Variation in Sex Expression in Canada Yew (Taxus Canadensis) Author(S): Taber D

Home , Strobilus, Taxus, Taxus canadensis

Variation in Sex Expression in Canada Yew (Taxus canadensis) Author(s): Taber D. Allison Source: American Journal of Botany, Vol. 78, No. 4 (Apr., 1991), pp. 569-578 Published by: Botanical Society of America Stable URL: http://www.jstor.org/stable/2445266 . Accessed: 23/08/2011 15:56

Your use of the JSTOR archive indicates your acceptance of the Terms & Conditions of Use, available at . http://www.jstor.org/page/info/about/policies/terms.jsp

JSTOR is a not-for-profit service that helps scholars, researchers, and students discover, use, and build upon a wide range of content in a trusted digital archive. We use information technology and tools to increase productivity and facilitate new forms of scholarship. For more information about JSTOR, please contact [email protected].

Botanical Society of America is collaborating with JSTOR to digitize, preserve and extend access to American Journal of Botany.

http://www.jstor.org AmericanJournal of Botany 78(4): 569-578. 1991.

VARIATION IN SEX EXPRESSION IN CANADA YEW (TAXUS CANADENSIS)1

TABER D. ALLISON2 JamesFord Bell Museumof Natural History and Departmentof Ecology and BehavioralBiology, Universityof Minnesota, Minneapolis, Minnesota 55455

Sex expressionwas measuredin severalCanada yew (Taxus canadensisMarsh.) populations of theApostle Islands of Wisconsinand southeasternMinnesota to determinethe extent of variationwithin and among populations. Sex expression was recorded qualitatively (monoecious, male,or female) and quantitatively (by male to female strobilus ratios or standardized phenotypic gender).No discernibletrends in differencesin sex expressionamong populations or habitats wererecorded. Trends in sexexpression of individuals within populations were complex. Small yewstended to be maleor, if monoecious, had female-biasedstrobilus ratios. Large yews were monoeciousbut had male-biased strobilus ratios. Phenotypic gender, recorded as relativemale- ness,however, was negatively,but weakly, correlated with plant size. Genderdistribution in fourof five populations was bimodal, suggesting that cosexual populations consist of male and femalemorphs. Strobilus ratios of individuals in ApostleIsland populations showed significant annualvariation, but genderfor these same plantswas significantlycorrelated from year to year.Annual adjustments in genderwere most pronounced in smallyews. The resultsindicate thatrelative investment in maleand femalereproductive structures by Canada yew individuals is responsiveto environmentalvariation, but sex expressionalso has a proximategenetic component.

Severalstudies have shownthat plants of McKone and Tonkyn,1986) in responseto cosexualspecies (containing individuals pro- environmentalvariation or changesin plant ducingboth pollen and ovules)vary widely in size orvigor. Such variations in genderamong theirrelative male and femalereproductive individualsare notnecessarily genetic. Alter- effort.Lloyd (1980) developedthe concept of natively,bimodal, or dimorphic,gender dis- standardizedphenotypic gender to providea tributionsalso have been recordedin angio- quantitativemeasure of this variation.Ac- spermswhere a cosexual plant population cordingto this measure, a plant'sgender ranges consistsof male and femalemorphs (Lloyd, from0.0 (female)to 1.0 (male)and is thecon- 1980).Female morphs, for example, may pro- tributionof male function (pollen) and female duceboth pollen and seeds,but they comprise function(ovules and seeds)to a plant'sfitness a discretegroup of the population that repro- relativeto othermembers of the population. ducesprimarily through seeds. The variationin genderamong individuals Dimorphicplant species are of special in- withina population,or genderdistribution, terest because they represent extremes in gen- suggestsmechanisms by which sex expression derspecialization. Lloyd (1980), forexample, in a plantpopulation is determined.For ex- has suggestedthat dimorphic or discontinuous ample, in manyplant speciesgender distri- genderdistributions indicate genetic differen- butionis unimodalor monomorphic;func- tiationwithin a populationreflecting incipient tionalgender may vary widely and continuously or stalledevolution of dioecy.Studying di- amongindividuals (Primack and Lloyd,1980; morphicplant species shouldtherefore illu- minate conditionsby which dioecy might 1 Receivedforpublication5 July 1990; revisionaccepted evolvefrom monoecy or vice versa. 28 December1990. To date,dimorphic gender has notbeen re- The authorthanks D. Thiede,S. Householder,and D. forany cosexual gymnosperm species, Blocksteinfor assistance in thefield; and P. Abrams,M. corded Davis, M. McKone,P. Morrow,P. Regal,A. Snow,and butthis could reflect the lack of application of anonymousreviewers for comments on themanuscript. Lloyd'sapproach to this taxonomic group. The Fundingwas providedby a CarolynCrosby Fellowship ratioof ovulate and staminatestrobili (referred fromthe Graduate School of the University of Minnesota, to hereafteras femaleand male strobili),how- The Dayton-WilkieNatural History Fund, the Minnesota ever, does vary widelywithin gymnosperm ZoologicalSociety, and SigmaXI. Denti, and 2 Currentaddress: Department of PlantBiology, The species (Sarvas, 1968; Schoen, OhioState University-Marion Campus, 1465 Mt. Vernon Stewart,1986), and these ratios vary according Avenue,Marion, OH 43302. to environmentaldifferences (Freeman et al., 569 570 AMERICAN JOURNAL OF BOTANY [Vol. 78

1981)and in responseto changesin plantsize temscan be tracedunder the litter layer, but and vigor (Matthews,1963; Smith, 1981; eventuallythese connections rot. In low-den- Whithamand Mopper,1985). In thispaper I sitypopulations, individual genets consisting describeannual and geographicvariation in of one or more branchesare readilydistin- sex expressionin Canada yew (Taxus cana- guished.In high-densitypopulations the ge- densisMarsh.), a shrubby,monoecious gym- neticrelatedness ofneighboring plants, or more nosperm.I recordedsex expressionin Canada appropriately,shoot system networks, is dif- yewqualitatively (e.g., male, female, and mon- ficultto determinedue to thislayering habit. oecious)as wellas quantitativelyby standard- In samplingreproductive effort in high-density ized phenotypicgender (Lloyd, 1980) to assess populations,I traced belowground connections the relativeusefulness of thesemeasures in to the rottedend. The sex expressionof an understandingsex expressionin Canada yew. individualgenet was estimatedby combining Speciesin the genus Taxus are principally the resultsfrom connected shoot systemsof dioecious,and the seeds are bird-dispersedthe plant. In high-densitypopulations, there- (Chamberlain,1966). Givnish(1980) hypoth- fore,a plantconstituted an unknownpropor- esizedthat dioecy in gymnospermswas linked tionof a genet. withbird-dispersal of seeds. Canada yew, how- Canada yewis a preferredwinter browse of ever,is monoecious,or cosexual,and is an white-taileddeer (Odocoileus virginianus Ra- exceptionto Givnish'shypothesis. Because of finesque)(e.g., Beals, Cottam, and Vogl, 1960). itsunique position in thegenus Taxus, it is of All dataon sexexpression in Canadayew were interestto examinevariation in sexexpression collectedfrom populations where current deer amongCanada yew individuals in orderto bet- browsingwas minimalor nonexistent.Brows- terunderstand the evolutionary and ecological ingby deersignificantly modifies sex expres- significanceof its matingsystem. sion in Canada yew(Allison, 1987). In addition,the studyof sex expressionin Canada yewis relevantto thedebate concern- ing the importanceof environmentalvs. ge- MATERIALS AND METHODS netic impactson gender.Taxus is cited as alteringsex expression in responseto environ- Reproductiveeffort in Canada yewwas es- mentalchanges (Freeman, Harper, and Char- timatedby countingthe number of male and nov, 1980). Carefulquantitative observations femalestrobili and seeds producedby indi- of the sex expressionof individualsover ex- vidual plants. This approach assumes that tendedperiods of timeis requiredto resolve variationamong individuals in thenumber of thisissue. pollengrains per strobilus or the mass of seeds Canada yewis a monoecious,or cosexual, is lowrelative to themean of these parameters evergreenshrub of the mixed conifer-hard-(e.g., Stanton,1984; Thompson,1984; Mc- woodforests of northeastern United States and Kone andTonkyn, 1986; McKone, 1989). The southeasternCanada (Martell, 1974). Male and size of male strobiliand seeds variedsignifi- femalestrobili are initiated during the summer cantlyamong individuals within populations, and typicallyreside singlyin leaf axils of butnot among the different populations sam- branchesproduced that year. The male stro- pled (Allison,1987). The rangeof variation bilusconsists of 5-14 peltatemicrosporophylls was small(C.V. = 13.3%for male strobiliand attachedto a centralaxis. Each microsporo- 11.1% forseed masses),however, indicating phyllcontains two to tenmicrosporangia (Du- thatcounts of male strobili,female strobili, pler,1 919). The femalestrobilus is uniovulate; and seedswere reasonable estimates for com- theovule is borneon a secondaryshoot of a parisonsofreproductive investment among in- shortprimary shoot (Dupler, 1920). After fer- dividualyews and yewpopulations. tilizationthe ovule develops into a stonyseed I sampledyew reproductive effort primarily thatis surroundedby a red, fleshy,aril-like at the Apostle Islands National Lakeshore, structure.Seed ripening in yew populations be- Wisconsin,USA (46?50'N latitude,90045'W ginsin lateJuly or earlyAugust and continues longitude;hereafter referred to as Islands).The for6 to 8 weeksinto early fall. Islands comprisea 21-islandarchipelago lo- Canada yewindividuals reproduce vegeta- catedon the southwesternshore of Lake Su- tivelyby layering when their arching branches perior.The Islands are edaphically similar; soils arepressed to the ground surface and take root. have developedfrom lacustrine deposits and Side branches,growing in differentdirections glacialtill derivedfrom Lake Superiorsand- thanthat of the main stem axis, cause the plant stone(Brander, 1983). As partof a largerstudy to spreadalong the forestfloor. Connections ofsexual reproduction in Canadayew (Allison, betweenthese rooted branches or shoot sys- 1987)1 chosethree different island populations April 1991] ALLISON -CANADA YEW SEX EXPRESSION 571 forstudy: Rocky Island, Otter Island, and Out- wheremi equals the numberof male strobili er Island. producedby planti, fiequals the numberof On eachisland, I systematicallylocated tran- femalestrobili or seeds producedby planti, sectsand recordedreproductive effort in yew and M and F equal thetotal number of male plants(defined above) locatedat 10-minter- strobiliand femalestrobili or seeds,respec- vals alongthese transects. On RockyIsland, tively,produced by thewhole population. As yewswere widely spaced (nearest neighbor dis- calculatedhere, gender is a measureof the tance= 3.13 m; Allison,1990). Consequently, malenessof a plantand rangesfrom 0.0 (fe- I was able to sampleentire genets on Rocky male) to 1.0 (male). Island. Populationson the othertwo islands I useda teststatistic proposed by Engleman weredense. Beginning in thespring of 1982 at and Hartigan(1969) (cited in McLaughlin, theIsland populationsI countedthe number 1989) to testwhether observed gender distri- ofmale and femalestrobili produced by each butionsfor all populationswere unimodal, rep- plant.Seed productionwas recordedfor each resentingmonomorphic gender. Gender for plantin late summer.Each plantwas resam- thesepopulations was sortedfrom lowest to pled in 1983, 1984,and 1985,and additional highestand thendivided up intotwo clusters plantswere sampled on each island.A maxi- N - 1 timeswhere N is thenumber of plants mumof 98 plantswas sampledon RockyIs- in thepopulation. For example: land,57 on OtterIsland, and 92 on OuterIs- land.Yew plantswere also sampledon Rocky (X1)(X2 . . . XN) Islandin 1986.In 1984 I estimatedthe size of . . . eachplant by measuring shoot length and basal (X1, X2)(X3 XN) diameter.The latterwas measuredat a point immediatelybelow the lowest living branch of (X1 . . . XN-1)(XN) eachshoot system since this criterion could be For each clusterpair, I calculatedB/W which appliedconsistently to all plants. is theratio of between- to within-clustersums Geographicvariation in yewsex expression of squares: was estimatedby additional sampling with the pop- same methodsin two high-densityyew B/W= n,n2(x -X2)2 ulationslocated in southeasternMinnesota: [s,2(n -1) + s22(n2 - 1)](n, + n2) NorthGrey Cloud Island (44?46'N latitude, 92?56'Wlongitude) (sampled in 1984and 1985) wheren, and n2 are the samplesizes forthe and Marion,Minnesota, along the Root River twoclusters, xl and x2are thesample means, (43?55'N latitude;92?22'W longitude)(sam- and S12 and s2 are the samplevariances. To pled in 1983 and 1984). testthe null hypothesisof a unimodaldistri- Allcensuses in the different populations were bution(gender is monomorphic),the maxi- made on plantsgrowing in the forestunder- mum B/W is used. When N > 8, ln(B/Wmax+ storyto minimizedifferences in reproductive 1) is normallydistributed with a mean of effortresulting from gross differences in light [- ln(l -2/r) + 2.4/(N-2)] and a variance of and mineralresources. To estimatethe influ- 1/(N-2) (McLaughlin,1989). The probability ence of resourceson sex expressionon Outer of obtaininga maximumvalue of B/W by Island,I randomlyselected plants within large chancecan be estimatedby scalingln(B/W + gapscreated by beaver cutting. The samemea- 1) (Snedecorand Cochran,1967) and com- surementswere made on plantsat thissite as paringit to tablesfor the standard normal dis- at othersites. I hereafterrefer to thissite as tribution.If B/Wmaxhas a low probabilityof OuterIs. (gap);yew plants sampled in theun- occurrence(P < 0.05), thissuggests that the derstoryof Outer Island are referred to collec- populationconsists oftwo discrete gender clus- tivelyas OuterIs. (closed). ters,or male and femalegender groups, i.e., All datawere analyzed by one-way ANOVA genderis bimodal. withsite as themain effect. Analyses were based Kendall'stau (Conover,1980) was calculat- onlog10(Y + l)-transformedvalues to equalize of individualyews ed to measureyear-to-year gender correlation, variances.Strobilus ratios or concordance,for individual yews. Pheno- were transformedbefore ANOVA by log10- is an [(malestrobili + 1)/(femalestrobili + 1)]. typicgender, as definedby Lloyd (1980), I calculatedstandardized phenotypic gender estimateof the relative genetic contribution of forplants in each population(Lloyd, 1980; a cosexualplant from male and femalefunc- Lloydand Bawa, 1984).Gender was calculated tion.Consequently, a plant's gender is depen- as denton thesex expressionof othermembers of the population.Change in the relativeal- mi/M locationto pollenand seedsby one plant,for (fi/F+ mi/M) example,can change the gender ofanother plant 572 AMERICAN JOURNAL OF BOTANY [Vol. 78

TABLE 1. Means ? 1 SE ofyew plants for Canada yewreproductive and vegetativeparameters by site. Means within a columnhaving different letters differ significantly at P < 0.05 usingTukey's Honest-Significant Difference. All meansare on a perbranch basis and werelog,0(Y + 1)-transformedprior to ANO VA. Male andfemale strobilus production,seed production, and basal diameterare presentedas arithmeticmeans. Male/female strobilus ratios arepresented as back-transformedmeans

Male/female Site Na Male strobili Female strobili Seeds strobilusratio Nb Basal diameter(mm) 1983 Root River 21 296.3 ? 68.4a 22.0 ? 7.6a 5.5 ? 2.8a 11.9 ? 1.4a OuterIsland (closed) 50 87.3 ? 23.3b 5.6 ? 1.0a 2.5 ? 0.5a 10.5 ? 1.2a OtterIsland 55 16.9 ? 4.0c 4.8 ? 2.2b 1.1 ? 0.3 3.2 ? 1.2b RockyIsland 54 10.7 ? 5.3d 0.8 ? 0.4c 0.2 ? 0.1 4.6 ? 1.3b 1984 RootRiver 26 296.9 ? 94.3ab 31.9 ? 7.9a 8.0 ? 2.8a 5.6 ? 1.3ab 32 11.4 ? 0.7b GreyCloud 28 156.6 ? 33.Oa 27.1 ? 5.lab 7.0 ? 2.1a 10.7 ? 1.3a 42 9.9 ? 0.4bc OuterIsland (gap) 14 560.6 ? 113.6a 58.5 ? 17.4a 23.9 ? 8.1 15.5 ? 1.6a 21 17.0 ? 0.8a OuterIsland (closed) 50 113.6 ? 30.8b 9.2 ? 2.Ob 3.6 ? 0.8ab 7.9 ? 1.2a 99 9.2 ? 0.4c OtterIsland 54 22.9 ? 5.4c 8.5 ? 1.2b 2.0 ? 0.5b 2.6 ? 1.2b 77 6.3 ? 0.3d RockyIsland 54 10.7 ? 5.3d 1.7 ? 0.4c 0.6 ? 0.lc 7.8 ? 1.2a 54 6.2 ? 0.4d 1985 GreyCloud 31 146.8 ? 28.2ab 19.0 ? 4.2b 6.0 ? 1.3ab 8.5 ? 1.2a OuterIsland (gap) 25 289.8 ? 50.7a 62.1 ? 10.7a 13.2 ? 2.4a 4.5 ? 1.2ab OuterIsland (closed) 92 77.2 ? 15.1b 9.9 ? 2.4bc 3.9 ? 1.0b 7.1 ? 1.2a OtterIsland 57 10.7 ? 2.2c 5.3 ? 1.0c 2.9 ? 0.7b 1.8 ? 1.2b RockyIsland 98 12.3 ? 5.lc 0.8 ? 0.3d 0.2 ? 0.1c 6.2 ? 1.2a a Numberof plants sampled. b Numberof branches sampled. even thoughthe latter has not alteredits ab- icantly(repeated measures ANOVA; P < 0.001 soluteinvestment in male and female function. forall comparisons).Finally, there was a sig- To estimatethe extentof genderchange in nificant,positive correlationbetween male ApostleIsland yews,I arbitrarilydesignated strobilusproduction and seed productionin plantsthat showed a gendershift of >0.4 gen- fiveof six populations,but the strength of the der units(on a scale of 0 to 1) as showing correlationvaried widely (r2 = 0.14 to 0.86). "flexible"gender. By using a shiftof this mag- Canada yewplants (as definedin Materials nitude,I ignoredthe minor changes in gender and Methods)are eithermale, female,mon- ofplants that resulted primarily from changes oecious,or nonflowering(Table 2). In 1983 in sexallocation of other members of the pop- and 1984the proportion of these types did not ulation. differsignificantly among five of thesix pop- ulationssampled (Table 2). The vastmajority RESULTS of individualssampled in thesepopulations weremonoecious. Only 3% ofall plantssam- Geographicvariation in sex expression -Male pledwere female. Outer and OtterIsland pop- and femalestrobilus production, seed produc- ulationshad significantlymore males in 1985 tion,and plantsize variedsignificantly among thanNorth Grey Cloud and OuterIs. (gap), populations(Table 1). ANCOVA withbasal but monoeciousplants were still in the ma- diameteras thecovariate still showed signifi- jority(72% ofreproducingindividuals) in these cant siteeffects (P < 0.001) on strobilusand twopopulations. A higherproportion of male seedproduction. There is lessvariation among yewsoccurred on RockyIsland, but this island populationsin strobilusratios; only -Otter Is- was notincluded in thestatistical analysis be- land in all 3 yearsand RockyIsland in 1983 cause RockyIsland yews included whole gen- had significantlylower ratios of male to female ets. Plantsin the otherfour populations rep- strobilithan the otherpopulations (Table 1). resentedan unknownproportion of a genet. Year-to-yearproduction of strobiliand stro- Mean basal diametersof the different breed- bilusratios of individualplants varied signif- ingtypes were significantly different by ANO- April 1991] ALLISON -CANADA YEW SEX EXPRESSION 573

TABLE 2. Variationin qualitativesex expressionwithin _ monoecious and among differentCanada yewpopulations. Rocky 13 24 51 48 44 70 El single sex Islandyews were not included in the statistical analysis 1040 E non-flowering forreasons as describedin theResults section. G-sta- tisticstest for homogeneityof distributionof repro- ducingindividuals among populations and weread- 80 justedforWilliamson's correction

Monoe- Non- 60 Site Males Females cious flowering c 1983

Root River 2 1 17 1 20 OuterIs. (closed) 8 0 37 5 OtterIsland 11 1 36 8 RockyIsland 9 2 15 28 20 Gadi. = 3.35; P > 0.50 1984 1 2 3 4 5 6 Root River 2 1 23 0 GreyCloud 4 0 24 0 1 22 31 30 5 9 OuterIs. (gap) 3 0 11 0 100- OuterIs. (closed) 10 1 39 0 OtterIsland 8 3 37 7 RockyIsland 17 0 17 20 80- Gadj = 2.19; P > 0.50 1985 -60- GreyCloud 1 0 30 0 OuterIs. (gap) 1 0 24 0 CL 40- OuterIs. (closed) 20 1 65 6 OtterIsland 13 4 33 7 RockyIsland 28 2 21 47 20- Gadj = 19.04; P < 0.005 0 1 2 3 4 5 6 VA (P < 0.001) forboth Otter and OuterIs- landsand RockyIsland. Small yews tended to diameterclass be nonreproducing,large yews were generally Fig. 1. Breedmngtypeof shoot systems by diameter size monoecious,and yews ofintermediate size were class,for Apostle Island yews. Diameter classes increase single-sexedand typicallymale (Fig. 1). The in 2-mmincrements beginning with 2 mm:class 2 =2-4 smallestyews on Outerand OtterIslands were mm;class 3 = 4-6 mm;class 4 = 6-8 mm;class 5 = 8- an exceptionto thistrend. 10 mm;and class6 > 10mm. (a) Outer(closed) and Otter Yewsusually produce only male strobili when Islandscombined-multiple shoots per plant (N = 250); and (b) RockyIsland-whole genet (N = 98). Numbers firstreaching reproductive maturity. During 4 above columnsare numbersof plantsin each diameter yearsof censusingon RockyIsland, 20 yews class. reproducedfor the first time; 18 ofthese plants weremale, one plant was female, and oneplant yewpopulations in 1985 are portrayedin Fig. was monoecious. 2; thesepopulations had the largestsample sizesof reproducing individuals. Distributions Gendervariation within populations-The forApostle Island populations were similar in vastmajority (87%) ofyews in all populations showinga preponderanceof male or male-bi- combinedproduced more male strobilithan ased yewsand an additional,but smaller, peak femalestrobili, although there was consider- below0.5. able variationwithin populations. Untrans- Thereis no significantrelationship between formedstrobilus ratios (male/female) of in- genderand thenumber of strobili produced by dividual plants rangedfrom 1.0 to >200. a plant.Gender is negativelycorrelated with Analysisof B/Wma,,indicates that gender dis- plantsize on Rocky Island and Outer Is. (closed) tributionin theRoot River population in 1984 (e.g.,Fig. 3), butnot on OtterIsland. The cor- and in threeof fourCanada yewpopulations relationis weak(r2- 0.03 to 0.19) and appears in 1985was bimodal (Table 3); onlythe gender due largelyto theabundance of smallmales. distributionof GreyCloud did not differsig- Whenmales are excludedthere is a significant nificantlyfrom a unimodaldistribution. Gen- but weak,positive correlation between plant derdistributions for Otter and Outer Is. (closed) size and the ratioof male to femalestrobili 574 AMERICAN JOURNAL OF BOTANY [Vol. 78

TABLE 3. Mean genderand varianceof presumed male and female clusters (or "morphs")for Apostle Islands popu- lationsand GreyCloud in 1985 and RootRiver in 1984. B/Wma_,was calculatedas describedin Materialsand Methodsand teststhe null hypothesis of a unimodaldistribution

Cluster Outer Island OtterIsland Rocky Island Grey Cloud Root River Female x 0.2570 0.2625 0.2112 0.2658 0.1390 s2 0.0246 0.0338 0.0825 0.0556 0.0123 N 40 20 8 12 10 Male x 0.9278 0.8728 0.9214 0.7568 0.7038 s2 0.0071 0.0281 0.1041 0.0660 0.0235 N 46 25 22 19 16 B/Wmax 7.53a 3.15b 4.25b 3.82c 4.24b a p < 0.002. b p < 0.01. Cp < 0.10.

(e.g., OuterIsland (closed): r2 = 0.13; P < changedbreeding status 44% ofthe time com- 0.001),and strobilusratios and thenumber of paredto 15%of the time for monoecious yews. strobiliproduced by a plant (e.g., OuterIs. Most of thechanges in thebreeding status of (closed)r2 = 0.29; P < 0.001). single-sexyews were to monoecyor nonflow- eringstatus. "Sex reversal"(changes from male Annual variationin sex expression-The to femaleor vice versa)in thesetwo popula- breedingtype of individual plants changed an- tionsoccurred four times in 4 years;two of nually,and mostchanges were in single-sexed thesewere from female to male. Changesin individuals.For example, on OuterIs. (closed) monoeciousplants were typically to male (23 and OtterIsland (all yr),male and femaleyews of30 changes)and lessfrequently to female(6 of 30). One plantbecame nonflowering. Male plantsbecame monoeciousin 19 of the 28 Outer Island changesthat occurred in thisgroup. Only one plantremained female for 2 consecutiveyears 30- (out of eightoccurrences); male plants re- mainedmale in consecutiveyears 61% ofthe I20 time. Genderbased on male and femalestrobilus 20 10 _ 1.0 ai aina n o

0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1.0 * * 0.8 * ;

U Monoecious n Male Otter Island 0.6.

C,~ ~ * U 20 0.4 . *

c 155 0.2 . E. * C-10-

0.0 .. 5- 0 50 100 150 200 250

Length(cm) 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1.0 gender Fig. 3 Gendervs. lengthof plant(as definedin Ma- terialsand Methods),for monoecious and male Canada Fig.2. Frequencydistribution of gender for (a) Outer yewplants on OuterIsland (closed)- 1985. Genderwas Island- 1985 and (b) OuterIsland- 1985. Genderwas calculatedas definedin Materials and Methods and ranges calculatedas definedin text(1.0 = male only). from0.0 (female)to 1.0 (male). April 1991] ALLISON-CANADA YEW SEX EXPRESSION 575

TABLE 4. Kendall'snonparametric correlation of gender yewsthat produced only one or two strobili fordifferent years for Rocky Island, Otter Island, and peryear. OuterIsland (closed). Gender was calculated using male andfemale strobilus production as indicesof paternal and maternaleffort for all populations.Correlations DISCUSSION forOuter Island werealso performedon gendercal- culatedusing seeds produced as theindicator offemale Canada yew,despite being listed as mon- effort.N refersto thenumber of plants on each island oecious (e.g., Chamberlain1966), comprises includedin theanalysis single-sexedindividuals as well.The latter were

1982 1983 1984 1985 recordedon RockyIsland, where entire genets weresampled, indicating that the presence of RockyIsland (N = 16) single-sexedCanada yew individuals in natural 1983 0.1333 populationsis notan artifactof sampling only 1984 0.3162 0.4462a a in high-densitypopula- 1985 0.3917a 0.3233a 0.2751a a portionof genet 1986 0.2333 -0.0254 0.2624a 0.0936 tions.When quantitative measures are taken intoaccount, however, the complexity of sex OuterIsland (closed) - (femalestrobili) (N= 46) expressionin Canada yewmakes generaliza- 1983 0.4412a of sex expres- 1984 0.3158 0.3034a tionsabout underlying patterns 1985 0.6013a 0.4038a 0.5691a sion in thisspecies difficult. Some trendsare apparentin the data, but Outer Island (closed) - (seeds) (N = 46) theseare weakor oftencontradicted by alter- 1984 0.3393a native measuresof sex expression.For ex- 1985 0.2949a 0.4141a ample,qualitative measures in Island popu- OtterIsland (N = 44) lations indicate clearly that femaleness 1984 0.3179a (recordedas a shiftfrom male to monoecy) 1985 0.3922a 0.4684a increaseswith plant size. Female gender is only a Indicatessignificant concordance at P < 0.05. weaklycorrelated with plant size, however, and if strobilusratios are used as the measureof sexexpression, the trend is reversed.The measuresof plant size used in thisstudy may not productionshowed significant concordance in accuratelyreflect plant status, but it is apparent 15 of 21 year-by-yearcomparisons; gender thatgender variation among yew individuals concordancein OtterIsland, Outer Is. (closed), cannotbe explainedsimply by size differences. Root River,and GreyCloud yewswas signif- Flexibilityin yewgender is suggestedby sig- icantin 10 of 11 cases (Table 4). Kendall'stau nificantannual variation in strobilusratios of forGrey Cloud and Root Riverwere 0.6217 individualyews indicating that a plant'ssex (1984 to 1985) and 0.6000 (1983 to 1984), allocationis responsiveto environmentalvari- respectively(P < 0.001), and thesewere not ation.Significant concordance of gender in in- includedin Table 4. Concordancewas typically dividualyews, however, also suggestsa prox- low in all populations(e.g., Table 4). Calcu- imategenetic component to yew sex expression lationsusing the numberof ripe seeds as an (e.g.,Primack and McCall, 1986). The largest index of femaleeffort for Outer Is. (closed) genderfluctuations occurred in smalleryews; genderyielded results similar to thosewhen largeyews showed relatively little gender flex- femalestrobili were used (Table 4). Significantibility. yearto yearcorrelation in genderof individual The populationwith the lowest gender con- yewsis in contrastto the significantannual cordancewas RockyIsland. Yews in thispop- variationin strobilusratios described above. ulationwere significantly smaller on average Sevenpercent to 18%of plants within a pop- thanyews in all otherpopulations, and con- ulationhad majorgender shifts (as described sequentlyone would predict greater gender ad- in Materialsand Methods)between years. On justmentsin RockyIsland yews. A smallyew OuterIs. (closed)and OtterIsland, plants that producesfew strobili, and theaddition of one showed"flexible" gender between 1983 and or twostrobili (male or female)can have large 1984 (theinterval with the greatest number of effectson thatplant's gender. In general,it is changes)were significantly smaller than plants notclear whether large shifts in genderreflect thatshowed relatively little or no genderchange truegender adjustments or random changes in (one-wayANOVA; P < 0.005) betweenthose allocationpatterns. Greater yearly variation in 2 years.Gender shifts from male (gender= the genderof small or youngplants also has 1.0) to female(gender = 0.0) and viceversa in been observedin the oil palm (Elaeis gui- Canada yewoccurred in <2% of all observa- neensis)(Williams and Thomas, 1970) and tions,but in all casesthese shifts involved small lodgepolepine (Pinus contorta) (Smith, 1981). 576 AMERICAN JOURNAL OF BOTANY [Vol. 78

Sex-reversal,or genderphase change (sensu ficultto interpretbecause of thegender flexi- Lloydand Bawa, 1984),has beenreported in bilitydiscussed earlier. Perhaps individual yews dioeciousTaxus and has been cited as evidence havea geneticallydefined tendency toward male forenvironmental determination of sex ex- or femalegender, but have a limitedability to pressionin thisgenus (Freeman, Harper, and modifygender that declines with age. It will Charnov,1980). Those casesof sex reversal in be necessaryto separatethe effects of age and dioeciousTaxus, however, typically involved plantsize fromgenetic components of sex ex- the appearanceof male brancheson female pressionto determineconclusively that gender plants(i.e., chimeras) or the reverse, on hybrid in Canada yewis bimodal. yews(Keen and Chadwick,1954). This sug- If Canada yewgender is bimodal,it would geststhat "sex-reversal" in thisgenus involves be thefirst-reported case ofdimorphic gender a disruptionof the geneticsex-determining in gymnosperms.The potentialfor dimorphic mechanism(sensu Lloyd and Bawa, 1984), genderin Canadayew is particularlyintriguing perhapsresulting from hybridization, rather giventhe prevalence of dioecy in Taxus.Geo- thanenvironmental effects. Evidence for en- graphicdistributions in the genus(Ferguson, vironmentaldetermination of sex in Taxus is 1976) and cladisticanalysis of gymnosperms weakand inconclusive. (Donoghue,1989) supportthe hypothesis that The principaleffort in thisstudy was to col- dioecyis theancestral trait in Taxus and that lectdata in similarhabitats: Outer Is. (gap)was cosexualityin Canada yewis derived. an exceptionto this.Yews in the latterpop- Dioecy in manyplant species is genetically ulation, where light intensitywas higher, determined(see Chamov, 1982). If Canada showed a trendtoward more female-biased yew'sancestor was dioeciousand if dioecyis gender.This was observedin otherspecies geneticallydetermined in Taxus,it could ex- (Freemanet al., 1981;Charnov, 1982 and ref- plain the geneticcomponent of Canada yew erencescited therein), although the difference gender. Bimodal gender in Canada yewwould herewas notsignificant, possibly due to small be a remnantofthis species' dioecious heritage. samplesizes. No geographicdifferences (south- In turn,it wouldsuggest that in theevolution easternMinnesota vs. northernWisconsin) in of a cosexualyew species froma dioecious strobilusproportions were detected, and only ancestor,there was selectionfor the ability of OtterIsland had consistently different strobilus individualsto modifygender, i.e., gender flex- ratios(more female-biased) than other popu- ibility.Cosexual individualshave been re- lations.This resultis hardto interpret,how- portedas rarein naturalpopulations of Eu- ever, as currentenvironmental differences ropean yew (Taxus brevifolia)(Pridnya, 1984). among Island populationsare not obvious. Occasional productionof male brancheson Historicalfactors such as pastland-use or deer females,or vice versa, in hybridyews, suggests browsinghistory of the Islands may be im- a proximatemechanism for the evolutionof portant(Brander, 1983; Allison,1987). cosexualityin Taxus. The mostinteresting result of thisstudy is Close examinationof the ecologicaldiffer- thebimodal gender distributions recorded in encesbetween Canada yew and other members most studypopulations. This resultsuggests of the genus could suggesthypotheses con- thatCanada yew populationsare dimorphic cerningthe evolutionof Canada yewfrom a consistingof male and femalemorphs. Bi- dioecious ancestor.Givnish (1980), for ex- modal, or dimorphic,gender would indicate ample,hypothesized, in part,that dioecy is genderspecialization and a pronouncedgenetic prevalentin bird-dispersed gymnosperms, like componentto sex expressionin Canada yew Taxus,because the concentration of resources (Lloydand Bawa, 1984). As Fig. 3 indicates, into"fruit" would have a disproportionateef- thedifference between morphs is notdramatic, fecton a plant'sfemale fitness. The increased and the bimodaldistributions may be an ar- efficiencyof seed dispersalwith large "fruit" tifactof thepreponderance of male yews(0.9 crops would resultin a convex femalegain to 1.0 gender)in ApostleIsland populations. curvefavoring the evolutionof genderspe- Manyof these plants are small,and theyshow cialization(e.g., Charnov, 1982). thegreatest gender fluctuations. Although the Severalcharacteristics of Canada yewmay abundanceof males is relativelyconstant from reducethe hypothesizedfemale fitness gains yearto year, the male plants are not necessarily associatedwith gender specialization. Individ- the same individuals.Gender distribution in ualsof this species are the smallest in the genus, theRoot Riverpopulation also was bimodal, theirseeds ripen over an extendedperiod of 6 however,and did nothave thehigh frequency to 8 weeks,and pollen occasionally limits their of male yews. seed production(Allison, 1990). Canada yew In general,the bimodal distributions are dif- also layersextensively; individuals increase in April 19911 ALLISON-CANADA YEW SEX EXPRESSION 577 sizehorizontally rather than vertically, and they * 1920. Ovuliferousstructures of Taxuscanaden- become fragmentedas undergroundconnec- sis. BotanicalGazette 69: 492-520. ENGELMAN, L., AND J. A. HARTIGAN. 1969. Percentage tionsrot. Any fitness advantages gained by gen- pointsof a testfor clusters. Journal of the American der specializationto improveseed dispersal StatisticalAssociation 64: 1647-1648. maybe counteractedby any or all ofthe above FERGUSON,D. K. 1976. Somecurrent research on fossil characteristicsof thisspecies. and recenttaxads. Review of Palaeobotany and Pal- Understandingthe selective mechanisms for ynology26: 213-226. a hypothesizedreversion to cosexuality in Can- FREEMAN,D. C., K. T. HARPER,AND E. L. CHAmNov.1980. ada yewis relevantto thecurrent controversy Sex changein plants:old and newobservations and newhypotheses. Oecologia 47: 222-232. concerningthe evolution and maintenanceof , E. D. McARTHuR,K T. HAluER,AN A. C. BLAU- dioecy.This debatehas typicallyfocused on ER. 1981. Influenceof environment on thefloral sex pathwaysfrom cosexual to fully dioecious taxa; ratioof monoecious plants. Evolution 35: 194-197. hypothesesfor this pathway involve outcross- GIwmsH,T. J. 1980. Ecologicalconstraints on theevo- ingadvantages, sexual selection, and ecological lutionof breeding systems in seedplants: dioecy and correlates(e.g., Givnish, 1980; Thomsonand dispersalin gymnosperms.Evolution 34: 959-972. KEEN, R., AND L. C. CHADWICK.1954. Warnpropagators Barrett,1981; Bawa, 1982; Willson, 1982). to watchfor sex reversalin Taxus.American Nurs- Studyingthe ecologicalcircumstances where eryman100: 13-14. dioecybreaks down, such as thehypothesized LiLox', D. G. 1980. Sexual strategiesin plantsHI. A case ofCanada yew,could resolve some of the quantitativemethod for describing the gender of plants. questionssurrounding this debate, although to NewZealand Journal of Botany 18: 103-108. myknowledge this evolutionary pathway has a AND K. S. BAWA.1984. Modificationof the gen- rarelybeen considered.Cosexual, or monoe- derof seed plants in varyingconditions. Evolutionary Biology17: 255-338. cious,taxa occurin otherpredominantly di- MARTELL, A. M. 1974. Canada Yew. In J.D. Gill and oeciousgymnosperm genera such as Ephedra W. M. Healy[eds.], Shrubs and vines for northeastem andJuniperus. Cosexual species in these genera wildlife,158-160. USDA Forest Service General Tech- mayhave bimodalgender, and theymay pro- nicalReport NE-9, Northeastern Forest Experiment vide additionalecological correlates to deter- Station,Upper Darby, PA. 180 pp. minethe mechanismsinvolved in theevolu- MArrHEws,J. D. 1963. Factorsaffecting the production ofseed by forest trees. Forestry Abstracts 24: 1-13. tionof monoecy from dioecy and vice versa. McKoNE,M. J. 1989. Intraspecificvariation in pollen yieldin bromegrass (Poaceae: Bromus).American LITERATURE CITED Journalof Botany 76: 231-237. ,ANDD.W.ToNKYN. 1986. Intra-populationgen- dervariation in commonragweed (Asteraceae: Am- Can- ALLISON,T. D. 1987. The reproductivebiology of brosiaartemisiifolia L.), a monoecious,annual herb. ada yew(Taxus canadensisMarsh.) and its modifi- Oecologia70: 63-67. cationby herbivory. Ph.D. thesis,University of Min- McLAuGHLIN,R. L. 1989. Search modes of birds and nesota.Minneapolis. lizards:evidence for alternative movement patterns. 1990. The influenceof deer browsingon the AmericanNaturalist 133: 654-670. reproductivebiology of Canada yew (Taxus cana- PRIDNYA,M. V. 1984. Phytocenoticstatus and structure densisMarsh.). II. Pollenlimitation: an indirecteffect. ofthe Khosta Common-yew population in theCau- Oecologia83: 530-534. casusBiosphere Reserve. Ekologiya 1: 3-8. BAWA,K. S. 1982. Outcrossingand the incidenceof PRIMACK, R. B., AND D. G. LLOYD. 1980. Sexualstrat- dioecismin islandfloras. American Naturalist 119: in IV. The distributionsofgender in two 866-871. egies plants monomorphicshrub populations. New Zealand Jour- BEALS,E. W., G. COTTAM,AND R. J.VOGL. 1960. Influ- 109-114. ence of deer on vegetationof the ApostleIslands, nal ofBotany 18: 24: 68- , AND C. McCALL. 1986. Gendervariation in a Wisconsin.Journal of WildlifeManagement a 80. red maple population(Acer rubrum: Aceraceae): of a species. BRANDER, R. B. 1983. Environmentalassessment: nat- seven-yearstudy "polygamodioecious" uralresources inventory and management-Apostle AmericanJournal of Botany 73: 1239-1248. IslandsNational Lakeshore, Wisconsin. Apostle Is- SARvAS,R. 1968. Investigationson the floweringand lands NationalLakeshore, Department of Interior, seed cropof Picea abies. CommunicationesInstituti Wisconsin,unpublished report. ForestalisFenniae 67: 1-84. CHAMBERLAIN,C. J. 1966. Gymnosperms,structure and SCHOEN,D. J.,D. DENTIs,AND S. C. STEwART. 1986. Stro- evolution.Dover Publications,New York. bilus productionin a clonal whitespruce seed orchard: CHARNov,E. L. 1982. The theoryof sex allocation. evidencefor unbalanced mating. Silvae Genetica 35: PrincetonUniversity Press, Princeton, NJ. 201-205. CONOVER,W. J. 1980. Practicalnonparametric statistics. SMrrH,C. C. 1981. The facultativeadjustment of sex Wileyand Sons,New York. ratioin lodgepole pine. American Naturalist 11 8: 297- DONOGHUE,M. J. 1989. Phylogeniesand the analysis of 305. evolutionarysequences, with examples from seed SNEDECOR, G. W.,AND W. G. COCHRAN. 1967. Statistical plants.Evolution 43: 1137-1156. methods,6th ed. Iowa StateUniversity Press, Ames, DUPLER, A. W. 1919. Staminatestrobilus of Taxus can- IA. adensis.Botanical Gazette 68: 345-364. STANTON,M. L. 1984. Seed variationin wild radish: 578 AMERICAN JOURNAL OF BOTANY [Vol. 78

effectof seed size on componentsof seedling and adult WHITHAM, T. G., AND S. MOPPER. 1985. Chronic her- fitness.Ecology 65: 1105-1112. bivory;impacts on architectureand sexexpression in THOMPSON, J.N. 1984. Variationamong individual seed pinyonpine. Science 228: 1089-1091. massesin Lomatiumgrayi (Umbelliferae) under con- WILLiAMS, C. N., AND R. L. THOMAS. 1970. Observations trolledconditions: magnitude and partitioningofthe on sexdifferentiation inthe oil palmElaeis guineensis variance.Ecology 65: 626-631. L. Annals of Botany 34: 957-963. THOMSON, J.D., AND S. C. H. BARRETT. 1981. Selection WILLSON, M. F. 1982. Sexual selectionand diclinyin foroutcrossing, sexual selection, and theevolution of angiosperms.American Naturalist 119: 579-583. dioecyin plants. American Naturalist 118: 2443-2449.