Starting from SeedGenetic Issues in Using Native Grassesfor Restorationby Eric E. Knapp and Kevin J. Rice

Increase the odds for

successful restoration

by obtaining the

best seed.

H ’istorical records of the botanical.composition of grasslands in their

pristine state are often incomplete or lack-ing, yet it is evident that many native spe-cies were at one time much more wide-spread than they are today. Changingland use, including agricultural conver-sion, urbanization, overgrazing, fire sup-pression, and introduction of non-nativeweedy annuals, have all dramatically al-tered the landscape and the flora. In re-cent years, the public has become betterinformed about the decline of our naturalheritage due to human influence, and theimportance of biodiversity is now widelyunderstood. Together with and perhaps inpart due to this increasing public aware-ness has been a mounting interest in res-toration using native species, includinggrasses.

Native grasses are being incorporatedinto the landscape in numerous innova-tive ways. These include seeding into pas-tures to increase forage quality and dura-tion of the grazing season; planting alongroadsides to suppress weeds and reduceerosion and herbicide use; revegetatingafter fire to prevent soil erosion; and re-establishing native grasses in nature pre-serves and parks in efforts to enhance bio-diversity. Native grass species have alsofound a place in the urban landscape as awater-conserving alternative to the tradi-tional lawn.

To meet demand, commerciallygrown native grass seed is becoming in-creasingly available. However, in manycases, availability of plant materials haspreceded the synthesis of knowledge nec-essary for successfully conducting a resto-ration project. We are now faced with tre-mendous opportunities, but also burdened

with the possibility of failure of a planting,as well as the risk of altering existing na-tive grass populations through contami-nation, when so little is known about howbest to proceed.

Few in-depth, species-specific studiesdirectly relevant to genetic issues facingthe restorationist have been conductedwith native grasses. However, inferencesbased on existing studies conducted withother species and on known genetic prin-ciples can improve the chances for successof a restoration planting, and such infor-mation can help avoid making unforeseenmistakes: Many steps in the restorationprocess can make use of genetic principles,including sampling of genetic variation inthe collection of germplasm, determiningwhere to look for germplasm if the speciesis not present on the site to be restored,and following proper methods for increas-ing seed under agronomic conditions. It isimportant to recognize that the tremen-dous variability among grass species incharacteristics such as mating system,number of chromosome sets (ploidylevel), and life-span will, in many cases,require species-specific recommendations.This paper will discuss, in a broader sense,some genetic issues that may be useful toconsider when collecting and utilizing na-tive grasses for restoration.

How to Collect SeedWhen collecting plant material, take careto adequately sample the genetic variationwithin a population. Genetic variation isnecessary for populations to evolve andundergo adaptive change in response tochanging environmental conditions, andmaintaining genetic variation can there-

40 RESTORATION ~ MANAGEMENT NOTES 12:1 Summer 1994

fore be critical for the long-term survivalof a population ( Lacy, 1987). Populationsunable to respond to selective pressuresdue to the lack of genetic variation maybecome extinct. For example, a geneti-cally variable population may containplants that differ in drought tolerance,and those that possess the necessary com-bination of genes may survive and pro-duce seed even in an extreme drought. Incontrast, a population with limited ge-netic variation may contain no plantswith the necessary combination of genes,and the same environmental conditionswould result in loss of the entire popula-tion.

Points to consider when collectingseed, in order to enhance the probabilityof adequately sampling existing geneticvariation, include:

1. Population sizeThe number of plants from which seedshould be collected will vary dependingupon numerous population parameters,including amount of genetic variation,the distribution of this genetic variationamong and within populations, and thebreeding system, but such information isoften lacking for native grasses. It is diffi-cult to quantify actual numbers of plantsnecessary for adequately sampling the ge-netic variation of a population. Modelsfrom which minimum population size es-timates are calculated are subject to over-simplification, as the genetic architectureof a population is generally site-specific,influenced by localized selection, and of-ten too complex to model accurately.However, it is obvious that the moreplants from which seeds are harvested, thegreater the chance that the collection willcontain potentially important genes, andthat these genes will be represented in thesame frequencies as in the original popu-lation. Several hundred to a few thousandplants may be necessary for a low proba-bility of loss of alleles, especially for allelesin low frequency in the population (Nam-koong, 1988), but fewer plants may beadequate if interest is in sampling andmaintaining only the most common ge-netic variation, or if the population to besampled is relatively genetically homoge-neous.

In addition, the more plants are sam-

pled, the smaller the chance that matingwill occur among related plants in subse-quent generations. Mating among rela-tives can cause inbreeding, which mayresult in reduced vigor or "inbreedingdepression" (Falconer, 1988), especiallyin species that are cross-pollinating. Har-vesting the same amount of seed fromeach parent will also reduce the risk ofmating among relatives by preventing thepotential dominance of a few parentalgenotypes. This equalization of the con-tribution from each parent potentially in-creases the amount of genetic variationsampled as well.

2. Relationship between plantsHarvesting seed from plants relatedthrough ancestry can reduce the geneticvariation sampled and may cause inbreed-ing. Because seed and pollen dispersal isoften spatially limited, neighboring plantsmay be close relatives. Sampling along atransect or any other systematic meansthat allows some distance between sam-pled plants is one way to choose plantswith a reduced probability of relatedness.If strong prevailing winds occur at a siteduring pollen and seed dispersal, samplingperpendicular to the wind direction mayreduce chance of relatedness with dis-tance.

3. Sampling local variationMajor differences in the physical environ-ment and correspondingly different selec-tion pressures may exist within a site fromwhich seed is collected. For example,plants growing on south- and north-facingslopes, under trees and in full sun, on ser-pentine soils, clay soils, or mine railingsare all exposed to different selection pres-sures and may therefore contain genecomplexes conferring adaptation to thatspecific micro-environment (Jain andBradshaw, 1966). To capture such poten-tial adaptive variation, harvest seeds fromplants growing in micro-environmentssimilar to those found on the restorationsite.

4.Conscious (and unconscious)selection

Seed collectors will often harvest seedfrom the most robust plants, assuming

that the seed will result in a similarly ro-bust planting. However, such selectionmay decrease the amount of genetic vari-ation sampled. Natural populations gen-erally consist of a mixture of plants vary-ing phenotypically in size, height, and soon, due to the effect of both genotype andenvironment, and the robustness of aplant at one given time may therefore notbe a good measure of its genetic value.Plants growing in poor environments,while not appearing as vigorous, may pos-sess desirable gene combinations. Simi-larly, a robust phenotype may be due togrowth in a favorable environment, andnot due to any genetic superiority. In ad-dition, genotype-by-environment interac-tions are common in plant populations.The most robust plants in one year maynot perform well in another year in whichthe environmental conditions are differ-ent. For example, a particularly &ought-tolerant genotype may look the best in ayear with sub-optimal moisture, but maynot be as vigorous as other less drought-tolerant genotypes in years of plentifulmoisture.

Collecting plants of only a certainphenotype (for example, ignoring plantsof small size) may be a mistake, as a mix-ture of plants differing in size and structurewill likely maximize genetic variation, aswell as habitat value and potential biodiv-ersity.

In most cases, it is best to chooseplants from which to harvest seed in arandom fashion, trying to avoid both con-scious and unconscious selection by pay-ing little attention to plant phenotype.This will enhance the probability of in-cluding genes for survival under varyingconditions, not just the conditions thatcontributed to the phenotype and thevigor of a plant at the time the seed washarvested.

Collection of seed from a populationat one sampling time may result in inad-vertent selection for uniformity in traitssuch as flowering time. Natural popula-tions often contain a mixture of plantsmaturing at different times, with such var-iability likely having adaptive impor-tance. Seed collectors concerned withmaintaining this source of potential ge-netic variation may wish to sample a sitemore than once.

RESTORATION & MANAGEMENT NOTES 12:1 Summer 1994 41

Where to Collect SeedOne decision that may profoundly influ-ence the success of a restoration project isdetermining where best to collect seed.Considerations will differ depending uponwhether or not the species already existson the site to be restored.

1. Species already occurs on siteIf the species already exists on or in closeproximity to the site, the seed of theseplants should ideally be used in the res-toration of that site. Plants that have beenexposed to selection pressures inherent toa particular site would be expected to bebetter adapted than plants from otherenvironments, as native germplasm rep-resents an integrated response to past se-lection events and possesses a "geneticmemory" of historical environmental var-iability. Plants on a site are progeny ofsurvivors of extreme droughts, floods, pro-longed heat spells, freezes, and disease out-breaks that have occurred over the lasthundreds or thousands of years or more.The most likely place to find plants withgene combinations best suited for survivalon a restoration site is therefore on thesite itself. Without good evidence to thecontrary, it is best to assume that localadaptation has occurred, and introductionof non-native germplasm should be re-sisted.

Even if non-native germplasm is in-troduced to a site and performs well, onecannot say that the introduced germplasmis well-suited to the site, unless the yearsin which the germplasm was evaluated en-compasses the entire range of yearly en-vironmental variability possible. In recip-rocal transplant studies, where accessionsfrom different locations are planted andcompared in each other’s environments,local genotypes often outperform non-local genotypes over the long term. Millarand Libby (1989) describe an examplewhere in a test plot of douglas fir (Psue-dotsuga men~esii) trees, initial rapidgrowth of a few non-local accessions ledto their planting on a wide variety of sites.An unusually harsh freeze, but an envi-ronmental extreme that was characteristicof the region, later killed the fast-growingintroduced stock.

Using germplasm from another loca-

tion, in spite of the presence of the samespecies on the site to be restored, risks ge-netic contamination or the introductionof non-adapted genes into the population.This can not only alter the genetic integ-rity of the native population, but couldresult in the loss of locally adapted genes.The potential impact of this genetic con-tamination is not well understood, but themagnitude may be dependent upon sev-eral factors, including the degree of differ-ence of the native and introduced germ-plasm in traits determining fitness, the

If seed collected fromplants growing in adifferent environment isused for restoration,there is a danger thatthe restoration plantingwill be poorly adapted,resulting in failure ofthe planting or in plantswith lower vigor andcompetitive ability thananticipated and desired.

relative numbers of native and introducedplants on the site, and the rate of cross-pollination. If the number of introducedplants is large in comparison to the nativepopulation, native plants may beswamped with pollen or overwhelmed bynumbers of competing plants. This willlower the frequency of "native" oradapted genes in the population, andgenes in low frequency are particularlyvulnerable to loss through genetic drift,caused by random fluctuation in gene fre-quencies. Introduction of non-nativeplants can therefore result in the loss ofadapted genes, despite any fitness advan-

tage the adapted genes may confer. Inter-crossing of the introduced plants with thenative plants may also break up adaptivegene complexes, thereby lowering the fit-ness of the native population.

Genetic contamination may occurbeyond the site of introduction of thenon-native germplasm, due to gene flow,which is a function of seed and pollen dis-persal. Seeds generally fall within closeproximity to the plant, but pollen maytravel much longer distances. Estimates ofcross-pollination rates offer some insightinto the propensity of pollen to disperse.Data on rates of cross-pollination ingrasses is best developed for species of ag-ronomic importance, and such informa-tion is unfortunately not available formost native grass species. Known rates ofcross-pollination indicate considerablevariability among different species, rang-ing from predominantly outcrossing [forexample, ryegrass (Lolium multiflorum)(Polans and Allard, 1989), and corn (Zearnays) (Brown and Allard, 1970)] to pre-dominantly selfing [for example, Festucamicrostachys (Adams and Allard, 1982)].

2. Species does not occur on siteThe lack of existing populations of thespecies to be planted on a site necessitatesthe introduction of germplasm not nativeto the site. In such cases, care must betaken in choosing proper germplasm. Ifseed collected from plants growing in adifferent environment is used for restora-tion, there is a danger that the restorationplanting will be poorly adapted, resultingin failure of the planting or in plants withlower vigor and competitive ability thananticipated and desired.

Many native grass species are quitewidespread, with the same species growingin many different environments andtherefore subject to different selectionpressures. Not surprisingly then, grass ac-cessions from different geographic regionsoften show considerable genetic differen-tiation. Our preliminary work with purpleneedlegrass (Nassella pulchra) indicatesthat populations collected from differentzones in C~difornia differ in alleles presentas well as allele frequency, at several iso-zyme loci. Studies have shown that traitswith possible strong adaptive significance,including flowering time or plant height,

42 RESTORATION & MANAGEMENT NOTES 12:1 Summer !994

can vary dramatically depending on sitecharacteristics such as aridity (Rice andMack, 1991), or temperature (Imam andAllard, 1965). Grass populations fromsites that differ in grazing pressure havebeen shown to differ in their ability towithstand defoliation (Detling andPainter, 1983).

Collecting seed from a populationgrowing in a similar environment as thetarget environment, on both a regionaland local scale, can increase the odds thatwell-adapted germplasm will be selectedfor a restoration planting. Without knowl-edge of the gene combinations that deter-mine adaptation to a certain environ-ment, the best we can do is to collectmaterial from environments with poten-tially similar selection pressures as theplanting site.

A. Regional adaptationIn order to sample plants with potentiallywell-adapted genes, the restorationist maymake an effort to collect seed from withina certain radius of where it will be planted.This is better than nothing, but regions ofenvironmental homogeneity rarely followsuch simple contours. The establishmentof zones based on broad ecological and ge-ographic similarities, between which thetransfer of grass seed is to be avoided, maybe more useful. The U.S. Forest Serviceuses such "seed zones," on which guide-lines for movement of conifer seed for re-planting efforts are based (Kitzmiller,1990). According to these guidelines, re-planting must be from stock native to thesame seed zone. Seed zones are further di-vided into sub-zones that are designed tofollow clinal or ecotypic contours withineach seed zone. The seed zone conceptwas developed in response to observationsof considerable genetic variation withinconifer species across latitudinal, eleva-tional, temperature, or moisture gradients.Before the importance of local adaptationwas understood, plantings were oftenmade without adequately considering thesource of the seed, which in some casesresulted in poor stand vigor, reduced pro-ductivity, or failure of the stand (Millarand Libby, 1989).

Patterns of genetic variation over thelandscape are species-specific. Some spe-cies are relatively genetically homogene-ous over large distances, while others vary

greatly over short distances. Seed zonesand guidelines for seed transfer will there-fore ideally be tailored to each species.Data on the spatial patterning of geneticvariation is helpful for delineating "seedzones," but is not yet available for nativegrasses. Until such data are collected, itmay be useful to create preliminary zones,based on existing regional environmentaldata. Horticultural or geographic zonesare readily available (see, for example,Sunset Publishing Corporation’s WesternGarden Book or Hickman and Roberts,1993), and may be very useful for the res-torationist. Reasonable regional adapta-tion can also be attained by just selectingplant material from sites with similar ele-vation, latitude, climate, and so on as thesite to be restored.

B. Local adaptationLess obvious, but perhaps equally impor-tant, genetic variation may be associatedwith local edaphic gradients, biotic fac-tors, or microclimate. Plants growing inheavy clay soils and those growing insandy well-drained soils are potentially ex-posed to very different soil pathogens andmay consequently carry different diseaseresistance genes. Plants growing on soilswith different chemistry and mineral con-tent may possess variation in metal toler-ance and some populations may thereforebe better suited for restoring sites withdifficult soils, such as mine railings, thanothers. In addition, the plants with whicha grass population is associated may selectfor different genes that influence compet-itive ability. Aspect of a site is anotherpotential strong local selective force, withplants growing on a sunny, southern ex-posure more likely to contain genes fordrought stress than plants growing justover the hill on more shaded northerlyslopes.

These are but some of the numerousexamples of how environment and corre-sponding genetic composition of plantpopulations can vary on a local scale.Gene flow may cause some dilution of lo-cal genetic differences, but plants growingin different local microenviroments stillpotentially contain adapted gene com-plexes. Matching not only regional, butalso local, environment of the collectionsite to the target site is therefore advisa-ble. Seed should be collected from sites

with soil type, soil chemistry, aspect, plantassociations, and so on similar to the siteto be restored.

Another issue for the restorationist toconsider is that the current environmentat a restoration site may be quite differentthan the original condition. Just becausewe know a species grew on a site histori-cally does not mean that the same geno-types, or even the same species, will thriveif re-introduced (see Millar and Libby(1989) for discussion). Top soil loss, min-ing-induced changes in soil chemistry,grazing pressure, and the presence of in-troduced species that compete withgrasses for the same limiting resources arejust some examples of how the environ-ment for establishing native grasses mightnow be different. When seeking plant ma-terial for restoring particularly disturbedand altered sites, don’t overlook germ-plasm growing in similarly "difficult" andnon-pristine environments, such as alongroadcuts or surviving in weed-chokedfields. Should the plants have adapted toharsh soil conditions or for competitiveability among introduced weeds, such ma-terial may in fact contain the only geno-types that can now successfully grow inthe environment to be restored.

Production of SeedIn many cases, the amount of seed col-lected from a site will be insufficient forthe restoration effort. Seed-increase plant-ings done under agronomic conditionshave become commonplace, and can beuseful for production of an adequate sup-ply of seed, but also present the opportu-nity for deleterious genetic shifts. If theenvironment under which the seed is in-creased is different from the eventual tar-get environment, selective pressures maygreatly alter the genetic composition ofthe seed.

The agronomic literature aboundswith examples of genetic shifts caused byseed production environment, includingone in which rye (Secale cereale) seedgrown for years in widely different envi-ronments was no longer adapted to theoriginal environment for which it was in-tended. ’Balbo’ rye was released in Ten-nessee, but was later grown as far west asColorado and as far north as Michigan.After several years, scientists suspected

RESTORATION ~t MANAGEMENT NOTES 12:1 Summer 1994 43

that seed produced in many of these dif-ferent regions was no longer well adaptedto the original Tennessee environment,even though is was still labeled and mar-keted as the same ’Balbo’ variety. Whenseed of these different sources was plantedtogether in an experiment in Tennessee,striking differences were found in growthtype, plant height, and heading date, in-

The outlook for theemerging field ofgrassland restoration isbright, but until more isunderstood, a degree ofcaution should prevail.

dicating that natural selection had indeedaltered the genetic composition of theoriginal population (Hoskinson and Qual-set, 1967).

One might expect genetic shiftswhen a population is grown in a differentenvironment for many generations, butgenetic shifts in a single generation ofseed-increase have been documented aswell. Stanford et al. (1960) found thatseed lots produced by the same clones oflading clover (Trifolium repens) planted atDavis, Calif., Aberdeen, Idaho, and Pros-ser, Wash., differed in genetic composi-tion. Some clones produced more seedand/or more pollen in one environmentthan in other environments, so that rela-tive genetic contribution of a clone to theseed lot was dependent on the environ-ment in which the seed was produced.Bulking an equal amount of seed fromeach parent would have partly reducedthe genetic shift, but still would not haveoffset the environmentally influenced dif-ferences in relative pollen contribution.

Genetic shifts are not always undesir-able, and in fact could be used to improveadaptation of germplasm. One strategymight be to plant a mixture of seed col-lected from different regions in a seed-increase environment simulating the tar-

get environment, and allow naturalselection to take its course, hopefully pro-ducing favorable genetic shifts in the seedproduced.

However, the seed-increase environ-ment will in many cases be quite differentthan the target environment. Agronomicenvironments typically contain an abun-dance of soil moisture, nitrogen, and othernutrients. Such conditions can be favor-able for high seed yields, but are likelyquite different from the environment inwhich the seed is destined to be planted.

As mentioned earlier, genotype-by-environment interactions (whereby cer-tain genotypes perform best in one envi-ronment, while other genotypes performbest in different environments) are com-mon in plant populations. It is extremelyunlikely that one genotype will be bestsuited to all environmental conditions.Genotypes that do best under seed-increase conditions may not be the samegenotypes that do best under conditionsat the restoration site. Genetic shiftscaused by different selective pressures inthe agronomic environment could there-fore potentially result in seed adapted tothat particular agronomic environmentand no longer adapted to the likelyharsher and more variable conditions inthe target environment.

Other selection pressures may prevailin an agronomic environment as well.Whereas natural ecosystems are a com-plex assemblage of niches with varying en-vironmental conditions and thereforevarying selection pressures, agronomicconditions tend to be more uniform, andcan select for uniformity of seed produced.For example, mechanical or single dateharvesting will select for uniformity oftime to maturity, as seeds from phenol-ogically extreme plants (seeds that are im-mature or have already shattered) are notharvested.

Practicality will necessarily limit op-tions for avoiding all possible undesirablegenetic shifts, but among points thatmight be considered are:

1. Environment for seed increaseThe best way to avoid genetic shifts is toincrease seed in an environment as similaras possible to the target site. This is per-haps best accomplished by increasing seed

in close proximity to the target site. Inaddition, the seed-increase planting canbe managed to simulate natural condi-tions. Selective mortality can be a strongforce altering the genetic composition ofa population. Opportunities for selectionand subsequent genetic shifts in the seedproducing environment may be reduced ifagronomic conditions are such that plantmortality is low. Excessive mortality ofgenotypes not well adapted to the seed-increase environment can be avoided byseeding at lower densities to reduce inter-plant competition.

2. Duration of seed increaseThe longer plants are exposed to differentselection pressures, the greater the possi-bility of genetic shifts. Minimizing thenumber of seasons in which seed from aplanting of perennial grass is harvestedwill reduce the risk of genetic shifts. Inaddition, upon replanting a seed-increasefield, always use originally collected seed,not seed from a previous seed increase.Seed from a previous increase has alreadypotentially undergone genetic shifts andanother cycle of increase using this seedwill contribute to the shift.

3. Isolation distancePollen flow among adjacent seed-increasefields may dilute desirable genetic combi-nations. Recommended isolation distancesnecessary to minimize contaminationhave been established for many crop spe-cies, and similar guidelines may be appli-cable to seed-increase fields of nativegrasses. Isolation distance will vary de-pending on the mating system of thespecies. Highly selfing species may notrequire any isolation distance, whereasoutcrossing species may require from 300to 1,600 meters between fields to main-tain reasonable genetic purity (Briggs andKnowles, 1967). Unfortunately, as men-tioned previously, data on the mating sys-tems of many native grass species are notyet available.

Importance of RecordKeepingMeticulously label native grass seed in-tended for use in restoration, giving sitespecifications that indicate where the

44 RESTOI~TION & MaNAGEIVmNT NOTES 12:1 Summer 1994

original seed was collected, as well aswhere it was grown. This is often the bestor only data available on which to basedecisions about the germplasm bestadapted and most likely to grow and thriveon a particular site. Monitoring restora-tion plantings over time and compiling adatabase on the specifics of both successesand failures of plantings from each seedlot can provide much needed informationfor future restoration efforts.

Restoration is an inherently complexissue, as successfully recreating an ecosys-tem is dependent upon at least some un-derstanding of how such a complex assem-blage of genotypes and species functionedtogether in the first place. The suggestionsfor sampling and utilization of native grassgermplasm in restoration outlined in thispaper will undoubtedly seem a bit strin-gent to some. However, the genetic prin-ciples we touch on here are based mainlyon observations from other species and onexisting genetic theory, and research isnecessary to verify the validity of theseprinciples to native grasses. It should alsobe recognized that generalizations such asthose in this broad treatment of the sub-ject may rarely directly apply withoutmodification, as grasses are a tremen-dously variable group, and every speciesand every case will be different. The ul-timate use of the seed should be an im-portant criterion in deciding the impor-tance of consideration of the geneticissues. For instance, restoration of an eco-system in which the intent is to perma-nently establish viable and biologicallyimportant populations that are not onlywell adapted, but also possess the geneticvariation for continued evolution, will un-doubtedly require a more conservative ap-proach to seed collection and increasethan would providing plant material forurban landscapes. Obtaining the best

germplasm for restoration will in manycases be most easily accomplished bymatching the seed collection and increaseprocess to a particular restoration project.

The outlook for the emerging field ofgrassland restoration is bright, but untilmore is understood, a degree of cautionshould prevail. We would do better to beoverly meticulous in establishing a suc-cessful restoration project than to havethat project fail as a result of inadequateprecautions. Consideration of genetic is-sues in the collection and use of germ-plasm can increase the odds for successfulrestoration efforts, which will ultimatelybenefit both the business and the scienceof grassland restoration.

ACKNOWLEDGMENTSMany of the points presented here were madeat a conference entitled "Using our NativeGrasses: A Genetic Perspective" sponsored bythe California Native Gross Association, heldon Nov. 13, 1992, in Sacramento, Calif., andthe input of those participating in related dis-cussions before, during and after that meetingis gratefully acknowledged. In addition, wewould like to thank Doria Gordon, Oren Pol-lak, and Connie Millar for their helpful com-ments on an earlier version of this manuscript.

REFERENCESAdams, W.T., and R.W. Allard. 1982. Mating

system variation in Festuca microstachys.Evolution 36:591-595.

Briggs, F.N., and P.F. Knowles. 1967. Introduc-tion to Plant Breeding. Pg. 383. ReinholdPublishing Corp. New York.

Brown, A.H.D., and R.W. Allard. 1970. Esti-mation of the mating system in open-pol-linated maize populations using isozymepolymorphisms. Genetics 66:133-145.

Detling, J.K., and E.L. Painter. 1983. Defolia-tion responses of western wheatgrass pop-ulations with diverse histories of prairie doggrazing. Oecolo~a 57:65-71.

Falconer, D.S. 1981. Introduction to Quantita.tive Genetics, 2nd ed. Longman, New York.

Hickman, J.C., and W. Roberts. 1993. Horti-cultural information in the Jepson manual.In J.C. Hickman, ed. The Jepson Manual:Higher Plants of California. Univ. of Calif.Press. London.

Hoskinson, P.E., and C.O. Qualset. 1967. Ge-ographic variation in Balbo rye. TennesseeFarm and Home Science Progress Report 62:8-9.

Imam, A.G., and R.W. Allard. 1965. Popula-tion studies in predominantly self-polli-nated species. VI. Genetic variabilitybetween and within natural populations ofwild oats from differing habitats in Califor-nia. Genetics 51:49-62.

Jain, S.K., and A.D. Bradshaw. 1966. Evolu-tionary divergence among adjacent plantpopulations. 1. The evidence and its theo-retical analysis. Heredity 21:407-441.

Kitzmiller, J.H. 1990. Managing genetic diver-sity in a tree improvement program. ForestEcol. and Mgmt. 35:131-149.

Lacy, R.C. 1987. Loss of genetic diversity frommanaged populations: Interacting effects ofdrift, mutation, immigration, selection, andpopulation subdivision. Conserv. Biol.1:143-158.

Millar, C.I. and W.J. Libby. 1989. Disneylandor native ecosystem: Genetics and the res-torationist. Restoration and ManagementNotes 7(1):18-24.

Namkoong, G. 1988. Sampling for germplasmcollection. Hortsci. 23:79-81.

Polans, N.O., and R.W. Allard. 1989. An ex-perimental evaluation of the recovery po-tential of ryegrass populations from geneticstress resulting from restriction of popula-tion size. Evolution 43:1320-1324.

Rice, K.J., and R.N. Mack. 1991. Ecologicalgenetics of Bromus tectorum. III. The de-mography of reciprocally sown populations.Oecologia 88:91-101.

Stanford, E.H., H.M. Laude, and J.A. Enloe.1960. Effect of harvest dates and locationon the genetic composition of the synl gen-eration of Pilgrim ladino clover. Agron. J.52:149-152.

Eric E. Knapp is a post-doctoral researcher andKevin J. Rice is an associate professor in the De-partment of Agronomy and Range Sci., Univ. ofCalifornia-Davis, Davis, CA 95616. Phone(916) 752-8529; FAX (916) 752-4361.

RESTORATION ~ MANAGEMENT NOTES 12:1 Summer 1994 45