Islam MR and Shepard KW, 1991. Present status ofgenetics of rust resistance in flax. Euphytica 5:255-268.

Jordan MC and McHughen A, 1988. Giyphosphate-tol-erant flax plants from Agrobacterium-mediated genetransfer. Plant Cell Rep 7:281-284.

O'Malley D, Wheeler NC, and Curies RP, 1980. A man-ual for starch gel electrophoresis. University Wiscon-sin-Madison Staff Paper 11. Madison, Wisconsin: De-partment of Natural Resources.

Prior, T, 1988. The origin and structure of fungal dis-ease resistance genes in plants. Trends Genet 3:217-221.

Schneeberger RG and Cullis CA, 1991. Specific DNAalterations associated with the environmental induc-tion of heritable changes in flax. Genetics 128(3):619-630.

Shaw CR and Prasad, 1969. Starch gel electrophoresisof enzymes: a compilation of recipes. Biochem Genet4:297-320.

Suiter KA, Wendel JF, and Case JS, 1983. Llnkage-1: aPASCAL computer program for the detection and anal-ysis of genetic linkage. J Hered 74:203-204.

Young ND and Tanksley SD, 1989. RFLP analysis ofthe size of chromosomal segments retained aroundthe Tm-2 locus of tomato during backcross breeding.Theor Appl Genet 77:353-359.

Zhan X-C, Jones DA, and Kerr A, 1988. Regenerationof flax plants transformed by Agrobacterium rhizo-genes. Plant Bio 11:551-559.

Behavior of ParentalGenomes in the HybridHordeum vulgare xH. bulbosum

K. Anamthawat-Jonsson,T. Schwarzacher, andJ. S. Heslop-Harrison

In situ hybridization of labeled total genomicDNA with unlabeled blocking DNA enabledthe parental origin of all chromosomes to beestablished in root tips of the mature sexualhybrid plant Hordeum vulgare x H. bulbo-sum. The parental genomes tended to remainspatially separated throughout the cell cycle,with the chromosomes of H. vulgare originlying in a more central domain than those ofH. bulbosum origin. During anaphase andtelophase, chromosomes of H. bulbosum or-igin tended to lag. Although the chromatidsusually separated, they did not have the V

shape characteristic of anaphase chroma-tids. Aneuploid nuclei, missing H. bulbosumorigin chromosomes, arose when the laggingchromatids were not incorporated into thedaughter nuclei, although most cells re-mained diploid. Some interphase cells con-tained micronuclei, all of which were of H.bulbosum origin. Information about chromo-some disposition and movement is importantto enable the understanding of chromosomestability.

Interspecific hybrids between cereals andtheir wild relatives are important becausethey enable the transfer of genes from onespecies into another, and hence broadenthe genetic base of cereal crops (e.g., Blan-co etal. 1986;Lapitanetal. 1986). In barleybreeding, the cross Hordeum vulgare (cul-tivated barley) x H. bulbosum (a wild spe-cies) is used widely to produce haploidbarley, because the H. bulbosum genomeis eliminated in the first divisions of thezygote when some genotypes are used; thechromosome number in the haploid plantcan then be doubled to produce a true-breeding variety in a single step (Kashaand Kao 1970). In crosses between geno-types such as those studied here, the H.bulbosum genome is usually retained, andthe hybrid provides a good model forstudying genome interactions and chro-mosome behavior.

The spatial positioning of chromosomesin mitotic metaphases has been studied inboth spread preparations and reconstruc-tions of hybrid plants (see Heslop-Harri-son and Bennett 1990; Linde-Laursen andJensen 1991). Metaphase studies per seare important because aberrations leadingto aneuploidy occur during division, andboth the three-dimensional position andidentity of all chromosomes can be estab-lished in reconstructions (Schwarzacheret al. 1992). The study of chromosome dis-position at interphase is required becausechromosomes are active in gene expres-sion and DNA replication (Jackson 1991)at this stage of the cell cycle, and thus anyinfluence of position on chromosome ac-tivity would be important then.

In our present work, we aimed to un-derstand aspects of the behavior of thetwo parental genomes in the hybrid H. vul-gare x H. bulbosum. First, we improved insitu hybridization methods to enable theidentification of the parental origin ofchromosomes in the hybrid between thetwo closely related species in the sametaxonomic section of the genus (von Both-mer et al. 1991); second, we used the meth-od to demonstrate where the chromo-somes lay in the nuclei of the hybridthroughout the cell cycle; and third, westudied the physical behavior of the chro-mosomes, including their loss from themain nucleus during division.

Materials and Methods

We studied (1) H. vulgare L. cv. Tuleen 346(barley; 2n = 2x= 14), (2) H. bulbosum L.clone L6 (In = 2x= 14), and (3) the sexualF, hybrid (2n = 2^:= 14, code number C24480/15) between the two species (kindlygiven by Dr. R. A. Finch). Ramets of thehybrid plant used here were maintainedin growth cabinets and glasshouses forabout 10 years before fixation. We trans-ferred plants from soil to hydroponic cul-ture for 3 to 4 days and used new root tipsfor experiments. Fresh leaves were col-lected from glasshouse-grown plants forDNA extraction.

For in situ hybridization, we pretreatedroot tips in ice water for 24 h, fixed themin ethanol: acetic acid (3:1), partially di-gested them with 2% cellulase (Calbio-chem) and 20% pectinase (liquid from As-pergillus niger, Sigma) for 60-75 min at 37°C,and squashed them in 45% acetic acid ontoglass slides, as described by Schwarz-acher et al. (1989). We refixed the prep-arations onto slides with 3:1 fixative for 10min and dehydrated them through an eth-anol series before treating slides withRNase (100 Mg/ml) at 37°C for 1 h, washingin 2 x SSC (0.3 M sodium chloride and 0.03M sodium citrate), and dehydratingthrough an ethanol series again.

We used genomic in situ hybridization(Le and Armstrong 1991; Schwarzacher et

Figure 1. Chromosome preparations from root tips of the sexual F, hybrid Hordeum vulgare (barley) x H. bulbosum (a wild barley) (2n = 14) after genomic in situhybridization using labeled H. bulbosum DNA as a probe. Sites of probe hybridization were detected by yellow-green fluorescence, while the propidium iodide counterstainfluoresces orange. (A) Without blocking DNA, hybridization to chromosomes of both parental sets is detected. Less hybridization is found in the paracentromeric (closedarrow) and the nucleolar organizing (open arrow) regions of the chromosomes originating from the H. vulgare parent. (B-H) In situ hybridization with the addition ofunlabeled blocking DNA from H. vulgare. Chromosomes of H. vulgare origin fluoresce orange with the counterstain since probe hybridization is greatly reduced. Atmetaphase (B and C), seven labeled (yellow-green) chromosomes from H. bulbosum and seven orange chromosomes from H. vulgare can be distinguished. Genomicprobing allows parental genomes to be identified at anaphase (D and E), interphase (F), and prophase (G and H). The two parental genomes tend to be spatially separatedat all stages of the cell cycle (B-H), with a tendency for the labeled chromosomes of H. bulbosum origin to lie toward the periphery of the nucleus (C, F-H). A yellowmicronucleus of H bulbosum origin is present in C. At late anaphase (E), all the chromosomes of H. vulgare origin are at the spindle poles, while most of the 12 H. bulbosumorigin chromatids are lagging. The other two chromatids may have been lost during preparation. The prophase (H) has six yellow-green chromosomes from H. bulbcsumand probably entered division missing one chromosome. Scale bar: 10 pm.

78 The Journal of Heredity 1993:84(1)

at University of L

eicester on Novem

ber 30, 2011http://jhered.oxfordjournals.org/

Dow

nloaded from

http://jhered.oxfordjournals.org/

Brief Communications 79

at University of L

eicester on Novem

ber 30, 2011http://jhered.oxfordjournals.org/

Dow

nloaded from

http://jhered.oxfordjournals.org/

al. 1989), with blocking DNA (Anamtha-wat-J6nsson et al. 1990) and digoxigeninprobe labeling and detection (Leitch U etal. 1991). For use as a probe, we shearedtotal genomic DNA from H. bulbosum togive 3- to 10-kb fragments, and labeledthem with digoxigenin- 11-dUTP (Boeh-ringer) using a standard nick-translationprotocol. Blocking DNA was autoclavedgenomic DNA from H. uulgare (15 psi, 5min, to give DNA fragments between 200and 500 bp long). Our probe hybridizationmixture (50 fi\ per slide) included 200- to300-ng labeled probe, 20-40 times thatamount of blocking DNA, 60% formamide,2 x SSC, 10% dextran sulfate, and 0.2% SDS(sodium dodecyl sulfate). We denaturedthe hybridization mixture at 70°C for 10min, placed it on the slide with a plasticcoverslip over it, and denatured the mix-ture and preparation in a humid chamberat 90°C for 10 min. After hybridizationovernight at 37CC, we washed the slides in50% formamide in 2x SSC at 40°C for 10min and two or three times more in 2xSSC. These conditions allowed sequenceswith more than 80% homology to form sta-ble hybrids. We detected sites of probehybridization by incubating slides in sheepanti-digoxigenin conjugated to fluorescein(FITC; Boehringer, 20 tig ml1; 37°C, 1 h),washed in 4x SSC with 0.2% Tween 20,and amplified signal in rabbit anti-sheepsecondary antibody conjugated to FITC(Dakopatts, 10 ng ml"1)- We counters-tained DNA with propidium iodide (5 tigml"1) and examined the preparations byepifluorescence microscopy. Photographswere taken on Fujicolor 400 print film.

Results

Root tip chromosome spreads from thehybrid H. vulgare x H. bulbosum followingin situ hybridization of labeled genomicDNA from H. bulbosum are shown in Figure1. A hybridization signal of similar strengthwas detected as yellow-green fluorescenceon all 14 chromosomes when no blockingDNA was used (Figure 1A). Some chro-mosomes showed unlabeled bands thatwere stained only with the orange-fluo-rescing propidium iodide counterstain. Theaddition of unlabeled blocking DNA fromH. vulgare revealed seven chromosomesthat showed a strong yellow-green in situhybridization signal and seven with theorange counterstain predominant (Figure1B,C). The chromosomes that originatedfrom the H. vulgare cv. Tuleen 346 parenthave morphologies unlike those of H. bul-bosum (see Schwarzacher et al. 1992).

Orange chromosomes that showed littlelabeling with H. bulbosum genomic DNAhad the H. vulgare morphology, whereaschromosomes that were labeled stronglyhad morphologies that corresponded tochromosomes originating from the H. bul-bosum genome.

In most metaphases, the two parentalgenomes were not intermixed but occu-pied spatially separated domains. For ex-ample, in Figure IB, the genomes lie nextto each other (side-by-side), while in Fig-ure 1C the labeled chromosomes from H.bulbosum lie around the seven H. vulgareorigin chromosomes. At other stages ofthe cell cycle (Figure 1D-H), genomicprobing with blocking enabled labeled, H.bulbosum origin, and unlabeled, H. uulgareorigin, chromatin or chromosomes to bedistinguished clearly by yellow-green ororange fluorescence, respectively. At ana-phase, the labeled H. bulbosum-originchromosomes separated into chromatids,but many chromatids did not have the Vshape characteristic of anaphase (Figure1D,E). In a late anaphase (Figure IE), theH. vulgare-origin chromosomes were at thespindle poles, while many H. bulbosum-origin chromosomes were lagging. At in-terphase, domains of labeled and unla-beled chromatin could be distinguished(Figure IF). At prophase (Figure 1G,H),chromosomes were recognizable, with theyellow-green chromosomes of H. bulbo-sum origin tending to be spatially sepa-rated from the orange chromosomes of H.vulgare origin.

Some chromosome preparations hadfewer than 14 chromosomes. Of these, 30metaphases and prophases from five slideshad a mean of 5.5 chromosomes of H. bul-bosum origin and 6.7 chromosomes of H.vulgare origin. At prophase, where the nu-clear envelope was still largely intact,chromosomes are unlikely to be lost dur-ing spreading; hence the presence of fewerthan 14 chromosomes indicated that thenucleus was aneuploid as it entered divi-sion. In Figure 1G, one H. bulbosum-originchromosome has been lost. In some meta-phases, Figure 1C, a micronucleus of H.bulbosum origin was present. Thus cellsthat were aneuploid or contained micro-nuclei could enter division.

Discussion

Genomic ProbingIn situ hybridization of labeled total ge-nomic DNA from H. bulbosum without ad-dition of competitive blocking DNA to

metaphases of the hybrid H. bulbosum xH. vulgare, labeled all chromosomes (Fig-ure 1A). Most chromosomes of H. vulgareorigin (identified by their morphology)showed unlabeled bands, which presum-ably consisted of tandemly repeated se-quences that were not represented in thelabeled DNA probe from H. bulbosum.Paracentromeric and some intercalary Cbands in H. vulgare cv. Tuleen 346 (Finchand Bennett 1982) correspond to the un-labeled bands. Many species have tandemrepeats at the centromere, and these re-peats tend to be divergent in sequence be-tween closely related species (e.g., Malu-szynska and Heslop-Harrison 1991)because of their rapid evolution. In Hor-deum, too, the centromeric and noncen-tromeric tandem repeats have apparentlyevolved and diverged between the two re-lated species, giving the unlabeled bands,whereas interspersed sequences show highlevels of homology between the genomes.

At stages of the cell cycle other thanmetaphase, the parental origin of the chro-matin could not be identified without ad-dition of blocking DNA in chromosomespreads. When blocking DNA from H. vul-gare was added to the hybridization mix,the parental origin of all chromosomes inthe hybrid H. vulgare x H. bulbosum couldbe easily distinguished throughout the cellcycle (Figure 1B-H). The DNA sequencesthat are common to both species, andhence are blocked, presumably consistlargely of interspersed and tandemly re-peated sequences. About 5% of the Hor-deum genome consists of one interspersedsequence family, BIS 1, which is disperseduniformly along most of the chromo-somes, excluding the paracentromeric,nucleolar organizing, and telomeric regions(Moore et al. 1991). Members of the BIS 1sequence family presumably are repre-sented by the hybridization signal detect-ed in the unblocked chromosomes (Figure1A), but are probably blocked in the otherspreads (Figure 1B-H). The H. bulbosumgenome must include a substantial pro-portion of essentially species-specific se-quences since the chromosomes are uni-formly labeled from telomere to telomerein the blocked spreads except for a slightlyless labeled region at the centromere. Theregions with strong cross-hybridization inFigure 1A are likely to include groups ofgenes that are conserved, but the abilityof the blocking DNA to allow discrimina-tion of the two species shows that the con-served sequences are interspersed withsequences essentially specific to one ge-nome.

80 The Journal of Heredity 1993:84(1)

at University of L

eicester on Novem

ber 30, 2011http://jhered.oxfordjournals.org/

Dow

nloaded from

http://jhered.oxfordjournals.org/

Nuclear ArchitectureThe micrographs of chromosomes follow-ing in situ hybridization show that the nu-cleus is not randomly organized, but thatcomplete parental sets of chromosomes,detected by presence or absence of an insitu hybridization signal, occupy spatiallyseparate domains not only at metaphase(Figure 1 B,C) but throughout the cell cycle(Figure 1D-H). We had expected the resultat metaphase based on knowledge fromreconstructions (Schwarzacheretal. 1992):the metaphase in Figure IB shows side-by-side parental genome separation, whileFigure 1C shows the chromosomes of H.bulbosum origin surrounding the chro-mosomes of H. vulgare origin. Analysis ofthe intergeneric hybrid H. vulgare x Secaleafricanum has established that the im-pression of interphase genome separation(Figure 1) is not an artifact but is presentin reconstructions of serially sectionednuclei (Leitch AR et al. 1991). Thereforewe can conclude that, in the intragenerichybrid H. vulgare x H. bulbosum, parentalgenome separation is present at all stagesof the cell cycle, and the results presentedhere are likely to represent the in vivosituation.

To examine nuclear architecture, knowl-edge of both the identity and position ofall chromosomes is required. In hybridsbetween species that are closely related,in situ hybridization is the method ofchoice to identify each chromosome, andit is likely that multiple labeling systems(Leitch IJ et al. 1991) will be developed toallow identification of genomes in hybridsas well as of individual chromosomes. Thiswill enable detailed studies of the higherorder organization of individual chromo-somes within the nucleus (Heslop-Harri-son 1991).

Stability of the Hybrid andChromosome EliminationIn many hybrids between cereals, includ-ing Hordeum species, a complete parentalgenome may be eliminated during earlyembryo development, so that embryo cul-ture regularly recovers plants with a com-plete haploid genome from one parent andno chromosomes from the other parent(Kasha and Kao 1970). The degree of suchinstability varies within a given pair of spe-cies and depends on the genotypes in-volved (Simpson et al. 1980). For example,crosses between diploid H. vulgare anddiploid H. bulbosum clone LI gave almost100% haploid and no diploid progeny, butH. vulgare x H. bulbosum clone L6 (used

in our study) gave about 70% diploid hy-brids in its progeny.

Chromosome counts in divisions indi-cated that many cells had remained dip-loid, with seven chromosomes from eachparent (Figure 1A-C), over the 10 yearssince the hybrid was produced. In divi-sions, lagging chromosomes were oftenobserved (Figure IE). Such chromosomesmight not be incorporated into daughternuclei at telophase, and thus they give riseto micronuclei. The micronuclei mayeventually degrade, but in some cases theyremain at least until the next metaphase(Figure 1C); however, most aneuploid cellsmay not divide or cycle more slowly thaneuploid ones. Despite the numbers of ab-errant nuclei, it is unlikely that the plantis becoming increasingly aneuploid sinceit has not become haploid over 10 years.

The H. bulbosum chromosomes may beunstable because they fail to initiatecongression at metaphase or to migrate tothe poles at anaphase. In mammalian cells,a similar differential behavior of chro-mosomes has been reported. In hamster-human cell fusion hybrids, Zelesco andMarshall Graves (1988) found that the hu-man chromosomes were preferentially lostduring division (segregant) and tended tobe central within a ring of hamster chro-mosomes. The authors suggested that thehuman centromeres attached aberrantly,or simply less efficiently, to the spindle inhybrid cells. If so, then differential cen-tromere function alone would be unlikelyto cause the parental genome separationseen in both the plant and mammalian hy-brids, since the less stable chromosomeset is peripheral in one case and centralin the other. In the human disease Rob-erts syndrome, Jabs et al. (1991) reportedthat aneuploidy and micronuclei (in 5%-11% of cells) arise as a direct result oflagging chromosomes, although all chro-mosomes congress onto the plate. Thuscongression and anaphase movementseem to be independent events, althoughdefects in microtuble attachment might beexpected to affect both processes.

Maintenance of Genome Separationand Centromere ActivityWide hybrids enable analysis of genomeinteractions and the genetical control ofchromosome behavior. In the hybrid H.vulgare x H. bulbosum, the differences be-tween the two genomes may enable themaintenance of genome separationthroughout the cell cycle, along with dif-ferential expression and functioning of thetwo sets of centromeres. The activity of

the centromeric structures, and the rate ortiming of their becoming active, must beunder genetic control since the two pa-rental genomes show disparate behaviorin the hybrid. However, the lagging of oneparental set of chromosomes is not theonly mechanism for maintaining genomeseparation, since the hybrid between H.vulgare and 5. africanum shows genomeseparation but the chromosomes segre-gate together (Leitch AR et al. 1991). Com-parative, timed studies of chromosome be-havior in different hybrids and quantitativemeasures of centromere activity will ad-vance understanding of the control ofchromosome elimination and may lead tomore general conclusions about centro-mere activity and chromosome move-ment.From the Karyobiology Group, Department of Cell Bi-ology, John Innes Centre for Plant Science Research,Norwich NR4 7UJ, U.K. Dr. Anamthawat-Jonsson isnow at the Agricultural Research Institute, Keldnaholt,IS-112 Reykjavik, Iceland. The work was enabled byBP and Venture Research International. Address re-print requests to Dr. Heslop-Harrison at the addressabove.

The Journal of Heredity 1993:84(1)

ReferencesAnamthawat-Jonsson K, Schwarzacher T, Leitch AR,Bennett MD, and Heslop-Harrison JS, 1990. Discrimi-nation between closely related Triticeae species usinggenomic DNA as a probe. Theor Appl Genet 79:721-728.

Blanco A, Fracchiolla GV, and Greco B, 1986. Inter-generic wheat x barley hybrid. J Hered 77:98-100.

Finch RA and Bennett MD, 1982. The karyotype ofTuleen 346 barley. Theor Appl Genet 62:53-58.

Heslop-Harrison JS, 1991. The molecular cytogeneticsof plants. J Cell Sci 100:15-21.

Heslop-Harrison JS and Bennett MD, 1990. Nucleararchitecture in plants. Trends Genet 6:401-405.

Jabs EW, Tuck-Muller CM, Cusano R, and Rattner JB,1991. Studies of mitotic and centromeric abnormalitiesin Roberts syndrome: implications for a defect in themitotic mechanism. Chromosoma 100:251-261.

Jackson DA, 1991. Structure-function relationships ineukaryotic nuclei. BioEssays 13:1-10.

Kasha KJ and Kao KN, 1970. High frequency haploidproduction in barley QHordeum vulgare L). Nature 225:874-876.

Lapitan NLV, Sears RG, Rayburn AL, and Gill BS, 1986.Wheat-rye translocations: detection of chromosomebreakpoints by in situ hybridization with a biotin-la-beled DNA probe. J Hered 77:415-419.

Le HT and Armstrong KC, 1991. In situ hybridizationas a rapid means to assess meiotic pairing and detec-tion of alien DNA transfers in interphase cells of widecrosses involving wheat and rye. Mol Gen Genet 225:33-37.

Leitch AR, Schwarzacher T, Mosgoller W, Bennett MD,and Heslop-Harrison JS, 1991. Parental genomes areseparated throughout the cell cycle in a plant hybrid.Chromosoma 101:206-213.

Leitch IJ, Leitch AR, and Heslop-Harrison JS, 1991.Physical mapping of plant DNA sequences by simul-taneous in situ hybridization of two differently labelledfluorescent probes. Genome 34:329-333.

Brief Communications 81

at University of L

eicester on Novem

ber 30, 2011http://jhered.oxfordjournals.org/

Dow

nloaded from

http://jhered.oxfordjournals.org/

Linde-Laursen I and Jensen J, 1991. Genome and chro-mosome disposition at somatic metaphase in a Hor-deum x Psathyrostachys hybrid. Heredity 66:203-210.

Maluszynska J and Heslop-Harrison JS, 1991. Local-ization of tandemly repeated DNA sequences in Ara-bidopsis thaliana. Plant J 1:159-166.

Moore G, Cheung W, Schwarzacher T, and Flavell R,1991. BIS 1, a major component of the cereal genomeand a tool for studying genomic organization. Ge-nomics 10:469-476.

Schwarzacher T, Heslop-Harrison JS, Anamthawat-Jonsson K, Finch RA, and Bennett MD, 1992. Parentalgenome separation in reconstructions of somatic andpremeiotic metaphases of Hordeum vulgare x H. bulbo-sum. J Cell Sci 101:13-24.

Schwarzacher T, Leitch AR, Bennett MD, and Heslop-Harrison JS, 1989. In situ localization of parental ge-nomes in a wide hybrid. Ann Bot 64:315-324.

Simpson E, Snape JW, and Finch RA, 1980. Variationbetween Hordeum bulbosum genotypes in their abilityto produce haploids of barley, Hordeum vulgare. ZPflanzenzuchtg 85:205-211.

von Bothmer R, Jacobsen N, Baden C, Jorgensen RB,and Linde-Laursen I, 1991. An ecogeographical studyof the genus Hordeum. Systematic and EcogeographicStudies on Crop Genepools 7. Rome: International Boardfor Plant Genetic Resources.

Zelesco PA and Marshall Graves JA, 1988. Chromo-some segregation from cell hybrids: IV. Movement andposition of segregant set chromosomes in early-phaseinterspecific cell hybrids. J Cell Sci 89:49-56.

Table 1. Progeny distribution in the fruit skin color trait

Progeny phenotype

Genetics of Skin Color,Flowering Group, and AniseScent in AvocadoU. Lavi, E. Lahav, C. Degani, andS. Gazit

The genetic information available on subtrop-ical fruit trees in general, and on avocado inparticular, is quite limited. The genetics of skincolor, flowering group, and anise scent in av-ocado (Persea americana Mill) have beenstudied. Progeny distribution of seedlingsoriginating from crosses between all possiblephenotypes in the above-mentioned threetraits have been presented. The results ruleout a model of one or two loci for any of thesetraits. It is quite probable that these traits arecoded by several loci having several allelesin each locus. The various phenotypes prob-ably result from various heterozygous com-binations in several loci. This knowledge isrelevant for both the genetics and breedingof avocado.

Avocado (Persea americana Mill) is a dip-loid having 2n = 24 chromosomes. Fruitskin color in most commercial cultivars isgreen or purple including light and darkhues of both colors. Flower groups areclassified according to Stout (1923) into Aand B. Group A cultivars exhibit the first

Parent phenotypeNo. of

Green Purple families Green/purple

Green x green (sellings)Green x green (crosses between cultivars)Green x green (total)

Green x purplePurple x greenPurple x purple (selfings)

TotalReciprocal crosses

Green x green (Ettinger x Tova)Green x green (Tova x Ettinger)Green x purple (Ettinger x Rosh-Hanikra II)Purple x green (Rosh-Hanikra II x Ettinger)

121273

1420

919

8.613.6

394

10715

34

4484

28

131

11.6

2.51.51.2

480 90 33 5.3

371021013

134

12

1111

37.034.0

2.51.1

N.S.

N.S.

opening of functional female flowers dur-ing the morning hours. These flowers closenear midday and reopen in the functionalmale stage the following afternoon. In Bcultivars, the female opening occurs in theafternoon, and the male (second) openingoccurs the following morning. Anise scentin the leaf is present in the Mexican raceand absent from Guatemalan and West In-dian races.

Bergh (1975) reported several studiesaimed at understanding the genetics ofseveral traits. Skin color was concluded tobe inherited as a typical polygenic char-acter, and flowering group was affected bysegregation at a number of loci. Data con-cerning the genetics of leaf anise scent,skin color, or flowering group traits werenot reported. This article reports the ge-netics of these traits.

Materials and Methods

The parent and progeny plots were locat-ed at the Akko Experiment Station in Is-rael. We collected seed from crosses andselfings by caging trees under a net, usingbees as the pollen vector (Lavi et al. 1991).The harvested seed was sown in a nursery,and one year later we transplanted theseedlings into breeding plots. The prog-eny of each cage were planted randomlyin one block. Progenies of different cageswere randomized in the orchard. The ju-venile period was shortened by the use ofautumn girdling (Lahav et al. 1986).

We recorded fruit skin color, floweringgroup, and leaf anise scent traits over a2-year period. Skin color was classified asgreen or purple, flowering group as A orB, and leaf anise scent as present or ab-sent.

The cultivars used were Anaheim,Fuerte, Irving, Nabal, Regina, Rincon, Wurtz

(Rounds 1950), Ettinger (Storey and Bergh1963), Hass (Griswold 1945), Pinkerton,Reed (Platt 1976), Rosh-Hanikra II (Laviet al. 1991), Horshim, and Tova (Slor andSpodheim 1971-1972).

To distinguish between hybrids and self-pollinated seedlings, we characterized theprogeny by isozyme analyses of leaf tissue(Degani and Gazit 1984). However we can-not rule out the possibility that a few in-dividuals were wrongly classified.

The number of observations in the var-ious crosses varied from four to 387, for atotal of 1,688 seedlings.

Significance was determined by the chi-square tests.

Results

Fruit Skin ColorThere were nine selfing crosses of thegreen x green type. The ratio of green/purple among the progeny varied from 3-20 to one, (3-20:1) and on the average was8.6. There were also 19 crosses betweenvarious cultivars with green skin color. Theratio of green/purple among these prog-eny varied from 2-37 to one (2-37:1) andon the average was 13.6. No significant dif-ferences were found between the two ra-tios (8.6 and 13.6) (Table 1).

There was only one cross of green xpurple, resulting in 10 green and four pur-ple progeny. There were three crosses ofthe purple x green type, resulting in aprogeny distribution ranging from 1.1:1 to2.2:1 green to purple with an average of1.5:1. One selfing represented the purplex purple cross, resulting in a progeny dis-tribution of 1.2:1. The ratio of green/purplein selfings green x green (8.6) was sig-nificantly different (P= .013) from the sameratio for purple x purple selfings (1.2).

82 The Journal of Heredity 1993:84(1)

at University of L

eicester on Novem

ber 30, 2011http://jhered.oxfordjournals.org/

Dow

nloaded from

http://jhered.oxfordjournals.org/