In total, 71 anchor markers were available for the comparison between the diploid genomes of Prunus and Fragaria. These were: a) 40 RFLPs from the same number of probes found polymorphic out of 65 single-copy probes already mapped in Prunus (Table 1), b) 13 Fragaria ESTs obtained after selecting from 135 of them highly homologous to Prunus mapped ESTs out of the 515 ESTs of the Prunus transcript map (Table 2); these markers were selected by their position on the map as they fell in regions not covered with RFLPs or that were judged to be of interest for the genome comparison, c) eight Prunus ESTs selected from 13 ESTs of the transcript map that did not have sequence homology with known Fragaria ESTs (Table 2), d) eight Fragaria STSs obtained from genes of known function previously mapped in Fragaria 21 (Table 2), and e) two SSRs (Table 2).

Two mapping strategies were used in the placement of novel markers on both the T×E and FV×FN linkage maps (Table 3). For 55 markers: all 40 RFLPs, the eight Fragaria STSs, five of the Fragaria ESTs that displayed intron length polymorphisms and the two SSRs, were mapped by genotyping all individuals of the FV×FN population. The 16 remaining markers were mapped using the bin mapping approach where a set of six plants (the bin set) permit wselective' or 'bin' mapping using the diploid strawberry mapping population 22 . Only ten new markers were mapped in the T×E population: the eight Fragaria ESTs were bin mapped 15 and the two SSRs were mapped using the whole population.

The Fragaria map, constructed with MapMaker as described in the materials and methods, included 228 markers: 172 of the previous map 18 , and 55 new markers studied here in the whole population. Three of the markers mapped in the map by Sargent et al. 18 , CFVCT028, CFVCT05 and UFFxa03B05, could not be located using the conditions set for mapping in this work. The seven expected linkage groups (FG1–FG7) were detected, and the total genetic distance covered by the FV×FN map was of 568.8 cM with an average density of 2.5 cM/marker and a ratio of 0.30 Mb/cM. The resulting map shows minor rearrangements in comparison to the original obtained by Sargent et al. 18 most of which occurred in the upper part of the FG2. This map is presented in Additional file 1: 'FV×FN reference map' that is included as supplementary information. The Prunus map included a total of 564 markers, the 562 described in 14 plus two SSRs mapped here. The map covers a distance of 511,3 cM with an average density of 0.91 cM/marker and a ratio of 0.59 Mb/cM.

The distribution of the 71 anchor markers across the seven linkage groups of the strawberry genome was relatively even (Figure 1), ranging from seven markers on FG1 and FG5 to sixteen markers on FG2, with average marker densities ranging from 10.6 cM/marker on FG1 to 5.5 on FG2. The map distance covered by these markers was 441.3 cM, a 78% of the map constructed with all markers. Only three large gaps of >30 cM were observed, one at the end of FG3 (32.0 cM), one in the middle of FG2 (31.7 cM) and the longest on the lower part of FG7 (37.2 cM). The distribution of these 71 markers across the Prunus map was similar, with a maximum of 17 markers on PG1 and a minimum of four on PG4 and marker densities per linkage group ranging from 15.6 (PG4) to 5.0 (PG2) cM/marker. Map coverage with anchor markers was of 399.5 cM, a 78% of the complete map. Longest gaps between markers were smaller in Prunus, with only four exceeding 20 cM: 21.7 on PG7; 21.9 on PG6, 20.8 on PG8 and 28.6 on PG5. All the gaps on both the Fragaria and Prunus maps had at least one marker bin-mapped between the extreme markers of the gap, except for those on PG6 and FG3, suggesting that these were the longest gaps of each map.

The overall pattern of synteny is summarized in Figure 2, where it can be seen that the majority of linkage groups of either species have most markers in one or two linkage groups of the other. The simplest case is G5 of both species which have all markers in common, and the most complex is FG6 which contains markers from five different Prunus linkage groups, although most of the markers on this linkage group belong to PG3 and PG7. Considering only the Prunus linkage groups, five of them contain most or all markers from only one group of Fragaria (PG2-FG7, PG3-FG6, PG4-FG3, PG5-FG5 and PG8-FG2), and the other three of two groups (PG1-FG2 and FG4, PG6-FG1 and FG3, and PG7-FG1 and FG6).

The presence of markers from two or more linkage groups of one species in a linkage group of the other would suggest that a fission/fusion, or a translocation event has taken place between the two species since their divergence from a common ancestor. Nine of these rearrangements would have taken place between the Fragaria and the Prunus genomes, based on our results.

Colinearity of markers within syntenic regions of Prunus and Fragaria was only partial. The two most colinear groups FG5 and PG5 require only one inversion event to place all markers in the same order, but many more are required in other linkage groups. As shown in the materials and methods section, FG1 would require two inversions for the markers present on that linkage group to be in the same order as on PG7 and PG6 of Prunus, FG3 and FG4, three inversions, FG7 four inversions, FG6 six inversions and FG2 eight inversions. In total, at least 27 inversions are needed to account for the differences in the marker order of the two genomes. The estimated lower boundary of the total number of breakpoints that separate the Fragaria and Prunus genomes is thus 36 (nine translocations and 27 inversions).

A scheme of the possible evolution of the Fragaria and Prunus genomes from a hypothetical ancestral genome of x = 9 23 is presented in Figure 3. In this scheme we have only considered the major chromosomal rearrangements, i.e. fusions/fissions and reciprocal translocations involving more than two of the markers used for the comparison. According to this scenario, the ancestral genome underwent two fusions (A5/A6 and A7/A8, corresponding to FG6 and FG1, respectively) and a reciprocal translocation between part of the A8 chromosome in FG1 and A9, which resulted in FG3, to become the x = 7 strawberry genome. For the Prunus x = 8 genome, the ancestral genome was submitted to three fusions, A1/A2 to form PG1, A6/A7 to form PG7 and A8/A9 to form PG6 and two fissions, part of A1 in the A1/A2 chromosome to form PG8 and part of A9 in the A8/A9 chromosome to form PG4.

Some inferences about the evolution of certain chromosomes can be formulated based on the comparison between the two maps (Figure 1), the proposed evolution of both genomes from an ancestral one (Figure 3), and the limited information available on cytogenetics and map comparison of the Rosaceae. The simplest comparison is that of FG5 and PG5, which display remarkable levels of structural conservation, the marker order between Prunus and Fragaria differing by just a single inversion event. Most markers of FG7 and PG2 are in common, suggesting their origin from a single ancestral chromosome (A4), but FG7 includes a short fragment of PG6, and PG2 is not completely included in FG7, with a small region being located on FG4. PG1 is a long linkage group that contains many more markers than the rest and that is likely to coincide with chromosome 1 of Prunus 14 , which is clearly longer than the other chromosomes of this genus 24 . Based on karyotype observations, such a long chromosome does not exist in Malus 25 or in Fragaria 26 . The map comparison between Malus and Prunus 14 provided additional evidence that the long Prunus chromosome may be split into two in at least one of the constituent genomes of the amphidiploid Malus chromosome complement. Our data suggest that PG1 arose from the fusion of two ancestral chromosomes, A1 and A2, which correspond to FG4 and FG2, respectively, in strawberry. The region of PG1 where the fission/fusion occurred in the Prunus-apple comparison (38–44 cM from the top of PG1) is also compatible with that of the Prunus-Fragaria (estimated to be in the region of 25–45 cM from the top of PG1). Interestingly, most of PG8, one of the Prunus chromosomes with the fewest number of markers and with the smallest genetic distance, appears to be integrated into FG2, with five of the six anchor markers studied being in the same order in PG8 and FG2. This situation may be the result of the fission of A1 in Prunus but not in Fragaria.

The parents and progeny of two mapping populations were used for comparing the Prunus and Fragaria genomes, those of the F₂interspecific Prunus reference mapping population (N = 82) derived from the F₁ cross between the almond (P. dulcis) cultivar 'Texas' and the peach (P. persica) cultivar 'Earlygold' (abbreviated T×E) and the F₂ interspecific diploid Fragaria reference mapping population (N = 76) derived from the F₁ cross of F. vesca 815 × F. nubicola 601 (abbreviated FV×FN). The T×E map is composed of 562 markers (185 SSRs, 361 RFLPs, 11 isoenzymes and 5 STSs) 14 , whilst the FV×FN map currently consists of 182 markers (175 SSRs, 6 gene specific markers and 1 SCAR) 18 . Most markers used were selected from T×E and subsequently mapped in FV×FN.

For Fragaria, one gram of young expanding leaves of each individual was collected and kept at -80°C before DNA isolation. Genomic DNA was isolated from the leaf samples using the CTAB method of Doyle and Doyle 42 , followed by DNeasy miniprep kit purification (Qiagen). DNA concentrations were measured using a Gene Quant II spectrophotometer (Pharmacia Biotech). For Prunus, DNA was extracted as described in 28 .

A total of 65 probes from various Prunus species and apple were used for RFLP analysis (Table 1). These probes were selected from those used in the construction of the Prunus reference map 14 to be single-copy and covering the whole genome at approximately even distances of 10–25 cM. RFLP analysis was performed with the procedure of Viruel et al. 28 . DNA probes were first hybridized in the parents of the FV×FN population and those that detected RFLPs were studied later in all F₂ individuals.

An additional set of markers was developed from the Prunus transcript map, i.e. the collection of EST unigenes that are located on the same BAC or the same BAC contig of the peach physical map as mapped markers (usually RFLPs) of the reference linkage map 43 . We analysed 515 T×E ESTs currently anchored in the Prunus transcript map (found in the Genome Database for Rosaceae) to find homologous EST sequences in Fragaria by using the TBLASTX program 44 . The selected sequences (only sequences with an e-value <1.00E-15) (Table 2) were then analyzed with TBLASTN, and the most homologous regions were selected for primer design. Primers were designed to include putative Fragaria introns based on the Arabidopsis genome sequence 45 and to give an expected amplicon size >800 bp, using the program Primer3 46 .

For SNP detection, Fragaria ESTs were PCR-amplified in the parents of the Fragaria mapping population in a temperature-gradient PCR to obtain the optimum annealing temperature for each EST primer pair. The PCR reactions were performed as described in 47 and amplification products were sequenced with an ABI PRISM 3700 DNA Analyzer and SNPs were detected between the two parental lines through sequence alignment using the STADEN package 48 and verified by visual inspection of the DNA chromatograms. The same procedure was followed for Prunus ESTs of the transcript map for which we did not find homologous Fragaria ESTs.

The sequences of the primers used for the development of these markers can be found in Table 4. The terminology used for the EST derived markers was: E (for EST), a two letter code for the species where the EST was obtained (Fv, Fa or Pp for F. vesca, F. x ananassa and P. persica, respectively), two additional letters for the place where the EST was obtained (i.e. NH for University of New Hampshire, VB for Virginia Bioinformatics Institute, UF for University of Florida, TR for Trisaia Research Center and CU for Clemson University) and four numbers that correspond to the last four digits of the EST EMBL accession number.

Eight further gene specific loci that had previously been mapped in the FV×FN population 21 were mapped in the T×E population. The primer pairs used for amplification of four of these genes (CEL-2, PES, ANS and APX) were previously described by Sargent et al. 21 whilst primer pairs for the remaining four (EKO, F3H, DFR and ADH) were designed following the procedures of Sargent et al. 21 from gene sequences deposited in the EMBL database and are listed in Table 4. The primer pairs were used to amplify products from the parents of the Prunus mapping population following the procedure of Sargent et al. 47 and SNPs were detected as described above.

One SSR (ACO) developed in the 5' UTR of Prunus 1-aminocyclopropane-1-carboxylate oxidase gene (ACC oxidase; AF129073) was used to locate this gene, which was already mapped in Fragaria 21 , onto the T×E map. The primers used are listed in Table 4. Another apricot SSR, AMPA112, previously described by Hagen et al. 49 , was polymorphic in both reference populations and was also included in the map comparison. PCR reactions were performed as described by 50 .

The T×E map of Dirlewanger et al. 14 , constructed with MapMaker/EXP v. 3.0 51 , was used as a standard for map comparisons. Data for the FV×FN map, originally constructed with Joinmap 3.0 52 by Sargent et al. 18 , was used for the mapping of novel loci in Fragaria. However, in order to make both maps comparable, the same data of the FV×FN map plus those obtained here were used, but the map was reconstructed with MapMaker. The Kosambi mapping function was used to convert recombination units into genetic distances. The mapping procedure followed the guidelines of previous maps constructed in Prunus 29 . The usual notation for the eight linkage groups of Prunus is G1 to G8 and for the seven Fragaria groups is I-VII. In order to facilitate the map comparison we have used on this occasion the terminology PG1–PG8 for Prunus and FG1–FG7 for Fragaria.

A subset of six plants of the T×E mapping population having a high number of recombination breakpoints and a uniform distribution across the Prunus genome was selected by Howad et al. 15 . The genotype of this set of plants (the bin set) identified 64 fragments (bins) of the Prunus map with average size 7.8 cM. Following a similar approach, a bin set of six plants was selected from the FV×FN mapping population by Sargent et al. 22 . The Fragaria bin set detected 46 fragments of its genome with an average length of 12.6 cM. Using the bin sets of both species allowed us to establish the position of some of the markers with lower cost and effort than mapping with the whole population.

Table 3 summarizes the mapping strategy employed for all novel markers mapped in this investigation. All RFLPs used were already mapped in Prunus using the entire T×E progeny and were mapped in Fragaria using the entire FV×FN population. The Fragaria and Prunus ESTs found to be polymorphic in the parents of the Fragaria mapping population were either sequenced in the Fragaria bin set and bin-mapped, or mapped in the whole FV×FN population when clear intron length polymorphisms were detected. The eight Fragaria ESTs, previously mapped in FV×FN, were bin mapped in Prunus. The SSRs were mapped in Prunus and Fragaria using all individuals of these populations. The data for each marker were scored independently by two researchers. Conflicting results were re-examined and in case of disagreement, the most conservative option was taken.

To estimate the number of chromosomal rearrangements that have occurred between Prunus and Fragaria since they diverged from a common ancestor, we elaborated a list of the markers of each of the Fragaria linkage groups with their correlative position on the Prunus linkage map. Taking FG1 as an example, this linkage group has seven anchor markers with order 7.2 (PG7, position 2), 7.3, 7.5, 7.4, 6.10, 6.6, 6.7. Then, we deduced the minimal number of mutations that would place the positions of these markers in the same order as in Prunus. In the example of FG1, we counted one translocation (between PG7 and PG6) and two inversions, one involving the 7.4–7.5 fragment and the other the 6.6–6.7 fragment, to give the final marker order: 7.2, 7.3, 7.4, 7.5, 6.10, 6.7, and 6.6. An additional mutation (probably caused by a translocation) had to occur to explain the gap between 6.10 and 6.7. This mutation was considered when analyzing the chromosome that received the translocated fragment. We counted one breakpoint per translocation and one per inversion. This is a lower boundary, as inversions may require one breakpoint if they involve the distal part of a chromosome, or two if they correspond to an internal fragment of a chromosome. Finally, we considered as translocations only those regions comprising two or more markers, as single-locus translocations are more likely to be spurious.