This document is posted to help you gain knowledge. Please leave a comment to let me know what you think about it! Share it to your friends and learn new things together.
Saccharum spontaneum L. 'SES 208' genetic linkage map combining RFLP- and PCR-based markers
Jorge da Silva 2'3, Rhonda J. Honeycutt *, William Burnquist 3, Salah M. A1-Janabi ,,4, Mark E. Sorrells 2, Steven D. Tanksley 2 and Bruno W. S. Sobral *'* *California Institute of Biological Research, 11099 N. Torrey Pines Road, Suite 300, La Jolla, CA 92037, USA (* author for correspondence); 2Department of Plant Breeding and Biometry, Cornell University, Ithaca, NY 14853, USA," 3 Copersucar Technology Center, Caixa Postal 162, 13.400 Piracicaba, S~to Pau[o, Brazil," 4 Current address: Agronomy Department, Baghdad Institute of Agricultural Technology, Iraq
Received 20 July 1994; accepted in revised form 18 November 1994
A 527 marker linkage map of Saccharum spontaneum L. 'SES 208' (2n = 64) was established by analyzing 208 single-dose (SD) arbitrarily primed PCR polymorphisms, 234 SD RFLPs, 41 double-dose (DD) and one triple-dose (TD) polymorphisms. A map hypothesis constructed using these markers (minimum LOD = 4.00, 0 = 0.25 M) had 64 linkage groups with 13 SD, nine DD, and one TD markers unlinked. Eight chromosome homology groups were identified by using DD fragments as well as SD RFLPs that identified more than one linkage group. Linkages in repulsion phase were absent from the map, as found in two previous genetic studies of this species. Together, these data demonstrate that SES 208 displayed polysomic segregation, a genetic behavior typical of autopolyploid species. As with previous studies, it was concluded that SES 208 behaved like an auto-octoploid, which was also in agreement with the number of homology groups observed. A Z 2 was used to test whether the 527 markers were randomty distributed throughout the genome: both arbitrarily primed PCR markers and RFLPs had a distribu- tion that was statistically indistinguishable from random. The integrated arbitrarily primed PCR-RFLP map had a predicted genomic coverage of 93 To (considering only 442 SD polymorphisms) and an average interval between markers of 6 cM. SD markers were used to estimate the genome size of SES 208 at ca. 33 00 cM.
Introduction
DNA markers have been used as a fundamental tool for rapid and detailed genetic dissection of higher plants. Among the different applications of information obtained with such markers is the construction of genetic linkage maps. Once a
dense map has been constructed, flanking mark- ers can be used to more precisely and quickly detect the presence of genes or quantitative trait loci (QTL). Therefore, high-density linkage maps have direct application in plant breeding due to the high probability that any gene of interest will be tightly linked to at least one marker. Such
166
linkages are useful for marker-assisted selection in breeding programs [9]. Furthermore, high- density maps can provide starting points for map- based gene cloning [34]. When using linkage maps for detection and characterization of QTL, in 'genome scanning approaches', the chance of success is a function of the density of markers in the map [17, 22]. However, alternative, map- independent QTL detection methods exist which help to overcome this problem in certain cases [6, 29-311.
Restriction fragment length polymorphisms (RFLPs) were the first DNA markers proposed for construction of genetic maps [5]. Since then, RFLPs have been used to construct linkage maps in a variety of species of animals and plants [re- viewed in 21 ]. More recently a new type of DNA marker [33, 35], based on the polymerase chain reaction (PCR) [24], has been used for the con- struction of genetic maps and other applications [reviewed in 26].
Linkage maps constructed with DNA markers have been of diploid species. However, a large proportion of angiosperms are polyploid, espe- cially within the grasses [27]. Mapping polyploid species is a more complex task due to the diffi- culty in identifying genotypes by the banding phe- notype. One way to overcome this problem is to use wild diploid relatives, as done to create a RFLP map for potato [4]. Mapping of polyploids with no known diploid counterparts lagged be- hind until Wu et al. [36] proposed a method for direct mapping of any polyploid that has strict bivalent pairing at meiosis. In this case, mapping is based on the segregation of single-dose (SD) or simplex loci [4, 23]. Using this method, Da Silva et al. [11] constructed a SD RFLP (also known as single-dose restriction fragments, SDRFs) linkage map, with 216 SD loci comprising 44 linkage groups, for Saccharum spontaneum 'SES 208' (2n = 64), which is a wild relative of culti- vated sugarcane. Simultaneously, AI-Janabi et al. [1 ] constructed a linkage map of the same acces- sion using arbitrarily primed PCR [33]; the arbi- trarily primed PCR map contains 208 SD poly- morphisms distributed in 41 linkage groups. Estimates of genome size and coverage of these
two studies were in close agreement. Both con- cluded that this species presented polysomic seg- regation, typical of autopolyploids.
Da Silva [10] presented a methodology for mapping duplex (DD) and triplex (TD) loci in autopolyploids and used this methodology to ex- tend the S. spontaneum RFLP map to over 300 markers. Mapping output is higher with PCR- based markers than with RFLPs, and full auto- mation of marker acquisition and analysis is pos- sible with PCR, thereby making PCR-based markers more amenable to use in breeding pro- grams [1, 25, 26]. However, because of the lack of knowledge about the homologous relationships among PCR markers obtained with the same primer, unless Southern hybridization data are acquired, SDRFs are preferred to obtain infor- mation about homology of linkage groups, par- ticularly when using SDRFs in autopolyploids, as S. spontaneum SES 208. Otherwise, the map will be a homolog map, with 2n linkage groups [ 1, 10, 11].
In this study we integrated the arbitrarily primed PCR map [1] with the SDRF map [11] of a wild relative of sugarcane, S. spontaneum, to facilitate the application of DNA markers to sug- arcane breeding. Highly polymorphic SDRF probes and DD and TD markers were used to organize the 64 linkage groups into 8 groups of pairing homologues.
Materials and methods
Plant materials
A common population of 71 progeny derived from a cross between Saccharum spontaneum 'SES 208' (2n = 64) and its doubled haploid, 'ADP85-0068' [12, 20], was used for mapping with both arbi- trarily primed PCR and RFLP markers [see also 1, 11].
Probes and markers
RFLP probes were obtained from a random low- copy sugarcane genomic library (SG probes [8],
and three sugarcane cDNA libraries, CSB, CSC and CSR probes (CSB, cDNAs from sugarcane buds; CSC, cDNAs from sugarcane cell cultures; CSR, cDNAs from sugarcane roots). In addition, heterologous probes from maize (UMC and BNL probes), rice (RZ probes), oats (CD probes) and barley (BC probes) were used [ 11 ]. Methods used for probe preparation, Southern hybridization and autoradiography have been reported [ 11 ]. Arbitrarily primed PCR markers were obtained following the protocol of Sobral and Honeycutt [25]. Methods of DNA extraction [14], primer screening and scoring of polymorphic fragments have been described [ 1, 25].
Map construction
To construct a map hypothesis we used Map- Maker v 3.0 for IBM and v 2.0 for Macintosh [18]. Three different ordering strategies were used: (1) 442 SD markers were analyzed and pairs of linked markers were identified using two- point analysis with LOD -- 4.0 and maximum re- combination fraction (0)= 0.25 M, followed by subsequent ordering of markers using three-point and multi-point analyses; (2) a framework map of 234 SDRFs [11] was used for integration of 208 SD arbitrarily primed PCR polymorphisms [1] to the SDRF map; and (3) 442 single-dose polymorphisms were grouped (LOD = 4.0, 0 = 0.25 M), but first a 'framework' map was estab- lished using multi-point analysis with pairs of markers with LOD scores > 9.0 and 0>0.10 M, after which remaining markers were ordered one- at-a-time followed by matrix correlation analysis after the addition of each new marker. Map units (in cM) were calculated using the Kosambi map- ping function. To map double- and triple-dose RFLPs, we used the methodology presented by Da Silva [10]. First DD and TD fragments were placed on the SDRF map, then linkages between DD and TD fragments and single-dose arbitrarily primed PCR polymorphisms were investigated. Genome size was estimated according to Hulbert et al. [151, which is based on linkages observed on an existing genetic map. The expected propor-
167
tion of the genome covered by this linkage map with n markers randomly placed, was calculated according to Bishop et al. [3].
Results and discussion
A total of 527 DNA markers, comprised of 208 PCR-based SD markers, 234 SDRFs, 41 DD (or 82 markers) and one TD (3 markers) polymor- phism, were mapped onto the sugarcane genome, yielding 64 linkage groups (Figure 1), with 12 (3~) unlinked SD markers (7 SDRFs and 5 SD arbitrarily primed PCR markers), 9 (11~o) un- linked DD markers, and 3 (100~o) unlinked TD markers. Figure 1 shows all SD markers and DD markers for linkage groups 45, 49, and 51. All remaining DD markers have their linkage rela- tionships described in Table 1. Markers were not uniformly distributed across linkage groups: the number of SD markers per linkage group varied from 2 to 16 (Fig. 1). Saccharum spontaneum cln'omosomes are of different sizes [ 16] and these differences may account for much of the differ- ences in marker distribution, as suggested for to- mato chromosomes [28]. Alternatively, there may be genomic regions that are heavily biased toward multiple-dose alleles, which are underrepresented in our data, and these regions may be lacking in cogerage in this map.
Four linkage groups are comprised of only PCR-based markers and nine linkage groups have only RFLP markers. However, a Z 2 goodness of fit test with 63 degrees of freedom showed no significant deviation from a random distribution of PCR-based (63.60 ~S) and RFLP (58.00 ns) markers among the different linkage groups, in- dicating that both kinds of markers are appropri- ate for mapping, in terms of genomic coverage. Therefore, our results suggest that sample sizes of 208 PCR-based markers or 319 SD, DD, and TD RFLPs were not adequate to identify all 64 link- age groups. The use of only SD markers limits mapping to those regions of the genome in which such markers are found. This may increase the number of markers required to detect all linkage groups. If genomic regions exist in which higher-
dosage markers are common, DD and TD frag- ments will be necessary for mapping such regions.
The number of RFLP-specific linkage groups was twice as large as the number of PCR-specific ones. This may be caused in part by only having mapped SD arbitrarily primed PCR polymor- phisms, as three of the RFLP-specific linkage groups are comprised of DD fragments. How- ever, in theory, mapping of DD and TD arbi- trarily primed markers is also possible because the definition of these marker classes is based exclusively on segregation data. It is possible that some DD and TD arbitrarily primed PCR mark- ers might include polymorphic 'fragments' com- posed of more than one molecular species of similar length, thereby inflating the apparent seg- regation ratios so that they appear to character- ize multiple dose markers. This is more of a prob- lem when using agarose gels than when using high-resolution sequencing gels to visualize arbi- trarily primed PCR products. If co-migration of similar-sized amplification products occurs then those markers will show abnormal linkage with other markers. To test this, we performed two- point linkage analysis on a random subset of ten of the 40 arbitrarily primed PCR polymorphisms that showed an 11:3 segregation (characteristic of DD markers). In these analyses, nine out of the ten DD arbitrarily primed PCR markers ap- peared unlinked, suggesting that many result from amplification of unlinked loci giving products of similar molecular size, as ascertained by agarose gel electrophoresis. The methodology used in this study cannot distinguish these kinds of markers from the unlinked true DD markers unless South- ern hybridization is used to determine homology or higher resolution gels could be used to diminish the chance of observing comigration of amplified products of slightly different sizes.
Probes that detected multiple SDRFs (highly polymorphic probes) allowed the identification of putative homologous pairing groups by establish- ing homologous linkage blocks on different chro- mosomes. Pairing homology was also established by multiple-dose markers, once the linkage rela- tionships between these and SDRFs were calcu- lated (see Table 2). For instance, linkage groups
LG1
+.'m Zl .~: 2.8 ~ CSR153C 8~'~,_ k ~OI2A 1.'~j~ "~- UMC027A 2.5~'. ,~ CDO~gA 7. I ~ ~ B
n'n.,,,'-l'k¢" BC 1 7 1 6 B . N _ _ L19. IS ~ 721 ] ' - - ~~---'11 ~ CD 1380B AIS:Is
8 7--Jl • ~ l l "U- - CD1056B "U-- B18,2$
LG54 LG 58
" ~ O15.1s " ~ K11.2.s
" 11 '7'°-II C$C024B A = ~ ' I t ' - M18.6s
" ' = M16.4s 5.7 ~01.~ 8.8 ~ . j ~ CD1380D
LG 28
- - R19.1s
24.9"--
5 . 8 " -- KIO.4s 8.7"-'-:" -- A7.3s
9.3_.~" -- R2,1s " -- CSR0858
9 , 3 ~ " ~ UMC047B
14.5---
CSRI33A
25.2--
~ g17..3s 13,8-"
• n g17,4s 1 0 , 2 - -
• - - 18.2s 11.7--
• ~ H I | . l s 9 . 9 ~ 5.8"--: -- F4.SS
- - G4.3S
LG 61
8.3 " ~ G|7.1s ^ ~*tt--" L20.2s
o.,~ :hE- TI6.2S
9 ' 9 - - ' U - - CSRO85A
LG 48
I 0' I m J J " - CD1380A
LG 49
6.6 . ~ CSR29A
T CSC42A.1
171
172
H O M O L O G Y GROUP 7
LG45 LGSI
0.0 . ~ C~oC037A.1 ~ j . CSC037A.2
H O M O L O G Y GROUP 8
LG 15
- -- CSC005A 10.7--"
" -- $C-004A 10.4-'-
" -- BNL911B 9 , 7 ~
-- CSR015A
16.5-'-
-- SG099A
36.7--
-- ~ 7 B
LG 22
14.1~L H4.1s
8 2~"~" C.SC005B 2'2--::I:L'- CSC024A ' II ~ CD1345C
24. I ~ I I
"U" SGO~B
H O M O L O G Y UNASSIGNED LG 39 LG 47 LG 52 LG 55
~r-oi,.8 = . l - i f ~ - CDO385B I H P9,2s
~ ~60 ~62
17. ' i ~ SG021A SGI40A °1'6--"[I
-b'--- K11.Is
Fig. 1. Map hypothesis for SES 208 constructed by analysis of 442 single-dose (SD) polymorphisms grouped using a minimum LOD score of 4.0 and 0 = 0.25 M (ordering strategy (1) in Materials and methods), and 41 double-dose (DD) polymorphisms (see Table 1 for their linkage relationships). Sixty-four linkage groups were detected using these parameters; these were coalesced into eight homologous groups using DD markers and SD RFLP markers that identified more than one chromosome (Table 1; see also Materials and methods).
X and Y were considered homologous when X contained SDRFs from probes A and B, and Y also contained different SDRFs from the same probes A and B. Conversely, a DD marker show- ing linkage to SDRFs from both groups X and Y also provides evidence of homology. The under- lying assumption for homology detection is that homologous segments are only present on ho- mologous chromosomes, that is, the absence of gene duplication involving non-homologous chro- mosomes. This assumption is likely to be violated in some cases and some false linkages can be expected to result from homologous segments ex- isting on non-homologous chromosomes due to duplications. Table 1 summarizes linkage rela- tionships of highly polymorphic probes and DD markers supporting homology detection. Linkage
equilibrium involving multiple-dose markers was declared by means of a X 2 test for linkage dis- equilibrium. This was performed by arranging the four gametic frequencies in a 2 x 2 contingency table and comparing each frequency to the product of the row and column totals, divided by the overall total [32]. Linkages giving r>0 .5 (a total of 101) were considered false and may be explained by the large number of two-point tests performed: 41 x 442 -- 18,122. At c~ = 0.005, the probability level used, one expects 0.005 x 18 122 = 91 false linkages resulting by chance alone.
Using highly polymorphic D N A probes, we found agreement between homologous groups identified by probes that gave multiple SDRFs, and multidosc markers (DD and TD). With these
Table 2. Linkage analysis and map positions of double dose restriction fragments in relation to single dose markers
173
DDI{F Single I~)ose ~ • = O LG u CSB03IE II9.gs 10.88 0,350 2 ~C004A 18.3s 8.73 0,212 46
only [ 11], showing that these approaches are con- cordant and complementary in identification of pairing partners during meiosis. Furthermore, this genetic approach using molecular markers allows
the basic chromosome number of any polyploid species to be determined, independent of origin or type of chromosome pairing and assortment.
* L G = linkage g roup (see Fig. 1); S D = segregat ion dis tor t ion; )~2 1 df, e = 0.005 = 7.88.
and assortment (random or preferential, typical of auto or allopolyploids, respectively), we com- pared the expected and observed number of link-
ages in repulsion. Under disomic segregation (complete preferential pairing or allopolyploidy), the expected number of linkages in repulsion de-
176
tectable with SDRFs or arbitrarily primed PCR markers (both are dominant, by definition) in a backcross population is a function of r ( = 0 = the recombination fraction). Linkages in repulsion at r = 0 cannot be detected because ADP 85-0068, the female parent, is a doubled haploid derived from the male parent, SES 208, and thus repre- sents one of its gametes. In this situation only one of the markers involved in the linkage will be polymorphic, whereas the other will be present in both parents, that is, not available for mapping. For repulsion linkages at r > 0, the probability to detect the linkage is that of having polymorphic markers involved in the linkage, which can only result from recombination during the formation of the gamete that gave origin to ADP 85-0068. Because only one of the two resulting recombi- nation classes satisfies the condition for linkage detection (i.e., both markers are polymorphic), that is, absence of both markers in ADP 85-0068, that probability is r/2. To arrive at the expected proportion of linkages in coupling:repulsion, one needs to calculate the probability to detect link- ages in coupling. In this situation, the event that will render either marker involved in the linkage unavailable for mapping would be a recombina- tion between one of the markers occurring during the formation of the gametes that gave origin to ADP 85-0068. Because in this case both recom- binant classes would preclude the detection of coupling linkages (one marker would be present in both parents), the probability of linkage detec- tion is 1-r. The probability function for detection of each kind of linkage for 0 < r < 0.25 (the maxi- mum recombination value used in this study), is given by Da Silva [10]:
and
0.25
R = f r2/4 ( 1 )
o
0.25
C= f r - r2/2 (2)
0
where R and C are probabilities for detection of linkages in repulsion and coupling, respectively.
Applying the R:C ratio to the total number of two-point linkages observed, one gets the ex- pected proportion of the two kinds of linkage under the completely preferential (allopolyploid) hypothesis, which can then be used for a Z 2 test of the hypothesis.
Under the random pairing (i.e., autopolyploid) hypothesis, the expected number of linkages in repulsion with a sample size of 71 progeny, as used in this study, is zero [36]. This value was used for the Z 2 test of the autopolyploid hypothe- sis.
To determine the relative proportion of markers in each linkage phase (coupling and repulsion), 590 two-point linkages were examined. All link- ages detected were in coupling phase. No case was detected in which the presence of a band was associated with the absence of another (repul- sion). If SES 208 were an allopolyploid, or ex- hibited disomic segregation (i.e., complete pref- erential pairing, functionally a diploid situation), one would expect to detect r/2 repulsion phase linkages, considering r as the recombination frac- tion for any two markers. Applying equations 1 and 2, one obtains the expected number of repul- sion linkages under the allopolyploid hypothesis. The Z 2 values for the autopolyploid (0 ns) and al- lopolyploid (41.8"*) hypotheses support poly- somic or non-preferential pairing and segregation, indicating that SES 208 behaves like an autopoly- ploid, in agreement with previous studies [ 1, 11 ].
A marker ordering strategy that relied on pair- wise combinations of markers associated with high LOD scores (> 9.0, see Materials and meth- ods for ordering strategies) gave results that agreed with ordering of the entire marker data- base (at LOD = 4.00, 0 = 0.25 M). Map orders were slightly different, for some closely linked markers, if ordering of markers was based on a framework of SDRFs [11]. We accepted the order based on analysis of the entire database of markers and note that for very closely linked markers precise positioning will require crosses between informative individuals.
Using the method proposed by Hulbert et al. [15], two estimates were made, considering SD markers at less than 10 cM and less than 20 cM,
to calculate the size of the genome. The average estimate was 6600 cM, but because we did not detect linkages in repulsion, this figure should be divided by two, giving an average estimate of 3300 cM. Applying the equations proposed by Bishop et al. [3] to our cross population, consid- ering 64 chromosomes with 442 mapped single- dose markers, a genome of size 3300 cM, and setting 'x' to 25 cM (one marker at every 25 cM in the genome), the expected proportion of the genome covered by single-dose markers was 93 ~o.
Considering the estimated genome size, this S. spontaneum linkage map had on average one marker every 6 cM. However, four gaps between 30 and 38 cM are still present on the map. The chance of each 30 cM gap is 0.991527 = 0.008, as- suming that marker placement in the genome is random and using the binomial distribution. If we multiply this chance by (527-30 = 497), we get four, which is the total number of adjacent 30 cM gaps we expect. Therefore, the number of ob- served gaps is as expected under non-preferential placement of markers throughout the genome and binomial distribution. Techniques that allow saturation of selected regions of the genome [13, 19] may be used to saturate these regions.
From an evolutionary perspective, detailed study of the SES 208 map, along with cytologi- cal observations, allowed inference of basic chro- mosome number and type of chromosome pair- ing and assortment. The potential importance of autopolyploidization in plant evolution has only recently been appreciated [27]. However, we are uncertain to what extent the nature of this cross has influenced our conclusions and a further assessment of S. spontaneum crosses is war- ranted. Furthermore, the S. officinarum-S, robus- turn genotypes of Saccharum may display dif- ferent chromosomal behavior at meiosis and possibly even a different basic chromosome complement [2].
Finally, from the breeder's perspective, modern sugarcane breeding progeny are usually three to four generations removed from the seminal inter- specific hybridizations between S. spontaneum (rarely S. robustum) and S. officinarum in Java. In these controlled crosses, it is possible that both
177
preferential and random pairing occurs. One might speculate that in such hybrids chromo- somes might pair randomly with other chromo- somes from the same genome but not at all with chromosomes from the other genome, although they are in the same nucleus. Genetic control of bivalent pairing may be important to increase via- bility of hybrid offspring during the evolution of the Saccharum polyploid complex. However, dur- ing breeding, progeny of interspecific crosses might be selected based on the presence or ab- sence of certain chromosomes, as aneuploidy is commonly observed in artificial hybrids [7].
Acknowledgements
This work is dedicated to The Clear Perception, in memoriam: thank you for showing happiness in your every smile and may we remain forever chil- dren and forever have ever-clearing perceptions of the world that surrounds us.
This work was sponsored by grants from the International Consortium for Sugarcane Biotech- nology to B.W.S.S. and M.S. The rave data used to construct this map have been deposited in the GrainGenes cereal database and are available on Internet. We thank numerous friends that have contributed to this work, in particular Jean- Claude Glaszmann (CIRAD, Montpellier) for fruitful discussions, Gavin Sills (CIBR, La Jolla) and Michael McClelland (CIBR, La Jolla) for critical comments.
References
1. A1-Janabi SM, Honeycutt RJ, McClelland M, Sobral BWS: A genetic linkage map of Saccharum spontaneum (L.) 'SES 208'. Genetics 134:1249-1260 (1993).
2. A1-Janabi SM, Honeycutt RJ, Sobral BWS: Chromo- some assortment in Saccharum. Theor Appl Genet 89: 959-963 (1994).
3. Bishop DT, Cmmings C, Skolnick M, Williamson JA: The number of polymorphic clones required to map the human genome. In: Weir BS (ed) Statistical Analysis of DNA Sequence Data, pp. 181-200 (1983).
4. Bonierbale MW, Plalsted RL, Tanksley SD: RFLP maps based on a common set of clones reveal modes of chro- mosomal evolution in potato and tomato. Genetics 120: 1095-1103 (1988).
5. Botstein D, White RL, Skolnick M, Davis RW: Con- struction of a genetic linkage map in man using restriction fragment length polymorphisms. Am J Hum Genet 32: 314-331 (1980).
6. Bridges W, Sobral BWS: A genealogical approach to identification of quantitative trait loci in plants. In: Sobral, BWS (ed) The Impact of Plant Molecular Genetics. Birkh/iuser, Boston, in press (1996).
8. Burnquist WB: Development and application of restric- tion fragment length polymorphism technology in sugar- cane (Saccharum spp.) breeding. Ph.D. dissertation, Cor- nell University, Ithaca, NY (1991).
9. Burr BJ, Evola SV, Burr FA, Beckmann JS: The appli- cation of restriction fragment length polymorphism to plant breeding In: Setlow JK, Hollaender A (eds) Genetic Engineering, vol. 5, pp. 45-59. Plenum Press, New York (1983).
10. Da Silva J: A methodology for genome mapping of auto- polyploids and its application to sugarcane (Saceharum spp.). Ph.D. dissertation, Cornell University, Ithaca, NY (1993).
11. Da Silva J, Sorrells ME, Burnquist WL, Tanksley SD: RFLP linkage map and genome analysis of Saccharum spontaneum. Genome 36:782-791 (1993).
12. Fitch MM, Moore PH: Haploid production from anther culture of Saceharum spontaneum L. Z Pflanzenphysiol S: 197-206 (1983).
13. Giovannoni J, Wing RA, Ganal MW, Tanksley SD: Iso- lation of molecular markers from specific chromosomal intervals using DNA pools from existing mapping popu- lations. Nucl Acids Res 19:6553-6551 (1991).
14. Honeycutt RJ, Sobral BWS, Keim P, Irvine JE: A rapid DNA extraction method for sugarcane and its relatives. Plant Mol Biol Rep 10:66-72 (1992).
15. Hulbert SH, Ilot TW, Legg EJ, Lincoln SE, Lander ES, Michelmore RW: Genetic analysis of the fungus Bremia lactucae using restriction fragment length polymorphisms. Genetics 120:947-958 (1987).
16. Janaki-Ammal EK: Cytogenetic analysis of Saccharum spontaneum L. 1. Chromosome studies in Indian forms. Indian J Agric Sci 6 :1-8 (1936).
18. Lander ES, Green P, Abrahamson J, Barlow A, Daly MJ, Lincoln SE, Newburg L: Mapmaker: an interactive com- puter package for constructing primary genetic linkage maps of experimental and natural populations. Genomics 1:174-181 (1987).
19. Michelmore RW, Paran I, Kesseli RV: Identification of markers linked to disease resistance genes by bulked seg- regant analysis: a rapid method to detect markers in spe- cific genomic regions using segregating populations. Proc Natl Acad Sci USA 88:9828-9832 (1991).
20. Moore PH, Nagai C, Fitch MMM: Production and evaluation of sugarcane haploids. Proc Intl Soc Sugar Cane Technol 20:599-607 (1989).
21. O'Brien SJ: Genetic Maps: Locus Maps of Complex Ge- nomes. Cold Spring Harbor Laboratory Press, Cold Spring Harbor, NY (1990).
22. Paterson AH, Lander ES, Hewitt JD, Peterson S, Licoln SE, Tanksley SD: Resolution of quantitative traits into Mendelian factors using a complete linkage map of restriction fragment length polymorphisms. Nature 335: 721-726 (1988).
23. Ritter E, Gebhardt C, Salamini F: Estimation of recom- bination frequencies and construction of RFLP linkage maps in plants from crosses between heterozygous par- ents. Genetics 125:645-654 (1990).
24. Saiki RK, ScharfS, FaloonaF, Mullis KB, Horn GT, Erlich HA, Arnheim N: Enzymatic amplification of beta- globin genomic sequences and restriction site analysis for diagnosis of sickle cell anemia. Science 230:1350-1354 (1985).
25. Sobral BWS, Honeycutt R J: High output genetic map- ping in polyploids using PCR-generated markers. Theor Appl Genet 86:105-112 (1993).
26. Sobral BWS, Honeycutt RJ: Genetics, plants, and the polymerase chain reaction. In:. Mulli~ KB, Ferre F, Gibbs RA (eds) The Polymerase Chain Reaction, pp. 304-320. Birkhauser, Boston, MA (1994).
27. Soltis DE, Soltis PS: Molecular data and the dynamic nature of polyploidy. Crit Rev Plant Sci 12:243-273 (1993).
28. Tanksley SD: Mapping polygenes. Annu Rev Genet 27, in press (1993).
29. Templeton AR, Boerwinkle E, Sing CR: A eladistic analysis of phenotypic associations with haplotypes in- ferred from restriction endonuclease mapping. I. Basic theory and analysis of alcohol dehydrogenase activity in Drosophila. Genetics 117:343-351 (1987).
30. Templeton AR, Sing CF, Kessling A, Humphries S: A cladistic analysis of phenotypic associations with haplo- types inferred from restriction endonuclease mapping. II. The analysis of natural populations. Genetics 120:1145- 1154 (1988).
31. Templeton AR, Crandall A, Sing CF: A cladistic analysis of phenotypic associations with haplotypes inferred from restriction endonuclease mapping. III. Cladogram esti- mation. Genetics 132:619-633 (1992).
34. Wicking C, Williamson B: From linked markers to genes. Trends Genet 7:288-293 (1991).
35. Williams JGK, Kubelik AR, Livak KJ, Rafalski JA, Tingey SV: DNA potymorphisms amplified by arbitrary primers are useful as genetic markers. Nucl Acids Res 18: 6531-6535 (1990).
t79
36. Wu KK, Burnquist W, Sorrells ME, Tew TL, Moore PH, Tanksley SD: The detection and estimation of linkage in polyploids using single-dose restriction fragments. Theor Appl Genet 83:294-300 (1992).