brief communications - oxford academic

71

Brief Communications

Low Microsatellite Variationin Laboratory Gerbils

K. Neumann, S. Maak, I. W.Stuermer, G. von Lengerken, andR. Gattermann

The Mongolian gerbil has become a mod-el organism of increasing importance forthe understanding of aging, epilepsy, theprocess of domestication or sociobiologi-cal questions. We report the developmentand characterization of the first nine poly-morphic dinucleotide repeat loci in thisspecies. Average observed heterozygos-ity and allele number of laboratory animalsmeasured 0.136 (SE 5 60.065) and 1.78(SE 5 60.278) compared to 0.761 (SE 560.025) and 9.2 (SE 5 60.57) found fora reference group of wild gerbils. The ex-treme low genetic variation observed inlaboratory animals is caused by severalsevere population size bottlenecks due tothe initial founder event and the later es-tablishment of subpopulations. Reducedlevels of allelic polymorphism in experi-mental animals hamper genetic mappingor parental studies. Therefore experimentsrelying on kinship analyses have to be car-ried out on wild animals. Estimates of ge-netic identity and parental exclusion werecalculated as Pid 5 2.8 3 10212 and Pex .0.999 in wild gerbils. Laboratory gerbilstrains show the expected high degree ofgenetic similarity. However, significant al-lele frequency differences (P , .001) be-tween American and European gerbils atsome microsatellite loci may still allow dis-crimination between breeding lines.

The Mongolian gerbil (Meriones unguicu-latus Milne Edwards, 1867) is a commonrodent inhabiting the open steppe andsemidesert regions of Mongolia and China.Its history as a laboratory animal startedin 1935 when 20 pairs were captured inMandschuria and brought to Japan for

breeding. From there, 11 pairs of gerbilswere imported into the United States in1954, of which five females and four malesat Tumblebrook Farm Ltd. formed thebreeding stock for laboratories in the Unit-ed States and Europe (Marston 1972;Schwentker 1963). Since then the Mongo-lian gerbil has become an extensively usedexperimental animal, for example, in neu-roscience, auditory research, and physi-ology. The species serves as an importantgenetic model for aging (Spangler et al.1997) and limbic epilipsy (Scotti et al.1998). However, despite being an estab-lished model organism, the genetics of theMongolian gerbil has hardly been ex-plored (Gray and Wong 1990).

A number of behavioral and reproduc-tion studies focus on the social organiza-tion and the cooperative breeding systemof M. unguiculatus (e.g., Elwood 1979; Wein-andy and Gattermann 1999), but most datawere obtained from captive populations oflaboratory animals and only a few fieldstudies have been carried out (e.g., Agrenet al. 1989). Studies under laboratory andseminatural conditions (Agren 1976, 1984)have shown that gerbils live in familygroups and defend their territories. Withinfamilies there is usually a single reproduc-ing pair suppressing reproduction of sub-ordinated group members. Despite the for-mation of stable pair bonds, females oftenseek extrapair mating opportunities out-side their territories to prevent inbreeding(Agren 1990; Agren et al. 1989).

The ongoing reproductive isolation ofexperimental gerbil strains has led to anumber of anatomical, physiological, andbehavioral pecularities not found in wildanimals. Brain size reduction, seizures,and different learning performance areconsidered as evidence for a new case ofdomestication in a rodent species(Stuermer et al. 1997).

The low number of founder individualsin the Tumblebrook population has led toa substantial degree of inbreeding among

captive gerbils. Previous data describingthe genetic variability in laboratory bredanimals are restricted to isozyme studies.Electrophoretic analysis of lactate dehy-drogenase and alkaline phosphatase re-vealed no variation between different col-or morphs (Shimizu et al. 1996). Acomparative survey of four gerbil strainskept in Japan found them indistinguish-able at 23 protein loci except for liver acidphosphatase Acp2 (Okumura et al. 1995).These data support the view of a low in-ter- and intrastrain diversity among exper-imental animals.

Because of the obvious lack of geneticdata and the need to evaluate the prog-nostics of genetic mapping experiments orparental studies, we have developed andcharacterized a set of microsatellite loci.Based on the data derived from nine lociwe have estimated the genetic diversity oflaboratory and wild gerbils.

Materials and Methods

The isolation procedure for microsatelli-tes followed standard protocols with mod-ifications (e.g., Neumann and Wetton1996). DNA was extracted from six malegerbils (three wild and three laboratory)and doubly digested with restriction en-zymes AluI and HaeIII. A genomic librarywas established in XL-1 Blue MRF9 (Stra-tagene) transformed with size fractionated(0.3–0.8 kb) DNA fragments ligated intothe SmaI site of pUC18 (Pharmacia). Ap-proximately 1500 recombinant cloneswere transferred to microtiter plates, rep-lica plated onto nylon membranes, andscreened with a cocktail of DIG-labeled oli-gonucleotides (CA)15, (TC)15, (AGG)5, and(GGAA)4. Hybridized colonies were visu-alized using the CSPD system (Roche).Thirty positive clones were sequencedcontaining 18x d(CA), 5x d(CT), 2xd(GGAA), 1x d(GGA), and 4x mixed d(CA/CT) motifs. Primers were designed for 13loci using OLIGO 5.0 (MedProbe, Norway).

72 The Journal of Heredity 2001:92(1)

Table 1. Microsatellite loci in Meriones unguiculatus

LocusGenBankaccession no. Primer sequence (59-39) Repeat type

Annealingtemperature

Mungm1 AF200939 F:TGTGGCTGGCATCCTA d(GT)12 d(GA)21 54–568CR:AAGCAATTCTGTCTCTGTCTG

Mungm2 AF200940 F:AGCCTTTATAGATGAGCAAGT d(TC)23 548CR:GCCTACTAATGGTGAACTGA

Mungm3 AF200941 F:CAGGCACCCCCAGTTT d(AC)17 55–578CR:GTCTACACAGGCTGAGGATGT

Mungm4 AF200942 F:GGCTCCTGATTCTACATTTCT d(TG)22 54–568CR:CAACCATTGGCAACTCTC

Mungm5 AF200943 F:GCTGGGCTTTAATGTTTATTT d(GT)19 548CR:GGTGGCTCACACTTTCTGT

Mungm6 AF200944 F:TTTCTGGGGTCTCTTTCTCTC d(TC)19 548CR:CCATTCTGCAAGACTCCTCT

Mungm7 AF200945 F:AGTCCCTATTACATCCACAAG d(GA)16 54–568CR:TTATCCTGCAAAGCCTAAG

Mungm8 AF200946 F:TGGGTCCTTTGGAAGA d(GT)21 54–568CR:TGGCTTAAAATGAATCACTTA

Mungm9 AF200947 F:GACAGAGTGGGAGGGGTATGT d(CA)21 54–568CR:TGGCAAGTTTGGTTTGTTTGA

Repeat type refers to the motif obtained from the cloned allele. Only the longest repeat arrays (.10) are presented.

Polymerase chain reaction (PCR) was per-formed using the Ready-To-Go-system(Pharmacia). Thirty picomoles of eachprimer (one was labeled with Cy5 Amiditeas fluorescent dye) and 0.01–0.1 mg of ge-nomic DNA were added to a total volumeof 25 ml. After an initial denaturation stepof 180 s at 948C the amplification proceed-ed for 35 cycles as follows; 60 s at 948C,60 s annealing at primer specific tempera-ture (Table 1) and 120 s at 728C (Thermo-cycler UNO II, Biometra). PCR productswere electrophoresed through 6% denatur-ing polyacrylamide gels on an automatedsequencer A.L.F. express II (Pharmacia).The amplification of nine dinucleotide re-peat loci Mungm1–Mungm9 produced un-ambiguous allele patterns.

Inheritance and linkage of microsatelli-tes were examined in seven gerbil familiesconsisting of laboratory and second gen-eration wild animals (2 laboratory 3 lab-oratory, 3 laboratory 3 wild, and 2 wild 3wild). Linkage analysis was carried out us-ing the sequential LOD score method(Morton 1955) compiled over all informa-tive families.

To estimate the level of genetic polymor-phism, 45 laboratory and 40 wild gerbilswere analyzed. DNA was prepared from eth-anol fixed liver tissues and frozen ear sam-ples using a commercial kit system (E.Z.N.A.Tissue DNA Kit II, peqlab BiotechnologieGmbH). The laboratory animals comprisefive different strains. Sixteen gerbils came di-rectly from Tumblebrook Farm Inc./USA in1995. A further 17 animals were supplied bythe Leibniz Institute of Neurobiology in Mag-deburg/Germany and are crossbred animalsfrom the Charles River corporation (strain:CRW/Mon) and the Technical University ofDarmstadt/Germany. Five animals belonged

to a second strain kept in Magdeburg whichoriginated from Molegart-breed/France. Asingle individual came from a laboratory atthe University of Cologne/Germany and sixanimals represent the laboratory strain Zoh:CRW bred at the Institute of Zoology Halle/Germany. The latter population derivedfrom three pairs supplied by Charles RiverWiga (Sulzbach/Germany) stock label CRW/(Mon)BR in 1992.

Wild gerbils were captured during ajoint expedition of the Leibniz Insitute ofNeurobiology Magdeburg/Germany andthe State University of Mongolia to CentralMongolia in 1995. Animals were trapped atsix different locations about 130–140 kmsouthwest and 100 km west of Ulaanbaa-tar.

All laboratory and wild animals werepooled into two separate groups definedas ‘‘laboratory’’ and ‘‘wild’’ because of thesmall number of individuals representinga particular strain or population. Expectedheterozygosity (HETexp) was calculatedfrom Hardy–Weinberg assumptions foreach locus HETexp 5 1 2 Sqi

2. Deviationsfrom Hardy–Weinberg equilibrium (HWE)were tested by the chi-squared test to de-tect potential genetic heterogeneity withinthe pooled samples. A test of genic differ-entiation combined over all polymorphicloci (Fisher exact test) was performed be-tween American (Tumblebrook) and Eu-ropean laboratory gerbils using the com-puter program GENPOP, version 3.1d(Raymond and Rousset 1995). Based onthe observed allele frequencies (f) we cal-culated the probabilities of profile identity(two random individuals share the samealleles 5 Pid) and parental exclusion (theprobability of detecting an incorrectly as-signed parent 5 Pex) combined over all

loci, according to Gundel and Reetz (1981)and Bruford et al. (1992).

Results

Primer sequences and characteristics ofnine microsatellites in the Mongolian ger-bil are presented in Table 1. SequentialLOD score analyses produced no evidenceof close linkage (z , 22 for all u , 0.05and z , 0 for all u , 0.3) between thesemarkers. The two pure laboratory pedi-grees provided no information because allparents proved homozygous at each lo-cus. Allele segregation among all offspringfollowed Mendelian inheritance. No mea-surable frequencies of null alleles were de-tected.

A genetic variability comparison of lab-oratory and wild gerbils documented adramatic reduction of allele numbers andobserved heterozygosity in the domesticanimals (Table 2). Four loci (Mungm1, 2, 3,9) were monomorphic and allele numbersat the five remaining loci ranged from 2(Mungm4, 5, 8) to 3 (Mungm6, 7) in oursample, giving a mean allele number of1.78 (SE 5 60.278). This differed signifi-cantly (P , .001, Mann–Whitney) from thehigh diversity of these loci in wild gerbilsexhibiting an average allele number of 9.2(SE 5 60.57). The mean observed hetero-zygosity in laboratory animals was 0.136(SE 5 60.065) compared with 0.761 (SE 560.025) in wild gerbils (P , .001, Mann–Whitney). Significant differences in alleledistribution were detected between ani-mals from Tumblebrook Farm (USA) andEuropean laboratories (P , .001). Privatealleles restricted to the American or Eu-ropean strains were found at loci Mungm6and 7 in rather low frequencies. Alleles C(f 5 0.011) and F (f 5 0.011) at locusMungm6 were confined to Europe and al-lele E (f 5 0.022) at locus Mungm7 to Tum-blebrook Farm (USA).

Two loci Mungm8 and 9 showed devia-tion from HWE in wild gerbils (P , .05)due to heterozygosity deficiencies. An ex-cess of observed homozygotes was alsoresponsible for the violation of the HWEin laboratory animals at locus Mungm8 (P, .05). Probability estimates of geneticidentity were calculated as Pid 5 0.056 andPex 5 0.467 in laboratory and Pid 5 2.8 310212 and Pex . 0.999 in wild gerbils.

Discussion

To evaluate the genetic variation in labo-ratory bred and wild Mongolian gerbils wehave isolated a set of nine dinucleotide re-

Brief Communications 73

Table 2. Allele frequencies, heterozygosity values, tests of Hardy–Weinberg equilibrium (HWE) fornine dinucleotide repeat loci in wild and laboratory gerbils

Locus

Alleleno. wild/lab Frequency wild/lab HETobs (HETexp) wild/lab

HWEtest

Mungm1 10/1 A (0.013/–), B (0.100/–), C (0.125/–), D (0.163/–),E (0.013/–), F (0.125/1), G (0.288/–), H (0.138/–),I (0.025/–), J (0.013/–)

0.775/– (0.828/–) ns/–

Mungm2 10/1 A (0.013/–), B (0.100/–), C (0.175/–), D (0.138/–),E (0.150/–), F (0.300/1), G (0.013/–), H (0.050/–),I (0.013/–), J (0.050/–)

0.800/– (0.823/–) ns/–

Mungm3 8/1 A (0.163/–), B (0.050/1), C (0.013/–), D (0.175/–),E (0.325/–), F (0.150/–), G (0.050/–), H (0.075/–)

0.825/– (0.804/–) ns/–

Mungm4 7/2 A (0.113/–), B (0.063/–), C (0.163/0.433),D (0.200/0.567), E (0.113/–), F (0.238/–), G (0.113/–)

0.800/0.422 (0.835/0.491) ns/ns

Mungm5 12/2 A (0.025/–), B (0.050/–), C (0.025/–), D (0.100/–),E (0.075/0.011), F (0.125/0.989),G (0.188/–),H (0.063/–), I (0.163/–), J (0.025/–), K (0.088/–),L (0.075/–)

0.875/0.022 (0.886/0.022) ns/ns

Mungm6 9/3 A (0.038/–), B (0.100/–), C (0.013/0.011),D (0.363/–), E (0.188/0.978), F (0.050/0.011),G (0.138/–), H (0.100/–), I (0.013/–)

0.750/0.044 (0.790/0.043) ns/ns

Mungm7 9/3 A (0.025/–), B (0.063/–), C (0.350/0.278),D (0.050/0.700), E (0.200/0.022), F (0.100/–),G (0.063/–), H (0.100/–), I (0.050/–)

0.700/0.467 (0.804/0.432) ns/ns

Mungm8 11/2 A (0.375/0.311), B (0.013/–), C (0.075/–),D (0.088/–), E (0.050/–), F (0.038/–), G (0.050/–),H (0.125/0.689), I (0.063/–), J (0.113/–), K (0.013/–)

0.650/0.267 (0.807/0.429) s/s

Mungm9 7/1 A (0.013/–), B (0.200/–), C (0.300/–), D (0.125/–),E (0.138/–), F (0.088/1), G (0.138/–)

0.675/– (0.809/–) s/–

ns 5 not significant; s 5 significant for P , .05.

peat loci. Their polymorphic nature andMendelian inheritance was confirmed. Allmarkers may be considered as geneticallyindependent because no evidence for link-age was obtained. Pronounced differencesin microsatellite variation between twopooled groups of laboratory and wild ger-bils correspond with previous biochemi-cal data and reflect the genetic history ofthe captive gerbil population. The low de-gree of polymorphism is a consequence ofthe initial founder event and following bot-tlenecks due to the establishment of sub-populations at different breeding locations.The genetic diversity of the investigated lab-oratory gerbil strains is below those ob-served in inbred mouse or rat strains (Loveet al. 1990; Serikawa et al. 1992). In contrast,the data of wild gerbils fall in the range ofvariation found in other outbred wild Crice-tidae populations (e.g., Cricetus cricetus; Neu-mann K, unpublished data).

All alleles present in the studied labo-ratory strains could be traced back to al-leles in wild gerbils and have probablyoriginated from early founder animals. In-dications for this are also the high fre-quencies of most of these alleles in thewild gerbil sample. However, the increasein single low-frequency alleles because ofmutations cannot be excluded. Smalllength differences by one repeat, for ex-ample, at locus Mungm7, are in agreementwith common stepwise mutations in mi-crosatellites.

Observed deviations from HWE at twoloci Mungm8, 9 (P , .05) in wild and onelocus Mungm8 (P , .05) in laboratory ger-bils are most likely explained by geneticdifferences between sampled populationsor strains, since a general trend for in-creased proportions of homozygotes isobserved at almost all loci.

Laboratory gerbil strains show the ex-pected high degree of genetic similarity.Four microsatellite loci were homozygousfor the same alleles in all tested domesticanimals. This is in concordance with thebreeding and distribution history of ger-bils in North America and Europe andproves their common origin. Significant al-lele frequency differences (P , .001) at theremaining five loci between animals fromTumblebrook Farm (USA) and Europeangerbils may still allow the discriminationbetween breeding stocks. The detection ofprivate alleles in the two laboratory gerbilpools could be a result of the small samplesizes but may equally account for true ge-netic strain diversity. Unfortunately ourmaterial did not contain laboratory gerbilsfrom Japan. Their genetic compositioncould be more heterogeneous due to thelarger number of unrelated founders (butsee Okumura et al. 1995).

Mapping and linkage studies of geneti-cally important loci in Mongolian gerbilsare possible but require a larger set of in-formative markers than in most other ex-perimental animals. However, the low lev-

el of genetic variation among strains andthe existence of obvious phenotypic dif-ferences, for example, seizure-sensitiveand seizure-resistant strains may ease theidentification of affected genetic loci.

High combined probability estimates ofgenetic identity and low parental exclu-sion strongly interfere with kinship analy-ses in laboratory gerbils. Therefore genet-ic studies concerning mating system andkinship have to be carried out on wild an-imals. The microsatellite markers reportedhere provide a powerful system to accom-plish this aim.

Laboratory Mongolian gerbils are an ex-ample for a thriving captive rodent popu-lation despite its apparent lack of substan-tial genetic variation. The existence ofsome distinguishing features such as anincreased susceptibility to cerebral infarc-tions in domestic versus wild animals maybe caused by genetic alterations, for ex-ample, the fixation of rare alleles andcould therefore present inbreeding effects.

From the Institute of Zoology, Martin-Luther-UniversityHalle-Wittenberg, Domplatz 4, D-06108 Halle (Saale),Germany (Neumann and Gattermann), Institute of An-imal Breeding and Husbandry with Veterinary Clinic,Martin-Luther-University Halle-Wittenberg, Halle (Saa-le), Germany (Maak and von Lengerken), and FederalInstitute of Neurobiology, Magdeburg, Germany(Stuermer). We thank M. Strobel for technical supportand Dr. K. Seidelmann for useful discussions and helpwith the statistics. K. Williams is acknowledged forchecking the English. The authors like to thank threeanonymous reviewers for their useful comments on themanuscript. Address correspondence to Karsten Neu-mann at the address above or e-mail: [email protected].

q 2001 The American Genetic Association

References

Agren G, 1976. Social and territorial behaviour in theMongolian gerbil (Meriones unguiculatus) under semi-natural conditions. Biol Behav 1:267–285.

Agren G, 1984. Pair formation in the Mongolian gerbil.Anim Behav 32:528–535.

Agren G, 1990. Sperm competition, pregnancy initiationand litter size: influence of the amount of copulatorybehaviour in Mongolian gerbils, Meriones unguiculatus.Anim Behav 40:417–427.

Agren G, Zhou Q, and Zhong W, 1989. Ecology and so-cial behavior of Mongolian gerbils, Meriones unguicu-latus, at Xilinhot, Inner Mongolia, China. Anim Behav37:11–27.

Bruford MW, Hanotte O, Brookfield JFY, and Burke T,1992. Single locus and multilocus DNA fingerprinting.In: Molecular genetic analysis of populations. A prac-tical approach (Hoelzel AR, ed). Oxford: IRL Press;225–269.

Elwood RW, 1979. Maternal and paternal behaviour ofthe Mongolian gerbil: a correlation study. Behav NeuralBiol 25:555–562.

Gray A and Wong R, 1990. Influence of coat color geneson seizure behavior in Mongolian gerbils. Behav Genet20:481–485

Gundel H and Reetz I, 1981. Exclusion probabilities ob-tainable by biochemical polymorphisms in dogs. AnimBlood Groups Biochem Genet 12:123–132.


Table 1. Occurrence of Mongolian gerbil mutants

Phenotype Occurred Allele Reference

Acromelanistic albino 1970 ch Robinson (1973)

Nonagouti 1971 aCramlet et al. (1974)Waring and Poole (1980)

White spot 1968 SpSilverstein and Silverstein (1976)Waring et al. (1978)

Gray 1976 g Leiper and Robinson (1985)Pink-eyed dilution 1977 p Henley and Robinson (1981)Hairless 1978 — Swanson (1980)Albino 1985 c Matsuzaki et al. (1989)Chinchilla medium 1994 cchm This article

Love JM, Knight AM, Mcaleer MA, and Todd JA, 1990.Towards construction of a high resolution map of themouse genome using PCR-analyzed microsatellites. Nu-cleic Acids Res 18:4123–4130.

Marston JH, 1972. The Mongolian gerbil. In: The UFAWhandbook on the care and management of laboratoryanimals. Essex: Longman; 257–268.

Morton NE, 1955. Sequential tests for the detection oflinkage. Am J Hum Genet 7:277–318.

Neumann K and Wetton JH, 1996. Highly polymorphicmicrosatellites in the house sparrow Passer domesticus.Mol Ecol 5:307–309.

Okumura K, Maeda Y, Koresawa Y, and Hashiguchi T,1995. Study of biochemical gene markers in the Mon-golian gerbil (Meriones unguiculatus). Exp Anim (To-kyo) 43:719–723.

Raymond M and Rousset F, 1995. Genepop (version1.2): population genetics software for exact tests andeumenicism. J Hered 83:239.

Schwentker V, 1963. The gerbil, a new laboratory ani-mal. Illinois Vet 6:5–9.

Scotti AL, Bollag O, and Nitsch C, 1998. Seizure patternsin Mongolian gerbils subjected to a prolonged weeklytest schedule: evidence for a kindling-like phenomenonin the adult population. Epilepsia 39:567–576.

Serikawa T, Kuramoto T, Hilbert P, Mori M, Yamada J,Dubay CJ, Lindpainter K, Ganten D, Guenet J-L, LathropGM, and Beckman JS, 1992. Rat gene mapping usingPCR-analyzed microsatellites. Genetics 131:701–721.

Shimizu M, Iida K, Yoshida H, and Shichinohe K, 1996.Electrophoretic study of lactate dehydrogenase and al-kaline phosphatase isoenzymes of the Mongolian gerbil(Meriones unguiculatus). J Vet Med Sci 58:401–406.

Spangler EL, Hengemihle J, Blank G, Speer DL, Brzo-zowski S, Patel N, and Ingram DK, 1997. An assessmentof behavioral aging in the Mongolian gerbil. Exp Ger-ontol 32: 701–717.

Stuermer IW, Plotz K, Wetzel W, Wagner T, Leybold A,and Scheich H, 1997. Reduced brain size and faster au-ditory discrimination learning in laboratory gerbilscompared to wild Mongolian gerbils (Meriones ungui-culatus). Soc Neurosci 23:2067.

Weinandy R and Gattermann R, 1999. Parental care andtime sharing in the Mongolian gerbil. Z Saugetierkd 64:169–175.

Received January 10, 2000Accepted October 31, 2000

Corresponding Editor: Robert Wayne

A Second AcromelanisticAllelomorph at the AlbinoLocus of the MongolianGerbil (Merionesunguiculatus)

F. Petrij, K. van Veen, M. Mettler,and V. Bruckmann

A new autosomal recessive coat color mu-tant in the Mongolian gerbil (Meriones un-guiculatus) is described: chinchilla medi-um (symbol cchm). The mutant has typicalacromelanistic features similar to those ofseveral acromelanistic c locus mutants ofother species of mammals. Previously amore severe form of acromelanism (chch)

has been described in the Mongolian ger-bil. The new allele shows to be allelic withthis form. On a nonagouti backgroundcompound heterozygotes (aacchmch) showan intermediate phenotype that is verysimilar to that of the Siamese mouse (Musmusculus) and rat (Rattus norvegicus).Homozygotes (aacchmcchm) display a verydark acromelanistic phenotype reminis-cent of that of the sable rabbit (Oryctola-gus cuniculus). The gray phenotype (gg)in the Mongolian gerbil resembles the al-bino locus phenotype chinchilla (cchcch) inmice. We show that the new mutant is notallelic with gray. Fertility and viability of thenew mutant are within normal range.

In 1866 the French Father Armand Davidsent some specimens of an unknown ro-dent species to the Musee d’Histoire Na-turelle in Paris. At that time ‘‘pere Ar-mand’’ was staying in ‘‘la Mongoliechinoise,’’ perhaps what is now known asnorthwestern Shansi, China (Allen 1940).In 1867 Milne Edwards described thesespecimens as Gerbillus unguiculatus (MilneEdwards 1867). Later, in 1908, they werereassigned by Thomas to Meriones and de-nominated Meriones unguiculatus (Thomas1908). Popular names are Mongolian gerbiland clawed jird (Gulotta 1971). Observa-tions of these animals in their natural en-vironment are described by Agren et al.(1989).

Mongolian gerbils have been bred inlaboratories since 1935, when Dr. C. Ka-suga caught 20 pairs in the basin of theAmur River in eastern Mongolia. Theywere sent to the Kitasato Institute with theintention of using them for rickettsial stud-ies. Miss Michiko Nomura (Central Labo-ratories for Experimental Animals) ob-tained some animals from the KitasatoInstitute in 1949, and in 1954 she sent fourpairs of their offspring to Dr. VictorSchwentker at Brant Lake, New York. Heestablished the first commercial colony inAmerica at the Tumblebrook Farm. At thattime many if not all investigators obtained

their stock from Dr. Schwentker’s colony(Rich 1968). Since then Mongolian gerbilsare used in scientific research all over theworld.

Mongolian gerbils can be managed quiteeasily and thrive on a diet of commercialrodent pellets. They are curious, gentle,easy to handle, and stress resistant. Theycan be transported without problems, re-produce throughout the year and are eco-nomical with water consumption, conse-quently their urine production is low.Their cages remain relatively dry andodorless unlike those of mice (Mus mus-culus) and rats (Rattus norvegicus). Mon-golian gerbils can tolerate high populationdensities in captivity. All these features,combined with the fact that Mongoliangerbils are active both day and night, havemade them very popular as pets and forfancy breeding mainly in Europe and theUnited States.

Gerbils have been used in pharmacolog-ical, parasitological, endocrinological, andcancer research. Their susceptibility todifferent types of bacteria and viruses hasalso been studied (Gulotta 1971; Rich1968). They seem to have an unexpectedlyhigh resistance to radiation (Chang et al.1964). Because of their naturally high in-cidence (1 in 5 animals) of seizures(Thiessen et al. 1968) they have been usedin many types of neurological research(Ellard et al. 1990; Gray-Allan and Wong1990; Loskota et al. 1974a,b). Probably be-cause of their distinct social behavior,they were used in many types of behav-ioral studies as well (Kaplan and Hyland1972; Thiessen and Yahr 1977; Walter et al.1963).

Surprisingly enough, Mongolian gerbilshave rarely been used in genetic research.Their diploid number of chromosomes is44, containing 22 metacentric, 10 subme-tacentric, and 10 acrocentric autosomes, alarge submetacentric X and a smaller sub-metacentric Y chromosome (Nadler andLay 1967). Between 1970 and 1985 sevenmutants arose in scientific laboratory and


Figure 1. Phenotypic effects of gerbil c locus mutations on a nonagouti background shown in three adult animals:(A) black chinchilla medium (aacchmcchm), (B) Siamese (aacchmch), (C) dark-tailed white (aachch).

pet populations (see Table 1). Recently wediscovered four new coat color mutants inthe Mongolian gerbil, three in pet popula-tions and one in a laboratory population.Two of them are allelic and possibly relat-

ed to the extension locus (Petrij F, unpub-lished data). The third one seems to bethe dilution mutation (Pund T and PetrijF, unpublished data), a well-known coatcolor mutation, which is described in

many other mammalian species (Searle1968). The fourth one has proven to beallelic to the albino locus and will be dis-cussed here.

Description of the Mutant

In November 1994 a British fancier discov-ered animals with two new phenotypes ina pet shop in the Republic of Ireland. Bothphenotypes were of an acromelanistictype, although one was lighter than theother. He brought one dark- and threelight-colored animals to the United King-dom (Barker J, personal communication).About 1 year later offspring of these ani-mals were imported to The Netherlandsand Germany by fanciers. These animalsformed the nucleus of our breeding exper-iments.

The darker version has a light brownbody (see Figure 1A). The ventral side isslightly lighter in color in comparison tothe dorsum. No clear demarcation line canbe determined. The color is clearly lighterand more creamy than that of the dark se-pia (Leiper and Robinson 1985) Mongoliangerbil (aagg). Compared with the typicalbrown coat color (aabb) of other rodents,such as mice (Searle 1968; Silvers 1979),the coat color of the new mutant is a light-er shade of brown. Nose, ears, and feet arecovered with dark sepia hairs and the tailhairs are almost black. The scrotum isdark colored and covered with hairs thatresemble the color of the ventral side. Thecolor slowly fades out toward proximal.The nails are dark. The eyes are almostblack but under bright illumination theyshow a dark red glow. The typical whitepatches on the upper lip, under the chin,and across the front feet of nonagouti an-imals are also recognizable in this newmutant. This, together with the nonagoutitail and the absence of the typical whitebelly produced by the agouti allele in thisspecies, makes it likely that this pheno-type has a nonagouti background.

The lighter version (see Figure 1B) canbe described as a diluted form of thedarker one and is reminiscent of the phe-notype of the Siamese mouse (Green 1961)and rat (Moutier et al. 1973). The bodycolor is a very creamy type of brown andthe extremities are light brown. The tail,however, is quite dark and of a dark sepiacoloration. The eyes are dark ruby and areclearly darker than those of pink-eyed di-lution (pp) and pink-eyed white (chch) ani-mals.

Juveniles don’t show the full acromelan-istic features like the adults do (see Figure


Figure 2. Phenotypic effects of gerbil c locus mutations on a nonagouti background shown in three juvenile animals (age 3 weeks): (A) black chinchilla medium (aacchmcchm),(B) Siamese (aacchmch), (C) dark-tailed white (aachch).

2). Until their first molt the dark pheno-types have a more grayish-brown colorwith only a darker-colored tail. Nose, ears,and feet are the same color as the body.Similarly the juvenile of the lighter versionalso shows no dark coloration except forthe tail. Both phenotypes develop dark ex-tremities, ears, and nose as they reach ma-turity. Under the influence of low temper-atures (,158C) the mid-dorsal area tendsto become darker than the rest of thebody. Altogether both phenotypes areclearly acromelanistic.

Another acromelanistic phenotype ofthe Mongolian gerbil has previously beendescribed (Leiper and Robinson 1984;Robinson 1973). In that case the pigmen-tation is more severely affected. The ani-mals of this phenotype are pink-eyedwhites with some light brown hairs on thetail (AAchch). On a pink-eyed dilution back-ground (chchpp) these animals are pseudo-albinos. On a nonagouti background(aachch) the tail can become quite dark,therefore we would like to call this latterphenotype ‘‘dark-tailed white’’ (see Figure1C).

Complete albinism (cc) was reported byMatsuzaki et al. (1989). However, no sub-sequent reports on this phenotype havebeen published. The question ariseswhether this mutation has been lost. Itcannot be excluded that the pink-eyed di-lution mutation (pp) or other modifyinggenes were involved in this specific stock.

Breeding Data

At first the darker animals were checkedfor their assumed nonagouti background.Crossings with nonagouti produced onlynonagouti offspring (51 of 51), showingthat the dark version is actually a nona-gouti variant of the new mutant. In orderto show that the dark mutant breeds true,dark individuals were mated with eachother producing only dark variant F1 ani-mals (104 of 104). A similar experimentwith animals of the light phenotype, sub-sequently called Siamese, resulted in threedifferent phenotypes: dark variants, Sia-mese, and dark-tailed whites. This indi-cates that the Siamese were in fact com-pound heterozygotes of the new mutationand the earlier described more severeform of acromelanism.

To investigate the phenotypic effects ofthe new mutation on a wild-type agoutibackground, dark variants were mated toagouti animals and produced only agoutianimals (51 of 51), and F1 agouti animalswere inbred or backcrossed to dark vari-ant animals. F1 3 F1 matings resulted in 22agoutis, 6 blacks, 2 dark variants, and 4animals of a new phenotype. The new phe-notype can be described as a light grayagouti with a somewhat diluted wild-typeagouti tail (further referred to as chinchil-la medium). F1 agouti 3 dark variant ani-mals resulted in 14 agoutis, 5 blacks, 12dark variants, and 6 chinchilla medium an-imals. Both results are within the expected

ratios of 9:3:3:1 and 1:1:1:1, respectively(see Table 2).

In the chinchilla medium animals itseems that the pheomelanin, which is al-most completely washed out on the body,is less affected on the tail. The nose andears of some individuals are slightly dark-ened. On close examination the mid-dorsalbanded hairs (mainly awls) have a light se-pia base, a creamy white subapical band,and light sepia tips. When compared towild-type agouti it is clear that the newmutation dilutes pheomelanin more thaneumelanin. Because the different pheno-types are more pronounced on a nonagou-ti than on an agouti background, all fur-ther experiments were performed on anonagouti background.

Dark-variant animals were mated withpink-eyed white and gray agouti (gg) ani-mals. Since acromelanism is a typical fea-ture of the albino locus, pink-eyed whitewas chosen. For the same reason grayagouti was chosen: it mimics the chinchillamutation at the albino locus of other ro-dent species (Leiper and Robinson 1985).Crossings between dark variants and pink-eyed whites only produced animals withthe Siamese phenotype (118 of 118). Cross-ings between dark variants and gray agou-tis only produced agoutis (39 of 39). Sia-mese F1 animals were inbred to produce anF2 of 40 dark variants, 59 Siamese, and 42pink-eyed whites. Siamese backcrossed todark animals produced 56 dark and 48 Si-amese animals, and Siamese backcrossed


Table 2. Segregation ratios of crosses

Cross Total

Progeny phenotype

Agouti BlackChinchillamedium

Blackchinchillamedium Siamese

Pink-eyedwhite

Expectedratio x2

Black chinchilla medium 3 black chinchilla medium 104 104Black chinchilla medium 3 black 51 51Black chinchilla medium 3 pink-eyed white 118 118Black chinchilla medium 3 gray agouti 39 39Black chinchilla medium 3 agouti 51 51F1 agouti 3 F1 agouti 34 22 6 4 2 9:3:3:1 1.35 (P . .50)F1 agouti 3 black chinchilla medium 37 14 5 6 12 1:1:1:1 6.35 (P . .05)Siamese 3 Siamese 141 40 59 42 1:2:1 3.80 (P . .10)Siamese 3 black chinchilla medium 104 56 48 1:1 0.62 (P . .25)Siamese 3 pink-eyed white 24 12 12 1:1 0.00 (P 5 1.00)

Table 3. Comparison of acromelanistic pheno- and genotypes in rabbits, gerbils, mice, and rats

SpeciesFullcolor

Darkacromela-nistic

Mediumacromela-nistic

Lightacromela-nistic

Faintacromela-nistic Albino Reference

Rabbit CC cchmcchm cchmcchl chch cchlcchl cc Robinson (1958)cchmch cchmc cchlch

cchlcchc

Gerbil CC cchmcchm cchmch — chch cca Robinson (1973), Matsuzaki et al.(1989), this article

Mouse CC — chch — chc cc Green (1961), Silvers (1979)Rat CC — chch — chc cc Moutier et al. (1973)

All described phenotypes are on a nonagouti (aa) background.a Mutation possibly lost (see text).

to pink-eyed whites produced 12 pink-eyedwhites and 12 Siamese. Segregation ratiosand statistical analyses of all crosses aresummarized in Table 2.

Our breeding data indicate an autoso-mal inheritance: no significant differencesbetween distribution of phenotypes overthe sexes could be observed. Fertility andviability seem to be within normal range.

In view of the fact that the (darker) ho-mozygous phenotype has an acromelan-istic expression, which is a typical featureof mutations at the c locus, and also be-cause it is allelic with pink-eyed white (ch),this new mutation can be located at the clocus with confidence. In comparison tothe acromelanistic mouse and rat the newgerbil mutation differs in phenotype. Al-though similar acromelanistic phenotypesin rabbits (sable rabbits) show a darkercoloration, their color characteristics andinheritance pattern have more resem-blance with the here described mutationin the gerbil (see Table 3). Therefore wewould like to propose to designate thegene symbol cchm (chinchilla medium).

Discussion

The breeding data indicate that the mu-tation we describe is inherited as an au-tosomal recessive trait since all F1 animals

of mutant 3 agouti matings are of an agou-ti phenotype. In other words cchm is fullyrecessive to C. Since all F1 animals fromblack chinchilla medium 3 pink-eyed whitematings are of an intermediate phenotype,cchm is shown to be codominant to ch.

Because in the Mongolian gerbil the clocus is linked to the p locus (Leiper andRobinson 1986) this will have conse-quences for the distribution of pheno-types in the F2 progeny of, for instance, ablack chinchilla medium (aacchmcchmPP) 3argente golden (AACCpp) mating. Thesetypes of experiments were not performedby us, but it is surely worth testing in or-der to confirm the linkage. Another inter-esting question to investigate would bewhether pp has a dominance modifying ef-fect on the dominance of C over cchm, likeit has on the dominance of C over ch (Lei-per and Robinson 1984). Preliminary dataindicate that this is indeed the case. Boththe A-Ccchmpp and the aaCcchmpp animalsare of a lighter color than the correspond-ing CCpp animals. In comparison to the ar-gente creme (A-Cchpp) the A-Ccchmpp ani-mal has a richer cast of yellow. Incomparison to the silver (aaCchpp), theaaCcchmpp animal is of a more bluish cast.Both varieties have darker red eyes. Wepropose to call these phenotypes argentefawn and sapphire, respectively.

Furthermore, it would be interesting tosee whether cchmcchm has a dominancemodifying effect on the dominance of Pover p. In the mouse it has been reportedthat albinism (cc) and pink-eyed dilution(pp) have a reciprocal dominance modi-fying effect (Silvers 1979). In case of thepink-eyed whites (chch), this would be dif-ficult to test because of the severe dilutionof body color, only on the tail of dark-tailed white animals the difference be-tween aachchPP and aachchPp might bevisible. In chinchilla medium animals,however, the amount of remaining pigmen-tation will be easier to assess.

Mutations at the c locus affect the syn-thesis of tyrosinase (monophenol oxygen-ase). This copper-containing enzymeplays a major role in the synthesis of mel-anin by the catalyzation of the oxidationof tyrosine to L-dopa (3,4 dihydroxyphen-ylalanine) and the dehydrogenation of L-dopa to dopaquinone. These two catalyticreactions form the first two steps in themelanin biosynthetic pathway. Dopaqui-none is a common precursor for eumelan-in as well as pheomelanin. Certain tyrosi-nase mutations produce an extremeunstable and thermosensitive form of theenzyme (Halaban et al. 1988). In thesecases melanin production is greater incolder areas of the body, leading to thetypical acromelanistic phenotype seen inseveral mammalian species (Searle 1990),including humans (Giebel et al. 1991).

Because of the above described similarphenotype, inheritance pattern, and com-parable interaction with the earlier de-scribed severe form of acromelanism as-signed to the c locus, it is likely that thenew mutation occurred at the albino locusof the Mongolian gerbil. Ultimate proofhas to come from molecular studies,which will become possible as soon as thetyrosinase gene of the Mongolian gerbilhas been cloned.


Fred Petrij is affiliated with the Department of ClinicalGenetics, Erasmus University, Westzeedijk 112, 3016 AHRotterdam, The Netherlands. The authors wish tothank Julian Barker for critical reading. Address cor-respondence to Fred Petrij at the address above or e-mail: [email protected].


References

Agren G, Zhou Q, and Zhong W, 1989. Ecology and so-cial behaviour of Mongolian gerbils, Meriones unguicu-latus, at Xilinhot, Inner Mongolia, China. Anim Behav37:11–27.

Allen GM, 1940. The mammals of China and Mongolia.New York: American Museum of Natural History.

Chang MC, Hunt DM, and Turbyfill C, 1964. High resis-tance of Mongolian gerbils to irradiation. Nature 203:536–537.

Cramlet SH, Toft JD II, and Olsen NW, 1974. Malignantmelanoma in a black gerbil (Meriones unguiculatus).Lab Anim Sci 24:545–547.

Ellard CG, Stewart DJ, Donaghy S, and Vanderwolf CH,1990. Behavioral effects of neocortical and cingulate le-sions in the Mongolian gerbil. Behav Brain Res 36:41–51.

Giebel LB, Tripathi RK, King RA, and Spritz RA, 1991. Atyrosinase gene missense mutation in temperature-sen-sitive type I oculocutaneous albinism. J Clin Invest 87:1119–1122.

Green MC, 1961. Himalayan, a new allele of the albinoin the mouse. J Hered 25:73–75.

Gray-Allan P and Wong R, 1990. Influence on coat colorgenes on seizure behavior in Mongolian gerbils. BehavGenet 20:481–485.

Gulotta EF, 1971. Meriones unguiculatus. MammalianSpecies 3:1–5.

Halaban R, Moellmann G, Tamura A, Kwon BS, Kuklin-ska E, Pomerantz SH, and Lerner AB, 1988. Tyrosinasesof murine melanocytes with mutations at the albino lo-cus. Proc Natl Acad Sci USA 85:7241–7245.

Henley M and Robinson R, 1981. Non-agouti and pink-eyed dilution in the Mongolian gerbil. J Hered 72:60–61.

Kaplan H and Hyland SO, 1972. Behavioral develop-ment in the Mongolian gerbil (Meriones unguiculatus).Anim Behav 20:147–154.

Leiper BD and Robinson R, 1984. A case of dominancemodification in the Mongolian gerbil. J Hered 75:323.

Leiper BD and Robinson R, 1985. Gray mutant in theMongolian gerbil. J Hered 76:473.

Leiper BD and Robinson R, 1986. Linkage of albino andpink-eyed dilution in the Mongolian gerbil and otherrodents. J Hered 77:207.

Loskota WJ, Lomax P, and Rich ST, 1974a. The gerbilas a model for epilepsies. Epilepsia 15:109–119.

Loskota WJ, Lomax P, and Verity MA, 1974b. A stereo-taxic atlas of the Mongolian gerbil brain (Meriones un-guiculatus). Ann Arbor, MI: Ann Arbor Science Publish-ers.

Matsuzaki T, Yasuda Y, and Nonaka S, 1989. The genet-ics of coat colors in the Mongolian gerbil (Meriones un-guiculatus). Exp Anim 38:337–341.

Moutier R, Toyama K, and Charrier MF, 1973. Himalay-an allele at the albino locus in the Norway rat. J Hered64:303–304.

Milne Edwards MA, 1867. Observations sur quelquesmammiferes du nord de la Chine. Ann Sci Nat (Zool)7:375–377.

Nadler CF and Lay DM, 1967. Chromosomes in somespecies of Meriones (Mammalia:Rodentia). Z Sauge-tierk 32:285–291.

Rich ST, 1968. The Mongolian gerbil (Meriones ungui-culatus) in research. Lab Anim Care 18:235–243.

Robinson R, 1958. Genetic studies of the rabbit. In: Bib-liographia Genetica, Part XVII (Sirks MJ and GoddijnWA, eds). The Hague: Martinus Nijhoff; 229–510.

Robinson R, 1973. Acromelanic albinism in mammals.Genetica 44:454–458.

Searle AG, 1968. Comparative genetics of coat colourin mammals. New York: Academic Press.

Searle AG, 1990. Comparative genetics of albinism.Ophthal Paediatr Genet 11:159–164.

Silvers WK, 1979. The coat colors of mice. Berlin:Springer Verlag.

Silverstein A and Silverstein VB, 1976. Gerbils. Phila-delphia: J. B. Lippincott.

Swanson HH, 1980. The ‘hairless’ gerbil: a new mutant.Lab Anim 14:143–147.

Thiessen DD, Lindzey G, and Friend HC, 1968. Sponta-neous seizures in the Mongolian gerbil, Meriones un-guiculatus. Psychonomic Sci 11:227–228.

Thiessen DD and Yahr P, 1977. The gerbil in behavioralinvestigations: mechanisms of territoriality and olfac-tory communication. Austin: University of Texas Press.

Thomas O, 1908. The Duke of Bedford’s zoological ex-ploration in eastern Asia—IX. List of mammals from theMongolian plateau. Proc Zool Soc Lond 1908:104–110.

Walter GC, Pearl J, and Rogers JV, 1963. The gerbil asa subject in behavioral research. Psych Rep 12:315–318.

Waring AD and Poole TW, 1980. Genetic analysis of theblack pigment mutation in the Mongolian gerbil. J Her-ed 71:428–429.

Waring AD, Poole TW, and Perper T, 1978. White spot-ting in the Mongolian gerbil. J Hered 69:347–349.

Received April 14, 2000Accepted August 31, 2000

Corresponding Editor: Neal Copeland

A Linkage Map forCRINKLED PETAL: AHomeotic Gene of Clarkiatembloriensis (Onagraceae)

L. L. Y. Leong, K. G. Wilson, andN. L. Smith-Huerta

Homeotic mutations in flowers lead to thedevelopment of floral organs in abnormallocations. In most laboratory-induced ex-amples of this type of mutation, two adja-cent whorls of organs are affected, result-ing in two whorls of abnormal organformation. However, the crinkled petal mu-tant of Clarkia tembloriensis is interestingbecause it is a naturally occurring muta-tion and it affects only the second whorl oforgans, producing sepaloid petals. In thisstudy one wild-type population (CantuaCreek-2) and one crinkled petal mutantpopulation (Red Rocks) were comparedusing 181 different primers in random am-plified polymorphic DNA (RAPD) analysis.Bulk DNA from each parent population

and their subsequent crosses were usedto compare the genetic differences be-tween the two populations and to searchfor molecular markers linked with theCRINKLED PETAL locus. A linkage mapwas developed for the CRINKLED PETALgene, and markers were discovered whichflanked both sides of the locus.

Clarkia tembloriensis (Onagraceae) is awildflower found growing in discrete pop-ulations in the Temblor region of southernCalifornia. Within these populations, flow-ers with two distinct petal phenotypeshave been observed (Vasek 1964). Thetypical wild-type flowers of C. temblorien-sis possess four petals composed of asmooth-textured deltoid limb with a slen-der claw. These petals are bright pink incolor, with a maroon spot at the base ofthe limb. All four petals are expanded andvery prominently displayed. In strikingcontrast are the flowers from the homeot-ic mutant called crinkled petal (cp) (Smith-Huerta 1992, 1996; Vasek 1964). The cp mu-tant flowers also possess four ‘‘petals,’’but these organs are small and linear-lan-ceolate in shape, with abundant tri-chomes. They have a greenish-pink, wrin-kled appearance, lack the maroon spot atthe base, and are reflexed backward. Inprevious developmental studies, Smith-Huerta (1992) showed that the secondwhorl organs of the cp mutant are initiatedin the proper time and location for wild-type petals, but in terms of color, mor-phology, and growth rate, they closely re-semble wild-type sepals.

The crinkled petal phenotype occurs inat least 10 natural populations in Califor-nia (Vasek 1964) and is controlled by asingle recessive gene (Vasek 1966). Insome populations both the wild-type andcrinkled petal phenotypes occur together;in others, only one of the phenotypes ispresent (Vasek 1964). In this study, plantsfrom the wholly wild-type population atCantua Creek-2 (CC-2) were comparedwith plants from the cp mutant populationat Red Rocks (RR). These two populationsshow minor differences in germinationtime, growth rate, and branching pattern(Leong L, personal observation), as wellas the obvious major difference in floralmorphology.

In this study we attempted to locate theCRINKLED PETAL gene using random am-plified polymorphic DNA analysis (Welshand McClelland 1990; Williams et al. 1990)to examine the genetic differences be-tween two populations of C. tembloriensisthat display different petal phenotypes.


Table 1. Linkage groups for C. tembloriensis

Group 1 Group 2

UBC-1631040 UBC-211360

UBC-207425 UBC-211405

UBC-251410 UBC-281450

cp1 (wild-type petal)

Band names indicate the primer number along with theapproximate size of the band in base pairs. All othermarkers were determined to be unlinked.

Figure 1. Mature flowers of C. tembloriensis: WT onleft, cp mutant on right. Map of markers linked with thecp1 allele.

We used RAPDs with template DNA fromparental wild-type and mutant plants, andDNA from subsequent crosses [F3 for mu-tants; F3 and selfed backcross (BC2) forwild type] to examine the genetic varia-tion between the CC-2 wild-type petal andRR crinkled petal populations. We thenused the observed differences to searchfor markers linked with the gene respon-sible for the petal phenotype. Further-more, by comparing the linkage of poly-morphic bands with the segregation of thewild-type and cp mutant petal phenotypesin homozygous individuals (cp1/cp1 orcp/cp) of the F3 generation, a linkage mapwas developed for the CRINKLED PETALgene of C. tembloriensis.


Clarkia tembloriensis seeds from the wild-type CC-2 population and the RR mutantpopulation were sown on vermiculite andgerminated in a growth chamber under a12-h light/12-h dark cycle with a constanttemperature of 138C (Smith-Huerta 1992).Seedlings were transplanted into MetroMix (Scott’s) and grown in a growth cham-ber under a 16-h light/8-h dark cycle at aconstant temperature of 238C. Plants werewatered as necessary with a dilute fertil-izer solution (Liquid Miracle Gro, Scott’s).To produce F1, F2, F3, and backcross gen-erations, flowers were emasculated 1 dayprior to anthesis and hand pollinationswere performed. F3 populations were usedto determine the genotype of the wild-typeF2 parents relative to the cp locus.

Healthy bud, stem, and shoot tissueswere collected from plants of each typeand generation. DNA extraction was per-formed using a CTAB method based onDoyle and Doyle (1987) with a 43 CTABextraction buffer plus 5% PVP and 1% so-dium bisulfite (dissolved in the buffer be-fore tissue was added). DNA was resus-pended in sterile deionized water. Saltconcentrations were adjusted to 0.25 M,and then the protocol according to Mi-chaels et al. (1994) was used to removeRNA and clean up polysaccharide contam-

inants. DNA was stored in sterile water at2208C.

Random amplified polymorphic DNA-polymerase chain reaction (RAPD-PCR;Welsh and McClelland 1990; Williams et al.1990) was performed in a thermocycler(Perkin-Elmer GeneAmp PCR System 2400)using the following components for each12.5 ml reaction: 0.2 mM primer, 0.2 mM ofeach dNTP, 4 mM MgCl2, 103 Stoffel Frag-ment PCR Buffer (final concentration 10mM KCl, 10 mM Tris-HCl pH 8.3), 2%DMSO, 23 BSA, and 2 units of AmpliTaqDNA Polymerase Stoffel Fragment (Perkin-Elmer, Foster City, CA). The following pro-gram was used: 948C for 3 min, followedby 50 cycles of 948C for 1 min, 368C for 1min, and 728C for 2 min. The reactionswere then terminated by an extension of728C for 7 min and maintained at 48C untilloading for gel electrophoresis. PCR prod-ucts were run on 1.3% agarose in 13 TAEbuffer, and gels were then stained withethidium bromide and photographed un-der ultraviolet (UV) light. Photographyand analyses were performed using theAlphaImager camera system (Alpha Inno-tech, San Leandro, CA). Gels were com-pared visually for presence or absence ofpolymorphic bands. A total of 181 differ-ent random 10-mer primers (UBC-120–UBC-300; Nucleic Acid Protein Service,University of British Columbia, Vancouver,BC, Canada) were used in RAPD-PCR tocompare the banding patterns of the twopopulations.

For initial population comparisons,DNAs from nine plants of the parental CC-2 population were combined together, aswere DNAs from nine plants from the RRpopulation (called CC-2 bulk and RR bulk,respectively). The purpose of bulking the

DNAs from each population was to pro-vide an estimate of the variation betweenthe two populations and to disregard vari-ation between individuals within eachpopulation (based on the idea of ‘‘bulkedsegregant analysis’’ from Michelmore et al.1991). An ethanol precipitation from Mi-chaels et al. (1994) was used to clean thebulked DNAs again. Bulk CC-2 and bulk RRDNAs were used in RAPD-PCR with each ofthe 181 primers. Primers that resulted inpolymorphisms between the two popula-tion bulks were checked at least twice toconfirm the polymorphism. Those primersproducing a confirmed polymorphismwere tested for linkage with the petal phe-notype by running PCRs with DNA fromvarious F3 individuals bearing the mutantphenotype (genotype cp/cp). Primers re-sulting in a majority of mutant F3 individ-uals showing the same band phenotype asone of the parental types were used againon F3 and BC2 (selfed backcross) individ-uals known to be homozygous for thewild-type cp1 allele (genotype cp1/cp1).

Segregation data for RAPD markers as-sociated with the petal phenotype wereanalyzed using the genetic analysis pro-gram MapMaker/Exp version 3.0 (Landeret al. 1987; Lincoln et al. 1992).

Results

One hundred eighty-one 10-mer primerswere screened through PCR with bulkDNAs from the CC-2 and RR populations,resulting in a total of 1060 scorable bands;1018 bands were present in both popula-tions (96.0%) and 42 were present in onlyone population (4.0%). To check furtherfor linkage with the petal phenotype, PCRwas performed on template DNA from in-dividuals of the F3 generation using thoseprimers that produced polymorphicbands. Of the bands that were found to bepolymorphic between the two popula-tions, 30 showed segregation of the bandin the F3 population and were used to cre-ate a linkage map for the cp gene. Thosebands that did not segregate at all in theF3 were considered uninformative for map-ping purposes.

Two linkage groups were determined us-ing MapMaker/Exp version 3.0 (Table 1).Group 1 is significant because it demon-strates linkage between the cp locus andthree different molecular markers. Thelinkage map for this group is shown in Fig-ure 1. Primer sequences for the markersused in mapping were as follows: 59-CCCCCC AGA T-39 (primer UBC-163); 59-CATATC AGG G-39 (primer UBC-207); 59-CTT


GAC GGG G-39 (primer UBC-251). Themarker UBC-1631040 correlated especiallystrongly with the wild-type allele, mappingonly 4.6 cM apart. This band occurs in thewild-type CC-2 bulk and most of the F3 andBC2 individuals homozygous for the wild-type allele. It does not occur in the mutantRR bulk nor in any of the F3 mutant indi-viduals except F3-25-08, indicating a tightlinkage with the wild-type allele.

Discussion

When no sequence information is avail-able for an organism, one way to deter-mine the location of a gene is to identifymarkers that are physically close to it ona chromosome. The closer the proximityof two markers on a chromosome, themore likely they will segregate together,and the less likely that they will cross overseparately during meiosis. Three bandswere determined to be linked with thewild-type petal allele: UBC-1631040, UBC-207425, and UBC-251410. In fact, the markerUBC-1631040 occurs in all scorable in-stances of the wild-type phenotype in theF3 and BC2 generations, with a very lowfrequency of occurrence in the F3 mutantphenotype plants. Preliminary attempts toclone this marker were unsuccessful, butthis may be a very useful marker for futureattempts to locate the position of the cplocus. The likelihood of finding the cp lo-cus in the future is enhanced by the factthat linked markers have been located onboth sides of this locus: UBC-1631040 onone side and UBC-251410 on the other. As ademonstration of the potential usefulnessof this finding, Thorlby et al. (1997) wereable to determine a fine-scale molecularmap position for the Arabidopsis male ste-rility gene ms1 by measuring the recom-bination frequency of two molecular mark-ers which flank both sides of the ms1 gene.A similar mapping study could be per-formed on the cp gene with its flankingmarkers. Furthermore, by creating andsearching a genomic DNA library for C.tembloriensis, it might be possible to findcontiguous DNA fragments that span theregion between the two flanking markers.A chromosome walk across this regioncould then lead to a more accurate map-ping of the cp locus and would make se-quencing the cp gene a viable option.

To explain the mechanism of floral or-gan development, Bowman et al. (1991)have suggested a series of three overlap-ping floral factors or functions called A, B,and C which, when expressed in differentcombinations, specify the development of

the four different floral whorls. The ex-pression of factor A alone results in se-pals, factors A and B together producepetals, B and C make stamens, and C alonestimulates the production of carpels. Ac-cording to this ‘ABC model,’ a mutation inone of these factors should have an im-pact on two adjacent whorls of organssince each factor is required for the de-velopment of two whorls. Indeed, this pre-diction is consistent with many studies de-scribing homeotic mutations that disruptthe development of two adjacent floralwhorls in a variety of plant systems (Coenand Meyerowitz 1991). However, not allhomeotic mutants behave exactly as thismodel predicts. Exceptions such as the su-perman mutation of Arabidopsis, which af-fects the carpel whorl only (Bowman et al.1992; Sakai et al. 1995), and the green pet-als mutation of Petunia which affects thepetals only (Angenent et al. 1992; Halfteret al. 1994; van der Krol and Chua 1993;van Tunen and Angenent 1991), do not fol-low the pattern of the basic ABC model. Inthe same way, the cp mutant does not fitthe normal model, as it produces a mutantpetal whorl, but otherwise appears to beunaffected anywhere else. However, thisdeviance from the model does not neces-sarily imply that the ABC model is not rel-evant for C. tembloriensis or other one-whorl mutants. Although the cp mutant isconsidered a class B mutant (Smith-Huer-ta 1992), it may be that the CRINKLED PET-AL gene is not responsible for the produc-tion of the ‘‘B’’ factor of C. tembloriensis,but is involved instead in the response ofthe plant to the presence of the A 1 B fac-tor product.

The crinkled petal mutant of C. temblo-riensis is significant not only because itrepresents a deviation from the more com-mon two-whorl mutant effect, but also be-cause it is not a laboratory-induced mu-tation as is the case of most of thecommonly studied floral mutants. The cpmutation has been observed in at least 10populations in the wild (Vasek 1964). Oth-er studies have reported apparently ho-mologous sepaloid petal phenotypes inother species of Clarkia (crinkled petalform in C. exilis, Vasek 1966; bicalyx mu-tant in C. concinna, Ford and Gottlieb1992) and in other genera of the Onagra-ceae including Epilobium and Oenothera(cruciata mutants; Renner 1959). All ofthese mutants are fully fertile and appearto be unaffected in the other three floralwhorls. The existence of these mutationssuggests that the single-whorl mutant con-dition, while by no means a common oc-

currence, is not as unusual in nature asmight be anticipated, at least in the Ona-graceae. Further pursuit of the cp locus inC. tembloriensis could thus prove very use-ful to other studies. Although it wasshown by Vasek (1966) that the CRINKLEDPETAL genes of C. tembloriensis and C. ex-ilis are homologous, it is not knownwhether the same is true for the rest ofthe sepaloid petal mutations in the Ona-graceae, or for other sepaloid petal mu-tants in other plant families. Continued in-vestigation of the crinkled petal mutant ofC. tembloriensis and other similar mutantswill help us to compare and contrast therelationships between the different floralwhorls in different plant systems and mayhelp to provide insight into the evolutionof the flower itself.

From the Department of Botany, Miami University, Ox-ford, Ohio. We thank A. Huerta for assistance in prep-aration of the figures, and N. Money and two anony-mous reviewers for helpful comments on themanuscript. This project was supported by AcademicChallenge grants from the Botany Department of MiamiUniversity, and by a grant-in-aid of research from SigmaXi, the Scientific Research Society. This article repre-sents a portion of a master’s thesis submitted by L. L.Y. Leong to the Department of Botany, Miami Univer-sity, Oxford, Ohio. Address correspondence to LaurelLeong, Department of Botany, Miami University, Ox-ford, OH 45056, or e-mail: [email protected].


References

Angenent GC, Busscher M, Franken J, Mol JNM, and vanTunen AJ, 1992. Differential expression of two MADSbox genes in wild-type and mutant petunia flowers.Plant Cell 4:983–993.

Bowman JL, Smyth DR, and Meyerowitz EM, 1991. Ge-netic interactions among floral homeotic genes of Ar-abidopsis. Development 112:1–20.

Bowman JL, Sakai H, Jack T, Weigel D, Mayer U, andMeyerowitz EM, 1992. SUPERMAN, a regulator of floralhomeotic genes in Arabidopsis. Development 114:599–615.

Coen ES and Meyerowitz EM, 1991. The war of thewhorls: genetic interaction controlling flower develop-ment. Nature 353:57–59.

Doyle JJ and Doyle JL, 1987. A rapid DNA isolation pro-cedure for small quantities of fresh leaf tissue. Phyto-chem Bull 19:11–15.

Ford VS and Gottlieb LD, 1992. Bicalyx is a natural hom-eotic floral variant. Nature 358:671–673.

Halfter U, Ali N, Stockhaus J, Ren L, and Chua N-H,1994. Ectopic expression of a single homeotic gene, thePetunia gene green petal, is sufficient to convert sepalsto petaloid organs. EMBO J 13:1443–1449.

Lander E, Green P, Abrahamson J, Barlow A, Daley M,Lincoln S, and Newburg L, 1987. MAPMAKER: an inter-active computer package for constructing primary ge-netic linkage maps of experimental and natural popu-lations. Genomics 1:174–181.

Lincoln S, Daly M, and Lander E, 1992. Constructinggenetic maps with MAPMAKER/EXP 3.0, 3rd ed. White-head Institute technical report. Cambridge, MA: White-head Technical Institute.

Michaels SD, John MC, and Amasino RM, 1994. Removalof polysaccharides from plant DNA by ethanol precip-itation. BioTechniques 17:274–276.


Michelmore RW, Paran I, and Kesseli RV, 1991. Identi-fication of markers linked to disease-resistance genesby bulked segregant analysis: a rapid method to detectmarkers in specific genomic regions by using segregat-ing populations. Proc Natl Acad Sci USA 88:9828–9832.

Renner O, 1959. Somatic conversion in the heredity ofthe cruciata character in Oenothera. Heredity 13:283–288.

Sakai H, Medrano LJ, and Meyerowitz EM, 1995. Role ofSUPERMAN in maintaining Arabidopsis floral whorlboundaries. Nature 378:199–203.

Smith-Huerta NL, 1992. A comparison of floral devel-opment in wild type and a homeotic sepaloid petal mu-tant of Clarkia tembloriensis (Onagraceae). Am J Bot 79:1423–1430.

Smith-Huerta NL, 1996. A comparison of venation pat-tern development in wild type and a homeotic sepaloidpetal mutant of Clarkia tembloriensis (Onagraceae). AmJ Bot 83:712–716.

Thorlby GJ, Shlumukov L, Vizir IY, Yang C-Y, MulliganBJ, and Wilson ZA, 1997. Fine-scale molecular genetic(RFLP) and physical mapping of a 8.9 cM region on thetop arm of Arabidopsis chromosome 5 encompassingthe male sterility gene, ms1. Plant J 12:471–479.

van der Krol AR and Chua N-H, 1993. Flower develop-ment in Petunia. Plant Cell 5:1195–1203.

van Tunen AJ and Angenent GC, 1991. How general arethe models describing floral morphogenesis? FlowerNewslett 12:34–37.

Vasek FC, 1964. The evolution of Clarkia unguiculata de-rivatives adapted to relatively xeric environments. Evo-lution 18:26–42.

Vasek FC, 1966. A case of gene homology in Clarkia.Evolution 20:243–244.

Welsh J and McClelland M, 1990. Fingerprinting ge-nomes using PCR with arbitrary primers. Nucleic AcidsRes 18:7213–7218.

Williams JGK, Kubelik AR, Livak KJ, Rafalski JA, andTingey SV, 1990. DNA polymorphisms amplified by ar-bitrary primers are useful as genetic markers. NucleicAcids Res 18:6531–6535.

Received February 21, 2000Accepted November 3, 2000

Corresponding Editor: Susan Gabay-Laughnan

Production of WheatDoubled Haploids byPollination With Job’s Tears(Coix lachryma-jobi L.)

K. Mochida and H. Tsujimoto

Wheat (Triticum aestivum L.) haploidswere produced by crossing with Job’stears (Coix lachryma-jobi L.) as the pollenparent. Pollination was followed by 2,4-Dtreatment, detached tiller culture, and em-bryo culture, as described for maize pol-lination. The frequency of embryo forma-tion was similar to that obtained bycrossing wheat with maize pollen. Job’stears is a perennial plant which forms sev-eral stalks and its pollen can be collectedthroughout the year when the plant ismaintained in a controlled environment.Our results indicate that Job’s tears can be

used as the pollen parent for wheat cross-es for haploid production without requiringsynchronization of flowering dates.

Haploid production followed by chromo-some doubling results in the creation ofgenetically pure lines within a short peri-od of time. This is an important procedureto shorten the time for crop improvement.In addition, a set of double haploid linesis useful for analyses of quantitative traitloci (QTLs) (Cadalen et al. 1998; Yan et al.1999).

In wheat, haploids can be produced byinterspecific crosses. In this procedure,haploid embryos appear as a result of se-lective elimination of the alien chromo-somes during embryogenesis. Hordeumbulbosum was first used as the pollen par-ent for this purpose (Barcley 1975). How-ever, subsequent studies indicated thatthis species was inappropriate for wheatcultivars due to the presence of dominantKr genes, which reduce crossability (Sna-pe et al. 1979). When maize pollen iscrossed with wheat stigma, the sperm nu-clei fertilize the wheat egg (Laurie andBennett 1986; Zenkteler and Nitzsche1984). However, the maize chromosomesare eliminated at early embryonic stages(Laurie and Bennett 1989). When hexap-loid wheat or durum wheat is crossed withmaize pollen, the efficiency of haploid pro-duction is not affected by Kr genes (Al-mouslem et al. 1998; Inagaki 1997; Laurieand Bennett 1987, 1988b). Other speciesrelated to maize, such as Teosinte (Zeamays L. spp. mexicana) and eastern gam-agrass (Tripsacum dactyloides (L.) L.), arealternative pollen donors for wheat hap-loid production (Li et al. 1996; Riera-Lizar-azu and Mujieeb-Kazi 1993; Suenaga et al.1998; Ushiyama et al. 1991).

Cytological studies have provided evi-dence for the successful fertilization andelimination of paternal chromosomes fromhybrid zygote in sorghum (Sorghum bicol-or (L.) Moench) and pearl millet (Penni-setum glaucum (L.) R. Br.) crosses, andthus these species are also possible polli-nators for wheat haploid production (Ah-mad and Comeau 1990; Inagaki and Mu-jeeb-Kazi 1995; Laurie 1989; Laurie andBennett 1988a). Here we report a new po-tential pollinator, Job’s tears (Coix lachry-ma-jobi L.). Job’s tears is a vigorous peren-nial grass of the tribe Maydeae, subfamilyPanicoideae of Poaceae. It includes bothcultivated and wild plants of SoutheastAsian origin. The plant has thick rosettingstems. The top of the stem divides intoseveral branches, and the plant is about

60–100 cm high. The cultivated type (ssp.mayuen) is used as food and medicine ineastern and southern Asian countries. Thewild species inhabits wet ground such asriverbanks and the flowers bloom in sum-mer to autumn. One male inflorescencecontains several florets, and thus collec-tion of the pollen is easy.

In this study we describe the potentialuse of Job’s tears as a novel pollinator forwheat double-haploid production.


Plant MaterialCommon wheat (Triticum aestivum L.) cv.Chinese Spring was used as the femaleparent. Wheat plants were grown in pot-ting soil in a temperature-controlled green-house [108C/238C (min/max)]. Job’s tearsplants were collected from the Maioka Riv-er close to our institution and were trans-ferred to pots. As a control, the maize line919J, which was kindly provided by theMaize Stock Center, was used as the pollenparent. Both pollen parents were grownin a temperature-controlled greenhouse[188C/278C (min/max)] with a light:darkcycle of 16:8 h. Flowering of Job’s tearswas maintained by growing lateral buds.

CrossingThe size of the anthers of Job’s tears is sim-ilar to that of maize anthers (5–7 mm) (Fig-ure 1). Several anthers could be obtainedfrom the flowering florets in the male inflo-rescence. Fresh anthers of Job’s tears col-lected by forceps and placed on the palmof the hand readily dehisced and were usedfor fertilization of the emasculated wheatspikes. Maize pollen was used as a controlto pollinate wheat, using the method de-scribed previously by Inagaki (1997).

Detached Tiller CultureAfter pollination, wheat tillers were cut offand cultured for 14 days in a solution con-taining sucrose (4 %) and 2,4-D (100 mg/l). They were incubated in a growth cham-ber set at a constant temperature of22.58C, with a 16:8 h light:dark cycle and80–90% relative humidity. The liquid me-dium was regularly changed with fresh me-dium throughout the cultivation period.

Double-Haploid ProductionCulture embryos, plant rescue, and chro-mosome doubling were performed accord-ing to the procedures described by Inagaki(1997). The seeds were removed from thespikes and immersed in 70% ethanol for afew seconds then sterilized for 10 min in


Figure 1. Male inflorescences of (A) Job’s tears and(B) maize. Bar 5 5 mm. (C) Job’s tears inflorescencesand (D) whole plant in a pot.

Figure 2. Comparison of wheat embryo formed (A)by crossing wheat and Job’s tears and (B) by crossingwith maize. Bar 5 1 mm.

Figure 3. Somatic chromosomes (2n 5 3x 5 21) of aplant derived by crossing wheat and Job’s tears.

Table 1. Frequencies of haploid formation following crossing between wheat and Job’s tears or maize

CrossNo. of crossedflorets

No. of formedembryos (%)

No. of haploidplants (%)

Wheat 3 Job’s tears 218 23 (10.6) 6 (2.8)Wheat 3 maize 1196 125 (10.5) 33 (2.8)

a solution containing 1–2% sodium hy-pochloride and a few drops per liter ofTween-20. After rinsing twice with sterilewater, the seeds were dissected on steril-ized filter papers, and the embryos wereexamined under the stereoscopic micro-scope. The embryos were transferred ontoMurashige and Skoog medium (M-5519one pack per liter; Sigma Chemical Co., St.Louis, MO) with 2% sucrose and 0.6% aga-rose in plastic dishes. The embryo rescueprocedure was performed under asepticconditions on a clean bench. The embryoswere cultured at 22.58C in a room set witha light:dark cycle of 16:8 h. After about 1month the new plants grown from the em-bryos were transplanted to potting soiland grown in a greenhouse set at 228C

with relative humidity of 80% and 16:8 hlight:dark cycle.

The plant somatic chromosomes wereexamined in squashed preparations ofroot-tip cells stained with acetocarmine.To induce chromosome doubling, theplants were treated with colchicine solu-tion [0.05% colchicine, 2% dimethyl sulf-oxide (DMSO), 15 drops/l Tween-20] for12–15 h in darkness at room temperature,rinsed with water, and transferred to soil.Self-pollinated seeds were obtained fromthe plants. The likelihood of embryo andhaploid formation was estimated using thefollowing formulas:

Frequency of embryo formation5 Number of embryos carrying cary-

opses4 Total number of pollinated florets

Efficiency of haploid formation5 Total number of regenerated plants

4 Total number of pollinated florets

These values were subjected to analysis ofvariance (ANOVA) after arc sine transfor-mation.

Results and Discussion

Wheat florets crossed with the pollen ofJob’s tears bore caryopses at 14 days afterpollination that were smaller than thoseobtained by self-pollination. The caryop-ses were filled with an aqueous solutioninstead of the solid endosperm normallyfound in the wheat, and some containedimmature embryos (Figure 2). The ob-tained embryos grew well and 30% formednormal shoots and roots. We examinedthe somatic chromosomes of all six regen-

erated plants. The root-tip cells of theseplants carried 21 chromosomes of wheat,indicating a complete elimination of thechromosomes of Job’s tears (Figure 3).ANOVA results indicated that the frequen-cies of embryo and haploid formationwere not significantly different from thoseobtained by crossing with maize pollen(Table 1). Plants were treated with colchi-cine and diploid plants were obtained. Theseeds of the double plants showed 42 nor-mal chromosomes of common wheat.

Collection of pollen from Job’s tears isnot a straightforward process when com-pared to collection of maize pollen. How-ever, Job’s tears is a perennial plant thatforms lateral buds, and thus pollen can beobtained throughout the year (Figure 1).In the present study, although only a fewplants of Job’s tears were grown in thegreenhouse, we were able to cross the pol-len whenever wheat flowered. Thus wewere able to easily produce wheat hap-loids without synchronization of the pol-len donor. Consequently Job’s tears can beused as a suitable pollen donor for wheathaploid production using the wide-crossmethod.

From the Graduate School of Integrated Science and Ki-hara Institute for Biological Research, Yokohama CityUniversity, 641-12 Maioka-cho, Totsuka-ku, Yokohama244-0813, Japan. The authors thank Dr. M. Inagaki, Min-istry of Agriculture, Forestry and Fishery Japan for tech-nical advice about wide hybridization. Address corre-spondence to Hisashi Tsujimoto, at the address above.


References

Ahmad F and Comeau A, 1990. Wheat 3 pearl millethybridization: consequence and potential. Euphytica50:181–190.

Almouslem AB, Jouhar PP, Peterson TS, Bommineni VR,and Rao MB, 1998. Haploid durum wheat productionvia hybridization with Maize. Crop Sci 38:1080–1087.


Barclay IR, 1975. High frequencies of haploid produc-tion in wheat (Triticum aestivum) by chromosome elim-ination. Nature 256:410–411.

Cadalen T, Sourdille P, Charmet G, Tixier MH, Gay G,Boeuf C, Bernard S, Leroy P, and Bernard M, 1998. Mo-lecular markers linked to genes affecting plant heightin wheat using a doubled-haploid population. TheorAppl Genet 96:933–940.

Inagaki MN, 1997. Technical advances in wheat haploidproduction using ultra-wide crosses. JIRCAS J 4:51–62.

Inagaki MN and Mujeeb-Kazi A, 1995. Comparison ofpolyhaploid production frequencies in crosses of hex-aploid wheat with maize, pearl millet and sorghum.Breed Sci 45:157–161.

Laurie DA, 1989. The frequency of fertilization in wheat3 pearl millet crosses. Genome 32:1063–1067.

Laurie DA and Bennett MD, 1986. Wheat 3 maize hy-bridization. Can J Genet Cytol 28:313–316.

Laurie DA and Bennett MD, 1987. The effect of thecrossability loci Kr1 and Kr2 on fertilization frequencyin hexaploid wheat 3 maize crosses. Theor Appl Genet73:403–409.

Laurie DA and Bennett MD, 1988a. Cytological evidencefor fertilization in hexaploid wheat 3 sorghum crosses.Plant Breed 100:73–82.

Laurie DA and Bennett MD, 1988b. The production ofhaploid wheat plants from wheat 3 maize crosses.Theor Appl Genet 76:393–397.

Laurie DA and Bennett MD, 1989. The timing of chro-mosome elimination in hexaploid wheat 3 maize cross-es. Genome 32:953–961.

Li DW, Qio JW, Ouyang P, Yao QX, Dawei LD, Jiwen Q,Ping O and Quingziao Y, 1996. High frequencies of fer-tilization and embryo formation in hexaploid wheat 3Tripsacum dactylodes crosses. Theor Appl Genet 92:1103–1107.

Riera-Lizarazu O and Mujeeb-Kazi A, 1993. Polyhaploidproduction in the triticeae: wheat 3 Tripsacum crosses.Crop Sci 33:973–976.

Snape JW, Chapman V, Moss J, Blanchard CE, and MillerTE, 1979. The crossabilities of wheat varieties with Hor-deum bulbosum. Heredity 42:291–298.

Suenaga K, Morshedi AR, and Darvey NL, 1998. Evalu-ation of teosinte lines as pollen parents for wheat hap-loid production. Cereal Res Commun 26:119–125.

Ushiyama T, Shimizu T, and Kuwabara T, 1991. High fre-quency of haploid production of wheat through inter-genetic cross with teosinte. Jpn J Breed 41:353–357.

Yan J, Zhu J, He C, Benmoussa M, and Wu P, 1999. Mo-lecular marker-assisted dissection of genotype 3 en-vironment interaction for plant type traits in rice (Ory-za sativa L.) Crop Sci 39:538–544.

Zenkteler M and Nitzsche W, 1984. Wide hybridizationexperiments in cereals. Theor Appl Genet 68:311–315.

Received January 13, 2000Accepted August 30, 2000

Corresponding Editor: Prem P. Jauhar

A Recessive Allele InhibitingSaponin Synthesis in TwoLines of Bolivian Quinoa(Chenopodium quinoa Willd.)

S. M. Ward

Quinoa cultivars currently grown in NorthAmerica and Europe require removal ofbitter-tasting saponins from the grain prior

to human consumption. This need for pos-tharvest processing is a barrier to expand-ing production of the crop outside its An-dean area of origin. Grain saponin contentin quinoa shows continuous variation andis considered to be a quantitative trait.However, segregation for the presence orabsence of grain saponin in F2 genera-tions derived from crosses between high-and low-saponin parents indicates a majorgene effect, with plants homozygous for arecessive allele sp1 having no detectablegrain saponin. Variation in saponin levelsamong F2 plants with detectable grainsaponin was consistent with polygenic in-heritance. It appears that grain saponinlevel in quinoa is both qualitatively andquantitatively controlled, with saponin pro-duction requiring at least one dominant al-lele at the Sp locus and the amount ofgrain saponin being determined by an un-known number of additional quantitativeloci. Introgression of sp1 into day-neutrallines will facilitate the development ofshort-season ‘‘sweet’’ quinoa cultivarswhich do not require postharvest process-ing to remove grain saponin.

Quinoa (Chenopodium quinoa Willd.) is atraditional Andean grain with many usefulattributes. The species is drought andfrost tolerant, will grow in poor soils athigh elevation, and the grain is rich in es-sential amino acids (Koziol 1992). Forthese reasons quinoa is attracting atten-tion as an alternative crop outside its areaof origin, especially in the western UnitedStates, Canada, and northern Europe (Ja-cobsen et al. 1996; Ortiz et al. 1998). A sig-nificant barrier to expanding quinoa pro-duction in these areas, however, is thegrain saponin content of short-season day-neutral varieties which can be grown athigher latitudes. Plant saponins are triter-penoid glucoside compounds found inmany genera; most plant saponins have anintensely bitter flavor and all are poten-tially toxic if ingested in large quantities(Koziol 1992). Mizui et al. (1988; 1990)identified 13 different saponins in quinoabran, 10 of them previously unknown. Qui-noa saponins are concentrated in the ex-ternal layers of the grain, which is botan-ically a fruit with a tightly adheringpericarp covering two seed coat layers(Varriano-Marston and DeFrancisco 1984).The tissue containing saponins is there-fore of maternal origin, and grain saponincontent reflects the genotype of the plantfrom which the grain is harvested. Sapo-nins are traditionally removed from qui-noa by washing the grain before cooking

it, although mechanical dehulling by abra-sion is also effective (Chauhan et al. 1992;Reichert et al. 1986). This method is usedby U.S. quinoa producers, but it increasescosts and is a major deterrent to potentialgrowers who do not have access to theappropriate equipment. The key to devel-oping commercial quinoa cultivars forNorth America and Europe is to combineearly maturity and high yield with a suffi-ciently reduced grain saponin content toeliminate the need for postharvest pro-cessing (Jacobsen et al. 1996; Johnson andWard 1993).

Estimates of grain saponin content indifferent quinoa cultivars range from 0.00to 11.2 mg/g (Koziol 1992; Wahli 1990).Some Ecuadorian and Bolivian cultivarshave extremely low levels of saponin inthe grain, which can be consumed directlywithout washing or milling; however,these low-saponin or ‘‘sweet’’ lines (‘‘qui-noas dulces’’) are late maturing and per-form very poorly at higher latitudes.There has been little investigation of thegenetic basis of quinoa saponin content.Gandarillas (1979) proposed that the traitwas controlled by two alleles at a singlelocus, with ‘‘bitter’’ (high saponin) domi-nant to ‘‘sweet’’ ( low saponin). More re-cently researchers have observed thatgrain saponin content in quinoa is a con-tinuously distributed variable and is there-fore more likely to be polygenically con-trolled and quantitatively inherited(Galwey et al. 1990; Jacobsen et al. 1996).

Gas chromatography and spectropho-tometric techniques have been used tocharacterize and quantify some individualquinoa saponins (Ng et al. 1994). Suchmethods are unsuitable for screening totalgrain saponin content in large numbers ofsamples, however, as they are time con-suming and expensive, and at least 13 dif-ferent saponins must be quantified, not allof which have been fully characterized. Af-rosimetric methods, based on the quantityof foam generated when saponins areshaken in water, cannot quantify the indi-vidual saponins present but do offer theadvantage of providing a rapid and eco-nomical estimate of total grain saponincontent. Koziol (1991) has described astandardized afrosimetric test for estimat-ing grain saponin levels in quinoa. Thismethod has some minor disadvantages: itunderestimates very high levels of grainsaponin, and results may be affected bythe presence of other surfactants. How-ever, the test readily identifies ‘‘sweet’’quinoa lines that contain no detectable sa-ponins, as these produce no measurable


Table 1. Grain saponin content (in mg/g) of parent plants and 20 F1 progeny from each of fourdifferent quinoa crosses

CrossFemaleparent

Maleparent

F1

Progenymean

Standarderror

Progenyrange

Amachuma 3 Tango 6.37 1.65 9.33 0.36 7.16–11.03Calcha 3 Baer 6.52 2.50 6.70 0.19 5.41–8.13Isluga 3 Sajama 6.50 0.00 7.67 0.44 4.18–10.13Amachuma 3 Sayana 6.25 0.00 10.26 0.05 10.13–10.64

foam. Coefficients of variation obtainedfor repeated tests using this method arewithin acceptable limits, and the standard-ized test has proved satisfactory whenscreening for saponin content of individ-ual plants within quinoa populations (Ja-cobsen et al. 1996; Ward 1994).

Genetic analysis of economically impor-tant traits in quinoa has been hamperedby the difficulty of making controlledcrosses. Quinoa is a predominantly self-pollinating allotetraploid with large num-bers of very small (3 mm diameter) per-fect flowers clustered on an inflorescence.This floral structure makes artificial hy-bridization very difficult and hinders theproduction of segregating generationslarge enough for analysis. The researchdescribed here used cytoplasmic malesterile (CMS) quinoa lines as female par-ents in four different crosses, two of whichused low saponin ‘‘sweet’’ quinoa cultivarsas pollen parents. Subsequent F1, F2, andF3 generations were screened for saponincontent using Koziol’s standardized afro-simetric test. The goal of this research wasto characterize genetic control of grainsaponin content in quinoa, especially inlow-saponin lines.


All quinoa plants used in this study weregrown in a greenhouse in Fort Collins, Col-orado, in 1998 and 1999. Five plants weregrown in each 25 cm diameter pot, usingcommercial potting compost supplement-ed by liquid fertilizer. Greenhouse temper-ature was maintained at 258C and plantswere grown under broad-spectrum halo-gen lamps with a 16 h photoperiod main-tained throughout the experiment. Fourindividual crosses were made using CMSquinoa plants as female parents (Table 1).The CMS lines were derived from a Peru-vian accession (PI 510536) from the USDA-ARS Chenopodium collection; this CMSsystem and the associated maintainer andrestorer lines have been described previ-ously (Ward 1998). The pollen parentswere selected from these restorer lines to

generate fully fertile F1 generations. Thefour crosses were made by enclosing amale sterile inflorescence with exsertedstigmas together with a male fertile inflo-rescence at anthesis within a single waxedpaper pollination bag for 5 days; this alsoenabled simultaneous selfing of the pollenparent. Grain saponin content of all parentplants was determined by using Koziol’s(1991) standardized afrosimetric test onthe F1 seed harvested from the female par-ents and on the S1 seed harvested fromthe male parents. Twenty F1 progeny fromeach of the four crosses were grown andself-pollinated by enclosing the inflores-cence in a waxed paper pollination bagthroughout anthesis, and the grain sapo-nin content of the F1 plants was deter-mined by using Koziol’s test on the F2 seedharvested from each F1 plant. Details of thefour initial crosses are given in Table 1.

Seed from a single F1 plant from eachcross was planted to produce an F2 of atleast 160 plants. Three of the four F2 gen-erations segregated for presence or ab-sence of the nuclear restorer allele, and inthese cases all male sterile F2 plants werediscarded at flowering. Each F2 plant wasself-pollinated by enclosing the inflores-cence in a waxed paper pollination bagthroughout anthesis, and F2 grain saponincontents were determined by testing theF3 seed harvested from each F2 plant. F3

families of 40 plants were raised from eachof 10 selected zero-saponin F2 parents, 5of which were derived from cross 3 and 5from cross 4. These F3 plants were self-pol-linated and tested for grain saponin con-tent as already described. Chi-square testsfor goodness-of-fit were performed on thetwo F2 generations segregating for pres-ence or absence of detectable levels ofgrain saponin, using Yates’ continuouscorrection factor (Gomez and Gomez1984).

Results

All of the CMS plants used as female par-ents had high grain saponin levels, whiletwo of the male parent plants had medium

saponin levels and two were effectivelyzero saponin, with grain saponin levels be-low the limits of detection using afrosi-metric methods (Table 1). All F1 plantswere high saponin, although variation insaponin levels among plants within indi-vidual F1 generations was observed inthree of the four crosses. All F2 offspringfrom cross 1 (Amachuma 3 Tango) andcross 2 (Calcha 3 Baer), where a medium-saponin male parent was used, had mea-surable levels of grain saponin, althoughconsiderable variation was observed be-tween plants within F2 families. F2 progenyfrom cross 3 ( Isluga 3 Sajama), where themale parent lacked saponin, segregated148:41 for the presence or absence of sap-onin. This corresponds to a ratio of 3:1 (x2

5 1.02, P 5 .34). F2 progeny from cross 4(Amachuma 3 Sayana), where the maleparent also lacked saponin, segregated139:27. This ratio falls between 3:1 (x2 56.53, P 5 .013) and 15:1 (x2 5 29.97, P ,.01). In both cross 3 and cross 4, variationamong F2 plants that contained measur-able levels of grain saponin was observed.Frequencies of plants containing differentlevels of grain saponin are summarized forall four crosses in Figure 1. None of the 40F3 progeny grown from each of the 10 se-lected zero-saponin F2 plants had detect-able grain saponin levels.

Discussion

Segregation for a continuous range of sap-onin levels was seen in the F2 populationsfrom crosses 1 and 2, where both parentshad either high or medium saponin levels.This is consistent with the hypothesis thatgrain saponin content in quinoa is a quan-titative trait controlled at multiple loci.The high saponin levels in the F1 genera-tions from all four crosses and the skewedfrequency distributions for F2 progenyfrom crosses 1 and 2 also indicate thatdominance effects may be significant con-tributors to the observed phenotypic vari-ance. F1 saponin levels exceeded parentalvalues in all four crosses, which could bea heterotic effect due to overdominance.This suggestion should be treated withcaution, however: as a quantitative trait,grain saponin level may also be affectedby the environment, and it is possible thateven in a greenhouse growing the parentand F1 plants at different times of the year(parents in late summer 1998, F1 plants inwinter 1998–1999) generational differ-ences in grain saponin content may be ob-served. The nature and extent of environ-mental influences on grain saponin in


Figure 1. Frequency distributions of grain saponin content of F2 progeny from four different quinoa crosses.

quinoa and the possibility of heterosis forthe trait clearly need further investigation.The amount of between-plant variation ob-served in three of the F1 generations indi-cates heterozygosity at the relevant mul-tiple loci at one or both of the parentsused in those crosses. This was not un-expected: although quinoa is largely self-pollinating, at least 1–2% outcrossing typ-ically occurs between adjacent plants(Espindola G, personal communication).The male fertile restorer lines used in thisstudy were maintained as individual bulkpopulations, isolated from other lines butwithout bagging individual plants, so com-plete homozygosity of all plants cannot beassumed. As an allotetraploid, quinoa mayalso be subject to fixed heterozygosity atsome loci, even when self-pollinated.

The F2 progeny in crosses 3 and 4,where zero-saponin pollen parents wereused, show frequency distributions forthose plants with measurable quantities ofgrain saponin consistent with quantitativeinheritance and similar to those fromcrosses 1 and 2. A striking difference inthe F2 generations from crosses 3 and 4,however, is the clear segregation for pres-ence or absence of grain saponin, with aratio of 3:1 in cross 3 and a distorted seg-regation falling between 3:1 and 15:1 incross 4. As quinoa is an allotetraploid,multiple genotypes are possible: the vari-ation in F2 progeny ratios between crosses

3 and 4 reflects different genotypes at theSp locus in the CMS female parents, andconsequent differences in genotype be-tween the F1 plants selfed to produce thesegregating F2 generations. The F2 ratiosobserved here are typical of allotetraploidsegregation at a single locus present inboth genomes, where partial pairing be-tween homologous chromosomes produc-es a range of erratic segregations. Similarsegregation ratios have been observed forother single-gene traits in quinoa (Ward2000). This segregation for presence or ab-sence of grain saponin is consistent withthe observation made by Gandarillas(1979) that a cross between ‘‘bitter’’ and‘‘sweet’’ Bolivian quinoas produced a high-saponin or ‘‘bitter’’ F1 and an F2 generationthat segregated for presence or absence ofgrain saponin (‘‘bitter’’ or ‘‘sweet’’) in a 3:1 ratio. Gandarillas did not report any F3data; however, the 400 F3 plants derivedfrom 10 zero-saponin F2 individuals in thisexperiment were all uniformly zero-sapo-nin, indicating that both they and their F2

parents were homozygous at the relevantlocus.

These results suggest that Sayana andSajama, the Bolivian cultivars used as pol-len parents in crosses 3 and 4, possess arecessive allele that inhibits saponin pro-duction when homozygous. This allelemay be a mutated or null version of a geneencoding either an enzyme critical for sap-

onin synthesis or a regulator of the sapo-nin synthetic pathway. Crossing thesezero-saponin cultivars with saponin-con-taining lines produces F1 plants in whichsaponin synthesis is restored, possibly be-cause at least one functional copy of thekey gene is now present. The between-plant variation in grain saponin content inthree of the four F1 generations, and themore extensive variation seen among F2

plants containing saponin, indicate that inplants with at least one functional copy ofthe key gene, grain saponin levels will de-pend on alleles present at an unknownnumber of additional loci. Variation forthis trait as seen in the F2 distributions(Figure 1) does not indicate simple addi-tive effects, which is not surprising giventhe number of different saponins known tobe present in quinoa. More extensive bio-chemical analysis is needed to determinethe relative contributions of the differenttriterpenoid glucosides in quinoa to thetotal grain saponin content.

Similar combinations of qualitative andquantitative variation for a single traithave been reported elsewhere: in maizesilks, for example, maysin concentration iscontrolled at a major locus p1 coding fora transcription activator, with additionalminor loci only active in the presence of afunctional p1 allele (Byrne et al. 1996). An-other possibility is that zero-saponin qui-noa plants result from the accumulationand fixation of null alleles at the relevantquantitative loci, regardless of any addi-tional major gene effect. However, no zero-saponin plants were identified in a total of263 segregating F2 plants from crosses 1and 2, suggesting that if this does occur itis a relatively rare event.

An unresolved question is whether the‘‘sweet’’ quinoa plants identified in this ex-periment are totally lacking in grain sap-onin or are failing to produce only certaintypes of saponin compounds. As de-scribed earlier, Mizui et al. (1988; 1990)identified more than 13 different saponinsin quinoa bran. Ng et al. (1994) reportedthat using thin-layer chromatography andmass spectrometry they were able to de-tect saponins containing phytolaccagenicacid and oleanolic acid in three quinoalines identified as ‘‘sweet,’’ while ‘‘bitter’’lines also contained the sapogenol heder-agenin. However, the total saponin con-tents reported by these researchers forthe ‘‘sweet’’ lines tested ranged from 1.3to 3 mg/g, levels which would actuallytaste bitter to most humans and which areabove the 1.1 mg/g threshold for ‘‘sweet’’as defined by Koziol (1991). These results


indicate that different types of saponinsare found in varying proportions in differ-ent quinoa varieties, again suggesting thatmultiple loci are involved. As discussedearlier, Koziol’s afrosimetric test cannotdistinguish between different types of sa-ponins, but it does distinguish effectivelybetween quinoa plants containing measur-able levels of some form of saponin andthose which do not. Seed from all theplants classified as zero saponin in this ex-periment failed to generate any measur-able foam when shaken with water, andsuch seed also had no detectable bitter-ness when tasted.

This recessive allele, provisionally des-ignated sp1, offers a number of advantagesto breeding programs developing commer-cial quinoa varieties for North America andEurope. Introduction of this allele into theearly maturing day-neutral lines via con-ventional backcross techniques should bestraightforward and less time consumingthan attempting to manipulate grain sapo-nin content as a quantitative trait. Plantsthat are homozygous for sp1 can be iden-tified easily and inexpensively using afro-simetric methods, will produce uniformlyzero-saponin progeny, and will eliminatethe need for postharvesting processing ofthe grain. A potential disadvantage is thatzero-saponin quinoa lines homozygous forsp1 must be grown in isolation if seed isto be saved, as even low levels of cross-pollination from saponin-containing lineswill result in the loss of the ‘‘sweet’’ traitin subsequent generations. It is also likelythat the presence of saponins in quinoagrain provides protection against bird andinsect attacks, and such damage in thefield may be a greater problem with zero-saponin quinoa lines. These disadvantag-es, however, are offset by eliminating theneed for postharvest grain processing,which will increase the potential for qui-noa as an alternative crop for non-Andeancountries.

From the Department of Soil and Crop Sciences, Colo-rado State University, Fort Collins, CO 80523. This re-search was supported by Colorado State University co-operative extension project no. 0168529.


References

Byrne PF, McMullen MD, Snook ME, Musket TA, TheurJM, Widstrom NW, Wiseman BR, and Coe EH, 1996.Quantitative trait loci and metabolic pathways: geneticcontrol of the concentration of maysin, a corn earwormresistance factor, in maize silks. Proc Natl Acad Sci USA93:8820–8825.

Chauhan GS, Eskin NAM, and Trachuk R, 1992. Nutri-ents and antinutrients in quinoa seed. Cereal Chem 69:85–88.

Galwey NW, Leakey CLA, Price KR, and Fenwick GR,1990. Chemical composition and nutritional character-istics of quinoa (Chenopodium quinoa Willd). Food SciNutr 42:245–261.

Gandarillas H, 1979. Genetica y origen. In: Quinua yKaniwa (Tapia ME, ed). Bogota: Instituto Interameri-cano de Ciencias Agricolas; 45–64.

Gomez KA and Gomez AA, 1984. Statistical proceduresfor agricultural research. New York: John Wiley & Sons.

Jacobsen S-E, Hill J, and Stolen O, 1996. Stability ofquantitative traits in quinoa (Chenopodium quinoa).Theor Appl Genet 93:110–116.

Johnson DL and Ward SM, 1993. Quinoa. In: New crops(Janick J and Simon JE, eds). New York: John Wiley &Sons; 222–227.

Koziol MJ, 1991. Afrosimetric estimation of thresholdsaponin concentration for bitterness in quinoa (Che-nopodium quinoa Willd). J Sci Food Agric 54:211–219.

Koziol MJ, 1992. Chemical composition and nutritionalevaluation of quinoa (Chenopodium quinoa Willd). JFood Comp Anal 5:35–68

Mizui F, Kasai R, Ohtani K, and Tanaka O, 1988. Sapo-nins from brans of quinoa (Chenopodium quinoa Willd)I. Chem Pharm Bull 36:1415–1418.

Mizui F, Kasai R, Ohtani K, and Tanaka O, 1990. Sapo-nins from brans of quinoa (Chenopodium quinoa Willd)II. Chem Pharm Bull 38:375–377.

Ng KG, Price KR, and Fenwick GR, 1994. A TLC methodfor the analysis of quinoa (Chenopodium quinoa) sa-ponins. Food Chem 49:311–315.

Ortiz R, Ruiz-Tapia EN, and Mujica-Sanchez A, 1998.Sampling strategy for a core collection of Peruvian qui-noa germplasm. Theor Appl Genet 96:475–483.

Reichert RD, Tatarynovich JT, and Tyler RT, 1986. Abra-sive dehulling of quinoa (Chenopodium quinoa): effecton saponin content as determined by adapted hemo-lytic assay. Cereal Chem 63:471–475.

Varriano-Marston E and DeFrancisco A, 1984. Ultra-structure of quinoa fruit (Chenopodium quinoa Willd).Food Microstruct 3:165–173.

Wahli C, 1990. Quinua: hacia su cultivo comercial. Qui-to: Latinreco SA.

Ward SM, 1994. Developing improved quinoa varietiesfor Colorado (PhD dissertation). Fort Collins, CO: Col-orado State University.

Ward SM, 1998. A new source of restorable cytoplasmicmale sterility in quinoa. Euphytica 101:157–163.

Ward SM, 2000. Allotetraploid segregation for single-gene morphological characters in quinoa (Chenopodi-um quinoa Willd). Euphytica 116:11–16.

Received March 12, 2000Accepted August 31, 2000

Corresponding Editor: David B. Wagner

Parental Effects in theInheritance of Nonnodulationin Peanut

M. Gallo-Meagher, K. E. Dashiell,and D. W. Gorbet

A nonnodulating line (M4-2) and three nor-mal nodulating lines (UF 487A, PI 262090,and Florunner) of peanut (Arachis hypo-gaea L.) were crossed in full diallel to in-vestigate the inheritance of nodulation.Data from F1, F2, F3, F1BC1, and F2BC1 gen-

erations indicated that three genes controlnodulation at three independent loci, withnodulation being a product of two genesand inhibited by a third gene when it isdominant and the others are homozygousrecessive. A genetic model has been pro-posed that describes the nonnodulatedgenotypes as n1n1n2n2N3N3 or n1n1n2n2N3n3

and all other genotypes as normally no-dulated except n1n1N2n2N3p, which has re-duced nodulation when the n1n2N3 malegamete unites with the n1N2p female ga-mete or when the n1n2n3 male gameteunites with the n1N2N3 female gamete.

Most legumes, when infected by the ap-propriate Rhizobium strains, form rootnodules that are capable of N2 fixation.However, nonnodulation mutants derivedeither spontaneously or through mutagen-esis have been reported from eight spe-cies that normally nodulate. A single re-cessive gene, rj1, controls nonnodulationin a spontaneously derived soybean (Gly-cine max L. Merr.) mutant (Williams andLynch 1954) and two ethyl methanesul-phonate (EMS)-induced soybean mutants(Mathews et al. 1989). However, a thirdEMS-derived soybean mutant is controlledby another single recessive gene differentfrom rj1 (Mathews et al. 1989). In wild pea(Pisum spp.), a homozygous recessivesym-2 gene controls nonnodulation (Holl1975), and in an EMS-induced pea mutantnonnodulation is conditioned by sym-5(Kneen and LaRue 1984). In common bean(Phaseolus vulgaris L.), a single recessivegene, nnd-2, controlled nonnodulation inan EMS-derived mutant (Park and Buttery1994). In nonnodulating mutants of sweet-clover (Melilotus alba Desr.), a single re-cessive gene is responsible for the trait;however, allelism tests demonstrated thatat least five different genes, sym-1, sym-2,sym-3, sym-4, and sym-5, are involved in no-dulation (Miller et al. 1991). In three chick-pea (Cicer spp) mutants derived by g ra-diation, nonnodulation is controlled bydifferent single recessive genes, rn1, rn2,and rn3 (Davis et al. 1986). Later, sponta-neous nonnodulating chickpea mutantswere identified that also were controlledby different single recessive genes (Singhet al. 1992; Singh and Ruplea 1998). Non-nodulation in red clover (Trifolium pra-tense L.) is controlled by a recessive gene,r, which is affected by a cytoplasmic factorand the presence of zygotic and postzy-gotic lethals (Nutman 1949). In a sponta-neous mutant of alfalfa (Medicago sativaL.), nonnodulation is controlled by two te-


Table 1. Cross description, parental genotype, and nodulation classification of field-grown F1 plants

Cross

/ ? GenotypeNo. of plantsNodulated Few Nonnodulated

UF 487A 3 M4-2 N1N1n2n2 N3N3 3 n1n1n2n2N3N3 32 0 1M4-2 3 UF 487A n1n1n2n2 N3N3 3 N1N1n2n2 N3N3 26 0 0PI 262090 3 M4-2 n1n1N2N2n3n3 3 n1n1n2n2 N3N3 1 24 8M4-2 3 PI 262090 n1n1n2n2 N3N3 3 n1n1N2N2n3n3 30 0 0Florunner 3 M4-2 N1N1N2N2n3n3 3 n1n1n2n2 N3N3 51 1 0M4-2 3 Florunner n1n1n2n2 N3N3 3 N1N1N2N2n3n3 31 0 0UF 487A 3 PI 262090 N1N1n2n2 N3N3 3 n1n1N2N2n3n3 17 0 0PI 262090 3 UF 487A n1n1N2N2n3n3 3 N1N1n2n2 N3N3 28 0 0Florunner 3 PI 262090 N1N1N2N2n3n3 3 n1n1N2N2n3n3 25 0 0PI 262090 3 Florunner n1n1N2N2n3n3 3 N1N1N2N2n3n3 18 0 0Florunner 3 UF 487A N1N1N2N2n3n3 3 N1N1n2n2 N3N3 29 0 0UF 487A 3 Florunner N1N1n2n2N3N3 3 N1N1N2N2n3n3 20 0 0

trasomically inherited recessive genes, nn1

and nn2 (Peterson and Barnes 1981).Gorbet and Burton (1979) were the first

to describe a nonnodulating peanut (Ar-achis hypogaea L.), which originally wasidentified in the F3 generation from the hy-bridization of UF 487A, a University ofFlorida breeding line, with PI 262090. Ni-gam et al. (1980) also identified nonnodu-lating peanut plants from the cross PI259747 with NC 17 and NC Ac 2731. Theyreported that two independent duplicategenes control nodulation, with nonnodu-lated plants reported as homozygous re-cessive at both loci (n1n1n2n2) (Nigam etal. 1982). However, Dutta and Reddy(1988) proposed a three-gene modelwhere two genes produce nodulation anda third gene inhibits nodulation when it isdominant and the other genes are homo-zygous recessive (n1n1n2n2N3p). In addition,Essomba et al. (1991), in studies of the in-heritance of stem color and nonnodula-tion in peanut, also reported that nonno-dulation may be inherited by threeindependent genes. However, in that studythe three genes showed additive effects,and nonnodulation was conditioned byany two of the three genes being in a ho-mozygous recessive state. In their model,plants with an n1n1n2n2n3n3 genotypewould be nonnodulating, whereas in theDutta and Reddy (1988) model that geno-type would be nodulated. Therefore the in-heritance of nonnodulation in peanut re-mains unclear.

The objective of this study was to ex-amine the inheritance of nonnodulation inpeanut using the nonnodulating peanutline, M4-2, selected from the original non-nodulating germplasm described earlier(Gorbet and Burton 1979). Our resultssupport the three locus model of Duttaand Reddy (1988). However, crosses withn1N2p female gametes fertilized by n1n2N3

pollen or n1N2N3 female gametes fertilized

by n1n2n3 pollen resulted in reduced ornonnodulation, whereas the reciprocalcrosses produced normal nodulation.


Three normal nodulating lines, UF 487A(University of Florida line), PI 262090 (Ro-bore, Bolivia), and Florunner (cultivar un-related to the above peanuts), and a non-nodulating line, M4-2, derived from thecross UF 487A 3 PI 262090, were used asparents in a diallel cross. F1 plants werebackcrossed to M4-2 or PI 262090. The F1,F2, F3, F1BC1, and F2BC1 generations werefield grown at the University of Florida Ag-ricultural Research and Education Centerat Marianna, Florida. Recommended agro-nomic practices were utilized, includinginoculation of seed at planting with cow-pea-type Rhizobium sp.

Leaf color ratings of individual plantswere taken from a representative leaf fromeach plant just prior to digging. Each plantwas tagged so that foliage color could becompared with nodule characteristics.Plants were dug using a conventional pea-nut digger-inverter, cutting roots 20–25 cmbelow the soil surface. Immediately afterdigging, nodulation of roots of individualplants were rated according to size andnumber as nodulated (equivalent to thenumber and size of nodules found in Flo-runner), few ( less than 50 nodules whosediameters were almost twice the size ofFlorunner nodules), and nonnodulated.Pods were hand picked from individualplants that were to be progeny tested. Inclassifying the nodulation trait in crossessubsequent to the F1 generation, the fewclass was considered as nodulated. Thisclassification scheme is consistent withthose used by others examining nodula-tion in peanut (e.g., Dutta and Reddy1988). Data were analyzed by chi-squaretests for goodness-of-fit to the two-locus

(Nigam et al. 1982) and three-locus mod-els for nodulation (Dutta and Reddy 1988;Essomba et al. 1991).


Field observation indicated three nodulecategories: nonnodulated, few nodules,and normally nodulated. Leaf colors wereyellow-green or dark green. Yellow-greenplants either had few or no nodules, anddark green plants had normal nodulation.Most plants with few nodules also hadlarger nodules than a normally nodulatedplant. Peanut plants with few, very largenodules also were observed by Nambiarand Dart (1980) in populations that weresegregating for nodulating and nonnodu-lating plants.

Reciprocal crosses in the F1 involvingthe nonnodulating mutant, M4-2, showeddifferences in nodulation (Table 1). Thethree crosses with M4-2 as the female par-ent resulted in all progeny being normallynodulated. However, the three crosseswith M4-2 as the male parent exhibited re-duced nodulation in the F1. This effect wasparticularly pronounced in F1 plants gen-erated from PI 262090 3 M4-2. Of the 33 F1

plants rated, 8 were nonnodulated, 24 hadfew, and only 1 was normally nodulated.

F2 data were analyzed by chi-squaretests for goodness-of-fit to the expectedratios for the two-locus (Nigam et al. 1982)and three-locus models for nodulation(Dutta and Reddy 1988; Essomba et al.1991). Only the three-locus model pro-posed by Dutta and Reddy (1988) fit theobserved F2 data (Table 2). The results in-dicate that three gene loci are operating inthese crosses. The genotypes proposedfor the parents are n1n1n2n2N3N3 for M4-2,n1n1N2N2n3n3 for PI 262090, N1N1n2n2N3N3 forUF 487A, and N1N1N2N2n3n3 for Florunner(Table 1). The cross UF 487A (N1N1n2n2N3N3)3 M4-2 (n1n1n2n2N3N3) and the reciprocalcross segregated 3 nodulated:1 nonnodulat-ed. This indicates that the N1 allele is com-pletely dominant to n1 and results in nodu-lation. The cross PI 262090 (n1n1N2N2n3n3) 3M4-2 (n1n1n2n2N3N3) segregated 10 nodulated:6 nonnodulated because half of the plantsthat were heterozygous at the N2 locus(n1n1N2n2N3p) were formed as a result of ei-ther the union of an n1N2N3 female gameteand an n1n2p male gamete or an n1N2n3 fe-male gamete and an n1n2N3 male gametewhich would produce plants that werenonnodulated as a result of the parentaleffect described earlier.

In accordance with this model of inheri-tance, the crosses Florunner (N1N1N2N2n3n3)


Table 2. F2 data analyzed by chi-square test for goodness-of-fit to the proposed model

Cross

/ ?

Number of F2 plants

Nodulateda Nonnodulated df x2 P

UF 487A 3 M4-2 1354 426 1 1.08 (3:1) .30M4-2 3 UF 487A 1440 485 1 0.03 (3:1) .86PI 262090 3 M4-2 1462 855 1 0.36 (10:6) .55M4-2 3 PI 262090 1077 668 1 0.48 (10:6) .49Florunner 3 M4-2 2082 237 1 2.03 (58:6) .15M4-2 3 Florunner 1698 198 1 2.48 (58:6) .12UF 487A 3 PI 262090 1094 124 1 0.97 (58:6) .32PI 262090 3 UF 487A 943 83 1 1.94 (58:6) .16Florunner 3 PI 262090 913 1PI 262090 3 Florunner 744 0Florunner 3 UF 487A 803 0UF 487A 3 Florunner 459 0

a Normally nodulated plus few nodules.

Table 3. F1BC1 data with each cross grouped according to the genotype of the parents and analyzed bychi-square test for goodness-of-fit to the proposed model

Cross

/ 3 (/ 3 ?) and (/ 3 ?) 3 ?

Nodulation

Nodulateda Nonnodulated

Chi-square test

df x2 P

M4-2 3 (UF 487A 3 M4-2) ERb 1 1M4-2 3 (M4-2 3 UF 487A) O 38 33 1 0.35 .55(UF 487A 3 M4-2) 3 M4-2 ER 1 1(M4-2 3 UF 487A) 3 M4-2 O 39 34 1 0.49 .48M4-2 3 (PI 262090 3 M4-2) ER 1 1M4-2 3 (M4-2 3 PI 262090) O 16 12 1 0.57 .45M4-2 3 (Florunner 3 M4-2) ER 3 1(M4-2 3 (M4-2 3 Florunner) O 40 13 1 0.01 .92PI 262090 3 (M4-2 3 PI 262090) ER 3 1PI 262090 3 (PI 262090 3 M4-2) O 8 3 1 0.03 .86(M4-2 3 PI 262090) 3 PI 262090 ER 1 0 0(PI 262090 3 M4-2) 3 PI 262090 O 14 0

a Normally nodulated plus few nodules.b ER 5 expected ratio, O 5 observed frequency.

3 M4-2 (n1n1n2n2N3N3), UF 487A (N1N1n2n2N3N3)3 PI 262090 (n1n1N2N2n3n3), and reciprocalsproduced the same F2 segregation ratiossince they had the same F1 genotypes(N1n1N2n2N3n3), with an expected ratio of58 nodulated:6 nonnodulated. All N1p p p p pand p pN2N2 p p genotypes were normallynodulated. Half of the plants that weren1n1N2n2N3N3 and n1n1N2n2N3n3 would bephenotypically classified as nonnodulatedand the other half as normally nodulated,for the same reason mentioned in the pre-vious cross.

The crosses Florunner (N1N1N2N2n3n3)3 PI 262090 (n1n1N2N2n3n3), Florunner(N1N1N2N2n3n3) 3 UF 487A (N1N1n2n2

N3N3), and their reciprocals produced allnormally nodulated plants since bothparents were homozygous dominant atthe N1 and/or N2 locus. All but one of2920 segregates were nodulated. Thisone easily could have been nonnodulat-ed due to a number of factors includingenvironmental conditions or mutations.The chi-square values for all F2 data hadprobabilities above the 10% level,strongly supporting the three-locusmodel ( Table 2).

The number of F2 families (classified F3

plants) derived from nodulated F2 plantswas determined by calculating the ratio ofeach expected F2 genotype that had nor-mal nodulation and then determining theexpected segregation ratio of the progenyfor each genotype, and the results sup-ported the three-locus model of nodula-tion. For example, from the cross UF 487A3 M4-2 and the reciprocal, one would ex-pect one-third of the nodulated plants toproduce only nodulated progeny and two-thirds to segregate 3 nodulated:1 nonno-dulated. These expected ratios were ob-served (P 5 .49 and .54).

The F1BC1 data also provided strong ev-idence to support the proposed model(Table 3). For example, in the last two en-tries involving PI 262090 and M4-2 (Table3) the genotypes of the parents aren1n1N2N2n3n3 and n1n1N2n2N3n3, when the fe-male gamete was n1N2n3 and the male ga-mete was n1n2N3, the resulting F1BC1 plantwas nonnodulated with a genotype ofn1n1N2n2N3n3. However, when the femalegamete was n1n2N3, and the male wasn1N2n3, the F1BC1 plant was nodulated andhad the same genotype as the previous ex-

ample, n1n1N2n2N3n3. So, the genotypes arethe same, but their nodulation differs rel-ative to inheritance from the maternal orpaternal parent. Similar results were alsoobtained for F2BC1 data.

Our proposed model for inheritance ofnonnodulation in peanut suggests that thephenotype of peanut plants can be differ-ent even though the genotypes are thesame because of a parental effect. Mouliand Patil (1975) reported a similar modeof inheritance for an X-ray-induced folia-ceous stipule mutant in peanut. Theyfound that F1 plants showed normal stip-ules when the mutant was used as the fe-male parent, but when the mutant wasused as the pollen parent all F1 plants hadfoliaceous stipules. Similar to our results,parental influence on expression also wasevident in their F1BC1. A parental effect inpeanut has also been described in a spon-taneously derived shriveled seed mutant(Jakkula et al. 1997b). Jakkula et al.(1997a) found that when the shriveledseed mutant was the male parent, theprogeny segregated in the expected Men-delian fashion, 3 normal:1 shriveled in theF2 derived from a cross with a wild-typepeanut. However, when the mutant wasused as the female parent, only normalseed was produced in F1, F2, and F3 seedprogenies.

In this study, results of our F2 and back-cross data make it appear unlikely that ex-tranuclear inheritance plays a role in non-nodulation. One possible interpretation ofour results may be that the inheritance ofnodulation in peanut is controlled by pa-rental or gametic imprinting (for a reviewsee Matzke and Matzke 1993). In such ge-nomic imprinting, either the maternally orpaternally derived allele is actively ex-pressed, while the other is silent. Imprint-ing involves an epigenetic mechanismsuch as DNA methylation. In the mamma-lian genome, it has been shown that thereare allele-specific regions of differentialmethylation in imprinted genes, and atrend for imprinted genes to be clustered(for a review see Constancia et al. 1998).In our study, it is possible that imprintingof the N2 locus results in reduced expres-sion when inherited through the egg. How-ever, when inherited through the male ga-mete, imprinting of the N2 locus is erasedand this allele becomes fully active. Sincea parental effect has been detected forseveral different characters in peanut, itsupports the hypothesis that peanut mayregulate expression of a number of genesthrough some form of genomic imprinting.


From the Agronomy Department, P.O. Box 110300, Uni-versity of Florida, Gainesville, FL 32611-0300 (Gallo-Meagher), IITA, PMB 5320, Ibadan, Nigeria (Dashiell),and Agricultural Research and Education Center, Uni-versity of Florida, Marianna, FL 32446 (Gorbet). The au-thors thank Drs. D. S. Wofford and R. L. Smith for crit-ical review of the manuscript. This work was approvedfor publication as journal series no. R-07333 by theFlorida Agricultural Experiment Station.


References

Constancia M, Pickard B, Kelsey G, and Reik W, 1998.Imprinting mechanisms. Genome Res 8:881–900.

Davis TM, Foster KW, and Phillips DA, 1986. Inheritanceand expression of three genes controlling root noduleformation in chickpea. Crop Sci 26:719–723.

Dutta M and Reddy LJ, 1988. Further studies on genet-ics of nonnodulation in peanut. Crop Sci 28:60–62.

Essomba NB, Coffelt TA, Branch WD, and Van ScoyocSW, 1991. Inheritance of stem color and non-nodulationin peanut. Peanut Sci 18:126–131.

Gorbet DW and Burton JC, 1979. A non-nodulating pea-nut. Crop Sci 19:727–728.

Holl FB, 1975. Host plant control of the inheritance ofdinitrogen fixation in the Pisum-Rhizobium symbiosis.Euphytica 24:767–770.

Jakkula LR, Knauft DA, and Gorbet DW, 1997a. Inheri-tance of a shriveled seed trait in peanut. J Hered 88:47–51.

Jakkula LR, O’Keefe SF, Knauft DA, and Boote KJ, 1997b.Chemical characterization of a shriveled seed trait inpeanut. Crop Sci 37:1560–1567.

Kneen BE and LaRue TA, 1984. Nodulation resistant mu-tant of Pisum sativum (L.). J Hered 75:238–240.

Mathews A, Carroll BJ, and Gresshoff PM, 1989. A newnonnodulation gene in soybean. J Hered 80:357–360.

Matzke M and Matzke AJM, 1993. Genomic imprintingin plants: parental effects and trans-inactivation phe-nomena. Annu Rev Plant Physiol Plant Mol Biol 44:53–76.

Miller JE, Viands DR, and LaRue TA, 1991. Inheritanceof nonnodulating mutants of sweetclover. Crop Sci 31:948–952.

Mouli C and Patil SH, 1975. Paternal inheritance of X-ray induced foliaceous stipule in the peanut. J Hered66:28–30.

Nambiar PTC and Dart PJ, 1980. Studies on nitrogenfixation by groundnut at ICRISAT. In: Proceedings of theInternational Workshop on Groundnuts, Patancheru,A.P., India, October 13–17, 1980; 110–124.

Nigam SN, Arunachalam V, Gibbons RW, Bandyopa-dhyay A, and Nambiar PTC, 1980. Genetics of non-no-dulation in groundnut Arachis hypogaea L. Oleagineux35:453–455.

Nigam SN, Nambiar PTC, Dwivedi SL, Gibbons RW, andDart PJ, 1982. Genetics of nonnodulation on groundnut(Arachis hypogaea L.). Studies with single and mixedRhizobium strains. Euphytica 31:691–693.

Nutman PS, 1949. Nuclear and cytoplasmic inheritanceof resistance to infection by nodule bacteria in red clo-ver. Hered 3:263–291.

Park SJ and Buttery BR, 1994. Inheritance of nonnodu-lation and ineffective nodulation mutants in commonbean (Phaseolus vulgaris L.). J Hered 85:1–3.

Peterson MA and Barnes DK, 1981. Inheritance of inef-fective nodulation and non-nodulation traits in alfalfa.Crop Sci 21:611–616.

Singh O and Rupela OP, 1998. A new gene that controlsroot nodulation in chickpea. Crop Sci 38:360–362.

Singh O, van Rheenen HA, and Rupela OP, 1992. Inher-

itance of a new nonnodulation gene in chickpea. CropSci 32:41–43.

Williams LF and Lynch DL, 1954. Inheritance of a non-nodulating character in the soybean. Agron J 46:28–29.

Received January 24, 2000Accepted August 31, 2000

Corresponding Editor: Halina T. Knap

Genetic and LinkageAnalysis of Cleistogamy inSoybean

R. Takahashi, H. Kurosaki, S.Yumoto, O. K. Han, and J. Abe

Early maturing cultivars of soybean [Gly-cine max (L.) Merr.] native to the shores ofthe Sea of Okhotsk (Sakhalin and Kuril Is-lands) and eastern Hokkaido (northern Ja-pan) have been used in breeding for chill-ing tolerance. These cultivars have astrong tendency to produce cleistoga-mous flowers throughout their bloomingperiod. This study was conducted to de-termine the genetic basis of cleistogamyin an early maturing cultivar, Karafuto-1,introduced from Sakhalin. Genetic analy-sis was performed using F1 plants, the F2

population, and 50 F3 families producedby crossing between Karafuto-1 and achasmogamous cultivar, Toyosuzu. F1

plants had chasmogamous flowers, indi-cating that chasmogamy was dominant tocleistogamy. Analysis of F2 populationsand F3 families generated segregationdata that was close to a two-gene modelwith epistatic interactions, although a por-tion of the pooled F3 data on the frequencyof chasmogamous segregants from cleis-togamous families significantly deviatedfrom the model. The results suggestedthat a minimum of two genes with epistaticeffects were involved in the genetic controlof cleistogamy. Furthermore, cleistogamywas associated with early flowering in theF2 and F3 populations. A gene for cleistog-amy was linked to one of the recessivegenes responsible for insensitivity to in-candescent long daylength.

Cleistogamy, production of open (chas-mogamous, CH) and closed (cleistoga-mous, CL) floral forms by one species, iswidespread among the angiosperms. Lord(1981) classified cleistogamy into four cat-egories: preanthesis cleistogamy in whichpollination occurs followed by anthesis;pseudocleistogamy in which no morpho-logical differences occur between CL andCH flowers other than a lack of expansion

of petals and anthesis in CL flowers; com-plete cleistogamy in which species pro-duce only CL flowers; and true cleistoga-my in which floral dimorphism resultsfrom divergent developmental pathwaysin a single species or individual. Pseudo-cleistogamy is occasionally induced by en-vironmental factors such as drought andlow temperatures (Uphof 1938).

Cleistogamy has been described inmembers of the genus, Glycine. Newell andHymowitz (1980) and Kenworthy et al.(1989) reported cleistogamy of Glycine ta-bacina, G. tomentella, and G. arenaria.Cleistogamy has also been observed incultivated soybean [Glycine max (L.)Merr.] and its wild relative [G. soja Sieb. &Zucc.]. Soybean usually produces both CHand CL flowers on an individual plant; fer-tilization occurs without opening of petalsin CL flowers. Thus, based on the classifi-cation of Lord (1981), soybean is pseudo-cleistogamous.

The production of CH and CL flowers onindividual plants depends on developmen-tal stages. Miyashita et al. (1999) studiedthe dynamics of CH and CL flowers in fiveG. max cultivars and two G. soja popula-tions. The soybean cultivars produced pri-marily CH flowers at the early stage offlowering, whereas CL flowers were pro-duced almost exclusively at the laterstage. Among these cultivars, the propor-tion of CL flowers produced by an individ-ual plant ranged from 22.6 to 66.5% of thetotal number of flowers.

In contrast, early maturing landraces(maturity group 000–00) native to theshores of the Sea of Okhotsk (Sakhalin andKuril Islands) and eastern Hokkaido(northern Japan) usually produce only CLflowers when cultivated in Hokkaido. Inthese cultivars, fertilization occurs with-out opening of petals, or without any ap-pearance of petals at anthesis. However,they have been observed to produce CHflowers at the early flowering stage duringyears with high temperatures.

The early maturing cleistogamous land-races have been used in breeding for chill-ing tolerance in Japan and Sweden (Holm-berg 1973; Sanbuichi 1979). Soybeans aresensitive to low temperatures at variousstages from germination to maturation(Raper and Kramer 1987). Low tempera-tures at the flowering stage, which is mostsensitive to chilling stress, induce flowerand pod abortion (Holmberg 1973; Humeand Jackson 1981, Takahashi and Asanu-ma 1996) or discolored and cracked seedcoats (Sunada and Ito 1982; Takahashi andAbe 1994, 1999). Minimum temperatures


Table 1. Maximum, minimum, and mean temperatures of June 26–July 25 in 1998 and 1999 and long-term average at Memuro

Year

Temperatures

Jun 26–30 Jul 1–5 Jul 6–10 Jul 11–15 Jul 16–20 Jul 21–25

1998MaximumMinimumMean

23.812.417.8

22.214.017.5

23.313.718.0

20.110.214.1

24.212.217.6

21.916.618.8

1999MaximumMinimumMean

18.911.314.9

18.812.515.1

18.912.114.7

23.116.519.3

23.715.819.3

31.318.725.0

AverageMaximumMinimumMean

21.611.716.3

22.812.117.0

22.012.716.9

22.113.817.5

23.414.418.4

24.416.019.6

required for good flowering and pod setvaried among cultivars (Holmberg 1973;Hume and Jackson 1981). The early ma-turing landraces and some of their descen-dants, such as the cultivar Fiskeby V, havesomewhat lower critical temperatures,and flowering and seed formation weregenerally not interrupted by low temper-atures (Holmberg 1973; Hume and Jackson1981).

Cleistogamy may be advantageous un-der severe environmental conditions be-cause the energetic investment in fertiliza-tion costs (i.e. sepals, petals, pollen, andnectar) of CL flowers appears to be con-siderably lower than CH flowers. Schem-ske (1978) evaluated the energetic costs ofCH and CL flowers in pseudocleistoga-mous plant species, Impatiens pallida, andfound that CH flowers had an energetic in-vestment of more than 100 times higherthan that of CL flowers. Further, CL flow-ers possibly shelter pollen from chillingtemperatures at fertilization because chill-ing temperatures hindered pollen forma-tion, anther dehiscence, and fertilizationin soybean (Goto and Yamamoto 1972).Detailed physiological and genetic studiesare necessary to evaluate possible roles ofcleistogamy in relation to chilling toler-ance. This study was conducted to inves-tigate the genetic basis of cleistogamy inthe early maturing soybean cultivars.


Plant MaterialA CL cultivar, Karafuto-1, was pollinatedby a CH cultivar, Toyosuzu, in 1997 and1998. Toyosuzu is a cultivar (maturitygroup II) developed at the Tokachi Agri-cultural Experiment Station that has graypubescence and a yellow hilum (IIrrtt). To-yosuzu has chasmogamous flowers irre-spective of environmental conditions, ex-cept for the later stage of flowering.Karafuto-1 is a pure line selected from anearly maturing landrace introduced fromSakhalin (maturity group 00); it has brownpubescence and brown hilum (i-ii-irrTT).Karafuto-1 has cleistogamous flowersthroughout flowering time, except foryears with high temperatures. The hybrid-ization of F1 plants was ascertained bybrown pubescence color.

Genetic and Linkage AnalysisTwenty seeds from each parent and 120 F2

seeds were sown in a field at Tokachi Ag-ricultural Experiment Station, Memuro,Hokkaido, Japan (428539N, 1438059E) onMay 20, 1998. Twenty seeds from each par-

ent, 4 F1 seeds, and 20 seeds from each of50 F3 families were sown at the same lo-cation on May 19, 1999. Seeds of parentswere sown in duplicate and thinned afteremergence. Cleistogamy of each plant wasdetermined by assessing flowers that fer-tilized at the first date of anthesis (R1;Fehr et al. 1971), because parental differ-ences in floral forms were evident at anearly period of flowering. The date of an-thesis in CL plants was visually evaluatedby the size of flower buds, and it was con-firmed by comparing sizes of developingpods between CL and CH plants 5 days af-ter anthesis. Floral forms were classifiedas follows; CH included open flowers withfully expanded petals, whereas CL includ-ed flowers that did not open fully, that is,those with slightly expanded petals thatprotruded out of calyxes, or those whosepetals did not come out of calyxes.

Flowering response to incandescentlong daylength ( ILD) was also evaluatedfor parents and 98 F3 families at HokkaidoUniversity, Sapporo, Hokkaido, Japan(438039N, 1418209E) in 1999. Toyosuzu issensitive to ILD and flowering was retard-ed under ILD, whereas Karafuto-1 is insen-sitive to ILD and its flowering was not af-fected by daylength. Twenty seeds foreach parent and 98 F3 families were plant-ed in a field where natural daylength wasextended to 20 h using incandescentlamps. The response to ILD for individualplants was evaluated by growth stages at60 days after planting. Classification ofgrowth stage followed Fehr et al. (1971).


Inheritance of CleistogamyTable 1 shows temperatures from June 26to July 25 in 1998 and 1999 at Memuro. All10 plants of Karafuto-1 and Toyosuzu hadCL and CH flowers at anthesis in both

years. Date of anthesis of the F2 plants in1998 and that of the F3 plants in 1999ranged from July 15 to 25 and July 11 to21, respectively. High temperatures fromJuly 21, 1999, apparently had no affect oncleistogamy.

Due to virus infection, only two F1 plantsand 98 F2 plants grew normally. The two F1

plants had CH flowers, indicating thatchasmogamy was dominant to cleistoga-my (Table 2). Ninety-eight F2 plants wereclassified into 77 CH and 21 CL plants. Theresult agreed with the dominance relation-ship observed in the F1 plants. F2 segre-gation fitted both a ratio of 3:1 under asingle recessive gene model (x2 5 0.67, .4, P , .5) and a ratio of 13:3 under a singlerecessive gene (a) and a dominant gene(B) model, in which the former is epistaticto the latter (x2 5 0.46, .4 , P , .5).

Segregation was further evaluated using10 plants each from 50 F3 families includ-ing 9 F3 families derived from F2 CL plantsand 41 F3 families derived from F2 CHplants. Four of the nine CL families pro-duced only CL plants, whereas the remain-ing five families segregated into CH and CLplants. This result evidently contradictsthe single-gene control model, but itseemed to fit a segregation ratio of 0:2:1for CH-fixed family:segregating family(aaBb):CL-fixed family (aaBB) under thetwo-gene model with epistatic interaction(x2 5 0.50, .7 , P , .8). On the other hand,the 41 families derived from CH F2 plantsproduced 28 CH-fixed families and 13 seg-regating families, with the segregation ra-tio in agreement with the expected ratio of7:6:0 ratio for CH-fixed family (AA–, Aabb,aabb):segregating family (AaBB, AaBb):CL-fixed family (x2 5 3.44, .1 , P , .2).

Taking into account a possible bias re-sulting from the small number of plantstested in each F3 family, segregation of


Table 2. Cleistogamy of Toyosuzu (P1), Karafuto-1 (P2), and their F1, F2, and F3 plants

Generation Year

Number of plants

Total CH CL

P1

P2

F1

F2

F3 (CL)a

(CH)

19981999199819991999199819991999

101010102

9889

406

1010002

7734

376

00

10100

215530

a F3 (CL) and (CH) represent pooled F3 plants derivedfrom CL and CH F2 plants, respectively.

Table 3. Association of cleistogamy with ILDresponse

ILD response of F3 family

Number ofF2 plants

CL CH

Karafuto-1 typeKarafuto-1 and intermediate typeKarafuto-1, intermediate, and

Toyosuzu typeKarafuto-1 and Toyosuzu typeIntermediate typeToyosuzu and intermediate typeToyosuzu type

81

33020

04

15104

2810

Figure 1. Date of anthesis in CL and CH plants of F2 ( left) and F3 population (right). Differences in frequencydistribution were significant at the 1% level by Kolmogorov–Smirnov test in the both populations.

cleistogamy in F3 was evaluated on an in-dividual-plant basis. A total of 495 plantscould be evaluated due to lack of germi-nation in some of the F3 families. The 41CH families produced 376 CH plants and30 CL plants, closely fitting the expectednumber of 367 CH plants (47/52) and 39CL plants (5/52) (x2 5 2.30, .1 , P , .2).The nine CL families, however, produceda surplus of CH segregants that was sig-nificantly different from the expected num-ber of 15 (1/6) based on the model (x2 529.9, P , .01). These results suggestedthat cleistogamy in Karafuto-1 is not con-trolled by a single gene, and a minimumof two genes with epistatic effects may beinvolved in genetic control. However, theeffects of different climatic conditions be-tween the two years or the microclimaticdifferences due to the different floweringdates within the segregating populationswere not factored in the present study.Neither penetrance nor expressivity weretaken into account as factors influencingthe expression of cleistogamy in the seg-regating populations, although parents ex-hibited constant and clear differences inour experiments. A further characteriza-tion of cleistogamy and/or experimentsunder controlled environments may there-fore be necessary to exactly determine thecontributions of genetic and environmen-tal factors on the expression of cleistoga-my in Karafuto-1.

Association of Cleistogamy with ILDInsensitivityToyosuzu is sensitive to ILD and its flow-ering was retarded under ILD, whereasKarafuto-1 is insensitive to ILD and itsflowering was not delayed. At 60 days afterplanting, Karafuto-1 reached R4, whereasToyosuzu still remained vegetative. F3

plants were classified into the followingthree types: Karafuto-1 type, Toyosuzutype, and intermediate type whose growthstage reached R2–R3. Segregation of inter-mediate type in F3 families may indicatethat dominance of genes for ILD sensitivityof Toyosuzu was not complete, unlike E3or E4 (Buzzell 1971; Buzzell and Voldeng1980). Of the 98 F3 families tested, 88 fam-ilies with a minimum of 15 plants eachwere classified into seven classes: (1) fam-ily fixed for Karafuto-1 type, (2) family seg-regating for Karafuto-1 and intermediatetype, (3) family segregating for all of thethree types, (4) family segregating for Kar-afuto-1 and Toyosuzu type, (5) family fixedfor intermediate type, (6) family segregat-ing for Toyosuzu and intermediate type,and (7) family fixed for Toyosuzu type

(Table 3). When all segregating or inter-mediate phenotypes (2–6) are consideredas one phenotypic class, the observednumber of Karafuto-1 class, segregating orintermediate class, and Toyosuzu classwas close to a two-gene (1:14:1) model (x2

5 5.45, .05 , P , .1), suggesting the in-volvement of two recessive genes in ILDinsensitivity of Karafuto-1. As shown in Ta-ble 3, all of the 8 F3 families fixed for ILDinsensitivity were derived from CL F2

plants, while all of the 10 F3 families fixedfor ILD sensitivity were derived from CH F2

plants. Segregation of cleistogamy and ILDinsensitivity was thus significantly associ-ated (x2 5 40.1, P , .005). The resultsstrongly suggested that one of the genesfor ILD insensitivity was closely linkedwith a gene for cleistogamy. CL plantsflowered 3 days earlier than CH plants inboth F2 and F3 populations probably dueto the close linkage (Figure 1).

Seven loci have so far been reported tocontrol time to flowering and maturity insoybean: E1 and E2 (Bernard 1971), E3(Buzzell 1971), E4 (Buzzell and Voldeng1980), E5 (McBlain and Bernard 1987), J(Ray et al. 1995), and an unnamed gene forILD response (e(t)) (Abe et al. 1998). TheE3, E4, and e(t) loci are known to be in-volved in the initiation of flowering underILD. The recessive alleles e3 and e4 or e(t)jointly confer insensitivity to ILD (Abe etal. 1998; Buzzell 1971; Buzzell and Voldeng1980). When combined with e3 and e4, E1markedly retards flowering under ILD ornatural daylength relative to e1 (Cober etal. 1996). Of these genes, E1 was linked toT with a recombination frequency of 3.9%(Weiss 1970). The association between thegenotype at the T locus and ILD insensitiv-ity in the F2 population was not significant(data not shown), suggesting that E1 maynot be a gene responsible for the varietal


differences in ILD insensitivity. Linkageanalysis using DNA markers may be usefulto identify the maturity gene associatedwith cleistogamy.

This study revealed that cleistogamy inan early maturing soybean cultivar wasprimarily genetic and a few major geneswere involved. Near-isogenic lines regard-ing cleistogamy should be developed tohelp clarify the relationship between cleis-togamy and chilling tolerance.

From the Legume Breeding Laboratory, National Agri-culture Research Center, Kannondai 3-1-1, Tsukuba, Ibar-aki, 305-8666 Japan (Takahashi), Tokachi Agricultural Ex-periment Station, Memuro, Hokkaido, Japan (Kurosakiand Yumoto), Faculty of Agriculture, Dankook University,Chonan, Korea (Han), and the Graduate School of Agri-culture, Hokkaido University, Sapporo, Japan (Abe). Wethank Dr. T. Narikawa for useful advice and Dr. JosephG. Dubouzet (JIRCAS) for critical reading of the manu-script. Address correspondence to R. Takahashi at theaddress above or e-mail: [email protected].


References

Abe J, Komatsu K, and Shimamoto Y, 1998. A new genefor insensitivity of flowering to incandescent long day-length ( ILD). Soybean Genet Newslett 25:92.

Bernard RL, 1971. Two genes for time of flowering insoybeans. Crop Sci 11:242–244.

Buzzell RI, 1971. Inheritance of a soybean flowering re-

sponse to fluorescent-daylength conditions. Can J Cy-tol 13:103–106.

Buzzell RI and Voldeng HD, 1980. Inheritance of insen-sitivity to long daylength. Soybean Genet Newslett 7:26–29.

Cober ER, Tanner JM, and Voldeng HD, 1996. Geneticcontrol of photoperiod response in early-maturing,near-isogenic soybean lines. Crop Sci 36:601–605.

Fehr WR, Caviness CE, Burmood DT, and PenningtonJS, 1971. Stage of development descriptions for soy-beans, Glycine max (L.) Merrill. Crop Sci 11:929–931.

Goto K and Yamamoto T, 1972. Studies on cool injuryin bean plants. Part 3. Abnormalities in the reproduc-tive processes relating to pod dropping as affected bycool temperature before anthesis in soybean plants [inJapanese]. Res Bull Hokkaido Natl Agric Exp Stn 100:14–19.

Holmberg SA, 1973. Soybeans for cool temperate cli-mates. Agric Hort Genet 31:1–20.

Hume DJ and Jackson AKH, 1981. Pod formation in soy-beans at low temperatures. Crop Sci 21:933–937.

Kenworthy WJ, Brown AHD, and Thibou GA, 1989. Vari-ation in flowering response to photoperiod in perennialGlycine species. Crop Sci 29:678–682.

Lord EM, 1981. Cleistogamy: a tool for the study of flo-ral morphogenesis, function and evolution. Bot Rev 47:421–449.

McBlain BA and Bernard RL, 1987. A new gene affectingthe time of flowering and maturity in soybean. J Hered78:160–162.

Miyashita K, Matsuda H, Ohara M, Misawa T, and Shi-mamoto Y, 1999. Flowering and fruiting dynamics ofchasmogamous and cleistogamous flowers in wild andcultivated soybeans [in Japanese]. Res Bull Univ FarmHokkaido Univ 31:41–48.

Newell CA and Hymowitz T, 1980. Cytology of Glycinetabacina. J Hered 71:175–178.

Raper CD and Kramer PJ, 1987. Stress physiology. In:Soybeans: improvement, production and uses, 2nd ed(Wilcox JR, ed). Agronomy Monograph 16. Madison,WI: American Society of Agronomy; 588–641.

Ray JD, Hinson K, Manjono JEB, and Maro MF, 1995.Genetic control of long- juvenile trait in soybean. CropSci 35:1001–1006.

Sanbuichi T, 1979. Studies on cool weather tolerance insoybean breeding [in Japanese]. Rep Hokkaido PrefectAgric Exp Stn 28:1–57.

Schemske DW, 1978. Evolution of reproductive charac-teristics in Impatiens (Balsaminaceae): the significanceof cleistogamy and chasmogamy. Ecology 59:596–613.

Sunada K and Ito T, 1982. Soybean grain quality as af-fected by low temperature treatments in plants (colorof hilum, seed coat cracking) [in Japanese]. Report ofthe Hokkaido Branch, Crop Science Society of Japanand Hokkaido Branch, the Japanese Society of Breeding22:34.

Takahashi R and Abe J, 1994. Genetic and linkage anal-ysis of low temperature-induced browning in soybeanseed coats. J Hered 85:447–450.

Takahashi R and Abe J, 1999. Soybean maturity genesassociated with seed coat pigmentation and crackingin response to low temperatures. Crop Sci 39:1657–1662.

Takahashi R and Asanuma S, 1996. Association of Tgene with chilling tolerance in soybean. Crop Sci 36:559–562.

Uphof JCT, 1938. Cleistogamic flowers. Bot Rev 4:21–49.

Weiss MG, 1970. Genetic linkage in soybeans. Linkagegroup I. Crop Sci 10:69–72.

Received June 6, 2000Accepted October 31, 2000

Corresponding Editor: Reid G. Palmer

brief communications - oxford academic

Documents