Bulletin of' the Torrey Botanical Club 123(1 ), 1996, pp. 7-15
Reproductive biology of the monoecious clonal shrub Tax us canadensis 1
Paul Wilson,2 Michelle Buonopane3 and Taber D. Allison4 'Department of Biology, California State University, 18111 Nordhoff Street, Northridge, CA 91330-8303 'Department of Biology, Bates College, Lewiston, ME 04240 "Department of Forestry and Wildlife, University of Massachusetts, Amherst, MA 01002
WILSON, P. (Department of Biology, California State University, 18111 Nordhoff Street, Northridge, CA 91330-8303), M. BUONOPANE (Department of Biology, Bates College, Lewiston, ME 04240) AND T D. ALLISON (Department of Forestry and Wildlife, University of Massachusetts, Amherst, MA 01002). Reproductive biology of the monoecious clonal shrub Taxus canadensis. Bull. Torrey Bot. Club 123: 7-15 1996.-Strobilus production, pollination, and maturation were studied in Taxus canadensis, a procumbent clonally-spreading shrub. Plants that produced many cones had greater shoot growth; no obvious tradeoff between sexual effort and vegetative growth was observed. T. canadensis is monoecious, but plants showed considerable variation in phenotypic gender. Because related species are dioecious, we hypothesize that T. canadensis evolved from a dioecious ancestor. Monoecy may have been favored as a means of assuring pollination via selfing; the proportion of ovules pollinated was positively correlated with the number of male cones on a plant. Sclfing may be at the cost of inbreeding depression; 26r/r; of the female cones aborted, about twice the percentage of the dioecious Taxus cuspidata. Application of nutrient fertilizer did not reduce the abortion percentage. Seed and aril matu ration in T. canadensis occurred over a three month period, much more gradually than in T. cuspidata. Removal of the seeds, probably by antagonistic rodents, was very rapid and thorough in T. canadensis. Asynchronous ripening and monoecy may help reduce seed predation by rodents by reducing the conspicuousness of "fruiting" displays relative to the dioecious species. Key words: monoecy, phenotypic gender, pollination limitation, reproductive allocation, seed abortion, seed consumption, Taxus canadensis.
Detailed studies of the reproductive biology sis adds a distinct perspective on these issues, of wind-pollinated woody plants have lagged be which we will speculatively relate to each other hind those of showy meadow herbs. In the pres by outlining an integrated shift in the reproduc ent study, on Taxus canadensis Marsh., we pro tive system of T. canadensis away from that of vide data relating to several topics that have re other Taxus species. ceived much attention from authors who were Taxus canadensis is a monoecious evergreen mainly focusing on other types of plants: (I) shrub native to mixed conifer-hardwood forests strategies of monoecy versus dioecy (Bawa of the northeastern United States and southeast 1980, 1994; Muenchow 1987; Thomson and ern Canada (Allison 1991). It is the only mon Brunet 1990), (2) life history allocation to oecious member of its genus, and monoecy in growth and reproduction (Delph 1990; Primack this case almost certainly arose from dioecy (see and Hall 1990; Cippollini and Whigham 1994), Discussion). Male and female cones are initiated (3) ecological variation in the degree of polli during the summer in the axils of leaves, and nation limitation (Zimmerman and Pyke 1988; they overwinter as buds. In the spring, ovules Young and Young 1992; Wilson et al. 1994), (4) are fertilized by wind-dispersed pollen. Unpol the abortion of immature seeds (Stephenson linated ovules remain small and become whitish 1981; Willson and Burley 1983), and (5) seed yellow, thus distinguishing them from develop predation versus effective seed dispersal (Janzen ing nascent seeds. Throughout the summer some 1971; Price and Jenkins 1986). Taxus canaden- cones abort, with first the receptacle and then the seed coat becoming shriveled and yellow. 1 We thank S. Kinsman, B. Smith, J. Thomson, and The seed, receptacle, and pedicel of these abort B. Tomlinson for criticism, and the Hurwitzes of Pe ed cones eventually fall off the plant. Mature tersham for permision to study their Taxus cuspidata trees. Financial support was provided by the Mellon female cones consist of a single seed surrounded Foundation, the National Science Foundation, and the by a fleshy red aril. New England Consortium for Undergraduate Science Taxus canadensis spreads clonally. Procum Education. This work was done under the auspices of bent branches take root resulting in plants that Harvard Forest. Received for publication February 7, 1995 and in expand horizontally. Branch by branch, the revised form July I, 1995. clone spreads along the forest floor, sometimes
7 8 BULLETIN OF THE TORREY BOTANICAL CLUB [VoL. 123
Table I. List of study sites.
Males Females Fertilizer Site N' counted counted experiment Site description Templeton 51 Yes Yes Yes Roadside, mixed hardwood-coni- fer canopy, dense patch. Riceville Road 72 Yes Yes No On a steep hill, dense, next to pasture, in a pine stand. Riceville Pond 50 No Yes No Open, relatively high light, road- side, mixed canopy. Nelson Road 44 No Yes Yesh On a hill, mixed hardwood-coni- fer canopy. Nelson Brook 50 No Yes No Next to brook, moist, mixed hardwood-conifer canopy.
'Number of "plants" sampled = rooted branches (sec text). h Possible effects on fruit maturation were not followed.
forming huge patches. In our study, a "plant" ipers twice, with the measurements taken per was considered to be one shoot system emanat pendicular to each other. The mean of these two ing from the forest litter. Our study plants were numbers was used to represent plant size. New well spaced, but some of them may have been ( 1994) growth was measured as the length of the separate ramets of the same genet. five longest sprig tips, which were lighter green Taxus canadensis, though poisonous to hu in color and softer in texture than previous mans, is a winter forage for deer and moose (Al years' growth. The average of these five num lison 1990a). Birds are assumed to act as dis bers was used as an indication of the amount of persal vectors, eating the red aril and voiding the growth. seed at some distance from the parent plant. Seed development was closely monitored on Studies on other species of Taxus have shown the twelve plants at the Templeton site that had that seeds fed to birds and spit out or excreted ten or more ovules. To test for nutrient limitation can later germinate (Barkowiak 1978). Not of seed maturation, half the plants were given much information is available concerning seed about one pint of Rapid GrowCRJ fertilizer every predation by rodents, but Bartkowiak ( 1978) two weeks from mid-June until the end of cites squirrels as eating Taxus seeds, and Allison "fruiting" in September. At the beginning of (pers. obs.) has often noted rodent teeth marks July, we tagged each branchlet that bore a fe and has observed chipmunks, squirrels, and male cone, and we recorded the condition of the mice eating Taxus canadensis seeds. cone every two to three days. Cones were scored as (I) green and healthy but not yet with a red Methods. We studied Taxus canadensis at aril, (2) having a red aril, (3) removed leaving five sites in north-central Massachusetts, as de only the receptacle or pedicel stub, or (4) yellow scribed in Table I. At each site, approximately and aborting. Very few were unaccounted for. fifty plants (i.e., rooted shoot systems) were cho For comparison, we also followed seed matura sen at random distances along parallel transects tion and removal on a stand of cultivated Taxus and tagged. In late May, we recorded for each cuspidata, which is a dioecous upright tree. The plant the number of male cones, the number of same procedure for marking and scoring female female cones, and the number of withering, pre cones was used. sumably unfertilized ovules, which we distin In order to evaluate whether or not rodents eat guished from pollinated immature seeds on the Taxus arils and seeds, we live-trapped a Pero basis of color. Due to the late date of sampling, mvscus maniculatus underneath a clone of T. the counts of male cones are only estimates canadensis. This was done after the peak in seed some of the male cones may have already fallen maturation but while some seeds were still rip off-and at three of the five sites the number of ening. We placed the animal in a bucket with males cones could not be determined with sat five mature T. canadensis cones and periodically isfactory certainty (Table I). observed what had been eaten. In late June, we collected data on size and growth for each plant at the five sites. The basal Results. GE~DER. Phenotypic gender was cal diameter at ground level was measured with cal- culated following Lloyd ( 1980; Lloyd and Bawa 1996] WILSON ET AL.: REPRODUCTION IN CANADA YEW 9
Templeton (Fig. I, lower panel). At both sites, intrapopu lation differences are evident in the percentage of male and female plants. The two sites also differed in gender, with Templeton having a higher percentage of male plants and Riceville >, u Road a higher percentage of female plants. c 8
0.21 1.8 • • ...... 0.18 Q) ...... 1.5 Q) Q) • ...... E • Eo.1 s ro • ro • 0 1.2 0 Cii '· Cii 0.1 2 Ill • Ill • • • ~ 0.9 ro • Ee 0.09 • • In ' :. ,...... , • ' • • Ill • • Q) ...... • • 4• • • "* 0.6 • E • ••• .,• .. ... • ~ 0.06 , Q) • ~ . • • u.. •• •• t: • • ~0.3 • • • • t5f 0.03 • • • • C"" •• • • • (./) • • • • • .r. ~~ 0 I ·- 0 I 5 1 5 25 35 45 55 65 75 85 10 20 30 40 so 60 70 80 New Growth New Growth
Riceville Road r8=0.42* (n=SS) Riceville Road r8 =0.059 (n=72)
Riceville Pond r8 =0.1 3 (n=SO) males only r;;=0.351* (n=39)
Nelson Road r8 =0.18 (n=44)
Nelson Brook r8 =0.26t (n=49)
Templeton r8 =0.24 (n=47) Templeton r;;=0.345* (n=47)
Fig. 2. Comparison of reproductive output and growth for Taxus canadensis at Riceville Road. Sqrt ~ Square root transformed. Spearman's Rank correlation coefficients arc given for all five populations UP < 0.1; *P < 0.05). Open circles indicate male individuals on the left panel and female individuals on the right panel.
(Table 2). The sites also differed in the amount number of male cones might experience high or of variation in pollination success among plants low pollination success (Fig. 4 ). within each population. For example. Templeton showed little variation with most plants experi SEED MATURATION. ABORTION, AND PREDA encing high pollination success. Riceville Road. TION. Aril and seed ripening was very gradual however, displayed very broad variation, with a in Taxus canadensis (Fig. 5, upper panel). At the bimodal distribution of success. There were Templeton site, seed ripening extended over a many plants with a pollination success of less period of three months. There was an early wave than 5% and many with a pollination success of maturation at the beginning of July, but mat close to I 00%, as well as variation between the uration soon slowed, and the last aril did not turn extremes (Fig. 3). Pollination success was posi red until the end of September. In addition to tively correlated with the number of male cones maturing gradually, the ripe "fruits" were spa produced (Templeton and Riceville Road com tially scattered within and among ramets and did bined r, = 0.57; N = 93; P < 0.001). A plant not produce a showy display. The abortion rate with a high number of male cones will have a was 26% overall; it ranged from 50% to 8%, and high pollination success, but a plant with a low there was no significant effect of nutrient fertil-
Table 2. Among-site variation in pollination success.
No. ovules pollinated Median Total no. of ovules pollination successd N''
Templeton "'%49 ~ 94'1<: !.0 (5 I) 2 Riceville Road " 'Y12 u ~ 27% 0.6 (72) Riceville Pond '"Ym ~ 60r/r; 0.5 (47) 66 Nelson Road Y712 = 93('::0 0.96 (37) Nelson Brook 17 '1, ~ 75% 0.74 (50)
'Kruskal-Wallis x' ~ 45.13 (4 df); P < 0.00!. h Number of "plants" sampled ~ rooted branches (see text). 1996] WILSON ET AL.: REPRODUCTION IN CANADA YEW 11
Templeton 1.1,------, 30.------~------. 0.9 25 ,. "'~ 0.7 u u 0 >-2 ::l u 6 c ~ 0.5 0 0 ~ 1 ·.;::; CT .. -: 12 Taxus canadensis i'? 1 0 c 1,/l 250 ~ 8 "c Ripened and Eaten CT 8 zoo ~ 6 u.. ro" ~ 150 4 u.. 0 100 Immature 2 ~ " 26% 0 ~ 50 0.05 0.25 0.45 0.65 0.85 z" Pollination Success 6 July 1 August 1 September Census Date Fig. 3. Frequency distribution of pollination suc cess for Taxus canadensis for Templeton and Riceville Road. :g"' 1 8 ro" izer addition on the proportion aborted (angu E 1 u.." larly transformed; t = 0.777 with 10 df; P = 0 0.455). Most of the abortion happened well after i3"' .0 the first arils of other seeds on a plant had turned E z red. In our study (unpubl. data) of the dioecious " species Taxus cuclpidata, maturation was much more simultaneous with most arils on a plant turning red in the period of a few days (Fig. 5, Fig. 5. Cumulative seed maturation and abortion. For Taxus canadensis, there were 13 cones that were lower panel). Also the cones were produced in eaten before they matured, and they are not included dramatic clusters on the trees. The abortion rate in the graphed data. "Ripe" means having an aril that was lower, 13%, and abortion occurred before was seen as having some red or in a few cases esti the other cones matured-also most of this abor mated as likely to turn red before the next census; it tion occurred on one plant, unlike the situation does not mean the arils and seeds had reached full size, which for T. canadensis most never did. All numbers in T. canadensis. Only three of the 189 cones on the graph are based on pooling all ccnsuscd plants that we followed were removed by animals. in the population, but there were not major differences Most arils became mushy or rotten several among plants in T. canadensis. For Taxus cuspidata, weeks after becoming red, and then the cones the several declines in number of immature cones are a result of different individuals maturing their cones Taxus canadensis fell to the ground. In when an on different weeks; even when these individuals arc aril turned red, the cone was generally removed pooled the simultaneous maturation is dramatic com very quickly by animals: the median time to re- pared to T. canadensis. 12 BULLETIN OF THE TORREY BOTANICAL CLUB [VOL. 123 moval was 2.3 days (one census unit), the mean around. Two non-mutually exclusive theories 3.8 ± 2.72 SD (N = 74). Fifty-one percent of seem reasonable for Taxus canadensis: (I) mon cones were removed when they had been green oecy may have been favored because it allowed in the previous census (N = 184). These typi for self-pollination in plants that had become cally were well developed, and some of them frequently pollen limited (Allison 1990b); and may have been removed after turning red but (2) dioecy may have been no longer favored be before we recorded them as such. We could not cause the ecology of seed disperseral and sur tell directly whether cones were being removed vival no longer made it advantageous to have as by birds that might disperse the seed or by ro many "fruits" as possible concentrated on half dents, but in numerous instances we found the plants (Bawa 1980; Givnish 1980). Both gouged arils with the seed removed and pedicel chronic pollen limitation and chronic seed pre stubs that seemed to have been cut by a sharp dation may in turn be due to the prostrate, out tooth, not broken by merely having the aril wardly spreading, clonal habit. plucked. Our captive deer mouse ate all five Life history theory typically assumes that cones that we provided, leaving only the intact there is a tradeoff between reproductive effort receptacles. The first was eaten after I hour and and vegetative growth (Willson 1983; Stearns 22 minutes, and when we left the mouse alone 1992; Reekie and Bazzaz 1992). Our results do and returned some 4 hours later, the remaining not support this assumption. We found positive four cones were eaten, again with the receptacles correlations between cone production and the left intact. length of new shoots, even holding plant size constant. This could be accounted for by a num Discussion, Our results on gender distribu ber of mechanisms (Reekie and Bazzaz 1987). tion confirm the findings of Allison ( 1991) There seems to be more variance in general vig from sites in the Apostle Islands of Lake Su or affecting both growth and reproduction than perior. In both regions, and hence probably for in any tradeoff between growth and reproduc the species a whole, there is a dioecious ten tion. Of course, there may still be a negative dency toward bimodal gender. This tendency genetic or evolutionary correlation, but most of may be a phylogenetic holdover from a dioe the variance among plants seems to be in how cious ancestry. All other species of Taxus seem much resource they have acquired, not in how to be usually dioecious (T. baccata and T. cus they allocate it (van Noordwijk and de Jong pidata-pers. obs.; T. brevifolia-Taylor and 1986 ). This suggests that the plants that are best Taylor 1981; T. jloridana-Godfrey 1988; T. at spreading clonally may also be those that are globosa-Standley and Steyermark 1958; T. best able to afford producing large numbers of sumatrana-de Laubenfels 1978; T. chinensis, male and female cones. Such a complex positive T. wallichiana, and T. yunnanensis-Botanical correlation cannot be proven without a con Institution Chinese Academy of Sciences, Bej trolled experiment, but the census data do point ing, 1972, and Cheng et al. 1978). Furthermore, in that direction. The plants that grew more, and related genera are dioecious such that a phy probably have spread out more, also produced logenetic analysis of character evolution un more cones and probably sired and set more ambiguously infers the three hypothetical com seed. mon ancestors below Taxus to have been di Our pollination-success results indicate that oecious (Hart 1987; Donoghue 1989). It is, some of our populations were experiencing pol therefore, almost certain that Taxus canadensis len limitation (see also Allison 1990b, 1990c, is secondarily monoecious. 1993), but, interestingly, some parts of those What ecological conditions might be func populations showed high pollination success. tionally associated with the change in breeding Considerable variation must therefore exist in system toward monoecy? Most discussions of the availability of airborne pollen at a very lo the strategic stability of monoecy versus dioecy cal scale. Substantial site-to-site differences in have focused on evolution in the opposite direc pollination limitation have been found in insect tion, that is, from cosexuality to dioecy. For in and bird pollinated species (see references in stance, the arguments presented by Bawa Wilson et al. 1994, p. 285; Wilson 1995). Our (1980), Givnish (1980), Muenchow (1986), and results provide an example of such variation in Charlesworth and Charlesworth ( 1978) all center a species with wind dispersed pollen. Some of on possible reasons for the sexes becoming sep this variation among sites may be explained by arated. The arguments, however, can be turned variation in levels of selfing associated with the 19961 WILSON ET AL.: REPRODUCTION IN CANADA YEW 13 degree of monoecy. As the number of male pollination success was higher presumably be cones increased pollination success also in cause of selfing; see also Allison 1993). His creased. Selfing may have prevented pollen experiment did not test whether or not there is limitation in some plants and in some sites. inbreeding depression given equal pollination Specifically, pollination did not limit female by self and outcross pollen. Monoecy may yet function at Templeton where plants were most be shown to be at the cost of increased abortion ly hermaphrodites or males, whereas pollina due to inbreeding depression. tion was limiting for some plants at Riceville Seed predation was exceedingly high in our Road where there were large numbers of essen study of Taxus canadensis. Although we have tially female plants. It is quite possible that, if not discovered a way of directly distinguishing Taxus canadensis were still dioecious, it would between removal of seeds by birds versus ro be chronically pollen limited. The architecture dents, we believe rodent attack was over of the species seems to act against adequate whelming, as is quite possibly the case for cross-pollination: layering increases the size of many understory plants (Heithaus 1981 ). Be clones horizontally rather than vertically, and cause T. canadensis plants are procumbent, the plants are very close to the ground as well as cones are produced close to the ground where protected by a dense overstory such that pollen they are easily accessible to mice and chip grains may not have far to fall in air that is munks. We found dozens of instances of what rarely turbulent. Individuals that were able to we interpreted as teeth marks. And we know become cosexual may have been favored after that rodents will eat seeds and arils of T. can the evolution of this growth form. adensis. Some seeds are even eaten before they The high rates of seed abortion that exist in mature, and most do not stay on the plant for many plants are a curious feature to explain more than a few days after developing a red (Stephenson 1981 ). Several explanations have aril. It would be interesting to do an experi been proposed. Conceivably it is because plants mental study of how distance from the ground are hedging their bets in favor of stupendously and number of ripe cones affect removal rates good years when resources are so plentiful that by rodents and birds. There is no indication that they are able to mature all their seeds (Ehrlen large "fruit" crops would be especially attrac 1993). Or perhaps it is because mother plants tive to birds who might disperse the seeds. are choosing to mature the best embryos from There is no indication that birds are attracted to a larger field of candidates (Casper 1988; Ber the seeds of T. canadensis at all. Thus, Giv tin and Peters 1992). Or perhaps some embryos nish's (1980, 1982) theory for the evolution of are genetically inviable on their own, either be dioecy could work in reverse for the evolution cause of inbreeding or genetic bad luck (Lee of monoecy in T. canadensis: allocating energy 1988). Seed abortion in Taxus canadensis was to male function probably does not lower the 26% at our study site in Massachusetts, and this fruit crop below any threshold of attractiveness confirms the high levels of abortion observed to beneficial birds but it might to antagonistic by Allison (unpubl. data) in Minnesota and mammals (on the quality of mammals as seed Wisconsin. Almost all of the abortion happened dispersers/predators see Smith et a!. 1986; Dav after many of the cones that would develop had idar and Morton 1986; Carr 1992). The very already completely matured. Light limitation gradual and scattered presentation of "fruit"- may play a role. Aborted cones are a bit smaller much more gradual and scattered than in Taxus than fully mature cones-the seed is perhaps cuspidata-might well reflect an ecology in half as large, and there is no aril. Still it is hard which a lone cone has a greater chance of not to believe that not maturing them is energeti being eaten than a cone on a heavily laden cally imposed. In addition, we found no evi plant, in which case the female gain curve dence from our nutrient fertilizer study that would be convex, not concave, and monoecy abortion was due to lack of mineral nutrients. would be favored (Charnov 1982). The explanation that seems the most reasonable and would be related to differences between the Literature Cited species in mating system is seed failure due to selfing. Allison (1990c) did an experiment ALLISON, T. 1990a. The influence of deer browsing on the reproductive biology of Canada yew (TciXus showing that female cones on emasculated canadensis Marsh.). I. Direct effect on pollen, plants (where selfing was prevented) has simi ovule, and seed production. Oecoiogia 83: 523- lar abortion rates to monoecious plants (where 529. 14 BULLETIN OF THE TORREY BOTANICAL CLUB [VOL. 123 1990b. Pollen production and plant density GIVNISH, T. J. 1980. Ecological constraints in the evo affect pollination and seed production in Taxus lution of breeding systems in seedplants: dioecy canadensis. Ecology 71: 516-522. and dispersal in gymnosperms. Evolution 34: 959- ---. 1990c. The influence of deer browsing on the 972. reproductive biology of Canada yew (TciXus cana 1982. Outcrossing versus ecological con densis Marsh.). II. Pollen limitation: An indirect straints in the evolution of dioecy. Am. Nat. 119: effect. Oecologia 83: 530-534. 849-865. ---. 1991. Variation in sex expression in Canada GODFREY, R. K. 1988. Trees, shrubs and woody vines yew (TciXus canadensis). Am. J. Bot. 78: 569-578. of northern Florida and adjacent Georgia and Ala ---. 1992. The influence of deer browsing on the bama. University of Georgia Press, Athens, Geor reproductive biology of Canada yew (Taxus cana gia. densis Marsh.). III. Sex expression. Oecologia 89: HART, J. A. 1987. A cladistic analysis of conifers: 223-228. preliminary results. J. Arnold Arb. 68: 269-307. 1993. Self-fertility in Canada yew (Taxus HEITHAL:s, E. R. 1981. Seed predation by rodents on canadensis Marsh.). Bull. Torrey Bot. Club 120: three ant-dispersed plants. Ecology 62: 136-145. 115-120. JANZEN, D. H. 1971. Seed predation by animals. BARTKOWIAK, S. 1978. Seed dispersal by birds. Jn S. Annu. Rev. Ecol. Syst. 2: 465-492. Bialobok [ ed. [, The Yew-laxus baccata L. For LEE, T. D. 1988. Patterns of fruit and seed production, eign Scientific Publications Department of the Na pp. 179-202. In J. Lovett Doust and L. Lovett tional Center for Scientific, Technical and Econom Doust [eds.], Plant reproductive ecology. Oxford ic Information, Warsaw, Poland. University Press, New York. BAWA, K. S. 1980. Evolution of dioecy in flowering LLOYD, D. 1980. Sexual strategies in plants. III. A plants. Annu. Rev. Ecol. Syst. 11: 15-40. quantitative method for describing the gender of ---. 1994. Pollinators of tropical dioecious angio plants. New Zeal. J. Bot. 18: 103-108. sperms: A reassessment? No, not yet. Am. J. Bot. LLOYD, D. G. AND K. S. BAWA. 1984. Modification of 81: 456-460. the gender of seed plants in varying conditions. BERTIN, R. I. AND P J. PETERS. 1992. Paterna] effects Evol. Bioi. 17: 255-338. on offspring quality in Campsis radicans. Am. Nat. Ml:ENCHOW, G. E. 1987. Is dioecy associated with 140: 166-178. fleshy fruit"' Am. J. Bot. 74: 287-293. BOTANICAL INSTITUTION, CHINESE ACADEMY OF Sci PRICE, M. V. AND S. H. JENKINS. 1986. Rodents as seed ENCES, BEJING. 1972. Iconographia cormophyto consumers and dispersers, pp. 191-235. In D. R. rum Sinicorum, Vol. I. Murray [ ed. [, Seed dispersal. Academic Press, New CASPER, B. B. 1988. Evidence for selective embryo York. abortion in Cryptantha Jlava. Am. Nat. 132: 318- PRIMACK, R. B. AND P HALL. 1990. Costs of repro 326. duction in the pink lady's slipper: A four-year ex CARR, D. E. 1992. The relationship between crop size perimental study. Am. Nat. 136: 638-656. and fruit removal and its implication for the evo REEKIE, E. G. AND F. A. BAZZAZ. 1987. Reproductive lution of dioecy. Can. J. Bot. 70: 1784-1789. effort in plants. 3. Effects of reproduction on veg CHARLESWORTH, D. AND B. CHARLESWORTH. 1978. A etative activity. Am. Nat. 129: 907-919. model for the evolution of dioecy and gynodioecy. --- AND ---. 1992. Cost of reproduction as Am. Nat. 112: 975-997. reduced growth in genotypes of two congeneric CHARNOV, E. L. 1982. The theory of sex allocation. species with contrasting life histories. Oecologia Princeton University Press, Princeton, New Jersey. 90: 21-26. CHENG, W.-C., L.-K. FU AND C.-D. CHU. 1978. lax SMITH, B. H., M. L. RONSHEIM AND K. R. SWARTZ. us, pp. 438-448. ln. Flora Republicae Popularis 1986. Reproductive ecology of Jeffersonia diphv! Sinicae, Vol. 7. /a (Berberidaceae). Am. J. Bot. 73: 1416-1426. CIPPOLLINI, M. L. AND D. F. WHIGHAM. 1994. Sexual STANDLEY, PC. AND J. A. STEYERMARK. 1958. F:ora dimorphism and cost of reproduction in the dioe of Guatemala. Fieldiana 24: 60-62. cious shrub Lindera benzoin (Lauraceae). Am. J. STEARNS, S. C. 1992. The evolution of life histories. Bot. 81: 65-75. Oxford University Press, Oxford. DAVIDAR, P AND E. S. MoRTON. 1986. The relation STEPHENSON, A. G. 1981. Flower and fruit abortion: ship between fruit crop sizes and fruit removal rates Proximate causes and ultimate functions. Annu. by birds. Ecology 67: 262-265. Rev. Ecol. Syst. 12: 253-279. DE LAL:BENFELS, D. J. 1978. The taxonomy of Phil TAYLOR, R. L. AND S. TAYLOR. 1981. Tax us brnij(J!ia ippine Coniferae and Taxaceae. Kalikasan 7: 117- in British Columbia. Davidsonia 12: 89-94. 152. THOMSON, J.D. AND J. BRl:NET. !990. Hypotheses for DELPH, L. F 1990. Sex-differential resource allocation the evolution of dioecy in seed plants. Trends Ecol. patterns in the subdioecious shrub Hebe subalpina. Evol. 5: 11-16. Ecology 71: 1342-1351. VAN NOORDWIJK, A. J. AND G. DEJONG. 1986. Acqui DoNOGI!l:E, M. J. 1989. Phylogenies and the analysis sition and allocation of resources: Their influence of evolutionary sequences, with examples from on variation in life history tactics. Am. Nat. 128: seed plants. Evolution 43: 1137-1156. 137-142. EHRLEN, J. 1993. Ultimate functions of non-fruiting WILLSON, M. F. 1983. Plant reproductive biology. Wi flowers in Lathyrus vernus. Oikos 68: 45-52. ley, New York. ENGLEMAN, L. AND J. A. HARTIGAN. 1969. Percentage ---AND N. Bl:RLEY. 1983. Mate choice in plants. points of the test for clusters. J. Am. Statistical As Princeton University Press, Princeton, New Jersey. soc. 64: 1647-1648. WILSON, P 1995. Variation in the intensity of polli- 1996] WILSON ET AL.: REPRODUCTION IN CANADA YEW 15 nation in Drosera trac}·i: Selection is strongest Ym:NG, H. J. AND T. P YOUNG. 1992. Alternative out when resources are intermediate. Evol. Ecol. 9: 1-15. comes of natural and experimental high pollen ---,J.D. THOMSON, M. L. STANTON AND L. PRIG loads. Ecology 73: 639-647. NEY. 1994. Beyond floral Batemania: gender bi ZIMMERMAN, M. AND G. H. PYKE. 19RR. Reproduction ases in selection for pollination success. Am. Nat. in Polemonium: Assessing the factors limiting seed 143: 283-296. set. Am. Nat. 131: 723-738.