comparison of size-dependent reproductive effort in two dandelion ( taraxacum...

Comparison of size-dependent reproductive

effort in two dandelion (Taraxacum officinale)

populations

Clive V.J. Welham and Robert A. Setter

Abstract: Reproductive effort in dandelions (Taraxacum officinale Weber) from two different habitats was compared. Onedandelion population occupied a 5-year-old alfalfa (Medicago spp.) field, an environment subject to regular disturbance butwith a relatively low density of neighbours. Individuals from the second population were derived from a number ofundisturbed sites where the density of neighbouring grasses was high. Three hypotheses were evaluated with respect to theobserved patterns of reproductive effort. One hypothesis, that reproductive effort was a function of differences betweenhabitats in resource availability, did not provide an explanation for our results. A second hypothesis considered patterns ofreproductive effort when mortality rates varied with degree of disturbance and neighbour density, which was a consequence ofeach habitat representing a different successional environment. A third hypothesis used a life-history approach to predictreproductive effort when mortality schedules were size dependent. Both of these hypotheses received support for theirpredictions. There was more than a sixfold variation in reproductive effort among individuals from the alfalfa field and afourfold variation on the undisturbed sites. Much of this variation, however, was attributable to a size-dependent relationshipbetween reproductive effort and vegetative mass. Total reproductive effort (total seed plus scape mass) in both populationsincreased linearly with vegetative mass, but the slope for the population from the alfalfa field was significantly higher. Incontrast, proportional reproductive effort (total seed plus scape mass per vegetative mass) showed a curvilinear increase forthe alfalfa field population but was linear and negative for the undisturbed population. There were also important differencesbetween the populations in reproductive morphology. Larger plants on the alfalfa field had longer average scape lengths,produced more flower heads (capitula) per plant, had greater seed production, and had a lower ratio of seed mass per pappusarea; only mean scape and mean seed mass did not differ significantly. We suggest that dandelions on the alfalfa field have adifferent reproductive morphology to facilitate colonization of open areas on the field.

Key words: Taraxacum, dandelion, reproductive effort, biomass, life history, neighbour density, agriculture, weed.

Résumé: Les auteurs ont comparé l’effort reproductif chez des pissenlits (Taraxacum officinale Weber) provenant de deuxhabitats différents. Une des populations de pissenlit occupait un champ de luzerne (Medicago spp.) âgé de 5 ans, unenvironnement sujet à des perturbations régulières mais avec une densité relativement faible de voisins. Les individus de ladeuxième population ont été obtenus à partir d’un nombre de sites non-perturbés où la densité d’herbacées voisines étaitélevée. Trois hypothèses ont été examinées selon les patrons d’efforts reproductifs observés. Une hypothèse, à l’effet quel’effort reproductif serait fonction des différences de disponibilité de ressources entre les habitats, n’a pas réussi à expliquerles résultats. Une seconde hypothèse considére les patrons d’effort reproductif lorsque les taux de mortalité varient avec ledegré de perturbation et la densité des voisins, conséquence du fait que chaque habitat représente un environnement successifdifférent. Une troisième hypothèse se base sur une approche de cycle vital pour prédire l’effort reproductif lorsque lesséquences de mortalité dépendent de la dimension. Les prédictions selon ces deux hypothèses ont été supportées. On trouveune variation plus de six fois plus grande dans l’effort reproductif entre les individus du champ de luzerne et quatre fois pluspour les sites non perturbés. Une bonne partie de cette variation est cependant attribuable à une relation de dépendance sur ladimension entre l’effort reproductif et la masse végétative. L’effort total de reproduction (masse totale graine plus hampe)dans les deux populations augmente de façon linéaire avec la masse végétative mais la pente pour la population venant duchamp de luzerne est significativement plus élevée. Au contraire, l’effort reproductif proportionnel (masse totale graine plushampe par masse végétative) montre une augmentation curvilinéaire pour la population venant du champ de luzerne mais estlinéaire et négative chez la population non perturbée. Il y a également des différences importantes entre les populations entermes de morphologie reproductive. Les plus grandes plantes du champ de luzerne montraient des hampes plus longues quela moyenne, produisaient plus d’inflorescence (capitules), avaient une plus forte production de semences et un rapport masse

Can. J. Bot. 76: 166–173 (1998)

Received April 24, 1997.

C.V.J. Welham1 and R.A. Setter.2 Department of Biological Sciences, Simon Fraser University, Burnaby, BC V5A 1S6, Canada.

1 Author to whom all correspondence should be addressed. e-mail: [email protected] Present address: BugBusters Pest Management Inc., P.O. Box 1750, Prince George, BC V2L 4V7, Canada.

166

© 1998 NRC Canada

B97-176.CHPFri Mar 13 11:37:40 1998

Color profile: DisabledComposite Default screen

séminale par surface de l’aigrette plus faible; seules les masses moyennes des hampes et des graines ne diffèrent passignificativement. Les auteurs suggèrent que les pissenlits du champ de luzerne auraient une morphologie reproductivedifférente pour faciliter la colonisation des surfaces ouvertes dans le champ.

Mots clés : Taraxacum, pissenlit, effort reproductif, biomasse, cycle vital, densité des voisins, agriculture, mauvaise herbe.[Traduit par la rédaction]

Introduction

The principle of allocation states that if the supply of a criticalresource is limited then its distribution within an organismshould reflect a trade-off between the benefits of investment inone structure versus investment elsewhere (Cody 1966; Gadgiland Bossert 1970). One application of this principle concernsthe proportion of a plant’s total resources allocated to repro-duction (hereafter termed reproductive effort (RE); seeThompson and Stewart 1981). Estimates indicate that RE inplants may be substantial but shows considerable intra- andinter-specific variation. Soule and Werner (1981), for exam-ple, reviewed RE in 14 plant species. Estimates among speciesvaried by as much as 52% (range in variance 4–56%) andthough within-species values were lower, in some cases theywere still highly variable (range 5–36%).

A number of hypotheses have been proposed to explainpatterns of RE in plants (reviewed by Evenson 1983; Willson1983). One hypothesis is that differential resource allocationis a consequence of underlying differences in resource avail-ability (Gadgil and Bossert 1970; Pianka 1972). When re-sources are scarce, allocation to RE should be reduced in favorof vegetative structures that may enhance the prospects forresource acquisition. While there is evidence to support thisidea (e.g., Snell and Burch 1975; Pitelka et al. 1980; Bell andQuinn 1987), other studies report either no effect or an effectin the direction opposite to that predicted (see Evenson 1983).A second hypothesis is that variation in RE reflects differencesin the stability or predictability of the environment (MacArthurand Wilson 1967; Gadgil and Solbrig 1972). In an unpre-dictable environment, density-independent mortality is highrelative to that caused by density-dependent factors, and selec-tion should favor a large investment in RE to ensure successfulcolonization of habitats that become temporarily available. Astable environment, in contrast, ought to be characterized byhigh density-dependent mortality. Therefore, successful or-ganisms will be those that invest proportionately more intovegetative growth at the expense of reproductive activities,thereby increasing their competitive ability. Numerous studieshave sought to test this hypothesis by examining patterns ofRE along a successional gradient where the severity of den-sity-dependent mortality presumably increases in relation tosuccessional stage (Abrahamson and Gadgil 1973). Evenson(1983), for example, listed 15 studies whose results providedat least some support for this prediction (see also Hancock andPritts 1987; Bazzaz and Ackerly 1992); only one study showedno effect of succession on RE. A third hypothesis concernspatterns of RE due to life-historical factors. One application ofthis idea addresses differences in RE between semelparous(annual) and iteroparous (usually perennial) plant species, theexpectation being greater RE in the former group (see Hancockand Pritts 1987; Stearns 1992; Bazzaz and Ackerly 1992). An-

other consideration, however, is differences in the lifetimescheduling of RE as determined by mortality risks that are agespecific (see Stearns 1992, for further details). Willson (1983)reviewed studies of reproductive investment in plants from alife-history perspective. She concluded that concordance withlife-history predictions was variable and generally not upheldfor comparisons within species.

While each of these hypotheses have at least some support-ing evidence, few studies have actually evaluated these ideasdirectly. Here we report on RE in two populations of the com-mon dandelion (Taraxacum officinale Weber) and comparethese results with predictions derived from the three hypothe-ses. One population consisted of individuals colonizing a 5-year-old alfalfa (Medicago spp.) field. The density ofestablished plants (alfalfa plus dandelions) on this site waslow, and there was an abundance of open spaces available fornew dandelion plants to colonize, particularly between therows of alfalfa. Annual harvesting activities also removedmuch of the biomass from the field thus permitting consider-able light penetration into these open areas. The second samplewas drawn from populations that had colonized a number ofundisturbed habitats (areas that were not presently subject tohuman activities). In this habitat, dandelions were found inter-spersed among tall and very dense populations of grasseswhere few open spaces were available for seedlings to becomeestablished.

Methods

Study areaThe study took place in a large valley near Creston, B.C., Canada(49°14′N, 116°38′W) in May and June 1989. The valley runs in anorth–south direction, is approximately 8 km wide at the location ofthe study area, and is bisected by a large river. Prior to the 1950s, itwas subject to large-scale spring flooding and soils throughout theregion are rich in organic content.

Of the two groups of dandelions selected for comparison, the firstoccupied a 32-ha agricultural field on the floor of the Creston Valleythat had been seeded to alfalfa in 1985, at a rate of 1.8 kg/ha. Acompanion crop of oats (Avena sativa) was also sown into the field inthat first year. Herbicides were applied after the initial sowing but notthereafter, and this field had not been subject to fertilizer applicationfor at least 8 years prior to the present study. Alfalfa is an importanteconomic crop in the area, and most fields can sustain three cuts in anaverage growing season. Though these harvesting events are usuallyfollowed by short bursts of flowering activity by the dandelions, thevast majority of dandelions reproduce only once per season, prior tothe first cut (C.V.J. Welham, unpublished data). The results reportedhere are confined to these individuals.

The second group of dandelions was selected from 18 separatesites that were not presently subject to agricultural activities (hereafterreferred to as undisturbed sites). These sites were grassy fields alsosituated on the valley floor and to the west of the alfalfa field. Fivewere located on gentle slopes along the bottom edge of the valley.

Welham and Setter 167

© 1998 NRC Canada

B97-176.CHPFri Mar 13 11:37:41 1998


This east–west orientation of the two types of habitat was consideredimportant because wind direction in the valley is predominantlynorth–south, and so seed dispersal between the agricultural and undis-turbed sites was likely rare. However, their close proximity meant thatall sites were subject to similar patterns of rainfall and sunshine. Al-though the undisturbed sites had once been in agricultural production(at least 20 years prior to this study), they now constitute part of theCreston Valley Wildlife Management Area. In this respect, each sitewas originally fertilized and seeded to grasses in an effort to provideforage for wild ungulates. These grasses (Agropyron repens, A. spi-catum, Bromus tectorum, and Hordeum jubatum) are the dominantherbage, forming a thick, uniform stand across each field. There waslittle evidence of herbivore activity in these sites, however, probablydue to their proximity to human activities and dwellings.

Plant selectionRandomly selected pairs of coordinates were used to locate individualdandelion plants, which were then marked with a small plastic stake.Any given pair of dandelions was separated by a minimum of 8 m(usually much more) to minimize the chance of including two relatedindividuals. Of the 40 plants identified using this protocol, 8 of thesmallest plants did not produce flowers during the study and, there-fore, were eliminated from subsequent analysis. Hence, a total of32 plants were used from the alfalfa field. In a few cases, very largedandelions were selected from the vicinity of a coordinate pair(though they may not have been the closest plants) to maximize thesize range of the sample. Our results, therefore, do not provide a truerepresentation of the frequency distribution of plant size within thispopulation.

All plants were checked daily for flowering and to determine thestage of seed development in the heads. Just prior to development ofthe seed plumes, heads were tied off with small pieces of cheesecloth.The cheesecloth was large enough to prevent seed dispersal but stillallow the head to open normally. Each flower was harvested as itmatured by clipping the scape at its proximal end (i.e., closest to thebody of the plant). The flower head was then separated from thescape, and each stored in a paper bag for subsequent analysis (seebelow). Plants that had not produced any new heads for 14 consecu-tive days were assumed to have stopped reproducing. These were thendug up, cleaned with a damp cloth, and stored in a paper bag.

From a total of 40 plants marked on 18 undisturbed sites, 29 wereselected for subsequent analysis (a pair from each of 13 sites andsingle plants from the remainder) using the same criteria as describedfor alfalfa-field dandelions. In two other cases, plants were damaged

while attaching cheesecloth, and these were eliminated from theanalysis. The same harvesting protocol was used as for dandelions

Fig. 1. Reproductive effort (RE) in relation to vegetative biomass(root + leaf mass) for dandelions colonizing an alfalfa field (opensquares; N = 32) and those occupying undisturbed sites (solidcircles; N = 29). (A) Total RE (total scape + seed mass). Theregression equations are y = –1.58 + 1.10x, r2 = 0.92, P < 0.001,and y = 1.10 + 0.36x, r2 = 0.66, P < 0.001, for the alfalfa andundisturbed sites, respectively. (B) Proportional RE (total RE pervegetative biomass). The regression equations are y = 0.16 + 0.13x

–0.005x2, r2 = 0.63, P < 0.001, and y = 0.73 – 0.02x, r2 = 0.17, P =0.025, for the alfalfa and undisturbed sites, respectively. (C) Meanseed mass. There was no significant difference between the twopopulations (see text), and so the regression equation describestheir combined relationship (broken line): y = 0.46 + 0.02x, r2 =0.57, P < 0.001, N = 61. (D) Seed mass per pappus area. Each pointis a mean derived from a sample of 25 seeds per plant. Equation ofthe line is y = 3.70 + 0.11x, r2 = 0.54, P < 0.001, and applies to theundisturbed sites only. The regression for the alfalfa field was notsignificant (y = 4.41 – 0.02x, r2 = 0.03, P = 0.55).

Can. J. Bot. Vol. 76, 1998168

© 1998 NRC Canada

B97-176.CHPFri Mar 13 11:37:43 1998


from the alfalfa field. We chose dandelions from a variety of locationsbecause, although soil conditions and drainage patterns are consid-ered relatively uniform throughout the study area (W. Eastman, per-sonal communication), differences between the alfalfa field and anyone particular undisturbed site could have affected the analysis.

Seed analysis was conducted in a sealed area within the laboratory.We first counted the total number of seeds per plant and randomlyselected a sample of 25 seeds from all that were produced. We thenweighed each of these seeds and determined the maximum diameterof its pappus. The latter measurement was used to calculate the areacircumscribed by the pappus. A single seed was also selected at ran-dom from each plant. Each of these seeds was then dropped down thecenter of a wooden box that was enclosed on three of its four sides(dimensions 0.2 m depth × 0.2 m width × 3 m height). The interior ofthe box was coated with a commercial spray (Static Guard®) de-signed to eliminate static electricity. The use of a wooden box wasimportant because attempts to eliminate static electricity from othertypes of materials were unsuccessful. Each seed was required to travelunimpeded down the length of the box before the elapsed time overthe last 1-m distance was determined by stopwatch. Previous experi-mentation using a composite sample of 15 seeds (different from thosereported here) had demonstrated that this was the maximum height forwhich there was less than a 5% difference in 10 repeated drop timesfor each seed. Dividing these times by the 1-m distance gave an esti-mate of the maximum velocity for a given seed (thereafter referred toas the terminal velocity). These estimates of terminal velocity wereused as an index of the seed’s dispersal potential; the higher the ter-minal velocity, the lower the potential for dispersal.

Prior to statistical analysis, all plant parts were dried at 70°C for72 h. Intrapopulational trends were analyzed by regressing variousmeasures of RE against vegetative biomass. When necessary, logtransformations were used to normalize distributions. In all cases, wetested for both linear and polynomial effects though the latter areshown only when higher order effects are significant (P < 0.05). Theuse of vegetative rather than total biomass as the independent variableis in contrast to many previous studies of RE but is preferable statis-tically, since the latter measure can introduce artifactual autocorrela-tion into the regression analysis (Samson and Werk 1986).Differences in RE between populations were determined by multipleregression using vegetative biomass and habitat type as independentvariables, and testing for a significant interaction (see Sokal and Rohlf1981). In cases where the slopes did not differ significantly, we alsotested for equality of intercepts as determined by the significancevalue of the habitat variable in the multiple regression. We did notcompare overall (i.e., mean) differences in RE between the two popu-lations, since in cases of size dependency, this comparison is onlyvalid when each population has identical size distributions (Samsonand Werk 1986), a criterion that our data did not satisfy (data notshown).

Results

Intrapopulation effectsEach population showed a positive and highly significant lin-ear relationship between vegetative mass (root + leaf mass)

Fig. 2. Log(reproductive effort) in relation to log(vegetativebiomass) for dandelions colonizing an alfalfa field (open squares; N

= 32) and those occupying undisturbed sites (solid circles; N = 29).(A) Mean scape mass. There was no significant difference betweenthe two populations (see text) and so the regression equationdescribes their combined relationship (broken line): y = 0.58 +0.20x, r2 = 0.63, P < 0.001. (B) Mean scape length. The regressionequation for the alfalfa field dandelions is highly significant; y =23.32 + 2.43x, r2 = 0.66, P < 0.001. Regression for the undisturbedsites was not significant (y = 35.9 – 1.08x, r2 = 0.03, P = 0.572).(C) Total seeds per plant. The regression equations are highlysignificant: y = –28.4 + 54.1x, r2 = 0.77, P < 0.001, and y = 59.2 +25.9x, r2 = 0.58, P < 0.001, for the alfalfa and undisturbed sites,respectively.


© 1998 NRC Canada

B97-176.CHPFri Mar 13 11:37:45 1998


and total RE (total seed + scape masses; Fig. 1A). When ex-pressed as a proportion (seed + scape mass per vegetativemass), RE showed a significant linear decline on the undis-turbed sites (Fig. 1B). In contrast, RE increased significantlyin relation to vegetative mass on the alfalfa field and, in thiscase, the second-order polynomial term was significant(Fig. 1B). Both mean scape mass and mean scape length in-creased significantly in relation to vegetative biomass for thealfalfa field population, but only the relationship for meanscape mass was significant in plants from the undisturbed sites(Figs. 2A and 2B). The total number of seeds per plant(Fig. 2C) and mean seed mass (Fig. 1C) increased significantlywith vegetative biomass in both populations. Seed mass perpappus area was constant in alfalfa field dandelions but in-creased significantly with vegetative biomass on the undis-turbed sites (Fig. 1D).

Interpopulation differences in REThe range in vegetative biomass between the two populationswas very similar (Fig. 1). However, the slope for total RE wassignificantly higher for alfalfa field dandelions compared withthat from the undisturbed sites (P < 0.001; Fig. 1A). The sta-tistical relationships describing proportional RE differed be-tween the two populations. In the case of the alfalfa field,proportional RE increased nonlinearly with plant size but waslinear for the undisturbed sites (Fig. 1B), thus precluding anydirect statistical test for interpopulational differences. Never-theless, the differences between the populations were strikingwith larger dandelions on the alfalfa field generally allocatingproportionately greater biomass into reproduction, while pro-portional allocation declined with plant size on the undisturbedsites (Fig. 1B).

There was no significant difference between populations ineither the slopes or intercepts for mean scape mass (P = 0.349and P = 0.246, respectively; Fig. 2A). However, the slopes formean scape length did differ significantly (P < 0.001). Thismeant that larger dandelions on the alfalfa field had longerscapes than their counterparts from the undisturbed sites(Fig. 2B).

The two populations differed significantly in total seed pro-duction per plant (P < 0.001). Relative to plants from the un-disturbed sites, alfalfa-field dandelions produced fewer totalseeds per plant when they were smaller, but this trend wasreversed for larger plants (Fig. 2C). One reason why largerdandelion plants from the alfalfa field produced relativelymore total seeds is because they tended to produce more headsper plant (Fig. 3). Neither the slopes nor the intercepts formean seed mass differed significantly between populations(P = 0.165 and P = 0.222, respectively), and there was a sig-nificant linear increase in mean seed mass for the combinedregression (Fig. 1C).

Terminal velocity (an index of seed dispersal potential; seeMethods) showed a significant increase with the ratio of seedmass per pappus area (Fig. 4). The ratio of seed mass per pap-pus area was significantly different between the two popula-tions (P < 0.001; Fig. 1D) suggesting important differences indispersal potential.

Discussion

Patterns of RE in plants have received considerable attention,

but there is, as yet, no clear understanding of their ecologicalsignificance. One factor contributing to this problem was thesubstantial variation in published estimates of RE (see Souleand Werner 1981; Evenson 1983; Willson 1983), which in ourcase, was more than sixfold in the alfalfa population and four-fold in the undisturbed sites (Fig. 1B). This variation has madeit difficult to establish, and thus interpret, trends in RE acrosspopulations or species. It appears, however, many plant popu-lations can be characterized by a linear relationship betweenreproductive and vegetative biomass (Thompson et al. 1991)

Fig. 3. Proportion of dandelions producing one or two heads (darkgrey bars) and more than three heads (light grey bars) in relation tovegetative biomass (root + leaf mass) for plants located inundisturbed sites and the alfalfa field. Number of plants in eachcategory are shown across the top.

Fig. 4. Terminal velocity of 61 dandelion seeds from the twopopulations (N = 32 and 29, for the alfalfa field and undisturbedsites, respectively) in relation to the ratio of seed mass per pappusarea. The regression equation is highly significant: y = 11.4 + 7.5x,r2 = 0.47, P < 0.001.

Can. J. Bot. Vol. 76, 1998170

© 1998 NRC Canada

B97-176.CHPFri Mar 13 11:37:48 1998


and that much of the variation in RE may be a function of thisintrinsic size dependency (Samson and Werk 1986; Weiner1988). Our results provide further support for this idea(Fig. 1A). Recently, Clauss and Aarssen (1994) have drawnattention to the distinction between size- and stage-dependentRE, the latter referring to the allocation of meristems betweenvegetative and reproductive modules as a function of environ-mental conditions and chronological age. They question thevalidity of analyses based solely upon size dependency andargue that the effect of size alone on RE is only meaningful ifplants are harvested upon completing development. Clearly,this will not be feasible for long-lived perennials (as was thecase here), and so, our harvest criterion was restricted to sizeeffects at a specific stage in the annual cycle (i.e., individualsthat had completed reproductive activities; see Clauss andAarssen 1994).

While size dependency may dictate patterns of RE withineach dandelion population, it does not explain why differencesoccur between the populations in the nature of this relation-ship. Given that total and proportional RE was generallygreater on the alfalfa field, one hypothesis is that this occurredbecause the level of available resources on this site was higher.It was apparent that light penetration to ground level wasgreater on the alfalfa field relative to the undisturbed sites.Nutrient levels could also have been higher on the agriculturalfield, since alfalfa fixes nitrogen (Goplen et al. 1980) and fer-tilizer application was more recent (see Methods). If resourceswere indeed more abundant on the alfalfa field then the growthrate of its resident dandelions should have been comparativelyhigher (e.g., see Ford 1985), and this could have resulted in anincrease in the minimum size at which reproduction was initi-ated (see Clauss and Aarssen 1994). There was, however, nodifference between the populations in size at first reproduction(see Fig. 1). Furthermore, proportional RE for the smallestdandelions on the undisturbed sites was equivalent to or higherthan comparably sized plants on the alfalfa field (Fig. 1B), aresult that would not be expected if RE was simply a functionof resource availability. We conclude, therefore, that differen-tial resource availability does not provide an explanation forthese results.

A second hypothesis is that patterns of RE resulted fromselection pressures imposed by mortality rates that varied withthe disturbance regime and its effect upon the density of neigh-bours. This can be tested by comparing RE among differentsuccessional stages ( Evenson 1983; Hancock and Pritts 1987),the prediction being that RE will decline with successionalstage (Abrahamson and Gadgil 1973). Given that RE was gen-erally lower in the undisturbed areas, these sites should possessthe characteristics of later succession (i.e., a more stable andcompetitive environment; see Introduction) compared with thealfalfa field (higher disturbance and lower neighbour density).We suggest the alfalfa field represents an early successionalenvironment, since it was subject to regular, large-scale distur-bances, and these constituted the major sources of dandelionmortality. Our marked dandelions had much of their above-ground biomass removed when the alfalfa was harvested andmany of these plants were also compacted (by the wheels ofthe harvester) or dislocated from the soil. A second and cata-strophic disturbance also occurs at intervals of approximately6–8 years when the alfalfa field is ploughed and reseeded(W. Eastman, personal communication), thus destroying all

vegetation. The nonagricultural sites, in contrast, had the char-acteristics of a later successional environment. These areas hadremained undisturbed for at least several decades, their lack ofspace offered little opportunity for further recruitment by dan-delions, and competition for resources with the dominantgrasses was likely severe. Other studies have compared RE indandelions occupying habitats analogous to those describedhere, and they report similar results. Molgaard (1977) foundthat when competing with high densities of grasses, dandelionsinvested proportionally less resources into reproduction thanplants growing in open, disturbed (by mowing) communities.Similarly, Solbrig and Simpson (1974) identified four differentbiotypes of dandelion in which the biotype occupying the mostdisturbed area (biotype A) invested more effort into reproduc-tion than the dominant biotype occupying the more competi-tive habitats (biotype D). Furthermore, in a series ofgreenhouse experiments (Solbrig and Simpson 1974, 1977),they were able to demonstrate that biotype D was a better com-petitor than biotype A in undisturbed plots but not when plantswere subject to removal of aboveground parts.

Although consideration of successional stage may explainoverall differences between the populations, an important ex-ception is the RE exhibited by those dandelions from undis-turbed sites but with the smallest vegetative mass. Contrary toprediction, proportional RE from these dandelions was equiva-lent to or, in some cases, considerably higher than that meas-ured for their counterparts from the alfalfa field. It is alsodifficult to explain from the previous hypothesis why therewere differences between populations in the shape of thecurves describing proportional RE (see Fig. 1B). A third hy-pothesis we consider then is that RE represents a strategy thatminimizes the impact of mortality rates that are size specific.If, for example, mortality in large, mature plants is high com-pared with the (smaller) juvenile or early reproductive stages,selection should favor a proportionally higher investment ineach reproductive episode by the larger individuals. The con-verse should occur when the mortality schedules are reversed(Willson 1983). It should be noted that predictions derivedfrom this life-historical approach are usually made with refer-ence to an organism’s chronological age rather than size (seeStearns 1992), but this information was unavailable for theplants used here and our observations suggested that mortalitywas as much a function of vegetative biomass as age (thoughthe two may have been correlated). Many dandelions from thealfalfa field incurred considerable damage during the harvest-ing process, but this was most often restricted to larger plants.Smaller plants usually avoided damage because the majorityof their aboveground biomass was located below the minimumheight of the harvesting equipment. From a life-history per-spective, the increase and essentially asymptotic relationshipin proportional RE exhibited by dandelions from the alfalfafield (Fig. 1B) can thus be interpreted as a response to mortal-ity rates that were positively related to plant size. If, in contrast,mortality was inversely related to plant size for the nonagri-cultural sites, this could explain why the smaller plants ofteninvested proportionately more into RE than either larger plantsfrom the same habitat, or equivalently sized plants from thealfalfa field. Though we did not estimate mortality for the dif-ferent habitats, other studies have analyzed size-related pat-terns of dandelion mortality for habitats that differ incompetitor density. Their results showed that whereas overall


© 1998 NRC Canada

B97-176.CHPFri Mar 13 11:37:50 1998


mortality rates were generally higher with increased density(Ford 1981a, 1985), this was particularly evident for seeds andseedlings when they must compete with dominant grasses. Forexample, Ford (1981a) estimated that the chances of seedlingestablishment were 23 times lower in sites with lush grassvegetation compared with more open sites, a result he ascribedto the inability of many seeds to penetrate the canopy and thusgerminate on the soil surface. Molgaard (1977) reported simi-lar findings and suggested that dandelion seedling estab-lishment is strongly inhibited when grass cover is densebecause of insufficient open ground and light penetration.

Aside from reproductive effort, there were important dif-ferences between the populations in reproductive morphology.Larger plants on the alfalfa field had longer average scapelengths, produced more flower heads (capitula) per plant, hadgreater seed production, and had a lower ratio of seed mass perpappus area; only mean scape and mean seed mass did notdiffer significantly between the two populations. Other dande-lion populations exhibit similar variation in reproductive ecol-ogy (Ford 1981b), either because of differences betweenpopulations or because a given population is comprised of amixture of biotypes (Solbrig and Simpson 1974; Molgaard1977; Ford 1981b, 1985). Furthermore, it appears the distur-bance regime associated with a given environment is importantin maintaining this diversity (Solbrig and Simpson 1974),which may provide an explanation for the variation in mor-phology reported here. In the case of the alfalfa field, the har-vesting process regularly removed much of the abovegroundbiomass and thus provided ample opportunities for successfulcolonization by dandelions. Therefore, those clones depositingthe most seed onto the field were likely to experience the high-est recruitment rates and, after the field was sown into a newalfalfa crop, to maximize the prospects for re-establishment.The phenology of alfalfa itself may also have been importantin shaping scape and seed morphology, in the following way.One important feature of alfalfa is its ability to initiate andmaintain growth early in the season (Goplen et al. 1980) withthe result that most dandelions were overgrown by the alfalfabefore they reached the flowering stage. Hence, if air currentswere to be effective as a mechanism for seed dispersal thenscapes would have to be long enough that they overtop thealfalfa (e.g., see Sheldon and Burrows 1973; Ford 1985;McEvoy and Cox 1987). That these plants also tended to pro-duce seeds with lower seed mass per pappus area ratios (com-pared with the nonagricultural dandelions) also suggests thatadequate dispersal had important fitness benefits. The lack ofspace in the undisturbed habitats, in contrast, offered little op-portunity for recruitment by dandelions, and in establishedplants, competition for resources with the dominant grasseswas likely severe. This environment is likely to have favoredclones that accumulated a large vegetative biomass by invest-ing relatively few resources into each reproductive episode (asreflected in seed production, for example; see above). Moreimportantly, at the time of dandelion flowering, the grasses hadgrown to heights well above those observed in the alfalfa andbeyond the maximum scape length attainable by these dande-lions. This precluded any potential for dandelions to alter thedispersal potential of their seeds through changes in scape orseed morphology.

Our study revealed striking differences between these twodandelion populations in both their reproductive effort and

morphology. It is evident, however, that no single hypothesiscan provide an adequate explanation for these results. Com-parison of each habitat with respect to disturbance regime(and, by implication, its successional stage), for example,could account for many of the overall differences, but thisapproach did not specify how reproductive allocation within apopulation might change in relation to plant size. On the con-trary, size-dependent reproductive tactics were readily inter-pretable within a life-historical context. This suggests that bothparadigms were important for interpreting the patterns of REreported here.

Acknowledgements

We thank Larry Dill, Don Hugie, Roy Turkington, Ron Yden-berg, and two anonymous reviewers for comments on an ear-lier version of the manuscript. Patti Chen and Craig Olsonwere valuable field assistants. Bob Drinkwater kindly supplieddata on alfalfa production in the Creston Valley. We are grate-ful to the farmers in the area (particularly Walter Eastman) fortheir willingness to participate in discussion and permission towork on their land. This work was supported by a NaturalSciences and Engineering Research Council (NSERC) operat-ing grant to R.C. Ydenberg and an NSERC postgraduate schol-arship to C.V.J.W.

References

Abrahamson, W.G., and Gadgil, M. 1973. Growth form and repro-ductive effort in goldenrods (Solidago, Compositae). Am. Nat.107: 651–661.

Bazzaz, F.A., and Ackerly, D.D. 1992. Reproductive allocation andreproductive effort in plants. In The ecology of regeneration inplant communities. Edited by M. Fenner. CAB International,Oxon, U.K. pp. 1–26.

Bell, T.J., and Quinn, J.A. 1987. Effects of soil moisture and lightintensity on the chasmogamous and cleistogamous components ofreproductive effort of Dichanthelium clandestinum populations.Can. J. Bot. 65: 2243–2249.

Clauss, M.J., and Aarssen, L. 1994. Patterns of reproductive effort inArabidopsis thaliana: confounding effects of size and develop-ment stage. Écoscience, 1: 153–159.

Cody, M.L. 1966. A general theory of clutch size. Evolution,20: 174–184.

Evenson, W.E. 1983. Experimental studies of reproductive energyallocation in plants. In Handbook of experimental population bi-ology. Edited by C.E. Jones and R.J. Little. Van Nostrand Rein-hold Co., New York. pp. 249–275.

Ford, H. 1981a. The demography of three populations of dandelion.Biol. J. Linn. Soc. 15: 1–11.

Ford, H. 1981b. Competitive relationships amongst apomictic dande-lions. Biol. J. Linn. Soc. 15: 355–368.

Ford, H. 1985. Life history strategies in two coexisting agamospeciesof dandelion. Biol. J. Linn. Soc. 25: 169–186.

Gadgil, M., and Bossert, W.H. 1970. Life historical consequences ofnatural selection. Am. Nat. 104: 1–24.

Gadgil, M., and Solbrig, O.T. 1972. The concept of r- and K-selec-tion: evidence from wild flowers and some theoretical considera-tions. Am. Nat. 106: 14–31.

Goplen, B.P., Baenziger, H., Bailey, L.D., Gross, A.T.H., Hanna,M.R., Michaud, R., Richards, K.W., and Waddington, J. 1980.Growing and managing alfalfa in Canada. Agric. Can. Rep.No. 1705.

Hancock, J.F., and Pritts, M.P. 1987. Does reproductive effort vary

Can. J. Bot. Vol. 76, 1998172

© 1998 NRC Canada

B97-176.CHPFri Mar 13 11:37:52 1998


across different life forms and seral environments? A review of theliterature. Bull. Torrey Bot. Club, 114: 53–59.

MacArthur, R.H., and Wilson, E. 1967. The theory of island bio-geography. Princeton University Press, Princeton, N.J.

McEvoy, P.B., and Cox, C.S. 1987. Wind dispersal distances in di-morphic achenes of ragwort, Senecio jacobaea. Ecology, 68:2006–2015.

Molgaard, P. 1977. Competitive effect of grass on establishment andperformance of Taraxacum officinale. Oikos, 29: 376–382.

Pianka, E.R. 1972. r and K selection or b and d selection? Am. Nat.106: 581–588.

Pitelka, L.F., Stanton, D.S., and Peckenham, M.O. 1980. Effects oflight and density on resource allocation in a forest herb, Asteracuminatus (Compositae). Am. J. Bot. 67: 942–948.

Samson, D.A., and Werk, K.S. 1986. Size-dependent effects in theanalysis of reproductive effort in plants. Am. Nat. 127: 667–680.

Sheldon, J.C., and Burrows, F.M. 1973. The dispersal effectivenessof the achene–pappus units of selected Compositae in steadywinds with convection. New Phytol. 72: 665–675.

Snell, T.W., and Burch, D.G. 1975. The effects of density on resourcepartitioning in Chamaesyce hirta (Euphorbiaceae). Ecology,56: 742–746.

Sokal, R.R., and Rohlf, F.J. 1981. Biometry. W.H. Freeman, NewYork.

Solbrig, O.T., and Simpson, B.B. 1974. Components of regulation ofa population of dandelions in Michigan. J. Ecol. 62: 473–486.

Solbrig, O.T., and Simpson, B.B. 1977. A garden experiment on com-petition between biotypes of the common dandelion (Taraxacumofficinale). J. Ecol. 65: 427–430.

Soule, J.D., and Werner, P.A. 1981. Patterns of resource allocation inplants, with special reference to Potentilla recta L. Bull. TorreyBot. Club, 108: 311–319.

Stearns, S.C. 1992. The evolution of life histories. Oxford UniversityPress, Oxford, U.K.

Thompson, B.K., Weiner, J., and Warwick, S.I. 1991. Size-dependentreproductive output in agricultural weeds. Can. J. Bot. 69: 442–446.

Thompson, K., and Stewart, A.J.A. 1981. The measurement andmeaning of reproductive effort in plants. Am. Nat. 117: 205–211.

Weiner, J. 1988. The influence of competition on plant reproduction.In Plant reproductive ecology: patterns and strategies. Edited byJ. Lovett Doust and L. Lovett Doust. Oxford University Press,New York. pp. 228–245.

Willson, M.F. 1983. Plant reproductive ecology. John Wiley, NewYork.


© 1998 NRC Canada

B97-176.CHPFri Mar 13 11:37:52 1998


comparison of size-dependent reproductive effort in two dandelion ( taraxacum...

Documents