Euphytica 38 : 2 47-260 (1988)© Kluwer Academic Publishers, Dordrecht - Printed in the Netherlands

Genetic strategies to determine the mode of 2n egg formation in diploidpotatoes

D.S. Douches and C.F. QuirosDepartment of Vegetable Crops, University of California, Davis, CA 95616, USA

Received 28 April 1987 ; accepted in revised form 15 July 1987

Key words: Solanum, potato, half-tetrad analysis, isozymes, 2x-4x, megasporogenesis .

Summary

In this study, nineteen diploid potato clones (Solanum spp . 2n = 2x = 24) were identified as 2n eggproducers on the basis of fruit set in 2x-4x crosses . The segregation of three genes mapped close to thecentromere, Got-1 (1 .1 cM), Pgm-2 (2.0 cM), and Sdh-1 (8 .3 cM), were analyzed in the tetraploid offspringin these 2x-4x crosses to discriminate between First Division Restitution (FDR) and Second DivisionRestitution (SDR) modes of 2n egg formation . The co-dominant nature of these markers lead to moreprecise estimates of the recombinational frequencies as a result of completely classifying the segregatingprogenies . 2x-4x data revealed a predominance of SDR mechanisms occurring in 20 of the 21 familiesanalyzed . With a SDR mode established, half-tetrad analysis (HTA) of four distal loci, 6-Pgdh-3, Mdh-1,Pgi-1, and Aps-1, revealed two SDR segregation patterns in some of the families . One pattern fit theexpectations for the distal arm position. The gene-centromere map distances based upon SDR modes in thefamilies following this pattern, were generally close to 4x-2x (FDR) estimates suggesting similar recombina-tion rates between micro- and mega-sporogenesis . Heterozygosity transmission, on average, was 39 .1% . Inthe other segregation pattern, in which the diploid parents were derived from S. chacoense PI 230580, higherthan expected homozygosity levels were found in the female 2n gametophyte populations . A post-meioticdoubling of the reduced megaspore, which generates homozygous 2n eggs, is suggested to operate in threefamilies. The common genetic background of the diploid clones suggested a heritable nature of thismechanism. Pooled data from these three deviant families calculated that 1 .8% of the heterozygosity wastransmitted to the tetraploid progeny .

It is concluded that utilization of seven enzyme-coding loci, with previously established gene-centromeremap distances, in 2x-4x crosses improved half-tetrad analysis (HTA) as a means to determine the mode of 2ngamete formation in megasporogenesis and megagametogenesis of diploid Solanum species .

Introduction

synthesize highly heterozygous clones and to pro-duce uniform true seed potato cultivars (Mendibu-

Numerically unreduced or 2n gametes offer a ru & Peloquin, 1977 ; Peloquin, 1982). It is impor-unique approach to maximizing heterozygosity in tant to identify the mechanisms of 2n gamete for-polyploid species. Their discovery in diploid potato mation occurring in the diploid parent since thespecies (Group Tuberosum 2n = 2x = 24) has led level of heterozygosity transmitted can be extreme-to the proposal of various breeding schemes to

ly different . In general, 2n gametes formed by first

248

division restitution (FDR) are considered superiorto those formed by second division restitution(SDR) because, in the former, most of the hetero-zygosity that is present in the diploid parent istransmitted to the tetraploid progeny (Mok & Pe-loquin, 1975) . This superiority of FDR 2n pollenhas been well established in 4x-2x crosses (Quinn& Peloquin, 1973; Mok & Peloquin, 1975; Mendi-buru & Peloquin, 1977; Tai & DeJong, 1980 ;McHale & Lauer, 1981) .

Cytological identification of the meiotic mecha-nisms which generate diplandroids, is routine anddiagnostic, hence, most of the information con-cerning 2n gamete mechanisms has been derivedfrom studies of microsporogenesis (Mok & Pelo-quin, 1975 ; Ramanna, 1979 ; Veilleux, et al ., 1982) .Direct information concerning the mechanisms of2n egg formation has been limited because of thedifficulties of cytologically studying megasporoge-nesis. However, recently improved cytologicaltechniques have begun to examine megasporoge-nesis and megagametogenesis, and mechanisms in-volved in diplogynoid formation (Stelly & Pelo-quin, 1984; Jongedijk, 1985) .As an alternative, Mendiburu and Peloquin

(1979) proposed that if a chromosome markercould be identified close to its centromere, themode of 2n egg formation could be determinedthrough half-tetrad analysis (HTA) of the segre-gating marker in the progeny . This approach re-sulted in varying degrees of success in the potato(Ross & Langton, 1974; Mok & Peloquin, 1975 ;Iwanaga & Peloquin, 1979 ; Mok, 1981; Stelly &Peloquin, 1986). In these cases, the limitations ofHTA were strongly influenced by the chromosomearm position of the segregating marker in the 2x-4xcrosses. To improve HTA as a means to studydiplogynoid formation, new markers closer to thecentromere need to be mapped . Douches & Quiros(1986b) recently determined gene-centromere mapdistances for ten co-dominant electrophoreticmarkers in the potato . These enzyme-coding locioffer a more precise testing of the theoretical ex-pectations for FDR and SDR segregations in 2x-4xcrosses .

In this paper we report on the identification of anumber of a 2n egg producing diploid clones that

were heterozygous for these markers . The segre-gation of three proximally positioned enzyme-cod-ing loci were used to discriminate between FDRand SDR modes of 2n egg formation through 2x-4xcrosses, while the segregation of the more distalloci were also examined to gain insight into thesemechanisms. In addition, heterozygosity levelstransmitted through SDR 2n eggs were comparedto theoretical expectations .

Material and methods

2x-4x crosses : The diploid clones P129-4 (S. tuber-osum L. X S. berthaltii Hawkes) and P178-1 (S .tuberosum x S. tarijense Hawkes) kindly suppliedby Dr. S. Jansky (North Dakota State University)and DM56-4 supplied by Dr. F.L. Haynes Jr .(North Carolina State University), were crossed tothe tetraploid clones NDD277-2, Nooksack, andmaintained by Dr . R.E. Voss (University of Cali-fornia, Davis) . Additional diploid parents werespecies selections chosen for their ability to set seedin 2x-4x crosses . These included S . chacoense Bitt .,S. kurtzianum Bitt. et Witt ., S . phureja Juz. etBuk., and S. tarijense which were obtained fromthe Potato Introduction Station, Sturgeon Bay,Wisconsin . In 1985, ten diploid selections wereidentified as 2n egg producers . In 1986, three addi-tional 2n egg producing clones were selected fromthe cross 106-2-1 x DM56-4, while four otherswere selected from an open pollinated berry of111-2-1 . Table 1 lists these selections .

Seed from 2x-4x crosses were treated with 1000ppm gibberellic acid to break dormancy . Germi-nated seedlings were transplanted into trays (50plants per tray) . Chromosomes of off-types werecounted to rule out diploids and triploids (Raman-na, 1979) . In four to six weeks, leaf samples fromhealthy and vigorously growing seedlings weresampled for starch gel electrophoretic analysis . Toassay for Aps-1, a tuber specific locus, the seedlingswere allowed to tuberize in the trays . Small maturetubers were harvested in three to four months foranalysis .

We assayed the diploid and tetraploid parentalclones in 11 enzyme systems which revealed 14

enzyme-coding loci: Glutamate oxaloacetate trans-aminase (GOT), Phosphoglucomutase (PGM),Peroxidase (PRX), Triose phosphate isomerase(TPI), Diaphorase (DIA), Shikimic acid dehydro-genase (SDH), 6-Phosphogluconic acid (6-PDGH), Malate dehydrogenase (MDH), Phos-phoglucose isomerase (PGI), Isocitrate dehydro-genase ((IDH), and Acid phosphatase (APS) asdescribed by Quiros & McHale (1985) and Douch-es & Quiros (1986a ; 1986b) . Gene-centromere mapdistances were previously estimated by Douches &Quiros (1986b), except for Tpi-1 . Specific electro-phoretic and enzyme staining procedures are de-scribed by Quiros (1981) and Vallejos (1983) re-spectively. Table 1 lists the diploid clones and theheterozygous loci identified for each individual .

Half-tetrad analysis : HTA is ideally ascertained bya locus tightly linked to its centromere (0% recom-bination frequency) . The expectations of FDR andSDR segregation in diploids are the inverse of

Table 1 . 2n egg producing diploid Solanum selections with genotypes for eight isozyme loci .

Diploid clone

Parentage'

Heterozygous isozyme loci

106-1

chc, PI 230580

+

-

-

+

+106-1-9

chc, PI 230580

+

+106-2-1

chc, PI 230580

+

-

-

+

+106-2-6

chc, PI 230580

-

-

-

+

+106-3-4

chc, PI 230580

+

+

-

+

+108-1-5

chc, PI 265576

+

-

-

+

-111-2-1

chc, PI 320283

-

-

+

+

-111-3-1

chc, PI 320283

-

-

+

+

-113-1-1

chc, PI 320294

-

-

+

+113-2-5

chc, PI 320294

-

-

+

+

-P129-4

hap x tar

+

-

-

+

+P178-1

hap x her

-

+

-

+

-86SD34-2

106-2-1 x DM56-4 +

-

+

+

+86SD34-6

106-2-1 x DM56-4 +

-

+

-

+86SD34-8

106-2-1 x DM56-4 +

-

+

-

+86SD43-1

111-2-1 OP

-

-

+

-

+86SD43-3

111-2-1 OP

-

-

+

+

+86SD43-5

111-2-1 OP

-

-

+

+

+86SD43-6

111-2-1 OP

-

-

+

+

+

'Species abbreviation in Huaman & Ross (1985) .+ = heterozygous .- = homozygous .

Got-1

Pgm-2

Sdh-1

6-Pgdh-3 Mdh-1

Tpi-1

each other, under this model . If FDR occurs, alltetraploid progeny would be simplex heterozy-gotes. When SDR occurs, the progeny would beequally divided between the nulliplex and duplexclasses . A mixture of the three genotypes for thislocus would be found if both FDR and SDR 2n eggswere operating .

Three loci, Got-1, Pgm-2, and Sdh-1, were iden-tified by 4x-2x (FDR) crosses to be proximal totheir centromeres (Douches & Quiros, 1986b) (Ta-ble 2) hence, approaching the ideal HTA case .Observed multinomial data for these three loci re-sulting from 4x-2x (FDR) crosses, were used toestablish probabilities for the genotypic classes,shown in Table 2 . The validity of these probabil-ities rests upon the assumption that the gene-cen-tromere map distances were well-estimated andthat recombination levels are similar in both sexes .From pooled family sizes of 624, 1122, and 396 weestimated the centromeric linkages for Got-1,Pgm-2, and Sdh-1, respectively, and considered

+

-

+

Pgi-1

Aps-1

249

250

these to be unbiased estimates (Douches & Quiros,1986b, 1988) . Chi-square tests were performed totest the segregation ratios observed at these loci in2x-4x crosses under FDR and SDR modes . Forsmall family sizes (n-8), Baye's Rule was used asan alternative to calculate posterior probabilitiesfor the SDR and FDR segregations from the ob-served data (Mendenhall, 1986) .

To calculate gene-centromere map distanceswith an SDR mode operating, the following formu-la was used :Gene-centromere map distance =

.50-freq(nulliplex + duplex)/2 x 100 cM(adapted from Mendiburu and Peloquin, 1979 .)

Transmission of heterozygosity : Estimates werebased upon the segregation patterns of the largerfamilies, which were divided into two subpopula-tions : 1) expected SDR segregation, and 2) deviantSDR segregation. The percent heterozygosity wascalculated for each locus within these subpopula-tions. To obtain a value of average amount of het-erozygosity transmitted through SDR 2n eggs, wecombined the data over all loci as follows :average heterozygosity =Simplex genotype/(Nulliplex + Duplex + Sim-plex) .

Results

Selection of 2n egg producing diploid clones : Ap-

Table 2. Expectations of genotypic frequencies for FDR and SDR modes of 2n gamete formation in 2x-4x crosses for three loci locatedclose to the centromere .

'Based upon 4x-2x (FDR) crosses (Douches & Quiros, 1987b, 1987c) .

proximately 100 seedlings from 20 diploid potatospecies accessions (S . chacoense, S. kurtzianum, S.phureja, S. tarijense) and other selected clones de-scribed in materials and methods were sampled for2n egg production using fruit set in 2x-4x crosses asa criteria for selection . Fruit set ranged from 65 tozero berries ; however, fruit set was not a goodindicator of seed yield (r = 0 .264) . In general, seedyield on a seed per fruit basis was low with a rangeof 5 .1 to 0.1 seeds per fruit and a mean of 1 .6 (Table3) .

Assays for 14 enzyme-coding loci revealed ninepolymorphic loci in the selected diploid parents :Aps-1, Got-1, Idh-1, Mdh-1, 6-Pgdh-3, Pgi-1,Pgm-2, Sdh-1, and Tpi-1 . When heterozygosity wasascertained for these loci, most of the respectiveclones were used directly for 2x-4x crosses . Of thediploid clones tested, 95% were heterozygous forat least one of the three proximally positioned loci .Got-1, Pgm-2, and Sdh-1 were heterozygous, on anindividual basis, in 43%, 14% and 52% of theclones, respectively. The 6-Pgdh-3 and Tpi-1 lociwere the most polymorphic with 67% of the diploidclones being heterozygous, while 53% were heter-ozygous for Mdh-1 . The Pgi-1 locus was hetero-zygous in only 14% of the plants, while the tuberspecific locus, Aps-1, was heterozygous in two of 13(15%) parents assayed . The Idh-1 locus was heter-ozygous in 14% of the plants, but all the tetraploidparents were also heterozygotes, thus, the segre-gation of this locus was not amenable to HTA in the2x-4x crosses .

Locus Gene-centromere map distances Mode Genotypes

Nullipex Duplex Simplex

Got-1 1 .10 FDR .005 .005 .99SDR .495 .495 .010

Pgm-2 2.00 FDR .010 .010 .980SDR .490 .490 .020

Sdh-1 8.30 FDR .041 .041 .918SDR .459 .459 .082

Segregation of proximal loci in large families : Eight2x-4x families were analyzed (Table 4) . The Got-1,Pgm-2, and Sdh-1 loci were segregating in 4, 1, and5 of the families respectively . In addition, two fam-ilies (86SD35 and 87SD4) segregated simultane-ously for two of these loci . Specific triallelic condi-tions were found for each of these three segregatingloci and are described below .

For the Pgm-2 locus (Fig . 1), a monomeric en-zyme, the triallelic condition resulting after the2x-4x cross was more ideal than the typical diallelicmodel that is referenced for HTA (Mendiburu &Peloquin, 1979; Mok, 1981) . Unlike HTA utilizinga dominantly controlled marker, the use of thePgm-2 locus insured complete classification of thesegregating progeny . In family 86SD35, the diploidparent was heterozygous for alleles Pgm-2' andPgm-23 , while NDD277-2 was homozygous for thePgm-22 allele . Considering the proximity of thislocus to the centromere (2 .0 cM), a preponderanceof three-banded phenotypes would indicate FDR,while a 1 : 1 ratio of the duplex genotypes

Table 3 . Fruit set and seed yield of 2n egg producing diploids in 2x-4x crosses .

(Pgm-2'2'2222 and Pgm-22222323 ) recognized as two-banded phenotypes would manifest SDR (Fig .1B) . A comparison of chi-square probabilities forthe FDR and SDR expectations demonstrated bythe high frequency of homozygous gametes that 2neggs generated by the diploid clone 106-3-4 orig-inate via SDR (Table 4) .The Got-1 locus (Fig . 2), which has the tightest

linkage to a centromere (1 .1 cM) of all the proximalloci tested, segregated in four 2x-4x families :86SD35, 86SD40, 86SD41, and 87SD4 (Table 4). Atriallelic segregation was found in the resulting 4xprogenies . The tetraploid parent of all these cross-es, NDD277-2, was heterozygous for this locus(Got-1 31 31 41 4 ), while all four diploid parents had aheterozygous genotype Got-1 41 5 . The three expect-ed genotype classes for the gametes of the diploidparents were Got-1 41 4, Got-1 515 , and Got-1 41 5 . Witha common allele between the two parents, discrimi-nation between the progeny classes was based uponthe presence, absence, or dosage of the Got-I5 al-lele . The genotype Got-1 41 5 would be expected if an

NDD = NDD277-2; NA = Not available ; mean = 1.6 Seeds/Fruit ; Standard error = 0 .32 ; r = 0 .514 8 .'Correlation between fruit set and seed set.

25 1

Family Cross Fruit Seed Seed/Fruit

86SD74 P178-1 x NDD 20 9 0.4586SD73 P129-4 x NDD 20 8 0.4086SD77 P129-4 X Nooksack 3 2 0.6786SD70 106-1-3 x NDD 33 3 0.0986SD36 106-1-9 x NDD 12 4 0.3386SD40 106-2-1 x NDD 55 195 3 .5586SD41 108-1-1 x NDD 4 12 3 .0086SD71 106-2-6 x NDD 4 14 3.5086SD35 106-3-4 x NDD 18 46 2.5686SD75 111-2-1 x NDD 65 14 0.2286SD72 111-3-1 x NDD 15 7 0.4786SD22 113-2-5 x NDD NA 13+ NA86SD42 113-1-1 x NDD NA 76+ NA87SD 1 106-1 x NDD 6 7 1 .1787SD2 86SD34-2 x NDD 5 6 1 .2087SD3 86SD34-6 x NDD 17 17 1 .0087SD4 86SD34-8 x NDD 23 60 2.6187SD5 86SD43-1 x NDD 5 6 1.2087SD6 86SD43-3 x NDD 16 13 0.8187SD8 86SD43-5 x NDD 9 18 2.0087SD9 86SD43-6 x NDD 16 81 5.06

252

FDR mode of 2n egg formation was occurring,while a SDR segregation would give in an equalproportion of Got-1414- and Got-151 5-genotypes .Chi-square tests for fit to FDR expectations havehighly significant deviations (P- .0001). Mean-while, chi-square tests for fit to SDR expectationsdid not deviate significantly in three of the fourfamilies. The lack of fit for the family 86SD40 wasdue to an excess of the Got-1515-genotype in com-parison to the Got-1 41 4-genotype (Fig . 2B) . Hence,SDR was concluded to occur in these clones . Het-erogeneity between these four families precludedpooling segregation data for the Got-1 locus, how-

Table 4 . Segregation of various isozyme markers in 2x-4x crosses in large families and its applications to determine the mode of 2n eggformation and gene-centromere map distances .

ever, the lack of the Got-141 5-genotype (none in 254offspring) also reaffirmed a tight centromeric link-age for the Got-1 locus .

Five 2x-4x families (86SD22, 86SD42, 86SD42,87SD4, 87SD8, and 87SD9) segregated for theSdh-1 locus (8 .3 cM) (Fig . 3) . In all crosses,NDD277-2 was used as the pollen parent . The het-erozygous nature of the Sdh-1 locus (Sdh-121 2151 5)in the tester parent did not hinder out ability todiscriminate between the homozygous and hetero-zygous gametes produced by the diploid parents .Each family ratio fit SDR expectations while at thesame time deviating significantly from FDR expec-

Obtained from Douches & Quiros (1986b ; 1988c) ; b Nulliplex and duplex refer to genotypes formed from homozygous 2n eggs ; simplexrefer to genotypes formed from heterozygous 2n eggs ; `Probabilities based upon chi-square tests ; d Simplex and duplex classes werepooled .

Family Diploidparent

Genotype 4x-2x(FDR) Mapdistances

Genotypes° ProbSDR`

ProbFDR

Mode of2n eggformation

Mapdistance

Nulliplex Duplex Simplex

86SD22 113-2-5 Sdh-1'1 3 8 .3 19 23 0 0.91 0.00 SDR 0.006-pgdh-3'3 2 30 .1 16 13 49 30.9

86SD41 108-1-5 Got-1 3 15 1 .1 3 6 0 0.82 0 .00 SDR 0.006-pgdh-3'3 2 30 .1 2 0 7 38 .9

86SD42 113-1-1 Sdh-1'1 3 8 .3 7 6 0 1 .00 0.00 SDR 0.006-pgdh-3'3 2 30 .1 2 1 13 40 .6

87SD8 86SD43-5 Sdh-1 1 1 3 8 .3 9 3 0 0.40 0 .00 SDR 0.006-pgdh-3'32 30 .1 7 8 22 29 .7Mdh-1 3 1 6 33 .5 5 8 23 31 .9

87SD9 86SD43-6 Sdh-1'3' 8 .3 21 15 0 0 .73 0.00 SDR 0.006-pgdh-3'32 30 .1 7 8 22 29 .7Mdh-1 3 16 33 .5 5 8 23 31 .9

86SD40 106-2-1 Got-131 5 1 .1 65 118 0 0.002 0.00 SDR 0.006-pgdh-3'32 30 .1 109 58 4 1 .2Mdh-1 3 1 5 33 .5 86 85 3 0 .9Tpi-1'1 3 98 81 4 1 .1Aps-1'12 13 .4 65 51 2 1 .1

86SD35 106-3-4 Pgm-21 2 3 2 .0 19 12 0 0.60 0.00 SDR 0.00Got-131 5 1 .1 12 14 0 0.99 0.00 SDR 0.006-pgdh-3'32 30.1 17 11 2 3 .3Mdh-1 31 5 33 .5 9 6 1 1 .9Tpi-1 1 12 7 17 2 3 .8

87SD4 86SD34-8 Got-13 1 5 1 .1 12 24 0 0.20 0.00 SDR 0.00Sdh-1 1 13 8.3 15 21 0 0.73 0.00 SDR 0.00Pgi-1 2 13 26.0 13 23 0 0.00Mdh-1 31 5 33 .5 23 13d 0.00Tpi-1'1 2 19 14 3 4.17

253

Fig. 1 . Segregation of Pgm-2 locus (2 .0 cM) . Anodal direction is above . A . FDR segregation is family 86SD30. P = tetraploid parent(NDD277-2) (Pgm-2222222 2 ), while the rest are simplex genotypes . B . First five lanes showing SDR segregation with 1 :1 ratio of theduplex genotypes Pgm-2 22223 2 3 (DI) and Pgm-2'2'2222 (D2 ) in the family 86SD37 . In Figs 1-3 the upper photograph represents a 4x-2x(FDR) segregation for each enzyme-coding locus (Douches, 1987) . The origins of the 2x-4x families are found in Table 3 .

Fig. 2. Segregation of Got-1 locus (1 .1 cM) . Anodal direction is above . A . FDR segregation in 86SD19 . The tetraploid (Got-1 3 13 1414NDD277-2) and diploid parents (Got-1 4 1 5 84SD22) are in the lanes to the far right respectively. The rest are simplex genotypes . B . SDRsegregation in 86SD40 for the Got-1 5 allele . In Figs . 1-3 the upper photograph represents a 4x-2x (FDR) segregation for eachenzyme-coding locus (Douches, 1987) . The origins of the 2x-4x families are found in Table 3 .

tations (Table 4) (Fig. 3B) .The five families that segregated for the Sdh-1

locus were pooled to estimate its centromeric link-age (Table 6) . A comparison to SDR expectationsdeviated significantly (X2 = 7.5 P<.005), indicat-ing a tighter centromeric linkage than expected .

Segregation of more distally positioned loci in largefamilies : The 6-Pgdh-3 locus, 30.1 cM from its cen-tromere, segregated in seven of the eight families .These tetraploid progenies were classified intothree genotypic classes : nulliplex, duplex, and sim-plex, based upon the diallelic segregation pattern .

254

Fig. 3 . Segregation of Sdh-1 locus (8.3 cM) . Anodal direction is above . A. FDR segregation in 85SD41. R denotes recombinantoffspring . B . SDR segregation in 87SD8 . Note the 1 : 1 segregation of Sdh-1 3 allozyme band (slowest migrating) . T denotes a Sdh-1 3 131 21 5genotype, while D is an example of Sdh-1 2 1 5-phenotype . In Figs . 1-3 the upper photograph represents a 4x-2x (FDR) segregation foreach enzyme-coding locus (Douches, 1987) . The origins of the 2x-4x families are found in Table 3 .

Two types of segregation patterns were observedfor this locus : 1) a SDR segregation typical for itsdistal arm position, and 2) SDR segregation as if itwas proximally positioned . The first segregationpattern was observed in families 86SD22, 86SD41,86SD42, 87SD8, and 87SD9, which are known toproduce 2n eggs by SDR as suggested by the segre-gation of the proximal loci (Table 4) . A pooledestimate of 152 progeny allowed us to detect a30.9 cM gene-centromere distance for 6-Pgdh-3,based on SDR expectations (Table 6) . This valuewas not significant from the distance determined by4x-2x (FDR) crosses (X 2 = 0 .177 P = .90) . Whenusing the 6-Pgdh-3 locus to attempt to predict themode of 2n egg formation, the segregation dataalso marginally fit FDR (X2 = 5 .08 P = .08) . In thefamilies 86SD35 and 86SD40, this locus segregatedin a SDR pattern typical of proximal position .Hence, the data fit poorly FDR and SDR expecta-tions assuming normal recombination levels . Agene-centromere recombination frequency, basedupon SDR, was calculated to be 1 .5% in the twodeviant families (Table 6) .The Mdh-1 locus, which is 33 .5 cM from its cen-

tromere, segregated in five 2x-4x families (Table4) . Two new Mdh-1 alleles were identified in thediploid parents of these crosses . The clones 106-2-1and 106-3-4 carried the Mdh-15 allele, while86SD43-5 and 86SD43-6 carried the Mdh-16 allele

(Fig. 4). The upper band of the two-banded Mdh-I5allele migrated more slowly than the upper band ofMdh-13, while the upper band of the two-bandedMdh-16 allele migrated further than the upper bandof the Mdh-13 allele . NDD277-2 was homozygousfor the Mdh-12 allele, which overlaps with the slow-er-migrating band of Mdh-13 , thus a triallelic segre-gation pattern was observed in the tetraploid pro-genies . Analysis of the progenies for this locus wassimilar to the reasoning used for the Pgm-2 locus .The gametic types were completely classified onthe presence and absence of Mdh-1 alleles in thefour families generated from these parents .

The segregation patterns of Mdh-1 for these fam-ilies were separated into two groups in the samemanner as that for 6-Pgdh-3 . The Mdh-1 locus inthe pooled data from the two families, 87SD8 and87SD9, segregated according to SDR expectations .A 29 .2 cM centromeric distance was estimatedfrom a total progeny size of 48 (Table 6) . Thisestimate was not significantly different from thoseobtained by 4x-2x (FDR) crosses . The segregationof the Mdh-1 locus (33 .5 cM) could not discrimi-nate between the two modes since expectations arepractically the same for both FDR and SDR (X 2 =2.5 P = .33) . In contrast, from the pooled segre-gation data of progenies 86SD40 and 86SD35, a1 .1% recombination rate was observed for theMdh-l-centromere chromosome segment on thebasis of SDR (Table 6) .

255

Two other loci that are loosely linked to their through 4x-2x (FDR) crosses . The Pgi-1 locus,respective centromere were segregating in these expected to be 26.0 cM from its centromere, segre-families and had a similar behavior as 6-Pgdh-3 and gated in a SDR manner in the family 87SD4, how-MA-1 . The Aps-1 locus segregated in the family ever, no simplex genotypes were recovered in 3686SD40 . A 1.1% recombination frequency was ob-

offspring .served for this locus, despite having previously esti-mated a 13 .4 cM gene-centromere map distance

Table 5 . Segregation of various isozyme markers in 2x-4x crosses in small families and its application to determine the mode of 2n eggformation .

Obtained from Douches & Quiros (1986b ; 1988c) ; b Nulliplex and duplex refer to genotypes formed from homozygous 2n eggs ; simplexrefer to genotypes formed from heterozygous 2n eggs ; `Probabilities based upon chi-square tests ; d Simplex and duplex classes werepooled .

Family Diploidparent

Genotype 4x-2x (FDR)Map distances

Genotypesb ProbSDR°

ProbFDR

Mode of 2negg formation

Nulliplex Duplex Simplex

87SD1 106-1 Got-1 3 1 5 1 .1 0 2 0 0.9999 0.0001 SDR87SD2 86SD34-2 Sdh-1'13 8 .3 1 2 0 0.9993 0.0007 SDR

Pgi-1 2 13 26 .0 1 2 087SD3 86SD34-6 Got-1 3 1 5 1 .1 3 0 0 0.9997 0.0003 SDR

Sdh-11 1 3 8 .3 2 1 0Pgi-1 2 13 26 .0 1 2 0Mdh-1 31 5 33 .5 2 1 0

87SD5 86SD43-1 Sdh-1'1 3 8 .3 4 2 0 0.9999 0.0001 SDRTpi-1'12 3 0 3Mdh-1 3 1 6 33 .6 4 2d

87SD6 86SD43-3 Sdh-1 1 13 8 .3 3 2 0 0.9993 0.0007 SDRMdh-1 31 6 33 .5 1 0 4Tpi-1'1 2 1 3 16-Pgdh-3'3 2 30 .1 2 0 3

86SD37 106-3-4 Pgm-2'2 3 2 .0 3 2 0 0.9999 0.0001 SDR86SD74 P178-1 Pgm-2'22 2 .0 2 3 0 0.9999 0.0001 SDR

Tpi-1'12 1 1 36-Pgdh-3'3 2 30.1 1 2 2

86SD36 106-1-9 Pgm-2'2 2 2 .0 1 1 0 0.9997 0.0003 SDRGot-131 5 1 .1 1 1 0

86SD75 111-2-1 Sdh-1 1 13 8 .3 2 4 0 0.9999 0.0001 SDR6-Pgdh-32 3 3 30 .1 2 0 3Tpi-1'1 2 2 0 3

86SD72 111-3-1 Sdh-1'13 8 .3 2 3 0 0.9993 0.0007 SDRTpi-1'1 2 3 8 0

86SD38 106-2-6 6-Pgdh-3`3 2 30 .1 7 3 0 0.4999 0.5001 SDR?Mdh-1 3 15 33 .5 6 5 0Tpi-1'1 2 3 8 0

86SD39 106-2-6 6-Pgdh-3'32 30.1 1 3 2 0.4244 0.5756 SDR?Mdh-1 3 15 33 .5 5 2 0Tpi-1'1 2 5 0 2

86SD73 P129-4 6-Pgdh-3'32 30 .1 2 0 2 0.6666 0.3334 Mixture?Tpi-1'1 2 3 0 1Mdh-1 2 1 4 33 .5 2 2d

256

HTA of sparse family data : Thirteen 2x-4x familiesof small size were analyzed using Bayes' Rule (Ta-ble 5) . Three families each segregated for thePgm-2 and Got-1 loci respectively, one of thesefamilies segregating for both (86SD36) . The Sdh-1locus segregated in six of the families while onesegregating for both Got-1 and Sdh-1 (87SD3) .

In ten of the 13 families, the discrimination be-tween SDR and FDR modes was straightforwardand simple. All these probability values were basedupon the segregation of at least one of the threeproximally positioned loci . A SDR mode was con-cluded for these ten families with the data summa-rized in Table 5 .

The diploid parent, 106-2-6, involved in families86SD38 and 86SD39, was homozygous for thethree proximal loci, but was heterozygous for thedistal loci 6-Pgdh-3 and Mdh-1 . The 2x-4x segre-

Table 6 . Heterozygosity transmission through 2x-4x crosses and comparison of gene-centromere distances estimated by 4x-2x and 2x-4xcrosses .

a) Normal 2x-4x families

b) Deviant 2x-4x families (86SD40, 86SD35, 87SD4)

Obtained from Douches & Quiros (1986b, 1988c) . NA = not available .

gations based upon these two loci could not dis-criminate between a SDR and FDR mode in thefamilies 86SD38 and 86SD39 (P = .4999 and .4244respectively) .

Pooled family data for the 6-Pgdh-3, Pgi-1, andMdh-1 loci revealed a progeny distribution throughthe three genotypic classes (Table 5) . The progenysizes were too small to calculate accurate gene-centromere values, however, two families, 87SD2and 87SD3, were unique in that they lacked sim-plex genotypes for the 2x-4x segregations of thePgi-1 and Mdh-1 loci .

Gene-centromere relationship of Tpi-1 locus : Segre-gation of the Tpi-1 locus in six 2x-4x families gavesome insight into the centromeric relationship ofthis previously unreported enzyme-coding locus(Fig. 5) . The pooled data from 865D72, 86SD75,

Locus Number offamilies

Nulliplex Duplex Simplex Percentheterozygosity

(SDR) 2x-4xMap distance

4x-2x (FDR)Map distancea

Got-1 3 89 156 0 0 .0 0.0 1 .1Pgm-2 1 19 12 0 0.0 0.0 2 .0Sdh-1 1 15 21 0 0 .0 0.0 8 .3Aps-1 1 65 51 2 1 .1 0.6 13 .4Pgi-1 1 13 23 0 0 .0 0.0 26.06-Pgdh-3 2 126 69 6 3 .0 1 .5 30.1Mdh-1 2 118 114 4 1 .7 0.9 33 .5Tpi-1 3 124 112 9 3 .7 1 .9 NA

Total 569 558 21 1 .83

Locus Number offamilies

Nulliplex Duplex Simplex Percentheterozygosity

(SDR) 2x-4xMap distance

4x-2x (FDR)Map distances

Got-1 1 3 6 0 0.0 0 .0 1 .1Sdh-1 4 56 47 0 0.0 0.0 8.36-Pgdh-3 5 30 28 92 61 .3 30.9 30.1Mdh-1 2 8 12 28 58 .3 29 .2 33 .5Tpi-1 1 4 2 6 50 .0 25 .0 NA

Total 101 95 126 39.1

86SD74, 87SD5, 87SD6, and 87SD8 suggests amore distal chromosome arm position based upona SDR mode of 2n egg formation, however, thesmall total progeny size of 33 precluded any precisechromosome arm location . Tpi-1 also segregated inthe three deviant families 86SD35, 86SD40, and87SD4 . Pooled data for the Tpi-1 locus had a segre-gation pattern that was similar to the other distalloci in these deviant families . Few simplex geno-types were observed . Pooled data of 255 offspringestimated a 1 .8% recombination frequency on thebasis of SDR .

Heterozygosity transmission through 2x-4x crosses :Estimates of gene-centromere map distances aresimply another way of expressing the recombina-tion frequency for that chromosome segment . The2x-4x segregation data from this study offered anopportunity to examine the ability of the female 2ngametophyte to transmit its heterozygosity throughan SDR mechanism. This analysis was limited to

257

Fig . 4. SDR segregation for Mdh-1 locus (33 .5 cM) . Anodal direction is above . A. Progeny from 86SD40 segregating 1 :1 forMdh-1 2 12 1 51 5 (M) and Mdh-1 2 121 31 3 (M2 genotypes) . B. Progeny from 87SD9 segregating in a SDR manner. S, T, and F denoteMdh-1 2 12 1 31 3 , Mdh-12121 3 1 6, and Mdh-1 21 2 1616 genotypes respectively .

Fig. 5. SDR segregation for Tpi-1 locus in 87SD4 . D and N denote the Tpi-1'1'121 2 and Tpi-1 212 1212 genotypes respectively.

the large 2x-4x families. The isozyme data weresubdivided into two gametic populations for thisanalysis: normal recombination vs . deviant SDRsegregations (Table 6) .

The gametic population in which the large fam-ilies showed normal recombination levels was be-lieved to be the least biased population since mostcentromeric linkages were within expectations .The SDR gametes transmitted on average 39 .1% oftheir heterozygosity (Table 6) . The other largefamilies (86SD35, 86SD40, and 87SD4) whichshowed abnormal SDR segregations had a muchdifferent gametic population structure . For theeight chromosome segments sampled, only 1 .83%of the heterozygosity was transmitted on average,indicating an almost completely inbred populationof 2n gametes . Almost all the heterozygosity wasfound between rather than within the gametes .

258

Discussion

We conclude from the segregation of these isozymeloci that SDR is more prevalent than FDR in 2n eggformation . Similar conclusions were drawn by Stel-ly & Peloquin (1986) when a different population ofdiplogynoid producing parents was analyzed in2x-4x crosses . Prevalence of SDR mechanisms ispredicted as a result of sequential type of embryosac development that the female gametophyte fol-lows in Solanum (Jongedijk, 1985) . In alfalfa,which has a similar embryo sac development, SDR2n eggs also have been identified cytologically(Pfeiffer & Bingham, 1983) . FDR 2n eggs havebeen previously identified through HTA in potato(Ross & Langton, 1974 ; Mok & Peloquin, 1976 ;Iwanaga & Peloquin, 1979 ; Mok, 1981), however inthese instances, the frequency was very low or incombination with a SDR mechanism. A FDRmechanism comparable to the parallel/fused spin-dles during metaphase II in meiosis during micros-porogenesis is not feasible in megasporogenesisbecause of successive divisions associated with itsdevelopment. Hence, a reduction in chromosomepairing is a prerequisite for FDR 2n egg formationthrough abnormal megasporogenesis (Jongedijk,1985) . Under normal synaptic conditons as seen inthis study, SDR 2n eggs should prevail .

Five of the eight large 2x-4x families segregatedin an expected manner . A SDR mechanism such assecond division omission or premature cytokinesiscould account for the observed results . In thesefamilies, recombination levels were generally closeto expectations extrapolated from 4x-2x (FDR)data, suggesting similar recombination levels in mi-cro- and megasporogenesis . Gene-centromeremap distance estimates were within expectationsfor Pgm-2, Got-1, 6-Pgdh-3, and Mdh-1, while theSdh-1 locus in megasporogenesis showed a slightlytighter linkage than expected (0 .0 vs . 8 .3 cM) .

HTA of the more distal loci estimated severelyreduced recombination rates in the families86SD35, 86SD40, and 87SD4. These discordantresults indicate that a different SDR mechanismother than second division omission was occurring .It is unlikely that a low frequency of simplex geno-types would result from a reduction in recombina-

tion . For instance, a strong expression of a desy-naptic gene would lead to univalents at metaphase Iand an unbalanced first division (Jongedijk, 1985 ;Hermsen, 1984) . Sterility would be extremely highif a SDR mechanism was exclusively operating inthis synaptic genotype, while, only FDR-type ga-metes would be expected to be fertile under thesedesynaptic conditions . The segregation data for theproximal loci in this study did not support an FDRmode of 2n egg formation . Under less extremedesynaptic conditions, a mixture of SDR and FDRgametes could be possible, however, the data didnot support this contention either .

Jongedijk (1985) has suggested that a post-meiotic doubling of reduced megaspores takingplace during gametogenesis could lead to the for-mation of homozygous 2n eggs . Stelly & Peloquin(1986) also noted this mechanism as a potentialsource of excess homozygous genotypes in certain2x-4x crosses . To account for this heterogeneouspopulation of almost completely homozygous 2neggs in our study, it is likely that two separatemechanisms of 2n egg formation occur in thesediploid parents : 1) a high frequency of post meioticdoubling of the reduced megaspore generatingcompletely homozygous 2n eggs and, 2) an omis-sion of the second division generating SDR-typegametes with normal meiosis occurring at a lowfrequency in the gametophytic population .

Transmission of heterozygosity through SDR mech-anisms : The SDR gametes, produced in non-de-viant families, transmitted on the average 39 .1% oftheir heterozygosity (Table 6) . In comparison, the-oretical expectations, based upon a number of cyt-ological assumptions, estimated SDR gametes cantransmit 35-40% of their heterozygosity (Peloquin,1981; Hermsen, 1984) . Our observed value, basedupon six chromosome segments, supports theoret-ical expectations that SDR mechanisms are unableto transmit heterozygosity at high levels .A SDR mechanism which promotes homozy-

gous 2n gametes is less desirable for transmittingheterozygosity in 2x-4x crosses than SDR gametesformed through second division ommission of themegaspore. A mechanism such as post-meioticdoubling of the reduced megaspore, prevents any

heterozygosity from being transferred through thefemale 2n gametophyte (Jongedijk, 1985) . SDRgametes of this type appear undesirable, consid-ering the importance of heterozygosity whenbreeding a tetrasomic autoploid crop like the pota-to. FDR gametes transmit a large percentage of theparental genome intact (81 .5-98.0%), which tendstoward homogeneous populations of highly hetero-zygous 2n gametes (Douches & Quiros, 1988) .SDR eggs generated through a post-meiotic dou-bling mechanism produce a population inbred 2ngametophytes .

The feasibility of producing inbred lines throughSDR mechanisms has been studied in several orga-nisms . Matromorphoric embryos generatedthrough SDR mechanisms have been identified inBrassica spp. (Eenick, 1974a, 1974b) and in S. an-digena (Taylor, 1978), however, heterozygosity hasbeen found in them . In rainbow trout, diploid gy-nogenesis, a potential method to produce inbredlines, has been suggested through retention of thesecond polar body (Allendorf et al ., 1986), or bysuppression of the first mitotic cell division (Pur-dom, 1983) . A system to generate completely in-bred potato clones would be established if these 2ngametes produced through post-meiotic doubling,could be induced to develop parthenogenetically .Parthenogenetic seed development has been rou-tine in generating potato haploids (2n = 2x = 24)from tetraploid clones (Hougas & Peloquin, 1957 ;Montelongo-Escobedo & Rowe, 1969) and has al-so been extended to generating monoploids (n =x = 12) (Van Breukelen et al., 1975). The gener-ation of inbred diploid clones could have importantgenetic and breeding implications ; i .e . inheritancestudies and the use of F t hybrid and double-crossbreeding methodologies . The testing of the efficacyof generating inbred diploid clones through ha-ploid pollinator and tissue culture techniqueswould be of interest .

Value of HTA : The limitations of HTA, in the past,were mainly influenced by the markers available .For example, the dominant nature of chromosomeposition of the yellow tuber flesh locus (Y) limitedits value to determine the composition of the 2ngametophyte populations (Stelly & Peloquin,

259

1986) . The enzyme-coding loci used in this studyexpand HTA as a means to study megasporogene-sis and megagametogenesis . Their co-dominant na-ture allowed the segregating progeny to be com-pletely classified to obtain more precise estimatesof segregation patterns through 2n gamete mecha-nisms. In addition, the identification of markerswith tighter centromeric linkage has improved ourability to discriminate between SDR and FDRmodes even in small progenies . The analytic pow-ers of HTA were greatly improved by following thesegregation of both proximal and distal loci in a2x-4x family. Initially, the segregation of proximalmarkers was used to determine the mode of 2n eggformation. Secondly, the segregation of a distallocus provided information concerning recombina-tion levels within the 2n gametic pool, which hasbeen extremely difficult to assess cytologically . Inthis study, a post-meiotic doubling of the reducedmegaspore was suggested to be operating in at leastthree of the 19 clones, leading to homozygous SDRgametes. Hence, caution should be exercised whenSDR mechanisms are utilized in 2x-4x and 2x-2xbreeding schemes, since heterozygosity transmis-sion can vary greatly (39.1% vs . 1 .8%) .

An analytic system as described above would beuseful in: 1) discriminating between FDR/SDR andexclusive FDR populations of female 2n game-tophytes, and 2) quantifying the asynapsis associ-ated with FDR 2n gametophytes, if sufficient prog-eny sizes can be generated .

References

Alleendorf, F.W., J .E . Seeb, K.L. Knudsen, G .H. Thorgaard& R.F. Leary, 1986 . Gene-centromere mapping in rainbowtrout . J Heredity 77 : 307-312 .

Douches, D.S ., 1987. Development of genetic strategies todetermine the mode of 2n egg formation in diploid potatoes .Ph.D. Thesis, University of California, Davis .

Douches, D .S. & C.F . Quiros, 1986a . Additional isozyme lociin tuber-bearing Solanums : Inheritance and linkage relation-ships . J Heredity (in press) .

Douches, D .S. & C .F . Quiros, 1986b . Use of 4x-2x crosses todetermine gene-centromere map distances of isozyme loci inSolanum species . Can J Genet Cytol (in press) .

Douches, D .S . & C .F . Quiros, 1988 . Genetic recombination ina diploid synaptic mutant and a Solanum tuberosum x S.

260

chacoense diploid hybrid . Heredity 60 : 183-191 .Eenink, A.H., 1974 . Matromorphy in Brassica oleracea L. IV.

Formation of homozygous and heterozygous diploid productsof gametogenesis and qualitative genetical research on matro-morphic plants . Euphytica 23 : 719-724 .

Eenink, A . H . 1975 . Matromorphy in Brassica oleracea L. VII .Research on products of microsporogenesis and gametogene-sis from prickle pollinated plants . Euphytica 24 : 45-52.

Hermsen, J .G .Th., 1984 . Mechanisms and Genetic Implica-tions of 2n-Gamete Formation . Iowa State Journal of Re-search 58(4) : 421-434 .

Hougas, R.W. & S .J . Peloquin, 1957 . A haploid plant of thepotato variety Katahdin . Nature 180 : 1209-1210 .

Huaman, Z. & R. W. Ross, 1985 . Updated listing of potatospecies names, abbreviations and taxonomic status . Am Pota-to J 62 : 629-641 .

Iwanaga, M . & S .F . Peloquin, 1979 . Synaptic mutant affectingonly megasporogenesis in potatoes. J Heridity 70 : 385-389 .

Jongedijk, E ., 1985 . The pattern of megasporogensis and mega-gametogenesis in diploid Solanum species hybrids ; its rele-vance to the origin of 2n eggs and the induction of apomixis.Euphytica 34 : 599-611 .

McHale, N.A. & F .I . Lauer, 1981 . Breeding value of 2n pollendiploid from hybrids and S . phureja in 4x-2x crosses in pota-toes . Am Potato J 58 : 365-374 .

Mendenhall, W., R.L. Schaeffer & D.D. Wackerly, 1986.Mathematical Applications . Third Edition . Duxberry Press,Boston, 62 p .

Mendiburu, A .O. & S .J . Peloquin, 1977a . The significance of2n gametes in potato breeding . Theor Appli Genet 49: 53-61 .

Mendiburu, A .O. & S .J. Peloquin, 1977b . Bilateral sexual pol-yploidization in potatoes . Euphytica 26 : 573-583 .

Mendiburu, A.O. & S .J . Peloquin, 1979 . Gene-centromeremapping by 4x-2x matings in potatoes . Theor Appl Genet 54:177-180 .

Mok, D.W.S. & S .J . Peloquin, 1975a . Breeding value of 2npollen (diplandroids) in tetraploid x diploid crosses in pota-toes . Theor Appl Genet 46 : 307-314.

Mok, D .W.S . & S .J . Peloquin, 1975b . Three mechanisms of 2npollen formation in diploid potatoes . Can J Genet Cytol 17 :217-225.

Mok, D.W.S ., S .J . Peloquin & A .O. Mendiburu, 1976. Geneticevidence for mode of 2n pollen formation and S-locus map-ping in potatoes . Potato Res 19 : 157-164.

Mok, I ., 1981 . Sexual polyploidization and protein diversity inpotatoes . Ph .D. Thesis, University of Wisconsin, Madison .

Montelongo-Escobedo, H. & P.R. Rowe, 1969 . Haploid in-duction in potato : Cytological basis for the pollinator effect.Euphytica 18 : 116-123 .

Peloquin, S .J .,1981 . Breeding methods for achieving phenotyp-ic uniformity. In : The International Potato Center. Report of

the Planning Conference on the Production of Potato fromTrue Seed . Manila Philippines, pp . 151-155 .

Peloquin, S .J .,1983 . New approaches to breeding for the potatoin the year 2000 . In : Proceedings of the International Con-gress on `Research for the Potato in the year 2000' . Interna-tional Potato Center, Lima, Peru, pp . 32-34 .

Pfeiffer, T.W. & E.T. Bingham, 1983 . Abnormal meiosis inalfalfa, Medicago sativa : cytology of 2n egg and 4n pollenformation. Can J Genet Cytol 25 : 107-112 .

Purdom, C.E., 1983 . Genetic engineering by the manipulationof chromosomes . Aquaculture 33 : 287-300.

Quinn, A .A. & S .J . Peloquin, 1973 . Use of experimental tetra-ploids in potato breeding. Am Potato J 50 : 415-420.

Quiros, C.F ., 1981. Starch gel electrophoresis techniques usedwith alfalfa and other Medicago species . Can J Plant Sci 61 :745-749 .

Quiros, C.F. & N. McHale, 1985 . Genetic analysis of isozymevariants in diploid and tetraploid potatoes . Genetics 111 :131-145 .

Ramanna, M.S ., 1979. A re-examination of the mechanisms of2n gamete formation in potato and its implications for breed-ing . Euphytica 28 : 537-561 .

Ross, H. & F .A. Langton, 1974 . Origin of unreduced embryosacs in diploid potatoes . Nature 247 : 378-379 .

Stelly, D .M ., S .J . Peloquin, R .G. Palmer & C.F . Crane, 1984 .Mayer's Hemalum-Methylsalicylate : A stain-clearing tech-nique for observations within while ovules . Stain Technology59(3) : 155-161 .

Stelly, D.M. & S .J . Peloquin, 1986 . Diploid female gameto-phyte formation in 24-chromosome potatoes : Genetic evi-dence for the prevalence of the second meiotic division resti-tution mode . Can J Genet Cytol 28 : 101-108 .

Tai, G.C.C . & H . DeJong, 1980. Multivariate analysis of potatohybrids . I . Discrimination between tetraploid-diploid hybridfamilies and their relationship to cultivars . Can J Genet Cytol22:227-235 .

Taylor, L.M ., 1978. Variation patterns of parthenogeneticplants derived from unreduced embryo sacs of Solanum tuber-osum subspecies andigena (Juz . et Buk .) . Theor Appl Genet52 : 241-249 .

Vallejos, C .E ., 1983. Enzyme activity staining . In : Isozymes inplant genetics and breeding (eds . Tanksley, S .D. and T.J .Orton) Vol A, Elsevier, Amsterdam, pp . 469-516 .

Van Breukelen, E.W.M ., M.S. Ramanna & J .G.Th. Hermsen,1975. Monoploids (n = x = 12) from autotetraploid Solanumtuberosum (2n = 4x = 48) through two successive cycles offemale parthenogenesis . Euphytica 24 : 567-574.

Veilleux, R.E ., N.A. McHale & F .I . Lauer, 1982 .2n gametes indiploid Solanum: frequency and types of spindle abnormal-ities . Can J Genet Cytol 24 : 301-314 .