Original article

Intra- and interpopulational genetic variationin juvenile populations of Quercus robur L

and Quercus petraea Liebl

G Müller-Starck S Herzog HH Hattemer1

1 Institut für Forstgenetik und Forstpflanzenzüchtung der Universität Göttingen,3400 Göttingen-Weende, Germany;

2 Eidgenöss, Forschungsanstalt für Wald, Schnee und Landschaft,Zürcherstrasse 111, 8903 Birmensdorf, Switzerland

Summary — In each of 5 2-year-old populations of Quercus robur and Q petraea (single and multi-population samples), genetic variation was quantified with respect to 13 polymorphic enzyme codinggene loci. Genetic control and inheritance of isoenzymes was verified beforehand by means of anal-yses of full-sib families. The observed average heterozygosities were 21.3% Q robur and 21.9% forQ petraea (conditional heterozygosities of 56.6 and 56.7 respectively). The mean number of allelesper locus is 3.2 for Q robur and 3.1 for Q petraea. The relatively small genetic diversities indicate mi-nor polymorphisms. The genetic distances between pairs of samples indicate remarkable differenc-es between populations. Most of the single population samples could be proven to share a smallerproportion of the entire gene pool than each of the multipopulation samples. There is a strong genet-ic similarity between Q robur and Q petraea in terms of common alleles. It is concluded that, morethan in other species, large genetic variation must be incorporated into oak population in order tomaintain the ability of these species to adapt to heterogeneous environments.

Quercus robur / Q petraea / alloenzymes / heterozygosity / diversity / genetic distahce / gene-tic differentiation

Résumé — Variabilité génétique intra- et interpopulation dans de jeunes populations de Quer-cus robur L et de Quercus petraea Liebl. La variablité générique a été estimée dans 5 popula-tions de Quercus robur et 5 populations de Q petraea à partir de 13 loci polymorphes contrôlantl’expression d’enzymes. L’hérédité mendélienne des isozymes a été au préalable vérifiée par l’étude

* Send all correspondence to address 2.

de ségrégation dans les croisements contrôlés. Les valeurs moyennes de l’hétérozygotie observéesont de 21,3% chez Q robur et 21,9% chez Q petraea. Les mêmes valeurs pour les hétérozygotiesconditionnelles sont de 56,6% et 56,7%. Le nombre moyen d’allèles est de 3,2 pour Q robur et 3,1pour Q petraea. Les diversités génétiques sont relativement peu élevées. Les distances génétiquesentre populations indiquent de très fortes différences entre elles. Les populations prises individuelle-ment partagent une partie plus faible de l’ensemble du pool génétique que les populations regrou-pées entre elles. Les allèles communs indiquent une très forte similarité entre Q robur et Q petraea.En conclusion, il est recommandé de conserver une variabilité génétique élevée dans les chênaiesde manière à maintenir leur aptitude à s’adapter à des milieux hétérogènes.

Quercus robur / Quercus petraea / allozymes / hétérozygotie / diversité / distance génétique /différenciation génétique

INTRODUCTION

Quercus robur L (penduculate oak) andQuercus petraea Liebl (sessile oak) be-

long to the major deciduous tree speciesin Germany. Like Fagus sylvatica L, oaksare carrier tree species of complex forestecosystems which range from the low-

lands to the submountainous or even themoutainous regions. Oaks are extremelylong-lived species with forest rotation cy-cles of 200 or more years. Oaks are ex-

posed to more heterogeneous environ-ments over time than any other

predominant tree species. In the study ofgenetic variation and its implications onthe ability of tree populations to survive incomplex environmental situations, oaks

may function well as model organisms.The objective of the present study was

to proceed in the description of the geneticvariation in oak populations and thus in thecharacterization of the natural variability offorest ecosystems. Data on patterns of ge-netic variation will contribute to a better un-

derstanding of principles of adaptation andsurvival of oaks and are needed as criteriafor the choice of reproductive material, forsilvicultural treatment as well as for declar-ation and conservation of genetic resourc-es.

MATERIALS AND METHODS

Samples

For each species, 5 populations were grownfrom commercially utilized seed lots (see table I)which are commonly used for artificial regenera-tion in Germany. Two categories of commercialreproductive material are involved: 1) mixturesof seed lots which originate from harvest in dif-ferent stands which all together belong to thesame region of provenance (’multipopulationsamples’); and 2) material which originates fromsingle oak stands which cover areas of 50-100ha. All stands are supposed to be predominantlyindigenous. A total of 1605 individuals were

genotyped at the age of 2 yr. For location of theorigin of the studied samples, see Müller-Starckand Ziehe (1991).

Genotyping

Genetic control and inheritance of isoenzymeswas verified beforehand by utilizing full-sib fami-lies and their parents of Q robur and Q petraea(Müller-Strack and Hattemer, 1990). For extrac-tion of bud and leaf tissues, enzymes were sep-arated by starch-gel electrophoresis and isoelec-tric focusing mean of genotyping see Müller-Starck and Ziehe (1991). Enzyme systems withenvironmentally dependent expression of isoen-zymes, such as acid phosphatases or esteras-es, were excluded from further studies. Ten en-

zyme systems were studied (abbreviations andEC No / in brackets): Aminopeptidase (AP,3.4.11.1, leucine-and alanine-AP), diaphorase(DIA, 1.8.1.4), glutamate-oxaloacetate transa-

mininase (GOT , 2.6.1.1 (= aspartate amino-transferase, AAT)), isocitrate dehydrogenase(IDH, 1.1.1.42), malate dehydrogenase (MDH,1.1.1.37), peroxidase (PER, 1.11.1.7), 6-

phosphogluconate dehydrogenase (GPGDH,1.1.1.44), phosphoglucose isomerase (PGI,5.3.1.9), phosphoglucomutase (PGM, 5.4.2.2),shikimate dehydrogenase (SKDH, 1.1.1.25).Genotypes were scored at 13 polymorphic geneloci: AP-A, DIA-A, GOT-B,C, IDH-A, MDH-B,C,PER-B, 6PGD-A,B, PGI-B, PGM-A, SKDH-A.

Measurement of genetic variation

Intrapopulational variation was measured bymeans of the observed and the conditional het-

erozygosities (Ha,Hc), (Gregorius et al, 1986),the number of alleles and of genotypes per lo-cus, and the genic (allelic) diversities (Gregori-us, 1987). Interpopulational variation was quanti-fied by genetic distance (Gregorius, 1974) andpopulation differentiation (Gregorius and Ro-

berds, 1986); for a summary see Müller-Starckand Gregorius (1986).

RESULTS AND DISCUSSION

Intrapopulational variationin Quercus robur and Q petraea

Average degrees of heterozygosity

Results are summarized in tables II and III;The conditional heterozygosities (Hc) aregiven in addition to the observed heterozy-gosities (Ha) because the latter values canbe biased as a consequence of their depen-dency upon the underlying gene frequen-cies. For Ha, the given multilocus mean is

arthmetic; for Hc it is equal to the ratio of thesummed Ha values to the summed maxi-mum attainable heterozygosities.

The Ha values showed substantial varia-tion among the gene loci. Loci reflecting

large heterozygosities were: AP-A, DIA-A,IDH-A, PER-B and PGM-A. Variation

among the samples was particularly indi-cated by the gene loci PER-B and SKDH-A. The mean Ha value for the 2 oak spe-cies were nearly the same: 21.3% for Q ro-bur and 21.9% for Q petraea. The averagevalues of the multipopulation sampleswere slightly smaller than the speciesmean (21.1 % for Q robur) or identical to it.

Deviating trends in the Hc values (seetable III) as compared to the Ha valueswere a consequence of differences in the

gene frequencies among samples and

among loci within samples. For instance,AP-A, PER-A and PGM-A reflected greatHa values but small Hc values. This differ-ence leads the Ha values to appear largebut to be small in reality, if the potential toform heterozygotes is taken into considera-tion. The opposite trend was revealed, forinstance, by the loci GOT-B and 6PGDH-B: large Hc values demonstrated that the

extraordinarily small Ha values could nothave been much larger due the underlyingallele frenquecies (1 frequent and a fewvery rare alleles). Loci AP-A, DIA-A, IDH-A, PER-B and PGM-A have 2 or 3 alleles.The variation with respect to MDH-B andMDH-C was primarily a consequence of

gene frenquency distributions close to fixa-

tion in most of the samples. The multilocusmean values showed little deviation amongthe samples. The overall mean values ofthe species were nearly identical (56.6%and 56.7%, respectively).

Generally, pronounced species-specificeffects were lacking and the samplingmode (multipopulation samples vs locallyseparated ones) did not seem to affect theheterozygosities considerably. The ob-served heterozygosities did not deviatemuch from those reported for other decidu-ous tree species, such as Fagus sylcatica:corresponding values (2-yr old plants from5 multipopulation samples genotyped at 12 gene loci) were; Ha = 22.2% and Hc =

52.4% (Müller-Starck and Ziehe, 1991).

Genetic multiplicity

Genetic multiplicity in terms of averagenumber of alleles or genotypes per locus issummarized in table IV. There is no allelewhich occurs in all samples of one speciesbut not in any sample of the other species.Alleles which are represented in some ofthe samples of only one species are so rare(≈ 1 %) that sample size may account for thenon-representation in the other species.

The largest mean numbers of alleleswere revealed by the gene loci PER-B

(5.3) and AP-A (4.7), the smallest ones byMDH-B (1.5) and MDH-C (1.8). The multi-locus means of the samples from multi-

populations compared to single popula-tions (3.4 vs 2.9 for Q robur, 3.3 vs 2.8 forQ petraea. The overall means of both spe-cies are nearly indentical (3.2 and 3.1 al-

leles locus).The mean number of genotypes per lo-

cus varied more among the samples thanthe gene number did. This finding is not

only a consequence of the sample sizes:the sizes of samples 5,7 and 10 are ex-traordinarily small (between 72 and 96 in-dividuals) (see table I) but these samplesdid not reveal the smallest number of

genotypes per locus. In contrast to the

mean number of alleles per locus, the cor-responding values for genotypes were

smaller in Q robur than in Q petraea, ie,5.1 vs 5.4 genotypes per locus (for Q ro-bur, this is equivalent to 76% of the maxi-mum attainable mean number of geno-types; for Q petraea to 85%). For the timebeing it is suggested that characteristics ofthe reproductive system of the respectiveparental populations may contribute to

these phenomena.

Genic (allelic) diversity

The multilocus values (see table V) werecalculated as harmonic means. The hypo-thetical gametic multilocus diversity (HGMD)

is equal to the number of genetically differ-ent gametic types which hypothetically canbe produced by individuals of each sampleon the basis of their 13-locus genotypes

where vl is the diversity at locus l) (Gregori-us et al, 1986). This measure quantifiesthe potential for creating genetic variationin the next generation and is therefore animportant determinant of the adaptability offorest tree populations.

In most cases, the single locus genic di-versities reflected trends similar to thoseobserved in the genic multiplicities (eg,largest values for PER-B and AP-A). Thiswas not true in cases of deviating distribu-tions of allele frenquencies; for instance,the locus DIA-A reflected on the averagesmaller numbers of alleles but larger genicdiversities than the GOT-C locus (DIA-A:3.4 vs 2.1; GOT-C: 4.0 vs 1.5). The reasonfor this difference is the greater deviationof the allele frequencies from the state ofeveness in the case of GOT-C comparedto that of DIA-A.

The variation among the multilocus

means of the samples was smaller thanthat among the mean number of alleles perlocus (see table IV). In both species, therewas not more deviation between the val-

ues from multipopulation samples andthose from single populations.

The hypothetical gametic multilocus di-versities revealed a large variation amongthe samples. The sample size did not

seem to affect these values substantially,because 2 of the smallest samples (5, 7)show quite large multilocus values. In thecase of Quercus robur, the values of the

single population samples did not deviatemuch from those of the mixed samples (onthe average 136.9 vs 137.4), but are con-

siderably smaller in the case of Quercus

petraea (164.9 vs 193.9).

lnterpopulational variationin Quercus robur and Q petraea

Genetic distances

In table VI, results are given for 2 out of 13 loci: AP-A represents those loci which re-

veal in both species, at least for some pairsof samples, considerably large genic (alle-lic) distances; SKDH-A, showed on the av-erage, large values for Quercus petraea butremarkably small values for Q robur. In thissense, SKDH-A reveals greater deviationsamong the 2 species than any other locus.

From the results of a statistical analysisbetween samples (log likelihood ratio test ofhomogeneity in contingency tables), it can

be stated that genic distance values greaterthan 0.1 will reveal significant deviations inmost cases (at least at a significance levelof 0.05). Genic distances larger than 0.2can be expected to indicate substantial in-

terpopulational genetic variation. For manypairs of samples, these values are exceed-ed, eg, in case of populations 8 and 9 (AP-A). Each of these populations can be dis-criminated easily from the other populationsbut not from each other. The similarity be-tween populations 8 and 9 is a surprisingobservation because these 2 samples origi-nate from single populations and should re-lect more specific genetic information thanmixed samples. As can be seen from thesame table, this trend was not confirmed bySKDH-A. The highly specific monitoring ef-fect of adaptive loci may account for this

phenomenon.

Genetic differentiation

Genetic differentiation was quantified bymeans of the genetic distances between

one sample an the remaining ones whichwere combined in order to form the re-

spective complement population. In this

way genetic variation was measured as awhole and not only in pairs.

Figures 1 and 2 illustrate genetic diffe-rentiation for 5 out of 13 gene loci (for asummary of numerical multilocus values,see Müller-Starck and Ziehe, 1991). Foreach species, the graphs refer to the allelefrequencies. In each graph, the radius ofthe dotted line is equal to the average lev-el of differentiation at that particular locus.The given scale measures the averageproportion of genes by which any samplediffers from the remainder. For each

graph, the radii of the sample-specific sec-tors are equal to the proportion of genesby which one sample differs from the

pooled remainder (the angles of the sec-tors correspond to the sample size). Themore the sector radii approach the center,the more representative of the remainderis the genetic information of this sample,and is equivalent to reduction of sample-

specific information. In the combined pres-entation (see graph ’gene pool’), loci-

specific effects can be compensated.The single locus graph show that the

average level of differentiation varied con-

siderably among the loci and among the

species. For both species, a large level ofdifferentiation was found for instance, forPER-B. Deviations between the 2 specieswere particularly evident in the case ofSKDH-A and to a certain extent also thoseof AP-A and GOT-C. The gene pool graphindicates greater differentiation among thesamples of Quercus petraea than amongthose of Q robur. Within the first species,samples 4 and 5 were differentiated aboveaverage, although the differences amongthe samples were small. In the case ofQuercus petraea, all samples except no 6were differentiated above average. The

samples which originated from seed collec-tions in single populations (no 4 and 5, and8, 9 and 10) tended to be more differentiat-ed than the remaining samples which de-scend from population mixtures. This ob-

servation means that single populationsamples share a smaller proportion of theentire genetic information than the mixedones.

CONCLUSIONS

Enzyme gene marker reveals a substantialgenetic similarity between Quercus roburand Q petraea. None of the genes was

represented exclusively in only one of thespecies. Within each species, heterozy-gosities appear to be smaller than in sever-al other tree species (for comparison of pa-rameters and references, see Müller-

Starck, 1991). Because information aboutheterozygosities in adult stands is still lack-ing, it cannot be excluded that relativelylow heterozygosities might characterize

juvenile Quercus populations but not thesucceeding life stages. The genetic multi-plicity was very large in all samples investi-gated including single-population samples.This finding concurs with results of a re-

cent study in 32 European populations ofQuercus petraea (Kremer et al, 1991). Great intrapopulational variation could indi-cate a successful strategy of adaptationand survival of species which are extreme-ly long-lived and exposed to extraordinarilyheterogeneous environments. Relativelysmall genetic diversities indicate minor

polymorphisms, ie, a constellation with 1

predominant and a few rare alleles. The

genetic distances between pairs of sam-

ples indicate remarkable differences in

many cases. These values tend to be high-er than those of Kremer et al (1991). Theillustrated genetic differentiation reveals

that multipopulation samples tend to sharea larger proportion of the entire gene poolthan each of the single-population sam-ples. Gene loci reveal very different trendsbecause each of the adaptive gene loci

may be subject to different selective forcesand thus monitor genetic similarities or dis-

similarities among populations in a veryspecific way.

The results presented herein suggestthat forest tree breeding and silviculture ofQuercus robur and Q petraea need to takeinto account large genetic multiplicities.Such genetic heterogeneity seems to cor-respond to the tremendous environmentalheterogeneity to which long-lived oak pop-ulations are exposed. Particularly in oak

species, it appears that large genetic varia-tion should be incorporated in productivepopulations in order to maintain the poten-tial of these populations to adapt to and tosurvive in complex environmental situa-tions.

ACKNOWLEDGMENTS

The technical assistance of G Dinkel is greatlyappreciated. This study was financially support-ed by the Commission of the European Commu-nities, GD XII, Brussels, Belgium.

REFERENCES

Anonymous (1972) Erste Verordnung zur

Durchführung des Gesetzes über forstlichesSaat-und Pflanzgut (1 FSaat V). Bundesge-setzblatt I, 92, 1561-1586

Gregorius HR (1974) Genetischer Abstand zwis-chen Populationen. 1. Zur Konzeption dergenetischen Abstandsmessung. Silvae Gen-et 23, 22-27

Gregorius HR (1978) The concept of genetic di-versity and its formal relationship to heterozy-gosity and genetic distance. Math Biosci 41,253-271

Gregorius HR, Roberds JH (1986) Measure-ment of genetical differentiation among sub-populations. Theor Appl Genet 71, 826-834

Gregorius HR, Krauhausen J, Müller-Starck G

(1986) Spatial and temporal genetic differen-tiation among the seed in a stand of Fagussylvatica L. Heredity 57, 255-262

Kremer A, Petit R, Zanetto A, Fougere V, Du-cousso A, Wagner D, Chauvin C (1991) Nu-

clear and organelle gene diversity in Quer-cus robur and Quercus petraea. In: GeneticVariation in European Populations of ForestTrees (Müller-Starck G, Ziehe M, eds)Sauerländer-Verlag, Frankfurt-Am-Main, 141-166

Müller-Starck G (1991) Survey of genetic varia-tion inferred from enzyme gene markers. In:Genetic Variation of Forest Tree Populationsin Europe (Müller-Starck G, Ziehe M, eds)Sauerländer-Verlag, Frankfurt-am-Main, 20-37

Müller-Starck G, Gregorius HR (1986) Monitor-ing genetic variation in forest tree popula-

tions. In: Proc 19th IUFRO World Congr.Ljubljana, Yugoslavia, 1986, div 2, vol II, 589-599

Müller-Starck G, Hattemer HH (1990) Geneticsand Breeding of Oaks. Final Report, MA 1 B-0012-D, CEC, DG XII, Brussels, Belgium,22 pp

Müller-Starck G, Ziehe M (1991) Genetic varia-tion in populations of Fagus sylvatica, Quer-cus petraea, and Q robur in Germany. In: Ge-netic Variation of Forest Tree Populations inEurope (Müller-Starck G, Ziehe M, eds)Sauerländer-Verlag, Frankfurt-am-Main, 125-140