Life history strategies and biomass allocation the population dynamics of perennial plants in a regional perspective

Eelke Jongejans Promotiecommissie

Promotoren

Prof. dr. F. Berendse Wageningen Universiteit

Prof. dr. J.C.J.M. de Kroon Radboud Universiteit Nijmegen

Overige leden

Prof. dr. J. Ågren Uppsala Universiteit, Zweden

Prof. dr. E. van der Meijden Universiteit Leiden

Prof. dr. P.H. van Tienderen Universiteit van Amsterdam

Prof. dr. ir. P.C. Struik Wageningen Universiteit

Dit onderzoek is uitgevoerd binnen de onderzoeksschool SENSE. Life history strategies and biomass allocation the population dynamics of perennial plants in a regional perspective

Eelke Jongejans

Proefschrift ter verkrijging van de graad van doctor op gezag van de rector magnificus van Wageningen Universiteit, Prof. dr. ir. L. Speelman, in het openbaar te verdedigen op vrijdag 3 december 2004 des namiddags te vier uur in de Aula. Jongejans, E. (2004) Life history strategies and biomass allocation the population dynamics of perennial plants in a regional perspective

Ph.D.-thesis, Nature Conservation and Plant Ecology group, Department of Environmental Science, Wageningen University, The Netherlands, with summaries in English and Dutch

ISBN 90-8504-101-5 Contents

1. General introduction 1

2. Flexible life history responses to flower and rosette bud removal 11 in three perennial herbs Hartemink N, Jongejans E & de Kroon H

3. Size-dependent and -independent plant reproductive allocation 27 in response to successional replacement Jongejans E, de Kroon H & Berendse F

4. Bottlenecks and spatiotemporal variation in the sexual reproduction 49 pathway of perennials in meadows Jongejans E, Soons MB & de Kroon H

5. Space versus time variation in the population dynamics of Succisa 65 pratensis and two accompanying species Jongejans E & de Kroon H

6. Inbreeding depression significantly influences population dynamics: 89 a population-based model for a long-lived perennial herb Picó FX, Jongejans E, Quintana-Ascencio PF & de Kroon H

7. Effects of increased productivity on the population dynamics of the 103 endagered, clonal herb Cirsium dissectum Jongejans E, de Vere N & de Kroon H

8. A hierarchical population matrix analysis of the effect of nutrient enrichment: a synthesizing discussion 125 Jongejans E, Jorritsma-Wienk LD & de Kroon H

Summary 145

Samenvatting 149

Nawoord 155

Affiliations of the co-authors 157

Curriculum vitae 159 1 General introduction

Plant populations in a changing landscape

Human impact on ecosystems threatens the survival of many species world-wide: their habitats are destroyed or degraded and become more fragmented (Tilman et al. 1994; Hanski & Ovaskainen 2000). Habitat is lost to agriculture, infrastructure and urban areas to satisfy increasing human demands. The remaining habitat patches are often small and isolated, which increases the vulnerability to negative influences from surrounding areas: disturbance, pollution through air or water, changes in water level regime, and, especially in areas of intensive agriculture, influx of nutrients (Saunders et al. 1991; Roem & Berendse 2000). Nutrient enrichment increases the productivity of nutrient-poor ecosystems and more competitive species take over from species that are characteristic for these ecosystems (de Kroon & Bobbink 1997). Besides, species richness can be expected to decrease with productivity in areas that currently receive high levels of nitrogen deposition such as the Netherlands (van Oene et al. 1999). Large population sizes usually buffer environmental, genetic and demographic stochasticity, but the risk of extinction caused by these sources of year-to-year variation increases when populations are reduced to low numbers of individuals by habitat fragmentation and environmental factors (Menges 1998; Oostermeijer 2003). Small populations contain less genetic variation that may be necessary to adapt to changing environmental conditions (Ellstrand & Elam 1993). Demographic stochasticity in population dynamics is for instance the chance event of simultaneously a high death rate and a low establishment rate in one year. Besides, small populations may perform less just because they are small (e.g. because they fail to effectively attract pollinators), which is called the Allee-effect (Fischer & Matthies 1998; Colas et al. 2001), or because inbreeding starts to affect the performance of individuals. The increasing distances between populations amplifies these problems, as they diminish the probabilities of arrival of genetically-different seeds or pollen in areas with populations that are decreasing in size or are already extinct. Genetically impoverished populations are therefore unlikely to survive. Metapopulation theory predicts that the number of occupied habitats declines with habitat size and habitat connectivity (MacArthur & Wilson 1967; Ouborg 1993). Climate change increases the importance of connectivity as species have to disperse to keep up with possible change in the spatial distribution of climates they are adapted to (Thomas et al. 2004). Since most nations engaged the obligation to protect our plant species at the Rio convention on biological diversity (Myers 1993), the above-mentioned issues urge to answer the important question: can plant species survive in highly-fragmented landscapes such as we find in the Netherlands? 2. Chapter 1 General introduction abundance inthe19 1996). These moist grasslandswereformerlyusedashaymeadowsandincreasedin communities Pleistocene soilareasoftheNetherlands (Soons2003),whichincludetheplant The studiedhabitats wererestrictedtonutrient-poor grasslandfragments ofthe Study system:herbspeciesofnutrient-poor, species-richmeadows in changinglandscapes’. project at Alterra (Wageningen UR)was‘Integrationandapplication:regionalsurvival aim toalleviatetheisolationofremnantpopulations. The titleofhispost-doctoral study theeffectiveness ofecologicalcorridorsandagri-environmental schemes,which populations, whichareoften smallandisolated. genetic effects ofhabitat fragmentation ontheperformanceofremaining of smallisolatedpopulationsplants inaregionalcontext’.Sheinvestigatedthe ‘Inbreeding andoutbreeding:effects ofgeneflowandlocaladaptation onthesurvival history components oftheseplants inexperiments andpopulationmodels. dynamics ofnaturalplantpopulations,andtheimpact ofnutrientenrichmentonlife resulted inthePhDthesisfrontofyou. At Wageningen UniversityIstudiedthe isthetitleofsecondproject,that perennial plants inaregionalperspective’ realistic landscapes. dispersal ofseedsbetweenhabitat patches andthecolonizationabilitiesofplants in process inplants’. Herresearchfocusedontheeffects ofhabitat fragmentation onthe fragmentation andconnectivity. Spatial andtemporalcharacteristicsofthecolonization PhDstudents involvedandbyapost-doctoralmodellingproject. by the set ofmodelspeciesrespectively. The projects wereintegratedbyclosecooperation were started onthelandscapeecology, localpopulationdynamics,andgeneticsofa topic inthesamestudysystem:nutrient-poor, species-richmeadows. Three projects plant ecologists fromseveralinstitutesworkedtogetherondifferent aspects ofthe fragmented landscapes’.Inthisprogram,whichwaschairedbyJanvanGroenendael, Research (NWO-ALW) financedtheresearchprogram‘Survivalofplantspeciesin of plantspeciesinfragmentedlandscapes,theNetherlandsOrganizationforScientific To generatedata andknowledgeonthemechanisms thataffect thesurvivalchances The researchprogram‘Survivalofplantspeciesinfragmentedlandscapes’ cultivated (Buck-Sorlin 1993).Inthe20 remnants ofthesesemi-natural grasslandshavenowbecomenature reservesandare cultivation foragriculture bydrainageandfertilization(Buck-Sorlin 1993).Mostofthe 99% ofthesehabitats was lost(Soons2003),probablyduetointensification of Felix Knauerusedmodelexercisesandtheresults of thesethreeprojects to At theUniversityofNijmegenCarolinMixperformedaPhDstudyentitled ‘Life historystrategiesandbiomassallocation. The populations dynamicsof At UtrechtUniversityMerelSoonsfinishedherPhDthesisin2003:‘Habitat Molinietalia th century whenanincreasingareaofpeatbogs andfenswas and Caricetalia th century (especiallythe firsthalf)morethan (Schaminée etal . 1995; Schaminéeetal . 3. Chapter 1 General introduction ), and Figure 1. cl.v L. plants Moench. (through ), small vegetative rosettes ), small vegetative flow flow cl.f sdl Hypochaeris radicata ), vegetative clonal offspring ( clonal offspring ), vegetative flow Succisa pratensis lrg cl.v with six stage classes: seedlings ( with six stage ), flowering rosettes ( lrg ). The arcs denote the contribution probabilities of plants of one stage denote the contribution probabilities of plants The arcs ). cl.f sml Cirsium dissectum As model species we chose four species that occur in these meadows and that in these meadows that occur four species species we chose As model ), large vegetative rosettes ( ), large vegetative sdl sml ( are phylogenetically closely related (all belong to the families Asteraceae or Asteraceae belong to the families closely related (all are phylogenetically (i.e. sexual reproduction, the same life history options They have Dipsacaceae). life history they have distinctly different cf. Fig. 1), but propagation; survival and clonal and clonality: in longevity they differ strategies, i.e. managed by continuing the old land use practices: mowing and hay removal, in order mowing and hay use practices: the old land by continuing managed of disrupting effect and 2) the and trees, of shrubs 1) the encroachment to counter 1999). & Berendse enrichment (Bakker nutrient flowering clonal offspring ( flowering clonal offspring The life cycle of normally live only for a few years, whereas for a few years, whereas normally live only class to a particular stage class one year later. Transitions can be grouped as survival (solid lines), Transitions class one year later. stage class to a particular are rosettes The different (dotted lines). sexual reproduction (interrupted lines), or clonal propagation The drawings were made by Lidewij Keser. not depicted on the same scale of size. 4. Chapter 1 General introduction and but alsoindispersalability(Ehrlén&vanGroenendael1998). The seedsof dissectum grouped assurvival,sexualreproductionandclonalpropagation (thelifecycleof cycle ofperennialplantspeciesconsists ofseveralalternativepathways whichcanbe conditions, unfavorableseasons,competitionandherbivory(Grime1979). The life 1992). Important factorstowhichspeciescanadaptareforinstance harshabiotic defined asthecomplexadaptation ofallaspects ofthelifecycleaspecies(Stearns canbe propagation) canpotentiallysurvivefordecades. The term‘lifehistorystrategy’ rosette survival), dispersal bywind.Besides, (Soons &Heil2002;Mixetal above-defined habitats, whereas to nutrientenrichment andsuccessionofthevegetation (Abrahamson 1980). components, arealsohypothesized torespondchangesingrowingconditions due allocation patterns, i.e. the relativeinvestmentofresourcesindifferent lifehistory and Heil(2002)reported higherseedproductioninmorenutrient-rich sites. The 1990). resources andtheactivationofmeristemsis often time-limited(Watson 1984;Geber manipulations needstobeconsideredinsuch experiments, asthereallocationof (Stearns 1992;Ehrlén1999). The flexibilityofplants intheirresponsesto component canshowanegativeimpact of one componentonothercomponents exist. Manipulativeexperiments thatprohibit investment inaparticular lifehistory fact thatcosts oflifehistorycomponents arenot apparent doesnotmeantheydo plants havemoretospendonallcomponents compared to small plants (Fig.2). The dependent onplantsize(vanNoordwijk&deJong1986;Reznicketal trade-offs. Butthemostimportant oneisthatmostlifehistorycomponents are Watson 1989)orbecausehighvariabilityingrowingconditionsbetweenplants masks because floweringstemsbearinggreenleavescanbeselfsupporting(Benner& explanations whyoften positiveinsteadofnegativecorrelationsareobserved: In naturalconditionshowever, trade-offs arenotreadilyfound. There areseveral between lifehistorycomponents canbeexpectedbased onthistheory(Stearns 1992). for othercomponents, e.g.growthorclonalpropagation. Physiologicaltrade-offs invested inonecomponent,e.g.sexualreproduction,aretheorynolongeravailable history ofspeciesshouldalsoconsidertheirinterrelations.Resourcesthatare Since lifehistorycomponents arenotindependentofeachother, researchonthelife Life historycomponents, trade-offsandtheeffectofnutrientenrichment populations arethereforelargerinthefirsttwospecies(Soons2003). productive ormoredisturbedareas(vanderMeijden1996). The distances between C. dissectum The modelspecieswerealsoselectedbecausetheynotonlydiffer inlongevity Effects ofnutrient enrichment havebeenfoundinthemodelspecies:Soons is givenasanexampleinFig.1). have pappus, whichenablesthemtobecarriedfurtherbywind Centaurea jacea C. dissectum . 2003). The othertwospecieshavenoadaptation for C. jacea L. and and S. pratensis and Cirsium dissectum H. radicata are mostlyrestrictedtothe also occurinmore (L.) Hill(clonal . 2000): large H. radicata C. 5. Chapter 1 General introduction Figure 2. p e + p e Vegetative biomass - a) b) c) d)

Hypothesized allocationresponses succession to Negative or positive correlations between plant traits Sexual reproductive biomass reproductive Sexual Specifically, plants are thought to have two possibilities when the productivity of a the productivity when have two possibilities thought to are plants Specifically, to dispersal, or 2) through seed other sites 1) to escape to increases: vegetation 1980). Abrahamson 1974; position (Ogden their competitive by reinforcing persist of respectively expectations have clear hypotheses these long-standing Although never been to seed production, it has decreased resource allocation increased and in experiments. explicitly tested A the hypothesis that the type of used in this dissertation: allocation theories summary of two important depends on the variation in plant size and the variation in allocation, and correlation between plant traits solid of allocation patterns the hypothesized plant responses to succession. In these schematic models invest in either sexual reproductive biomass (flowers lines symbolize the proportion of biomass plants The upper diagrams show and stems). leaves biomass (roots, and seeds) as a function of vegetative biomass depends on variation in how the observed correlation between the sexual and vegetative between dotted lines) (van Noordwijk & de variation in plant size (distance and on allocation pattern little, but variation in allocation Jong 1986): a) negative correlations are found when plant size varies less between plants differ is large, b) positive correlations are found when allocation patterns patterns The bottom diagrams show how the hypothesized escape (e) and than the variation in their sizes. Sugiyama and Bazzaz persistence (p) strategies (Abrahamson 1980) alter the allocation patterns. size or d) by to vegetative (1998) show that sexual reproductive allocation can be changed c) relatively plant size. an absolute amount independent of vegetative 6. Chapter 1 General introduction (B), 8m determines extinction risksofsmallpopulations,canbequantified whenmultiple provide awealthofinformation. Demographicstochasticity, one ofthefactorsthat from yeartoand therateatwhichseedlingsandclonaloffspring establish, can them. Simpledemographic observations,i.e.monitoringthefate of individualplants assess theimportance ofparticular lifehistorycomponents, andofthefactorsaffecting Detailed data onthelifecycleofplantspeciesunder naturalconditionsisneededto Population dynamicsofperennialplantspecies for eachmeadowwas6m study onthefourspeciesatleft top(studiedspecies-sitecombinationsinbold). The areaofallplots Presence andabundanceofhigherplantspeciesinthepermanentplots usedforthedemographic Table 1. = 25-50%cover;450-75%575-100%cover. 6-50 specimens;2m=>50specimens,butcoverisstill<5%;2a5-12%cover;2b13-25%3 present isgiven.Braun-Blanquetabundancycodesareused:r=1specimen;+2-5specimens; Meijden (1996).Foreachspeciespresentinasitethemedianabundanceofplots inwhichitis and V2).Species lists weremadeinJulyand August, 1999and2000.Nomenclaturefollowsvander Euphrasia stricta tetralix Erica Hypericum pulchrum Hypericum perforatum Holcus lanatus Hieracium laevigatum Gentiana pneumonanthe Filipendula ulmaria Festuca pulicaris Carex Betula pendula Anthoxanthum odoratum Agrostis Species Equisetum palustre Equisetum arvense Carex panicea Carex oederi nigraCarex flacca Carex Calluna vulgaris Betula pubescens Drosera rotundifolia Danthonia decumbens maculata Dactylorhiza Cirsium palustre vulgaris Carlina Achillea millefolium Achillea Succisa pratensis radicata Hypochaeris Cirsium dissectum Centaurea jacea 2 p 12b12a2b 1 spp 2a111 spp in Leemputten(L),and32m +1 +r+ r +11 12b2a1 12b12a11 11 +r +1+1++ ++ x2a 2 1V pce 1V2 V1 L B O N Species V2 V1 L B O N 12a1 + 1 + 11 32a 3 111 in Konijnendijk(N),4m + + r + ++ 1 r 1 + 1 + 1 r r r r2a 2 respectively forthetwolocationsinVeerslootlanden (V1 2 Rumex acetosa Trichophorum cespitosum Thalictrum flavum officinale Taraxacum Sorbus aucuparia Sanguisorba officinalis salicaria Lythrum Salix repens Lysimachia vulgaris Rhinanthus angustifolius Rhamnus frangula Salix cinerea Plantago lanceolata sylvestris Pinus Phragmites australis sylvatica Pedicularis Parnassia palustris Viola arvensis serpyllifolia Polygala Molinia caerulea Ranunculus flammula Ranunculus acris Quercus robur vulgaris Prunella erecta Potentilla Peucedanum palustre in Koolmansdijk(O),5m Luzula multiflora Luzula campestris Linum catharticum Leucanthemum vulgare Leontodon autumnalis Juncus conglomeratus 2 in Bennekomsemeent b+2 a5+ 5 2a 2b + 2b a+2 + 1 + 2m + 2a ++ 1r +r +r +r+r r r r r ++ ++ +111 a1 2a +1 r + r a1 2a ++ ++ + + 1 r r r + +1 1 r r 7. Chapter 1 General introduction 1996). In general, 1996). . 1996; Menges & Dolan 1996; Menges . the costs of sexual reproduction and clonal the costs 2 1986). Elasticities allow comparison between very between allow comparison 1986). Elasticities . 1997; Caswell 2001). Moreover, the direct and indirect 1997; Caswell 2001). Moreover, . we test the hypothesis that individual plants can alter their we test the hypothesis that individual plants 3 deals with the variation in the sexual reproduction pathway within deals with the variation in the sexual reproduction pathway 4 Whether temporal variation can be substituted with spatial variation in the Whether temporal variation can be substituted with spatial Population matrix models are very suitable for analyzing demographic field data for analyzing demographic models are very suitable Population matrix In chapter Chapter dissimilar species or growing conditions due to their proportional nature (Silvertown et dissimilar species or growing contributed in life history components analyzing what differences al. 1993). However, in population growth rate requires variation decomposition to observed differences techniques (Horvitz et al Outline of this thesis and population dynamics of the four I study the allocation patterns In this dissertation and modeling of increased productivity in experiments model species and the effect chapter in exercises. Specifically, in the sexual try to answer the question which step and between species. We largest bottleneck from flowering adult to established is the reproduction pathway with the variability steps the survival probabilities in the different compare seedling. We as being steps these two analyses pinpoint different to see whether of those steps most important. 1998). Whether temporal and spatial variation in population dynamics are equal or not are equal or dynamics in population variation and spatial temporal 1998). Whether unknown as to good sites) is largely similarly respond to good years (whether plants investigation. and needs more classify plants These Lefkovitch matrices are used. mostly In plants (Caswell 2001). These than on age (Lefkovitch 1965). and their size rather stage on their phenological calculates the ways: elasticity analysis can be analyzed in different transition matrices growth rate the arcs in Fig.1) for the of each matrix element (i.e. relative importance (de Kroon et al of the population with hierarchical matrix models can be studied plant traits of underlying importance of those factors on the population the impact 2000). Issues concerning Tienderen (van of the natural the quantification be addressed after dynamics however can only factors on specific life history of external the effects population dynamics and of components. continuously removing flower buds and rosette buds. are investigated by propagation we study the flexibility of the allocation patterns, from the altered biomass Apart responses to bud removal in this one-season garden experiment. biomass and height of the surrounding vegetation when the allocation patterns but the previous chapter, increases. In the same garden experiment as reported in with the studied species and tussocks to plots now for three years, we added nutrients change their biomass allocation grass, and we investigate how plants of a tall plants. to unfertilized control compared survival becomes more important at the expense of growth and fecundity when growth and expense of at the important becomes more survival et al (Oostermeijer less favorable conditions are growing populations are followed over several years (Oostermeijer et al years (Oostermeijer over several are followed populations 8. Chapter 1 General introduction Ellstrand NC&ElamDR1993. Populationgeneticconsequencesofsmallpopulation size:implications Ehrlén J&vanGroenendael JM1998. The trade-off between dispersabilityandlongevity-animportant Ehrlén J1999.Modellingandmeasuringplantlifehistories. -In:Vuorisalo TO &MutikainenPK(eds), de KroonH,Plaisier A, van GroenendaelJ&CaswellH1986.Elasticity:therelativecontributionof de KroonH&BobbinkR1997.Clonalplantdominance underelevatednitrogendeposition,withspecial Colas B,OlivieriI&RibaM2001.Spatio-temporal variationofreproductivesuccessand conservation Caswell H2001.Matrixpopulationmodels.Construction, analysis,andinterpretation. -Sinauer Buck-Sorlin G1993. Ausbreitung undRückgangderEnglischenKratzdistel- 1989.Developmental ecologyofmayapple:seasonalpatterns ofresource &Watson MA Benner BL &BerendseF1999.Constraints intherestorationofecologicaldiversitygrasslandand Bakker JP (ed),Demography Abrahamson WG1980.Demographyandvegetative reproduction.-In:SolbrigOT References population dynamicsofperennialplantspeciesisthecentralquestionchapter years. whether thelifehistoriesofthesespeciesrespondsimilarlytobadsitesas 1) andusematrixprojectionmodelsvariationdecompositionanalysestoexamine We studiedthepopulationdynamicsofthreemodelspeciesinfivesites(Table population growthrateindifferent populations. variation inthesamelifehistorycomponents underliestemporalvariationinthe to quantifythetemporalvariationinpopulationgrowthrate,andseewhether populations overagradientofproductivity. We usedvariationdecompositionanalyses whether theallocationpatterns foundinthethirdchapterarealsonatural Cirsium dissectum pratensis dependency ofthepopulationdynamicsarestudiedinalong-livedherb population projectionmatrices. The impact ofinbreedingdepressionanddensity enrichment onplanttraits andlifehistorycomponents. effect onvariationinthepopulationgrowthratebyincorporating theeffects ofnutrient matrix modelstostudywhichlifehistorycomponents and planttraits havethelargest of thelifehistoriesstudiedspecies.We presentanexercise withhierarchical for plantconservation.- Annu RevEcolSyst24:217-242. aspect ofplantspeciesdiversity. - Appl Veg Sci1:29-36. Life historyevolutioninplants. Kluwer Academic Publishers,London. demographic parameters topopulation growthrate.-Ecology67:1427-1431. J (eds), The ecologyandevolutionofclonalplants. Backhuys Publishers, Leiden. reference to of thenarrow-endemic Associates, Inc.Publishers,Sunderland,Massachusetts. distribution insexualandvegetative rhizomesystems.-FunctEcol3:539-547. heathland communities.- Trends EcolEvol14:63-68. and evolutioninplantpopulations.BlackwellScientificPublishers,Oxford. - inNordwestdeutschland. - Tuexenia 13:183-191. In chapter In thefinalchapter, chapter Chapter with stochasticsimulationmodeling. 7 investigates thepopulationdynamicsofendangered,clonalherb Brachypodium pinnatum 6 , whichpropagates throughrhizomes.Special attentionispaid to measured levelsofinbreedingdepressionareincorporatedin Centaurea corymbosa 8 , wesummarizetheflexibilityandnaturalvariation in chalkgrassland.-In:deKroonH&vanGroenendael (Asteraceae). -BiolConserv99:375-386. Cirsium dissectum Succisa (L.) Hill 5 . 9. Chapter 1 General introduction Tussilago Gentianella in midwestern : negative genetic : negative Silene regia Polygonum arenastrum Polygonum , a rare perennial herb. - J Ecol 84: 153-166. , a rare perennial herb. - J Ecol 84: Gentiana pneumonanthe . - J Ecol 86: 195-204. . - J Ecol L. - J Ecol 62: 291-324. Population viability in plants. Springer-Verlag, Berlin. Springer-Verlag, Population viability in plants. - Oikos 66: 298-308. along the Dutch Rhine-system. approaches applied to vascular plants Ecol Evol 15: 421-425. Trends - grassland and heathland communities. - Biol Conserv 92: changes in plant species diversity in 151-161. - Conserv Biol 5: 18-32. a review. Press, Uppsala. Plantengemeenschappen van graslanden, zomen en droge heiden. - Opulus Opulus Press, Uppsala. Plantengemeenschappen van wateren, moerassen en natte heiden. - to the finite rate of increase in woody and herbaceous of life-cycle components importance perennials. - J Ecol 81: 465-476. - Utrecht University. colonization process in plants. dispersed grassland forbs. - J Ecol 90: 1033-1043. - Funct Ecol 12: 280-288. and genetic identity. competition availability, Miles AS, Midgley GF, L, Hughes L, Huntley B, van Jaarsveld A, Hannah Grainger Sigueira MF, OLA, Phillips Peterson Townsend MA, L, Ortega-Huerta & Williams SE 2004. Extinction risk from climate change. - Nature 427: 145-148. 65-66. germanica - Evolution 44: 799-819. and growth. between fecundity correlations 755-758. S & Caswell H Tuljapurkar - In: prospective and retrospective analyses. population growth: systems. Chapman & models in marine, terrestrial, and freshwater (eds), Structured-population York. Hall, New 1-18. Princeton, New Jersey. decade. Chapman & Hall, London. Conservation biology: for the coming geographic location, population with fire management, genetic variation, prairies: Relationships size and isolation. - J Ecol 86: 63-78. 25: 12-19. Technol - Seed quantifying fruit traits. farfara the demography of Oostermeijer JGB 2003. Threats to rare plant persistence. - In: Brigham CA to rare Threats Oostermeijer JGB 2003. & Schwartz MW (eds), The classical and metapopulation size and extinction: Ouborg NJ 1993. Isolation, population Reznick D, Nunney LA Tessier & of reproduction. 2000. Big houses, big cars, superfleas and the costs acidity and nutrient supply ratio as possible factors determining Roem WJ & Berendse F 2000. Soil CR 1991. Biological consequences of ecosystem fragmentation: Saunders DA, Hobbs RJ & Margules van Nederland. Deel 3. EJ 1996. De vegetatie AHF & Weeda Schaminée JHJ, Stortelder van Nederland. Deel 2. V 1995. De vegetatie EJ & Westhoff Schaminée JHJ, Weeda ASilvertown J, Franco M, Pisanty I & Mendoza plant demography: Relative 1993. Comparative and temporal characteristics of the Spatial and connectivity. fragmentation Soons MB 2003. Habitat in fragmented populations of wind- Soons MB & Heil GW 2002. Reduced colonization capacity of life histories. - Oxford University Press, Oxford. The evolution SC 1992. Stearns Sugiyama S & Bazzaz FA reproductive allocation: the influence of resource 1998. Size dependence of YC, Erasmus BFN, de A, Green RE, Bakkenes M, Beaumont LJ, Collingham Thomas CD, Cameron D, May RM, Lehman CLTilman & Nowak MA destruction and the extinction debt. - Nature 1994. Habitat Fischer M & Matthies D 1998. Effects of population size on performance in the rare plant in the rare size on performance of population Effects & Matthies D 1998. Fischer M Geber MA in limitation cost of meristem The 1990. Grime JP & Sons, Chichester. processes. - John Wiley Plant strategies and vegetation 1979. - Nature 404: of a fragmented landscape. capacity The metapopulation O 2000. Hanski I & Ovaskainen to of life-history stages The relative “importance” DW & Caswell H 1997. Horvitz C, Schemske Lefkovitch LP - Biometrics 21: organisms grouped by stages. The study of population growth in 1965. Press, - Princeton University The theory of island biogeography. Wilson EO 1967. MacArthur RH & risks in plant populations. - In: Fiedler PLMenges ES 1998. Evaluating extinction & Kareiva PM (eds), viability of populations of 1998. Demographic Menges ES & Dolan RW A Mix C, Picó FX & Ouborg NJ 2003. analysis for of stereomicroscope and image comparison extinction. - Biodiversity and Conservation 2: 2-17. Myers N 1993. Questions of mass The reproductive strategy of II. The reproductive strategy of higher plants. Ogden J 1974. variation in and spatial Temporal Boer ER & den Nijs HCM 1996. Oostermeijer JGB, Brugman ML, de 10. Chapter 1 General introduction asnM 1984.Developmental constraints: effect onpopulationgrowthandpatterns ofresource Watson MA van Tienderen PH2000.Elasticitiesandthelinkbetweendemographicevolutionarydynamics.- van OeneH,BerendseF&deKovelCGF1999.Modelanalysisoftheeffects ofhistoricCO van Noordwijk AJ &deJongG1986. Acquisition andallocationofresources:theirinfluenceonvariation floravanNederland.-Wolters-Noordhoff, Groningen. van derMeijdenR1996.Heukels’ allocation inaclonalplant.- Am Nat123:411-426. Ecology 81:666-679. nitrogen inputs onvegetation succession.-Ecol Appl 9:920-935. in lifehistorytactics. - Am Nat128:137-142. 2 levels and 2 Flexible life history responses to flower and rosette bud removal in three perennial herbs

Nienke Hartemink, Eelke Jongejans & Hans de Kroon 2004. Oikos 105:159-167.

Summary

In a garden experiment we investigated the response to continuous removal of either flower buds or rosette buds in three perennial grassland species (Hypochaeris radicata, Succisa pratensis and Centaurea jacea), which differ in longevity and flowering type. We distinguished two possible responses: compensation for lost buds by making more buds of the same type, and switching towards development of other life history functions. Both responses were demonstrated in our experiment, but bud removal had significantly different effects in each of the three species. The degree of compensation and the expression of trade-offs between life history functions differed markedly between species and seem related to longevity and developmental constraints. With respect to switching, our results suggest costs of reproduction and a trade- off between life history functions, at least for H. radicata and S. pratensis. For these species weight of new rosettes increased when resource allocation to flowering was inhibited. In H. radicata, we see that both compensation for lost flower buds and switching from lost rosette buds increased production of flower buds, underscoring the pivotal role of sexual reproduction in this short-lived species. The most prominent response seen in C. jacea is compensation for lost rosette buds, indicating that this long-lived species with monocarpic rosettes relies on rosette formation. Although S. pratensis does respond to bud removal, time is an important constraint in this species with long-lived rosettes and preformed flowering stalks. Trade-offs in S. pratensis seem to operate at a larger time scale, requiring long-lasting experiments to reveal them. We conclude that the response of these species to inflicted damage is likely to be linked to their longevity and developmental constraints.

Keywords: bud removal, Centaurea jacea, compensatory effects, costs of reproduction, garden experiment, herbivory, Hypochaeris radicata, meristem allocation, phenology, resource allocation, Succisa pratensis, switching 12. Chapter 2 Flexible responses to bud removal can eithertrytocompensate forthelossbyreplacinglostparts, orswitchtheir general, plants havetwo optionsofrespondingwhenplantparts areremoved; they development oftheremoved plantparts becomeavailableforother purposes.In Ehrlén 2002).Inbud removal experiments resourcesthatcannotbeused forthe relative importance oflifehistoryfunctions(Stearns 1989;Ehrlén1999;García& investigating theimportance oftrade-offs betweenlife-historyfunctions. limiting factoris,bothbiomassandmeristem allocationshouldbeconsideredwhen abortion offruits orseeds(Stöcklin 1997).Incaseitisunclearbeforehandwhat the over resourceflowsthatdeterminethesizes ofstructures,whichoften includesthe (Watson 1984;Duffy etal.1999). The second stepcomprisesthephysiologicalcontrol particular meristemintoavegetative orasexual structureortoremaindormant step process. The firststepinvolvesthedevelopmental decisiontodifferentiate a concomitantly. evidently important, stillfewstudieshavelookedatbiomassandmeristemallocation resources, e.g.Watson (1984)andGeber(1990). Although bothprocessesare reproduction arepaid inthecurrencyofmeristemsratherthanamountspent 1984; Olejniczak2001).Meristemallocationtheoryassumesthatthecosts of vegetative growthcanincreasethenumberofavailablemeristemslaterinlife(Watson meristems availableforvegetative growth.Conversely, dedicatingmeristemstowards & Aarssen 1996).Commitmentofmeristemstoreproductionreducesthenumber environments, meristemsmaybemorelimitingthanresources(Geber1990;Bonser strategies (Olejniczak2001),sinceithasbeenshownthat,especiallyinnutrient-rich also meristemallocationisregardedasanimportant featureininterpretingplant consequence trade-offs betweenfunctionsarenotnecessarily apparent. Nowadays, sources ofcarbon(Benner&Watson 1989;Vuorisalo &Mutikainen1999). As a that photosynthesizingplantparts donotonlyactasresourcesinksbutalso their abilitytobeself-sustaining. Inplantecologyithassincelongbeenrecognised function only, andthattrade-offs betweenthesefunctionscanthereforebeexpected. allocation theorystated thatresourcescanbeallocatedtowardsonelifehistory factors (Harper1967).Likeinanimallifehistorytheory, traditionalplantresource variation, lifespan, competitivedominance,modeofvegetative spreadandother species (van Andel &Vera 1977;Bazzaz1996),dependingonenvironmental allocation towardsthevariouslifehistoryfunctionsmaydiffer considerablybetween between floweringandseedset,productionofnewrosettes,growth.Resource this meansthatwithineachgrowingseasontheresourcesacquiredarepartitioned reproduction (Stearns 1989;Bazzaz&Grace1997).Inrosette-producingperennials the trade-off betweencurrentsexualreproductionandgrowth,survivalfuture Life historytheorypredicts thatorganismsoptimisetheirlifetimefitnessaccordingto Introduction Manipulating theeffort aplantmakestocertain functioncanrevealthe Prati andSchmid(2000)suggestedthatallocationinplants isessentiallyatwo- However, thisviewmaynotaccordwiththemodularcharacterofplantparts or 13. Chapter 2 Flexible responses to bud removal Figure 1.

, by Prati and a. The flower buds and rosette The flower a. RB FB and by Lehtilä and Syrjänen and by Lehtilä , and in the cantaloupe melon , and in the cantaloupe Helianthus tuberosus Centaurea jace Centaurea and RB FB Ranunculus reptans Gentianella campestris Succisa pratensis , species. In the latter experiment flower removal also led to latter experiment flower removal species. In the FB by El-Keblawy and Lovett Doust (1996). A (1996). by El-Keblawy and Lovett Doust between life natural trade-off Melampyrum Hypochaeris radicata Compensatory fruit production after main stem clipping was found by stem clipping was found main fruit production after Compensatory Switches in resources allocation after manipulation have been found by manipulation have after Switches in resources allocation RB Hypochaeris radicata Succisa pratensis Centaurea jacea Lennartsson et al. (1998) in et Lennartsson resource investments towards other functions. This choice probably depends on probably This choice functions. towards other investments resource etc. In the timing of damage, life span, the expected constraints, developmental for were found responses different so far, performed bud removal experiments various species. different (1995) in two compensatory seed production later in the season. Examples of overcompensation seed production later in the compensatory The release of apical 1995). (Watson as a result of herbivory are also known When behind overcompensation. to be one of the mechanisms dominance is assumed and start behind it will become activated has been removed, the apices the main apex can even be 1988) and the overall effect grow (Svensson & Callaghan to develop and an increase in flowering. the tuber-forming species (1993) in Westley Drawings of Juul Limpens RB respectively. buds that were removed in the experiment are indicated with FB and made the drawings. Schmid (2000) in the clonal Schmid (2000) in the clonal Cucumis melo 14. Chapter 2 Flexible responses to bud removal (Asteraceae), We performedagardenexperimentwiththreespecies: Species Materials andmethods radicata radicata become independentfromthemotherrosette.However, theydiffer inlongevity( as wellvegetatively bymeansofdevelopingnewrosettes,whicheventuallymay Succisa pratensis history functionsin allocation consistentwithatrade-off betweenlifehistoryfunctions? and siderosettebudremoval,respectively? Are changes in biomassandmeristem and towhatextentdotheyswitchotherfunctionsinresponseflowerbudremoval following questions: To whatextentdothesespeciescompensateforlostplantparts removal experimentwiththreeperennialgrasslandherbs( plant species. between compensationandswitchingtheexpressionoftrade-offs inherbaceous However, itremainsunclearwhichoftheseunderlyingfactorsdeterminethechoice damage (Lennartsson etal.1998)andplantarchitecture(Lehtilä&Syrjänen1995). damage, suchasmeristempreformation(Geberetal.1997),growthrate,timingof 2001). There areseveralconstraints toaplasticresponseofplantinflicted instance ifextragrowthalsoimpliesmoreflowerbudproduction(Avila-Sakar etal. following year. Insomecaseshowever, itishardtomakesuchadistinction,for would havebeenallocatedtogrowth,vegetative propagation orfloweringinthe They concludethatfloweringandfruitingdepletedstoredresourcesotherwise Fone 1989);its lifespan usuallydoesnotexceedtwoseasons. as wellsexualreproduction),buttheydiffer inlifespan andflowering type. grasslands (Soons&Heil2002),andhave similar lifehistorychoices(clonalgrowth (Dipsacaceae) (Fig.1). They arefoundincomparable habitats, rathernutrient-poor During andafter flowering,vegetative side-rosettesformontherootstock andappear years (Tamm 1956). shoots. Singleindividuals havebeenrecordedtosurviveuntilatleast theageoften until lateautumn(Fone1989). flowering stalks variesconsiderably(deKroonet al.1987).Floweringlasts fromspring stalks andnewrosettesareformedinthecentre ofthemainrosette;number In anattempttoincreaseourunderstanding ofthistopic,weperformedabud Hypochaeris radicata Centaurea jacea is short-lived,whereastheothertwoarelong-lived)andfloweringtype( and C. jacea Hypochaeris radicata and Tipularia discolor Centaurea jacea Centaurea jacea is arelativelylong-lived perennial,althoughithasmonocarpic flower apically, is arelativelyshort-livedspecies(deKroon etal.1987; has beenstudiedbySnowandWhigham(1989). L. (Asteraceae)and rosettes haveasingle apicalfloweringstalk. ). These threespeciesallreproducesexually S. pratensis laterally). We addressedthe Succisa pratensis Hypochaeris radicata Centaurea jacea H. radicata flowering Moench H. H. L. , 15. Chapter 2 Flexible responses to bud removal , Figure 2. Centaurea jacea Centaurea 300300 200200 100100 00 flower buds are on the right-hand were grown from seed in a were grown from Hypochaeris Hypochaeris Hypochaeris Hypochaeris H. radicata C. jacea and 2-8-002-8-00 2-9-00 2-9-00 3-10-00 3-10-00 2-8-002-8-00 2-9-00 2-9-00 3-10-00 3-10-00 flowers laterally and usually produces more produces laterally and usually flowers Succisa Succisa , can survive for many years (Adams 1955; years (Adams for many , can survive Succisa Succisa . Note that the Centaurea Centaurea Centaurea Centaurea C. jacea S. pratensis S. pratensis S. pratensis , like 55 00 55 00 1-5-001-5-00 1-6-00 1-6-00 2-7-00 2-7-00 1-5-001-5-00 1-6-00 1-6-00 2-7-00 2-7-00 Succisa pratensis Cumulative Number of Removed Rosette Buds per Plant Plant Rosette Rosette per per Buds Buds Number Number of Removed of Removed Cumulative Cumulative Cumulative Number of Removed Flower Buds per Plant Plant Flower Flower per per Buds Buds Number Number of Removed of Removed Cumulative Cumulative 1515 1010 1515 1010 and ± plants were grown from seed, germinating in March 2000. Plants were put were grown from seed, germinating in March 2000. Plants plants Succisa pratensis Succisa nature reserve in the vegetation in a Cirsio-Molinietum Seeds were collected than one flowering stalk. New rosettes also emerge laterally. This species, which is This emerge laterally. New rosettes also stalk. than one flowering al. et late July until October (Vergeer the Netherlands, flowers from rather rare in 2003). ‘Bennekomse Meent’ of a formerly This remnant 5°36’E). (52°01’N, near Wageningen August, it is not a year in the beginning of hayfield is mown once extensive communal Hoek and site description see van der herbivores. For a more detailed grazed by large of Plants Braakhekke (1998). at the soil surface alongside the main stem (personal observation). Flowering lasts Flowering observation). stem (personal the main surface alongside at the soil Meijden 1996). frost (van der until the first from June Hooftman & Diemer 2002). & Diemer 2002). Hooftman rosettes were cut from of the experiment. Side the start greenhouse one year before garden in May 2000. and transplanted into the experimental these one-year-old plants H. radicata Cumulative number of removed flower buds and rosette buds per plant of Cumulative number of removed flower buds and rosette buds per Hypochaeris radicata axis. Bars represent day. error of the number of removed buds per plant on that particular standard figure, but are included in Fig.3. Small rosette and flower buds found at harvest are not depicted in this 16. Chapter 2 Flexible responses to bud removal had ripened;harvestdatediffered betweenthespecies(8 by countingandweighingtheremovedparts. Harvesttookplacejustafter mostseeds each individual.Fortheremovaltreatments thiswasdonepartly duringtheexperiment neither flowerbudsnorrosettewereremoved. dried andweighedliketheflowerbuds. The thirdtreatmentwasacontrol,inwhich off after gentlyremovingthesoilaroundrootstock. Rosettebuds werecounted, forms side-rosettesontherootstock justbeneaththesurface, rosettebudswerecut removed apicallyandlaterallyfromthemainrosette,respectively. For and days andweighed. The secondtreatmentwasarosettebud removal.For experiment. The removedflowerbudswerecountedperplant, driedat70°Cfortwo removal wasappliedapproximatelythreetimeseachmonthuntiltheendof enough toberemovedwithoutdamagingtherestoffloweringstalk. Flowerbud photosynthetic capacity (Fig.1).Flowerbudswereremoved whentheywerelarge bud), sinceremovingthewholefloweringstalk wouldprobablyaffect aplant’s removal treatment.We removedonlythebudsofflowerheads(hereafter calledflower design wasused. randomisedblock consisted ofseedlingsthatdescendedfromoneflowerhead. A size wereused,astobeablestudytheeffect ofinitialsize.For from eachtriowererandomlydistributedoverthethreetreatments. Trios ofdifferent single parent, andnumberofleavesstems,wereselectedtoforma‘trio’.Plants three new-bornrosetteswithapproximatelythesameweight,descendingfroma per treatmentandspecies. To minimisetheeffect ofgeneticvariationandinitialsize, The experimentwasperformedduringthegrowingseasonof2000.We used15plants Experimental design caerulea at anintervalofapproximately0.5minnutrientpoorsoilahexagon flower budsandofrosette budswereanalysedwithanon-parametric WilcoxonSigned identical genotypeand sameinitialsize)asfactorsforthedryweights. Numberof We usedananalysisofvariancemodel withspecies,treatmentandtrio(plants of Data analysis the experiment,smallplasticwire-nettingbags wereplacedovertheflowerheads. heads withflowersandseedsthatotherwise wouldhavefallenfromtheplantduring were weighed. To beabletodeterminethenumberandweightoffull-grownflower flowering season. At harvestplants weredugout,andabovebelowgroundparts S. pratensis S. pratensis The numberofflowerbuds,rosetteflowersandrosetteswasrecordedfor The plants weresubjectedtooneofthreetreatments. The firstwasaflowerbud plantlets (5shoots each)inordertomimicnaturalconditions. and October3 , thatproducenewrosettebudsabovethesoilsurface,were rd for H. radicata ), since H. radicata th of August for of August has amoreprolonged H. radicata C. jacea C. jacea H. radicata , the‘trios’ Molinia , that and 17. Chapter 2 Flexible responses to bud removal able T

able 1). l a v o m e r B R T s Figure 3. e

s of identical

l a v o m e r B F ferent letter with the rio (plant T ight [g] s) were control (Control), e

W l o r t n o C art. Dry reatment and T s (with n=15 plant

s. Within each graph a dif l a v o m e r B R

art

l a v o m e r B F

fect on that plant p

l o r t n o C The treatment

0

l a v o m e r B R

est with Bonferroni correction (Fig. 3 a, b, d, e, g and h, see also

l a v o m e r B F T

l o r t n o r of Buds Numbe C 0

5 0

10 15

a e c a j a e r u a t n e C s i s n e t a r p a s i c c u S a t a c i d a r s i r e a h c o p y H art denotes a significant treatment ef As on the dry weight at harvest were performed with ANOV g. 3 c, f and I, see also genotype or seed family and same initial size) as explaining factors (Fi rosettes for RB removal should Note that the weight of the flowers for FB removal and the weight of the be interpreted as the weight of the removed plant p same plant p ivided into main rosette (black), Number and dry weight of rosettes and flowers per harvested plant, d new rosettes (white) and flowers (dashed). umbers of buds were analysed flower bud removal (FB removal) and rosette removal (RB removal). N with a Wilcoxon Signed Ranks 2). 18. Chapter 2 Flexible responses to bud removal rosettes inthesespecies. Mainrosetteweightwasslightlylower for (Fig. 3c,i).Flowerweight wasnotaffected bythesmallerbiomassinvestmentinside species (Fig.2). treatments. assigned to. The results arethusaconservativeestimateoftherealeffects ofthe therefore someplants couldnotreceivetheremoval treatmentthattheywere 2/3 ineverySpecies-Treatment combination)floweredorformednewrosettesand Most plants floweredandproducednewvegetative rosettes.Notallplants (butatleast Timing andextentoffloweringrosettebudformation Results the numberofcompared tests ( the Bonferronicorrectionwasapplied:significancelevelisloweredbydivision comparisons betweentheplants inonetrio.Inordertocompare allthreetreatments, were significantlynothomogeneous. This testwasusedtomakepair-wise Ranks Test, sincethedata werenotnormalorPoissondistributed,andtheirvariances with atwo-tailed significance of0.003. dry weightatharvest(n=45),resultingina Pearson’s correlationcoefficient of 0.437 pratensis increased (Fig.3f).Sincetriowasanimportant explainingfactorfordryweightin concomitant increaseintotal plantweight. The weights ofthemainandnewrosettes responses weresignificant(Fig.3i).Onlyin weight whenflowerbudswereremoved(Fig. 3c),butin 3). In The threespeciesrespondedmarkedlydifferent toflowerbudremoval (Table 1,Fig. Effect ofbudremovalondryweights was asecondperiodoffloweringinJuly, whenthesiderosettesstarted toflower. produced flowerbudsontheirmainsteminMayandthebeginningofJune,butthere June, whereastheproductionofrosettesstarted inJuly. The pratensis flowering andproducingsiderosettesuntiltheplants were harvested. The flower bud.Newrosetteswereformedin August andSeptember. of theanalyses. effect. Four H. radicata The emergenceofnewrosettesoccurredafter theonsetoffloweringinallthree Rosette budremoval decreased total dryweightin (Table 1),wetestedforbivariatecorrelation betweeninitialfreshweightand plants producedflowers(aboutthreeflowerheadsperplant)inMayand H. radicata Hypochaeris radicata we seeastrongtrend(p=0.065)ofanincrease intotal vegetative plants diedinthecourseofexperimentandwereleft out α = 0.05/30.017). There wasnosignificantblock produced severalnewonesforeveryremoved S. pratensis C. jacea flower budremovalledtoa H. radicata Centaurea jacea H. radicata none ofthebiomass H. radicata and continued C. jacea Succisa plants and S. 19. Chapter 2 Flexible responses to bud removal and 50% 50% (d)(d) Figure 4. controlcontrol controlcontrol / RB/ RB / RB/ RB 00%00% 1 1 Succ.Succ. Succisa pratensis Succ.Succ. Hypo.Hypo. FBremovalFBremoval RBremovalRBremoval ) RBRB RBRB Cent.Cent. Cent.Cent. Succ. , ( Hypo.Hypo. 0%0% 50% 50% 1 1 0%0% 100% 100% 200% 200% 300% 300% 400% 400%

0%0% 0%0%

50%50% 50%50% 00%00% 50%50% 50%50% 00%00%

11 11 11 11

control control FBremoval FBremoval RBremoval RBremoval control control / FB FB / / FB FB FB FB / FB FB / / Hypochaeris radicata (c)(c) (a)(a) (b) (b) ) Cent.Cent. controlcontrol controlcontrol Hypo. / RB/ RB / RB/ RB Succ.Succ. Hypo.Hypo. Succ.Succ. FBremovalFBremoval RBremovalRBremoval Number of Buds Dry Weight [g][g] Weight Weight Dry Dry ofof Buds Buds Number Number RBRB RBRB Cent.Cent. Hypo.Hypo. . In each diagram a treatment effect on flower allocation is plotted on the y-axis, . In each diagram a treatment effect 00%00% 200% 200% 300% 300% 400% 400% 500% 500% = flower (buds) of the control group = flower (buds) of the flower bud removal group = flower (buds) of the rosette bud removal group = new rosettes of the control group = new rosettes of the flower bud removal group = new rosettes of the rosette bud removal group 0%0% 1 1 0%0% 100% 100% 200% 200% 300% 300% 400% 400% 0%0%

0%0%

00%00%

50%50% 00%00% 50%50%

11

500%500% 400%400% 300%300% 200%200% 11 11 FBremoval FBremoval control control control control RBremoval RBremoval FB FB / FB FB / / FB FB / FB FB / /

Centaurea jacea Centaurea

)

Switching Compensation Switching Switching Compensation Switching Control FBremoval RBremoval Control FBremoval RBremoval Cent. ( FB FB RB RB RB Summary of the effects of continuous removal of either flower or rosette buds on the number and dry of continuous removal Summary of the effects weight of flowers and new rosettes of ( on new rosette allocation on the x-axis. Compensation for continuously removed and a treatment effect change in the number (a) and dry weight (b) of that bud type, buds is calculated as the percentage change in the number (c) are calculated as the percentage Trade-offs group. to the control compared The 100%-lines bud type, when the other type of buds is removed. and dry weight (d) of a certain errors on and bars represent species means and standard Dots of the treatments. indicate no effect basis of within trio comparisons. FB 20. Chapter 2 Flexible responses to bud removal 3b, e,h).Especiallyin side rosettesin the controlgroup,p=0.059)andshowedmore branchingofstems. The numberof many flowerbudswhenwerecontinuously removed. compensation forlost rosette budswasnotsignificant. plants newrosettescould notcontributetoflowering(Fig.3g). For buds wereremoved, and mostclearlysoin buds wereremoved. Flower productiondeclinedforallthreespecies whenrosette the flowerbudremovalgroupalsoproduced morefloweringstems(17.2 spectacular increasewasseenin species, althoughthiswasnotsignificant for Flower budremovalcausedanincreaseinthenumberofflowerbudsallthree Effect ofbudremovalonnumbersflowersandrosettes affect theweightofanypart ofthe significantly lowerfor radicata ANOVAs ondryweightatharvestforallspeciestogetherandeach speciesseparate ( Table 1. new rosettes,andflowerheadsbuds). Species x Treatment were usedasexplainingfactorsofthevariationindryweight(total, mainrosette, initial size)withinSpecies, Treatment (control,flowerbudremoval,androsetteremoval) C. jacea C. pratensis S. radicata H. All plants All ramn 2638* 3953**4922.3 9.2 4 *** 5.3 43.9 4 ** 3.8 62.6 4 Treatment x S ro(ihnS 23. . * 21. . *4 . 1.4 5.7 42 ** 2.2 18.1 42 *** 2.3 37.2 42 S) (within Trio aito fM fM fM fM F MS df F MS df F MS df F MS df Variation of Source pce S 1516. * 2. 75**28. 06**21379. *** 93.3 103.7 2 *** 20.6 81.9 2 *** 87.5 728.1 2 *** 68.4 1125.1 2 (S) Species ramn 6. 62**21721. * 601. * 842. *** 25.6 28.4 2 *** 16.6 66.0 2 *** 12.9 107.2 2 *** 16.2 266.2 2 Treatment ramn 691. *25654*23. . *2043.2 0.4 2 *** 11.0 ** 0.9 8.6 39.2 2 2 ** 6.3 * 5.4 3.3 *** 16.7 5.6 57.1 2 2 2 * 5.4 * ** 5.2 10.9 11.5 76.9 38.3 2 2 2 Treatment * 4.7 ** 7.4 16.6 170.5 2 2 Treatment ** 6.8 271.2 2 Treatment , Rosette budremovalinducedanincreasein thenumberofrosettebuds(Fig. Succisa pratensis ro 2712 . 5462 0.1 27 0.1 28 4.6 25 3.4 0.5 24 28 1.0 25 1.1 7.4 79 2.1 24 28 7.1 22 4.0 Error 77 23.1 3.6 24 28 Error 8.3 40.1 77 24 Error 16.5 74 Error ro1 1561**1 6276**1 . . 40446*** 4.6 0.4 14 1.6 0.9 14 1.9 *** 7.6 13.4 16.2 14 14 Trio *** 6.1 21.5 14 Trio 1.9 76.8 14 Trio S. pratensis Total plant C. jacea C. jacea and Centaurea jacea ( ( increased whenflowerbudswereremoved. * * ) ) , underrosettebudremoval,butthistreatmentdidnot , whichproducedfourtimesmorerosettebuds ifrosette 43. . 488121 . 2.0 1.9 6.7 2.0 14 14 1.2 8.8 14 1.6 36.1 14 H. radicata S. pratensis Main Rosette Main ). Species, Trio (plants ofidenticalgenotypeandsame that producedonaveragefourtimesas ( C. jacea * ) plants. C. jacea 475161 . 1.6 0.2 14 1.6 7.5 14 New Rosettes New , probablybecausefor these (Fig. 3a,d,g). The most ( ( * * ) ) H. radicata Flower Heads&Buds Flower 22422*** 2.2 2.4 42 631. *** 14.6 16.3 4 H. radicata vs. Hypochaeris plants in 10.7 for ( ( * * ) ) 21. Chapter 2 Flexible responses to bud removal S. texana and Succisa , showed a Gentianella ssp. Melampyrum H. radicata C. jacea showed a 3-4 fold increase Cucurbita pepo Cucurbita Hypochaeris radicata replaces flower buds and rosette replaces flower Hypochaeris radicata Succisa pratensis . The significantly different responses of the three The significantly different . shows a strong compensation for rosette buds and the for rosette strong compensation shows a (Lehtilä &d Syrjänen 1995), the biennial . C. jacea populations is highly depending on the sexual pathway, unlike populations is highly depending on the sexual pathway, C. jacea Centaurea jacea Centaurea M. sylvaticum (Lennartsson et al. 1998) and the annual et al. 1998) and (Lennartsson shows the strongest compensation for lost flower buds, but the weakest for but the weakest lost flower buds, for strongest compensation shows the and H. radicata and , but not in A in other words an production towards flowering, switch from rosette bud Plotting the effects of both treatments in the same graph (Fig. 4) can highlight graph (Fig. in the same of both treatments the effects Plotting Our results are consistent with compensatory effects shown in other with compensatory effects are consistent Our results (Avila-Sakar et al. 2001). Rosette bud removal led to a minor increase in number of (Avila-Sakar Compensation for lost flower All three species responded to flower bud removal by compensating buds, but not all to the same extent. buds at the same degree, it produces twice as many buds as it would have if buds as many buds as it would degree, it produces twice buds at the same was not effect this compensatory to develop (Fig 4a). However, had been allowed small size. buds were removed at a very (Fig 4b), since flower weights seen in the dry buds were removed, flower weight when rosette of flowers or total increase in number especially All species and in our experiment. did not occur (Fig. 4c and rosette buds were removed number and weight when decrease in flower to rosette production did occur in 4d). Switching from flowering pratensis decisions and biomass allocation reflect differences species with respect to meristem on allocation. and in constraints traits between the species in life history Discussion of flowering or development of new rosettes had Our study revealed that inhibition on the three perennial herbs impacts very different weakest for flower buds (Fig 4a). weakest for flower rosette buds, whereas buds, whereas rosette the hypothesised effects of flower or rosette bud removal, compensation or switching. compensation bud removal, or rosette of flower effects the hypothesised H. radicata campestris pratensis pratense experiments on short-lived species, such as the hemiparasitic annuals such as the hemiparasitic on short-lived species, experiments many other perennials. in the number of flower buds formed. This, together with the fact that this species had This, together with the fact that in the number of flower buds formed. in terms of biomass allocation to flower the largest proportional reproductive effort of seed production for heads in the control treatment, indicates the great importance since the urge for sexual short life span, This may be related to its this species. season is assumed to be larger for short-lived growing reproduction in a certain (1987) showed that the species (Ehrlén & van Groenendael 1998). De Kroon et al. growth of 22. Chapter 2 Flexible responses to bud removal flowers. sexual reproduction,what mightexplainthelowertendencytocompensate forlost strong andfastflowerbudcompensationthat conclude fromthehighbranchingrate,flexiblenumberoffloweringstemsand rosette budsin is loweredbydivisionthenumberofcompared tests ( between thethree Treatments (asinFig.3)theBonferronicorrectionisapplied:significancelevel (plants ofidenticalgenotypeandsameinitialsize).Notethattodeterminesignificantdifferences removal). Numberofflowerorrosettebudsproducedduringtheexperimentwerecompared per Trio Wilcoxon SignedRank Tests between Treatments (Control,FlowerBudremoval,andRosette Table 2. like compensate forlostrosettes; theplantdoesnotdependonnew rosettes forsurvival size ortogenotypiceffects. or lessidentical,wecannotconcludewhether variationbetweentriosisduetoinitial correlation test.However, sincewithineachtriobothinitialsizeandgenotypearemore between initialsizeanddryweightatharvest thatwasshownbytheresults ofthe programme, indicatedbythesymmetricgrowth formaswellthestrongcorrelation compared to acontrolsituation. This mightbelinkedtoamorestrictdevelopmental other twospecies.Whenbudsareremoved, itonlyproducestwiceasmanybuds circumstances maybemorefavourable. since duetothelonglifespan ofthegenetitcan waituntilnextyearwhen dies after flowering.For This monocarpicspeciesdependsonnewrosettesforsurvivalsincethemainrosette al. 1987). to contributecriticallyplantfitnessthroughenhancedseedproduction(deKroonet increase its vegetative weightandthenumberofnewrosettes,whichhasbeenshown soon after havingbeenmown. At thesametimethisshort-livedspeciestendsto adaptive responsemayalsobebeneficial,becausetheplantcanstart floweringagain case ofayearlymownnaturereserveliketheplaceoriginourplants, this respond quicklytodamageflowerbuds,forinstance inflictedbyherbivores.Inthe meristem availability. This, aswellthehighgrowthrate,enablesthisperennialto ubro oprsnZpZpZp Z 0.698 0.001 -0.387 -3.413 p 0.011 Z 0.194 -2.559 -1.299 0.065 0.065 0.102 -1.843 -1.848 0.017 -1.636 -2.397 p 0.005 Z -2.805 0.059 removal FB - Control -1.889 removal RB - Control 0.004 -2.900 0.045 -2.001 Rosettes Side removal FB - Control removal RB - Control Comparison Heads&Buds Flower of Number C. jacea Since themainrosetteispolycarp,there isnonecessitytoimmediately Succisa pratensis For C. jacea . Unlike Brmvl-R eoa 005092-.1 .6 3400.001 -3.410 0.360 -0.916 0.001 0.972 -3.415 -0.035 0.007 -2.674 removal RB - removal FB 0.001 -3.180 removal RB - removal FB H. radicata , compensationisshown,butonlywithrespecttothenewrosettes. H. radicata C. jacea shows somecompensation,buttoalesserextentthanthe , probablybecauseonlyfewrosetteswereremoved.We , thelong-livedrosette of new rosettesmaybemoreimportant thanflowering, H. radicata H. radicata α = 0.05/30.017). S. pratensis is notnoticeablylimitedby S. pratensis can postpone C. jacea 23. Chapter 2 Flexible responses to bud removal , H. may be H. radicata S. pratensis C. jacea Helianthus tuberosus . For this longer-lived species this . For this longer-lived species . New rosettes can therefore be seen . New rosettes , we see a tendency to shift from sexual to shift a tendency , we see H. radicata H. radicata H. radicata . Prati and Schmid (2000) found a similar increase in . Prati and Schmid (2000) found . In , flower bud removal caused an increase in the number of side , flower bud removal caused , flower bud removal did not cause a switch towards production of , flower bud removal did not cause S. pratensis S. pratensis C. jacea S. pratensis and In Timing has been recognised as an important factor determining the response has been recognised as an important Timing For In conclusion, we argue that both meristem and biomass allocation play an In conclusion, we argue that both meristem and biomass Switching functions in other life history towards in allocation shifts also caused Bud removal radicata reproduction towards vegetative growth and new rosettes when flower development is rosettes when flower development growth and new vegetative reproduction towards flower bud removal increased as a result of The number of new rosettes prevented. at no effect to increase, but there is of new rosettes also tended weight and the total A plant weight. total in an inflorescence allocation was demonstrated similar switch in forming species on the short-lived tuber bud removal experiment that new rosettes are very de Kroon et al. (1987) found 1993). However, (Westley in for current reproduction important as a way of switching and as a way of compensating indirectly for lost flowers. This indirectly for lost flowers. and as a way of compensating as a way of switching into the new rosettes, switched to an increased resource allocation means the plants for lost flower buds. and at the same time compensated explained by assuming that the lower part of the main stem serves as a storage organ, explained by assuming that the lower part buds. which is depleted by the formation of compensatory rosette to damage may even be of a plant to damage (Lehtilä & Syrjänen 1995). Response et al. 1998). In the season (Lennartsson period in restricted to a certain rosettes, even stronger than the switch in rosettes, even stronger than the in later years, whereas for the short-lived off investment will probably pay for lost flower buds is more immediate. Flower the relative benefit of compensation of the weight of the whole plant as well as the weights bud removal also increased the in parts separate bud removal for the clonal herb Ranunculus reptans, plant size as a result of flower the several functions did not change. Switches in while proportional allocation towards of sexual of costs size are an expression vegetative biomass allocation towards described by Ehrlén and van Groenendael (2001) and reproduction, and have been (1996) amongst others. El-Keblawy and Lovett Doust of the plant‘s resources could be that only a small part An explanation new rosettes. of a small flower bud to a flower head with seeds, and is used for the development becomes available when thus only a relatively small amount of extra resources As is shown by the large response in the rosette removal flowering is prohibited. The decrease of response. treatment, meristem preformation cannot explain the lack in in main rosette weight in response to rosette bud removal most of the rosette formation took place after the flowering peak. Thus, it is hardly the flowering peak. most of the rosette formation took place after removal of new rosettes on flowering was very small, of surprising that the impact development were since decisions concerning allocation to flowering and meristem This timing of allocation and of meristem made before side rosettes were removed. between allocation activity might be a major constraint on the expression of trade-offs to either function. 24. Chapter 2 Flexible responses to bud removal Adams AW 1955. Adams AW References Organization forScientificResearch(NWO-project80.33.452). Limpens madethedrawingsofFig.1. This researchwasfinancedbytheNetherlands Erik Simons,PetraSchepandMarkGroeneveldfortheirpracticalassistance. Juul and FransMöller, JanvanWalsem, HermanKlees,MariaVerkade, LiesjeMommer, Merel Soons,CarolinMixandPetrDostálfortheirusefulcomments onthemanuscript, We thankJanvanGroenendael,MaxineWatson, FrankBerendse,FelixKnauer, Acknowledgements developmental constraints. compensation andtheexpressionoflifehistorytrade-offs arerelatedtolongevityand between compensatingandswitching.Ourresults suggestthatthedegreeof important roleindeterminingtheresponsetobudremoval,termsofchoice ofmnDP&DiemerM2002. Effects ofsmallhabitat sizeandisolationonthepopulationstructure Hooftman DAP Darwinianapproachtoplantecology. -JEcol55:247-270. 1967. A Harper JL 1997.Organpreformationinmayapple asamechanismfor Geber MA,deKroonH& Watson MA 1990. The costofmeristemlimitation in Geber MA García MB&EhrlénJ2002.Reproductiveeffort andherbivorytiminginaperennialherb: Fitness comparative demographicstudyofannualandperennial 1989. A Fone AL &LovettDoustJ1996.Resourcere-allocationfollowingfruitremoval incucurbits: El-Keblawy A Ehrlén J&vanGroenendaelJM1998. The trade-off betweendispersabilityandlongevity-animportant Ehrlén J&vanGroenendael2001.Storage andthedelayedcosts ofreproductionintheunderstorey Ehrlén J1999.Modellingandmeasuringplantlifehistories.-In:Vuorisalo TO &Mutikainen PK(eds), & Aarssen LW 1999.Patternsofvariationinmeristemallocationacross Duffy NM,BonserSP andStephenson AG 2001.Growth andresourceallocationin Avila-Sakar G, KrupnickGA eKonH liirA&vanGroenendaelJM1987.Densitydependentsimulationofthepopulation de KroonH,Plaisier A newclassificationtheoryforadaptivestrategies & Aarssen LW 1996.Meristemallocation: A Bonser SP 1989.Developmental ecologyofmayapple: seasonalpatterns ofresource &Watson MA Benner BL &GraceJ1997.Plantresourceallocation.- Academic Press,SanDiego. Bazzaz FA 1996.Plants inchangingenvironment.Linkingphysiological,population,andcommunity Bazzaz FA historical effects ondemography. -JEcol85:211-223. correlations betweenfecundity andgrowth.-Evolution44:799-819. components attheindividualandpopulation levels.- Am JBot89:1295-1302. Patterns incantaloupe melons.-New Phytol134:413-422. aspect ofplantspeciesdiversity. - Appl Veg Sci1:29-36. perennial Life historyevolutioninplants. Kluwer Academic Publishers,London. genotypes andspeciesinmonocarpicBrassicaceae.-Oikos84:284-292. J Ecol77:495-508. pepo dynamics ofaperennialgrasslandspecies, in herbaceousplants. -Oikos77:347-352. distribution insexualandvegetative rhizomesystems.-FunctEcol3:539-547. ecology. -UniversityPress,Cambridge. ssp texana Lathyrus vernus Succisa pratensis : Effects offruitremoval.-Int JPlantSci162:1089-1095. . -JEcol89:237-246. Moench. ( Scabiosa succisa Hypochaeris radicata Polygonum arenastrum L.). -JEcol43:709-718. . -Oikos50:3-12. Hypochoeris : negativegenetic (Asteraceae). - Cucurbita 25. Chapter 2 Flexible responses to bud removal Chamaenerion (Orchidaceae). and Helianthus tuberosus species after damage. - Funct damage. species after . - J Ecol 91: 18-26. Tipularia discolor Tipularia Senecio sylvaticus Melampyrum Melampyrum Succisa pratensis . - Oikos 51: 331-342. . - Oikos 90: 442-456. . - Ecology 79: 1061-1072. . - Ecology (Onagraceae). - Am J Bot 84: 763-768. (Onagraceae). - in relation to mineral nutrition. - J Ecol 65: 747-758. in relation to mineral nutrition. - J Ecol Ranunculus reptans Lycopodium annotinum Lycopodium angustifolium PM (eds), The Netherlands. - In: Joyce CB & Wade in communities by water management John Wiley & Sons Ltd. management and restoration. European wet grasslands: biodiversity, of quality and population size on performance (eds), Life history evolution in plants. Kluwer Academic Publishers, London. Kluwer (eds), Life history evolution in plants. Am Nat 123: 411-426. allocation in a clonal plant. - Ecol Evol 10: 180-182. Trends interactions. - L., (Asteraceae). - Ecology 74: 2136-2144. of common wetland species. - Plant Biol 4: 720-728. - Plant Biol 4: wetland species. of common - Ecology 70: 1286-1293. forbs. - J Ecol 90: 1033-1043. dispersed grassland Epilobium dodonaei in 7: 273-292. Ecol 9: 511-517. campestris Gentianella Oikos 95: 156-160. plant van der Hoek D & Braakhekke WG 1998. Restoration of soil chemical conditions of fen-meadow plant van der Hoek D & Braakhekke WG van der Meijden R 1996. Heukels’ Groningen. flora van Nederland. - Wolters-Noordhoff, A Rengelink R, Copal of genetic variation, habitat P, The interacting effects & Ouborg NJ 2003. Vergeer Vuorisalo T Vuorisalo P & Mutikainen & Mutikainen PK TO life histories. - In: Vuorisalo 1999. Modularity and plant MAWatson of resource and patterns on population growth effect constraints: 1984. Developmental MAWatson plant-herbivore affect phenology in plant developmental 1995. Sexual differences in of inflorescence bud removal on tuber production effect The LC 1993. Westley Lehtilä K & Syrjänen K 1995. Compensatory responses of 2 Compensatory & Syrjänen K 1995. Lehtilä K Soons MB & Heil GW 2002. Reduced colonization capacity in fragmented populations of wind- in fragmented capacity GW 2002. Reduced colonization Soons MB & Heil evolution. - Funct Ecol 3: 259-268. in life-history Trade-offs SC 1989. Stearns and seed set in the perennial herb compensatory regulation of fruit J 1997. Competition and the Stöcklin dynamics dominance and the simulation of metapopulation Apical TV 1988. Svensson BM & Callaghan the survival and flowering of some perennial herbs, I. - Oikos CO 1956. Further observations on Tamm F 1977. Reproductive allocation in Andel J & Vera van Lennartsson T, Nilsson P Nilsson T, Lennartsson in the field gentian, Induction of overcompensation J 1998. Tuomi & Olejniczak P - in plants. and sexual reproduction vegetative allocation to stable 2001. Evolutionarily of the clonal within populations traits of life-history B 2000. Genetic differentiation Prati D & Schmid AASnow fruit production in of flower and DF 1989. Costs & Whigham 3 Size-dependent and -independent plant reproductive allocation in response to successional replacement

Eelke Jongejans, Hans de Kroon & Frank Berendse

Summary

Perennial herbs have been hypothesized to have two alternative options when they face the risk of being outcompeted in the course of succession: escape to other sites through seed dispersal, or persist by investing relatively more in vegetative growth. The hypotheses implicitly assume that costs of sexual reproduction prevent plants from escaping and persisting simultaneously. We test these hypotheses in a three-year garden experiment with four perennials (Hypochaeris radicata, Cirsium dissectum, Succisa pratensis and Centaurea jacea) by growing them in the midst of a tall tussock-forming grass (Molinia caerulea) that may successionally replace them in their natural habitat. Costs of sexual reproduction were significant since continuous bud removal enhanced total biomass or rosette number in all four species, while storage biomass increased slightly in the two most long-lived species. To mimic succession we added nutrients, which resulted in tripled grass biomass and higher death rates in the shorter-lived species. Three species increased their sexual reproductive allocation irrespective of size, thus supporting the escape strategy, but not at the expense of allocation to storage, which increased even in C. jacea. The strong size-dependency of both sexual reproduction and storage suggests that the two strategies collapse into a single syndrome: in response to succession plants have to increase in size for both survival and increased seed production.

Keywords: Centaurea jacea; Cirsium dissectum; Costs of sexual reproduction; escape strategy; Hypochaeris radicata; persistence strategy; sexual reproductive allocation; size dependency; succession; Succisa pratensis

Introduction

Succession is the gradual replacement of one plant community by another, and is characterized by the accumulation of plant biomass and soil nutrients (Crawley 1997). 28. Chapter 3 Reproductive allocation and succession production andvegetative reproduction.Costs ofsexualreproduction, i.e.any decreased vegetative reproductive effort under manipulatedhighplantdensities). been testedexperimentally (butseeHollerand Abrahamson (1977)foranexampleof However, thehypothesized phenotypicresponsesattheindividuallevelhaverarely characterized bylowcompetition(Werner &Caswell1977;deKroonetal.1986). high competitionbyherbaceousdicots andshrubs,relativetopopulations establishment probabilitiesanddecreasedindividual growthratesinpopulationswith escape hypothesisastheyfindhigher seedproductionbutlowerseedling Studies onthemonocarpicperennial individuals inconsecutivesuccessionalphases, frommeadowstowillowbushes. that severalwetmeadowspecieshave a decreasedpercentage offlowering periods (Marrsetal.2000). Analyzing within-speciesvariation Falinska(1991)showed populations, therebypreventingtheshift tonewsuccessionalstages forextendedtime aquilinum strategy (Prach&Pyšek1999).Species as many clonalperennialsmaybeinterpretedtoaconsequenceofthepersistence are relativeinnatureandnotallornothingdecisions. increasing sexualreproductionandseeddispersal. The predictedshifts inallocation but insuringlife-timereproductiveeffort. And thelatteralternativeisrealizedby storage andclonalgrowthattheexpenseofshort-termsexualreproductiveallocation increasing thelikelihoodofpersistencethroughincreasedplantsize, still inanearlysuccessionalstage (Ogden1974). The firstalternativeisrealizedby Alternatively plantspeciescanescapeinspace viaseed dispersaltoareasthatare species thataltertheirgrowthpattern couldsurvivetothenextsuccessionalstage. recruitment (Harper1977; Abrahamson 1979;1980). These authorsarguedthatplant more andimportant, attheexpenseofsexualreproductionandseedling average. Withincreasingsuccessionalstage vegetative growthandsurvivalbecome that duringsecondarysuccession,planttraits ofthecurrentspecieschangeon The rationalebehindthesealternativepredictionswasdeducedfromtheobservation predictions. three-year gardenexperimentwithfourherbaceousperennials,intendedtotestthese predictions withintheframeworkofallometricallocation,andpresentresults ofa has neverbeenexplicitlytested.Herewerephrasethishypothesisintogeneral persist, or2)toescapenewareas. To ourknowledgethislong-standing hypothesis 1974; Abrahamson 1980):1)toformdenseandnon-invadablepopulations gradually suppressedhavelongbeenhypothesizedtotwoalternatives(Ogden individuals respondtotheriskofsuccessionalreplacement.Perennialplants thatare et al.1992;Roem&Berendse2000).Itisunclearhoweverwhetherandhowplant are graduallyoutcompetedbytaller competitorsthataccumulatebiomass(Berendse go locallyextinct(Tilman 1987;Falinska1991),asearlysuccessionalplantspecies On thetimescaleofsuccessionplantpopulationsareephemeral,andwilleventually Underlying Abrahamson's (1980)hypotheses isatrade-off betweenseed Some circumstantial evidencedoesexistforthesehypotheses:thesuccessof are extremeexamplesofclonalspeciesthatcanformpersistent Dipsacus sylvestris Phragmites communis , bycontrast,supportthe and Pteridium 29. Chapter 3 Reproductive allocation and succession Figure 1. Centaurea = Cj ; SpSp from growing into other MM 5050 cm cm MM MM Succisa pratensis = were planted in a hexagon at 10 cm Sp TT BB BB ; Cirsium dissectum MM MM BB MM Molinia caerulea BB Cirsium dissectum = 1010 cm cm ; B = flower bud removal. SpSp Cj Cj Cj Cj Cd ; CdCd Cd Cd Hr Hr M. caerulea = M HrHr 2525 cm cmdeepdeep 1010 cm cmdeepdeep LawnLawn edging: edging: Hypochaeris radicata = Hr ; T; plant; = target reduction in fitness parameters like survival, growth, size or future reproduction due to or future reproduction growth, size like survival, in fitness parameters reduction hypotheses for these alternative are crucial sexual reproduction, allocation to resource of reproduction both modes could change plants of costs in the absence because history life between different that trade-offs studies have shown Recent independently. plant relationship between masked by the allometric size-dependent functions can be Jong 1986; (van Noordwijk & de into e.g. sexual reproduction size and investment Worley Ehrlén & van Groenendael 2001; 1994; Reznick et al. 2000; Ågren & Willson sexual which needs to be reached, after many species a threshold size et al. 2003). In 1988; plant size (Weiner increases with vegetative linearly often reproductive biomass a non-linear Klinkhamer et al. 1992 for Méndez & Obeso 1993; but see Hartnett 1990; between life history trade-offs for, when plant size is accounted increase). But that show by manipulative experiments be found, as is evidenced functions can 1993; Primack & another life history function (Westley switches from one function to et al. 2004) or by studies under natural, 1998; Obeso 2002; Hartemink Stacy Schematic design of part of the experiment. The subblock of eight plants represented here either The subblock of eight plants of the experiment. Schematic design of part two adjacent or served as control. Each of the twenty blocks contained received extra nutrients, eight species by bud removal subblocks with contrasting nutrient treatment. Within a subblock the The subblocks were fenced with 25 cm combinations were randomly distributed over the positions. cm Ten between nutrient treatments. deep lawn edging to prevent root growth or leakage of nutrients deep lawn edging (50 cm diameter) kept the rhizomatous plots. Around each target plant six cuttings of Around each target plots. intervals. jacea 30. Chapter 3 Reproductive allocation and succession outcompete theherbs (Berendseetal.1992;vanderHoek 2004). half oftheplots. Underfertilizedconditions thegrassspeciesisexpectedto availability duringnaturalsuccessionwithhigh atmosphericdeposition,wefertilized they co-occur. In ordertomimictheaccumulationofplantbiomassandnutrient In ourexperimentthetarget herbplants competewithadominant,tall grasswithwhich Crawley 1997),andespeciallysubordinate herb speciesarelost(Roemetal.2002). deposition astheyoccurintheNetherlands acceleratesuccession(Tilman 1987; dominant grassspecies.Lateralsoshrubs start todominate.Highlevelsofnitrogen and sometimesadditionalgrazing.Succession starts bybiomassaccumulationofthe (Schaminée etal.1996)andismaintained inthisstage bymowingandhayremoval Molinietum grasslandsharboraspecies-rich,early-successionalcommunity perennial herbsfromnutrient-poormeadowsintheNetherlands. These Cirsiodissecti- Our secondaimistotesttheshypothesesinathree-yeargardenexperimentwithfour plants ofequalsize. vegetative andstorageorganslesstoseedproduction,compared tounaffected their allometricrelationships insuchawaythattheyallocatemorebiomassto plants ofequalsize. (escape strategy)andlesstovegetative andstorageorgans, compared tounaffected relationships insuchawaythattheyallocatemorebiomasstoseedproduction neighboring plants increasesduringsuccession,perennial plants shift theirallometric hypothesis ofOgden(1974)and Abrahamson (1980): Consideration ofsize-dependencyleadstothefollowingreformulation phenotypic (plastic)responsesoftheallocationpattern tochangingconditions. allocated tosexualreproduction.Onlythelattertwoscenarioscanberegardedasreal shift inreproductiveallocation,orc)byanincreasetheproportionofresources relationship betweenreproductiveallocationandplantsize,b)byasize-independent can differ betweentwogroups ofplants: a)astheresultofasimpleallometric Sugiyama andBazzaz(1998)describedthreewaysinwhichreproductiveallocation shifts inreproductiveallocationpatterns. Acknowledging thissize-dependency offspring numberwhenfloweringandseedsetareinhibited. 2002) toinvestigateifplants switchtoincreasedsize,storage,andvegetative Ogden and Abrahamson aremet.We usethemethodofflowerbudremoval(Obeso sexual reproductioninordertoverifywhethertheprerequisitesforhypothesesof vegetative growth(Reekie&Bazzaz1987;Watson etal.1997). sexual reproductivestructuresareself-supportingorevenleadtoenhanced when thesecosts decreasewithincreasingnutrientavailability(Reekie1991)orwhen Økland 2002).However, costs ofsexualreproductionmaystillnotbecomeapparent undisturbed conditions(Snow&Whigham1989;WillemsDorland2000;Rydgren The allometricrelationships alsohaveconsequencesforthehypothesized For thefourperennialplantspeciesstudied,ourfirstaimistotestforcosts of Alternatively (persistencestrategy),perennialplants mayshift when overallbiomassof 31. Chapter 3 Reproductive allocation and succession . Figure 2. Centaurea jacea Centaurea and day of the month. C=control. th Centaurea jacea 2000 jaod fmamjjasondjfmamjjas fmamjjasondjfmamjjas mjjasondj mjjasondj jaod fmamjjasondjfmamjjas mjjasondj 00 7575 5050 2525 75 100100 100 Succisa pratensis (L.) Hill (Asteraceae), is a rhizome-forming , flowering stalks and new rosettes are formed flowering stalks L. (Asteraceae), is a relatively short-lived perennial L. (Asteraceae), is a relatively Cirsium dissectum Cirsium dissectum 2001 2002 , ), vegetative plant parts (flowering stems, stem leaves, rosette (flowering parts plant ), vegetative Hypochaeris radicata Hypochaeris radicata BB NBNB CC NN Hypochaeris radicata Hypochaeris radicata Succisa pratensis Succisa Hypochaeris radicata Cirsium dissectum 2000 2001 2002 C. dissectum jaod mm aod fmamjjas fmamjjas jasondj jasondj fmamj fmamj mjjasondj mjjasondj Meadow Thistle, jaod mm aod fmamjjas fmamjjas jasondj jasondj fmamj fmamj mjjasondj mjjasondj

00

7575 5050 2525 7575 100100 100100 Survival [%] Survival (Fone 1989). Study system and species system Study Cat's ear, leaves, and roots), and storage organs (the caudex, i.e. the persistent rootstock to caudex, i.e. the persistent rootstock and storage organs (the leaves, and roots), attached). are leaves, stems and roots which the rosette varies the number of the leafless flowering stalks clonally by branching of the taproot; in June and continues until et al. 1987). Flowering starts from one to several (de Kroon autumn. rosettes. Most daughter rosettes are formed at the end clonal plant, with monocarpic Allocation of biomass to four parts of the plants was studied: sexual reproductive of the plants to four parts Allocation of biomass (rhizomes; flower heads), clonal organs heads, seeds, and buds of structures (flower only in B=bud removal; N=nutrients added; N+B=both bud removal as nutrients added. added; N+B=both bud removal as nutrients B=bud removal; N=nutrients The hypotheses on costs of sexual reproduction and shifts in allocation pattern were pattern in allocation and shifts of sexual reproduction on costs The hypotheses perennials: experiment with four a three-year (2000-2002) garden investigated in Hypochaeris radicata Materials and methods Materials Survival percentages in time. A in time. Survival percentages for the 15 letter on the x-axis stands 32. Chapter 3 Reproductive allocation and succession dissectum species *twobudremoval treatments *tworesourcetreatments *20 replicates (Fig. transplanted intoanexperimental gardenofWageningen University. sizes ratherthanselecting forequally-sizedplants. The cuttingsandseedlingswere Chapter 1)wastaken intoaccountbystarting the experimentwitharangeofplant the lownutrients treatment. The size-dependencyofallocationresponses(Fig.2 group. Eachpair ofplants waseitherassignedtothenutrientadditiontreatmentor each pair was assignedtothebudremovaltreatment,otheruntreated collected atthesamelocality. Plants of (52°01'N, 5°36'E;vanderHoek&Braakhekke1998).Cuttingsof 'Bennekomse Meent',anutrient-poorgrasslandnearWageningen intheNetherlands Seeds ofthefourtarget specieswerecollectedin 1998inthenaturereserve Plant material Aerts etal.1990),especiallywhenfieldsareabandoned. caerulea forming, tall grass,whichoccursinnutrient-poorgrasslandsandgrassheaths. target specieshavecompositeflowerheads.Onlytheseeds stems thatgrowhorizontally forseveralcentimetersbeforegrowingvertically. All four surface alongsidethefloweringstem.Subsequentlytheserosettescanformnew vegetative side-rosettesareformedonthewoodyrootstock andappearatthesoil flowering stalk andflowersfromJuneuntilautumn.Duringafter flowering, or, incaseof seedlings weregroupedinpairs ofsimilarinitialsize,andthesamegeneticidentity, the budremovaltreatment(seebelow)and theundisturbedplants, allcuttingsand two-months-old seedlingswereused.Inorder toallowfordirectcomparisons between 2000 newlyformedrosettesoftheseplants werecarefully brokenoff. For grown fromseedinagreenhouseoneyearbeforethestart oftheexperiment.InMay to surviveatleasttenyears(Tamm 1956). perennial, althoughithasmonocarpicshoots. Singleindividualshavebeenrecorded caudex. and flowersfromJulytillSeptember. Newrosettesalsoemergelaterallyfromthe Hooftman &Diemer2002). polycarpic andcansurviveformanyyears(Adams1955;Safford etal.2001; Meijden 1996))duetothedeclineofits habitat (Lucassenetal.2003). Netherlands Cirsium dissectum of longrhizomes,butnewrosettescanalsobeformedatthebaseoldrosette. Purple Moor-Grass, The 320plants wererandomly placedinarandomizedblockdesign: four Knapweed, Devil's-bit Scabious, starts todominatewhennutrientdepositionishigh(Berendse& Aerts 1984; are plumedandadaptedtowinddispersal(Soons&Heil2002). C. dissectum H. radicata Centaurea jacea normally formsonlyone,apicalflowerheadinJune.Inthe , grownfromseedsofthesamemotherplant. Oneplantof , isarareandendangeredspecies(RedList2(vander Succisa pratensis Molinia caerulea Succisa pratensis L. C. jacea s.l. (Asteraceae), isarelativelylong-lived laterally formsuptofourfloweringstalks (L.) Moench(Poaceae),isatussock Centaurea jacea Moench (Dipsacaceae), rosettesare , S. pratensis and has asingleapical H. radicata C. dissectum M. caerulea H. radicata and Molinia were were C. 33. Chapter 3 Reproductive allocation and succession C. M. consist s added; a Figure 3. C. dissectum m H. radicat s; the upward error a u ant six arget pl e art s not shown). c s of ect a ss ea j di (result r u ium rs each t Around Centa Ci B D M. caerulea t these circles of lawn s indicated tha s of 0 cm. s was 5 al weight. C=control. B=bud removal; N=nutrient s), storage organs (caudex), and rhizomes ( ta andard errors of the mean weight of plant p ca s cm. th sides of 10 a hexagon wi were placed in s each ant arget pl di si s added. Please note that the N and NB group en t s ra a ri . onal plot s on 20 additi pr a hae c s only ccis s (leaves, stems and root po s increment fect the biomas y ole through the wh from growing atous species vent this rhizom s to pre r shoot s of fou andard error of the mean tot art Su H C A

plant

clump

i e w y r D ] g [ t h g

g i e w y r D ] g [ t h ative plant p he t rval between t The inte of two and three plant e rcle around th m-diameter ci eep in a 50 c ced 10 cm d dging was pla Lawn e m dissectu rement garden. Measu af edging did not 1). a caerule only). Downward error bars denote st bar denotes the st NB=both bud removal as nutrient buds, flower heads and seeds), Dry weight [g] at harvest divided into sexual reproductive tissue (flower veget 34. Chapter 3 Reproductive allocation and succession applied 200kgNha al. (2004)foundsignificantshifts inthevegetation compositioninthefieldwhenthey agricultural landscapes(Bobbinketal.1998;vanOene1999).Van derHoeket the target plants. The Hoagland'sstocksolutioncontained KNO experiment. Nutrientsolutionwasappliedtoacircularareaof50cmdiameteraround allowing theplants toestablish underthesameconditionsinfirstyearof Nutrient enrichmentwasappliedtohalfoftheplants inthesecondandthirdyear, Treatments Rank test(Lee&Wang 2003)perspecieswithbudremovaland nutrientenrichment Treatment effects onthesurvivalofplants wereanalyzedwithaKaplan-MeierLog Analysis and weighed. rhizomes ofthefourperennialsandgrass weredriedat70ºCforleast48hours caerulea hasthickroots (Taylor etal.2001). The stems, leaves,caudex,roots and The roots ofthetarget plantandthegrasswere relativelyeasytoseparate sinceM. percentage ofthefineroots waslostastheplants rootedshallowlyinthesandysoil. hexagon. The plots weredugoutatleast25cmdeep,andonlyaverysmall harvested inacircularareaof40cmdiameteraroundthecenter the individual.InSeptember2002allplants wereharvested.Belowgroundparts were the averageofeachspeciesandtreatmentcombinationwithflowerheadcount flowers, andseeds. Total flowerheadweightperplantwascalculatedbymultiplying experiment toestimateaverageweightofthesexualreproductivetissue:flowerhead, measured. Randomflowerheadswerebaggedafter floweringthroughoutthe and flowerheadswerecounted;stemheight,maximumleaflengthwidth measured attheendofeachthreegrowingseasons.Leaves,rosettes,stems, In additiontothemonitoringofsurvivaltarget plants, plantsizewas Measurements treatment; theotherhalfwasallowedtoflowerandsetseedsnaturally. throughout thethreeyearsofexperiment.Halfplants receivedthis by thesamelawnedging. group (Fig.1). The unfertilizedplants ofeachblockwerealsogroupedandenclosed 25 cm-deeplawnedgingtopreventnutrientleakagetheplants ofthelownutrients nutrient treatmentwereplacedtogetherineachofthe20blocksandsurroundedwith kg Nha within twomonthsatthebeginningofgrowingseasonandwasequivalentto120 and NH Flower budsonfloweringstalks wereremovedthree timeseachmonth 4 H -1 2 PO yr -1 4 , whichisaboutthreetimestheatmosphericdepositioninDutch (Gamborg &Wetter 1975). The solutionwasappliedinthreeportions -1 yr -1 . Inourexperimentallplants thatwereassignedtothehigh 3 , Ca(NO M. caerulea 3 ) 2 , MgSO 4 , 35. Chapter 3 Reproductive allocation and succession Figure 4. CNBNB CNBNB CNBNB CNBNB CNBNB CNBNB 200020002000 2001 2001 2001 2002 2002 2002 200020002000 2001 2001 2001 2002 2002 2002 Centaurea jacea jacea Centaurea Centaurea CirsiumCirsium dissectum dissectum CNBNB CNBNB CNBNB CNBNB D D B B 000 000 808080 606060 404040 202020 404040 303030 202020 101010 CNBNB CNBNB CNBNB CNBNB CNBNB CNBNB CNBNB CNBNB HypochaerisHypochaeris radicata radicata Succisa pratensis pratensis Succisa Succisa 200020002000 2001 2001 2001 2002 2002 2002 200020002000 2001 2001 2001 2002 2002 2002 CNBNBCNBNBCNBNB CNBNB CNBNB CNBNB CNBNB CNBNB CNBNB CNBNB CNBNB CNBNB A A CNBNB CNBNB CNBNB CNBNB CNBNB CNBNB CNBNB CNBNB CNBNB C C

000 555

000 444 888

101010

2.52.52.5 7.57.57.5 121212 161616 Number of rosettes rosettes of of Number Number Number of rosettes rosettes of of Number Number as explaining factors in different tests. Prior to statistical analysis the number of analysis the to statistical Prior tests. in different factors as explaining of Dry weights improve normality. to buds were ln-transformed flowers and rosettes, the to increase when necessary were ln-transformed plant parts and plants performed were ANOVAs III Type groups. of the tested of the variance homogeneity as fertilization and their interaction with bud removal, of plant parts on dry weights as random factor. within the fertilization treatments nested plant pair fixed factors and factor were factors and year as time with the same MANOVAs Repeated Measures with (Zar 1996) ANCOVAs I Type on flower and rosette numbers. used for the data to test for as covariate were performed leaves and stems) (roots, biomass vegetative Development of the number of rosettes (mean + standard error) of surviving plants. The extra nutrient error) of surviving plants. Development of the number of rosettes (mean + standard Flower buds were removed all three years. C=control. B=bud removal; June 2001. treatment started added. added; NB=both bud removal as nutrients N=nutrients 36. Chapter 3 Reproductive allocation and succession in survival ratesinthehigh nutrientgroupdecreasedto33%in 10.50, p=0.001andLog Rank=25.58,p<0.001respectively; Fig. 2). At harvest nutrient additiondecreased survivalin 279 g,n=320,F=1.47·104,p<0.001).While budremovaldidnotinfluencesurvival, The total biomassof Effects ofnutrientenrichment:increasesinsexual reproductiveeffort short floweringperiodinJune. removed in of flowerheadsincreasedstronglywhenbudswerecontinuously in efforts tocompensateforthelostflowerbuds. The numberofflowerheadsandbuds S. pratensis resulted inasmallincreasetherelativebiomassallocationtostorageorgans in theformationofadditionalrosettes. Apart fromsizeincreasesbudremoval also its rosette numberlinearlyintime(Fig.4B).Onlythelastyearbudremovalresulted in between years(Fig.5; Table 2). This compensationeffect wasalmostabsenthowever constant intimetheunmanipulatedplants. The rhizomatous vs. 4.9)inrosettenumberthethirdyear(Fig.4C),whereaswasvery first twoyears(Fig.4A).In increase after thefirstyear, andwaspositivelyaffected byflowerbudremovalinthe bud removalgroup(Table 1). The numberof the averagetotal dryweightatharvestafter threeyearswassignificantlyhigherinthe although notinallyears(Table 1and2).In clonal propagation. Significantcosts ofsexualreproductionwerefoundinallspecies, (Figs. 3and4),thusclearlyshowingcosts ofsexualreproductionforplantgrowthand Both total biomassandrosetteformationwereenhancedbycontinuousbudremoval Costs ofsexualreproduction Results collinear inthreespecies(Table 1). designed todiffer inplantsize,andthesetwoparameters indeedturnouttobe ANCOVAs. Plantpair wasnotincludedinthesemodelsbecausetheplantpairs were (flower headsandseeds).Budremovalfertilizationwerethefixedfactorsinthese differential biomassallocationtostorageorgans(caudex)orsexualreproduction C. dissectum H. radicata Besides theseswitchestootherlifehistoryfunctions,budremovalalsoresulted and H. radicata (90% and65%respectively forthelownutrientgroup). Two (Fig. 5B),probablybecauseithaspreformed flowerheadsandonlya C. jacea Molinia caerulea , , butnotintheothertwospecieswithshorter-livedrosettes. S. pratensis Succisa pratensis tripled inresponsetonutrientaddition(91 and C. dissectum Cirsium dissectum bud removalcausedalargeincrease(10 C. jacea Hypochaeris radicata , althoughthiseffect differed and H. radicata C. dissectum C. dissectum and Centaurea jacea rosettes didnot (Log Rank= increased and 13% C. jacea vs . , 37. Chapter 3 Reproductive allocation and succession Figure 5. CNBNB CNBNB CNBNB CNBNB CNBNB CNBNB the variance in total weight in total the variance Centaurea jacea jacea Centaurea Centaurea CirsiumCirsium dissectum dissectum 200020002000 2001 2001 2001 2002 2002 2002 200020002000 2001 2001 2001 2002 2002 2002 CNBNB CNBNB CNBNB D D CNBNB CNBNB B B 000 555 000 300300300 200200200 100100100 202020 151515 101010 S. pratensis S. pratensis CNBNB CNBNB CNBNB CNBNB CNBNB CNBNB plants survived. Only these two species were able to two species Only these survived. plants CNBNB CNBNB CNBNB CNBNB S. pratensis S. pratensis HypochaerisHypochaeris radicata radicata Succisa pratensis pratensis Succisa Succisa 200020002000 2001 2001 2001 2002 2002 2002 20002000 2001 2001 2002 2002 CNBNBCNBNB CNBNB CNBNB CNBNB CNBNB CNBNB CNBNB CNBNB CNBNB CNBNB CNBNB CNBNBCNBNBCNBNB CNBNB CNBNB CNBNB CNBNB CNBNB CNBNB CNBNB CNBNB CNBNB C C A A 000 00

200200 400400 600600 800800 100100100 200200200 300300300

Number of flowers and removed buds buds removed removed and and flowers flowers of of Number Number build up significantly more biomass when extra nutrients were given (Table 1). The 1). (Table were given when extra nutrients biomass significantly more build up and were additive biomass on total and bud removal enrichment of nutrient effects in interact. Especially did not these treatments plants died and all died plants was largest in the group of plants that were fertilized as well as having their buds fertilized as well as having that were the group of plants was largest in in size to were able to increase 3), indicating that not all individuals removed (Fig. no buds of which the group of plants by the grasses. Within prevent being dominated (p<0.002 the root-shoot ratio significantly nutrient enrichment lowered were removed, ratio of root in mean of the ln-transformed when testing differences in all species and leaf biomass). biomass by stem Development of the number of flowers and removed buds (mean + standard error) of surviving plants. Development of the number of flowers and removed buds (mean + standard June 2001. Flower buds were removed all three years. C=control. The extra nutrient treatment started added.. added; N+B=both bud removal as nutrients B=bud removal; N=nutrients 38. Chapter 3 Reproductive allocation and succession replacement wasfound inthreeoutoffourspecies,i.e.an increase insexual al. 2003). pratensis This isinagreementwithSwissfieldobservations: withincreasingsiteproductivity keep larger andtosecuretheirplaceinthevegetation. Onlytheselargeplants wereableto two species, forming plantcouldrunandescapefromthe increasinglydensetussocks.Intheother Mortality ratesin (Berendse etal.1987;deKroon&Bobbink 1997;vanderKrift &Berendse2002). when competingwithagrassspeciesthat accumulatesbiomasslike two specieshavearelativelyhighturnoverofleafbiomass,whichisdisadvantageous mortality ratesintheshort-lived and causedmoreintensecompetitionpressureonthetarget plants, resultinginhigh caerulea nutrients tograsslandperennialsthatweregrowninbetweentussocksof successfully mimickedthisessentialprocessofsuccessionalchangebyadding vegetation andsubordinatespeciesaregraduallydriventolocalextinction.We reproduction (persistencestrategy).Duringsuccessionnutrients accumulateinthe allocation tostorageandvegetative plantparts attheexpenseofallocationtosexual allocation tosexualreproduction(escapestrategy)ortrypersistbyincreasingthe and Abrahamson (1980)thatwithprogressingsuccessionplants mayincreasetheir In ourexperimentweexplicitlytestedthereformulatedhypothesesofOgden(1974) Discussion 6D). which speciesstoragebiomassincreasedslightlyperunitofvegetative weight(Fig. allocation tostoragebiomasswasonlyaffected bynutrientadditionin addition effect wasdifferent forlargerplants ascompared tosmallerplants. Relative of thetwo and nutrient additionincreasedsexualreproductivebiomassin vegetative biomassinallspecies(Figs.6and7; Table 3).Whilecontrollingforsize, sexual reproductiveandstoragebiomasswerehighlysignificantlycorrelatedwith increased averageflowernumberinallspeciesbut sexual reproductiveeffort tendtosupporttheescapehypothesis.Nutrientaddition for thepersistencehypothesisasformulatedinintroduction,whileresults on while therelativeallocationtostoragedidnotchangemuch. Thus thereislittlesupport S. pratensis M. caerulea One ofthehypothesized phenotypicallocationresponsesto successional The plantinthefertilizedplots allocatedrelativelymoretosexualreproduction, . As expectedthistall grassincreasedinbiomassafter nutrient enrichment density decreased,but plantsizeandseedproductionincreased(Billeter et C. jacea Succisa pratensis . (Figs.7A,7Band7C).Bycontrast,theslopesofregressionlines C. dissectum under controlandcouldbenefitfromtheextra nutrients themselves. groups differ significantly(Fig.7D),indicatingthatthenutrient were lowerthanin and Hypochaeris radicata Centaurea jacea H. radicata H. radicata , largerplants wereabletogrow and Cirsium dissectum H. radicata , becausethisrhizome (Table 2;Fig.5).Both , C. dissectum M. caerulea C. jacea . These Molinia , in S. 39. Chapter 3 Reproductive allocation and succession s that Figure 6. s, leaves and wer s had lo cted s intera fect um ative (root cea a ssect i a j ts r d s veget re a s added; N+B=both bud removal u p which suggest fected,

a t ium s r an ent atment ef , the tre Ci C pl B ve D tati e C. jacea g e v

f o

] ght [g i e ta w ca s y r si D adi r en s i Ln at r nt but smaller pla ive allocation, gher reproduct s had hi a pr hae s c i c po c y H Su A C s added.

n under nutrient addition. in elevated sexual reproductio s are involved

n L f o ] g [ t h g i e w y r D s n a g r o e g a r o t s , the ll four species contrast, in a ed controls. In than unfertiliz ctive allocation reprodu rdly af organs was ha tion to storage endent alloca size dep no cost pecies, . In the fourth s ctive allocation reprodu e: larger plant with siz it Dry weight [g] of the storage organs (caudex) plotted per plant against stems) dry weight [g] at harvest. C=control. B=bud removal; N=nutrient as nutrient 40. Chapter 3 Reproductive allocation and succession especially in bud removal.Inhibitionofflowerandseedproductionincreasedtotal biomass lived perennials,asisexemplifiedbyourresults ofthreeyearscontinuousflower Sexual reproductionhoweverdoeshavedemographiccosts inthelongrunlong- Costs ofSexualReproductionafter three Years ofBudRemoval interactions betweennutrient additionandbudremovalinourexperiment whichwould (Saxena &Ramakrishnan 1984;Loehle1987;Reekie1991), but therewereno availability couldbeanother explanationforincreasedsexualreproductive allocation the annual (2001) forinstance, foundastrongcorrelationbetweenseedoutputandstemmassin increased incomparison toallocationroots, whentheplots werefertilized.Cheplick production mayhavebeenincreasedbecause allocationtoallabove-groundtissues the resultofalowerroot-shootratiounder nutrientenrichedconditions:seed the allocationresponses,wecouldspeculate thattheincreasedseedproductionwas runners andlong-livedplantparts beforetheycanadoptthepersistencestrategy. strategy todealwithsuccession.Perhaps herbs needtheabilitytoformbothlong between theirlifehistories(Harteminketal.2004)donotmattertoomuchfor escape strategy. The similarresponsesofthesespeciessuggest thatthedifferences allocation irrespectiveofplantsizeafter nutrientaddition, whichisconsistentwiththe to sexualreproduction. Three ofourfourspeciesincreased theirrelativesexual succession duetotwofactors:anincreaseinplantsizeandallocation Our experimentrevealedincreasedseedproductionperplantundermimicked strongly size-dependent Sexual reproductiveallocationincreasesinresponsetothenutrienttreatmentandis sexual reproductiveallocationinrelationtootherlifehistoryfunctions. (1974) and Abrahamson (1980)intheseperennialherbsby studyingshifts inrelative and survivalindicatethatitisindeedmeaningfultotestthehypothesesofOgden 5). These demographictrade-offs betweensexualreproductionandvegetative growth resulted inlargerplants, whichhavehighersurvivalprobabilitiesinthefield(Chapter & vanGroenendael2001;Obeso2002;Harteminketal.2004). costs ofsexualreproductionasithasinotherstudies(Avila-Sakar etal.2001;Ehrlén this additionalinvestmentinnewflowerbuds,themethodsucceededrevealing compensation responsesbyactivationandproductionofnewflowerbuds.Inspite increased onlyslightly. Although themethodofflowerbudremovalalsoinduced dominance offloweringoverrosetteformation.Simultaneously allocationtostorage increased relativelymorethantotal biomass,probablyduetoareleaseofapical total biomassbutalsocausedmeristemicresponses:thenumberofrosettes S. pratensis Although theexperimentwasnotset-upto elucidate themechanismsbehind Both biomassandmeristemicresponsestofloweringinhibitioneventually Amaranthus albus and C. dissectum C. jacea . In and S. pratensis . Lowercosts ofseedproduction underhighnitrogen C. jacea , andincreasedrosettenumberespeciallyin bud removalnotonlycausedincreasesin 41. Chapter 3 Reproductive allocation and succession , s, C. Figure 7. ative (root ative weight of s veget um eget cea a ssect e ds. Nor can th s of see i a j roduction s of sexual rep ts r er omass are rath ative bi d re a u p . s are analyzed

a t s added. V ium s r an ent Ci C pl B ve D tati aking the number of rosettes during flowering e g e v

f o

] contrary t group: on the he high nutrien ality in t ght [g i e ta w ca s y r si D adi r en s i Ln at r a pr hae s c i c po c y ost removal treatment showed c s, while the bud H Su A C was set back to the moment of flowering by t

n the ed reproduction was found i e that no cost of the elevat It is remarkabl g [ t h g i e w y r D n L 0 2 n i s r e w o l f f o ] 2 0 ith high mort two species w s in the eget expense of v duction at the igher seed pro s with h g ion of survivin uted to select ation be attrib oductive alloc s sexual repr t toward plant plant urviving plant selected against when only s expected to be fertilized plant se the cost would decrea utrient addition cted in case n be expe shif nt against it Dry weight [g] of the flowers (flower heads and seeds) plotted per pla leaves and stems) dry weight [g] at harvest. C=control. N=nutrient dissectum and multiplying it with the average rosette weight at harvest. 42. Chapter 3 Reproductive allocation and succession statistic; (*)=p<0.10;*p<0.05;**p<0.01and***p<0.001 buds; 'Vegetative' =stems,leavesandroots; 'Storage' =caudex;dfdegreesof freedom;F= ANOVA factors. Data wereln-transformedwhennecessary(ln(W)).'Flowers'= flowers,seedsandremoved pratensis larger part ofthevariation insexualreproductionwasexplainedby plant sizethanby species andbetween different habitats or successional stages. Also inourstudya (1992) reportthatplant sizeexplainsmostofthevariationinseedoutput betweenherb (Sugiyama &Bazzaz1998; Reznicketal.2000;Worley etal.2003).ShipleyandDion between sexualandvegetative reproduction aslargeplants arebetterinboth contradiction isthatthesize-dependency ofseedproductionmaskstrade-offs in total biomassandinthenumberofside rosettes. The explanation forthis ANOVA's ondryweight(W)[g]atharvestof Table 1. trg nW nW ln(W) ln(W) ln(W) W ln(W) ln(W) ln(W) W Rhizomes W ln(W) Storage W W W Vegetative Flowers oa nW W F df ln(W) F df W F df W F df Effect Total Dry Weight and ro 23 36 1.6 18 1 38 0.1 ** 7.6 36 ** 1 3.6 12.7 1.5 1 1 1 29 *** 19.2 22 38 1 36 ** 2.9 *** 350.7 * 4.6 Error 38 4.0 1.8 1 B *N 7 (N) 1 Nutrients 3.9 1 *** 22 33 3.5 Bud (B) removal *** 18.3 *** 3.7 37 N) (within Pair 1 2227.2 Intercept 38 1 36 1 ** 2.4 ** *** 1 17.9 2.3 16.1 1.5 *** 160.0 * 1 38 1 4.8 Error 1 ** *** 37.1 1 36.5 20 1.7 B 7 (N) Nutrients 1 1 1 ** 22 *** 33 2.6 Bud removal (B) 106.8 * *** 4.7 38 N) (within Pair 210.6 1 Intercept 38 * 1 ** ** 1 13.0 17.0 0.9 1.2 0-10.210.110.6 3.3 ** 1 1 8.8 Error 1 *** 64.4 20 1.0 B N 1 7 Nutrients 1.8 1 1 22 33 Bud removal (B) 1.8 N) (within Pair 1 Intercept * 0.8 ** 1 22.4 0.1 0.9 0-10.810.010.1 * 1 1 4.5 Error 1 20 B N 7 (N) Nutrients 1 Bud removal (B) 0.6 N) (within Pair Intercept * 1 1.0 0-11.310.110.0Error 1 B N (N) Nutrients Bud removal (B) * 0-10.910.110.1 N (N)10.711.711.0113.4** necp 98* 26**135. * 9. *** 397.1 1 *** 3153.3 1 *** 42.6 1.7 1 33 ** 19.8 1.2 1 20 N) (within Pair Intercept Centaurea jacea with budremoval,nutrientadditionandpair ofplants asexplaining H. radicata H. MS=0.39 MS=3.52 MS=0.22 MS=7.98 ( * ) C. dissectum Hypochaeris radicata . . 76*** 37.6 1 * 5.0 1 2.7 1 MS=0.86 MS=116.60 MS=2.93 MS=1.29 MS=51.04 ( ( ( ( * * * * ) ) ) ) 837**3 . ** 2.7 38 *** 3.7 38 843**3 . *** 3.3 38 *** 4.3 38 S. pratensis 0. * 959*** 3995.9 1 *** 901.4 1 MS=0.30 MS=0.30 MS=0.23 MS=0.34 , Cirsium dissectum ( ( * * ) ) 15.2* *** 20.9 1 C. jacea C. MS=0.63 MS=2257.95 MS=0.21 MS=983.96 , Succisa 43. Chapter 3 Reproductive allocation and succession S. and C. dissectum , (Sugiyama & Bazzaz 1997) H. radicata . In order to produce more seeds that may . In order to produce Abutilon theophrasti C. jacea responds similarly to another type of stress, salinity. responds similarly to another that only survives in a small number of remnant Iris hexagona C. dissectum , or size-dependent like in , or size-dependent Increasing seed production as habitat conditions deteriorate may have Increasing seed production as habitat Consequently the hypotheses of plant responses to successional replacement hypotheses of plant responses Consequently the populations (Soons et al. 2003), seed production level can be a crucial limitation for be a crucial limitation populations (Soons et al. 2003), seed production level can Acolonization. of seed production may be the last sign of life before relative increase succession advances and a remnant population becomes a senile one in which seedlings no longer can establish. in which sexual offspring is the only way of increasing fitness with higher resource in which sexual offspring of other findings of increased sexual reproduction in we are not aware availability, et al. (2003) do show Zandt Van However, perennials under increasing competition. that the clonal plant that perennial and clonal species can adjust their Thus empirical evidence is emerging growing conditions, confirming model predictions life history strategy to adverse 1998; Gardner & Mangel 1999; Olejniczak 2003). (Sakai 1995; Saikkonen et al. implications for a species on a regional level. Such responses would have important dynamics, in which both persistence (patch great implications for metapopulation (production of diaspores for colonization of empty and sexual reproduction occupancy) (Eriksson 1996; Soons et al. 2003). Especially for a Red are key parameters patches) List-species as Implications from examples in annuals like Apart pratensis only do so has to survive, and it can a plant first favorable patches, in more establish Therefore the & Wiggerman 1997). size to avoid shading (Huber its by increasing one syndrome: the into essentially and persistence strategy collapse escape strategy by increasing their size, that manage to survive increased competition respond plants of their biomass to sexual an increased percentage while at the same time allocating are sexual reproduction the plants to increased there are costs Although reproduction. still larger. (Ogden 1974; Abrahamson 1980) need a reappraisal. Our data suggests that suggests Our data 1980) need a reappraisal. Abrahamson (Ogden 1974; to on plant size and that the responses is strongly dependent allocation in general as in be either size-independent succession may the mimicked succession. The tight relationships between storage weight and the storage weight between The tight relationships succession. the mimicked of response with the small is consistent which plant parts, of the vegetative weight less link and even developmental a strong removal, suggest allocation to bud storage reproduction. towards sexual storage than allocation towards for flexible opportunity 44. Chapter 3 Reproductive allocation and succession *** =p<0.001 analysis. df=degreesoffreedom;FWilk'sLambdastatistic; (*)=p<0.10;*p<0.05;**p<0.01and dissectum end ofthethreeconsecutiveyearsallocationexperimentwith Repeated measuresMANOVAs onrosettenumbersandof flowers orremovedbudsatthe Table 2. Adams AW 1955. Adams AW (ed),Demography Abrahamson WG1980.Demography andvegetative reproduction.-In:SolbrigOT Abrahamson WG1979.Patterns ofresourceallocationinwildflowerpopulations offieldsandwoods.- References research (NWOproject805-33-452). Gleichman. The NetherlandsOrganizationforScientificResearch fundedthis Jan vanWalsem, JaspervanRuijven,Juul Limpens,LouisdeNijsandMaurits we aregratefultoNienkeHartemink,Frans Möller, HenkvanRoekel, HermanKlees, helpful comments onearlierversionsofthismanuscript. Fortheirpracticalassistance We thankSusan Kalisz,JohanEhrlén,HeidiHuberandJaspervanRuijvenfortheir Acknowledgements ubro feto fFd fFd F df F df F df F df of Effect Rosettes of Number FlowersBuds orRemoved and evolutioninplantpopulations. BlackwellScientificPublishers,Oxford. Am JBot66:71-79. , Succisa pratensis Within subjects Between subjects Within subjects Between subjects . . 0.0 36 *** 88.3 1 2 * *** 6.3 55.6 0.2 38 2 1 1 ** 1.4 8.5 ** 9.4 * 4.9 ** 0.0 2.9 76 1 2 22 1 *** * 1 38 31.4 1.9 *** 3645.0 ** 2.5 1 2.7 1 66 0.0 - *** ** 2 2612.4 38 6.4 1 0.8 1 7 ** 2.9 0 40 *** 818.5 1 0.1 33 1 N) (within * Pair Y 1 *** 0.7 Year (Y) 120.1 1.3 1 1 Error 20 * B N Nutrients (N) Bud removal (B) Pair (withinN) Intercept . . 0.8 2 2.1 0.0 36 2 0.4 1 2 0.2 *** 80.0 0.7 38 * 2 2 4.7 ** 10.4 1 *** 560.3 2 1 - *** 2 0.4 40.1 1.1 ** 11.3 *** 1.2 107.9 22 1 2 0 1 1 2 *** 85.5 2 4.0 *** ** 37.6 0.2 8.5 1 - * 3.1 2 6.0 2 2.4 2 1 7 0.4 Y * * B N 2 0.5 3.0 2 0 40 Y* Nutrients (N) 2 0.3 0.2 Y* Budremoval (B) 1 1 1.4 N) (within * Pair Y ** 8.5 2 1 *** ** Year (Y) 420.0 35.5 2 1 2 1 Error *** - 12.7 * B N Nutrients (N) 2 1.8 Bud removal (B) 0 Pair 2 Intercept 1.8 2 0.2 Y * * B N 2 Y* Nutrients (N) Y* Budremoval (B) (withinN) 202.1 332.2*387.1***382.1* Succisa pratensis and Centaurea jacea Moench. ( H. radicata MS=1.61 MS=0.87 Scabiosa succisa ( ( * * ) ) . Data weretransformed(Ln(number+0.1))priorto 661.2 764.1***761.3 C. dissectum . 15**254** 5.4 2 *** 11.5 2 * 3.7 2 MS=0.87 MS=1.38 ( L.). -JEcol43:709-718. * ) 044**120. *** 2407.9 1 *** 5074.4 1 S. pratensis Hypochaeris radicata MS=0.21 MS=0.55 ( * ) 61.4 76 MS=0.42 jaceaC. MS=1.25 , Cirsium ( * ) 45. Chapter 3 Reproductive allocation and succession and Cirsium Fragaria : A field study (L) Moench as (Asteraceae). - Cucurbita pepo Cucurbita Geranium maculatum Geranium Hypochoeris . - Oikos 50: 3-12. Molinia caerulea Trifolium fragiferum Trifolium L and Hypochaeris radicata Erica tetralix in chalk grassland. - In: de Kroon H & van Groenendael in chalk grassland. - In: de Kroon H . - J Ecol 89: 237-246. : Relation to soil nutrients. - Int J Plant Sci 162: 807-816. : Relation to soil nutrients. : an experimental field study. - Oikos 70: 35-42. - Oikos field study. : an experimental Brachypodium pinnatum : Effects of fruit removal. - Int J Plant Sci 162: 1089-1095. - Int J Plant of fruit removal. : Effects (L.) Hill. Plant Ecol 165: 45-52. (Rosaceae). - Am J Bot 64: 1003-1007. (Rosaceae). - Lathyrus vernus texana gradient of nutrient availability. - Oikos 57: 310-318. - of nutrient availability. gradient affected by the availability of nutrients. - Acta Oecol Oecol Plant 5: 3-14. Acta - by the availability of nutrients. affected 73: 46-53. ecosystems. - Ecology the plant. - Oecologia 74: 174-184. production and nutrient losses from ecosystems; Litter Sci 6: 3-12. Appl Veg to abandonment. - meadow species J Ecol 86: 717-738. - European vegetation. diversity in natural and semi-natural Amaranthus albus reference to Backhuys Publishers, Leiden. and evolution of clonal plants. The ecology J (eds), growth rate. - Ecology 67: 1427-1431. to population demographic parameters species, dynamics of a perennial grassland J Ecol 77: 495-508. Saskatoon. clonal plant. - Ecology 80: 1202-1220. removal in three perennial herbs. - Oikos 105: 159-167. composites. - Oecologia 84: 254-259. metapopulations. - Oikos 77: 248-258. metapopulations. ssp sylvaticum G. perennial virginiana height. - Plant Ecol 130: 53-62. manipulated vegetation with experimentally Funct Ecol 6: 308-316. reproductive output in plants. Incorporated, Hoboken. 208. levels responsible for the decline of of low pH and high ammonium Interactive effects dissectum of common wetland species. - Plant Biol 4: 720-728. Aerts R, Berendse F, de Caluwe H & Schmitz M 1990. Competition in heathland along an experimental in heathland M 1990. Competition de Caluwe H & Schmitz F, R, Berendse Aerts Krupnick GA G, allocation in Growth and resource AG 2001. Avila-Sakar & Stephenson Berendse F, Elberse WTBerendse F, in grassland plants Competition and nitrogen loss from RHEM 1992. & Geerts AF & Bol J 1987. Oudhof Berendse F, on nutrient cycling in wet heathland study comparative DAPBilleter R, Hooftman responses of common fen and reversible & Diemer M 2003. Differential species on pollutants of air-borne nitrogen The effects M & Roelofs JGM 1998. Bobbink R, Hornung Cheplick GP of mass allocation and the allometry of reproduction in genetics Quantitative 2001. Blackwell Science, Oxford. Crawley MJ 1997. Plant Ecology. plant dominance under elevated nitrogen deposition, with special de Kroon H & Bobbink R 1997. Clonal Elasticity: the relative contribution of A, van Groenendael J & Caswell H 1986. de Kroon H, Plaisier Ade Kroon H, Plaisier of the population & van Groenendael JM 1987. Density dependent simulation LGamborg O & Wetter Culture Methods. National Research Council of Canada, Tissue 1975. Plant and translocation in a clonal offspring, in seeds, investments Gardner SN & Mangel M 1999. Modeling Harper JLAcademic Press, London. 1977. Population biology of plants. Hartemink N, Jongejans E Flexible life history responses to flower and rosette bud & de Kroon H 2004. reproduction in four clonal Hartnett DC 1990. Size-dependent allocation to sexual and vegetative reproduction in relation to density in Abrahamson WG 1977. Seed and vegetative Holler LC & Eriksson O 1996. Regional dynamics of plants: a review of evidence for remnant, source-sink and a review of plants: Eriksson O 1996. Regional dynamics Academic Publishers, Dordrecht. succession. Kluwer in vegetation Falinska K 1991. Plant demography AFone A 1989. and perennial demographic study of annual comparative Berendse F & Aerts R 1984. Competition between R 1984. Competition Aerts Berendse F & Ågren J & Willson MF 1994. Cost of seed production in the perennial herbs in the perennial Cost of seed production Willson MF 1994. Ågren J & reproduction in the understorey of and the delayed costs Storage Ehrlén J & van Groenendael J 2001. J 1992. On the analysis of size-dependent TJ & Weiner Klinkhamer PGL, Meelis E, de Jong Lee ET John & Sons, analysis. Wiley, methods for survival data JW 2003. Statistical & Wang a benefit-cost model. - Oikos 49: 199- in clonal plants: Loehle C 1987. Partitioning of reproductive effort PJM, Lamers LPM & Roelofs JGM 2002. van der Ven AJP, Bobbink R, Smolders Lucassen ECHET, The ecology of Mitchell RJ, Goddard D, Paterson S & Pakeman RJ 2000. Marrs RH, le Duc MG, Hooftman DAPHooftman on the population structure size and isolation of small habitat & Diemer M 2002. Effects Huber H & Wiggerman L in the clonal herb 1997. Shade avoidance 46. Chapter 3 Reproductive allocation and succession h aao r egtwr ntasomd f=dgeso reo;F=ACV statistic; The data ondryweightwereln-transformed.df=degreesoffreedom; F= ANCOVA weights oftheflowersin2002undisturbed(nobudremoval)plants wereanalyzedthesameway. addition asfixedfactorsandvegetative dryweight(roots, leavesandstems)ascovariate. The dry jacea p<0.10; *=p<0.05;**p<0.01and***p<0.001 Reekie EG1991.Costofseedversusrhizomeproduction in Primack R&Stacy E1998.Costof reproductioninthepinklady'sslipperorchid, 1999.Howdospeciesdominatinginsuccessiondiffer fromothers?-JVeg Sci10: Prach K&PyšekP 2003.Optimalallocationtovegetative andsexualreproductioninplants: theeffect of Olejniczak P Ogden J1974. The reproductivestrategy of higherplants. II. The reproductivestrategyof Obeso JR2002. The costs of reproductioninplants. NewPhytol155:321-348. Méndez M&ObesoJR1993.Size-dependentreproductiveandvegetative allocationin Safford HD,Rejmánek M&HadacE2001.Species poolsandthe"hump-back"modelofplantspecies Rydgren K&ØklandRH2002. Ultimatecosts ofsporophyteproduction intheclonalmoss Roem WJ,KleesH&BerendseF2002.Effects ofnutrientaddition andacidificationonplantspecies Roem WJ&BerendseF2000.Soilacidityandnutrient supplyratioaspossiblefactorsdetermining 2000.Bighouses,bigcars,superfleasand thecosts ofreproduction. & Tessier A Reznick D,NunneyL 1987.Reproductiveeffort inplants. 3.Effect ofreproductiononvegetative Reekie EG&BazzazFA ANCOVAs perspecies( Table 3. Dry Weight Dry Flowers organs Storage ) onthedryweight[g]ofstorageorgans(caudex)atharvestwithbudremovalandnutrient 1679. Orchidaceae): An eleven-yearexperimental studyofthreepopulations. Am JBot85:1672- 383-392. ramet density. -EvolEcol17:265-275. farfara (Araceae). CanJBot71:309-314. bracken: Its roleinsuccessionandimplicationsforcontrol.- Ann Bot85(SupplB):3-15. diversity: anempiricalanalysis atarelevantspatial scale.-Oikos95:282-290. splendens diversity andseedgermination inheathland.-J Appl Ecol39:937-948. 151-161. changes inplantspeciesdiversitygrasslandandheathland communities.-BiolConserv92: - Trends EcolEvol15:421-425. activity. - Am Nat129:907-919. 2683. . . . . * 7.0 1 0.6 0.2 1 1 0.0 *** 223.1 3.7 *** MS=0.44 49.5 35 2.5 0.5 1 1 1 1 MS=0.096 36 *** 958.9 1 1 *** 50.1 0.2 1 * 6.2 1.3 MS=0.41 25 1 F 0.0 0.1 1 * MS=0.082 4.6 70 MS=0.095 1 1 11 *** 134.0 * 4.3 1 1 df 1 1 MS=0.13 * 2.2 72 4.9 3.0 0.1 1 *** *** 95.9 166.5 1.9 1 MS=0.10 1 1 51 ** 1 1 10.6 1 *** 1 10780.7 F 3.6 1 df 1 MS=0.31 0.1 *** 23 138.8 0.2 Error 0.1 *** *** V *N 433.8 1 1617.2 1.4 1 1 F (N) Nutrients 1 1 0.6 1 1 (V) Vegetative Weight 1 *** 409.2 *** Intercept 318.5 1 df 0.1 1 1 *** 1 Error 27.9 *** 0.2 F 854.1 B *NV 1 1 N *V 1 B *V df *** 77.5 B *N (N) Nutrients 1 (B) Bud removal (V) Vegetative Weight Intercept Effect L. -JEcol62:291-324. . -Ecology83:1573-1579. Hypochaeris radicata H. radicata , ( Cirsium dissectum * ) 11.6 11.5 12.2 C.dissectum Agropyron repens , Succisa pratensis S. pratensis . -CanJBot69:2678- (Cypripedium acaule ( * ) and 39*** 13.9 1 Arum italicum Hylocomium C. jacea Centaurea Tussilago ( * ( * ) ) = , 47. Chapter 3 Reproductive allocation and succession Iris levels and Eupatorium 2 (Orchidaceae). Helianthus tuberosus Tipularia discolor Tipularia 1111. L. - Ecology 79: 1620-1629. L. - Ecology (L.) Moench. - J Ecol 89: 126-144. (L.) Chevall. - Plant Biol 2: 344-349. Molinia caerulea Potentilla anserina Potentilla Huds.). - Ecology 58: 1103- Spiranthes spiralis : implications for maintenance of genetic variation. - Oecologia 112: 35-41. variation. - Oecologia 112: : implications for maintenance of genetic L. in the secondary successional environments following slash and burn agriculture successional environments L. in the secondary . - J Ecol 91: 837-846. Dipsacus sylvestris Dipsacus availability, competition and genetic identity. - Funct Ecol 12: 280-288. competition and genetic identity. availability, 7: 273-292. - Ecology 70: 1286-1293. forbs. - J Ecol 90: 1033-1043. dispersed grassland Soons MBrich semi-natural grasslands. - In: and connectivity. fragmentation (ed), Habitat PhD thesis, Utrecht temporal characteristics of the colonization process in plants. and Spatial University. Abutilon theophrasti - Ecol Monogr 57: 189-214. gradients. in the Netherlands. - In: Joyce CBcommunities by water management PM & Wade (eds), John Wiley & Sons Ltd. management and restoration. European wet grasslands: biodiversity, Sci 15: 387-394. - J Veg meadow. in plant species composition in a species-rich fen shifts - Funct Ecol 16: 198-203. nutrient availability. Am Nat 128: 137-142. - in life history tactics. life-history traits. - Am Nat 161: 153-167. Am - life-history traits. nitrogen inputs on vegetation succession. Ecol Appl 9: 920-935. succession. Ecol on vegetation nitrogen inputs the clonal plant, consequences of salinity stress for the growth and reproduction of hexagona L. (Asteraceae). Ecology 74: 2136-2144. the perennial orchid of shoot bud fates: ways in which the developmental program modulates fitness in clonal plants. of shoot bud fates: ways in which the developmental - In: de Kroon H & van Groenendael JThe ecology and evolution of clonal plants. (eds), Backhuys Publishers, Leiden. L and strategies. Oxford University Press. (eds), Plant reproductive ecology: patterns teasel ( odoratum (Jhum). - Weed Res 24: 127-134. (Jhum). - Weed droge heiden. Opulus Press, Uppsala. van graslanden, zomen en Plantengemeenschappen 467-483. metals on resource allocation in on resource metals a spatially varying environment. - J Theor Biol 175: 271-282. - J varying environment. a spatially Sugiyama S & Bazzaz FA of resource 1998. Size dependence of reproductive allocation: the influence the survival and flowering of some perennial herbs, I. - Oikos CO 1956. Further observations on Tamm AP K, Rowland Taylor & Jones HE 2001. Soons MB & Heil GW 2002. Reduced colonization capacity in fragmented populations of wind- in fragmented capacity GW 2002. Reduced colonization Soons MB & Heil and connectivity of species- E & Heil GW 2003. Fragmentation Soons MB, Messelink JH, Jongejans Sugiyama S & Bazzaz FA and density in output in response to soil nutrients 1997. Plasticity of seed nitrogen experimental of plant dominance along Secondary succession and the pattern D 1987. Tilman 1998. Restoration of soil chemical conditions of fen-meadow plant van der Hoek D & Braakhekke WG and nutrient-driven Nutrient limitation AJEM & van Groenendael JM 2004. van der Hoek D, van Mierlo in differing of four grass species from habitats Berendse F 2002. Root life spans & TAJ van der Krift Groningen. van Nederland. Wolters-Noordhoff, van der Meijden R 1996. Heukels' flora on variation Acquisition and allocation of resources: their influence AJ & de Jong G 1986. van Noordwijk of historic CO effects van Oene H, Berendse F & de Kovel CGF 1999. Model analysis of the Worley AC, Houle D & Barrett SCH 2003. Consequences of hierarchical allocation for the evolution of Worley NJ, USA. analysis. Prentice Hall, Upper Saddle River, Zar JH 1996. Biostatistical Van Zandt PA, Tobler MA, Mouton E, Hasenstein KH & Mopper S 2003. Positive and negative MA, Tobler PA, Zandt Van of inflorescence bud removal on tuber production in The effect LC 1993. Westley and their relation to age in Willems JH & Dorland E 2000. Flowering frequency and plant performance Watson MA, Hay MJM & Newton PCD 1997. Developmental phenology and the timing of determination MA, Hay MJM & Newton PCD 1997. Developmental Watson of competition on plant reproduction. - In: Lovett Doust LThe influence J 1988. Weiner & Lovett Doust PAWerner models for & Caswell H 1977. Population growth rates and age versus stage-distribution Schaminée JHJ, Stortelder AHF & Weeda EJ 1996. De vegetatie van Nederland. Deel 3. van Nederland. Deel vegetatie EJ 1996. De AHF & Weeda Stortelder Schaminée JHJ, Am Nat 139: herbaceous angiosperms. - allometry of seed production in The J 1992. Shipley B & Dion AASnow fruit production in of flower and DF 1989. Costs & Whigham Saikkonen K, Koivunen S, Vuorisalo T Vuorisalo K, Koivunen S, Saikkonen Mutikainen P & heavy of pollination and effects 1998. Interactive Sakai S 1995. Optimal resource allocation to vegetative and sexual reproduction of a plant growing in of a plant and sexual reproduction vegetative allocation to Optimal resource Sakai S 1995. in allocation of resource and patterns PS 1984. Growth KG & Ramakrishnan Saxena 4 Bottlenecks and spatiotemporal variation in the sexual reproduction pathway of perennials in meadows

Eelke Jongejans, Merel B Soons & Hans de Kroon

Summary

Sexual reproduction is important for population expansion and maintenance of genetic diversity. Several steps are involved in the sexual reproduction pathway of plants: production of flowers, production of seeds and establishment of seedlings from seeds. In this paper we quantify the relative importance and spatiotemporal variability of the different steps involved. The grassland perennials studied were Centaurea jacea, Cirsium dissectum, Hypochaeris radicata and Succisa pratensis. We compared undisturbed meadows with meadows where the top soil layer had been removed as a restoration measure. Data on number of flower heads per flowering rosette, flower and seed number per flower head, and seedling establishment probabilities per seed were collected by field observations and experiments in several sites and years. Combination of these data shows that H. radicata and S. pratensis have higher recruitment rates (1.9 and 3.3 seedlings per year per flowering rosette respectively) than the more clonal C. dissectum and C. jacea (0.027 and 0.23 respectively). Seedling establishment is the major bottleneck for successful sexual reproduction in all species. Large losses also occurred due to failing seed set in C. dissectum. Comparison of the coefficients of variation per step in space and time revealed that spatiotemporal variability was largest in seedling establishment, followed closely by flower head production and seed set.

Keywords: Centaurea jacea; Cirsium dissectum; flower production; grassland; Hypochaeris radicata; seed predation; seed production; seedling establishment; Succisa pratensis

Introduction

The demography of plants comprises the survival and growth of individuals as well as their vegetative and sexual reproduction (Harper 1977). In perennial plants it is mostly survival and vegetative reproduction of existing individuals that determine year-to-year 50. Chapter 4 Seed production and seedling establishment Hartemink etal.2004). Specifically weaddressthefollowing questions:Whatarethe Schmid (1999)highlightedallsteps intheclonal subsequent steps ofthesexualreproductionpathway inthesamesystem.Meyerand investigated. Onlyalimitednumberofstudieshaveattemptedtocompare the bottlenecks forsexualreproductionisonlypossiblewhenallsteps inthesequenceare species thatco-occurin nutrient-poormeadowsintheNetherlands (Soonsetal and spatial variationinallsteps. As astudysystemwe usefourperennialplant sexual reproductionpathway to identify thebottlenecks,andcompare thetemporal is particularly lowforperennialforbs.Inthisstudy wequantifyallmajorsteps ofthe number ofstudiesthatsubsequentlyanalyze allreproductionsteps andtheirvariability will affect their paws (Haemodoraceae),whileKlinkhameret al. (1988)concludethatanyseedloss seed productionandseedlingestablishment was enhancedbyfireintwokangaroo seedling establishment isveryrare.Lamont and Runciman(1993)foundthatboth (Špacková etal the creationofgaps ofbaresoilinthevegetation canenhanceseedlingnumbers experimental treatments, concludingforinstance thattheremovalofamosslayer, and (Eriksson &Ehrlén1992;JakobssonEriksson2000).Othershavecompared central questionofwhetherseeds,microsites,orbotharelimitingseedlingnumbers & Potvin1995).Researchonseedlingestablishment hasmainlyfocusedonthe heads hasbeenshowntolimitseedlingestablishment in 2003), andnumber-masstrade-offs (Cheplick1995).Predationbyinsects onflower Woodward 2003),pollination(Menges1995;Herrera2000), seedabortion(Berg (Oostermeijer etal Research onseedproductionhasfocusedforinstance oneffects ofpopulationsize seedlings (Fig.1).Separately, theseaspects havebeenilluminatedbyseveralstudies. suitable forbothgerminationandestablishment, andsubsequentlythesurvivalof pollination offlowers,productionviableseeds,arrivalseedsatmicrositesthatare involves severaldifferent processes:productionofflowerheadsandflowers, establishment oftheseseedsasseedlingsandfinallyreproductiveadults. This pathway generallyconsists oftheproductionseedsbyfloweringplants, and low probabilitiesofsurvivalfromonesteptothenext. The sexualreproduction demography. This isduetothemultitudeofsteps involvedandthesometimesvery reproduction ismuchhardertoassessthantheotheraspects ofperennial colonize areasoutside therangeofvegetative growth(Crawley1997). maintains geneticdiversity, mayformseedbanks,andenablesthepopulationto sexual reproductionhasothermerits thancontributingtolocalplantnumbers:it growth rateinsomecases(Silvertownetal.1993;deKroonal 1989; Picó&Riba2002),butsexualreproductioncanmarkedlyincreasepopulation plants isrelativelylessimportant forpopulationgrowththanadultsurvival(Eriksson population dynamics.Manystudiesindicatethatsexualreproductioninlong-lived Assessment ofthefullsexualreproductionpathway andidentificationof Although suchstudiesaremorecommonin forestecology(Schupp1990),the Although important inbothalocalandregionalperspective,sexual Cirsium vulgare . 1998; Kotorová&Lepš1999;Isselsteinetal . 1998; Brysetal.2004),geographicdistribution(Jump& populations. Solidago altissima Cirsium canescens . 2002). . 2000). Inaddition, and concludethat (Louda . 2003; 51. Chapter 4 Seed production and seedling establishment s n art ant n o i o t i t a a d Figure 1. d n e ferent p e o i r t Pr (L.) Hill and ng o l a on P t i l a i t zi t a e y a a m rs t s i r i u e er G p mit e l / t n p s i abil s d i s o i ng vi ti i -d os d se r t en li r w o s ll s e- c o i on- o o r o M L M P Ab P N P mport iation and are i s of var Cirsium dissectum e L., t

ficient a sources in this study on the dif r a e set o s p r g er athway from flower production to seed set to seedling n s? Which als in meadow of four perenni athway i f atial coe mb er u w N o l f s s Hypochaeris radicat r ed ce e t r een ent a u eads eeds eads w ow m s t l i

h h ed f e r f r

so f e e er pr o a t o species s t a s ow ow on b t l l on- f D t) are indicated. f is n

r oun a on exp oun i C p it and C d m he ws and mown road verges: t nials of nutrient poor meado ared four peren d o a ed and C - athway (lef d at e d poral and sp the highest tem s have e e comp ablishment. Causes of losses of ovules (right) and dat S e pr tudy system three Asteraceae nd methods Materials a S W numbers? tion in seedling redicted varia for the p n p al reproductio cks in the sexu bottlene step Schematic overview of the sexual reproduction p est of the p 52. Chapter 4 Seed production and seedling establishment or rhizomatous,clonalgrowth( persistence: highrosettesurvival( to theotherspecies(potentiallydecades),whichhavedifferent strategiesof Hypochaeris radicata Flowering wasmonitoredinthreetoeightpermanentplots of1m Flower headproduction and theircodescanbefoundin Table 1onpage 58. listofallsites meadows andrestorationareaswherethetopsoillayerisremoved. A clustered infourregions. The rangeofsitesincludesnutrient-poor, undisturbed 1999; Soons&Heil2002;Mixetal.2003). dissectum one seedperflower. The seedshavesimilarweight,butthoseof flowers relativelylate. All fourspeciesproduceflowersinflowerheadswithpotentially radicata dissectum Centaurea jacea populations of of thenumberviable,emptyandpredated seedsin C. jacea tweezers) andthenumberofflowerswerecounted. Flowerswerecounteddirectlyin that seemedviable(testingforastrongembryo bysqueezinggentlywithapair of after collection,theflowerheadwasdiscarded from thedata set. The numberofseeds flowers orpappus ringswerestillpresent.Ifone ormoreofthesecriteriawerenotmet that noseedshaddisseminatedandincaseof selecting flowerheadswerethatseedshadtoberipeenoughdetermineviability, Flower headsofthefourspecieswerecollectedinseveralpopulations. The criteriafor Flower productionandseedset 1999 to2003. heads perfloweringrosettewerecountedonceeveryyearafter peakfloweringfrom V, NandO).Intheseplots thenumberoffloweringrosettesand numberofflower radicata flower heads. the meanseed-flower ratioofnon-predatedflowerheadswiththe meanratioofall insect larvae. The impact ofpredationon apopulationwasevaluatedbycomparing not. Signsofpredationwere:crumbledor partly eatenseeds,andthepresenceof Furthermore itwasevaluatedwhetherany seeds inaflowerheadwerepredatedor The studysitesareinthecentral-easternpart oftheNetherlandsandare , butestimatedbynumberofpappus ringsin (site L,NandO),infivemeadowpopulationsof and , arareandendangered(RedList)species,flowersinJune,whereas have pappus thatmayenhancedispersalbywind(Jongejans&Schippers C. jacea C. jacea L. s.l. is short-lived(seldommorethantwoorthreeyears)compared have amoreprolongedfloweringseason. (sites B,NandO), and theDipsacaceous species C. dissectum S. pratensis C. dissectum ), formationofnewrosettes( ) (Harteminketal.2004). C. dissectum C. jacea H. radicata Succisa pratensis (sites B,VandO) S. pratensis and , andbysummation 2 H. radicata Succisa pratensis and in threemeadow C. dissectum S. pratensis (sites B,L, C. jacea Moench. Cirsium and all C. H. H. ) .

53. Chapter 4 Seed production and seedling establishment

. s e d o c e t i s r o f 0 6 e g a p n o

t o l p t n e n a m r e p n i e t t e s o r g n i r e w o l f r e p s d a e h r e w o l f f o r e b m u n e g a r e v a e h T r a e y d n a , e t i s , s e i c e p s r e p s 1 e l b a t e e S . s n o i t a i v e d d r a d n a t s e t o n e d s r a b r o r r E .

. 2 e r u g i F 54. Chapter 4 Seed production and seedling establishment At siteBallspeciesbut jacea site Nallfourspecieswereused,but their scarcity. This seedadditionexperimentwasperformedatthreemeadowsites. At preceding summer. For seeds ofonespecieswereadded,whichcollectedatthesamesitein in rosettenumberssimilarsizedcontrolplots (noseedsadded).Ineachplot100 November 2000andcountedtheincreaseinrosettenumberscompared tochanges on themossorsoillayerinuntreatedplots of5by50cminNovember1999and To studyestablishment probabilitiesperseedweplaced seedswithinthevegetation Seedling establishment of thefirstseedadditionexperiment,butnumber experiment isalsodescribedbySoonsetal.(2003)). The set-up wasthesameasthat three meadowswherethetopsoilwasremovedbysod-cutting(thissecond addition experimentinautumn2000attensites:sevenundisturbedmeadowsand 1 mapart. of the10blocks(replicates)plots werelayedat20cmintervals. The blockswere made withspecies,sowingyear, andtreatment(controlorseedaddition).Withineach used becausetherewas onlyspatial andnotemporalvariationinthedata. For thesecalculations theresults ofthesecondseedadditionexperiment werenot product ofindependentvariables(Goodman 1960): successive steps wasdeterminedusingthefollowing equationforthevarianceof combinations. The standard deviationoftheresultwholesequence species separately, weusedthemeanof theaveragesofallavailablesite-year seeds thatestablish asone-year-oldseedlings.Foreachofthesesteps andforeach flower ratio,thefractionofseedsthatarenot predatedandtheaveragefractionof heads perfloweringrosette,themeannumber offlowersperflowerhead,theseed- consecutive steps ofthesexualreproductionpathway: themeannumberofflower of seedlingsproducedbyasinglefloweringrosette. To dothis,wemultipliedthe To assesstheoverallimportance ofallthesesteps, wecalculatedtheaveragenumber Pathway analysis the introducedseeds. this experimentweremovedallestablished plants topreventgeneticcontamination by was usedperspeciesforall20sites,sincethedidnotoccuratsites. After In contrasttothepreviousexperimentamixtureofseedscollectedfromdifferent sites plot wasincreasedto70,andthenumberofblockspersiteeightinsteadten. and In ordertoinvestigatealargerrangeofsitesweperformedsecondseed a(y a()a()xvry var(x) y var(y) x var(x)var(y) var(xy) H. radicata =++ were sown(bothyears).Persiteafullexperimental designwas C. dissectum H. radicata 22 C. jacea were sowninbothyears. And atsiteOonly only 50seedsperplotwereused,becauseof and H. radicata C. dissectum were notsownin1999. seeds ineach (1) C.

55. Chapter 4 Seed production and seedling establishment

. p o t n o s r a b d e h c t a h e h t y b d e k r a m s i , s d a e h r e w o l f d e t a d e r p t u o h t i w o i t a r r e w o l f - d e e s e g a r e v a e h t

f o s t n a l p f o , s r a e y d n a s i s n e t a r p a s i c c u S d n a a t a c i d a r s i r e a h c o p y H , m u t c e s s i d m u i s r i C , a e c a j a e r u a t n e C f o l e v e l e h T . s n o i t a i v e d d r a d n a t s e t o n e d s r a b r o r r E .

f i d n i d e t c e l l o c s d a e h r e w o l f p o t f o o i t a r r e w o l f - d e e s e h t d n a d a e h r e w o l f r e p s r e w o l f f o r e b m u n e g a r e v a e h T ) s e d o c e t i s r o f 0 6 e g a p n o 1 e l b a T e e s ( s e t i s t n e r e f

. 3 e r u g i F 56. Chapter 4 Seed production and seedling establishment years. In effect ofyear:in 1999 flowerheadscontained smallernumberofflowersthaninother in othersitesthatyearoryears site.In heads of (Fig. 3).Foreachspecies,site-yearinteractions weresignificant(Table 2): the flower standard deviationbetweenthesite-meansinyear s at leasttwoyears, year meansofaspecies-stepcombination, species, butespecially in radicata with flowernumberin allspecies,butthiscorrelationexistedonly asatrendin radicata than theotherthreespecies,and On average Flower productionandseedset many flowerheads. effects ofsiteandyearareonlysignificantin specific site-yearcombinationsinallspecies(Fig2.and Table 2).However, themain The numberofflowerheadsperfloweringrosettediffered significantlybetween Flower headproduction Results in whichforeachspecies-stepcombinationseperately compare theCVsinspace andtimeforeachstepspeciesseparately: temporal,m To analyzetheobservedvariationinsexualreproductionpathway, we The fractionofflowers thatsetseedinaflowerheadwaspositively correlated CV CV was intermediatetothelasttwospeciesand differed onlyfrom H. radicata (Table 2).Seedpredation significantlyreducedseed-flower ratios inall C. jacea the standard deviationbetweentheyear-meansinsite spatial temporal C. dissectum = = g 1 and 1 f ∑ n1 g = m1 ∑ x g for instance hadfewflowersinsiteO2000,but highernumbers f = s) (s means s) (s x the numberofyearswithobservationsinatleasttwosites, S. pratensis spatial,n means temporal,m C. jacea. had significantlylargernumberofflowersper flowerhead main differences betweensitesweresignificant. In thatspeciesinsect larvae regularlyconsumea C. jacea f the numberofsiteswithobservationsin C. jacea more than n . , inwhich2000wasayearwith C. dissectum x means S. pratensis is themeanofallsite- m there wasamain , and . C. dissectum Hypochaeris s spatial,n the (3) (2) H. 57. Chapter 4 Seed production and seedling establishment C. C. and . Cirsium and , Figure 4. U Y C. jacea F H Q C (Fig. 4). The first (Fig. 4). C. jacea K S. pratensis f h Centaurea jacea Centaurea m U Y F H. radicata H , but not in Q C and K f h m . The first experiment in the meadows B, N, and . U C. dissectum Y F S. pratensis H Q C K f h Succisa pratensis m and U Y and slightly so in F H Q B N O B N N B O N B N B O N C K f h C. jacea C. dissectum H. radicata S. pratensis H. radicata C. dissectum C. jacea TSR MeadowsTSR Meadows TSR Meadows TSR Meadows TSR m H. radicata 0 0 8 4 8 4

-4 -4 changes in rosette number in seeded plots were not statistically different were not statistically number in seeded plots changes in rosette

16 12 16 12

: 1999 was also a bad year for seed set in this species. year for seed also a bad : 1999 was Exp. 2 Establishment [%] Establishment 2 Exp. Exp. 1 Establishment [%] Establishment 1 Exp. only sown in 2000 were not seeded in site N in 1999. In the second experiment seeds were Hypochaeris radicata and , H. radicata Please note that negative in meadows with the top soil removed (TSR), and in undisturbed meadows. than in the in the control plots values can occur when the change in rosette number was more positive 60. 1 on page Table site codes see For seed addition plots. large part of the developing seeds. Significant site-year interactions and main effects interactions site-year seeds. Significant of the developing large part in only significant factor year was The in most species. were present of sites Seedling establishment was highest in Seedling establishment meadows of which which also included The second experiment, plots. from the control for all species. In the very open seed addition effects the top soil was removed, shows meadows for is higher than in undisturbed (top soil removed) establishment habitats C. jacea dissectum significant to meadows, revealed experiment, which was restricted seed addition In 2). two species (Table of seed addition only in these effects dissectum The average percentage of sown seeds that is established as seedling after 12 months per species, as seedling after of sown seeds that is established The average percentage The species are deviations. sites, and years. Error bars denote standard dissectum in 1999 (black) and 2000 (dark grey). Since no local seed were available O was started 58. Chapter 4 Seed production and seedling establishment pratensis seed onayearlybasis.In indicating thateachfloweringrosettewouldproduceoneormorenewrosettesvia that resultfromthesecalculationsarelargerthan1(respectively1.9and3.3), occurred inthenumberofflowerheadsperfloweringrosette temporal andspatial aspects ofseedlingestablishment, relativelyhighvariation temporal differences inseedlingestablishment werealsorelativelyhigh. Apart from meadows inthesecondexperimentwasashighthatfirst. The CVsofthe compared todata ontheothersteps, variationinestablishment probabilitiesbetween establishment. Although theyarebasedonasmallsampleofyearsandsites species (Fig.4). The highestCVswerethoseofthespatial differences inseedling between sites,bothcompared totheoverallmeanvalueofthatstepinaparticular the sexualreproductionpathway thehighestvariationoccurredbetweenyearsand than 1(0.23),andin seedling of1to4( growth. We calculatedthateveryfloweringrosettehasa chanceofproducinga recruitment ratein species-specific casessuchasseedsetin perennials wastheestablishment ofseedlings,followed byflowerproduction,and The mostimportant bottleneckinthesexualreproductionpathway ofallfourstudied Bottlenecks inthesexualreproductionpathway Discussion seed productionstep.In Beside seedlingestablishment, strikingdecreasesalsooccurin quantitatively (Fig.5),thelargestlossestake placeintheseedlingestablishment step. than differences inclonality. the differences inlongevityarelessimportant for thepattern ofsexualreproduction and when comparing species-averagesexualreproduction pathways (Fig.5), probability thatagenotypeisinvolvedinrecombination. Itisalsoremarkablethat, years andmayexistofmanyfloweringrosettes, therebyincreasingtheper-year- Cain 1990;Watkinson &Powell1993).Besides, ageneticclonemayliveformany population sizeand,moreimportantly, tomaintain geneticdiversity(Eriksson1989; were used.However, clonalspeciesneedlessrecruitmentperyeartomaintain this probabilityisstilllessthanonewhenthe highestpopulationmeansforeachstep H. radicata The temporalandspatial coefficients ofvariation(CVs)showinwhichstep When thedifferent steps ofthesexualreproductionpathway arecompared The lowprobabilities of seedlingestablishment canonlybeexplained fora and theseed-flowerratioin have verysimilarpatterns, whereastheydiffer inlongevity. Perhaps C. jacea C. dissectum C. dissectum H. radicata C. jacea ) or1to38( however thecalculatedseedlingnumberwaslower even 0.027. and and C. dissectum C. jacea S. pratensis C. dissectum C. dissectum which bothshowextensiveclonal . the averagenumberofseedlings ) peryear. Inthe latterspecies . We foundthelowest C. dissectum C. jacea S. pratensis and in the S.

59. Chapter 4 Seed production and seedling establishment

Below: The spatial and temporal coefficient of variance of each step is given per species (see materials and methods). and materials (see species per given is step each of variance of coefficient temporal and spatial The Below:

site-year mean. The low scenario in in scenario low The mean. site-year end with zero seedlings. zero with end dissectum C. and jacea C.

are progressing standard deviations (see methods), and the dashed lines represent scenarios in which all steps are either the l the either are steps all which in scenarios represent lines dashed the and methods), (see deviations standard progressing are owest or the highest observed highest the or owest

Above: The average (solid line) sexual reproduction pathway per species (above), based on the means of all available site-year site-year available all of means the on based (above), species per pathway reproduction sexual line) (solid average The Above: combinations. The error bars error The combinations.

Figure 5. Figure

150 150 150 150 150 150 150 150 150 150 150 150

pta CV CV CV Spatial Spatial Spatial

100 100 100 100 100 100 100 100 100 100 100 100

Temporal CV CV CV Temporal Temporal Temporal

CV CV CV CV

50 50 50 50 50 50 50 50 50 50 50 50

0 0 0 0 0 0 0 0 0 0 0 0 flower heads flower flower heads flower flower heads flower flower heads flower unpredated unpredated unpredated unpredated seedlings produced seedlings seedlings seedlings produced produced produced flowering flowering flowering flowering seeds seeds seeds seeds

rosette seeds

rosette rosette rosette seeds seeds seeds flowers flowers flowers flowers

0.01 0.01 0.01 0.01 0.01 0.01 0.01 0.01

Centaurea jacea Cirsium dissectum Hypochaeris radicata Succisa pratensis pratensis Succisa Succisa radicata radicata Hypochaeris Hypochaeris dissectum dissectum Cirsium Cirsium jacea jacea Centaurea Centaurea

0.1 0.1 0.1 0.1 0.1 0.1 0.1 0.1

1 1

1 1 1 1 1 1

Number Number Number Number

10 10 10 10 10 10 10 10

100 100 100 100 100 100 100 100

1000 1000 1000 1000 1000 1000 1000 1000 60. Chapter 4 Seed production and seedling establishment control sites. Top soilandvegetation removal createsopportunitiesforseedlings, but study, whentheycompared seedlingrecruitmentbetweentheirCzech sod-cutand probabilities. predation arethereforeprobablymoreimportant forexplainingtheselowrecruitment germination). Otherexplanationslikemicrosite limitation andpost-dispersalseed species tobelowerthan50%onlyin experiments. SoonsandHeil(2002)reportgermination percentages forthesamefour small part byinviabilityofintact seedsthatwereusedfortheseedaddition inthewinterof1997/1998. winter of1999/2000,‘h’ weresod-cutinthe and‘f’ Meadows aremownannually, roadvergesonceortwiceperyear. Sites‘m’ Code, name,andtypeofthesitesthatwereusedinthisstudy. Per regioncoordinatesaregiven. Table 1. Kotorová andLepš(1999) foundsimilarpatterns for Achterhoek (51°59' N, 6°25' E) Type Kop van Overijssel (52°37' N, 6°10' E) Veluwe (52°17' N, 5°44' E) Name Code Gelderse Vallei (52°01' N, 5°35' E) Region m De Maandag top soil removed soil top soil removed top soil removed top verge road verge road Maandag De meadow verge road heide Nijkampse verge road Stuifveen m meadow h Sonderenweg Nijkampseweg f meadow W Kooigootsweg meadow Eibergseweg J Veller ‘t G meadow meadow E Vildersveen meadow Stelkampsveld Y meadow meadow U Breedslatsweg (heather) S Koolmansdijk meadow R Koolmansdijk Q Konijnendijk Heksengat O heide N Nijkampse Stuifveen K H F idne road removed soil verge top meadow meadow C meadow Veerslootlanden Middenweg Veerslootlanden v Stadsgaten D Staphorst Olde V T P meadow meadow Leemputten meadow L meadow Groot Zandbrink Allemanskampje meent Bennekomse Z A B productive part) (more Koolmansdijk S. pratensis (for whichspeciestheyfound20% meadow S. pratensis as inour 61. Chapter 4 Seed production and seedling establishment . S. and H. radicata Abia sericea which is a very and C. dissectum C. jacea towards the edge of their , however, did benefit from , however, C. dissectum and was found at several sites. however seed set is low and only however seed C. jacea and S. pratensis C. dissectum Cirsium heterophyllum H. radicata is the caterpillar-like larva of the sawfly is the caterpillar-like larva of and still is expected to establish better in sod cut areas but better to establish still is expected , with more flowers, tended to have higher predation , with more flowers, tended S. pratensis seems to be a relatively good establisher in closed vegetations as closed vegetations in relatively good establisher seems to be a C. jacea C. dissectum Cirsium acaule seedling establishment for up to a year after sod cutting (Dorland et al. (Dorland sod cutting after for up to a year establishment seedling (18% of the seeds on average). In accordance with Fenner et al. (2002) larger (18% of the seeds on average). Seed production is not only important for within population dynamics, but seed Seed production is not only important Flower production per flower head largely determines the number of seeds per flower head largely Flower production well, whereas pratensis number, higher seedling of significantly the lack explain may partly This 2003). cutting. sod at least ten months after the seed addition experiment although we started Succisa pratensis in low numbers. still apparently also causes abiotic stress due to ammonium accumulation (de Graaf et al. 1998; (de accumulation due to ammonium abiotic stress also causes to hamper has been shown effect This et al. 2003). Dorland Temporal and spatial variation and spatial Temporal limited is much debated The question whether populations are seed or microsite recruitment is the (Eriksson & Ehrlén 1992; Coulson et al. 2001). Within a population, and a relative increase in one product of seed production and seedling establishment, increase in the other as long as density dependent as a relative is as important of relative between populations comparison processes can be ignored. However, sites, we find that meadows are more microsite can be made. Comparing limitation limited than sod-cut areas in the two more common species, However, all these losses are less severe than these in the seedling establishment all these However, step. dynamics, since it partly for regional population number is also a major parameter problems caused (Soons & Heil 2002). Considering determines colonization capacity due to displacements range possibly also by habitat but fragmentation, by habitat as be as important climate change, seed production and seed dispersal may therefore outside probabilities may be different establishment Because seedling establishment. of the different importance the site of the seed source, a regional evaluation of the would require modeling of both local and spatial of sexual reproduction steps dynamics (Higgins & Cain 2002). (McGee 2001), which feeds specifically on (McGee 2001), which feeds specifically probabilities (n=375; F=3.38; p=0.067). This effect was rendered insignificant however This effect p=0.067). probabilities (n=375; F=3.38; A into account. were taken when site and year differences cause of destruction of buds of flower heads in weakly correlated with flower production. Jump and Woodward (2003) report lower (2003) report production. Jump and Woodward weakly correlated with flower seed set for similar processes are at work in range in the UK. Perhaps Losses due to seed predation are most significant in rare species in the Netherlands. C. jacea flower heads of vegetation removal. Their ability to utilize open spaces may be part of the explanation of the may be part ability to utilize open spaces Their removal. vegetation species are relatively common. why these two flowering rosette. In produced per 62. Chapter 4 Seed production and seedling establishment adding 10(Fowleretal.1998). seed-flower ratiosarcsinetransformed,andtheincreasesinrosettenumberwerelogtransformedafter To increasehomogeneity amongthedata, thenumberofflowerheadswerearcsinhtransformed, year andsiteasrandomfactors,flowernumber, andpredation(absentorpresent)ascovariates. Statistics modelsperspecies.Whenreportedseedadditionisbuiltinasafixedfactor, oftypeIII ANOVA Table 2.

Centaurea jacea Cirsium dissectum Hypochaeris radicata Succisa pratensis Source df MS F df MS F df MS F df MS F # Flower heads per Flowering Rosette Site (S) 2 10.573 13.60 ** 2 0.276 0.75 2 1.210 2.05 4 10.631 1.67 Year (Y) 4 5.816 7.92 * 4 0.244 0.63 4 1.025 1.77 4 13.080 2.58 (*) S * Y 6 0.761 2.72 * 8 0.439 6.78 *** 7 0.755 2.25 * 15 7.578 15.77 *** ( * )

# Flowers per Flower head = p<0.1;*=p<0.05;**=p<0.01;***=p<0.001 Site (S) 4 12672 26.55 *** 9 2395 1.94 5 7209 5.01 (*) 12 7112 7.59 *** Year (Y) 4 697 1.34 4 12475 8.57 ** 3 4900 3.66 (*) 4 1126 1.75 S * Y 8 5313.29**1415264.04***5 14213.07* 2410367.44***

Ratio #Seeds / #Flowers Flower number 1 891 4.05 * 1 485 3.04 (*) 1 3328 23.16 *** 1 797 4.22 * Predation 1 36071 163.99 *** 1 8180 51.34 *** 1 438 3.05 (*) 1 1468 7.77 ** Site (S) 4 4087 6.68 ** 9 5346 8.35 *** 5 1020 6.40 * 12 3062 2.79 * Year (Y) 4 220 0.30 4 5726 7.73 ** 3 327 2.06 4 978 1.24 S * Y 8 749 3.41 ** 14 813 5.10 *** 5 160 1.11 24 1253 6.63 ***

Increase Number rosettes (Seed addition Experiment 1) Seed Addition 1 0.002 0.25 1 0.001 0.19 1 0.111 4.24 * 1 0.134 16.47 *** Site (S) 2 0.045 0.90 1 0.156 348.58 * 1 - - 1 0.002 0.02 Year (Y) 1 0.026 0.53 1 0.003 6.19 1 - - 1 0.000 0.00 S * Y 1 0.049 5.27 * 1 0.000 0.14 0 - - 1 0.097 11.88 **

Increase Number rosettes (Seed addition Experiment 2) Seed Addition (SA) 1 0.156 37.15 *** 1 0.009 9.45 ** 1 0.764 92.51 *** 1 0.655 71.92 *** Site within T 8 0.022 5.32 *** 8 0.004 4.75 *** 8 0.046 5.52 *** 8 0.076 8.39 *** Type (T) 1 0.004 5.96 * 1 0.019 4.28 (*) 10.370 8.12 * 1 0.027 0.36 SA * T 1 0.127 30.16 *** 1 0.004 4.58 * 1 0.162 19.68 *** 1 0.034 3.77 (*) 63. Chapter 4 Seed production and seedling establishment . - . High spatial Primula vulgaris S. pratensis and . - J Ecol 78: 27-46. C. jacea Solidago altissima . - Int J Plant Sci 156: 522-529. and flower head production in and flower head J Ecol 92: 5-14. Amphibromus scabrivalvis Appl Ecol 38: 204-216. in responses to management. - J dispersal versus microsite limitation Bot Neerl 47: 89-111. Acta of appropriate soil conditions. - The importance 607-618. - Ecology 81: limitations. Appl Ecol 40: 804-814. cutting hampers the restoration of degraded wet heathlands. - J Oecologia 91: 360-364. species: Aspecies: between two flower types. - Plant Biol 5: 194-202. comparison Reduced reproductive success in small populations of the self-incompatible Although the largest reductions took place in germination and seedling germination and place in reductions took the largest Although C. dissectum Cain ML 1990. Models of clonal growth in Cheplick GP mass, and allocation in clones of the perennial grass Plasticity of seed number, 1995. MJ & Pywell RF 2001. Colonization of grassland by sown species: Coulson SJ, Bullock JM, Stevenson - Blackwell Science, Oxford. Crawley MJ 1997. Plant Ecology. PJM, Bobbink R & Roelofs JGM 1998. Restoration of species-rich dry heaths: de Graaf MCC, Verbeek review of methods and model de Kroon H, van Groenendael JM & Ehrlén J 2000. Elasticities: a JTADorland E, Bobbink R, Messelink JH & Verhoeven sod 2003. Soil ammonium accumulation after - Oikos 55: 231-238. Eriksson O 1989. Seedling dynamics and life histories in clonal plants. of recruitment in plant populations. - Eriksson O & Ehrlén J 1992. Seed and microsite limitation Fenner M, Cresswell JE, Hurley RAT & Baldwin 2002. Relationship between capitulum size and pre- References success in four cleistogamous Berg H 2003. Factors influencing seed : ovule ratios and reproductive L, De Bruyn LTriest Hermy M, Rossum F, Van De Blust GDE 2004. & Brys R, Jacquemyn H, Endels P, CVs may further enhance local adaptation, whereas high temporal CVs may select whereas high temporal enhance local adaptation, CVs may further against specialization. Acknowledgements Berendse and the Nature Conservation and Plant Ecology are grateful to Frank We Keser of this manuscript. Lidewij on earlier drafts PhD-group for their useful comments rangers of State The park flower and seed counts. assisted with the numerous Ermelo kindly gave permission to perform this study in Forestry and the community of for research was funded by the Netherlands Organization This their meadows. 805.33.452). Scientific Research (NWO-project recruitment, it would be premature to conclude that this is the only important that this is to conclude be premature it would recruitment, within- high Rather, of perennials. pathway sexual reproduction in the local bottleneck sexual of the steps indicate which of variation coefficients temporal population et al. play grounds for selection (Koenig are the most important reproduction pathway within- that had the highest it was again seedling establishment 2003). In our study ratio considerable as well in seed-flower but temporal variability was population CVs, in 64. Chapter 4 Seed production and seedling establishment Picó FX&RibaM2002.Regional-scaledemographyof Klinkhamer PGL,deJong TJ &vanderMeijdenE1988.Production,dispersalandpredationofseeds Jump AS &Woodward EI2003.Seedproductionandpopulationdensitydeclineapproachingtherange- 1999.Modelingseeddispersalbywindinherbaceousspecies.-Oikos87: Jongejans E&SchippersP comparative studyofseednumber, seedsize,seedlingsizeand &ErikssonO2000. A Jakobsson A Isselstein J, Tallowin JRB&SmithREN2002.Factorsaffecting seedgerminationandseedling 2002.Spatially realisticplantmetapopulation modelsandthecolonization- Higgins SI&CainML Herrera CM2000.Flower-to-seedlingconsequencesofdifferent pollinationregimesinaninsect- Hartemink N,JongejansE&deKroonH2004.Flexiblelifehistoryresponsestoflowerandrosettebud 1977.Populationbiologyofplants. - Academic Press,London. Harper JL 1960.Ontheexactvarianceofproducts. -J Am Stat Assoc 55:708-713. Goodman LA Watkinson AR &PowellJC1993.Seedlingrecruitmentand the maintenanceofclonaldiversityinplant Špacková I,Kotorová I&LepšJ1998.Sensitivityofseedlingrecruitmenttomoss,litteranddominant Soons MB,MesselinkJH,JongejansE&HeilGW2003. Fragmentation andconnectivityofspecies- Soons MB&HeilGW2002.Reducedcolonization capacity infragmentedpopulationsofwind- 1993.Comparative plantdemography:Relative Silvertown J,FrancoM,PisantyI&Mendoza A Schupp EW1990. Annual variationinseedfall,postdispersal predation,andrecruitmentofa Oostermeijer JGB,LuijtenSH,KrenováZV&denNijsHCM1998.Relationships betweenpopulation comparison ofstereomicroscopeandimageanalysisfor Mix C,PicóFX&OuborgNJ2003. A Meyer AH &SchmidB1999.Seeddynamicsandseedlingestablishment intheinvadingperennial Menges ES1995.Factorslimitingfecundityandgerminationinsmallpopulationsof McGee K2001. 1995.Effect ofinflorescence-feedinginsects onthedemographyandlifetime Louda SM&PotvinMA Lamont BB&RuncimanHV1993.Firemaystimulateflowering,branching,seedproductionand Kotorová I&LepšJ1999.Comparative ecologyofseedlingrecruitmentinanoligotrophicwet meadow. Koenig WD,KellyD,SorkVL,DuncanRP, ElkintonJS,PeltonenMS&Westfall RD2003.Dissecting Biol 12:1042-1053. in thebionnial edge of 362-372. recruitment ingrasslandplants. -Oikos88:494-502. establishment offen-meadowspecies.-RestorEcol10:173-184. competition trade-off. -JEcol90:616-626. pollinated shrub.-Ecology81:15-29. removal inthreeperennialherbs.-Oikos105:159-167. dispersal seedpredationbyinsectlarvaeincommon Asteraceae. -Oecologia130:72-77. ouain:Acomputersimulation of populations: A removal inanoligotrophicwetmeadow. FoliaGeobot33:17-30. University. Spatial andtemporalcharacteristicsofthecolonizationprocessinplants. PhDthesis,Utrecht rich semi-naturalgrasslands.-In:SoonsMB(ed), Habitat fragmentation andconnectivity. dispersed grasslandforbs.-JEcol90:1033-1043. perennials. -JEcol81:465-476. importance oflife-cyclecomponents tothefiniterateofincreaseinwoodyandherbaceous Neotropical tree.-Ecology71:504-515. dynamics inapreglacialrelictspecies.-PlantEcol161:1-13. and habitat characteristicsandreproductionoftherare quantifying fruittraits. -Seed Technol 25:12-19. Solidago altissima (Caryophyllaceae), ararehummingbird-pollinatedprairieforb.- Am MidlNat133:242-255. fitness ofanativeplant.-Ecology76:229-245. seedling establishment intwoKangaroopaws (Haemodoraceae).-J Appl Ecol30:256-264. - JVeg Sci10:175-186. behavior. -Oikos102:581-591. components ofpopulation-levelvariationinseedproductionandtheevolutionmasting Cirsium Abia Sawflies atMillMeadow, DrakesBroughton.-Worcestershire Rec6. Cirsium vulgare species. -NewPhytol160:349-358. under different experimental treatments. -JEcol87:28-41. . -JEcol76:403-414. Ranunculus repens Ramonda myconi Gentiana pneumonanthe . JEcol81:707-717. : Remnantpopulation L. -Conserv Silene regia 5 Space versus time variation in the population dynamics of Succisa pratensis and two accompanying species

Eelke Jongejans & Hans de Kroon

Summary

Many plant species are currently restricted to small and isolated populations due to habitat destruction and habitat fragmentation. To obtain sufficient data for management temporal variation has often been substituted with spatial variation, but it is still largely unknown whether there are substantial differences between these two types of variation. In order to fill this gap we studied the demography of the declining perennial Succisa pratensis at five sites over four years (1999-2003). Furthermore, to contrast with the long-lived life cycle of S. pratensis, two coexisting species with similar life cycle components (fecundity, survival, growth and ramification) were studied in the same plots: the shorter- lived Hypochaeris radicata and the long-lived, but more clonal Centaurea jacea. The S. pratensis populations showed high elasticities for adult rosette survival. Life table response experiment analysis (LTRE) revealed that temporal and spatial variation in the life history components were qualitatively different: fecundity contributed most to variation in population growth rate (λ) between sites, but growth contributed most to variation in λ between years. In bad years most of the life history components were reduced, whereas bad sites had low lambdas for different reasons: all major components were reduced in the most productive site, while in two other bad sites the negative contributions to the variation in λ by some components (fecundity or growth) were partly compensated for by positive contributions by other components (growth or stasis). While the results of H. radicata were similar to those of S. pratensis, C. jacea showed a reverse pattern: in bad years negative contributions by some life history components were buffered by positive contributions by other components, while this buffering did not occur in bad sites. Our analysis shows that the population dynamics of perennial plants may respond differently to temporal than to spatial variation in site-specific environmental conditions. Moreover, co-occurring species with similar life history options responded differently to the same spatiotemporal variation. We conclude that temporal dynamics can not readily be exchanged with spatial dynamics in population viability analyses and management studies of a species. 66. Chapter 5 Space versus time in population dynamics have shownthattherelative importance, orelasticity, ofparticular matrix elements for importance ofdifferent lifehistorycomponents between sitesoryears.Severalauthors attempt tofillthisgap. covary inresponsetotemporalorspatial variation isstilllacking. This paper isan systematicanalysisofwhetherandhowthelifehistorycomponents ofaspecies A and spatial variationinlifehistorycomponents is interchangeable (Horvitzetal.1997). Ascencio etal.2003).However, veryfewstudieshaveinvestigatedwhethertemporal Damman &Cain1998;Valverde &Silvertown 1998;Bühler&Schmid2001;Quintana- tempting tosubstitutetemporalvariation by spatial variation(Silvaetal.1991; Valverde &Silvertown1998).Inorderto quickly gatherdata formanagementitis seem torelymoreonsurvival(Oostermeijer etal.1996;Menges&Dolan1998; associated withincreasedimportance ofseedproduction.Inbadyearsorsitesplants plants (Silvertownetal.1993).Furthermore,withinspecies,goodyearsorsitesare (see foranexamplePicóetal.2003),andsurvivalofindividualsmoreinlong-lived seedling establishment contributemoretothepopulationgrowthinshort-livedspecies (Oostermeijer etal.1996).Generalpatterns between specieshavebeenshown: knowledge isneededabouttherelativeimportance ofdifferent lifehistorycomponents 1996; Menges&Dolan1998;Watkinson etal.2000). spatiotemporal variationinsite-specificenvironmental conditions(Oostermeijeretal. and climaticfactorssuchasrainfall. The dynamicsofplantpopulations respondtothis succession, butdifferences betweenyearsdoexistduetovariation inmowingdate Soons etal.2003;Vergeer etal.2003). Annual mowingisastabilizing factorasithalts conditions, vegetation compositionandmanagementregime(Soons&Heil2002; Individual sitesdiffer infragmentation history, theincreaseinproductivity, abiotic 1992; Bakker&Berendse1999;Lucassenetal.2003;vanderHoek2004). have irreversibleoratleastlong-termeffects onthesemeadows(Berendseetal. al. 2003;OostermeijerVergeer etal.2003). The processesoutlinedabovemay Tilman etal.1994;Roem&Berendse2000;HanskiOvaskainenLucassen remaining populationshavebecomesmallerandmoreisolated(Saundersetal.1991; neighbouring agriculturalfieldscausenutrientenrichmentandacidificationthe and acidificationarecausedbynitrogendepositionenrichedgroundwaterfrom intensified landuse:thewatertables areloweredbydrainage,andnutrientenrichment grasslands arethreatenedbyenvironmental changesduetohabitat fragmentation and Many plantspeciesthatarecharacteristicofnutrient-poorspecies-richmoist Introduction Keywords: temporal variation Population matrixprojection modelsprovideapowerfultoolfor studying the In ordertofine-tunemanagementfordecliningplantspecies,thorough demography, elasticity , three-way fixed-factorvariancedecomposition , life historycomponents , , spatial and trade-off 67. Chapter 5 Space versus time in population dynamics Figure 1. (Silvertown et al. 1993; (Silvertown λ V V V L L L B B B Sites O O O ), is correlated with ), is correlated Centaurea jacea Hypochaeris radicata Succisa pratensis Succisa λ N N N , and bootstrapped 95% confidence intervals for all constructed , and bootstrapped λ

.75 .50 .25 .75 .50 .25 .75 .50 .25

1.75 1.50 1.25 1.00 1.75 1.50 1.25 1.00 1.75 1.50 1.25 1.00

Population growth rate, rate, growth Population λ the projected population growth rate ( growth rate population the projected Oostermeijer et al. 1996; de Kroon et al. 2000). However, differences in elasticities do in elasticities differences al. 2000). However, de Kroon et et al. 1996; Oostermeijer have caused that components in life history actual changes to coincide with not need local are only as elasticities to occur, growth rate in population the differences investigate what To Horvitz et al. 1997). matrix (Moloney 1988; certain properties of a Projected population growth rate, matrices. Site and year codes: N = Konijnendijk, O = Koolmansdijk, B = Bennekomse Meent, Lmatrices. Site and year codes: N = Konijnendijk, O = Koolmansdijk, = Leemputten, V = Veerslootlanden, = 1999->2000, >2003. = 2000->2001, = 2001->2002 and = 2002- 68. Chapter 5 Space versus time in population dynamics populations ofthesespeciessimultaneously with variation in stem (Harteminket al. 2004). rosettes areformedon thewoodyrootstock atthesoilsurfacealongside theflowering and flowersfromJune untilautumn.Duringandafter flowering, vegetative side- monocarpic shoots (Tamm 1956). stalks variesfromonetoseveral.Floweringstarts in Juneandcontinuesuntilautumn. formed clonallyinthecentreofmainrosette; thenumberofleaflessflowering perennial (deKroonetal.1987;Fone1989). Its floweringstalks andnewrosettesare one yearmostshortstolonsareseverednaturally. July tillOctober. Axillary budsoccasionallyformsiderosettesonshortstolons. After from axillarybudsofoldleavesandflowers areproducedinflowerheadsfromlate Leaves growinalternatingpairs andstay aliveforalmostayear. Flowering stalks arise characteristic ofCirsiodissecti-Molinietumcommunities(Schaminéeetal.1996). perennial withpolycarpicrosettes(Adams1955;Hooftman &Diemer2002)and nutrient-rich surfacewaterandrestoringtheupwellingofbase-richgroundwater. mowing andhayremoval,restoringhydrologicalconditionsbyhaltinginfluxof management isaimedatconservinglocalfloristicdiversityintwoways:annual cattle grazing(Pegtel1983;Grootjansetal.2002).Currentlyconservation In formerdaysthesewetgrasslandsweresolelyusedforhaymakingandextensive are nowrestrictedtosmallnaturereserves(Berendseetal.1992;Soons2003). in the20 Nutrient-poor, species-rich,moistmeadowsdeclinedinabundance bymorethan99% Study system Materials andmethods Hypochaeris radicata studied twoco-occurringbutlessdecliningspecies:theshort-livedperennial patterns ofthelong-lived determine thespatial andtemporalvariationin four years,weinvestigatewhethervariationinthesamelifehistorycomponents actually causedchangesin lived perennialherb, decomposition technique)contributionsandelasticitiesforanoveralldeclininglong- Caswell 2000). In thispaper wecompare LTRE (lifetable responseexperiment,avariance Centaurea jacea Hypochaeris radicata Succisa pratensis th century intheNetherlandsduetodrainage,cultivation,andfertilization, λ into contributionsoftheunderlyingcomponents (Horvitzetal.1997; Succisa pratensis and thelong-lived,clonal L. Moench (Devil'sbitScabious,Dipsacaceae), isalong-lived S. pratensis s.l. λ , techniqueshavebeendevelopedthatdecomposethe L. (Cat'sear, Asteraceae) isarelatively short-lived (Knapweed, Asteraceae), isalong-livedperennialwith Centaurea jacea Centaurea jacea with twodifferent lifehistorystrategies,wealso . After studyingits demographyinfivesitesfor λ . Furthermore,inordertocontrastthe S. pratensis Centaurea jacea has asingleapicalflowering stalk and H. radicata in asubsetofthesites. . We monitored have awider 69. Chapter 5 Space versus time in population dynamics ) t ) 2 S. able 1) s as Figure 2. alks were s was lef , the sites aurea jacea S. pratensis Cent ), survival of side s in the three-way C ter ( and a ) or af was studied in the same a ). in the same plot a R , as all flowering st ), ramification ( re studied from populations we G m N, and O respectively) of 1x1 , H. radicat C. jacea C. jacea Hypochaeris radicat C. jacea , ), growth ( F ), and retrogression ( S Hypochaeris radicat low to high productivity Arranged from ) grouped by the life history component q asis ( and three these plot border zone of 15 cm around a s: fecundity ( d e or disturbe ore productiv o occur in m and als ), st A T fect ( Succisa pratensis H. radicat in sites L, N, and O, and r grasslands in the annually mown, nutrient-poo All sites were . April 1999. ensis S. prat , three s ecies in a plot were mapped with 5 mm All rosettes of the studied sp ed before ( The other two species flower as mown unexpectedly early in 2001 (T in sites B, N, and O. Site O w S. pratensis ablished in S. pratensis our in sites B, L, V s (five, eight, three, six, and f s as 6°08' E). eerslootlanden (V) (52°36' N, atrices of the three species in TRE analysis of the variation in population growth rate between the m pratensis ition matrices of causing the loss of two trans removed. they were censused as in the other years. Permanent mowing in 2001 in site O, and plot V plot es for each year-to-year transition matric 3, resulting in 20, 12 and 10 1999 until 200 ctively species, respe st of the Netherlands. Centre and Ea N, 5°36' E), nnekomse Meent (B) (52°00' en (L) (52°17' N, 5°44' E), Be were Leemputt °33' E) and lmansdijk (O) (52°01' N, 6 ) (52°02' N, 6°26' E), Koo Konijnendijk (N L sites N and O over four years. were est undisturbed. t Permanent plo Five ion than distribut erges. ds and road v grasslan (a): Mean elasticity per species ( grouped by six life history component rosettes formed in the previous year ( (b): Positive and negative species ef 70. Chapter 5 Space versus time in population dynamics every yearduringflowering(see Table 1forexactcensusdates).For accuracy, andtheirestablishment, survival,growth,andfloweringwerestudiedonce distance to25,10and40mmforrespectively distance apart. Basedonfieldexperienceof the species,wearbitrarilysetthis nearest old,non-seedlingrosette,whenthey werenotfurtherthanaspecies-specific disturbance. Therefore weassignedallnewarisen,non-seedlingrosettestothe possible todeterminefromwhicholdrosetteanewclonaloriginateswithout have predominantlytransientseedbanks(Thompson etal.1997). (Bühler &Schmid2001).Noseedstage ismodelledasthesespeciesareknownto The firstfourclasseslargelycorrespondtoanearlierclassificationfor Seedlings( 1 status andtheformationofside-rosettes.We distinguishedsixclasses: Classification ofindividualrosetteswasbasedonacombinationsize,flowering Stage andsizeclasses disseminated. phenological stage atthemomentofcensusrecorded:bud,flowering,seeded,or and stemleaves.Finallythenumberofflowerheadswascountedtheir number andmaximumlengthofrosetteleaves,thefloweringstems (>250) rosettesweremeasuredintheremainingplots. Plantsizewasquantifiedasthe andNrespectively, becauseenough randomly omittedsixandthreeplots insitesL Smallvegetative rosettes( 2 Floweringside-rosettes ( 6 Vegetative side-rosettes( 5 Floweringrosettes ( Largevegetative rosettes( 4 3 (n=63 for In thosecaseswhereitwaspossibletodetermine connectionswithoutdisturbance stems. radius, because89% of the200knownramifiedrosetteswereattached toflowerings non-flowering rosetteswhennoflowering hadbeenpresentwithinthe40mm specific distance foreachspecies.In of thedistances betweenoldandnewrosetteswerenotlargerthanthespecies- and at leastonefloweringstem. sense, andthatmayactuallyalsobeolderverysmallrosettes. longer than5cm( no floweringstems. irrespective ofrosettesize. Ramification resultedinnewrosettesclosetooldrosettes.Itisnotalways S. pratensis H. radicata sdl ) aretinyrosetteswithnoleaveslongerthan2cm( , n=18for ) or5cm( S. pratensis flow ) haveatleastonefloweringstemandaregrouped side.flow side.veg H. radicata lrg sml C. jacea ) havelongerleavesthansmallrosettes. ) arelargerthanseedlingsbutwithnoleaves and C. jacea ) arenewlyformedbyramificationandhave ) arenewlyformedbyramificationandhave H. radicata .Nt htw s seln’inabroad ). Notethatweuse‘seedling’ and n=200for S. pratensis side-rosettes wereonlyappointedto ) or12.5cm( , C. jacea H. radicata C. jacea ), morethan92% S. pratensis and S. pratensis S. pratensis ). C. jacea we . 71. Chapter 5 Space versus time in population dynamics th ), j C asis t ), and S Figure 3. ferent life asis ( d s were classifie ), st ), ramification ( T G e nsition from th s the tra ), growth ( F ). Please note that the elasticities of a S+R ent ) repres ) and the summed elasticity of dif ij surviving rosette. S age class by a agation and survival of side rosettes formed in lement The 36 matrix e a . t+1 s: fecundity ( ), clonal prop G r category in yea ), growth ( F th i ( each element in which A from a smaller and h is defined as the transition able 2). Growt ), and survival and retrogression ( to the ) (T t C+T R s: fecundity ( arger or flowering st age class to a l = p<0.1; * = p<0.05; ** = p<0.01; *** = p<0.001 ) rization aramete * ( ative st veget ress survive lass, and rosettes that retrog osette remains in the same c means that a r in the hierarchy of classes. but get smaller ng life history component into the followi ction cted 6x6 proje ties we constru sition probabili ar-to-year tran calculated ye With the s of the form matrice r category in yea Matrix p Correlation between projected population growth rate ( cycle component retrogression ( in the previous year ( -rosettes that were formed survival of side the previous year ( fore negatively correlated. Each matrix sum to one, and that the x-axes of the four diagrams are there ssion lines are shown for each dot represent an individual transition matrix. Significant, linear regre species. 72. Chapter 5 Space versus time in population dynamics offspring classesasramificationwasarareeventin smaller thansixintheadultclasses in thesamepopulation. The numbersofobservationswasonlyin9%theinstances earlier onaverage. head budswerecountedaswell,sincethese earlierfloweringspeciesweremonitored would belostbymowinginthislateflowering species.Intheothertwospeciesflower in thiscalculation,becauseweassumedthattheremainingbudsofflowerheads flower headsthatfloweredorhadfinishedfloweringwhencensusedwereconsidered population averagenumberofseedspertopflowerhead(Table 3).In was estimatedastheproductofmeannumberobservedflowerheadsand years after seedadditionforeach seedlingpresentafter oneyear. The annual relative increaseinthe numberofsmallrosettesintheseedsadded plots from 1to2 plots. Seedlingstasis, experiment becauseverylownumbersofseedlings wereobservedinthepermanent seed ( addition wascompared tocontrolplots. Seedlingandsmallrosetteestablishment per number ofseedlingsandsmallrosettesjustbefore1224monthsafter seed O. Onehundredseedsweresownineachof10plots peryearandsite. The 1999 and2000,seedswereaddedto5by50cmrectangular plots insitesB,N,and since thedensityofseedlingswasmuchlowerthanthatadultplants. InNovember classes attime less thansixrosettesattime For thosefirstyearclasses( resulting inemptyclassesofnewclonally-producedrosettesthatwereformed1998. rosettes formedinthepreviousyear, becausenoobservationsweremadeinthe1998, in whichtheplots persite werepooled.Seedlinggrowth, or oneyearlater( order togetthesexualreproductionmatrixelements, by theaverageseedproductionoffloweringrosetteperpopulation,in seed additionplots (SA)orcontrolplots (NS)atthebeginning( where p Seedling fates, Seedling establishment andfatewerestudiedinaseedadditionexperiment In thefirstyear(1999to2000)itwasnotpossibleobservefatesofside- N S p p sdl sml 11 sdl is thenumberofseeds,seedlings( and = = = N N ()() ()() NNN NN NNN NN d t sdl SA d t sdl SA p d d d d t sdl t sdl t sdl t sdl AS SNS NS SA SA m m m m t sml t sml t sml t sml AS SNS NS SA SA t+1 ,2 sml ,1 = = 1, 1,0 ,1 ,0 ,1 1, 1,0 ,1 ,0 ,1 t=1 ======respectively) werecalculatedasfollows: was estimatedfromallobservationsovertheyearsonthatclass −−− −−− ). These perseedestablishment probabilities, S S 11 11 , wascalculatedasfollows: N N and seeds SA side.veg seeds SA t , thetransitionprobabilitiesofthatclasstoallother G 21 , werederivedfromthesameseedaddition and sml , side.flow lrg sdl and ) orsmallrosettes( flow ) andforstage classescontaining H. radicata F , butmoreoften intheclonal ij . Perrosetteseedproduction G 21 , wascalculatedasthe t=0 and ) oftheexperiment p sml S. pratensis , weremultiplied S. pratensis ) ineitherthe . only (3) (2) (1) 73. Chapter 5 Space versus time in population dynamics , 22 (4) S , 12 Figure 4. R , and the main f , and the sum o ferent scales. ferent species, and per life history arting year TRE) in dif mputed as: , was co 21 ). Black symbols and trend lines signify regressions with G ferent diagrams can have dif S+R ), ramification and survival of side rosettes formed in the previous G ), growth ( F ling growth, a. Seed for each site and st s were pooled asis and retrogression ( l rosettes. e survival probability of smal was the averag = 42 e, s between the deviation from the species mean population growth rat 21 G ), and st G C+T s of site and year in the variation decomposition (L ated from the for and estim so accounted settes was al te of small ro ality ra and > 0.50. Please note that the axes of dif 32 fect 2 G t in which the plo mort ent plot dat perman Relationship ef component: fecundity ( year ( R 74. Chapter 5 Space versus time in population dynamics element ( which isthedominanteigenvalueofamatrix. The elasticityof intervals wereadjustedforsmalldeviationsbetweenthemean the observedfloweringrosettes. The upperandlowerlimits ofthe95%confidence calculated after resamplingtheaccompanying data setofnumberflowerheads α matrix element(Caswell2001): main andinteractioneffects canthenbedecomposedintocontributionsfromeach for eachlevelofthemaineffects separately, whileignoringtheinteractionterm. The (Horvitz etal.1997). The maineffects canbeestimatedbyfillinginthisequation first eigenvalue ofthemeanallmatricesaspecies, population transitionmatrixwecalculatedtheprojectedgrowthrate, The Matlabstudenteditionof1996wasusedforallmatrixcomputations. Foreach Matrix analysis (82% in completely assignedtofecundity, whichformsthemajorpart ofthesematrixelements these matrixelements couldnotbedividedintothesetwocomponents, itwas rosettes isthesumofafecundityandretrogressioncomponent. As theelasticityof for eachmatrixseparately. The transitionfromafloweringstage tosmallvegetative one. We addedtheelasticityvaluesofalltransitionssamelifecyclecomponent Elasticity valuesquantifytherelativecontributionofeachelementto in whichagiven 2001): experiments (LTRE). The LTRE modelwithtwofactorswithinaspeciesis(Caswell constructed matricesandthe were resampledforallclassesseparately. Newsexualreproduction( 1992; Caswell2001). The survival( fates withwhichaparticular matrixwasconstructed(Efron 1982;Kalisz&McPeek bootstrapping methodbyresampling3,000timesthedata setofobservedrosette confidence intervalsfortheprojectedpopulationgrowthrates( m , themaineffect ofthe To decomposevariationin =− α= e λ≅λ+α+β+αβ ij )( () () () mm S. pratensis nmnmn n m mn e = ij ) (deKroonetal.1986)isgivenby: a λ∂ ∑ ij ij , ∂λ () a ⋅⋅ aa ij jij ij )() () λ ⋅⋅⋅ of the , 80%in m ∂ ∂λ a th n ij () th λ site and H. radicata 2 1 () of theoriginalmatrix(Caswell2001,page 306). AA year, ()() m ⋅⋅⋅ + T , β λ G n n th , andtheresidual'interaction'effect, , we appliedfactoriallifetable response S year iswrittenasthesumofdominant and 100%in , and R ) andramification( λ (··) , themaineffect of the C. jacea λ ). To construct95% λ of the3,000newly λ ), weappliedthe F C ) elements were for eachmatrix λ ) observations , andsumto m th αβ site, (mn) (7) (5) (6) λ , 75. Chapter 5 Space versus time in population dynamics ), F (8) (9) ) and 1 and S Figure 5. asis ( ), and the mean ), st m Succisa pratensis T is 1.176, 0.81 ) in fect ( ontribution The resulting c , site ef (bottom row). Contributions of e and negative ain both positiv tion matrix is a rough s of a contribu aurea jacea (Horvitz et al.1997). The overall species mean Cent s may cont fect TRE) in population growth rate ( arately grouped by life history component: fecundity ( is given. (mn) s are sep f the matrix by the sensitivity values o , are multiplied fect (middle row) and (··) fect of a factor on a A ), survival of side rosettes formed in the previous year ( C . ). For each site m and species the average site mean R f rresponding matrix element o h matrix element with the co ferences of eac Hypochaeris radicat ), ramification ( s in the decomposition of variation (L of the sum of all element The magnitude G fect absolute value of the interaction ef in which dif an matrix, the overall me retrogression ( 0.910 respectively Site ef summary measure of the ef ll mean matrix. trix of interest and the overa halfway the ma main and interaction ef matrices of the values. growth ( (top row), positive and negative matrix element 76. Chapter 5 Space versus time in population dynamics comparable elasticitymatrices astheelasticityvaluesofdifferent components Remarkably, thelong-lived history components ( populations ofallthreespecies. The referencematrix, we alsoperformedathree-wayLTRE onthematricesconstructedforNandO variation inpopulationgrowthrate. lambdas. Sterck etal.(2003)showedthatlifehistorytrade-offs maythusdampenthe Similar buffering occursbyelements withnegativecontributionsformatriceshigh some elements maybuffer thenegativecontributionsofotherelements partly. the sumofallLTRE contributionswillbenegative,butpositiveLTRE contributionsof stasis ( for fecundity( There weresignificantlypositivecorrelations between others. Notethatifagivenmatrixhas contributions bysomeelements werecompensatedforbynegativecontributionsof elements withinlifehistorycomponents inordertoseewhetherandhowpositive analysis (cf.Sterck etal.2003). Therefore weseparated thepositiveandnegative and positivecontributionswithinamatrixcanbeexpectedinvariancedecomposition whose variationcontributesconsistentlytothein contributions anddeviationsfromthemean and and retrogression( correlations. Projection matriceswithlowerlambdashadhigherelasticities forstasis Because theelasticitiesofeachmatrixsum toone,therewerealsonegative as themeanofthreespeciesmatrices. decrease in jacea pratensis other twospecies:55%ofthelambdasweresignificantlylargerthanunityin The projectedpopulationgrowthrates( Life cycleandelasticityanalyses Results the valuesoftheseelements foreachofthesixlifehistorycomponents ( the elements ofthecontributionmatrixislostinthismeasure. Therefore wesummed R populations didnotdiffer muchbetweenyears,therewasatendencyto ) foreachofthetransitionmatrices.PositivecorrelationsbetweenLTRE To investigatetherelativeperformanceofthreecoexistingspeciestogether, If trade-offs betweenlifehistorycomponents exist,acombinationofnegative Elasticity analysisshowedthehighestsummed elasticitiesforgrowth( However, informationonthemagnitudeanddirection(positiveornegative)of S ) in , vs.9%intheotherspecies(Fig.1).Whereaslambdasof λ S. pratensis F through timein ) andgrowth( S+R C+T ), butmoststrongly so in and ) werepositivelycorrelatedwith S. pratensis H. radicata S. pratensis G ) in S. pratensis λ λ and theshort-livedperennial ) werehigherin smaller thanthe and forramification( and especiallyin λ are expectedforthosecomponents and A S. pratensis H. radicata λ (··) Succisa pratensis and thesummedelasticities , inthiscasewascalculated λ λ of theoverallmeanmatrix among theyearsorsites. Hypochaeris radicata λ C in ) in whereas clonallife C. jacea and C. jacea H. radicata F H. radicata , G Centaurea than inthe , (Fig. 3). (Fig. 2). C G , ) and T . had , S S. . , 77. Chapter 5 Space versus time in population dynamics , of ter (mn) The was s are Af . ference Figure 6. fect arallel. ), survival of side The dif Succisa pratensis C ) in . ). For each year n and species R ), ramification ( G . ference of 1.6% ), growth ( F 1 and 0.910 respectively ) and retrogression ( S ecies fitted well. TREs within sp TRE) in population growth rate ( . Contributions of positive and negative matrix element asis ( ), and the mean absolute value of the interaction ef n ), st in contribution to the variation , the smallest T is 1.176, 0.81 of 6%. rger than 1%, with a maximum ference was la fect ( aurea jacea s of year Cent , year ef fect and a well with an average dif TRE also fitted s in the decomposition of variation (L The overall species mean omposition of the variation in (species, site and year) dec The three-way fect ment analyses sponse Experi able Re arately grouped by life history component: fecundity ( ear ef is given. cases the dif in six out of 42 three-way L factor ecies was the strongest main ites N and O revealed that sp populations in s te ef the intermedia Life T approximation model of L The first-order age, though s were less than 1% on aver served and modelled lambda between the ob nes are p e regression li nd because th both species a ilar ranges in have sim Y Hypochaeris radicat sep rosettes formed in the previous year ( the average site mean 78. Chapter 5 Space versus time in population dynamics O deviated littlefromthespeciesandsiteaverages ( ( performed lesswellinthatsite( fecundity ( stronger thanyearand interactioneffects (Table 4).Overallsitescontributions of interaction effects in thedifferent species.In performed better( followed themeanyeareffect mostclosely. Especiallyinthelastyear 0.015 subtracted S. pratensis effect. After subtractingtheLTRE contributionsofthemaineffects ofspeciesandsites, factors, however, wereaslargethemainsiteandalmost aslargethemainyear made bythefactorsite(Table 4). The effects oftheinteractionsbetweenmain Leemputten (L)andVeerslootlanden (V). and Census andmowingdatesforeachyearpopulationof Table 1. apparent thatespecially the components thatexplainedthedeviation in αγ αγ Centaurea jacea = -0.077for = -0.004).However, whenthemainspeciesandyearcontributionmatriceswere vs Within-species variationin . 0.081for F S. pratensis ) werepositivelycorrelated withsiteeffects ( performed betterthanaverageinsiteN( ieYear Site 012Ot1-u 23-Jul 9-Oct 1-Aug &2-Oct 25-Aug 23-Jul 22-Aug - 2-Aug 23-Jul 6-Jul 2-Aug 9-Jun 12-Jul 23-Jul 6-Jun 9-Jun 23-Aug 2003 2-Oct O 2002 16-Aug O 2001 9-Jun O 2000 O 1999 O 034Sp4Jl1-u 10-Sep 5-Sep 15-Oct 18-Aug 13-Sep 2-Sep 4-Jul 18-Oct 6-Sep 6-Jul 21-Aug 22-Jun 21-Sep 4-Sep 7-Jun 2-Sep 2003 7-Jun 17-Sep N 2002 13-Sep N 2001 21-Sep N 2000 N 1999 N 035Sp--15-Jun 15-Sep 15-Sep 15-Sep - 15-Sep ------5-Sep - 12-Aug 4-Sep 2003 22-Aug 5-Sep V 5-Aug 2002 7-Sep 22-Jul V 5-Aug 2001 7-Sep 26-Jul V 2000 15-Aug 31-Jul V 1999 - V 26-Jul - 20-Jul - - 22-Jul - 26-Jul 2003 31-Jul B 2002 26-Jul B 2001 20-Jul B 2000 B 1999 B 035Sp2-u - - 15-Nov - 15-Oct - - 23-Jul 11-Jul - 5-Sep 9-Jun 2003 3-Oct 23-Sep L 6-Oct L2002 2001 23-Sep 2000 L 1999 L 3-Oct - L 1-Aug - S. pratensis βγ . The sitesareKonijnendijk (N),Koolmansdijk(O),Bennekomsemeent C. jacea = +0.128)and showed thesmallestdeviationfromaverages (mean| .paessH aiaaC. jacea H. radicata S. pratensis and 0.055for and αγ F , = -0.038). The oppositepattern wasfoundinsiteO G αγ H. radicata λ and = +0.037for decomposed differently intosite,yearand S λ H. radicata S. pratensis for eachofthesitesseparately, itwas components displayed high positive worse ( αγ Succisa pratensis C. jacea = +0.002forsiteNand βγ ) indicatingthat αγ α = -0.121)thanaverage. m the mainsiteeffects were Mown = +0.046)while ) (Fig.4a). Analysing the ). Hypochaeris radicata , Hypochaeris radicata S. pratensis C. jacea C. jacea βγ | = 79. Chapter 5 Space versus time in population dynamics ) λ S G , both λ = +0.131; Fig. = +0.131; B α ) covaried strongly with did not contribute much ; Fig. 6a). Growth ( G F F was negatively correlated only slightly lower than the only slightly lower were not only smaller than were not only smaller F λ = +0.152 and = +0.152 N over the years balanced in the over the years balanced in (Fig. 1), and therefore the different α . Growth ( , Fig. 5a), S variation over years (Fig. 4b), but did variation over years (Fig. 4b), λ F λ ), made the largest contributions. But = -0.207) was due to an overwhelming was due to an = -0.207) S. pratensis G V contributions were buffered by a high by a high buffered contributions were | = 0.025 for and α β ( G G λ , which were both strongly positive in the first , which were both strongly positive S. pratensis S elements were partly compensated for by high partly were elements , site effects were smaller than year effects (Table were smaller than year effects , site effects | = 0.096 for F = -0.066) because the large negative contributions = -0.066) because L α ë and | = 0.060) in α β G element were actually higher than average resulting in element were actually higher (Fig. 5b). The largest contributions to the variation in The largest (Fig. 5b). R by life history components were more consistent between by life history components H. radicata | = 0.123), but also qualitatively different. In contrast to the different. also qualitatively | = 0.123), but λ showed little variation in α and between sites, but not between years. Within sites clonal variation over sites. Especially in the last year, which had the in the last year, variation over sites. Especially S = -0.059 and had large negative contributions. This indicates that rosettes grew had large negative contributions. λ λ between years (mean | O S α λ ( H. radicata ), and in site B also growth ( ) contributions had the strongest correlations with variation in ) contributions had the strongest correlations with variation λ C C elements; in site L in site elements; negative and G G , λ The year effects (mean | The year effects In the shorter-lived Centaurea jacea Centaurea in both cases and other components had less effect (Fig. 4c,d). In all three sites had less effect in both cases and other components (-0.267 on average) were compensated to a large extent by positive contributions were compensated to (-0.267 on average) O low values for (+0.204). In site values for contributions. whereas contributions of all life history components were predominantly either positive whereas contributions of all life history components especially the bad years. or negative in a given site (Fig. 5c), this was less the case in (Fig. 6c). simultaneously In such years positive and negative contributions occurred contributions in sites with higher in sites with contributions ( average than 5a), and almost all their matrix elements were higher than average. The reverse was The reverse than average. were higher matrix elements almost all their 5a), and the very low For this site, site V. true for effect of negative contributions (-0.239) with little positive contributions (+0.032). In contributions little positive (-0.239) with contributions of negative effect sites had an average the other two bad V, contrast to site overall mean differences between sites (mean | between differences the site effects (mean | the site effects to variation in lowest small, or even Because they stayed classes in this year. to larger less often retrogressed, some contributions, however, covaried positively with contributions, however, not correlate with positive LTRE contributions that buffered the predominantly negative contributions. buffered contributions that positive LTRE 4). The contributions to The contributions 4). λ of positive and negative contributions populations of site and year effects than they were in site and year effects year and strongly negative in the last year (Fig. 6b). year and strongly negative in between years were made by between years were made by 4). Clonal (Table being the most important were also smaller with site effects effects ( propagation in between sites and between years (Fig. 4e,f). Variation with variation in propagation ( propagation 80. Chapter 5 Space versus time in population dynamics The sixlifehistorycomponents arefecundity( = 1)arecompared by ramification( formed inthepreviousyear( transition matricesof those oftheshort-lived species( population growthrates ( radicata The averageelasticitypatterns suggestthat Life cycleandelasticityanalyses Discussion vegetative rosettes( classes are:verysmallvegetative rosettesorseedlings( Population matrixdividedintolifehistorycomponents andmeanmatrices perspecies. The sixstage Table 2. each meanthecoefficient ofvariation(CV)isgiven. S. pratensis C. jacea H. radicata have comparable life-cycles. This isprobably due tothefactthat side.flow side.veg flow lrg sml sdl side.flow side.veg flow lrg sml sdl side.flow side.veg flow lrg sml sdl side.flow side.veg flow lrg sml sdl side.veg lrg Succisa pratensis ), rosetteswithatleastonefloweringstem( S. pratensis .9 .2 3 .9 2-01952 0.199 52 - 0.050 68 52 0.222 52 0.199 CV 30 Mean 0.050 1.743 CV 37 Mean - 0.148 CV 66 30 Mean - 0.199 0.334 131 CV 0 0.026 75 Mean 0 30 0.399 CV 0.184 Mean 26 1.743 CV 0.498 90 CV Mean 0.572 Mean 0.994 CV 78 Mean - 0.111 CV 63 68 Mean - 0.260 0.750 126 CV 0 0.033 84 Mean 0 90 0.163 CV 0.295 Mean 84 0.994 CV 0.375 CV Mean 0.677 Mean CV 87 Mean - 0.075 CV 69 Mean 0.306 154 CV 0 0.033 Mean 0 0.370 CV Mean 0.489 CV Mean ), rosettesproducedbyramificationwithatleastonefloweringstem( G sdl S - - - - .0 30265 .6 3- - 27 43 0.272 33 0.522 0.268 84 61 59 0.016 0.276 0.098 54 42 - 0.591 0.390 - 155 0.012 74 0.000 170 58 0.081 0.035 0.266 126 34 0.109 93 148 0.318 0.000 0.111 0.109 0.000 63 - 148 300 0.240 0.111 0.007 - 0.000 94 193 - 0.000 0.005 118 0.107 71 - 0.025 30 0.190 105 0.349 0.011 95 0.000 0.000 231 38 0.135 0.015 0.479 147 126 127 0.033 76 0.200 40 - 0.180 0.282 0.000 0.680 184 - 90 0.033 112 45 - 0.013 0.155 89 0.451 42 - 0.105 222 50 0.002 0.471 104 365 0.413 0.038 52 0.002 112 25 0.280 0.041 0.566 156 0.060 56 - 0.416 - - - 11 21 T λ ij ) ofthelong-livedperennial ( ), stasis ( sml , 12of G G relies morestronglyon rosettesurvivalaswouldbe C C R S H. radicata 22 62 52 12 42 32 S ij ) andretrogression( F Hypochaeris radicata ij ), growth( G C C R S lrg - 33 63 53 23 43 ). Whenmatriceswith thesame G Succisa pratensis sdl ij ), ramification( F ), smallvegetative rosettes( 24 flow C C R S (+R F R 14 44 64 54 34 ij S. pratensis ). Matrixelements aremeansof20 24 flow and 10of ) ), vegetative rosettesproduced side.veg C C C T T T ij ), survivalofsiderosettes - 45 35 25 65 55 Centaurea jacea ) werehigherthan and side.flow F 26 Hypochaeris C C T T (+R F 46 36 16 66 56 sml 26 side.flow λ ) . Beside (e.g. ), large λ ). 81. Chapter 5 Space versus time in population dynamics , . C. jacea were both were much H. radicata , followed by λ was positively λ λ (Ehrlén 1995) site H. radicata and negatively with the and negatively C and can form extensive below-ground G Lathyrus vernus ), F ), also contribute most, by their variation, λ is related to the monocarpic nature of the is related to the C. jacea , which is consistent with patterns in other with patterns , which is consistent S . A. resemblance of the outcome of these different , and even less so by the shorter-lived λ C. jacea ) and ). This indicates that those life history components that This indicates that those life history components ). R F S. pratensis LTRE and elasticity analyses point at the same major players: growth and elasticity analyses point at the same major players: LTRE ) and fecundity ( S The ramification rates of the meadow populations of rates of the meadow populations The ramification S. pratensis ), stasis ( ), stasis G ( In Life table response experiments response Life table that the variation showed response experiment (LTRE) The three-way life table for the observed variation in important between the species was most correlated with the elasticities of fecundity ( the elasticities of fecundity correlated with retrogression ( elasticities of 2000). et al. 1996; de Kroon et al. et al. 1993; Oostermeijer species (Silvertown et al. 1994). al. 1987; van Groenendael of road-verges (de Kroon et lower than those of clonality in The importance as dykes and road verges (E. Jongejans, habitats woody branches in more productive personal observation). that co-occurring species, This suggests variation. year variation and then by site are relatively more synchronized in time than that they given their specific life history, to 2000) had The synchrony occurred because one year (1999 are similar in space. species, while the last year (2002 to 2003) was always the highest growth rates in all of the dry early spring and summer of 2003. But the below average, probably because year-to- between species: environmental differed magnitude of the temporal variation woody by the clonal life history of the year fluctuations are better buffered than by the long-lived rosettes that die after flowering. In this species clonal propagation, the sprouting of the sprouting clonal propagation, flowering. In this species after rosettes that die a flowering the only way to survive for the base of the old stem, is new rosettes at 2004). In these meadows the number of connected individual (Hartemink et al. whereas rosettes rarely exceeds three, contribute most to local population growth ( variation in to the spatiotemporal expected of a longer-lived species (Silvertown et al. 1993). This stresses that This et al. 1993). (Silvertown species of a longer-lived expected is needed when that caution matrix and of a particular are properties elasticities (Shea et al. species within and between matrices of different elasticities comparing our three species 2001). Within 2000; Caswell Kroon et al. 1994; de qualitatively (direction) and quantitatively (magnitude) different for site effects and year for site effects (magnitude) different (direction) and quantitatively qualitatively that year and site variation influence population dynamics in suggests This effects. ways. Like in the perennial herb very different Temporal and spatial variation in population dynamics and spatial Temporal to variation in Contributions of variation in life history components types of analyses was also found by Horvitz et al. (1997), Eriksson & Eriksson (2000), types of analyses was also found by Horvitz et al. (1997), Eriksson Kiviniemi (2002), Nordbakken et al. (2004) and Endels (2004). 82. Chapter 5 Space versus time in population dynamics effects werealsostrongerthanyeareffects in Leemputten (L)andVeerslootlanden (V). combination seperately. The sitesareKonijnendijk(N),Koolmansdijk(O),Bennekomsemeent flower headsperfloweringrosettewerederivedfrompermanentplots foreachsiteandyear selected topflowerheadsperpopulation(data fromallsampledyears arepooled). The numberof Succisa pratensis Mean numberofseedsperflowerheadandmeanheadsfloweringrosette Table 3. differed especiallyinthewaylowgrowthratesof population decline. vicinity oftheplots, thisisunlikelytobethe mainexplanationfortheobserved decline (Crawley1990),butaswedidnot observehighcolonizationratesinthe present atthebeginningofourstudywemay havebeenbiasedtofindpopulation species. Byselectingspots forpermanentplots inwhich Another explanationmaybefoundinthe higherspatial dynamicsofshort-lived year for population declinesinallthreesitesovertime. The lastyear, whichwasalsotheworst strength betweenthesiteandyeareffect. 2000). Inourcase,however, theinteractioneffects werealwaysintermediatein demographic variationovertimeisweaklycorrelatedamongpopulations(Menges study. et al.2004),butnodistinctionwasmadebetweenspatial andtemporalvariationinthat been reportedtocausehigh variation in smaller in realized. Whilethedifferences betweentemporalandspatial effects weremuch

C. jacea C. jacea C. jacea H. radicata H. radicata H. radicata pratensis S. pratensis S. pratensis S. pratensis S. pratensis S. Site Species Large interactioneffects betweensitesandyearswouldmeanthat S. pratensis H. radicata λ between sitesandnotyears.Highfecundityhasfrequently , 9,0,0 93 . .533 1.69 0.91 3.00 1.08 1.10 - 3.00 0.35 1.85 3.30 2.60 2.22 2.73 1.53 1.61 0.25 0.90 0.64 2.44 2.31 1.51 2.4 1.29 2.3 2.35 4.5 29 0.44 2.85 2.63 1.7 1.56 1.57 0.19 36 34 70 1.92 23 0.89 1.00 69 3.11 0.87 22 2.04 1.56 75 '99,'00,'02 2.57 0.39 3.2 '99,'00,'02 2.1 3.00 B 0.70 0.00 '00,'01,'02 53 O 2.36 45 1.9 N 4.44 0.00 53 1.6 '01 75 46 1.6 '99,'00,'01,'02 30 L 1.7 73 2.9 '00,'01,'02 63 O 80 48 N 39 70 '99,'00,'02 49 54 '99,'00,'01 V '99,'00,'02 L '00,'01,'02 B '01,'02 O N Hypochaeris radicata Years , maypartly beexplainedbytheabove-mentioned drought. and Flower heads sampled C. jacea λ , e.g.inthenon-clonal ense 9920 012002 2001 2000 1999 s.e. mean n and , alsointhesespeciesfecunditycovariedwith No. Seeds Centaurea jacea Hypochaeris radicata S. pratensis No. flowerNo. heads perfloweringrosette Centaurea corymbosa . Seedswerecountedinrandomly S. pratensis Permanent plots . Different sitesandyears H. radicata populations were showed marked rosettes were (Fréville 83. Chapter 5 Space versus time in population dynamics . F ) was and G G , S (caused by makes these ) seem to be expressed as F contribution. A C G ë ) when compared to ) when compared S . ) and survival of clonal H. radicata showed the same pattern G S. pratensis , irrespective the underlying λ H. radicata and especially performed below average, large negative below average, performed leads to lower population extinction risks (Sterck et al. 2003). (Sterck Lobularia maritima λ may be expected because Hartemink et al. may be expected because Hartemink G and plants stayed small, showed negative contributions small, stayed plants ). Interestingly F S. pratensis S. pratensis G+S+R may also be high due to favourable growing conditions in this may also be high due to favourable G S. pratensis Vouacapoua americana Vouacapoua indications of compensation by different life cycle components were life cycle components different indications of compensation by ) and compensating positive contributions of stasis ( stasis positive contributions of ) and compensating ). This would suggest that climatic variation influences plant developmental ). G T C. jacea contribution as Buffering of negative contributions to variation in Buffering In In the second bad site (O) a negative LTRE contribution of a negative LTRE In the second bad site (O) G only found in bad years. LTRE contributions of clonal propagation ( contributions of clonal propagation only found in bad years. LTRE contributions of especially growth ( by opposing buffered decisions rather than overall growing conditions for this species as decisions rather than overall growing conditions for this species varies little. Picó as the likely (precipitation) et al. (2002) also pointed at temporal variation in climate periods of in different explanation for the negative covariances of seedling production the flowering season of the perennial herb the other sites. This can be understood by the observation that rosette growth ( by the observation that rosette This can be understood the other sites. in of seed production found costs (2004) experimentally is unlikely to fully explain the such a trade-off growth. However, reduced vegetative high ( offspring the population dynamics as it reduces the variance in for mechanisms, is important population growth rate. Lower variance in LTRE contributions were almost completely buffered by positive contributions of other contributions by positive buffered almost completely were contributions LTRE of negative Buffering negative net effect. resulting in a small, life history components, life cycle seems to be a more of the elements of different by positive contributions were also life history components as negative covariances between general pattern These negative correlations et al. (2003). et al. (2002) and Sterck reported by Picó the most nutrient- history functions. For instance, between life may reflect trade-offs which poor site L, in of growth ( patterns, factors may underlie the specific LTRE somewhat drier site. Here, external to compared their importance would be required to compare and experimentation lower or rates were consistently In the third bad site (V) all vital those of trade-offs. reduced probably consistently vegetation unchanged. Here a dense grass synchrony in the spatial & Orzack 1980; Menges 1998). However the (Tuljapurkar dynamics of the populations of Flexibility in life history responses in life history Flexibility in which the three sites In two of species regionally more vulnerable to extinction (Harrison & Quinn 1989; Sutcliffe et & Quinn 1989; Sutcliffe species regionally more vulnerable to extinction (Harrison al. 1997; Heino et al. 1997; Matter 2001). mowing before seed set) was partly compensated for by a positive compensated for by a partly mowing before seed set) was reduced but not survival ( were found fates of surviving meristems between different in site L. Similar trade-offs in the tropical tree between demographic trade-off 84. Chapter 5 Space versus time in population dynamics for thethreespeciesseperately. effect aregiven(*100). The overallmean over fouryears. The mean andstandard deviation(sd)oftheabsolutevaluesalllevelswithinaLTRE together insiteNandOoverfouryears(three-way),eachspeciesseparately inallobservedsites species canbedifficult evenwhenthey possesscomparable lifehistories. cautiously appliedtoplants, asour results suggest thatgeneralizationsamongplant cases (seeHeppellet al.(2000)foranexamplewithmammals) should thereforebe to predictthepopulation dynamicsofpoorlystudiedspeciesfrom data ofwell-known third species( in growingconditions,themagnitudeofresponses differed considerably, whilethe S. pratensis may underliethevariationin various sourcesincludingsiteproductivityand vegetation composition. climatic fluctuationsandvariationinmanagement, whereasspatial variationhas that generatethevariation:year-to-yearvariation inmeadowsismainlycausedby between temporalandspatial variationarelikelytobetheresultofdifferent factors dynamics ofplants thisisanimportant pointtotake intoconsideration. The differences that badsitesmayalsodiffer amongsteachother. Formodelstudiesontheregional substitutable, aswehavefoundthatgoodyearscanbedifferent fromgoodsitesand Temporal andspatial variationinthedynamicsofpopulationsmaynotbereadily Implications population growthrate, Magnitude ofthedifferent effects offourvariationdecompositionanalyses(LTRE) ofvariationin Table 4. ie()| Site (S) effect LTRE | S * Y Species (P) Year (Y) Year S * Y * P Y * P Y * | S * P Our LTRE analysesshowthatvariationinvery different lifehistorycomponents and C. jacea H. radicata | | | | α αβ γ β αβγ βγ αγ q n m | | λ nq mq .204 23 .941 .275 2.48 7.53 1.22 4.16 6.19 12.31 0.47 mn 4.32 | : allspecies( mnq | .82.80 3.38 | .125 .661 .06536 2.22 3.68 6.5 8.40 6.11 8.86 2.52 4.41 | ) hardlyrespondedtoexactlythesamevariation atall. Attempts | ens ens ens ensd mean sd mean sd mean sd mean 58 8.31 15.86 Three-way .854 .943 37 36 .01.63 2.70 13.65 13.70 4.31 5.99 5.49 8.08 .24.97 5.32 .75.48 7.07 λ showed similarresponsestothespatiotemporal variation , evenamongspecieswithsimilarlifehistories. Although Succisa pratensis λ was 1.047,and .paessH aiaaC. jacea H. radicata S. pratensis , Hypochaeris radicata λ was 1.176,0.811 and0.910respectively and Centaurea jacea ) 85. Chapter 5 Space versus time in population dynamics (Asteraceae). - in semi-natural (Asteraceae). - Hypochoeris . - Oikos 50: 3-12. : II. Herbivory and population Plantago media Plantago L.). - J Ecol 43: 709-718. Centaurea corymbosa Centaurea Lathyrus vernus Hypochaeris radicata Scabiosa succisa : implications for vegetation monitoring. - J Appl Ecol 38: 689-698. - J monitoring. : implications for vegetation Moench. ( . - J Ecol 86: 13-26. Succisa pratensis Succisa pratensis heathland communities. - Trends Ecol Evol 14: 63-68. Trends heathland communities. - The Netherlands. - Biol Conserv 62: 59-65. rich meadows in structure of biology. - Ecology 81: 619-627. biology. Massachusetts. Associates, Inc. Publishers, Sunderland, 125-140. Asarum canadense growth rate. - Ecology 67: 1427-1431. to population demographic parameters dynamics of a perennial grassland species, limitations. - Ecology 81: 607-618. - Ecology 81: limitations. applied mathematics, Philadelphia, Pennsylvania. J Ecol 77: 495-508. demographic variability in the endemic plant species dynamics. - J Ecol 83: 297-308. thesis, University of Leuven. 245-252. Sci 11: grasslands. - J Veg Ecology 85: 694-703. the Netherlands. - Hydrobiologia 478: 149-170. 755-758. References Adams AW 1955. Bakker JP of ecological diversity in grassland and in the restoration & Berendse F 1999. Constraints Altena HJ & Elberse WT Oomes MJM, Berendse F, on the restoration of species- 1992. Experiments on the population The influence of management regime and altitude Bühler C & Schmid B 2001. We thank Frank Berendse, Johan Ehrlén, Linda Jorritsma-Wienk, Felix Knauer, Juul Knauer, Felix Linda Jorritsma-Wienk, Johan Ehrlén, Berendse, thank Frank We for their Soons and Jasper van Ruijven Mix, Xavier Picó, Merel Limpens, Carolin Forestry of State The responsible officers discussions. and fruitful comments helpful meadows. permission to work in their of Ermelo gave us their kind and the community (NWO Research funded this research Organization for Scientific The Netherlands project 805-33-452). Their roles in conservation retrospective perturbation analyses: Caswell H 2000. Prospective and - Sinauer models. Construction, analysis, and interpretation. Caswell H 2001. Matrix population R Soc Lond Ser B-Biol Sci 330: Trans - Philos The population dynamics of plants. Crawley MJ 1990. Damman H & Cain ML herb, Population growth and viability analyses of the clonal woodland 1998. Elasticity: the relative contribution of A, van Groenendael J & Caswell H 1986. de Kroon H, Plaisier Ade Kroon H, Plaisier JM 1987. Density dependent simulation of the population & van Groenendael Acknowledgements de Kroon H, van Groenendael JM & Ehrlén J 2000. Elasticities: a review of methods and model de Kroon H, van Groenendael JM & Ehrlén J 2000. Elasticities: a and other resampling plans. - Society for industrial and The jackknife, the bootstrap Efron B 1982. Ehrlén J 1995. Demography of the perennial herb and temporal A, Imbert E & Olivieri I 2004. Spatial Fréville H, Colas B, Riba M, Caswell H, Mignot Endels P a demographic approach. - PhD landscape elements: plant species in small 2004. Vulnerable Eriksson Å & Eriksson O 2000. Population dynamics of the perennial ALFone A 1989. demographic study of annual and perennial comparative in AJM & Kemmers RH 2002. Restoration of brook valley meadows Jansen Bakker JP, AP, Grootjans of a fragmented landscape. - Nature 404: capacity The metapopulation Hanski I & Ovaskainen O 2000. 86. Chapter 5 Space versus time in population dynamics Silva JF, RaventosJ,CaswellH& Trevisan MC1991.Populationresponses tofireinatropicalsavanna Shea K,ReesM&Wood SN1994. Trade-offs, elasticitiesandthe comparative method. -JEcol82: Schaminée JHJ,Stortelder AHF &Weeda EJ1996.Devegetatie vanNederland.Deel3. Saunders DA,HobbsRJ&MargulesCR1991.Biological consequencesofecosystemfragmentation: Roem WJ&BerendseF2000.Soilacidityandnutrient supplyratioaspossiblefactorsdetermining fire-explicit populationviabilityanalysisof Quintana-Ascencio PF, MengesES&Weekley CW2003. A Picó FX,Quintana-Ascencio PF, MengesES&Lopez-BarreraF2003.Recruitmentratesexhibithigh Picó FX,deKroonH&Retana J2002. An extendedfloweringandfruitingseasonhasfewdemographic Oostermeijer JGB,BrugmanML,deBoerER&denNijsHCM1996. Temporal andspatial variationin &SchwartzMW(eds), Oostermeijer JGB2003. Threats torareplantpersistence.-In:BrighamCA Nordbakken J-F, RydgrenK&ØklandRH2004.Demographyandpopulationdynamicsof 1988.Fine-scalespatial andtemporalvariationinthedemographyofaperennial Moloney KA Menges ES&DolanRW 1998.Demographicviabilityofpopulations Menges ES2000.Populationviabilityanalysisinplants: challengesandopportunities.- Trends Ecol &KareivaPM(eds), Menges ES1998.Evaluatingextinctionrisksinplantpopulations.-In:FiedlerPL Matter SF2001.Synchrony, extinction,anddynamicsofspatially segregated,heterogeneous Lucassen ECHET, BobbinkR,Smolders AJP, vanderVen PJM,LamersLPM&RoelofsJGM2003. Pegtel DM1983.Ecologicalaspects ofanutrient-deficientwetgrassland Kiviniemi K2002.Populationdynamicsof 1992.Demographyofanage-structuredannual-resampledprojection Kalisz S&McPeekMA Horvitz C,SchemskeDW&CaswellH1997. The relative“importance” oflife-historystages to &DiemerM2002.Effects ofsmallhabitat sizeandisolationonthepopulationstructure Hooftman DAP Heppell SS,CaswellH&CrowderLB2000.Lifehistoriesandelasticitypatterns: Perturbationanalysis Heino M,Kaitala V, Ranta E&LindströmJ1997.Synchronousdynamicsandratesofextinctionin Hartemink N,JongejansE&deKroonH2004.Flexiblelifehistoryresponsestoflowerandrosettebud Harrison S&QuinnJF1989.Correlatedenvironments andthepersistenceofmetapopulations. -Oikos grass, 951-957. Plantengemeenschappen van graslanden,zomenendrogeheiden.-OpulusPress, Uppsala. a review. -ConservBiol 5: 18-32. 151-161. changes inplantspeciesdiversitygrasslandandheathland communities.-BiolConserv92: Hypericum cumulicolainFloridarosemaryscrub.-Conserv Biol17:433-449. 69-74. elasticity andhightemporalvariationinpopulationsof ashort-livedperennialherb.-Oikos103: effects inaMediterraneanperennial herb.-Ecology83:1991-2004. Ges Ökol10:217-228. the demographyof Population viabilityinplants. Springer-Verlag, Berlin. anglica bunchgrass. -Ecology69:1588-1598. size andisolation.-JEcol86:63-78. prairies: Relationships withfiremanagement,geneticvariation,geographiclocation,population Evol 15:51-56. Conservation biology:forthecomingdecade.Chapman&Hall,London. populations. -EcolModel141:217-226. dissectum Interactive effects oflowpHandhighammoniumlevelsresponsibleforthedecline grassland species.-PlantEcol159:153-169. matrices, elasticityanalyses,andseedbankeffects. -Ecology73:1082-1093. Hall, New York. (eds), Structured-population models inmarine,terrestrial,andfreshwatersystems.Chapman& population growth:prospectiveandretrospectiveanalyses.-In: Tuljapurkar S&CaswellH of commonwetlandspecies.-PlantBiol4:720-728. for specieswithminimaldemographicdata. -Ecology81:654-665. spatially structuredpopulations.-ProcRSocLondSerB-BiolSci264:481-486. removal inthreeperennialherbs.-Oikos105:159-167. 56: 293-298. Andropogon semiberbis and (L.) Hill.-PlantEcol165:45-52. D. rotundifolia Gentiana pneumonanthe . -JEcol92:110-121. matrixmodelapproach.-JEcol 79:345-356. : A Agrimonia eupatoria , arareperennialherb.-JEcol84:153-166. and Geum rivale Silene regia (Cirsio-Molinietum) , twoperennial in midwestern Drosera . -Verh Cirsium 87. Chapter 5 Space versus time in population dynamics Primula : regional, patch Vulpia ciliata Vulpia . - J Ecol 91: 18-26. Succisa pratensis ) along the forest regeneration cycle: projection matrix analysis. - J Ecol 86: 545-562. ) along the forest regeneration cycle: importance of life-cycle components to the finite rate of increase in woody and herbaceous in woody and rate of increase to the finite components of life-cycle importance - J Ecol 81: 465-476. perennials. - J Ecol 90: 1033-1043. grassland forbs. dispersed connectivity. and fragmentation Habitat grasslands. - In: Soons MB (ed), rich semi-natural Ph.D. thesis, Utrecht plants. of the colonization process in and temporal characteristics Spatial University. 84: 2751-2761. analyzed with a matrix model. - Ecology tropical forest tree - Oecologia butterfly metapopulation. in a highly connected ringlet and population fluctuations 109: 235-241. 7: 273-292. - Cambridge University Press, density and longevity. 65-66. Theor Popul Biol 18: 314-342. rates and extinction. - vulgaris Sci 15: 387-394. - J Veg meadow. in plant species composition in a species-rich fen shifts matrices. - Ecology 75: 2410-2415. in population projection pathways of quality and population size on performance and local dynamics. - J Ecol 88: 1012-1029. Silvertown J, Franco M, Pisanty I & Mendoza A Pisanty I & Mendoza J, Franco M, Silvertown Relative demography: plant 1993. Comparative of wind- fragmented populations in capacity Reduced colonization & Heil GW 2002. Soons MB connectivity of species- and Fragmentation JH, Jongejans E & Heil GW 2003. Soons MB, Messelink Module responses in a M & de Kroon H 2003. During HJ, Martínez-Ramos FJ, Bongers F, Sterck Correlated extinctions, colonizations TJ & Greatorex-Davies JN 1997. Yates Thomas CD, OL, Sutcliffe I. - Oikos flowering of some perennial herbs, observations on the survival and CO 1956. Further Tamm K, Bakker JPThompson methodology, Europe: The soil seed banks of North West & Bekker RM 1997. RM, Lehman CL D, May Tilman & Nowak MA destruction and the extinction debt. - Nature Habitat 1994. I. Long-run growth SD & Orzack SH 1980. Population dynamics in variable environments. Tuljapurkar T Valverde in the demography of a woodland understorey herb ( J 1998. Variation & Silvertown van der Hoek D, van Mierlo AJEM & van Groenendael JM 2004. Nutrient limitationand nutrient-driven AJEM & van Groenendael JM 2004. Nutrient limitationand van der Hoek D, van Mierlo Loop analysis: Evaluating life history S 1994. Tuljapurkar S & van Groenendael JM, de Kroon H, Kalisz A Rengelink R, Copal of genetic variation, habitat The interacting effects & Ouborg NJ 2003a. P, Vergeer Watkinson AR, Freckleton RP Watkinson & Forrester L dynamics of 2000. Population 6 Inbreeding depression significantly influences population dynamics: a population-based model for a long-lived perennial herb

F Xavier Picó, Eelke Jongejans, Pedro F Quintana-Ascencio & Hans de Kroon

Summary

Although inbreeding depression usually reduces several fitness traits, the importance of inbreeding depression for the dynamics of plant populations still remains controversial. Assessing the demographic implications of inbreeding depression is essential to develop conservation strategies for plants in fragmented habitats where population sizes tend to be low, the degree of isolation high and inbreeding depression also high. We constructed a simulation model parameterized with demographic and genetic data for the long-lived herb, Succisa pratensis, whose populations are highly fragmented in many places across Europe. Demographic data were based on four years of field data from a declining and an increasing population while genetic data were obtained from a greenhouse experiment on inbreeding depression. Inbreeding depression was included into the model by using a relationship between outcrossing probability and the effective population size, and reducing those demographic traits that were significantly affected by inbreeding. Density- dependent recruitment was also incorporated into the model to simulate habitat limitation. Overall, the dynamics of the decreasing population were strongly affected by inbreeding depression, but not by density dependence since the population never reached the carrying capacity. The opposite pattern was exhibited by the increasing population that was less affected by inbreeding depression but limited by density-dependent recruitment.

Keywords: density dependence, extinction risk, inbreeding depression, matrix population models, perennial plants, population dynamics, stochastic simulations, Succisa pratensis.

Introduction

Inbreeding depression, i.e. the reduction in fitness of selfed progeny compared to outcrossed progeny, has traditionally been considered as the main genetic risk 90. Chapter 6 Demographic implications of inbreeding hand-pollination experiment. Becausethehabitat patches ofnaturalpopulations of inbreeding depression onlife-cycletraits throughoutthewholelifecycle basedona field data fromtwopopulations duringfiveyears.We alsoestimatedtheeffects of perennial plants acrossEurope. We constructed amatrixpopulationmodelbasedon species thereforerepresents agoodexample ofthecurrentstatus ofmanycommon species faceshighdegreesoffragmentation in manyplacesacrossEuropeandthe inbreeding depressionintheperennialherb depression onlife-cycletraits. empirical demographictraits includingthe demographic effects ofinbreeding it becomesessentialtodevelopspecies-specific conservationplansbasedon population viabilityfromthescarcecurrentknowledge onthisissuearerisky. Hence, (Silvertown etal.1993),generalizationsontheeffects ofinbreedingdepressionon contribution oflife-cycletraits topopulationgrowth rate dependsonlifehistory on life-cycletraits dependonmatingsystem(Lande& Schemske 1985),andthatthe life-cycle traits forthespeciesstudied.Giventhateffects ofinbreedingdepression parameters ofwildpopulationscombinedwitheffects ofinbreeding depressiononthe inbreeding depression,noneofthemwasbasedonempiricaldemographic extinction risks. consistent withthenotionthatincreasinginbreedingdepressionleadsto estimates ofheterozygosity(Saccherietal.1998).Overall,thesestudiesare term surveysofbutterflyspeciestorelateextinctionrisksnaturalpopulationwith (Brook etal.2002;Frankham2003).Finally, anotherstudyhasusedlarge-scalelong- assuming thatrecruitmentwasalwaysnegativelyaffected by inbreedingdepression modeling approachestosimulatepopulationgrowthofseveralendangeredspecies plant fitnessandgeneticvariability(Newman&Pilson1997).Otherstudieshaveused pulchela al. 2003). Another studycreatedexperimental populationsoftheannualplant compare thefitnessoflinesdiffering intheirgeneticload(Bijlsmaetal.2000;Reed example, modelsystems,suchas on thedemographicimplicationsofinbreedingdepressionisstillveryscarce.For & Laurenson1994). because theformeroccursfarbeforelatterbecomesevident(Lande1988;Caro could bemoreimportant thangeneticfactorsindrivingthefateofwildpopulations been arguedthattheimportance ofenvironmental stochasticityandcatastrophes 1998; Bijlsmaetal.2000;Brook2002;Frankham2003).However, ithasalso rates andincreaseextinctionprobabilities(Newman&Pilson1997;Saccherietal. inbreeding depressiononimportant demographictraits mayreducepopulationgrowth references therein).Empiricalandmodelingworkhasalsoshownthattheeffects of evidenced byseveralempiricalstudies(seeCharlesworth&1999and depression onseveraltraits throughoutthelifecycleofanimalsandplants havebeen affecting populationperformanceandviability(Soulé1987). The effects ofinbreeding The goalofthisstudywastoassess the demographicimplicationsof Although thesestudiesprovidedinsights intothedemographicimplicationsof Such acontroversycouldbecausedbythefactthatourempiricalknowledge with highandlowlevelsofinbreedingtoinvestigatetherelationshipbetween Drosophila melanogaster Succisa pratensis , hasbeenusedto (Dipsacaceae). This Clarkia 91. Chapter 6 Demographic implications of inbreeding , 44 s 1-p 55 ·p ·p when S S g g s * p * id g ), small * p ·id ·id Figure 1. * id sml 54 78 88 g s s sdl ·(1-p) ·(1-p) F * id 58 S R ). The matrix 18 28 F * id ·id ·id and F F 58 68 * p flow g sdl F F 42 56 43 r * id * p G R 87 67 77 G s - g * id S R G * id * id 34 65 r r s R G 64 ·id ·id F 56 33 66 * id Selfed plants S 76 86 S R 69 t F G G . p r ) or as demographic transitions between ij 66 F ·id 55 ------S sdl sml lrg flow 65 32 r ) and reproductive adults ( ) and reproductive adults 67 G lrg R * id 23 76 ·p·p S G R G g g 22 stage at time S ·id ·id 34 44 s s ·(1-p) ·(1-p) S R , increase in size). Inbreeding depression affects some of the , increase in size). Inbreeding depression affects flow ij 14 24 ·id ·id sml lrg flow G F F 54 64 77 * (1-p) S r F F for seed set, seed germination, and growth of 24 r 21 78 F G id * id R 43 23 33 , stasis; , stasis; - ij 86 lrg * (1-p) S R G S G 87 12 and 28 G R g F 32 42 12 22 ), large vegetative rosettes ( ), large vegetative id Outcrossed plants Outcrossed sml , S R G G s sml id The four stages are seedlings or very small vegetative rosettes ( are seedlings or very small vegetative The four stages 21 11 * (1-p) sdl sdl S G flow lrg sml sdl 14 F r---lrg--- - lrg - sdl sdl--- * (1-p) sml--- sml 11

88 , decrease in size; lw- - - flow--- flow -

ij S 18

S Selfed Outcrossed R F

t+1 stage at time time at stage ucosn cycle Outcrossing efn cycle Selfing demographic traits ( demographic traits stages ( stages Life cycle graph and population projection matrix including the outcrossing and selfing cycles for Life cycle graph and population projection matrix including the outcrossing Succisa pratensis. rosettes ( vegetative elements are marked as fecundity from reproductive adults ( are marked as fecundity from reproductive adults elements plants enter into the selfing cycle. The probability of producing outcrossed progeny is given by The probability of producing outcrossed enter into the selfing cycle. plants whereas the probability of producing selfed progeny is given by 92. Chapter 6 Demographic implications of inbreeding outcrossed seedproduced. We counted andseparated filledseedsfromtheaborted pollinations untilseed harvesting.InSeptember2002,wecollected allselfedand as arecipienthead. All treatedflower heads remainedbaggedduringandafter hand- flower headoveranother one.Oneheadalwaysactedasapollen donor andtheother heads wereemasculatedandhand-pollinations weremadebygentlyrubbingone obtain selfedandoutcrossedprogeny, respectively. Priortohand-pollination,flower two werecross-pollinatedwithpollenofdifferent plants withinthesamepopulationto progeny. Two flowerheadswereself-pollinatedwithpollenofthesameplant andother submitted Garden oftheUniversityNijmegenin1998 foranotherexperiment(Picóetal. Netherlands) thatwereraisedandmaintained inthegreenhouseofBotanic We selectedupto 29adult Experimental crosses seedlings thatappearedinsidetheplots werealsocounted,measuredandtagged. flowering stalks andthenumberoffloweringheads per stalk. Eachyear, allnew (dead oralive),thenumberofleaves,lengthlongestleaf, population ranged157-246plants. Foreachplantandyear, werecordedthestatus tagged andmeasured.Overall,thenumberofplants monitoredperyear Meent andthreeinVeerslootlanden) werelaidoutandallplants withintheplots were populations respectively).In April 1999,permanentplots (1×1m;fiveinBennekomse once ayearduringflowering(July/AugustandSeptember/Octoberforthetwo are mownonceayear. Populationsweremonitoredfrom1999 to2003andcensused on thedynamicsof (52°00' N,5°36'E)andVeerslootlanden (52°36'N,6°08'E), whicharepart ofastudy populations intheNetherlandswereselectedforthisstudy:BennekomseMeent history traits canbefoundelsewhere(Vergeer etal.2003;Hartemink2004). Two descriptionoftheplantandits life- meadows andcalcareousfensinEurope. A Succisa pratensis Study speciesandpopulationsampling Materials andmethods inbreeding depressionanddensitydependence. limitation. We finallysimulatedthedynamicsofpopulationsdiffering intheextentof dependent recruitmentfunction,whichisaconservativewayofincludinghabitat microsites forplants togrowin,acarryingcapacity wasimposedbyadensity- many plantspeciesinEuropearelimitedsizeandthenumberofavailable ). InJuly2002,foreachmaternalplant,we createdselfedandoutcrossed is along-lived,perennialherbwidelydistributedinheathlands,hay S. pratensis S. pratensis (chapter 5).Bothpopulationsarehaymeadowsand plants (fromoverallsixpopulationsacrossthe 93. Chapter 6 Demographic implications of inbreeding Figure 2. were analyzed Succisa pratensis Inbreeding level level Inbreeding Inbreeding No Low High No Low High (density dependence) (density No Low High S. pratensis No Low High (density independence) (density Increasing Population Population Increasing Increasing Population Population Increasing

(b)(b) (d)(d)

44 22 66 4 2

11 11

Plants Plants Final No. No. No. Final Final Final No. Plants Plants No. Final 1010 No. Final 10 1010 1010 10 1010 = 100 plants; circles) initial population sizes with = 100 plants; initial N Inbreeding level level Inbreeding Inbreeding (density dependence) (density (density independence) (density NoNo Low Low High High NoNo Low Low High High = 20 plants; quadrats) and large ( quadrats) = 20 plants; Decreasing Population Population Decreasing Decreasing Population Population Decreasing initial N 00 00

8080 6060 4040 2020 8080 6060 4040 2020 (c) (c) (a)(a)

100100 100100 to extinction ( ( extinction extinction to to ) ) ( ( yr yr Time to extinction extinction extinction to to Time Time yr) Time Time Time yr) yr) density independent and density dependent recruitment. SE values are not visible due to their narrow density independent and density dependent recruitment. SE values are was based on 100 runs. range of variation. Each computation using one-way ANOVAs, testing the effects of hand-pollination (selfing and of hand-pollination the effects testing ANOVAs, using one-way pooled all plants and plant growth. We seed set, seed germination outcrossing) on on any of the not have a significant effect because population did from all populations ones to calculate seed set estimated as the number of filled seeds divided by the total divided by the of filled seeds as the number set estimated calculate seed ones to cm) to calculate (15 × 15 in pots were sown Filled seeds of seeds produced. number to sowing. Up after one month treatment plant and pollination germination per percent randomly selected were pollination treatment plant and per maternal five seedlings transplanting, plant size after Three months 15 cm). (15 × sown in pots and individually longest leaf. leaves and the length of the the product of the number of was recorded as of on life-cycle traits of pollination treatment The effects Stochastic population statistics for the simulated decreasing and increasing for the simulated population statistics Stochastic (for the decreasing population) populations in three inbreeding scenarios. Mean ± SE time to extinction Simulations were performed and final number of individuals (for the increasing population) are given. for small ( 94. Chapter 6 Demographic implications of inbreeding calculated as in eachvital rateduetoselfing(anestimateofinbreedingdepression, seed productionperplant attime one growingseason, newly recruitedseedlings exhibitedfastgrowthratesandreached the per plantattime bank. Seedlingproductionwasestimatedas theproductofmeanseedproduction produced perreproductiveplantoverayear, 5cm; as < leaf S. pratensis made largelyaccordingtoBühler&Schmid(2001) wherethenon-floweringclassesof risaaye P>0.05inallanalyses;Picóetal. traits analyzed(P the simulations.MatrixentriescodesasinFig.1. experiment, andthepercentchangeinamatrixentryforlowhighinbreedingscenariosused the matrices,meanpercentdecreaseinvital rateobtained fromtheinbreedingdepression Summary ofdemographicvital ratesaffected byinbreedingdepression,thematrixentriesmodifiedin Table 1. ( population ofstudy. We constructedmatricesforfouryears(from1999-2000 to 2002-2003)foreach size-based Lefkovitchpopulationprojectionmatrices(seeCaswell2001fordetails). pratensis Based ondemographicandgeneticdata, weconstructed ademographicmodelfor The demographicmodel (Vergeer etal.2004). indicated thatadultstages of depression ongrowth,survivalandfecundityofadultstages, otherresearchalready mean ofmaternalplantvalues. Although wedidnotestimate theeffects ofinbreeding depression forseedset,germinationandplantgrowthwascalculatedasthe mean vital rateofselfedandoutcrossedprogeny, respectively. The meaninbreeding ri arxetyFitness change Matrix entry and small rosettes Growth seedlings of Seed germination Seed set Trait sml ), largerosettes( The matrixmodelincludedfoursizeclasses: seedlings( ≤ including theeffects ofinbreedingdepressiononlife-cycletraits. We used were differentiated bythemaximumleaflength( id =1-(W lrg t and theseedestablishment probabilityattime , leaf>5cm).Fecunditywasdefinedasthe numberofseedlings F F G sml lrg 54 54 65 ) andreproductiveadults ( ,F ,F ,G s production wasalsoestimated astheproductofmean / W 64 64 76 ,F ,F ,G S. pratensis o ) 58 58 86 (Ågren &Schemske1993). The ,F ,F t and the 68 68 are notaffected byinbreedingdepression sml 46.6% 13.6% 27.3% establishment probability attime S. pratensis submitted flow ). The sizeclassification was id id id inbreeding ). The percentage change g s r Low = 40% sdl =2%id = 20% = 10% has nolong-termseed ef 2cm; , leaf t+1 sdl W . Giventhatsome s ), smallrosettes sml and ≤ id id inbreeding class within W g s r = 40% High = 60% = 20% o id are the sml ) was t+1 , 2 S. . 95. Chapter 6 Demographic implications of inbreeding . e S. N y = (Chapter S. pratensis populations across S. pratensis production was due to production was as seed set and seed g sml id at each time step. Seedlings e and S. pratensis N and s sdl id (i.e. the number of reproductive plants in of reproductive plants (i.e. the number e N is a function of p, of 100 corresponded to an outcrossing probability of of 100 corresponded to an outcrossing e N (Table 1). Selfed adult plants can also produce two types 1). Selfed adult plants (Table r id coefficients to reduce vital rates (Table 1). rates (Table to reduce vital coefficients and the genetic variability of e id N ), where a -1 ] x · -0.05 included two cycles (Fig. 1). The matrix of the outcrossing cycle is based on The matrix included two cycles (Fig. 1). Inbreeding depression was included into the demographic model using two demographic model using was included into the Inbreeding depression The life-cycle graph, and the corresponding population projection matrix, of The life-cycle graph, and the the population) and the outcrossing probability were proportionally related. The were proportionally related. and the outcrossing probability the population) function ( followed a sigmoidal Ne and inbreeding probability relationship between [1 + 90 · e Matrix analyses and simulations on of inbreeding assess effects analyses to carried out three different We of progeny, selfed and outcrossed. The selfed progeny will remain within the selfing The selfed progeny will remain selfed and outcrossed. of progeny, by inbreeding depression whereas the outcrossed progeny will move cycle affected and juvenile fates are out to the outcrossing cycle (Fig. 1). It is assumed that seedling and that inbreeding plants, independent of the inbreeding history of parental the demographic assume that depression does not increase with inbreeding load. We those of the proportion of selfed individuals in the population are greater than effects caused by the inbreeding history of the individuals. and small rosettes produced from selfing enter into the selfing cycle where their and small rosettes produced from selfing enter into the growth is also reduced by (Vergeer et al. 2003). Second, those vital rates or components of vital rates that of vital or components rates Second, those vital et al. 2003). (Vergeer 1; see the Results (Table between pollination treatments showed significant variability The reduction in demographic transition matrices. section) were modified in the depression was conducted by multiplying the transition transitions due to inbreeding created two inbreeding scenarios to value. We reduced its id that by a coefficient of two sets by obtaining of inbreeding depression simulate low and high effects matrices with different Seedling and smalle rosette establishment probabilities were obtained from a field were obtained probabilities rosette establishment and smalle Seedling 10 site and year, In each in two meadows. in 1999 and 2000 conducted experiment and of seedlings and the number were set up 100 seeds each × 50 cm) with (5 plots The same probabilities 5). sowing (Chapter one year after recorded small rosettes in all years and temporal variability were used for temporal variability in seed production. To keep the model of the life cycle simple we model of the life cycle simple keep the To in seed production. temporal variability event anyway in for ramification, which is a rare did not account 5). size population effective criteria. First, the germination are the first components of fecundity affected by inbreeding depression of fecundity affected germination are the first components of selfing, The probability 1). (Table 0.5. With this function, effective population sizes higher than 250 have no inbreeding 0.5. With this function, effective study on the The function was based on a recent rates. on vital depression effects relationship between the Netherlands indicating that the inbreeding coefficient increased with decreasing increased with decreasing the inbreeding coefficient the Netherlands indicating that pratensis can produce selfed seeds which Adult plants matrices. observed population transition rate that is reduced by have an establishment 96. Chapter 6 Demographic implications of inbreeding matrices modifiedwithhighinbreedingdepression (see Table 1for of matricesmodifiedwithlowinbreedingdepression, andanotherusingthesetof recruitment andgrowth ofpre-adultclasses(deKroonetal.1987; Gillman etal.1993; dependence onvital rates. Assuming thatdensitydependence mostlyaffects individuals. population sizesof20(withselfingprobability for 30years. The effects ofinitialpopulationsizewereincorporatedbysimulating initial the total numberofindividualsinthepopulation was<1.Eachsimulationrunlasted populationwentextinctwhen the meantimetoextinctioncomputedover 100 runs. A For eachsimulatedpopulation,wecomputed themeanfinalnumberofindividualsor latter twosimulations, one usingtheobservedsetofmatrices(noinbreeding; observed matricesateachtimestep(Caswell2001).We ran themodelthreetimes: stochasticity intothemodelbyrandomlyselectingamatrixfrompooloffour conducted stochasticsimulationson growth rate(deKroonetal.1986;Caswell2001).Second,foreachpopulation,we matrices todeterminetherelativecontributionoflife-cycletraits tothepopulation inbreeding depression.We performedelasticityanalysison the observedpopulation rates, First, foreachpopulationandyear, wecomputeddeterministic populationgrowth are alsogiven. population growthrateisindicatedinparentheses. The mean(SE)populationgrowthratesoveryears their elements (see Table 1)duetolowandhighinbreedingdepressioneffects. The change(%) in Meent) Deterministic populationgrowthratesofthedecreasing(Veerslootlanden) andincreasing(Bennekomse Table 2. 02-20 .9 .3 1.3 1.048(19.07) 1.077(17.91) 1.134(12.43) 1.043±0.017 1.005(16.60) 1.162(11.43) 1.032(20.00) 1.124±0.021 1.075(10.79) 1.295 1.124(12.87) 1.275±0.028 1.312 1.205 0.833(2.46) High inbreeding 1.290 Low inbreeding 0.839(7.70) Mean ±SE Observed matrices 0.833(2.46) 0.829±0.030 2002 - 2003 0.885(16.51) 0.859(5.50) 2001 - 2002 0.760(15.65) 0.859±0.032 0.937(11.60) 2000 - 2001 0.854 0.808(10.32) 0.931±0.052 1999 - 2000 0.909 Year 1.060 High inbreeding 0.901 Low inbreeding Mean ±SE Observed matrices 2002 - 2003 2001 - 2002 2000 - 2001 1999 - 2000 Year Succisa pratensis λ Finally weexamined the jointeffects of inbreedingdepressionanddensity , fortheobservedmatrices,andsets ofmatriceswithlowandhigh populations foreachyearofstudyandresultingmatricesafter modifying p varied asafunctionof Decreasing population Increasing population S. pratensis p N = 0.97)and100( e populations includingenvironmental at eachtimestepofthesimulation. p= 0), anotherusingtheset p = 0.5)reproductive id values). Inthe 97. Chapter 6 Demographic implications of inbreeding 1,19 sml . With -1 ) x ) and 12 G -0.04 · and 11 S = 5.89, P = 0.018), (i.e. 1,65 (1 + 4000 · e sdl y = exhibited high survival rates (> exhibited high survival rates that decreases the matrix entry as total the matrix that decreases d , we used another sigmoidal function that function another sigmoidal , we used S. pratensis S. pratensis = 5.47, P (F = 0.026) and growth of pre-adult plants 1,33 ) were multiplied by ) were multiplied 24 G increases. Given that we did not have empirical data to estimate the empirical data did not have Given that we increases. and . The sigmoidal function had the form sigmoidal function had the form The . N 23 d G , 22 and S , N Llike other long-lived perennials, Llike other long-lived perennials, for the decreasing Density-independent stochastic simulations indicated that simulations For the decreasing population density-dependent stochastic 21 G density dependence function for dependence function density related this function, populations with more than 300 individuals were strongly affected by affected individuals were strongly with more than 300 this function, populations effects. density dependent population size population Inbreeding depression significantly affected seed set (F affected Inbreeding depression significantly P= 11.75, of decrease in seed set, percent germination and growth The = 0.003). 1). to inbreeding depression ranged 14 - 47% (Table seedlings and small rosettes due low recruitment rates (range 0.26 - 2.3 and 0.19 - 1.4 67.5% for all size classes) and per reproductive plant, respectively). On average stasis seedlings and small rosettes up to 42 - 84%, growth 14 - 43% and fecundity 2.2 - and retrogression accumulated by inbreeding Each demographic transitions affected elasticity. 15% of the total of pre-adult individuals) accounted for less than 7% depression (fecundity and growth rates of Population growth elasticity. 28.9% of the total that altogether summed up to population of study were 0.93 and 1.28, respectively the decreasing and increasing the matrices, deterministic 2). When inbreeding depression was included into (Table and decreasing population were reduced 2.5 - 11.6% population growth rates of the The 2). high inbreeding scenarios, respectively (Table 2.5 - 16.5% for the low and for the increasing reduction due to inbreeding depression was more pronounced rates of 10.8 - 12.9% population with reductions in the deterministic population growth 2). respectively (Table and 16.6 - 20.0% for the low and high inbreeding scenarios, 50% under the low and population the mean time to extinction was shortened about to the no inbreeding scenario (Fig. 2a). For high inbreeding scenarios as compared by reducing effect the increasing population, inbreeding also had a dramatic still exhibited an population growth with more than 60%, though simulated populations of small initial population size were more The effects increasing dynamics (Fig. 2b). (Fig. 2a and 2b). dramatic for the increasing than for the decreasing population from the density-independent stochastic as those obtained yielded the same results by density hardly affected simulations (Fig. 2c), due to the fact that recruitment was density however, dependence at low population sizes. For the increasing population depression decreased dependence strongly reduced population growth. Inbreeding density dependency (Fig. population growth slightly in the increasing population with Results Crawley 1997: 491), all demographic transitions involving transitions involving demographic 1997: 491), all Crawley (i.e. percentage germination (F percentage 98. Chapter 6 Demographic implications of inbreeding simulations for demographic andinbreedingdata isessentialtoobtain accurateresults. Our inbreeding data) whileusingroughestimationsforthe other. detailed data foronlyoneofthetwocomponents required(either demographicdata or be misleadingtomodelthepopulationdynamicsofaparticular speciesbasedon with hand-pollinationdata forinbreedingdepressioneffects onlife-cycle traits. Itcan 2002) byfullyparameterizing themodelwithfielddata fordemographicattributesand of otherdemographicgeneticmodels(Alvarez-Buyllaetal.1996;Oostermeijer Brook etal.2002;Frankham2003).However, ourstudydoes avoidacommonpitfall Saccheri etal.1998;Bijlsma2000;Groom&Preuninger Tanaka 2000; increasing theriskofextinction(Halley&Manasse1993;NewmanPilson1997; also indicatedthatinbreedingdepressioncanstronglyaffect populationdynamics mean timetoextinction,respectively. Overall,ourresults agreewithotherstudiesthat decreasing populationsgenerallyshowedimportant reductionsingrowthratesand consequences forthe The results showedthatinbreedingdepressionhasimportant demographic Discussion dependent functionwhereinbreedingrateswereverylow. increasing populationgrewuptothecarryingcapacity determinedbythedensity inbreeding hadhardlyanyeffect atallinpopulationgrowth(Fig.2d).Infact,the 2d). This wastrueforthesmallinitialpopulationbutnotlargewhere significantly increases withdecreasingpopulationsizein inbreeding depression onpopulationbehavior, astheinbreedingcoefficient demographic implications ofinbreedingdepression. quantifying theirproportionalchangebecome essentialtoaccuratelyestimatethe identifying thedemographictraits thataretruly affected byinbreedingdepressionand of inbreedingdepressionwouldhaveeven beenmoredramatic.Forthisreason, adult survival)wereaffected byinbreedingdepression,thedemographicimplications life-cycle traits moreimportant forthepersistence of altogether accumulatedlessthan30%ofthe total elasticity. This meansthatifother accounted forbycomponents oflife-cycletraits affected byinbreedingdepressionthat It mustalsobeemphasizedthatsuchareduction inpopulationperformancewas reduce thespeciespopulationfitnessespeciallyfordecreasingandsmallpopulations. 1996). Nonetheless,ourresults suggestthatinbreeding depressioncanstrongly populations topersistforlongtimeinagivenarea(i.e.remnantdynamics;Eriksson inbreeding depression. inbreeding whereasdecreasingpopulationsbecomemorepronetoextinctiondue In thecaseofassessingaspeciesconservationalstatus, usingdetailed Changes inpopulation sizerepresentthemechanismthattunes effects of S. pratensis Succisa pratensis Succisa pratensis clearly indicatedthatincreasingpopulationscancopewith populations studied. The increasingand has alonglifespan thatpermits extant S. pratensis S. pratensis populations (e.g. (Vergeer etal. 99. Chapter 6 Demographic implications of inbreeding S. S. pratensis as well as many other plant species in Europe, the as well as many other plant species S. pratensis population decline should include habitat restoration (Vergeer et al. 2003) as restoration (Vergeer population decline should include habitat Density dependence and inbreeding depression do not interact as long as the Density dependence and inbreeding Theoretically, the extinction of small populations could be prevented by the the extinction Theoretically, habitat size is large enough to harber sufficient numbers of inidividuals to eliminate the numbers size is large enough to harber sufficient habitat quality can interact size, habitat In contrast to habitat risk of inbreeding depression. has been shown that the performance of with inbreeding depression. It (Bakker & Berendese 1999) resulting in reduced is high extent of fragmentation This scenario limits populations (Soons & Heil 2002). of extant colonization capacity and poses a serious threat to the survival of the genetic rescue effect the effectiveness suggested that purging of many plant species in fragmented landscapes. It has been of inbreeding depression reduce the effects might of deleterious mutations could increase their (Charlesworth & Charlesworth 1999), so that small populations reviews to assess the extent of performance under high inbreeding levels. However, of purging is controversial (Byers & Waller purging have revealed that the importance not strong enough to be of 1999; Miller & Hedrick 2001) and that purging is probably et al. 1997; Koelewijn practical use in eliminating inbreeding depression (Dudash Although the link between 1998; Lacy & Ballou 1998; Frankham et al. 2001). persistence is still largely increasing genetic variation, mean fitness and population measures to prevent unknown (Ingvarsson 2001), studies focusing on conservation 2003). We have included two sources of variation that affected population size and population size affected variation that sources of included two have 2003). We stochasticity first, environmental depression: of inbreeding the extent subsequently second, density- and conditions variation in environmental year-to-year due to size limitation. mimicking habitat growth population that limits recruitment dependent carrying capacity stochasticity, studies that included environmental Other simulation of 1993) also stressed the importance depression (Halley & Manasse and inbreeding depression, of inbreeding the effects between these factors. In general, the interaction probability, size and outcrossing the relationship between population which depend on the showed that of the system. Our results on the carrying capacity depend in turn than by density by inbreeding depression was more affected decreasing population The opposite the carrying capacity. the population never reached dependence since by population that was less affected by the increasing was exhibited pattern by density-dependent recruitment. inbreeding depression but limited populations can be as much affected by habitat quality as by inbreeding depression quality by habitat affected populations can be as much population size quality leads to decreasing Deteriorating habitat et al. 2003). (Vergeer and inbreeding depression, factors that act in concert as well as increasing inbreeding support et al. 2003). Our simulations empirically increasing extinction risks (Vergeer populations, presumably due to poor habitat this view showing how decreasing the so-called ‘extinction vortex’conditions, can enter faster into (Gilpin & Soulé 1986) traits. of inbreeding depression on demographic due to the effects from surrounding populations that is, the arrival of immigrants genetic rescue effect, and reduce inbreeding depression (Ingvarsson that could replenish genetic variation 2001). In the case of pratensis to the loss of genetic well as reintroduction of individuals from other populations et al. 2004). variation and inbreeding depression (Vergeer 100. Chapter 6 Demographic implications of inbreeding Ågren J&SchemskeDW1993.Outcrossingrateandinbreedingdepressionintwoannualmonoecious Alvarez-Buylla ER,García-BarriosR,Lara-MorenoC&Martínez-RamosM1996.Demographicand References project(EVK2-1999-00004). the EuropeanUnion TRANSPLANT by NetherlandsOrganizationforScientificResearch(NWOproject805-33-451)and valuable logisticassistance atdifferent phasesofthisstudy. This studywassupported We thankthestaff oftheBotanic GardenoftheUniversityNijmegenwhoprovided Acknowledgments Bijlsma R,BungaardJ&Boerema AC 2000.Doesinbreedingaffect theextinctionriskofsmall &BerendseF1999.Constraints in the restorationofecologicaldiversityingrasslandsand Bakker JP ila ,BlokJ,Sletw la ilB19.Adensity-dependent modelof Gillman M,BullockJM,Silvertown J&ClearHillB1993. A 2001.Inbreedingandextinction: effects ofpurging. Frankham R,GilliganDM, Morris DR&BriscoeDA Frankham R2003.Geneticsandconservationbiology. -ComptesRendusBiologies326:22-29Suppl. Eriksson O1996.Regionaldynamicsofplants: areviewofevidenceforremnant,source-sinkand Dudash MR,CarrDE&FensterCB1997.Fivegenerations ofenforcedselfingandoutcrossingin Dudash MR&FensterCB2000.Inbreedingandoutbreeding depressioninfragmentedpopulations.- &vanGroenendaelJM1987.Densitydependent simulationofthepopulation de KroonH,Plaisier A de KroonH,Plaisier A, vanGroenendaelJ&CaswellH1986.Elasticity:therelativecontributionof Crawley MJ1997.Lifehistoryandenvironment.-In:(ed),PlantEcology. Blackwell Charlesworth D&B1999. The geneticbasisofinbreedingdepression.-GenetRes74: Caswell H2001.Matrixpopulationmodels.Construction,analysis,andinterpretation, Second edition. Caro TM &LaurensonMK1994.Ecological andgeneticfactorsinconservation:acautionarytale. - Brook BW, Tonkyn DW, O’GradyJJ&FrankhamR2002.Contributionofinbreedingtoextinction riskin &Waller DM1999.Doplantpopulationspurgetheirgenetic load?Effects ofpopulationsize Byers DL Bühler C&SchmidB2001. The influence ofmanagementregimeandaltitudeonthepopulation Boyce MS1992.Populationviabilityanalysis.- Annu RevEcolSyst23:481-506. herbs, tree species.- Annu RevEcolSyst27:387-421. genetic modelsinconservationbiology:applicationsandperspectivesfortropicalrainforest populations? predictionsfrom heatland communities.- Trends EcolEvol14:63-68. ouaindnmc sn il-siae aaee aus eooi 6 282-289. population dynamicsusing field-estimatedparameter values.-Oecologia96: - ConservGenet2:279-284. metapopulations. -Oikos77:248-258. 51: 54-65. Mimulus guttatus populations. CambridgeUniversityPress,Cambridge. In: Young AG &ClarkeGM(eds),Genetics,demographyandviabilityoffragmented demographic parameters topopulationgrowthrate.-Ecology67:1427-1431. Science, Oxford. 329-340. - Sinauer, Sunderland. Science 263:485-486. threatened species.-ConservEcol6(1): Art No16http://www.consecol.org/vol6/iss1/art16 and matinghistoryoninbreedingdepression.- Annu RevEcolSyst30:479-513. structure of dynamics ofaperennialgrasslandspecies, Begonia hirsuta Succisa pratensis : Inbreedingdepressionvariationatthepopulationand familylevel.-Evolution and B. semiovata : implicationsforvegetation monitoring.-J Appl Ecol38:689-698. Drosophila . -Evolution47:125-135. . -JEvolBiol13:502-514. Hypochaeris radicata . -Oikos50:3-12. Cirsium vulgare 101. Chapter 6 Demographic implications of inbreeding . Plantago coronopus Plantago . - J Ecol 91: 18-26. . - Evolution 51: 354-362. Succisa pratensis Clarkia pulchella . - J Evol Biol 14: 595-601. (Onagraceae). - Evol Ecol 14: 155-180. - Evol Ecol (Onagraceae). during generations of inbreeding. - Evolution 52: 900-902. during generations Drosophila melanogaster Clarkia concinna heterosis. Conserv Biol 18: 812-821. in plant conservation. - Biol Conserv 113: 389-398. 113: in plant conservation. - Biol Conserv prior environment, and lineage. - Evolution 57: 1822-1828. of inbreeding, effect - Nature 392: 491-494. in a butterfly metapopulation. of increase in woody and herbaceous to the finite rate of life-cycle components importance perennials. - J Ecol 81: 465-476. 90: 1033-1043. dispersed grasslands forbs. - J Ecol Popul Ecol 42: 55-62. versus quality and population size on the performance of population size: experimental populations of population size: experimental ME (ed), Conservation biology: the science of scarcity and diversity. Sinauer, Sunderland. Sinauer, of scarcity and diversity. the science Conservation biology: ME (ed), in depression depression. - Evol Ecol 7: 15-24. depression. - Evol herbs. - Oikos 105: 159-167. removal in three perennial Evol 16: 62-63. - Evolution 52: 692-702. Peromyscus polionotus I. Genetic models. - Evolution 39: 24-40. populations of Oostermeijer JGB, Luijten SH & den Nijs JCM 2002. Integrating demographic and genetic approaches Oostermeijer JGB, Luijten SH & den Reed DH, Lowe EH, Briscoe DA in a novel environment: Fitness and adaptation & Frankham R 2003. I 1998. Inbreeding and extinction Fortelius W & Hanski P, Vikman Saccheri I, Kuussaari M, Kankare M, AI & Mendoza Silvertown J, Franco M, Pisanty relative plant demography - 1993. Comparative in fragmented populations of wind- colonization capacity Soons MB & Heil GW 2002. Reduced - Cambridge University Press, Cambridge. populations for conservation. Soulé ME 1987. Viable by inbreeding depression under stochastic environments. 2000. Extinction of populations Y. Tanaka, Introduction strategies put to the test: local adaptation Sonderen E & Ouborg NJ 2004. P, Vergeer A Rengelink R, Copal of genetic variation , habitat The interacting effects P, & Ouborg NJ 2003. Vergeer Newman D & Pilson D 1997. Increased probability of extinction due to decreased genetic effective Newman D & Pilson D 1997. Increased Gilpin ME & Soulé ME 1986. Minimum viable populations: processes of species extinction. - In: Soulé extinction. - In: of species populations: processes Minimum viable & Soulé ME 1986. Gilpin ME magnitude of inbreeding influence the type can TE 2000. Population & Preuninger Groom MJ Halley JM & Manasse RS 1993. A1993. & Manasse RS Halley JM subject to inbreeding model for annual plants population-dynamics rosette bud life history responses to flower and E & de Kroon H 2004. Flexible Hartemink N, Jongejans Ingvarsson PK Ecol Trends rescue hypothesis. - of genetic variation lost - the genetic 2001. Restoration Koelewijn HP on progeny fitness in levels of inbreeding of different 1998. Effects Lacy RC & Ballou JD 1998. Effectiveness of selection in reducing the genetic load in populations of the genetic load in populations of selection in reducing JD 1998. Effectiveness Lacy RC & Ballou Lande R 1988. Genetics and demography in biological conservation. - Science 241: 1455-1460. and demography in biological Lande R 1988. Genetics of self fertilization and inbreeding depression in plants. The evolution Lande R & Schemske DW 1985. of inbreeding depression and fitness declines in bottlenecked Miller PM & Hedrick PW 2001. Purging 7 Effects of increased productivity on the population dynamics of the endangered, clonal herb Cirsium dissectum

Eelke Jongejans, Natasha de Vere & Hans de Kroon

Summary

Plant species with rhizomatous runners have the potential to effectively exploit and occupy their habitat. But even in this group rare plants are not uncommon, as is examplified by Cirsium dissectum, a characteristic clonal herb of nutrient- poor, wet grasslands, which is endangered by habitat destruction and nutrient enrichment. In this paper we aim to investigate how rosettes of C. dissectum respond nutrient enrichment and what the consequences are for the population dynamics. A previous experiment revealed that this species increases its allocation to seed production and not to vegetative biomass when plots with this species and a common grass competitor were fertilized. Here we further analyze these experimental allocation patterns and data on excavated plants in seven sites of increasing productivity, for underlying mechanisms. Both sexual and clonal reproduction increased with rosette size. An additional greenhouse experiment on rhizome formation confirmed the size-dependency of clonal propagation and showed no inherent differences between plants from different sites. Demographic data were recorded in permanent plots in three sites from 1999 till 2003. Elevated clonal propagation and rosette mortality were found in both the fertilized garden plots and the most productive site. The combination of faster rosette growth and higher biomass turn-over therefore explained the observed trend of higher percentages of flowering rosettes at high nutrient availability. Furthermore, increased rosette mortality rates may cause decreasing genetic diversity in the already small, remnant populations of C. dissectum, as seedling establishment seems to be limited to bare soil.

Keywords: Cirsium dissectum; clonal propagation; demography; life table response experiments; nutrient availability; projection matrix model; rosette turn-over; sexual reproductive allocation; size dependent reproduction 104. Chapter 7 Population dynamics of Cirsium dissectum above-mentioned objectives weuseasetofexperiments andfieldobservations: 1) To population dynamicsof populations. importance ofshifts inallocationdependsontheirimpact onthedynamics ofnatural potentially alsobeasourceofvariationinsexual reproductiveallocation.Besides,the a gradientofproductivity. Geneticdifferences betweenpopulationshowevermay allocation pattern orwhethersimilarpatterns canbefoundinnaturalpopulationsover more favorablesites.Itisunclearhowever whatmechanismcausedthisshift in allocation canbeseenasastrategyofescape inspace whenseedsmaydisperseto rosettes andtotal biomasswasnotdifferent. This increasedsexualreproductive fertilized plots compared totheplants inthecontrolplots, whilethenumberofliving percentage oftheclonaloffspring ofthesurviving the genetmortality rateof (Chapter 3). After threeyears thegrasshadbenefittedfromextranutrients, while the tall grass allocation experimentwasperformedinwhichthisspeciesgrowntogetherwith (Abrahamson 1980;Loehle1987). To testthishypothesisfor either wayofreproductionandonthedetectability ofenvironmental cues depending ontheirlifehistorystrategies,theexpectedrelativegainofincreasing vegetation biomassbyincreasingeithersexualreproductionorclonalpropagation enrichment speedsupthisprocessofsuccessioningrasslands. successional grasslandspeciesisreplacedbymorecompetitiveplants andnutrient Tilman etal.1994;Roem&Berendse2000;Soons2003). This early- reductions inthenumberandsizeof habitat fragmentation, desiccation,acidificationandnutrientenrichmenthaveledto grasslands hasan Atlantic distributioninNorthwestEurope. Habitat destructionand Cirsium dissectum clarification. & Klimesová2000). This paradox ofplantrarityandlongrunnersneedsmore longer than10cmisequaltothepercentage forothertypesofclonalgrowth(Klimeš success, sincethepercentage ofrarespeciesamongclonalplants withrhizomes Kroon &Bobbink1997;Marrsetal.2000).Butrunnersmaynotbeaguaranteefor under suitable conditionsandmayformmonoculturesthatholdupsuccession(de Brachypodium pinnatum contains notoriousexamplessuchas occupy space (Eriksson1986;deKroon&Knops 1990). This groupofplantspecies runners maybeaneffective waytoexploitheterogeneousresourcesorrapidly Clonal growthbylongstolonsorrhizomescanbeasuccessfullifehistorystrategyas Introduction In thispaper westudy the roleofclonalgrowthandseedproduction inthe In generalclonalplants havebeenhypothesizedto respondtoincreased An exampleofarhizomatousbutrare(RedList)speciesistheMeadowthistle, Molinia caerulea . This inhabitant ofrathernutrient-poor, species-rich,wet C. dissectum which canquicklyattain dominancethroughclonalgrowth C. dissectum in nutrientenrichedplots andunfertilizedcontrolplots C. dissectum in responsetonutrient enrichment. To meetthe Phragmites australis a ihri etlzdpos higher was higherinfertilizedplots. A C. dissectum populations (Saundersetal.1991; , Pteridium aquilinum plants floweredinthe C. dissectum and an 105. Chapter 7 Population dynamics of Cirsium dissectum C. ) ) * * ( ( Figure 1. in the allocation experiment: (a) number Source rosettes live No. Nutrients removalBud N*B 1Error 115.3 df 12.47 *** 1 1Source MS rosettes live No. 174.8Nutrients 0.0 18.90 *** removalBud F 0.00 1 53N*B 1Error 56.27 26.1 df 9.2 4.42 2.82 1 * 1Source MS rosettes live No. 9.49Nutrients 0.53 removalBud 0.74 107 0.04 F 1N*B 1Error 12.74 10.12 6.26 df 0.40 0.49 1 1 MS 100.52 0.08 3.96 0.00 F 1 7 56.56 25.41 2.23 SourceNutrients removalBud N*BError df 1 1Source 713.7 MS rosettes live No. 58.9 2.12 Nutrients 0.18 1 removal 55Bud F N*B 1 336.1 35.7Error 0.056 df 0.11 0.44 1 1Source MS rosettes live No. 0.006 0.076Nutrients 0.05 Error 0.60 F 1 53 1 0.000 0.126 0.296 df 0.00 7.52 ** 2 MS 0.224 33 5.69 ** F 0.039 Cirsium dissectum =23 =16 =6 =55 =23 =23 n n n n n n added ) * ( ** *** =36 =36 =20 =20 =6 =6 =57 =57 =36 =36 =36 =36 n n n p<0.10; * p<0.05; **p<0.01; *** p<0.001 n n n ) * ( Control Nutrients 8 6 4 2 0 5 0 0 5 0 0 0 20 15 10 15 10 0.5 0.4 0.3 0.2 0.1 0.8 0.6 0.4 0.2 30 20 10 per plot at harvest after dividing by the number of rosettes, (c) the percentage of rosettes in dividing by the number of rosettes, (c) the percentage per plot at harvest after Flowering rosettes inrosettes rosettes inrosettes September September September September September per rosette in rosette per in September Number of live grey=flowered) rosettes inrosettes June Number of dead percentage of live from the origin in (black=vegetative; individual rhizomes individual rhizomes Total dry weight [g] [g] weight dry Total Mean distance [cm] Mean length [cm] length Mean of (f) (e) (d) (c) (b) (a) of rosettes alive (i.e. with green leaves) at harvest, September 2002, (b) the total dry weight [g] of (b) the total of rosettes alive (i.e. with green leaves) at harvest, September 2002, Overview of the effect of nutrient addition on Overview of the effect dissectum [cm] of the (d) the mean distance without bud removal only,) that flowered June 2002 (of the plots rosette at the beginning of the rosettes at harvest measured from the position of the first, planted connected two rosettes in 12 experiment, (e) the mean length [cm] of all individual rhizomes that in which the rhizomes were revealed, and (f) the number of rosettes in those randomly selected plots ANOVA that were dead. 12 plots The number of are given on the right. on these six parameters results and bud removal as fixed live rosettes at the time of measuring is used as covariate, nutrient addition factors. N*B is their interaction. 106. Chapter 7 Population dynamics of Cirsium dissectum Meadow thistle, Study system Materials andmethods clonal propagation topopulationgrowth. productivity toquantifythecontributionsofrosettesurvival,sexualreproductionand population dynamicsof experiment withplants fromfivedifferent populations.4) And finallywestudythe differences ingrowthpatterns aregeneticorplasticweperformasmallgreenhouse between sevenfieldpopulationsoveragradientofproductivity. 3) To testwhether populations, wecompare plantsizes,sexualreproductionandclonalpropagation growth androsettemortality. 2) To checkiftheseallocationshifts alsooccurinnatural allocation experimentisfurtheranalyzedtoinvestigatetherelationwithofclonal better understand themechanismofobservedincreasedseedproduction, fertilization ofthesitestheyinhabit(Buck-Sorlin1993). and manysurvivingpopulationsweredecimatedinsize,duetodrainage dissectum (Kay &John1994;Buck-SorlinWeeda 2000).Fifty-seven percentofthe North-west Europe. These habitats havedeclinedsubstantially innumberandextent to oligotrophicwetgrasslandsandheathlandsoverits entiredistributionin Atlantic it canformdensestands byclonalpropagation. up tohalfwaythe60cmtall floweringstem. After seedsettherosettediesoff. Locally Flowering stemsareformedapically. Duringstemelongationrosetteleavesareraised winter theabove-groundrosetteisreducedtooneortwosmall,fleshyleaves. forms rosetteswithuptofivesoftly prickledleaves,butmostlyonlytwoorthree.In plots wereenrichedwith theequivalentof120kgNha plants thatrapidlyincrease biomasswhenmorenutrients areavailable.Half of the six clumps of Wageningen University (Chapter3).One we performedathree-yearexperiment in anutrient-poor, sandygardenof To investigatetheeffects ofnutrientenrichmentonallocationtosexualreproduction Allocation experiment 2003). currently beinginvestigatedbythefloristicfoundation FLORON(Rossenaar&Groen 1998; Lucassenetal.2003). The status oftheDutch rather sensitivetotheeffects ofacidification(deGraafetal.1997;de populations inGermanyareestimatedtohavebecomeextinctsince1930, Molinia caerulea Cirsium dissectum C. dissectum to mimicagrassland containing (L.) Hill Asteraceae, isarhizomatousherbthat for fouryearsinthreesiteswithcontrasting C. dissectum Cirsium dissectum Cirsium dissectum -1 rosette wasplantedbetween C. dissectum yr -1 during thesecondand is mostlyrestricted C. dissectum populations is plants are and C. 107. Chapter 7 Population dynamics of Cirsium dissectum Figure 2. ns ns clumps was clumps H n h V N B S =0.02 =0.41*** =0.09 2 2 2 B: R V: R R N: M. caerulea rosettes in the circular plots. rosettes in the Total ** Total C. dissectum Dry weight Caudex [g] Dry weight 0 0.05 0.1 0.15 0.2 0.25 0 0.05 0.1 0.15 0.2 0.25

5 4 3 2 1 0 0 1

No. Flowering stems Flowering No. No. Rhizomes No. At the end of the third growing season, just before the harvesting of the just before the harvesting of the third growing season, At the end of determined on the maps. Detailed observations were made on the below-ground observations Detailed the maps. determined on (nutrient four groups in each of the plots in three randomly selected rhizome network used to High-pressure air was removal treatment combinations). addition and bud to reveal rhizomes and decaying rosettes without remove the soil around the rosettes rosettes and rhizomes were also mapped. These leafless leaves. experiment we mapped the position of all experiment we third year of the experiment, while the other half received no fertilization. To quantify To no fertilization. other half received while the of the experiment, third year buds were growth the flower on vegetative reproduction of sexual the costs the addition and the nutrient in both the plants from half of removed continuously of on the number main results and the on the methods group. More details control chapter 3. are presented in plant parts biomass of different rosettes and the Relationships between flowering and the number of produced rhizomes with the dry weight [g] of the Relationships were randomly selected (see The 159 rosettes caudex of excavated rosettes in the field survey. (S), Bennekomse meent (B), heide (h and H), Stelkampsveld methods) in seven sites: Nijkampse The letters denote sod-cut areas. (V), in which non-capital Konijnendijk (n and N) and Veerslootlanden together: y = (exp (16.6 · x -3.70) ) / (1 + exp logistic regression on flowering was performed on all data Four The linear regression on rhizome number was performed for each site separately. (16.6 · x -3.70) ). are all significant. ** p<0.01; *** regression lines are not shown: they are intermediate to V and B, and p<0.001. In order to study the degree of spatial spread from the initial rosette position, the spread from the initial the degree of spatial In order to study of the six rosette to the geometrical centre of each distance 108. Chapter 7 Population dynamics of Cirsium dissectum containers werewatered daily. Everyfortnightthecontainers wereweighedand their was performedinan open, unheatedgreenhouseroofedwithtransparent foil. The dissectum established readily, and wascutregularlyat4cmheightto avoid shadingthe to mimicthedifference betweensod-cutareasandgrasslands. The poor sandysoil.Halfofthecontainers wereseededwiththegrass in themiddleofa34bycmsquareand 15 cmdeepcontainer filledwithnutrient- does notstart beforeMay. The roots wererinsedwithwater, andonerosetteplaced out fromeachsite. These rosetteswereatleastonewinteroldasclonalpropagation first halfofMay2001smallrosetteswithtwo leaveswererandomlyselectedanddug formation, plants fromthefivesitesweregrownunderequalconditions.During In ordertoinvestigatepossiblegeneticdifferences ingrowthrateandrhizome Rhizome formationexperiment next toeachplot. 70°C. The averagethicknessofthehumuslayerwasdeterminedfromthreesoilcores centimeters abovethesoilsurface,andweighedafter beingdriedfor48hoursat for eachexcavatedrhizome. The vegetation intheplotwasclippedataheightoffour The maximumdepthandthedistances betweenconnectedrosettesweremeasured rosettes werefollowedtotheirdestination,whichwasoutside theplots insomecases. plot wererandomlyselectedandcarefullyexcavated.Rhizomesconnectedtothese of thestemwasmeasured.Fiverosettes(andatleasttwoinsodcutareas)per the numberofleavesandtheirmaximumlength.Forfloweringindividualsheight were mappedwithanaccuracyof5mm. The sizeofeachrosette wasdeterminedas the sod-cutareassincethere contain aminimumoffiverosettes,thiswasreducedtotworosettesfor randomly laidoutineachofthesevenareas.Fornon-sodcutareasplots hadto stalks) andthepercentage vegetation cover. The positionsofall than theminimumnumberofrosetteswereomittedandnewplots chosenrandomly. are referredtoashandn. had beenremovedtoreducesoilfertility(sod-cutting)in1997. These sod-cutareas Common reed, population insiteVwasmowntwicesomeyearsordertohaltencroachmentof E). All sitesaremownonceperyearafter whichhay isremoved. The E),andVeerslootlanden (V) N,6°08’ (52°36’ N,6°26’ E),Konijnendijk(N)(52°02’ N,5°36’ E),BennekomseMeent(B) (52°00’ N,6°29’ E),Stelkampsveld (S)(52°07’ N,6°33’ (52°02’ site characteristicswasstudiedinfivegrasslands. The sitesare:Nijkampse heide(H) The relationshipbetweenrosettesize,floweringprobability, rhizomeformation,and Site survey Within eachplotweestimatedtheheightofvegetation (withoutflowering Between July26 rosettes andtofocus on below-groundcompetitiononly. The experiment Phragmites australis th and August 3 and August C. dissectum . The topsoil ofparts ofthegrasslandsHandN rd , 2001,five50bycmsquareplots were density waslower. Plots containing less C. dissectum A. capillaris Agrostis capillaris rosettes seeds C. 109. Chapter 7 Population dynamics of Cirsium dissectum . The Figure 3. A. capillaris rosette in an open greenhouse Cirsium dissectum S*GS*GErrorError 4 4 50 50 0.093 0.093 0.107 0.107 0.87 0.87 SourceSource weight weightDry Dry SiteSite treatment treatmentGrass Grass S*GS*G 1 1 1 1 df dfErrorError 0.351 0.351 0.147 0.147 16.64 16.64 MS MS *** *** 6.97 6.97 * * 4 4 F F 0.012 0.012 4 4 49 49 0.56 0.56 0.005 0.005 0.021 0.021 0.22 0.22 SourceSourceSiteSite treatment treatmentGrass Grass 1 1 47.277 47.277 440.10 440.10 df df *** *** MS MS 4 4 0.519 0.519 F F 4.83 4.83 ** ** plants separated from the separated plants . The error bars are standard errors. Different letters denote errors. Different The error bars are standard . bb . The number of clonal offspring and the distance and clonal offspring The number of . th NumberNumber Rhizomes Rhizomes of of Dry weight weightDry Dry [g] [g] abab C. dissectum bb ControlControl GrassGrass treatment treatment Agrostis capillaris Sites Sites and November 8 and November abab rd aa HSBNVHSBNV HSBNV HSBNV The experiment lasted five months with the final, block-wise harvest between harvest the final, block-wise five months with lasted The experiment 88 66 44 22 00 00 55 1010 1515 October 23 October weight was brought back to the original 14,000 g by watering. The experiment had a The experiment watering. 14,000 g by to the original brought back weight was and six replicates. two treatments site origins, five containers: design with 60 full block The dry weight [g] and number of rhizomes per with (grass treatment) or without sites were grown in pots experiment in which rosettes from different (control) a competitor, sifnificant difference in dry weight between sites (post-hoc Tukey test). The variation in rhizome number test). Tukey in dry weight between sites (post-hoc sifnificant difference dry weight as covariate, but is successfully explained by both the grass treatment as factor and plant heide Sites were in order of increasing productivity: Nijkampse site of origin was not a significant factor. (V). (S), Bennekomse meent (B), Konijnendijk (N) and Veerslootlanden (H), Stelkampsveld between new rosettes and the original one were determined. For each container the were determined. For each container and the original one between new rosettes and the soil was rinsed biomass was determined for each species after drying for 48 hours at 70°C. drying for 48 hours at 70°C. for each species after biomass was determined 110. Chapter 7 Population dynamics of Cirsium dissectum where calculated asfollows: compared tocontrolplots. Seedlingandsmallrosetteestablishment perseedwere seedlings andsmallrosettesjustbefore12monthsafter seedadditionwas 5x50cm plots inB,N,andONovember19992000 (Chapter4). The number of of seedlingsinthepermanentplots wasverylow. Fifty seeds wereaddedtoten Seedling establishment wasstudiedinaseedadditionexperimentsincethenumber were usedtoparameterize 6by 6projectionmatricesoftheform The demographicdata fromthe permanent plots andtheseedadditionexperiment Matrix parameterization one yearlater( additions plots (SA)andcontrolplots (NS)atthe beginning( represents thetransition elementfromthe Smallvegetative rosettes( Seedlings( 2 1 was usedtoclassifyrosettes.We distinguishedsixclasses(Fig.1ofChapter1): heads wascounted. maximum leaflength.Ifarosettehadproducedfloweringstem,thenumberofflower mapped andmeasured:wecountedthenumberofrosetteleavesdetermined year inJune,when plots wereestablished in April 1999atthreesites:fiveinB,sixNandtwoV. Each Floweringclonal offspring ( 6 Vegetative clonaloffspring ( Floweringrosettes ( 5 Largevegetative rosettes( 4 3 plots inwhichallrosetteswerefollowedfrom1999to2003. The 1bym To studythenaturalpopulationdynamicsof Demography census oftheprecedingyearandflowerinwhichtheyareformed. census oftheprecedingyearanddonotflower. than 12.5cm. combinationofplantsize,floweringstatus, andtheformationofside-rosettes A N p p sml sdl is thenumberofseeds,seedlings( = = ()() ()() NNN NN NNN NN t=1 d d d d t sdl t sdl t sdl t sdl AS SNS NS SA SA m m m m t sml t sml t sml t sml AS SNS NS SA SA 1, 1,0 ,1 ,0 ,1 1, 1,0 ,1 ,0 ,1 sdl === == ). === == −−− −−− C. dissectum ) aretinyrosetteswithnoleaveslongerthan2cm. flow N N seeds SA seeds SA ) aregroupedirrespectiveofrosettesize. sml lrg cl.f cl.v flowers, allrosettesinthepermanentplots were ) havelongerleavesthansmallrosettes. ) arenewrosettesformedbyrhizomessincethe ): largerthanseedlingsbutwithnoleaveslonger ) arenewrosettesformedbyrhizomessincethe j th C. dissectum sdl category inyear ) orsmallrosettes( we established permanent t=0 t ) oftheexperimentor to the sml A ij i th ) intheseed in which 2 category in permanent (2) (1) a ij , 111. Chapter 7 Population dynamics of Cirsium dissectum , cl.f and 11 or Cirsium S ), stasis Figure 4. ) rosette. T cl.v , cl.f flow and , lrg flow , V sml . t . Therefore we assigned new Therefore we assigned . t+1 ) or a flowering ( N ), survival of clonal offspring ( of clonal offspring ), survival Sites C cl.v and lrg , B sml , and bootstrapped 95% confidence intervals for all 95% , and bootstrapped λ .8 .6 , clonal propagation ( propagation , clonal ) (Table 3). Growth is defined as the transition of a smaller or transition of a defined as the 3). Growth is ) (Table 1.4 1.2 1.0

R Population growth rate, rate, growth Population G) λ , another Asteraceae species; Chapter 5). Secondly, because species; Chapter 5). Secondly, Asteraceae , another belonging to a certain class, and the average number of rhizomes belonging to a certain t ), growth ( ), growth F matrices. Site and year codes: Bennekomse Meent (B), Konijnendijk (N), Veerslootlanden matrices. Site and year codes: Bennekomse Meent (B), Konijnendijk . The 36 matrix elements were classified as different life history components: different classified as were elements The 36 matrix . clonally produced which new rosette in year clonally produced which new Three important assumptions were made during the construction of the during the construction of assumptions were made Three important t t+1 , were estimated conservatively high at 55% however (comparable to that of high at 55% however (comparable , were estimated conservatively ) and retrogression ( ) and retrogression 21 S clonal offspring to old rosettes of either of the stage classes rosettes of either of the stage to old clonal offspring produced by either a vegetative ( produced by either a vegetative according to a probability function that was based on two factors: the proportion of all according to a probability function rosettes in year based on the excavated each population separately, The latter was calculated for production of different No distinction was made in the rhizome in the site survey. plants Hypochaeris radicata rhizomes are invisible without excavations we could not determine which rosette in rhizomes are invisible without year G fecundity ( fecundity year Projected population growth rate, ( rosette. Stasis by a surviving class stage a larger or flowering class to stage vegetative survive and rosettes that retrogress remains in the same class means that a rosette as the product was calculated plants size. Fecundity of the flowering but decrease in production of the per seed, the average seed probabilities of the establishment of observed and the mean number rosette in each population, average flowering in year flowering rosette in a population flower heads per dissectum (V), = 1999->2000, = 2000->2001, = 2001->2002 and = 2002->2003. population matrices. Firstly, since only a small number of seedlings were present in number of seedlings were since only a small Firstly, population matrices. experiment, seedling fates, and the seed addition plots both the permanent 112. Chapter 7 Population dynamics of Cirsium dissectum (Caswell 2001p.306). in whichagiven within aspeciesis(Horvitzetal.1997;Caswell 2001): experiments (LTRE). The modelforsuchadecompositionanalysiswithtwofactors S New sexualreproduction( mean limits ofthe95%confidenceintervalswereadjusted forsmalldeviationsbetweenthe of numberflowerheadstheobserved floweringrosettes. The upperandlower occurred whenastage classcontained lessthansixrosettesattime matrix elements ofnewclonally-producedrosettesforthatyear. Similarproblems the previousyear, becausenoobservationsweremadein1998. This leadstoempty year (1999to2000)itwasnotpossibleobservefatesofclonaloffspring formedin probabilities hadtobeestimatedassomedata wereabsentornotsufficient: inthefirst similar assumptionsinhis determined theelasticityvalue rate, the accompanying righteigenvectorwereusedastheprojectedpopulationgrowth using theMatlabstudentedition(1996).Foreachmatrixdominanteigenvalueand The 12(threesitesforfouryears)constructedprojectionmatriceswereanalyzed Matrix analysis the cases. and observations overtheyearsforthatclassinsamepopulation. Apart fromthe the transitionprobabilitiesfromaclassincertain yearwereestimatedfromall times (Efron1982;Kalisz&McPeek1992;Caswell2001)inwhichthesurvival( method byresamplingthedata setofobservedrosettefateseachmatrix3,000 To construct95%confidenceintervalsforthelambdas,weappliedbootstrapping eigenvalue ofthemean ofallmatricesaspecies, of producedrhizomesandplantsizeclassifiedaseither vegetative sizeclassesastherewasnosignificantrelationshipbetweenthenumber Elasticity valuesquantifytherelativecontributionofeachelementto transition matrix(deKroonetal.1986;de2000): , and cl.f λ , andthestable-state distributionrespectively(Caswell2001).Furthermore,we To decomposevariationin λ R λ≅λ+α+β+αβ e classes inthefirstyear, observationnumbersweretoosmallinonly13%of ij )( () () () ) andramification( nmnmn n m mn of the3000newlyconstructedmatricesand the = a λ∂ ij ∂λ a ⋅⋅ ij λ of the m F C th ) elements werecalculatedafter resamplingthedata set () ) observationswereresampledforallclassesseparately. site and Oxalis acetosella e ij of eachmatrixelement λ n th we appliedfactoriallifetable response year iswrittenasthe sum ofthedominant model. Thirdly, sometransition λ (··) , themaineffect ofthe sml or λ a lrg of theoriginalmatrix ij of eachpopulation . Berg(2002)made λ t and sumtoone. . Inthesecases m th T site, , cl.v (4) (3) G , 113. Chapter 7 Population dynamics of Cirsium dissectum (5) (6) (7) . The ) ), clonal Figure 5. mn ( G αβ is 1.141, 1.043 λ ), growth ( F ). Mean R

FGCTSR FGCTSR ) and retrogression ( S

populations: Bennekomse meent (B), Konijnendijk Sites ⋅⋅ + , is +0.072, -0.025 and -0.097 for sites B, N and V mn ), stasis ( ), stasis ()() m FGCTSR FGCTSR AA , and the residual ‘interaction’, and the effect, () T 1 2 α n β + ij ⋅⋅⋅ a + m ∂λ ()() n ∂ AA ⋅⋅⋅ () () () AA 1 2 () 1 2 , are multiplied by the sensitivity values of the matrix , are multiplied by the sensitivity ⋅⋅ year, year, ij Cirsium dissectum (··) a ij th ∂λ a ∂ B N V A

n ∂λ ∂ mn m n () () ij ij aa , of three () FGCTSR FGCTSR λ ⋅⋅⋅ , ij ∑ () () ⋅⋅⋅ ij ij () () ij ij aa aa () 0.2 0.2 0.1 0.1 0 0 0.8 0.6 0.4 0.8 0.6 0.4

0.2 0 0.2 0

() -0.1-0.1 -0.2-0.2

, LTRE site effect site LTRE Elasticity ( ), survival of clonal offspring ij , mn ∑ ij () ∑ C ∼ mm nn αβ = − − α − β α= − β= − ⎛⎞ ⎜⎟ ⎝⎠ , the main effect of the effect , the main m main effects can be estimated filling in this equation first for each level of the main for each level equation first filling in this be estimated can main effects interaction effects The main and term. the interaction while ignoring separately, effects 2001): each matrix element (Caswell into contributions from can now be decomposed of of each matrix element with the corresponding matrix element in which differences the overall mean matrix, respectively. α and 0.968, and netto LTRE site effects, site effects, and 0.968, and netto LTRE Elasticity values (mean and standard error) and site effects in the decomposition (LTRE) of variation in in the decomposition (LTRE) error) and site effects Elasticity values (mean and standard population growth rate, matrix (V). Elasticity values and positive and negative contributions of (N) and Veerslootlanden grouped by life history component: fecundity ( are separately elements reproduction ( 114. Chapter 7 Population dynamics of Cirsium dissectum was removedhadlowerbiomass(64and98 gm m The grasslands(withnotopsoilremoval)ranged inmownbiomassfrom197to363g Site survey but its variationwasexplainedbyneitherthebudremoval northenutrienttreatment. years ofthestudy. This percentage increasedsignificantlywiththenumberofrosettes, by rhizomestoanotherrosette. The otherconnectionshad decayedoverthethree flowered (15% respectively), andanincreaseintheproportionofrosettesthatweredeadbuthadnot of rosettesthathadflowered(2.3% higher inthenutrientaddedplots. This wasduetobothanincreaseintheproportion higher turn-overrate:bothrosetteproductionandmortality hadverylikelybeen treatments (Fig.1e). The higher percentage ofdeadrosettes(Fig.1f)suggesteda explained bydifferential rhizomelength,asthisparameter didnotdiffer between away fromthepositionoffirst,plantedrosette(Fig.1d). This couldnotbe nutrients (Fig.1c). The rosettesinthenutrient-richplots werealso significantlyfurther those rosettesthatfloweredhoweverwassignificantlyhigherintheplots thatreceived also hadnoeffect onthedryweightofindividualrosettes(Fig.1b). The percentage of caerulea The nutrientadditiontreatmentincreasedthebiomassoftall grass Allocation experiment Results components ( negative values,whichwesummedseparately foreachofthesixlifehistory contribution matricesofthemainandinteractioneffects contain bothpositiveand halfway betweenthematrixofinterestandoverallmeanmatrix. The resulting heavier offloweringrosettes thanofvegetative rosettesinbothgrasslands and sod- between grasslands, but tendedtobelowerinthesod-cutareas. The caudexwas Seedlings wereonlyfound inthesod-cutareas. not correlatedwithrosettedensityorplot productivityineithervegetation type. productive grasslandandinonesod-cutarea. Whetherarosettefloweredornotwas for thepercentage ofrosettesthatfloweredper plot,whichwashighestinthemost between plots whichoverruledpossibledifferences betweensites. The sameistrue more baresoilthangrasslands. -2 and hadcomparable vegetation height(Table 1). The areasinwhichthe topsoil Average plantsizeasmeasuredthe dryweightofthecaudexdidnotdiffer , butnotthebiomassof F vs , G . 29%)respectively. Onaverage30%oftherosetteswereconnected , C , T , S and R Cirsium dissectum ). Cirsium dissectum vs . 19%)inthecontrolandfertilizedgroup -2 ), nohumuslayerandsignificantly (Chapter 3).Nutrientenrichment rosette densityvariedgreatly Molinia 115. Chapter 7 Population dynamics of Cirsium dissectum Figure 6. 1.0 in site N, and ) in the stable stage ) in the stable vs. cl.v + cl.f )) nsns ** (( =0.19 =0.19 22 =0.71 =0.71 22 V: R R V: V: , and the percentage of all rosettes that are , and the percentage V: R R V: V: 2.0 in site B, 0.29 2.0 in site B, 0.29 λ nsns vs. =0.20 =0.20 nsns 22 )) B: R R B: B: ** =0.30 =0.30 (( 22 nsns ClonalClonal offspring offspring[%][%] FloweringFlowering rosettes rosettes [%] [%] B: R R B: B: =0.71 =0.71 22 =0.14 =0.14 22 N:N: R R (b) (b) (a) (a) N:N: R R 000 255075100 255075100 255075100 000 5 5 5 10 10 10 15 15 15 20 20 20 111 111

) or are newly formed by clonal propagation ( ) or are newly formed by clonal propagation

0.80.80.8 0.60.60.6 1.41.41.4 1.21.21.2 0.80.80.8 0.60.60.6 1.41.41.4 1.21.21.2

Population growth rate, rate, rate, growth growth Population Population Population growth rate, rate, rate, growth growth Population Population λ λ λ λ = p<0.10 ) * ( 3.0 in site V, for respectively vegetative and flowering rosettes). Rhizome and flowering rosettes). for respectively vegetative 3.0 in site V, flow + cl.f vs. cut areas (Table 2 and Fig. 2). The number of rhizomes produced by an excavated by an of rhizomes produced The number Fig. 2). 2 and (Table cut areas and for site B in all sites except caudex weight its correlated with was positively rosette the within with plot productivity correlated also positively 2), and was N (Fig. more rhizomes produced had heavier caudices, rosettes, which Flowering grasslands. 2; e.g. 0.32 rosettes (Table than vegetative Correlation between projected population growth rate, 0.82 in the sod-cut area of between sites, but systematically did not differ length and depth and no long runners. shallow rhizomes were found site n only short distribution. The sites are: Bennekomse meent (B; open circles), Konijnendijk (N; grey circles) and distribution. with linear regression. ns = black circles). Within each site correlations are tested (V; Veerslootlanden not significant; flowering ( 116. Chapter 7 Population dynamics of Cirsium dissectum (Table 3). The seedadditionexperiment resultedinverylowseedlingandsmallrosette had bothahigherrate ofclonalpropagation andstrikinglylowerrosette survivalrates rosettes insiteNhad smallerleavesthanrosettesintheothertwo sites,andsiteV the threesitesoverlapped, thereweredistinctdifferences betweenthepopulations: rates aroundone( in themeadowsB,NandVresultedonaverage inprojectionmatriceswithgrowth The data onsurvival,rosettegrowthandclonal propagation fromthepermanentplots Population dynamics were foundinthenumberofrhizomesperunit dryweight. that rosetteproductiondependsstronglyonsize.Nodifferences betweensites was explainedfullybythegrasstreatmentandtotal factor 5.8irrespectiveofsiteorigin.Variation inthenumberofrhizomesproduced presence ofthegrass those fromsiteHbutnodifferences wereseenbetweenthe othersites(Fig.3). The Rosettes of Rhizome formationexperiment (S), Bennekomsemeent(B),Konijnendijk(nandN)Veerslootlanden (V). data wereln-transformedbeforebeingtested.Siteswere:Nijkampse heide(handH),Stelkampsveld test, destination. Foreachparameter different lettersstand forsignificantsitedifferences (Tukey’s post-hoc were randomlyselectedandcarefullyexcavated.Connectedrhizomesfollowedtotheir vegetation height,vegetation cover, thicknesshumuslayerandrosettedensity. Fiverosettesperplot Cirsium dissectum Overview ofthesitesurvey. At eachsitefiverandomlychosenplots of50x50cmwithatleastfive Table 1. Selected parameters Plot hzm et c]3a b3b 3 ab 2 b 17 ab b 10 3 ab 8 b 3 bc 62 ab 11 bc 69 b 4 b 16 bc 67 a 1 a 2 c 84 ab b 3 29 bc b 61 15 17d b 29 15d a 21 14cd ab 14cd 28 ab 11bc 36 b 32 5a rhizome depth[cm] 86b 8ab ab [cm] length rhizome 16 6b 93b No max. a 92b 15 16d [mg] caudex weight dry leaf 90b 12c ab length 16 10bc 92b [cm] 9bc % seedling 39a flowering % 48a 0a C.dissectum thicknessvegetation 0a [cm] height vegetation humusmown biomass [g m cover layer [%] [cm] o olrmvdysysn on ono no no no no yes yes soil removed top α . = 0.05).Inordertoimprovehomogeneityofvariancethevegetation height androsettedensity hzmsprrste04 b01 .8a04 b04 b03 .0b 1.00 a 0.33 ab 0.46 ab 0.42 a 0.28 a 0.11 ab 0.43 rosette per rhizomes C. dissectum C. dissectum C. dissectum rosette density [No. m C. dissectum rosette parameters rosette rosettes (graslands)ortwo(sodcutareas)weredescribed:mean -2 ] λ ranging from0.68to1.30;Fig.4). Although thevariationin Agrostis capillaris from sitesBandVhadsignificantlylowerbiomassthan -2 4a 8a13b8 9a 2 2ab 42 b 120 ab 49 b 89 b 133 a 18 ab 84 ] 4a9 b17ac23b 7 c35c33c 363 c 315 bc 278 bc 263 abc 197 ab 98 a 64 20a8a6a9a6a7a15a 3a15a0a0a0a0a0a h reduced averagebiomassproductionbya n H C. dissectum S B biomass, indicating N V λ of 117. Chapter 7 Population dynamics of Cirsium dissectum in site Cirsium λ ) and especially higher clonal G was on average 0.13%. The two- was on average 0.13%. λ , between sites and years, we decomposed the , between sites and years, we λ in site B and N, but that clonal propagation by one-year by but that clonal propagation in site B and N, by higher growth rates ( λ λ and the projected population structure differed between sites population structure differed and the projected λ with flowering percentage, but the percentage of clonal offspring was of clonal offspring but the percentage with flowering percentage, ). The other two sites showed the reverse pattern of contributions. The other two sites showed the reverse pattern ). C λ according to the LTRE techniques. The model fitted the data well as the well as model fitted the data The techniques. according to the LTRE λ had spread further in nutrient-rich plots and that more dead rosettes were had spread further in nutrient-rich plots Due to such low fecundity, seedlings made up less than 0.1% of the projected than 0.1% of made up less seedlings such low fecundity, Due to V as compared to the average reference matrix was mostly due to low survival of one the average reference matrix was mostly due to low survival to V as compared this was to a large extent compensated by (Fig. 5). However, year old clonal offspring positive contributions to Explanations for increased sexual reproduction under nutrient enrichment Explanations for increased sexual reproduction under nutrient for the increase in Higher biomass turn-over seems to be the most likely explanation The observation that seed production in the three-year allocation experiment. Discussion dissectum found in these plots strongly suggests that the biomass and clonal growth rate were strongly suggests found in these plots from the biomass and number of living rosettes did not differ whereas the total higher, stable population structure in all three populations, even while seedling survival was even while seedling survival structure in all three populations, population stable the rosette survival diagrams confirmed that The elasticity high. kept conservatively to had higher contributions establishment rates (0.00025 and 0.00050 per seed; Chapter 4). Only three seedlings Chapter 4). Only per seed; and 0.00050 rates (0.00025 establishment were plots the permanent summers in which the five were found during in total observed. survival was and no seedling censused old clonal offspring was most important in most productive site V (Fig. 5). The V (Fig. 5). in most productive site was most important old clonal offspring correlation between rate was positively in site V population growth years were compared: when the four that of rosettes but the percentage of flowering plants, the percentage correlated with the years at about 75% over constant was by clonal propagation were newly produced was only a weak was found in site N where there The opposite pattern (Fig. 6). correlation of were almost and interaction effects than site effects, revealed larger year way LTRE values of the site, year and the means of the absolute as large as the year effects: The relatively low and 10.22 respectively. 6.51, 11.18 were intereaction effect reproduction ( variation in Life Table Response Experiments Life Table could explain the variation in in matrix elements In order to analyze what differences rate, the projected population growth the observed and modeled between differences twice as high in good years, compared to bad years. Site B was intermediate to the to bad years. Site compared twice as high in good years, other two. 118. Chapter 7 Population dynamics of Cirsium dissectum density of 1991). The results ofoursitesurveyshowthat rosettebiomassdoesnotdiffer between bothmodes ofreproductionwasalsoinfluencedbylight levels(Reekie nitrogen availabilityalso lowersthecosts ofrhizomeallocation,although thetrade-off growth. Inhisstudy on therhizomatousgrass may altertrade-offs betweensexualreproduction,clonalpropagation, andplant favored. Reekie(1999)agreesthatvariation inthecosts ofmodesreproduction limitation ofseedproduction,sexualreproduction becomeslesscostlyandshouldbe Loehle (1987)hypothesizedthatwhenhigh nutrient availabilityalleviatesthenitrogen variable anddependingongoodyears. flowering rosettesintheprojectedpopulation structure,althoughfloweringwashighly gradient arestrikinglysimilarthough.Site Valsohadthehighestpercentage of are moresandy. The patterns ofthenutrientenrichmenttreatmentandsite productivity, butalsoinotherways.ForexamplesiteVispeaty, whereastheothers caution isneededasonlythreesiteswerecompared, thattendtodiffer notonlyin site (V)hadhigherclonalgrowthrateandlowerrosettesurvival.Ofcoursesome seems tobebalancedbyhigherrosettemortality. for flowering.Highgrowthratesalsoresultinmoreclonaloffspring, buttheirnumber growth ratesprobablyallowagreaterproportionofrosettestoreachthesizerequired fertility onrootdecompositionwasstudied(vanderKrift etal.2001).Inourcasehigher higher. Suchincreasesinmetabolism arealsofoundwhentheeffect ofincreased control groupatharvest. The latterwouldfurthersuggestthatthedecayrateisalso (0=no, 1=yes). The parameters ofthe50x50cmplots were:mownvegetation biomass[gm the dryweight[g]ofcaudex,numberrhizomesmadebyrosette,andits floweringstatus grasslands, and36rosettesintwoareasofwhichthetopsoilwasremoved. The plantparameters were dissectum Pearson correlationcoefficients betweencombinationsofparameters ofrandomlyselected Table 2. o hzms05 *0.31 ** ** 0.50 0.57 ** 0.43 Rosette density [No. m m [g biomass Mown rhizomes No. ** ** 0.27 0.43 Flowering Sod-cut areas Rosette density [No. m m [g biomass Mown rhizomes No. Flowering Grasslands Other explanationsforincreasedsexual reproduction arelessconvincing. The demographicfielddata seemtosupportthistheory:themostproductives C. dissectum rosettes andtheirplotcharacteristics,withintwovegetation types:123 rosettesinfive rosettes [No.m 2 2 03 01 -0.22 -0.16 * 0.20 * -0.37 0.01 ] 0.08 ] -2 -2 ] ] Caudex [mg] Caudex [mg] Caudex 00 01 01 0.00 -0.15 -0.10 -0.03 .501 .9*-0.32 * 0.39 0.16 0.25 -2 ]. Flowering Flowering ( * ) Agropyron repens No. rhizomes No. rhizomes No. Mown biomass Mown biomass however, high ( -2 * ) ] andthe Cirsium 119. Chapter 7 Population dynamics of Cirsium dissectum on its vital rates within and vital on its are almost entirely based on clonal C. dissectum is caused partly by poor seed production but by poor seed is caused partly populations in the Netherlands have been shown populations in populations is an area for further research. C. dissectum Cirsium dissectum C. dissectum C. dissectum C. dissectum The elasticity and the LTRE analyses both point to clonal propagation and propagation analyses both point to clonal The elasticity and the LTRE Differences between sites in allocation to sexual reproduction may also have a to sexual reproduction may between sites in allocation Differences The population dynamics of The population dynamics of to differ significantly in AFLP-markers, although variation within sites was larger than AFLP-markers, although variation within sites in significantly to differ in growth found some significant differences 2000). We between sites (Smulders et al. more populations with a suggestion that different rates between individuals from This may be due to the with a lower growth rate. productive sites may have individuals to soil of the cold greenhouse experiment, which is more similar sandy nutrient-poor, to local of adaptation indicating a degree vegetation, the sites with a low-productive in it does not give an explanation for higher rosette growth rates conditions. However, in the common sites. Neither does the variation in rhizome production certain was not influenced by site of origin, but was fully environment experiment, which of the grass treatment. (through plant biomass) effects explained by direct and indirect for the are unlikely to be the main explanation differences conclude that genetic We and clonal reproductive allocation between populations. of sexual patterns the exact role of genetics in Disentangling research. more detailed between populations requires of seedlings is very rare. Kay and John (1994) as survival and growth propagation seedlings we The three found no seedlings at all in their survey of British populations. This low per census. found in five years made up 0.1% of the observed rosettes number of seedlings in This (Chapter 4). stands rates within vegetation primarily by very low establishment in genetic diversity. low seedling recruitment rate could potentially lead to a reduction of clonal species can be very long-lived and require little On the other hand, genets and to maintain mortality seedling recruitment per year to compensate for genet Powell 1993), and modest & genetic diversity (Eriksson 1989; Cain 1990; Watkinson The (Suzuki et al. 1999). but repeated seedling recruitment can occur in clonal species of low seedling recruitment rates on the genetic diversity and the implications of effect this for the health of genetic basis. Distinct genetic basis. Distinct between meadows, but that relative clonal allocation is highest in the most productive is highest in allocation that relative clonal meadows, but between be but may also production, of rhizome lower costs be due to This may indeed site V. sinks for nutrient are stronger when rhizomes growth rates by higher explained and it could this distinction, to make not suitable are Our data than rosettes. allocation of reproduction, and lower costs higher biomass turn-over be that both explanations, of lowered the costs that nitrogen availability it seems unlikely are additive. However, production of seed not interact with costs as the nutrient treatment did seed production experiment:in the allocation of revealed similar costs the bud removal treatment treatment as in the low nutrient in the nutrient rich treatment sexual reproduction (Chapter 3). 120. Chapter 7 Population dynamics of Cirsium dissectum stasis ( Konijnendijk (N)andVeerslootlanden (V). and CVofthemortality probabilityofeachclassaregiven. The sitesare:Bennekomsemeent(B), coefficients ofthespatiotemporal variation(CV)ofeachmatrixelement.Underthemean history components: fecundity( flowering clonaloffspring ( ( Year-to-year populationtransitionmatriceswithsixclasses:seedlings( Table 3. ieVMa VMa VMa VMa VMa VMa CV Mean CV Mean CV Mean CV Mean CV Mean CV Mean CV Mean CV V Site Mean CV Mean CV Mean CV Mean CV Mean CV Mean CV N Site Mean CV Mean CV Mean CV Mean CV Mean B Site sml ), largevegetative rosettes( S ), andretrogression( flow lrg sml dead cl.f cl.v flow lrg sml dead cl.f cl.v flow lrg sml cl.f cl.v flow lrg sml sdl dead cl.f cl.v sdl sdl sdl .0 .0 000730 0.007 30 30 0.007 0.014 12 29 0.426 0 0.002 30 1.000 0.014 22 37 0.333 0.277 0 32 0.482 1.000 29 0 37 0.002 29 0 0.268 0.150 0.005 39 0.300 60 0.475 0.061 0 29 0.005 0.550 111 0.079 131 0.054 0 0 0.150 0.300 .0 .0 .0 9 0 0.003 1.000 7 0.778 0 9 9 1.000 41 0.006 0.003 0.660 83 39 0.020 0 0.327 9 0 1.000 0.006 15 0.550 84 0.391 0.071 0 96 0.062 1.000 0 27 0 0.382 0.150 29 0.300 0.368 0 0.550 sdl G S ------11 21 cl.f ). Above: eachexistingelementofthematrixisassignedtoonesixlife- .4 20582 .0 60472 .6 9 41 79 0.147 1.969 0.029 45 30 21 0.035 0.173 0.477 100 16 0.365 2.207 132 27 0.077 96 128 0.538 0.145 22 0.124 102 0 0.549 0.124 10 0 11 0.000 46 2.100 54 0.611 0.000 0.033 56 0.150 129 0.382 0.002 55 200 0.038 1.352 200 54 0.012 0.406 147 102 54 0.014 61 0.063 7 54 0.396 0.278 19 76 0.040 0.129 17 2.049 40 25 0.110 0.108 16 0.027 0.498 0.323 90 23 0.138 1.915 81 25 0.019 0.302 200 44 25 0.008 30 0.110 0.307 0.544 116 24 0.018 0.453 R F lrg ), growth( ). Below:foreach sml G G ), floweringrosettes( C C S - - - - 22 62 52 42 32 G ), clonalreproduction( G C C S R lrg - - - - 33 63 53 23 43 Cirsium dissectum flow ), vegetative clonaloffspring ( flow C C F F ------24 14 64 54 C ), survivalclonaloffspring ( sdl population themeansand ), smallvegetative rosettes cl.v C C T T T - - - - 45 35 25 65 55 cl.v cl.f C C F F ------26 16 66 56 ), and T ), 121. Chapter 7 Population dynamics of Cirsium dissectum (Kleijn & C. dissectum under conditions with Veratrum album album Veratrum a poor competitor, explaining its a poor competitor, Molinia caerulea Cirsium dissectum C. dissectum seldom deviate from a negative hyperbolic trajectory from a negative hyperbolic seldom deviate is restricted to a limited abiotic range due to its sensitivity to effects range due to its is restricted to a limited abiotic C. dissectum Drainage and nitrogen deposition are likely to have a further negative impact as negative impact Drainage and nitrogen deposition are likely to have a further Rhizome length is highly variable (from close to the parent plant up to 40 cm in plant up to 40 cm to the parent is highly variable (from close Rhizome length Nature management aimed at conserving populations of this endangered Nature management aimed at conserving populations Cirsium dissectum add that 1998; Lucassen et al. 2003). We of acidification (de Graaf et al. and that within the range of abiotic conditions is also a poor below-ground competitor on the population dynamics of this a profound effect it occurs, nutrient availability has does which in itself increases rosette turn-over, endangered species. High productivity size, but it is expected to decrease genetic not necessarily decrease population in genet This increase increases. of genet mortality diversity as the probability In fact, the increased is not balanced by sexual recruitment from seed. mortality such as biomass of competitive species Implications for the conservation of Implications for the conservation our study) and does not differ significantly between populations. Neither was it populations. Neither was significantly between does not differ our study) and studies (de Kroon which is consistent with other by increased nutrient levels, affected The thick 1999). Alaten 1994; Dong & Evans 1992; Dong & de Kroon 1990; & Knops rhizomes of tussocks of other species or to be hampered by roots through the soil and seem not observation). or soil types (E. Jongejans, personal survival of clonal offspring as the most important life-history components. The LTRE components. life-history the most important as of clonal offspring survival rosette whereas with clonal propagation, off survival trades that rosette suggest results of the percentage Although clonal propagation. with is positively correlated growth role plays no sexual reproduction years, variable between rosettes is highly flowering propagation, a higher rate of clonal But flowering rosettes do have in the LTRE. as e.g. in restricted to flowering plants, although it is not Steinger 2002). Steinger species should focus on restoring and maintaining hydrologically suitable conditions, hydrologically suitable species should focus on restoring and maintaining through mowing and hay and should counteract succession and nutrient enrichment of Negative effects removal. Sod-cutting seems necessary for seedling establishment. sod-cutting (Dorland et al. 2003) do not seem to limit initial ammonium peaks after long run, provided that seed sources are nearby. in the seedling establishment can be expected to be Increased seed production in nutrient enriched grasslands conditions. near such favorable establishment helpful increased nutrients may further decrease seedling recruitment due to a reduction in may further decrease seedling recruitment increased nutrients for establishment. gaps short stature its This inability and this species is unable to accumulate biomass. (Hegde & Ellstrand 1999) make The role of 1987). Tilman 1979; (Grime wet habitats restriction to nutrient-poor, new rosettes, as rhizomes in this clonal species seems to be restricted to spreading as storage organ or as a the rhizomes are relatively short-lived and do not function the initial phase. means of resource sharing between rosettes after 122. Chapter 7 Population dynamics of Cirsium dissectum Eriksson O1986.Mobility and space captureinthestoloniferous plant Efron B1982. The jackknife,thebootstrap andother resamplingplans.-Societyforindustrialand 2003.Soilammoniumaccumulationafter sod Dorland E,BobbinkR,MesselinkJH&Verhoeven JTA Dong M&deKroonH1994.Plasticityinmorphology andbiomassallocationin Dong M& Alaten B1999. Clonalplasticityinresponsetorhizomeseveringandheterogeneousresource de KroonH,vanGroenendaelJM&EhrlénJ2000. Elasticities:areviewofmethodsandmodel de KroonH,Plaisier A, van GroenendaelJ&CaswellH1986.Elasticity:therelativecontributionof de KroonH&Knops J1990.Habitat explorationthroughmorphologicalplasticity intwochalkgrassland de KroonH&BobbinkR1997.Clonalplantdominanceunderelevatednitrogendeposition,withspecial de GraafMCC,BobbinkR,Verbeek PJM&RoelofsJGM1997. Aluminium toxicityandtolerancein de GraafMCC,BobbinkR,RoelofsJGM&Verbeek PJM1998.Differential effects ofammoniumand Caswell H2001.Matrixpopulationmodels.Construction,analysis,andinterpretation. -Sinauer 1990.Modelsofclonalgrowthin Cain ML Buck-Sorlin G&Weeda EJ2000.Oecologieenplantensociologischepositievan Buck-Sorlin G1993. Ausbreitung undRückgangderEnglischenKratzdistel- Berg H2002.Populationdynamicsin Berendse F& Aerts R1984.Competitionbetween (ed),Demography Abrahamson WG1980.Demographyandvegetative reproduction.-In: Solbrig OT References project 805-33-452). The NetherlandsOrganizationforScientificResearchfundedthisresearch(NWO Dutch State Forestryfortheirkindpermission toperformthisstudyintheirmeadows. Gleichman fortheirpracticalassistance withtheallocationexperiment,andto Klees, JanvanWalsem, JaspervanRuijven,JuulLimpens,LouisdeNijsandMaurits We arealsogratefultoNienkeHartemink,FransMöller, HenkvanRoekel,Herman Ben Starink performedthegreenhouseexperimentandassistedwithsitesurvey. Acknowledgements 82-87. applied mathematics,Philadelphia, Pennsylvania. cutting hamperstherestorationofdegradedwetheathlands. -J Appl Ecol40:804-814. grass speciesformingstolonsandrhizomes.-Oikos 70:99-106. Plant Ecol141:53-58. supply intherhizomatousgrass limitations. -Ecology 81:607-618. demographic parameters topopulation growthrate.-Ecology67:1427-1431. perennials. -Oikos59:39-49. J (eds), The ecologyandevolutionofclonal plants. BackhuysPublishers,Leiden. reference to three heathlandspecies.-Water Air SoilPollut98:229-239. nitrate onthreeheathlandspecies.-PlantEcol135:185-196. Associates, Inc.Publishers,Sunderland,Massachusetts. Hill inOostfriesland. -Stratiotes 21. clonal, cleistogamousforestherb.-Ecography25:233-243. affected bytheavailabilityofnutrients. - Acta OecolPlant5:3-14. and evolutioninplantpopulations.BlackwellScientificPublishers,Oxford. - inNordwestdeutschland. - Tuexenia 13:183-191. Brachypodium pinnatum Oxalis acetosella Psammochloa villosa Solidago altissima in chalkgrassland.-In:deKroonH&vanGroenendael Erica tetralix : thesignificanceofsexualreproductionina . -JEcol78:27-46. in anInnerMongoliandune,China.- and L Molinia caerulea Potentilla anserina Cirsium dissectum Cirsium dissectum Cynodon dactylon (L) Moenchas . -Oikos46: (L.) Hill (L.) , a 123. Chapter 7 Population dynamics of Cirsium dissectum Cirsium genets in a genets . - Can J Bot 69: 2678- Festuca rubra . - J Ecol 81: 707-717. Agropyron repens , an unpalatable, long-lived, clonal plant species. - J long-lived, , an unpalatable, . - Oecologia 89:265-276. . - Oecologia Ranunculus repens Veratrum album Veratrum Hydrocotyle bonariensis Hydrocotyle (L.) Hill. - Plant Ecol 165: 45-52. mountain grassland: its relevance to genet establishment and dynamics. - J Ecol 87: 942-954. relevance to genet establishment grassland: its mountain - Ecol Monogr 57: 189-214. gradients. 65-66. 191-200. control over rhizodeposition and root decomposition. - Plant and Soil 228: Apopulations: computer simulation of 2683. Academic Publishers, London. Kluwer in plants. Mutikainen PK (eds), Life history evolution grassland and heathland communities. - Biol Conserv 92: changes in plant species diversity in 151-161. Gorteria 29: 22-28. - Conserv Biol 5: 18-32. a review. meadow species. - Plant Biol 2: B 2000. Genetic diversity and the reintroduction of & Vosman 447-454. and connectivity. fragmentation rich semi-natural grasslands. - In: Soons MB (ed), Habitat Ph.D. thesis, Utrecht process in plants. and temporal characteristics of the colonization Spatial University. dissectum species of California and the British Isles. - Int J Plant Sci 160: 1083-1091. and the British Isles. - Int J Plant species of California S & Caswell H Tuljapurkar - In: prospective and retrospective analyses. population growth: systems. Chapman & models in marine, terrestrial, and freshwater (eds), Structured-population York. Hall, New 73: 1082-1093. - Ecology analyses, and seed bank effects. matrices, elasticity Council for - Countryside lowland grassland and related habitats. of Welsh vascular plants UK Bangor, Science Report No. 93, Wales population structure of Ecol 90: 360-370. 208. of of low pH and high ammonium levels responsible for the decline Interactive effects morphology in morphology Tilman D 1987. Secondary succession and the pattern of plant dominance along experimental nitrogen of plant dominance along experimental pattern D 1987. Secondary succession and the Tilman D, May RM, Lehman CLTilman & Nowak MA destruction and the extinction debt. - Nature 1994. Habitat Kuikman PJ, Möller F & Berendse F 2001. Plant species and nutritional-mediated TAJ, van der Krift of clonal diversity in plant AR & Powell JC 1993. Seedling recruitment and the maintenance Watkinson Reekie EG 1999. Resource allocation, trade-off, and reproductive effort in plants. - In: Vuorisalo TO & TO - In: Vuorisalo in plants. and reproductive effort trade-off, Reekie EG 1999. Resource allocation, acidity and nutrient supply ratio as possible factors determining Roem WJ & Berendse F 2000. Soil - in het Landelijk Meetnet Flora-Aandachtsoorten. AJG & Groen CLG 2003. Veranderingen Rossenaar CR 1991. Biological consequences of ecosystem fragmentation: Saunders DA, Hobbs RJ & Margules A der Werf Korevaar H, van AG, Antonisse-De Jong RHEM, der Schoot J, Geerts Smulders MJM, Van and connectivity of species- Soons MB, Messelink JH, Jongejans E & Heil GW 2003. Fragmentation T Krahulec F & Hara T, Suzuki JI, Herben of pattern 1999. Size and spatial Reekie EG 1991. Cost of seed versus rhizome production in Reekie EG 1991. Cost of seed versus Grime JPChichester. Wiley & Sons, processes. - John strategies and vegetation 1979. Plant plant common flowering between rare and differences 1999. Life history & Ellstrand NC Hegde SG to stages of life-history "importance" The relative DW & Caswell H 1997. Horvitz C, Schemske MAKalisz S & McPeek projection age-structured annual - resampled 1992. Demography of an and declining ecology of some scarce R 1994. Population genetics and demographic Kay QON & John T Kleijn D & Steinger and genetic spatial of grazing and hay cutting on the effects 2002. Contrasting Klimeš L 9: 43-52. J 2000. Plant rarity and the type of clonal growth. - Z Ökol Naturschutz & Klimesová - Oikos 49: 199- a benefit-cost model. in clonal plants: effort Loehle C 1987. Partitioning of reproductive & Roelofs JGM 2003. PJM, Lamers LPM der Ven van AJP, Bobbink R, Smolders Lucassen ECHET, Eriksson O 1989. Seedling dynamics and life histories in clonal plants. - Oikos 55: 231-238. plants. histories in clonal dynamics and life O 1989. Seedling Eriksson Evans JP functional on ramet and clonal integration resource availabiltiy of local effect The 1992. 8 A hierarchical population matrix analysis of the effects of nutrient enrichment: a synthesizing discussion

Eelke Jongejans, Linda D Jorritsma-Wienk & Hans de Kroon

Summary

Nutrient enrichment decreases species richness in grasslands through accelerated secondary succession. A previous experiment on four grassland herbs showed that species with different life history strategies are differently affected in their survival rate, plant size and sexual reproductive allocation. In this paper we investigate the contrasting effects of fertilization: do negative effects on survival and recruitment outweigh positive effects on plant size and seed production? This was tested with a hierarchical population matrix model in which matrix elements were functions of life history components, which in turn were functions of the plant traits size, flowering threshold size and relative sexual reproductive allocation. The model results differed for species with short-lived and long-lived rosettes. For the former species (Hypochaeris radicata and Cirsium dissectum) high mortality rates overruled other effects and caused strong population declines, whereas in the long-lived species, in the absence of increased mortality, decreased recruitment had little affect on population growth rate (Centaurea jacea) or was compensated by increased flowering and seed production (Succisa pratensis). But also the long-lived species will face strong population declines when eutrophication lasts for more than a few years and even these rosettes will experience high mortality rates.

Keywords: biomass allocation, Centaurea jacea, Cirsium dissectum, elasticity, Hypochaeris radicata, life history components, LTRE, nutrient enrichment, plant trait, sensitivity, sexual reproductive allocation, Succisa pratensis, succession

Introduction

Herbaceous perennials in grasslands are outcompeted by grasses and more competitive herbs when the vegetation biomass increases due to natural succession. Human induced nutrient enrichment accelerates this process significantly. Eutrophication has already caused declines in species richness in many grasslands 126. Chapter 8 A hierarchical population matrix model all otherparts inconcert(Franco andSilvertown2004). dynamics itisnotpossible toevaluateoneaspectofthelifecycle without considering rate. However, wheninvestigatingthe effects ofnutrientenrichment onthepopulation reproduction reported inchapter3canbeexpectedtoincrease populationgrowth seedling establishment cancausepopulationdeclines,theenhancedsexual ground competitiondependingonthespecies studied.Whereasreducedsurvivaland decreased survivalandreducedseedling recruitmentthroughincreasedabove- affected byincreasing vegetation biomass. Thus nutrientenrichmentcanleadto meadows. The recruitment ofthetwomoresexuallyreproducingspecieswasless reduced tozeroathigher-productivitymeadows ascompared withnutrient-poor establishment ratesofthetwomainlyclonally reproducingspecieswerefurther crop ofgrasslandsmatterforrecruitment: Soonsetal.(2003)foundthatthe more seedlingsinsod-cutareasthangrasslands. Also differences inthestanding (Chapter 4).Especiallythetwomainlyclonallyreproducingherbspeciesestablished establishment, whichisanimportant aspectofthesexualreproductionpathway recruitment ratesassomespeciesdependonbaresoilforsuccessfullseedling with short-livedrosettes(Chapter3).Vegetation biomassalsostronglyinfluencesthe mechanisms areenhancedbynutrientenrichment. rather thandecreasedaswouldbeexpected. Thus both persistence andescape biomass ofsurvivingplants wasincreased,whiletheallocationtostorageincreased to seedproduction.Notwithstanding thedemonstratedcosts ofseedproduction,total larger thantheplants intheunfertilizedplots andthattheyincreasedtheirallocation main results ofthefertilizationtreatmentwerethatsurviving,fertilizedplants were succession byfertilizingplots withtheseherbsandcompeting grasstussocks. The (Chapter 3).Inthatexperimentwemimickednutrientenrichmentduringsecondary experiment revealedthatplants respondtonutrientenrichmentindifferent ways propagation whenfloweringwasinhibited.Furthermore, athree-yeargarden species), buttheywerealsoabletoswitchfromsexualreproductionclonal species) andforlostrosettebuds(clonalspeciesfasterthanmorerarelyramifying able tocompensateforlostflowerbuds(short-livedspeciesfasterthanlonger-lived in theirlife-historydecisions(Chapters2and3).Notonlyweretheseperennialherbs grassland herbsthatarestudiedinthispaper havebeenshowntoberatherflexible production inordertoincreasethecapacity tocolonizemorefavourablesites. The four can notadapt.Contrastingly, plants mayalsobeexpectedtoincreasetheirseed limitation isreducedbyfertilization,willformmorestable populationsthanplants that poor grasslandsthatareabletoenhancetheircompetitiveabilitywhennutrient how individualplants respondtothechanginggrowingconditions.Plants ofnutrient- how populationgrowthratesarereduced. components ofthelifecycletheseherbsareaffected bynutrientenrichmentand however maygainineffectiveness byabetterunderstanding ofhowexactly endangered andtargets ofnatureconservationandrestorationefforts. These efforts (Neitzke 2001;Stevens etal.2004).Manyofthedecliningspeciesarenow Plant mortality increasedduetonutrientenrichmentaswellinthetwospecies The effects ofnutrientenrichmentonaplantpopulationdependsstrongly 127. Chapter 8 A hierarchical population matrix model C. population S. pratensis some populations had lower Succisa pratensis have lower establishment rates and inbred seedlings rates and have lower establishment , which has short-lived rosettes, showed a contrasting response to , which has short-lived rosettes, Succisa pratensis Succisa pratensis population became more dynamic than the species-average population population became more dynamic A dynamics in the field revealed that the closer examination of the population To investigate the impact of succession, populations of the same species in of the same populations of succession, the impact investigate To dynamics: had a higher rosette turn-over and the temporal variation in population it between populations over a growth rate was high. Whether these field patterns of the plant individuals as gradient of nutrient richness are the result of the responses with nutrient enrichment remains untested. observed in our experiments of nutrient enrichment on how positive direct effects we aim to investigate In this paper increased competition with that work indirectly through and negative effects plant traits that aim we construct a To population growth rate. together affect neighboring plants of life history importance hierarchical matrix model that allows for the evaluation of the 2000). Tienderen on the population growth rate (cf. van and plant traits components and 7) from a nutrient-poor site (Chapters 5 therefore combine demographic data We of the nutrient richness experiment (Chapter 3). In this modeling and the results by nutrient is affected exercise we firstly test how the population growth rate ‘senile’enrichment and whether the population changes into a population (lower (higher proportion of proportion of seedlings), or into a more dynamic population of the Secondly we study how the relative importance seedlings and clonal offspring). change when nutrient enrichment is simulated traits and plant life history components dissectum lower growth rates, which reduces the population growth rate and population size growth rate and population which reduces the population lower growth rates, 6). Both in an extinction vortex (Chapter population becomes trapped even more: the of capacity also reduce the colonization and site productivity inbreeding depression herbs (Soons &the same grassland inbreeding depression and low Heil 2002) and parameters and reduce several fitness been shown to interact quality have habitat et al. 2003). (Vergeer and temporal variation to the same spatial differently species studied can respond very in growing conditions. In the long-lived was or because seed set the rosettes grew slower, population growth rates because partly increases in other life history components prohibited by early mowing. Whereas were growth, all components on the population effects compensated these negative The rhizomatous 5). biomass (Chapter vegetation lower in the site with the highest Cirsium dissectum the same productive site: in two other rate was higher than the clonal propagation These species therefore was also higher (Chapter 7). sites, but rosette mortality responses to nutrient enrichment: the displayed two different almost et al. 1996: consisting of adult plants became more senile (sensu Oostermeijer and with a low population growth rate), while the without seedling recruitment different successional stages (Oostermeijer et al. 1996) or of species characteristic for of species characteristic et al. 1996) or (Oostermeijer stages successional different The have been compared. et al. 1993) (Silvertown stages successional different of importance that the relative studies is from these that emerges pattern general strong succession. Furthermore, declines in time with secondary sexual reproduction in the long- of inbreeding. Inbred seeds size increase the risk reductions in population lived perennial 128. Chapter 8 A hierarchical population matrix model ramifying clonal Kroon etal.1987)).Sexualreproductionisnegativelycorrelatedwithclonality:the relatively lowadultsurvival(althoughhigherthaninmoreproductivehabitats (de Hypochaeris radicata flowering rosette. 5 and7). The lastspecieshoweverhadthehighestnumber ofclonaloffspring per rosettes washighestin strategies ingrasslandherbs. The year-to-yearsurvivaloflarge,non-flowering clonality andsexualreproductionrates,inordertocoverarangeofpossiblelifehistory The fourstudyspecieswereselectedfortheircontrastinglifetimeexpectancy, Study species Materials andmethods nutrient enrichmentthanlonger-livedspecies. we expectshort-livedherbspeciestosuffer morefromthenegativeindirecteffects of seedling recruitmenttobecomelessimportant forlocalpopulationdynamics. Thirdly, to becomemoreimportant forthepopulationgrowthrateinfertilizedpopulations,and in naturalpopulations.We expectplantsizeandtraits ofsexualreproductiveallocation traits (van Tienderen 2000). The lifehistorycomponents were seed production, life historycomponents, andtheselifehistorycomponents werefunctions ofplant formulated ahierarchical pathway (Fig.1a)inwhichmatrixelements were functionsof rosettes caneitherflowerorstay vegetative. new siderosettescanbecomeadults after one yearordie,andadults andside sexual reproduction,flowering,survivaland clonal propagation (Fig.1b):seedlingsor paper the5x5stage classificationisnotbased on plantsize,butentirelybased these consistedofastage classificationwhichwaspartly basedonplantsize.Inthis of plantsize,wehadtoreconstructthereported 6x6matrices(Chapters5and7)as these herbspecies.Becausetheaimofthis study wastoexplicitlyinvestigatetherole We usedstage-based matrices(Lefkovitch1965)tomodelthepopulationdynamicsof Population matrixconstruction from 1999to2003(Chapters5and7). the samepermanentplots intheDutchnaturereserveKonijnendijk(52°02'N,6°26'E) transitions foreachspecies. These demographicdata werecollectedforallspeciesin In ordertobeable investigate theimportance oflower-levelparameters we For thismodelstudyweusedthecombineddata offouryear-to-yearpopulation C. dissectum S. pratensis Centaurea jacea and was themostshort-livedperennialwithlittleramificationand and Succisa pratensis C. jacea H. radicata had muchlowerfecunditythantheonlyrarely was intermediateinbothsurvivalandclonality. . and lowestin Cirsium dissectum (Chapters 129. Chapter 8 A hierarchical population matrix model ). nd C age ) ) ) ) ) ) ))) CCC CCC FFF ( ( ( ( ( ( ...... (. (. ( s, matrix ))) ))) rrr Figure 1. ppp SSS SSS opopop opopop s of the 2 eee rrr rrr agation ( ( ( ( ( ( ( lll lll ppp ppp r r r lll aaa aaa The five st aaa vvv vvv iii iii alalal alalal uuu vvv vvv xxx rrr rrr s. ononon ononon lll lll SeSeSe SuSuSu SuSuSu CCC CCC ))) ))) ww w 777 888 ), the element 777 888 101010 oo o F ll l f ff www www -w-w-w -w-w-w www . .. 333 555 999 ee e (1(1(1 (1(1(1 d dd 333 555 www www www sisi si )w)w)w )w)w)w , 666 888 , , 666 888 egeg eg vv v www www -w-w-w -w-w-w . .. --- 222 444 e ee (1(1(1 (1(1(1 dd d 222 444 www www , age classes): ttt si sisi side.flow side.flow side.veg eee , , , )w)w)w )w)w)w , mmm s are assumed to be independent. ponent s to consecutively life history com iii and 777 888 ttt and and

777 888 101010 ttt w ww www www -w-w-w -w-w-w www fspring, o oo 333 555 flow veg e ae ae a l ll 999 (1(1(1 (1(1(1 f ff side.flow side.veg 333 555 www www w ww agagag side.flow fspring. ttt , , side.veg sss and and )w)w)w )w)w)w and The plant trait 666 888 888 666 veg flow www www -w-w-w -w-w-w flow ) chances and the bottom two rows represent clonal prop side.flow 444 222 side.veg (1(1(1 (1(1(1 S veg vegveg 222 444 www www s are: and www www and , agation by agation by sdl veg flow l ll , flowering clonal of 111 , non-flowering clonal of ------www sd sdsd ablishment rate per seed. est = new = new sdl egegeg = flowering rosettes older than one year s of the first row of the matrix represent sexual reproduction ( = non-flowering rosettes older than one year lowlowlow l ll = vvv = new seedlings, www = clonal prop = clonal prop = flowering probability of surviving = flowering probability of surviving = flowering probability of new = seed production per = survival of = survival of = survival of ... .f.f.f ooo eee 9 10 4 5 6 7 8 1 2 3 eee sd sdsd vegvegveg flflfl ddd w w w w w w w w side.veg side.flow w sdl veg flow w s and population growth rate. ddd row are rosette survival ( iii

sisisi sss

rd ))

)) 1 1 1 + + + t t t i i i t t t t t t a a a e e e g g g a a a t t t s s s e e e m m m (b(b (a(a The element and 3 The 10 life history component element ory component (b) Construction of a 5x5 population transition matrix with 10 life hist classes are (the x in the upper diagram aggregates all st ns of plant trait (a) Path diagram of the contributio 130. Chapter 8 A hierarchical population matrix model made betweenthesexualreproductionofseedlings( present foraccuratequantificationoftheirfate.Forthesamereasonnodistinctionwas non-flowering rosettes,andbecausenotinallspeciesenoughsiderosetteswere This wasbecausethemaindifferences infatewereobservedbetweenfloweringand reduction ofplantsizeinthefloweringprobability functionandwith in which survival andfloweringweusedbinarylogisticregressions(Childsetal.2004): indicate naturalpopulationparameters. Fortheboolean lifehistorycomponents rosette size(thenumberofleavestimesthemaximumleaflength). The vegetative plantsize. The thresholdsizeforflowering, traits shape therelationshipbetweenamountofsexualreproductionand plant size, value forthenaturalpopulationdynamics. The lastplanttrait,seedproduction perunit was sometimesaffected byexperimental nutrientenrichment,wedefined logistic regressionanalysisofthedemographic data. Butsincethefloweringthreshold establishment, demographic fielddata byperformingregressionmodelswithplantsize, as constants. In experiment (Chapters5and7)werenotconstructedasfunctionsofplantsizebut the fateoffloweringnewsiderosettes( ( of different classes(Fig.1b).However, thefateofnon-floweringnewsiderosettes seedling establishment, andthesurvival,floweringprobabilityclonalpropagation components (Fig.1b) the sameyeartheywereproduced.Intotal wedefined10different lifehistory and wasassumedthatallnewsiderosetteshadthesameprobabilityoffloweringin clonal offspring was toosmalltoallowforregressionanalyses. propagation byfloweringrosettesandtheconsecutiveprobabilityofnew forms onlyoneflowerheadperfloweringrosetteandbecausethedata setonclonal in which regressions oftheform: explaining factor. Forseedproductionandclonalpropagation weusedlinear relationship. This traitisequalbydefinition to seed production. side.veg The nextstepwastoquantifythese10lifehistorycomponents, Besides thetraitplantsize, w wazb a kk kk k NN NN N ) wasconsideredtobeidenticalthatofnon-floweringadults ( a k = =•+ k z and 3 e p ) 1exp( and , determinedtheslope ofthesexualreproduction-vegetative plantsize +•+ x() exp( w b .dsetmw C. dissectum 10 k b 1 azb are constants andinwhichvegetative plantsizewasestimated by k , andseedlingsurvival, kk NN azb are theconstants thatshapethelogisticregression.Seedling kk NN •+ 1 1 5 , w z 8 1 , twoothertraits weredefinedaswell. These two and side.flow w 9 were alsoconstants becausethisspecies w 1 , werederivedfromaseedaddition a 9 ) andfloweringadultrosettes( parameter inthelinearfunction for sdl z ) byeither 2 , wasnotestimatedinthe flow z 2 = 0asdefault or veg w k side.flow , withthe N ), aswas z -indices 2 z flow 1 as a , as (2) (1) ). , 131. Chapter 8 A hierarchical population matrix model , , Cj Hr ( ( 1 w 5 5 5 2. 2. 2. 10 w Figure 2. 2,3,4,5 w -values were f 2 2 2 Centaurea jacea Centaurea 5 5 5 1. 1. 1. in the natural population i Hypochaeris radicata , on the population growth f i f X X X 1 1 1 Sp Sp Sp Sp Hr Cj Cj , circles) and Sp 5 5 5 ( Cd Cd 0. 0. 0. Sp Sp (f) (e) (d) : rel. change in adult survival, : rel. change in establishment rate, : relative change in seedling survival, Hr Hr 5 Cj 6 4 Cd -values as found in experiments on nutrient -values as found in experiments f f f 0 0 0 f 8 6 4 2 8 6 4 2 8 6 4 2 8 6 4 2 8 6 4 2 8 6 4 2 0 2 1 0 2 1 0 2 1 0. 0. 0. 0. 1. 1. 1. 1. 0. 0. 0. 0. 1. 1. 1. 1. 0. 0. 0. 0. 1. 1. 1. 1. 5 5 5 2. 2. 2. z 2 Succisa pratensis Sp Sp were varied from 0 to 2.5 while the other five i f 3 z 1 z Cj 2 2 2 Perturbation analysis Perturbation Hr , squares), Hr 5 5 5 1. 1. 1. Cd ( Cj Sp Sp X X 1 1 1 Cd Cd 5 5 5 0. 0. 0. Cj Cirsium dissectum (c) (b) (a) : rel. change in rel. seed prod., : rel. change in flowering threshold, : relative change in plant size, Sp Sp Cd 3 2 1 Hr f f f 0 0 0 X

6 4 2 6 4 2 8 2 8 2 8 6 4 8 6 4 8 6 4 2 8 6 4 2 0 1 0 2 1 2 0 1 2

0. 0. 0. 1. 1. 1. 0. 0. 1. 1. 0. 0. 0. 1. 1. 1. 0. 0. 0. 0. 1. 1. 1. 1.

Population growth rate, rate, growth Population Population growth rate, rate, growth Population Population growth rate, rate, growth Population λ λ λ rate. In separate simulations one of the six rate. In separate kept constant at the default level of the natural populations (the values of kept constant Perturbation analysis on the impact of the six nutrient enrichment factors, Perturbation analysis on the impact dynamics are indicated on the x-axis with an ‘x’). The four species are: dynamics are indicated on the x-axis with an ‘x’). triangles), enrichment (see Table 2 for the exact values). Table enrichment (see diamonds). The arrows indicate the species-specific diamonds). 132. Chapter 8 A hierarchical population matrix model affected intercept intheregressionandmeanvegetative biomass.Sincenutrientaddition model. dynamics (eq.1),andthe in whichthe field data, weremodifiedbyincludingthefactor components inwhichplantsizewasasignificantparameter intheregressionson treatment compared tothenofertilizationtreatment. The functionsofthelifehistory estimated bytherelativeincreaseinvegetative biomassinthenutrientenrichment determined intheplants thatsurviveduntiltheendofthree-yearexperimentand the total vegetative biomass(roots, leavesandstems). The effect onplantsizewas Only thefactor z not taken intoaccountinequation4. flowering probabilityfunctions w factors, from theexperimentandthenimplementedinmatrixmodels(Table 1). The six situation, relativechangesinplanttraits andlifecyclecomponents werecalculated garden (Chapter3).Inordertotranslatetheresults ofthisexperimenttothefield four herbspecieswhiletheygrewamongstdominatinggrassesinanexperimental the effect ofnutrientenrichmentonthesurvival,sizeandallocationpatterns ofthese directly. These effects weredetermined inapreviousstudywhichweinvestigated considered: threeonunderlyingplanttraits andthreeonlifehistorycomponents Six different effects ofnutrientenrichmentonthepopulationdynamicswere Effects ofnutrientenrichment production, species (butnotin sexual reproductivebiomass onvegetative biomasschangedsignificantly in three of vegetative plantbiomass,inthefertilizedanduntreatedgroups separately. The value determined bytheinterceptoflinearregressionssexualreproductivebiomasson 1 k : f ( 2 f 2 was thenestimatedbytheincreaseinarelativemeasureof was zeroinallothercases): f f f w wazfb 2 1 3 f z i kk kk on thethresholdsizeforflowering( k N N FNN on plantsize( , bywhichnutrientenrichmentaffected traits andcomponents, were: F on seedproductionperunitplantsize( 2 == ••+=• only in w e p ep ) ) 1exp( ) ( ) 1exp( ( N 9 +•−++•−•+ x(())ep ) ) ( exp( ) ) ( exp( , wasmodifiedasfollows: -parameters arethevaluesinregressionsnaturalpopulation f 2 () is modeledhere,whiletheotherfactorsof nutrient enrichmentare C. jacea z bazzfb azzb 11 k k kkk N N N NNN N NNF z bazzfb azzb k k kkk N N N NNN N NNF − •−•+ •−+ C. dissectum z 1 2112 12 ). Intheexperimentvegetative plantsizewasestimatedby 2112 12 , F z -parameters arethenewvaluesinfertilizedmatrix 2 w was notmodeledintheotherspecies.For 6 and ). Inthesethreespecies thefunctionofseed w 8 were modifiedbysetting z 2 ). The thresholdsizeforfloweringwas f 1 ; forinstance inthelinearfunctionof z 3 ). The slopeoftheregression z z 2 2 to afraction : theratioof C. jacea f 2 the (3) (4) of 133. Chapter 8 A hierarchical population matrix model = F s of s (a/d), Figure 3. s (a/d; = flowering 2 z s, and that the s (e/h; see Fig. ation of the bars indicates , = plant size, 1 s of matrix element z s (e/h) and selection gradient s (i/l; The segment , to group The white bars represent the natural population arameter: elasticity values of matrix element agation), and to life history component , of the plant trait ain p s in single matrix element s. Please note that the elasticity values of the lower-level s of a cert = clonal prop art C ). k w = survival, , do not sum to one like the elasticity values of the matrix element r S z = sexual reproductive allocation). 3 and z k w s in single life history component t two columns: elasticity of the population growth rate, arameters, 1 for the definitions of the sexual reproduction, Right column: phenotypic selection gradient, threshold size, elasticity values of life history component trait p d by the trait mean. selection gradient of a trait is equal to the elasticity value of trait divide dynamics and the grey bars the fertilized population dynamics. the summation of the constituent p Lef 134. Chapter 8 A hierarchical population matrix model elasticity of (Morris andDoak2004). The sensitivityand function ofthemeanparameter on while theotherparameters arenotchanged.Elasticitieshowevermeasuretheeffect measures how and relativesensitivityorelasticity, the endproduct,populationgrowthrate, To investigatetheimportance ofallparts ofthehierarchicalpath diagram(Fig.1a)for Elasticity analysis estimate oftheeffect ofnutrientenrichment. 2) compared totheestablishment rateinlowproductivesites(habitat class1)asan Here weusetheratioofestablishment rateinhighproductivesites(habitat class establishment wasinvestigatedinaseedadditionexperiment(Soonsetal.2003). was modeledbymultiplyingtheaveragevalueofalifehistorycomponentwith because adultsurvivalunderliesthoselifehistorycomponents. The mortality factor rosette survival,clonalpropagation bytheserosetteswasaffected bythesamefactor extra nutrients onadultsurvivalandclonalpropagation. We assumedthatbesides of survivingplants intheunfertilizedgroupwasusedasanestimateofeffect of plants after threeyearsinthenutrientenrichmenttreatmentcompared tothenumber Again weonlyenteredthefactor production function the seedlingsurvivalratewasaffected inthesamewayasadultsurvivalrate. one-year oldseedlingswasnotinvestigatedseparately. Therefore weassumedthat λ when aparameter ischangedbyasmallfactor. Therefore elasticitiesareadirect f f f wafzb wwf e s 6 5 4 a a 99319 on seedlingsurvival( kk NNN FN ij on seedlingestablishment ( ij FN on adultsurvival( = = =••+ =• λ a () ∂ ∂λ λ∂ a ij to matrixelement ij λ ∂λ a ij changes whenacertain parameter ischangedbyasmallamount 5 . w 2 , w w 1 a ). The effect ofnutrientenrichmentonthesurvival 3 ij f , 2 is (deKroonetal.1986;Caswell2001): e w in thisequationtoillustrateits effect ontheseed , of w 4 10 and λ ). The effect ofhighproductivityon seedling to everypart ofthepath diagram.Sensitivity λ , wecalculatedtheabsolutesensitivity, w 5 ). The ratioofthenumbersurviving f 5 : (6) (5) (8) (7) s , 135. Chapter 8 A hierarchical population matrix model was TRE- Figure 4. ference in to variation in plant ared to the natural , is given, which can = sexual reproduction, , between the fertilized The dif F fect, by the magnitude of the trait s (a/d; s (e/h; see Fig. 1 for their meaning) using ation of the bars indicates the summation of the fect on the variation in TREs that decomposed variation in ference in population growth rate, The segment . of the fertilized population dynamics comp (see the discussion section for an explanation) we divided the L arately -values of the L fect. In each diagram the overall fertilization ef nd As the trait mean we used the mean of the trait values in the natural a agation) or life history component TRE ef = clonal prop Because the C s by the trait mean to rescale the ef s of each L ared to the deviation in , deviated much from r z TRE for each diagram sep s, t two columns: Decomposition of the dif = survival, fect of trait ef values (i,j,k,l; cf. Fig. 3i,j,k,l). rom 0 to - 0.0015. fertilized model. Please note that the scale of the y-axis in (k) ranges f Lef e matrix. matrix and the natural population matrix that was used as the referenc t decomposed into contributions by the variation in either matrix elemen S an L component be comp population dynamics. Right column: trait 136. Chapter 8 A hierarchical population matrix model Construction oflifehistorycomponents, Table 1. (maximum leaflengthtimesthenumberofleaves)asexplainingfactor. Survival( the regressionparameters andtheaverage ( relative impact offertilization, the parameters theychange.Planttraits areonlyconsideredtobeaffected whenp and seedproduction( w 6,7,8 w w w w w w w w w w w w w w w w w w w w w w w w w w w w w w w w w w w w w w w w w 10 9 8 7 6 5 4 3 2 1 10 9 8 7 6 5 4 3 2 1 10 9 8 7 6 5 4 3 2 1 10 9 8 7 6 5 4 3 2 1 k ) probabilitieswerefittedwithabinarylogisticregressionmodel(eq.2).Clonalpropagation ( ier284. .0 .8 .2 .9 .0 0.091 0.073 0.000 0.197 0.921 0.000 0.001 0.085 0.697 0.082 0.942 0.000 0.929 0.033 44.9 0.000 0.000 0.003 0.105 208 0.000 0.345 4.084 2.267 37.1 constant 74 -0.029 44.5 linear 44.9 0.008 195 35.4 208 logistic bin. 35.2 0.313 696 750 logistic bin. 0.040 0.028 0.030 logistic bin. 0.180 0.006 44.9 linear 0.774 -1.789 0.086 208 35.2 linear 0.005 750 0.020 logistic bin. 46.6 logistic bin. 0.001 89 constant 48.8 constant 17 51.2 46.6 linear 63 89 32.5 logistic bin. 31.1 339 465 logistic bin. logistic bin. 46.6 linear 89 31.1 linear 465 logistic bin. logistic bin. constant Regression ier163. .0 .6 .7 .0 .2 0.956 0.186 0.024 0.000 0.072 0.001 0.005 0.864 0.575 0.158 0.003 0.730 31.1 0.001 0.009 136 0.015 0.147 0.641 39.3 constant 195 linear 31.1 0.009 42.3 136 logistic bin. 41.3 255 350 constant 0.328 logistic bin. 0.012 linear 0.033 0.612 41.3 linear 0.411 350 constant logistic bin. -0.003 constant constant constant 28.9 33.9 constant 22 514 constant logistic bin. constant 32.8 linear 825 constant logistic bin. constant w 9 mean n ) werefitwithalinearregressionmodel(eq.1). The valuesandsignificanceof Rosette size f i , onthelifehistorycomponents andtheplanttraits isindicatedbehind · · · · · · · · · · · · · · · f f f f f f f f f f f f f f f 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 ·(1- Hypochaeris radicata Hypochaeris f Cirsium dissectum 2 Succisa pratensis Centaurea jacea Centaurea ) w w k k 005-.8 .0 .8 .8 0.334 0.081 0.689 0.002 -0.087 -0.015 0.186 .2 8700050050680.065 0.608 0.025 0.158 0.035 -8.730 0.034 0.229 0.000 1.490 0.000 0.058 0.147 0.124 -3.412 0.753 0.023 0.733 -0.072 0.218 0.032 0.053 0.022 0.001 0.009 0.002 0.012 -1.599 0.531 0.058 0.015 .2 2040000000120.306 0.112 0.000 0.000 -2.064 0.029 0.056 0.081 0.000 0.000 -4.369 0.625 0.045 0.015 0.950 0.003 -0.012 0.016 0.051 0.448 0.128 0.000 0.216 0.048 0.107 0.000 -4.423 0.000 0.156 0.000 1.099 -2.073 0.000 0.041 -2.961 0.042 0.047 (Fig. 1),withregressionmodelsusingrosettesize value ofthefieldpopulationaregiventoright. The ab · · · f f f Regression parameters Regression 3 3 3 36 .0 .1 .0 133.1 0.100 0.010 0.003 63.67 28 .0 .0 .9 28.669 0.195 0.000 0.000 22.88 71.08 0.028 0.310 0.015 21.73 p a p b R a w 2 < 0.05. 2,3 ) andflowering 0.023 0.858 0.019 0.539 0.008 0.000 0.000 0.897 0.001 31.88 0.006 0.000 2.481 0.000 0.450 w k · · · · · · · · · · · · · · · · · · · · f f f f f f f f f f f f f f f f f f f f · · · · f f f f 4 6 5 5 5 5 4 6 5 5 5 5 4 6 5 5 5 5 4 6 5 5 5 5 w 4,5 ) 137. Chapter 8 A hierarchical population matrix model , i f (9) (11) (12) (13) (10) is (van Tienderen between 0 and 1) r i f z was investigated both (van Tienderen 2000): (van Tienderen λ is (Caswell 2001; Franco is (Caswell r z k w between 1 and 2.5). i f to plant trait λ , served as the reference matrix. We per unit of k N λ to the life history components or plant to the life history components A λ kk ww ∂∂ ij a to life history component component to life history r ij ∂ z of the natural and enriched matrices into contributions a λ e ∂ to all matrix elements sum to one (Mesterton-Gibbons to all matrix elements λ ∂ ij ij λ aa ij ∂∂ ∑ k ∑ ⎜⎟ ⎜⎟ ⎝⎠ ⎛⎞ r ⎜⎟ ⎜⎟ ⎝⎠ ⎛⎞ z ij ij ∑∑ kk kij k (Caswell 2001). In order to be able to compare plant traits which plant traits able to compare (Caswell 2001). In order to be ∑∑ ∑ rr ww ∂λ rr r r λ zz kijkk zz z z ∂λ ∂λ ∂λ rijkrr kk waww rr k λ∂ λ ∂ ∂λ ∂λ zawzz ww ∂∂∂∂ λ∂ λ ∂ ∂λ ∂λ 11 λ∂ λ∂ zz w ∂∂∂∂∂ ======kij kij r rk r z zw zw wa wa ss es β= = = es ss Secondly we performed life-table response experiment analyses (LTRE) to response experiment analyses (LTRE) Secondly we performed life-table , which measure the proportional increase in , which measure the proportional 2000): can not be interpreted as relative and elasticities of lower-level parameters traits contributions to mean value we also calculated phenotypic selection gradients, strongly in their differ β 1993), this is not true for the elasticities of 1993), this is not true for the Whereas the elasticities of one by one, while keeping the others constant. The effect on The effect one by one, while keeping the others constant. in the range where the affected component or trait was lowered ( in the range where the affected estimated the LTRE effect of nutrient enrichment in each species with the following effect estimated the LTRE formulas (Horvitz et al. 1997; Caswell 2001; Cooch et al. 2001): The sensitivity and elasticity of and elasticity The sensitivity & Silvertown 2004): & Silvertown and in a range where the affected parameter is raised ( parameter and in a range where the affected decompose the differences in decompose the differences by the differences in the matrix elements or lower-level parameters. The matrix of the or lower-level parameters. in the matrix elements by the differences natural population dynamics of each species, Analyses of the impact of nutrient enrichment on the population dynamics Analyses of the impact values of the population dynamics to which the of the elasticity Besides a comparison two of enrichment were applied with that of undisturbed population dynamics, effects the six nutrient factors, we varied other types of analyses were carried out. Firstly, And consequently, the sensitivity and elasticity of the sensitivity and elasticity And consequently, 138. Chapter 8 A hierarchical population matrix model plant traits (eq.17). decomposed intothedifference inlifehistorycomponents (eq.16)orthedifference in half-way thefertilizedandnaturalmatrix.Inasimilarwaydifference in fertilized andthecontrolmatrix,sensitivityof matrix elements theproductofdifference betweenthematrixelements ofthe in whichtheeffect ofnutrientenrichment, three species,butnot in reproductive allocation andvegetative plantsizeofsurvivingplants both increasedin Nutrient enrichment had different effects on thefourspeciesstudied.Sexual reproduction Different effects ofnutrientenrichmentinspecies differinginlongevityandmodeof Results dissectum life historycomponents. Besidesthe The experimentally determinedvaluesoftherelativeimpact offertilization, Table 2. all speciesarealsogiven. f f f f f f f 6 5 4 3 2 1 i Establishment rate Establishment Adult survival Seedling survival Allspecies production Relative seed threshold Flowering Plant size trait/component Affected ≅− ≅− ≅− = −λ ϕ=λ  , ∑ ∑ ∑ Succisa pratensis k ij r () () () FN ww aa zz rr jij ij FN FN kk FN ∂ ∂ ∂λ ∂λ ∂ z a ∂λ r w ij w w w z z z k and 3 2 1 2 1 2 1 10 2,3,4,5 1 () () AA AA FN 2 1 FN + () + AA Centaurea jacea FN Cirsium dissectum + Field values Field f -values ofthefourspecies( .010 .100 0.00 1.00 1.00 0.00 2.02 0.53 0.48 0.53 1.44 0.31 1.00 1.00 0.00 1.00 0.81 1.00 2.48 0.15 0.00 0.15 1.34 1.00 1.93 1.00 0.00 1.00 1.68 1.00 0.00 1.00 .aiaaS rtni .iscu C.jacea C.dissectum S.pratensis H.radicata ϕ ), the , isestimated(eq.15)bysummingoverall f -values ofthenaturalpopulationdynamics (Table 2). As predicted,nutrient λ to thatelement,determined Fertilized values Fertilized Hypochaeris radicata f i , ondifferent planttraits and , Cirsium λ was (17) (16) (15) (14) 139. Chapter 8 A hierarchical population matrix model λ S. (Fig. (-6%; , was 6 Cirsium w , gained in 10 w S. pratensis (Fig. 3a,e). In and 9 w and , whereas only small , Centaurea jacea Centaurea 6 w , 1 shifted towards the relative shifted w λ = 0.96 to 0.33) and H. radicata λ H. radicata H. radicata . Although the trait elasticity values Although the trait elasticity . = 1.23 to 1.28). The decomposition = 1.23 to 1.28). λ , of the matrix elements resembled the resembled , of the matrix elements , and the probability that those rosettes , and the probability that those ϕ 3 w of the most short-lived species. This was species. short-lived of the most (+4%; λ H. radicata (-66%; from were consistent with our expectation that plant were consistent with our expectation , was the most important life history component life history important , was the most was very responsive to changes in survival rates changes in survival responsive to was very , the selection gradient of sexual reproductive 2 and especially λ w , was strongly reduced by the effects of nutrient , was strongly reduced by the effects λ . 7 C. jacea w H. radicata Succisa pratensis S. pratensis = 1.01 to 0.51) but only slightly affected in = 1.01 to 0.51) but only slightly affected λ Hypochaeris radicata closely (the difference was -1.2% on average). The LTRE-effects of the The LTRE-effects was -1.2% on average). closely (the difference λ (-50%; however, the emphasis of the elasticity of the emphasis however, Only the results of Only the results This is because the selecetion gradients only equal the elasticity values after This is because the selecetion gradients . = 0.98 to 0.92) and The importance of life history components and plant traits for population growth rate for population growth plant traits and of life history components The importance rosettes, Survival of non-flowering through increased plant sexual reproductive allocation and especially increased plant allocation and especially increased plant sexual reproductive through increased (Fig. for moderate increases in mortality the potential to compensate size did have 2a,c). in all four herbaceous perennials (Fig. 2e). The positive effects of fertilization on of fertilization effects The positive 2e). perennials (Fig. herbaceous in all four mostly caused by the observed effects on survival in the nutrient enrichment in the nutrient on survival observed effects caused by the mostly 2), as Table (Chapter 3; experiment changes took place in the elasticity values of life history components in the two clonal values of life history components changes took place in the elasticity of sexual reproduction, species. Life history components enrichment had a larger impact on the on the impact had a larger enrichment flower again one year later, flower again one year later, increased in the fertilized population of increased in the fertilized population of plant size were much higher than that of sexual reproductive allocation, their of plant size were much higher were of the same magnitude in selection gradients pratensis in all species (Fig. 3e/h). This component became less important in the fertilized became less important This component in all species (Fig. 3e/h). population matrix of declining, fertilized population of in the importance size would be more important in fertilized population than in natural populations (Fig. in fertilized population than in size would be more important due to of plant size decreased in magnitude 3i), while the selection gradients gradient of the selection (Fig. 3j,k,l). Especially the part fertilization in the other species rosettes, flowering probability of vegetative of plant size that involved the 3i,j) importance of survival of flowering rosettes, of survival importance enrichment in Changes in population dynamics due to nutrient enrichment The population growth rate, allocation was much higher than that of plant size in the natural population, but was allocation was much higher than that of plant size in the natural occurred at all. reduced to zero in the fertilized population where no recruitment dissectum λ division by the mean trait value, which is much larger in plant size than in sexual division by the mean trait value, which is much larger in increased both size reproductive allocation. In the third species in which fertilization and allocation plant traits, model fitted well as the summed LTRE-effect, model fitted well as the summed LTRE-effect, difference in difference 140. Chapter 8 A hierarchical population matrix model was reducedmoststronglyin contribution ofsexualreproductionto ( mostly duetoaloweradultsurvivalratebutalsodecreasedseedling effects ofplanttraits arestronglydependentonthescaleoftraits involved. but alsoduetodifferences inlifehistorycomponents directly. And lastly, theLTRE- between controlandfertilizedmatricesarenotonlyduetodifferences inplanttraits which theabove-mentioneddeviations,areone.Butitisalsobecausedifferences increased to45%. The relativenumberofseedlingswasmore constant in was reducedfrom6%to0%in increased rosettemortality rates(Table 2).Contrastingly, theproportionofseedlings much higherthaninthenaturalpopulation(71% proportion ofseedlingsinthestronglydeclining,fertilized stable stage structureofthepopulationschangedinthreefourspecies: they inhabitbecomemoreproductiveduetosuccessionornitrogendeposition. The ormoredynamicwhenthegrasslands these herbaceousperennialsbecome‘senile’ flowering rosette(positiveLTRE-effect of species thenegative effects ofincreased competitionbyneighborsprevailed lived herbaceousperennials: ourmodelingexerciserevealed that inshort-lived some lifehistorycomponents by nutrient enrichmentdiffered forshort-livedandlong- The balanceofpositive, directeffects on planttraits andnegative,indirecteffects on Discussion probabilities (positiveLTRE-effect of lower recruitmentratesarealmostentirelycompensatedforbybothhigherflowering seedling recruitmentalreadyunderunfertilizedconditionsandintheothertwospecies senile phasewithoutseedlingrecruitment,because traits didnotresemble in Coochetal.2001foradiscussiononthisproblem). The LTRE-effects oftheplant (see Appendix A parameters arecompared withpartial derivativesofmatrixelements due todeviationsthatarisewhensumofproducts ofpartial derivativesoflower-level life historycomponents deviatedmorefrom conditions (23% but inthatspeciesahigherpercentage oftheplants floweredundernutrientenriched other twospecies(Fig.4j,k). jacea reproductive allocationcontributedmoretothedifference in H. radicata (Fig. 4i,l),whilethedifference inplantsizecontributedrelativelymorethe The largereductionsof One ofourresearchquestionswastoinvestigatewhetherthepopulations The traitmean-standardized LTRE-effects showthatthedifference insexual ; Fig.4a,e)andlessclonalpropagation ( vs . 16%).Overall,onlythepopulationof ∆λ at all(+4,900%onaverage). This hasseveralreasons,of C. jacea λ C. jacea in thetwospecieswithshort-livedrosetteswere w ∆λ (Fig. 4d). 6 ; Fig.4e,f)andhigherseedproductionper w while theproportionofnewclonaloffspring increased onlyin 9 ∆λ , Fig.4e,f). (-3.5% onaverage),whichisprobably vs. 27%), probablyduetothestrongly C. dissectum C. dissectum H. radicata C. jacea H. radicata λ in H. radicata seems toentera ; Fig.4c,g). The had insignificant population was (Fig. 4a)and S. pratensis and in C. , 141. Chapter 8 A hierarchical population matrix model S. , all Gentiana while the λ may also be S. pratensis at the end of the and C. dissectum lower than one and almost λ after two years of nutrient after S. pratensis however showed the opposite H. radicata H. radicata . But the fecundity elements were also . But the fecundity elements λ H. radicata Centaurea jacea Centaurea , the data on the rare , the data and by fecundity. Plant size in comparison mattered much Plant size in comparison by fecundity. C. jacea λ and populations in closed vegetations have a populations in closed vegetations as it is also part of the survival and clonal propagation components. But components. and clonal propagation of the survival as it is also part Succisa pratensis Succisa pratensis S. pratensis after longer periods of nutrient enrichment in the field. Is this pattern of first pattern longer periods of nutrient enrichment in the field. Is this after λ In our previous experiment (Chapter 3) mortality rates were unchanged in the rates were unchanged experiment (Chapter 3) mortality In our previous experiment may indicate that some plants were going to die in the fertilized plots if the in the fertilized plots were going to die some plants experiment may indicate that of Billeter et al. (2003) reported lower adult survival experiment had lasted longer. pratensis other from by a decrease of adult survival different a reduction in recruitment followed clearly show that in the long-lived herb studies? Oostermeijer et al. (1996) pneumonanthe The importance of sexual reproductive allocation for the population growth rate of sexual reproductive The importance In the two mainly sexually reproducing species, consistent with this general pattern, when considering that seedling recruitment was when considering that consistent with this general pattern, of this rhizomatous of succession and that genets already prohibited in earlier phases The short-lived species are long-lived. than a reduction in nutrient enrichment induced relatively more mortality pattern: to increase rather than recruitment, and the proportion of seedlings was found decrease in declining populations. of the sexual reproduction that are part elasticity values of the life history components rosettes, seed production, seedling of vegetative (flowering probability pathway when fertilization was recruitment and seedling survival) increased to some extent limited, although the was of the allocation traits the importance simulated. Still seed production, which increased sexual reproductive allocation markedly increased This may be caused by (Chapter 4). can be a significant step in the sexual pathway the limited contribution to enrichment, but the increased variation in plant size in the increased variation in plant enrichment, but elasticity of fecundity was positively correlated with The no recruitment. with lower elasticity of survival increased long-lived strongly increasing mortality rate and consequently a lower population growth rate. In growth rate. a lower population consequently rate and increasing mortality strongly and the were unaffected, rates the mortality rosettes with longer-lived the species positive for by the were compensated rates of reduced establishment effects negative sexual and increased plant size through increased production on seed effects herbs are being gradually that when grassland This suggests reproductive allocation. rate by their population growth maintain succession they can first suppressed during But when eventually chances. for lower seedling establishment compensating also the biomass in vegetation go up by continued increases rates mortality will strongly decline. these longer-lived herb species populations of when analyzing only sexual reproduction, plant size has higher elasticity values than when analyzing only sexual reproduction, plant size has higher more for positively correlated with important survival elements, indicating that both recruitment survival elements, positively correlated with important These authors however also declining populations. and survival was reduced in these that decreases first during succession. Besides the suggested that it is recruitment on results 142. Chapter 8 A hierarchical population matrix model of aplantlike estimator (butoneofthemostfeasiblenon-destructiveoptions)biomass flowering thresholdseither, althoughitmaybethatrosettesizeisnotthebest regressions ofplantsizeonfloweringprobabilitiesinthefieldshowednoclear-cut size playsaminorroleandishardtodetermine(Klinkhameretal.1992). The logistic experiment (Chapter3)wentthroughtheorigin,indicatingthatafloweringthreshold relationship ofsexualreproductivebiomassandvegetative biomassinthegarden aspect ofthecolonizationcapacity ofthesespecies(Soonsetal.2004). regional populationdynamicsasthenumberofproducedseedsisaveryimportant be veryimportant whenlocallysmall-scaledisturbancesareintroducedandfor have limitedbuffering capacities fortheoverallnegativeeffects offertilization,itcan production (causedbyincreasesbothinsizeandallocationduetofertilization)may but alsodetermineswhetherrosetteswillflowerornot. Although theincreasedseed sexual reproductiveallocationbecauseitnotonlyinfluencesseedproductiondirectly, our species(Chapter2) wereinherenttothedemographicfielddata. were independentand thattheimportant trade-offs betweenlifehistorycomponents in traits (van Tienderen 2000).However, weassumedthat the planttraits inourmodel underlying planttraits, andtoevaluatedirectindirectelasticityvaluesofthese way tostudytrade-offs isprobablytouseknowledgeofthecorrelations (2002) furtherwarnsthattrade-offs mayinvolveseveralconsecutiveyears. The best trade-offs dueto theirrelativenature(Sheaetal.1994;deKroon2000).Ehrlén However, negative correlationsinelasticitypatterns cannotbedirectlyinterpretedas effect onthepopulation dynamics(van Tienderen 2000;Shefferson etal.2003). opens theopportunitytoinvestigatetrade-offs betweenmodelcomponents, andtheir (van Tienderen 2000;Coochetal.2001). This underlying lifehistorycomponents correlated variation)ofthematrixelements. based onthesamecomponents: thesensitivityvalueandmean(orpositively arepositivelycorrelatedbecausethey (Chapter 5)elasticityvaluesandLTRE observed variationin λ supplement eachotherinausefulway:elasticityvaluesindicateonwhichparameters matrices. Bothmethodsanswerdifferent questions(Caswell2000),butcan Ellison 2002).ElasticityandLTRE providedtwopowerfulmeanstoanalyzepopulation populations canbeevaluatedandthatlong-termeffects canbemodeled(Gotelli& experiments arecombinedthattheimportance oftheexperimental findingsfor The benefits ofamodelinwhichdemographicfielddata andresults ofmanipulative vital rates Transition matricesastoolsforstudyingthepopulationconsequencesofeffects on depends, whileLTRE analyseswhichdeviationsindifferent parameters causedthe Flowering thresholdwastheleastimportant traitinouranalyses. All butone Hierarchically constructedmodelsfurtherallowforsimilaranalysesofthe S. pratensis λ . Whenthenaturalvariationinpopulationdynamicsisstudied (Graaskamp 2003). 143. Chapter 8 A hierarchical population matrix model . - MSc thesis, . - Oikos 50: 3-12. Succisa pratensis Hypochaeris radicata . - Ecology 83: 2758-2765. (Fabaceae). - Perspect Plant Ecol Evol Syst 5: 145-163. meadow species to abandonment. - Appl Veg Sci 6: 3-12. Appl Veg - meadow species to abandonment. - Ecology 81: 619-627. biology. Associates, Inc. Publishers, Sunderland, Massachusetts. projection model. - Proc variable environment: construction and analysis of a stochastic integral R Soc Lond Ser B-Biol Sci 271: 425-434. A change: environmental Lesser Snow Goose example. - Ecol Monogr 71: 377-400. to population growth rate. - Ecology 67: 1427-1431. demographic parameters dynamics of a perennial grassland species, limitations. - Ecology 81: 607-618. - Ecology 81: limitations. Lathyrus vernus - Ecology 85: 531-538. Sarracenia purpurea University of Nijmegen. S & Caswell H Tuljapurkar population growth: prospective and retrospective analyses. - In: References DAPBilleter R, Hooftman and reversible responses of common fen & Diemer M 2003. Differential Their roles in conservation retrospective perturbation analyses: Caswell H 2000. Prospective and - Sinauer models. Construction, analysis, and interpretation. Caswell H 2001. Matrix population Childs DZ, Rees M, Rose KE, Grubb PJ & Ellner SP 2004. Evolution of size-dependent flowering in a of demographic responses to Cooch E, Rockwell RF & Brault S 2001. Retrospective analysis of A, van Groenendael J & Caswell H 1986. Elasticity: the relative contribution de Kroon H, Plaisier Ade Kroon H, Plaisier JM 1987. Density dependent simulation of the population & van Groenendael Acknowledgements Acknowledgements This Limpens and Pieter Zuidema for constructive discussions. are grateful to Juul We the Netherlands Organization for Scientific Research research was financed by (NWO-project 80.33.452). Conclusions of growth rate on the population influences strong negative enrichment has Nutrient a few years take species it may In long-lived in grasslands. perennials herbaceous will face serious also populations of these species but is affected before adult survival the system. continue to enter or nutrients when succession progresses extinction risks and establishment lower to buffer species have some capacity The investigated allocation higher sexual reproductive seed production through survival by increased growth rates drop further the population sizes, but when survival rates and larger plant life history production is only a feasible In the end increased seed will decline quickly. success. to a higher life time reproductive strategy if it leads de Kroon H, van Groenendael JM & Ehrlén J 2000. Elasticities: a review of methods and model de Kroon H, van Groenendael JM & Ehrlén J 2000. Elasticities: a Assessing the lifetime consequences of plant-animal interactions for the perennial herb Ehrlén J 2002. rates. based upon elasticities of vital demography of plants Franco M & Silvertown J 2004. Comparative AM 2002. Nitrogen deposition and extinction risk in the northern pitcher plant, Gotelli NJ & Ellison Graaskamp, E 2003. Environmental effects on the life history of effects Graaskamp, E 2003. Environmental Horvitz C, Schemske DW & Caswell H 1997. The relative "importance" of life-history stages to of life-history stages The relative "importance" Horvitz C, Schemske DW & Caswell H 1997. 144. Chapter 8 A hierarchical population matrix model Oostermeijer JGB,BrugmanML,deBoerER&denNijsHCM1996. Temporal andspatial variationin Neitzke M2001. Analysis ofvegetation andnutrientsupplyincalcareousgrasslandborderzonesto Morris WF&DoakDF2004.Buffering oflifehistoriesagainstenvironmental stochasticity: Accounting Mesterton-Gibbons M1993.Whydemographicelasticitiessumtoone:apostscript todeKroonetal.- 1965. The studyofpopulationgrowthinorganismsgroupedbystages. -Biometrics21: Lefkovitch LP Klinkhamer PGL,MeelisE,deJong TJ &Weiner J1992.Ontheanalysisof size-dependent Vergeer OuborgNJ2003. & The interactingeffects P, ofgeneticvariation,habitat RengelinkR,Copal A van Tienderen PH2000.Elasticitiesandthelinkbetweendemographic andevolutionarydynamics.- Stevens CJ,DiseNB,Mountford JO&GowingDJ2004.Impact ofnitrogendepositiononthespecies Soons MB,MesselinkJH,JongejansE&HeilGW2003.Fragmentation andconnectivityofspecies- Soons MB,HeilGW, NathanR&KatulGG2004.Determinants oflong-distance seeddispersalbywind Soons MB&HeilGW2002.Reducedcolonizationcapacity infragmentedpopulationsofwind- 1993.Comparative plantdemography:Relative Silvertown J,FrancoM,PisantyI&Mendoza A 2003.Lifehistorytrade-offs inarareorchid: The Shefferson RP, ProperJ,BeissingerSR&SimmsEL Shea K,ReesM&Wood SN1994. Trade-offs, elasticitiesandthecomparative method.-JEcol82: the demographyof determine criticalloadsfornitrogen.-Flora196:292-303. - Am Nat163:579-590. for aspuriouscorrelationbetweenthevariabilitiesofvital ratesandtheircontributionstofitness. Ecology 74:2467-2468. 1-18. reproductive outputinplants. -FunctEcol6:308-316. Hall, New York. (eds), Structured-population models inmarine,terrestrial,andfreshwatersystems.Chapman& quality andpopulationsizeonperformanceof Ecology 81:666-679. richness ofgrasslands.-Science303:1876-1879. University. Spatial andtemporalcharacteristics ofthecolonizationprocessinplants. Ph.D.thesis,Utrecht rich semi-naturalgrasslands.-In:SoonsMB(ed),Habitat fragmentation andconnectivity. in grasslands.-Ecologypress. dispersed grasslandforbs.-JEcol90:1033-1043. perennials. -JEcol81:465-476. importance oflife-cyclecomponents tothefiniterateofincreaseinwoodyandherbaceous costs offlowering,dormancy, andsprouting.-Ecology84:1199-1206. 951-957. Gentiana pneumonanthe , arareperennialherb.-JEcol84:153-166. Succisa pratensis . -JEcol91:18-26. Summary

This study aims to contribute to the knowledge of how plants respond to adverse influences of intensified land use. In particular, attention was paid to the ways in which life history strategies change in order to buffer environmental variation, and which important parts of the life cycle are affected most. Herbaceous plant species are often threatened when the land use in an agricultural landscape is intensified: habitat is destroyed for new fields, infrastructure or urban areas, or habitat becomes less favorable because water levels are lowered for agriculture and because of pollution and nutrient enrichment by the influx of particles through air and water from the neighboring fields. Nutrient enrichment enhances natural succession and causes more competitive species to take over from species that are characteristic for nutrient- poor grasslands. As a result of habitat reduction and deterioration, populations decline in size and are more prone to extinction. The distances between remaining populations also increases, which makes gene flow through pollen or seed dispersal less likely. Decreased gene flow may lead to inbreeding depression and again higher extinction risks.

Plant species may counterbalance some of these influences by flexible growth patterns and may adapt for instance to nutrient enrichment. For the conservation of the floral diversity of agricultural landscapes it is therefore important to know to what extent plant species are endangered by the different effects of intensified land use and to what extent these species can buffer the changes in their growing conditions. In this study I concentrate on four co-occurring perennial herbs in nutrient-poor, species-rich meadows. With three species from the Asteraceae (Cirsium dissectum (L.) Hill, Hypochaeris radicata L. and Centaurea jacea L.) and one from the Dipsacaceae (Succisa pratensis Moench), the species are phylogenetically related but have contrasting life history strategies as they differ in the life-span of their rosettes and in the rate of clonal propagation.

The flexibility of the allocation of biomass to different life history functions (plant growth, sexual reproduction and clonal reproduction) is investigated in chapter 2. In a one-season garden experiment we continuously removed either rosette buds or flower buds to study the responses to damage. All investigated plants compensated for the removed buds. The short-lived species (H. radicata) showed stronger compensation for lost flower buds than the longer-lived species (S. pratensis and C. jacea) and the species with a high clonal propagation rate (C. jacea) compensated more strongly for lost rosette buds than the infrequently ramifying species. However, two species (H. radicata and S. pratensis) also switched to increased rosette formation when flower 146. Summary four years. The sexualpathway ofthe life cyclewascompletedwithdata ofthe flower headproduction weremadeinpermanentplots inthreeorfive populationsover in chapter5.Demographic measurements onrosettesurvival,clonal propagation and The wholelifecycleis studied forthreespecies( pathway ofthelifecycle. set, indicatingthatselectionprocessescanbe expectedindifferent parts of thesexual closely followedinmagnitudebythecoefficients forflowerheadproduction andseed spatiotemporal coefficients ofvariationwerehighestforseedlingestablishment, but different steps werestudiedin several sitesandoveryears. The pathway, whileseedsetwasimportant tooinoneofthemoreclonalspecies. The ( on average. This numberwasbelowoneinthetwomoreclonally propagating species and and establishment asseedlings. The twoinfrequentlyramifyingspecies( involved steps: flowerheadproduction,seedset,predation seeds andbyperformingseedadditionexperiments wequantifiedforeachspeciesall further investigatedinchapter4.Bysamplingfloweringplants andflowerheadswith The sexualreproductionpathway, fromflowerproductiontoestablished seedlings,is survive andtoincreaseseedproduction. a singlesyndrome:inresponsetosuccessionplants must increaseinsizeorderto dependent onplantsize.Consequently, thetwohypothesizedstrategiescollapse into in smallplants. Allocation toeverylifehistoryfunctionwasstronglyandpositively the fourthspecies( allocation tostoragedidnotdecreasebutinsomespeciesevenslightlyincreased.In allocation toflowersandseedswhentheirplots wereenriched,whiletheirrelative means ofcompetingwiththegrasses. Three target speciesincreasedtheirrelative two target species( threefold increaseofthegrassbiomass.Mortality ratesincreasedunderfertilizationin of rosettes.Successionwasmimickedbyfertilizingtheplots, whichcauseda were foundinalltarget speciesaseitherreducedplantbiomassornumber Costs ofsexualreproductionwereinvestigatedbycontinuousflowerbudremovaland were growntogetherwithadominating,tall grass( but largelyuntestedhypothesesinathree-yeargardenexperiment. The target plants investing moreinvegetative growth.Inchapter3weaimtotesttheselong-standing, expense ofplantgrowthandfuturereproduction),oralternativelytopersistby their seedproductiontoescapeothersitesthroughdispersal(butatthe are graduallyoutcompetedduringsuccession:plants areexpectedtoeitherincrease Costs ofsexualreproductionunderliethehypothesesonresponseplants that propagation. weight, thatsexualreproductionhascosts intermsofgrowthsizeandclonal buds wereremoved. This switchshows,togetherwithincreasedvegetative plant C. dissectum S. pratensis and ) hadhighnumbersofseedlingsperfloweringrosette:morethanone C. jacea C. jacea H. radicata ). Seedlingestablishment wasthelargestbottleneck inthe ) seedproductionwasonlyhigherinlargeplants butnot and C. dissectum S. pratensis ) probablybecausetheylackeda Molinia caerulea , H. radicata (L.) Moench). and H. radicata C. jacea ) 147. Summary . ) ) and H. radicata S. pratensis H. radicata and ). The implication of this ). S. pratensis C. jacea ). Using life table response experiment response life table ). Using C. jacea ) was studied in chapter 7. The results of the The results ) was studied in chapter 7. C. dissectum most stable in the most clonal species ( clonal species in the most most stable to the same in their response found that the species differed analyses we responded Furthermore, species conditions. variation in environmental spatiotemporal to years in which low population growth rates than in sites in which they had differently species ( growth rates. In two they had low population previous chapter. We constructed 42 stage-based projection matrices to study the matrices to study projection 42 stage-based constructed We chapter. previous The calculated over time. sites in different the three species dynamics of population ( species in the most short-lived were lowest growth rates population below-average values of some life cycle components in bad sites were partly in bad sites were components values of some life cycle below-average but similar of other components, for by the above-average values compensated did appear contrast, this compensation did not occur in bad years. By compensation ( the third, more clonal species in bad years for whereas the total allocation experiment of the third chapter are further analyzed: by the fertilization treatment, the surviving was unaffected biomass of the plants of and more remnants rosettes were further away from the original planting location Field observations on excavated plants dead rosettes were found in the fertilized plots. and rhizome formation are and an additional experiment confirmed that both flowering of flowering rosettes in the that the high percentage This suggests size-dependent. The the fertilized rosettes. allocation experiment was the result of a high growth rate of rate of the non- the mortality because from the control plots biomass did not differ total of increased This pattern flowering rosettes was also higher in the fertilized plots. is consistent with the flowering and of increased turn-over of biomass and rosettes In the most productive site we found high study. of a demographic field results but lower rosette flowering probabilities and above-average clonal propagation competitor even This trend towards higher turn-over rates makes this bad survival. difference between spatial and temporal variation in population dynamics is that and temporal variation spatial between difference variation as has readily be substituted with spatial temporal variation can not time-limited studies. previously been done in some demographic implications of inbreeding depression in In chapter 6 we investigate the on several life history components of inbreeding effects order to evaluate if reported dynamics and population survival of significantly influence the population For that purpose we constructed a projection matrix model with an outcrossing and an For that purpose we constructed latter growth of seedlings and small rosettes were reduced in the The inbreeding cycle. The proportion using empirically based data. cycle, as was seed set and germination, was a function of the size of the population: more seeds of seeds entering either cycle a declining and an increasing compared We were inbred in smaller populations. was affected, of the life cycle Although only a part population in the simulations. to extinction in the decreasing population. Inbreeding inbreeding reduced the time population only when the initial population was small. Density the increasing affected increasing population. the size only affected dependency by limiting habitat most endangered herb of The demography of the most clonal (rhizomatous) and also the model species ( 148. Summary dynamics ofaspeciesinthelandscapethroughincreasedcolonizationprobabilities. population, thisflexiblelifehistoryresponsemayhavepositiveeffects onthe allocation tosexualreproductionleadsincreasedseedproductionforthetotal to beevenacceleratedbyinbreedingdepressions.However, whenincreased succession andnutrientenrichment. The declineofsmallpopulation canbeexpected populations willgoextinctingrasslandsthathavebecomemoreproductivethrough that althoughthelifehistoryresponsesoftheseperennialsareratherflexible,their and seednumbercannotcompensateforhighdeathrates.We thereforeconclude decline oncethemortality ratesofindividualsincrease.Positiveeffects onplantsize the modelresults alsosuggestthatinthesebetter competitorspopulationswill were smallorcompensatedbyincreasedseedproductionandplantsize.However, mortality wasstillunaffected, negativeimpacts ofreducedseedlingestablishment competitors ( fertilization onthepopulationgrowthratebyincreasedmortality ratesinthetwoweak the components inturnwerefunctionsofplanttraits. The negativeeffects of matrix inwhicheachelementwasafunctionoflifehistorycomponents andmanyof order tomodeleffects onlifehistorycomponents andplanttraits we builtahierarchical effects wereincorporatedintoamodeldescribingthenaturalpopulationdynamics.In seedling establishment, wewerealsointerestedintheoveralloutcomewhenall positive effects onplantsizeandseedproductionnegativeeffects onsurvivaland have thestrongestimpact onthedynamicsofpopulation. As fertilizationhadboth results ofthefertilizationexperiments toinvestigatewhicheffects on individualplants The lastchapterwasaimedtointegratethepopulationdynamicsmodelsand habitats. rate asseedlingrecruitmentseemsconfinedtoveryopen,early-successional increased seedproductionhasanegligiblecontributiontothelocalpopulationgrowth more pronetoextinctionwhenthenutrientenrichmentofameadowcontinues. The population declines.Intheothertwospecies( C. dissectum and H. radicata ) overruledothereffects andcausedstrong S. pratensis and C. jacea ), inwhich Samenvatting

Het doel van deze studie is bij te dragen aan de kennis over de wijze waarop wilde planten reageren op de negatieve effecten van een intensiever landschapsgebruik door mensen. Onderzocht is vooral de manier waarop planten hun levensstrategieën veranderen om de verandering in de groeiomstandigheden te bufferen, en welke belangrijke onderdelen van de levenscyclus het meest beïnvloed worden door het veranderde milieu. Kruiden worden vaak bedreigd wanneer de landbouw in een landschap wordt geïntensiveerd: habitat wordt vernietigd voor nieuwe akkers, infrastructuur of bebouwing, of de kwaliteit van de habitat gaat achteruit doordat de grondwaterstand wordt verlaagd voor de landbouw en doordat graslanden vervuild en verrijkt worden door de omgevende landbouw via het water en de lucht. Verrijking versterkt natuurlijke successie van de vegetatie en zorgt ervoor dat meer competitieve soorten gaan domineren ten koste van soorten die karakteristiek zijn voor nutriëntarme graslanden. Door habitatverkleining en -verslechtering worden plantenpopulaties kleiner, waardoor de kans op uitsterven groter wordt. De afstand tussen populaties wordt groter bij intensiever landgebruik, waardoor pollen en zaden minder makkelijk van de ene naar een andere populatie worden verspreid. Daardoor ontstaat inteelt, wat de vitaliteit van de planten kan benadelen en de kans op uitsterven van een populatie wederom vergroot.

Planten zouden een aantal van deze invloeden kunnen compenseren met hun flexibele groei en zich kunnen aanpassen aan voedselrijke omstandigheden. Voor het behoud van de floristische diversiteit in agrarische landschappen is het daarom belangrijk te weten hoezeer plantensoorten bedreigd worden door intensief landgebruik en in welke mate deze soorten de veranderingen in het milieu kunnen opvangen met aanpassingen. In deze studie staan vier kruiden centraal, die samen voorkomen in nutriëntarme maar soortenrijke hooilanden. De soorten zijn fylogenetisch verwant (drie Asteraceae: Spaanse ruiter (Cirsium dissectum (L.) Hill), Gewoon biggekruid (Hypochaeris radicata L.) en Knoopkruid (Centaurea jacea L.) en een Dipsacaceae: Blauwe knoop (Succisa pratensis Moench)), maar ze hebben contrasterende levensstrategieën: de levensduur van de rozetten en de frequentie van uitstoeling verschillen tussen de soorten.

De flexibiliteit van de biomassa-allocatie naar verschillende plantonderdelen, die van belang zijn voor de contrasterende levensstrategieën (groei van het rozet, seksuele reproductie en klonale vermeerdering), wordt beschreven in hoofdstuk 2. We verwijderden de knoppen van ofwel bloemhoofdjes ofwel zijrozetten gedurende een groeiseizoen in een experiment in een proeftuin om uit te vinden hoe de planten 150. Samenvatting uitstoelen, bloemhoofdjes endoorzaai-experimenten. Detweesoortendierelatiefweinig bloeiende plantentebemonsteren,doorbloemen enzadentetelleninafzonderlijke zaden, zaadpredatie endevestigingvanzaailingen) werdengekwantificeerddoor hoofdstuk 4. Alle faseninditproces(hetaanmakenvanbloemhoofdjes, bloemenen De seksuelereproductiebijplanten,vanbloei totzaailing,wordt verderonderzochtin worden omnogtekunnenoverlevenmaarook ommeerzadentekunnenproduceren. hypothesen samen:inlateresuccessiestadia van devegetatie moetenplantengroter zwaarder dandievankleineplanten.Daaromvallendetweecontrasterende planten afhankelijkvandegrootteplant:alleonderdelengrotezijn niet bijkleinere.Maarbovenalwasallocatienaardeverschillendeonderdelenvan de vierdesoort, opslagorganen nietverminderdemaarinsommigesoortenzelfslichtelijktoenam.Bij en zadenwanneerdeplots verrijktwerden,terwijlhunrelatieveallocatienaarde gras. Driedoelsoortenverhoogdenhunrelatievebiomassa-allocatienaardebloemen en het grasdriemaalzozwaarwerd. Tegelijkertijd namdesterfte toebijde of lagerrozetaantal. Successiewerdnagebootst doordeplotjesteverrijken,waardoor knopverwijdering enwerdenbijallesoortenaangetroffen als afgenomenplantgewicht De kostenvanseksuelereproductiewerdenonderzochtdoorcontinue hoogopgaande endominerendegrasPijpenstrootje( Individuele stekjesvandeviersoortenwerdengeplanttussenpollenhet oude, maarnogweiniggetestehypothesenineendriejarigexperimentdeproeftuin. door juistmeerteinvestereninvegetatieve groei.In hoofdstuk 3toetsten wedeze zaadverspreiding, ofweldeverhoogdecompetitiemetandereplantenteweerstaan koste gaatvangroeientoekomstigezaadproductie) omzoteontsnappen via worden verondersteldofwelhunzaadproductie teverhogen(hetgeenduswelten de responsvanplantendieinverdrukkingkomentijdenssuccessie: Deze kostenvanseksuelereproductievormenhetuitgangspunthypothesenover klonale vermeerdering. zien datseksuelereproductieduidelijktenkostegaatvandegroeihetrozeten met degrotereplantgewichtentenopzichtevannietbehandeldecontroleplanten, wanneer debloemknoppencontinuverwijderdwerden.Dezeverschuivinglaat,samen verwijderde knoppen.Maarhetkortlevende reageren opdergelijkebeschadiging. Alle plantencompenseerdenvoordecontinu soorten per bloeiendrozet.Dit getal kwamechterveellageruitdan1bij de meerklonale vestigingsfase inalle soorten, maarookbijdezaadontwikkelingin Hypochaeris radicata meer voorverwijderdezijrozetjesdandesoortenmetinfrequentereuitstoeling. jacea verwijderde knoppenvanbloemhoofdjesdandelangerlevende H. radicata , endesoortmetfrequenteklonalevermeerdering, C. jacea H. radicata , waarschijnlijkdoordatdezesoortennietkunnenconcurrerenmethet C. jacea en vooral en en , wasdezaadproductie alleenhogerbijgroteplanten,maar S. pratensis S. pratensis C. dissectum , haddengemiddeldmeer dan1zaailingperjaar maakten echterookmeernieuwerozetten . Degrootste verliezen tradenopinde H. radicata Molinia caerulea compenseerde meervoor C. jacea S. pratensis C. dissectum , compenseerde (L.) Moench). C. dissectum en . De C. 151. Samenvatting C. werden de wordt beschreven wordt H. radicata H. radicata en C. jacea . Dit verschil tussen ruimtelijke en . Daarvoor construeerden we een S. pratensis C. jacea H. radicata H. radicata , S. pratensis ) en het meest stabiel in de meest klonale soort ( ) en het meest stabiel S. pratensis H. radicata ). Variantie decompositie technieken voor matrices toonden dat de soorten decompositie technieken ). Variantie projectie matrix model met een kruisbestuivingscyclus en zelfbestuivingscyclus. De projectie matrix model met een kruisbestuivingscyclus en van zaailingen en kleine rozetten werd verkleind in de kieming en groei zaadproductie, zaden dat in de inteeltcyclus op basis van inteeltexperimenten. Het percentage meer zaden waren inteeltcyclus kwam was een functie van de populatiegrootte: we een afnemende ingeteeld in kleinere populaties. In de simulaties vergeleken deel een planten. Hoewel slechts populatie met een populatie die toenam in aantal in de de tijd tot uitsterven van de levenscyclus werd beïnvloed, reduceerde inteelt beïnvloed door inteelt bij afnemende populatie. De toenemende populatie werd enkel door kleine initiële populatiegrootte in de simulaties. Dichtheidsafhankelijkheid op de toenemende populatie. had enkel effect limiterende habitat-grootte lage waarden van sommige componenten van de levenscyclus in slechte gebieden lage waarden van sommige met hoge waarden van andere componenten, maar gedeeltelijk gecompenseerd niet gevonden in slechte jaren. Daarentegen werd dergelijke compensatie werd wel gevonden in compensatie in slechte jaren impliceert dat temporele variatie niet en temporele variatie in de populatiedynamiek met ruimtelijke variatie zoals soms gebeurt in tijd- zomaar vervangen kan worden gelimiteerde studies. van de negatieve In hoofdstuk 6 onderzoeken we de demografische implicaties van inteelt op van inteelt, om te evalueren of gerapporteerde effecten effecten levenscyclus een significante invloed hebben op de onderdelen van de bepaalde populatiedynamiek en -overleving van jacea dezelfde ruimtelijke en temporele variatie in verschillend reageerden op verschillend in gebieden Bovendien reageerden de soorten groeiomstandigheden. populatie-groeisnelheden hadden dan in jaren waarin waarin ze lager dan gemiddelde hadden. Bij ze lagere populatie-groeisnelheden bestudeerd in hoofdstuk 5. De overleving van rozetten, vegetatieve vermeerdering en vermeerdering rozetten, vegetatieve 5. De overleving van bestudeerd in hoofdstuk kwadraten in drie tot in permanente bloemhoofdjes werden bepaald de productie van van de De seksule-reproductiefase in vier achtereenvolgende jaren. vijf populaties hoofdstuk. In totaal uit het voorgaande completeerd met data levenscyclus werd om de klassen samen projectiematrices met phenologische stelden we 42 soorten te onderzoeken in de verschillende gebieden populatiedynamiek van de drie populatie-groeisnelheden waren het laagst bij de en jaren. De berekende soort ( kortstlevende verschillende fasen van de seksuele reproductie waren in verschillende jaren en jaren waren in verschillende reproductie de seksuele fasen van verschillende hoogst in de vestigingsfase, waren het De variatiecoëfficienten onderzocht. gebieden van het zaad. het ontwikkelen en bij de bloemhoofdproductie in de fase van maar ook van de seksuele fasen in verschillende kan plaatsvinden dat selectie Dit suggereeert reproductie. van De gehele levenscyclus 152. Samenvatting mortaliteit nietopvangen. Daaromconcluderenwedat,hoeweldelevenscyclus van op hetaantal geproduceerdezaden kunnendenegatieveeffecten vanverhoogde sterftesnelheid vanindividuen omhooggaat.Depositieveeffecten opplantgrootteen competitieve soorten depopulatiegroeisnelheidsnelzal dalen zodrade plantgrootte. Demodelresultaten suggereren echterweldatookvandezemeer zaailingen kleinofgecompenseerddoor detoegenomenzaadproductie en nog nietbeïnvloedwerd,wasdenegatieveimpact vaneenlagerevestigingskans kleiner. Bijdetwee anderesoorten( de populatie-groeisnelheiddoorverhoogde mortaliteit enwerdendepopulatiessnel ( functies warenvanplanteneigenschappen. Bij detweeweinigcompetitievesoorten een ofmeercomponentenvandelevenscyclus enwaarinveelvandiecomponenten bouwden weeenhiërarchischmatrixmodel,waarinelkelementfunctiewasvan afzonderlijke levenscycluscomponentenenplanteneigenschappentemodelleren, populatiemodellen diedenatuurlijkedynamiekbeschrijven.Omeffecten van bemesting wanneeralleafzonderlijkeeffecten meegenomenwordeninde vestiging vanzaailingen,warenweookgeïnteresseerdinhettotale effect van plantgrootte enzaadproduktie alseennegatieveinvloedopdeoverlevingen populatiedynamiek. Omdatbemestingzoweleenpositieveinvloedheeft op bemesting opindividueleplantendegrootste invloedhebbenopde het bemestingsexperimentteintegrerenomonderzoekenwelkeeffecten van In hetlaatste hoofdstukwordt geprobeerddepopulatiemodellenenresultaten van lijkt tezijntotgraslandenmetzeerveelopenplekken. effect opdelokalepopulatiegroeisnelheidomdathetvestigen vanzaailingenbeperkt zijn habitat toeblijft nemen.Deverhoogdezaadproductie heeft eenverwaarloosbaar weinig-competitieve soortnoggevoeligervooruitsterven wanneerdeverrijkingvan overleving vanrozetten.Dezetrendnaarhogereroulatierozettenmaaktdeze hoge bloeikansenenbovengemiddeldevegetatieve vermeerderingmaarlagere demografische studieinnatuurgebieden.Inhetmeestproductievegebiedvondenwe bloei enverhoogdebiomassaroulatieisconsistentmetderesultaten vaneen vegetatieve rozettenooksnellerstiervenindeverrijkteplots. Ditpatroon vanmeer biomassa van groeisnelheid indeverrijkteplots tenopzichtevandenietverrijkteplots. Detotale van bloeienderozetteninhetallocatie-experimentresultaat wasvaneengrotere afhankelijk zijnvandegrootteeenrozet.Ditsuggereertdathethogepercentage kasexperiment bevestigdendatzowelbloeialsdevormingvanwortelstokken verrijkte plots. Veldwaarnemingen aanuitgegravenplanteneneenextra de eersteplantstondenwarenermeerrestenvandoderozettengevondenin verrijkingsbehandeling, warendeoverlevenderozettenverderwegvanplekwaar soort: hoeweldetotale biomassavandeplantennietbeïnvloedwerddoor allocatie-experiment vanhetderdehoofdstukwerdenverdergeanalyseerdvoordeze soorten ( De demografievandemeestklonaleenookbedreigdebestudeerde C. dissectum C. dissectum and C. dissectum H. radicata ) wordt bestudeerdinhoofdstuk7.Deresultaten vanhet verschilde niettussendetweebehandelingenomdat ) overheerstendenegatieveeffecten vanbemestenop S. pratensis and C. jacea ), waarindemortaliteit 153. Samenvatting deze perenne soorten redelijk flexibel is, hun populaties ten dode opgeschreven zijn opgeschreven ten dode is, hun populaties redelijk flexibel soorten deze perenne door toeneemt de vegetatie van de biomassaproductie waarin in graslanden waarschijnlijk planten in aantallen de afname Bovendien zal en verrijking. successie indien een verhoogde van inteelt. Echter, de negatieve effecten worden door versneld van de in een verhoogde zaadproductie reproductie resulteert allocatie naar seksuele positieve van de levenscyclus kunnen deze flexibele responsen gehele populatie, door verhoogde op landschapsschaal hebben op de metapopulatiedynamiek effecten kolonisatiekansen. Nawoord

Het bedanken van alle mensen die ik erkentelijk ben voor hun bijdrage aan dit proefschrift en voor hun vriendschap in de tijd waarin ik aan dit proefschrift werkte, vormt een dilemma: ik wil bij een ieder stil staan om anekdotes op te halen en ook niemand vergeten. Dat zou een lang verhaal worden. Waarschijnlijk wordt het meer een opsomming en zullen vele mooie verhalen overblijven voor andere gelegenheden! Veel dank gaat allereerst uit naar mijn twee promotoren. Behalve de initiator en begeleider van mijn aio-project bleek Hans de Kroon nog een aardige vent ook. De inspirerende discussies en het zeer grondige maar altijd opbouwende commentaar op mijn schrijfsels heb ik enorm gewaardeerd. Frank Berendse overzag mijn werk binnen zijn leerstoelgroep en gaf waardevol commentaar bij het plannen, analyseren en opschrijven van de experimenten. Dat mijn project binnen een groter programma viel (zie pagina 2) was een grote meerwaarde. De samenwerking met Merel Soons, Carolin Mix, Felix Knauer en Philippine Vergeer tijdens bijeenkomsten, over de e-mail en in het veld was zeer aangenaam en leerde mij veel over hun onderzoeksvelden. Ook de discussies de FLOS-meetings met Jan van Groenendael, Jana Verboom, Carla Grashof, Gerrit Heil en Joop Ouborg hielpen zeer om mijn demografische deel te plaatsen binnen de gehele ecologie van grasland soorten in versnipperde landschappen. Het begeleiden van afstudeervak-studenten was zeer plezierig en leverde een schat aan ideeën en data op. Een belangrijke bijdrage aan dit onderzoek is dan ook geleverd door Koen van der Velden, Jeroen van den Bogert, Nienke Hartemink, Ernst- Jan van Haaften, Ben Starink, Esther Graaskamp, Kim Weijtmans en Floris Breman. De medewerkers van Staatsbosbeheer en de gemeente Ermelo bleken goede gastheren met veel interesse voor het onderzoek dat ik in hun grasland reservaten heb uit mogen voeren. Onder andere Johan Wensink, Frits van Wijngeren, Jeroen Bredenbeek en Wim Klein Tijssink hebben met veel toewijding hun terreinen getoond. Voor hun informatie en enthousiasme ben ik hen zeer dankbaar. Wetenschap gedijt het best in een werkomgeving waar onderzoekers elkaar veelvuldig raadplegen en er veel gediscussieerd wordt in overleggen of gewoon bij de thee. Op TON trof ik deze belangrijke randvoorwaarden dankzij mijn collega-ecologen: Wim Braakhekke, Dick van der Hoek, Karlè Sýkora, Jan Bokdam, Elmar Veenendaal, Siny ter Borg, Monique Heijmans, Harmke van Oene, David Kleijn, Ruben Smit, Herbert Prins, Arend Brunsting, Ignas Heitkönig, Sip van Wieren, Fred de Boer, Frank van Langevelde, Pieter Ketner en Christien de Jong. In de TON-barak was het bovendien ook erg gezellig met mede-aio's Bart Fransen, Fulco Ludwig, Liesbeth Bakker, Juul Limpens, Michael Drescher, Barend van Gemerden, Irma Wynhoff, Ivo Raemakers, Jasper van Ruijven, Christiaan van der Hoeven, Jantineke Zijlstra, Eric 156. Nawoord weten hoevenappeltaarten ikondertussen indeminsta. nagekeken. Lidewij,ookbedanktvoordetekeningen inditboekje!Ikwilgeloofikniet mijn boekje.Jasper, Juul, Anne enLidewij hebbenbelangrijkedelenvandetekst idee voordeomslagkomtnatuurlijkvanMerel, enJolandaHiddinkwerktehetuitvoor maar iniedergevalvoorWieger:gefeliciteerd metje88 Wim, Hanneke,Mien, Anne, RienenWouter. Driedecemberwordt hopelijkeenfeest, alleen eenfantastische jeugdbezorgdenmaar diemeooknuzeerdierbaarblijven: Verder wilikgraag nogevenstilstaan bijdemenseninBovenkarspeldiemijniet Bogich, MelanieNorthrupandKatrionaShea). Shea lab(OlavSkarpaas, EmilyRauschert,ZeynepSezen, JessicaPeterson, Tiffany make youfeelathome.FortheirwarmwelcomeIreallywanttothankeveryoneinthe Menno, Arnoud, Maja,Jasjaennogeenheleboelanderen. baraksgrenzen somsvervagenzouditrijtjenietcompleetzijnzonderMaurice,Nansie, Pauline, Willeke,Merijn, Annelieke, Kim,Ieme,Lidewij, Sabrina enGuus.Daar Lies, Eiso,Roberta, Rene,Linus,Valentijn, Mark,Maria,Jan,Marieke,weerMatthijs, Caroline, Judith,Gijs,Esther, Jasper, Liesje,Linda,Eric,Irene,Mark, André, voorgeschoteld: Ilse,Carolien,Hadewych,Bas,Peter, Matthijs, Sjoerd,Henny, Joost, culinaire hoogstandjes dieiktijdensmijnstudie(vaakuiteigentuin)kreeg Lieve Mona,BahChi-racenDivineSixty-Ninemoetikopzijnminstbedankenvooralle anders. MijnbaraksgenotenopachtereenvolgendKleinHilton,Sad&Valleysealley 95, wonderbaarlijke plannendiesomsookuitgevoerdworden,vinddatmaarsnelergens dieren, slapenophetdakofbijkampvuurtjesenmensenomjeheenmet onkruid engroteaantallen (hoewelheternegenjaargeledentochechtmeerwaren) mogelijkheid ommevandeDroevendaalseVrijstaat lostescheuren.Leventussenhet reisje naarNijmegendemoeitewaard. Tamara vanMölkenendevrolijkemeelevendheidJoséBroekmansmaaktenelk Pierik, CorienJansen,Coco,LingWang, JuliaWenisch, JelmerWeijschedé en Heidi Huber, JosefStuefer, JohnLenssen,Werner vanEck,LiesjeMommer, Ronald Steeg, GerardBögemann,HanniedeCaluwe, Annemiek Smit-Tiekstra, EricVisser, dat ikereenhoopcollega'sbijkreeg.DeinformelewerkoverleggenmetHarryvande Heren, dankzijjullieinzetwarentweeovervolleoogstwekennogleukook! van Walsem, FransMöller, HenkvanRoekel,LouisdeNijsenMaurits Gleichman. Oeveren, MarrietteCoops, GerdaMartinenWillemienSchouten.EnnatuurlijkvanJan Grunsven enBrianMantlana.PraktischeondersteuningwaseraltijdvanHerman Geerten Hengeveld,GabrielaSchaepman,BjornRobroek, Angela Breeuwer, Royvan Verkaik, LindaJorritsma-Wienk, JasjaDekker, NicolHeuerman, Thomas Groen, Aan ditboekjezelftenslottehebbenverscheidene mensenmeegewerkt.Het Moving toanothercountrybecomesmucheasierwhenpeopledotheirbest Verhuizen naareenverlandleekmeopgegevenmomentnogdeenige Dat HansenLindanaarNijmegen'verdwenen'betekendeeigenlijkalleenmaar University Park,PA, 7 oktober2004 ste verjaardag! Eelke Jongejans Current affiliations of the co-authors

Frank Berendse is Full Professor in the Nature Conservation and Plant Ecology group of Wageningen University. [email protected]. Bornsesteeg 69, 6708 PD Wageningen, the Netherlands.

Hans de Kroon is Full Professor in the Experimental Plant Ecology group of the Radboud University of Nijmegen. [email protected]. Toernooiveld 1, 6525 ED Nijmegen, the Netherlands.

Natasha de Vere is a PhD-student at the University of Plymouth. Natasha.deVere@paigntonzoo. org.uk. Paignton Zoo Environmental Park, Totnes Road, Paignton, Devon TQ4 7EU, United Kingdom.

Nienke Hartemink is a PhD-student in the Medical Statistics and Bio-Informatics group of Leiden Univerisity. [email protected]. Centre for Informatics and Methodological Counselling (IMA) of the National Institute for Public Health and the Environment (RIVM), Anthonie van Leeuwenhoeklaan 9, 3721 MA Bilthoven, the Netherlands

Linda Jorritsma-Wienk is a PhD-student in the Experimental Plant Ecology group of the Radboud University of Nijmegen. [email protected]. Toernooiveld 1, 6525 ED Nijmegen, the Netherlands.

F. Xavier Picó is Post-Doc at the department of Molecular Genetics of Plants at the National Center for Biotechnology (CSIC). [email protected]. Campus de la Universidad Autónoma de Barcelona, 28049 Cantoblanco (Madrid), Spain.

Pedro F. Quintana-Ascencio is Senior Lecturer at the Department of Biology of the University of Central Florida. [email protected]. 4000 Central Florida Blvd., Orlando, FL 32816-2368, USA.

Merel B. Soons is Post-doc in the Landscape Ecology section of the department of Geobiology of Utrecht University. [email protected]. Sorbonnelaan 16, 3584 CA Utrecht, the Netherlands. Curriculum vitae

Eelke Jongejans was born in Bovenkarspel, the Netherlands, on November 21st 1975. After attending the Murmellius gymnasium in Alkmaar for his secondary education, he studied biology at Wageningen University from 1994 till 1999. His master's project was on seed dispersal by wind in grassland plants (supervised by Dr. Peter Schippers). As a follow-up of the lab experiments and model exercises he went to Stockholm University to study seed dispersal in the field (supervised by Dr. Anders Telenius). His Ph.D. project on the establishment, dynamics and extinction of populations of perennial plants in agricultural landscapes started in 1999 and was part of the 'Survival of Plant Species in Fragmented Landscapes’ program funded by the Netherlands Organization for Scientific Research (NWO). In this project he combined demographic field observations and experiments on biomass allocation and seedling establishment with projection matrix model exercises (supervised by Prof. Frank Berendse of the Nature Conservation and Plant Ecology group of Wageningen University and by Prof. Hans de Kroon of the Experimental Plant Ecology section of the Radboud University Nijmegen). He participated in several Ph.D. courses on life history theory, spatial modeling and population dynamics and conservation. He gave oral presentations of his research at large conferences in Basel, Prague and York and presented posters at several others. During his Ph.D. project he supervised eight students on their M.Sc. thesis and assisted in teaching several courses for undergraduate students. In 2004 he started a two-year research project on the demography and dispersal of invasive Carduus thistles at the Pennsylvania State University (at the laboratory of Dr. Katriona Shea). His main scientific interests are in processes at the plant level, and in the influence of those processes on the population dynamics of these plants, both at the local and at the regional scale. A few examples of such processes are adaptation to local environmental conditions competition, herbivory, infection by mycorrhizal fungi, plant developmental programs, resource allocation and seed dispersal. 160. Curriculum vitae van RuijvenJ,Haaften E-J,JongejansE,BaarJ&BerendseF. Effects of AMF on Jongejans E,Weijtmans K&vanRuijvenJ. The influenceofsodcuttingandsoilbiota Jongejans E,Jorritsma-Wienk LD&deKroon H.Populationmatrixanalysisofthe Jongejans E,deVere N&deKroonH.Responsestoincreased productivityinthe Papers inpreparation Soons MB,MesselinkRJH,JongejansE&HeilGW. Fragmentation andconnectivity Picó FX,JongejansE,Quitana-Ascencio PF&deKroonH.Inbreedingdepression Knauer F, GrashofC,Verboom J,SoonsMB&JongejansE.Conservinggrassland Jongejans E,SoonsMB&deKroonH.Bottlenecksandspatiotemporal variationinthe Jongejans E,deKroonH&BerendseF. Size-dependentand-independentplant Jongejans E&deKroonH.Space versus timevariationinthepopulationdynamicsof Submitted papers &JongejansE.Releasethresholdsstronglyenhancetherangeofseed Schippers P Hartemink N,JongejansE&deKroonH2004.Flexiblelifehistoryresponsestoflower 2001.Fieldexperiments onseeddispersalbywindinten Jongejans E& Telenius A 1999.Modelingseeddispersalbywindinherbaceous Jongejans E&SchippersP Publications plant diversityingrasslandsonsandysoil. on theestablishment ofseedlings effect ofnutrient enrichment onlifehistorycomponents andplanttraits. threatened herb of species-richsemi-naturalgrasslands. long-lived perennialherb. significantly influencespopulationdynamics:apopulation-basedmodelfor plants indynamiclandscapes-thematteroftemporalscale. sexual reproductionpathway ofperennialsinmeadows. reproductive allocationinresponsetosuccessionalreplacement. Succisa pratensis dispersal bywind.-UndermoderaterevisionforEcolModel. and rosettebudremovalinthreeperennialherbs.-Oikos:105:159-167. umbelliferous species(Apiaceae).-PlantEcol152:67-78. species. -Oikos87:362-372. Cirsium dissectum and twoaccompanying species. : clonalgrowthandpopulationdynamics. Cirsium dissectum and Succisa pratensis .