67 U T4 DNA ligase (400 units/μl; NE BioLabs) were mixed in equal volumes (30 μl), totaling 60 μl. Ligation was performed at 16°C for 16 h, and terminated by heat inactivation at 65°C for 20 min. The ligation products (diluted to 5 ng/μL) as template DNA were used to amplify selected fragments anchored to the retrotransposon sequence. A final volume 20 μl reaction mixture contained 4 μl of ligation products, 300 nM of each primer (a primer complementary to the adapter, and a primer complementary to the retrotransposon sequence, Table 5, Figure 2), 1.5 mM MgCl₂, 0.2 mM each dNTP, 1 X buffer enzyme, and 2 U Taq DNA polymerase (Fermentas). PCR was conducted under the following conditions: an initial denaturation step at 94°C for 5 min; followed by 8 cycles at 94°C for 45 s, 58°C (− 1°C per cycle) during 50 s, and 72°C for 1 h:15 min; 25 cycles at 94°C for 45 s, 50°C for 50 s, and 72°C for 1 min; and a final extension at 72°C for 10 min. After PCR, the protocol for preparing AFLP gels was completed, followed by electrophoresis, and gel staining.

^a D = A, G or T; R = A or G; W = A or T.

Sixteen restriction enzyme and retrotransposon complementary primer combinations (for each sugarcane retrotransposon) were examined using the DNA from both parents in duplicate reactions. Combinations that exhibited clear band distribution over the gels, and revealed polymorphism rates ≥ 15% were selected for genotyping the F₁ population.

The nomenclature adopted for AFLP primers followed the Keygene standard primer list 80 followed by the corresponding molecular size (in bp) of the band. A primer code was adopted for EST-SSR loci (ESTA, ESTB, ESTC and CV) followed by the molecular size of the allele (in bp). The retrotransposon-based locus nomenclature was an abbreviation that indicated the enzymes used in the digesting reaction (AI for AluI, DI for DraI, SI for SspI, and RI for RsaI) followed by the primer code (Table 5), and the molecular size of the fragment (in bp). Wu’s 31 loci-segregation pattern notations follow all marker abbreviations e.g., D1-E35M47510 (Figure 1). We assumed the presence of a band, denoted by A, dominant over all 0 or null alleles (or simplex configuration with eight homo(eo)logous copies in the sugarcane genome), independent of marker type. Loci were denoted as “D1” when ‘IAC66-6’ was heterozygous for band presence, and the other parent was homozygous for band absence (A0000000 x 00000000). Loci were denoted as “D2” when ‘TUC71-7’ was heterozygous and ‘IAC66-6’ was homozygous. These loci were expected to segregate in a 1:1 fashion in the F₁ population. “C” loci were heterozygous in both parents (A0000000 x A0000000), and exhibited a 3:1 segregation ratio. Differences between observed and expected proportions were compared using Chi-square test, assuming a polyploid model based on single-dose markers (SDM) for analyzing segregation in outcrossing species 14. For minimizing problems caused by multiple comparisons, the Bonferroni correction was performed. Chi-square tests and Bonferroni adjustments for the effects of multiple comparisons were performed using the software R, v. 2.13.0 81. Loci with segregation distortion were not included in linkage analysis.

All linkage analyses were performed using OneMap software 82. Version 2.0-1 was preferred to construct a multipoint maximum likelihood linkage map applying a Hidden Markov Model approach 83.

Firstly, co-segregation groups (CGs) were established using a LOD score ≥ 6.0, and a recombination fraction ≤ 0.35. For groups with six or less markers, the best order was obtained by comparing all possible orders choosing the one with highest likelihood using the algorithm implemented in the command named “compare”. To obtain the best order for larger groups (more than six markers), the command “order.seq” was applied. In this case, the likelihood was the criteria used to place the markers along the CGs under a multipoint approach, as validated by Mollinari et al.84. Additionally, the “ripple” command was used to check for alternative orders, as well a visual inspection on the matrix containing the pairwise recombination fractions and LOD scores (heatmaps) for the CGs. The commands “compare”, “order.seq” and “ripple” were similar to those in the MAPMAKER/EXP software 17. Finally, multipoint estimates of recombination fractions were calculated and converted into linkage distances using the Kosambi map function 85. Map drawings were generated using MapChart 2.2 86.

Due to the multiallelic nature and known polymorphisms, EST-SSR loci are valuable in recognizing homo(eo)logous groups (HGs) in sugarcane. Initially, CGs were assigned to HGs if they contained at least two of the same EST-SSRs. In addition, CGs were putatively added if they contained an EST-SSR locus in common with the HG 28303254. Using this practice, two HGs were established, I and IV. Then due to the insufficient amount of SSRs, in a number of instances only one locus was used to assign CGs to the following putative HGs, II, III, V, VI and VII 3032. We applied Roman numerals to denote HGs; within each HG, CGs were classified in a descending order according to size (in cM). The unassigned groups were designated U, and also classified according to their size.

Finally, to test if scIvana_1-based markers have a tendency to be clustered along the genome, we used an approach similar to the one presented by Echt et al.87. The genetic map was divided in 10 cM bins and the number of scIvana_1-based markers in each interval was recorded. If markers were randomly distributed, they would follow a Poisson distribution 47, defined as P(x) = λ^xe^-λ/x!, where P(x) is the probability function; x is the number of markers observed in the intervals (ranging from 0 to 3), λ is the distribution parameter calculated as average number of markers per interval in the map. A Chi-square goodness-of-fit test was performed, with 2 degrees of freedom (df = c −1- r, where c is the number of classes and r is the number of estimated parameters, r = 1).

Verification that amplified DNA fragments were derived from retrotransposon templates was conducted by excising DNA fragments from polyacrylamide gels, eluting DNA in a TE solution (10:1), and re-amplifying the DNA fragments. Five μl of the diluted DNA mixture was added to 50 μL of the same solution used for retrotransposon-based marker generation, however primer concentration was 30 nM, and Taq DNA polymerase was 5 U. The PCR program was simplified and conducted under the following conditions: an initial denaturation step at 94°C for 5 min; 30 cycles of 94°C for 30 s, 55°C for 30 s; 72°C for 42 s; and a final extension at 72°C for 10 min. For sequencing, PCR fragments were resolved on agarose gels, purified with the QIAEX II Gel Extraction kit (QIAGEN), and cloned into pMOS Blue Blunt-Ended Cloning Kit (GE Healthcare Life Sciences). Inserts were sequenced in the forward direction. Sequencing reactions were performed according to Sanger et al.88 using DYEnamicTM ET Dye Terminator Cycle Sequencing Kit (Amersham Pharmacia Biotech, Inc.) on an ABI 3730 system (Applied Biosystems). Sequence quality was examined using the Phred/Phrap/Consed package 89. Nucleotide sequences were compared to reference data available at Genbank by BLAST analysis 90.