How does surrounding vegetation affect the course of succession:A ve-year container experiment
Lanta, Vojtech1,2 & Leps, Jan3,4
1Institute of Botany, Czech Academy of Sciences, 37005 Trebon, Czech Republic;2Section of Ecology, University of Turku, 20014 Turku, Finland;
3Department of Botany, Faculty of Science, University of South Bohemia, Branisovska 31, 37005 Ceske Budejovice,
Czech Republic; E-mail email@example.com;4Biology Centre, Institute of Entomology, Czech Academy of Sciences, Branisovska 31, 37005 Ceske Budejovice,
Czech Republic;Corresponding author; Fax1420 384 721 136; E-mail firstname.lastname@example.org
Question: How does location and time of insertion affectthe course of succession in experimental containers?
Location: Benesov nad Lipou, Ceskomoravska vrchovina(Czech-Moravian uplands), Czech Republic
Methods: We designed a 5-year container experiment inwhich plant succession started from scratch. Soil conditionswere constant and all containers were lled with homoge-neous substrate containing no propagules. We placed thecontainers in two contrasting habitats (meadow and ood-plain) under identical climatic conditions but differing insurrounding vegetations and hence seed input. New con-tainers were installed (and hence succession started) in twosubsequent years, twice in each year (spring and autumn).We assume that the individual dates would lead to differ-ences in propagule input and weather conditions.
Results: Although both year and season of successioninitiation considerably affected the initial species compo-sition, we observed a pronounced convergence within theset of containers located in each habitat. However, thesimilarity of containers initiated at the same time butlocated in different habitats decreased over the course ofsuccession. Final composition of the meadow and ood-plain containers was therefore mostly determined bypermanent seed input from their nearby neighborhood.
Conclusions: This study demonstrated that propaguleavailability is an important determinant of the course ofsuccession, and that differential seed input leads to differ-ent pathways of succession, even when all otherenvironmental conditions are equal.
Keywords: Convergence; Divergence; Habitat effect; Pro-pagule availability; Seed rain.
Nomenclature: Kubat et al. (2002).
The course of succession within any regiondiffers among individual habitat types. The threebasic determinants of the successional pathway aresite conditions, composition of propagules arrivingto the newly opened site or already present in thesoil, and climatic/weather conditions. Under naturalconditions, these three sets of factors are closely re-lated and, consequently, their effects are difcult todistinguish. Moreover, the disturbance that opens anew area and starts succession in that area onlyrarely completely destroys that habitat and, even inthe case of a major disturbance, succession does notstart from complete scratch (so that biotic interac-tions play an important role from the verybeginning). In the case of secondary succession, soilbiota are always present and can affect the course ofsuccession.
It is well known that seedling establishment is oneof the most sensitive phases of the plant life cycle(Fischer & Matthies 1998; Isselstein et al. 2002) andthat there is large variability in seedling establishmentamong individual years (e.g. Spackova & Leps 2004;Pakeman & Small 2005). The regeneration niche the-ory (Grubb 1977), that seedling establishmentrequirements differ among individual species, meansthat different availability of safe sites (Harper et al.1965) could determine establishment of individualspecies. Similarly, the time when succession startsmight, to a large extent, determine the whole course ofsuccession. Both weather conditions and seed rainchange between seasons over the course of a year andbetween individual years. Although differences be-tween seasons (in both weather and seed rain) aremore or less predictable, the differences between in-dividual years are not.
Journal of Vegetation Science 20: 686694, 2009& 2009 International Association for Vegetation Science
Classical successional theory (Clements 1916)suggests that community composition shouldconverge toward climax, as determined by environ-mental conditions (the monoclimax theory). Incontrast, Eglers (1952) Initial Floristic Composi-tion concept predicts that the course of succession is,to a large extent, determined by the species that wereinitially dominant; thus we would expect divergence.Leps & Rejmanek (1991) predict that convergent ordivergent behaviour can be determined by thevariability in habitats observed with respect to fac-tors limiting the distribution of early and latesuccessional species, which are differentially sensi-tive to environmental factors. If the habitats arehighly variable in factors to which the early succes-sional species are sensitive (typically the type ofsuccession initiating disturbance) and rather homo-geneous with respect to factors to which the latesuccessional species are sensitive (typically soil orclimatic conditions), we expect convergence. We ex-pect divergence under the opposite scenario.Examples of both types of development have beenrepeatedly reported from natural communities(Matthews 1979; Barbour & Minnich 1990; Sjors1990; Nilsson & Wilson 1991; Inouye & Tilman1995; del Moral 2007); however, we are not aware ofan experimental study able to disentangle the effectof site conditions, weather conditions and differ-ences in seed rain (propagule input).
We cannot manipulate climate conditions orkeep them constant from year to year, but in ex-periments on a small spatial scale (e.g. in plasticcontainers), we are able to keep soil conditions con-stant. We can change seed input simply by sowingseeds, but we do not have a clear idea how thequantity or composition of seed rain might differbetween two relatively close sites or among in-dividual seasons and years, i.e. what is the real seedinput. Consequently, we designed a container ex-periment in which we started the succession fromscratch. We placed the containers in two contrastinghabitats under identical climatic conditions. Thecontainers in the two habitats were thus subjected totwo differing but realistic types of seed rain. We in-serted sets of containers in two seasons in each oftwo subsequent years. From this we hoped to de-monstrate that the effects of seed rain are notaffected by variations in soil conditions or seasonaland annual variations during the course of succes-sion. The effect of insertion time includes both theeffect of weather in a season/year and varying seedrain between seasons/years.
Our experiment enabled the study of experi-mental succession over 5 years where initial soil
conditions were kept constant, and any variabilitywas caused only by differences in propagule inputand weather conditions. Our aim was to separatethese two effects. By following the course of succes-sion for several years, we can examine which factorsdetermine the initial variability. Specically, we ex-amined:
(1) How the course of succession was affected by thedifferential availability of propagules in differenthabitats.
(2) How the course of succession was affected by thetime of succession initiation (in various seasonsand various years).
(3) Whether the initial differences increase or de-crease during succession, i.e. whether and underwhich conditions the course of succession isconvergent or divergent.
The experiment was conducted within an or-ganic farm at Benesov nad Lipou, a site in thesoutheastern part of the Czech Republic, in theCeskomoravska vrchovina (Czech-Moravian up-lands, 491920N, 151000E, 665m a.s.l.). This area hasa temperate climate, with a mean annual tempera-ture of 6.71C and annual precipitation of 759mm(Cernovice meteorological station).
The experiment was established in two habitatswithin the farm. The rst (here called meadow)was in a strip that formed the border between twoarable elds and an extensively grazed pasture. Thisstrip was 3-m wide and faced northwest-southeast.Young trees of Sorbus aucuparia, Salix caprea andother woody species were planted there at regulardistances of 4-5m. The herbaceous vegetation be-tween the trees formed permanent grasslandcontaining several ruderals or arable weed species.Common species included the grasses Agropyron re-pens, Poa trivialis, Phleum pratense, Dactylisglomerata and the forbs Taraxacum sect. Ruderalia,Trifolium repens, Artemisia vulgaris and Veronicaarvensis. The second habitat (here called ood-plain) was located in the oodplain of theVcelnicka rivulet. (However, oods are rare and theplot was not ooded during our experiment.) Plantstypical of moderately wet meadows and pasturesprevailed (grasses Phalaris arundinacea, Festuca ru-bra, Agrostis tenuis, Holcus lanatus and the forbs
- How does surrounding vegetation affect the course of succession - 687
Ranunculus acris, Myosotis nemorosa, Galium uligi-nosum, Anthriscus sylvestris). Both habitats wereunmanaged during the course of the experiment.The whole area of meadow is rather at with a slopeof o51, the oodplain is completely at. The twosites were about 500m from each other and sepa-rated by a small wood; the difference in elevationwaso10m. Consequently, the habitats share nearlythe same weather conditions. (Originally, the ex-periment included a third habitat type: a newlyabandoned eld, but all the containers were stolenand we therefore had to restrict ourselves to just twohabitat types.)
In each habitat type, we selected one location(locations within habitats are not replicated). Theexperiment does not intend to represent the succes-sional course in the respective habitats ourintention is only to demonstrate the differences insuccessional pathways due to differential seed input.Consequently, we selected two locations that areclose enough not to differ in climatic conditions, butfar enough apart to receive different seed rain.
Circular plastic containers (0.55m dia-meter0.25m deep) lled with garden substrate(without any seeds) containing active humus (pH inwater suspension between 5.5 and 7.0; 5% of parti-cles bigger than 20mm) were inserted in the eld. Ateach starting time, 10 containers (ve with and vewithout perforated bases) were inserted into the soilwithin each of the two habitats, dated and coded:April 2002 (T1), September 2002 (T2), April 2003(T3) and September 2003 (T4). As the water regimediffered between the two sites, containers with per-forated bases enabled vertical water movement toprevent desiccation and introduce differences in soilmoisture. This was insurance in the case of an ex-tremely dry period that could have led to completedrying out of containers with non-perforated bases.Fortunately, no such conditions occurred. Further,in the analyses, the effect of perforation was alwaysnegligible and non-signicant.
Containers were inserted into the soil at regularintervals (i.e. 2m) with the upper margins of a con-tainer being level with the soil surface. In total, 80containers were inserted in the two habitats (20 per-forated and 20 non-perforated in each habitat) overthe rst 2 years of the experiment. The use of con-tainers helped prevent penetration of rhizomes andstolons of surrounding vegetation into the gardensoil. In a limited number of cases (ca. 10) in theoodplain, we found aboveground stolons of plants
tended to spread into the containers. These were re-moved from the closest neighbour to the containers.Consequently, the succession in each container re-ected establishment only of plants that hadgerminated from the seed rain.
The cover of all present plant species was re-corded twice per year, in the spring and late summer,in ve subsequent years, 2002-2006.
The changes in total species composition wereanalysed using multivariate ordination methods(Canoco for Windows, ter Braak & Smilauer 2002).Data on species richness were analysed using ANO-VA models. Cover data from each time census wereused as response data for ordination, yielding a ma-trix of 600 samples and 88 plant species. Values ofspecies cover (in %) were log(X11)-transformedprior to analyses. Each vegetation record was char-acterized by the location of the container (meadowor oodplain), year of container insertion, season ofcontainer insertion (spring or summer), year of datarecording, and season of data recording. All of thesefactors were considered as environmental variables(in Canoco terminology).
Data were rst subjected to detrended corre-spondence analysis (DCA) in order to assess theoverall variation pattern in species composition,with the environmental variables projected passivelyto the ordination plane. The interaction among fac-tors, time of container insertion (T; four dummyvariables)position of container (P; two dummyvariables)census for data recording (R; sevendummy variables) was passively projected to the or-dination plain (yielding 4275 56 centroids).Note that the ordination is solely based on speciesdata. The passive projection of interaction displays,where the centroids of containers inserted in in-dividual habitats and at individual times are atindividual census dates in ordination space, was de-ned by species composition. Successional trendsare displayed as lines connecting the time series ofthe centroids of the same position and insertiontime. Canonical correspondence analysis (CCA) wasthen used to evaluate the species-environment re-lationship. Environmental variables were subjectedto forward selection (FS), mainly to see the sequenceof contributions of individual variables to explana-tion of species composition. The amount ofvariability explained by individual variables wascalculated by dividing corresponding eigenvalues bytotal inertia. The signicance of each variable wasevaluated using the Monte Carlo permutation test
688 Lanta, V. & Leps, J.
(499 permutations). Four related CCA analyseswere used: CCA 1 for the whole dataset, CCA 2 for1-year-old containers (recorded in spring of the yearfollowing insertion, i.e. the set contains records fromvarious years), CCA 3 for the 3-year-old containers(similarly to the previous, but recorded 2 years la-ter), and CCA 4 for data from the nal census(summer 2006). In all analyses, the Position, Year ofcontainer insertion and Season of container inser-tion were used (all binary variables); in addition, inCCA1, the Year and Season of data recording wereused. Year was in this case used as quantitativevariable and explains the successional trend. InCCA 2 and CCA 3, containers were of the same age,but recorded in various years, so the differencesmight be caused either by habitat, or by year of in-sertion, which was not confounded by thesuccessional age, but might be affected by weather inthe year of recording. In CCA 4, time of insertionwas confounded with successional age of the con-tainer, but all the samples were recorded at the sametime.
Changes in species richness were analysed byrepeated measures ANOVA. Because the analysisassumes the same number of observations for all theobserved units, separate analyses were done for eachtime of insertion.
The effect of container perforation was not de-tected in any analyses; therefore, we pooled theperforated and non-perforated container data andthis factor was disregarded in further analyses.
DCA analysis of all the recorded releves withpassively projected positions of centroids of treat-ments at individual census times (Fig. 1) revealedseveral interesting patterns of species compositiondynamics. First, within each habitat, all the startingpoints converge towards a similar species composi-tion. This means that, in the last census, the speciescomposition was determined mainly by habitat, andtime of insertion had a minor effect. Second, thestarting points of succession differ considerablyamong insertion dates within the habitats but areremarkably similar between habitats, particularlyfor containers inserted in spring in both years (T1and T3). The ordination diagram suggests strongconvergence within habitats where the containerswere inserted from remarkably different startingpositions at various times of insertion. However, di-vergence for containers inserted at the same time
was rather similar at the beginning, but was nallydifferentiated according to habitat.
The development of vegetation on individualinsertion dates started with similar plant composi-tion in both meadow and oodplain containers(start of lines is positioned at a similar level; Fig. 1).For example, in both meadow and oodplain T1containers, succession starts with annual ruderals(Echinochloa crus-gali and Chenopodium album),while in T3 containers it starts with a high input ofTaraxacum diaspores, forming a nearly mono-dominant community (many of these seedlings laterdisappeared). Over the course of succession, speciesof the respective habitats gradually prevailed. Mea-dow successional seres (left part of Fig. 1) mainlycontained ruderal annual (e.g. Cerastium holos-teoides) and ruderal perennial plants, such areTaraxacum sect. Ruderalia, Phleum pratense, Poa
Fig. 1. The DCA for the complete dataset of 600 relevesand 88 species. The biplot displays species with highestweight in the analysis and centroids of environmentalvariable interactions. The recording time series for con-tainers T1 (inserted in April 2002), T2 (September 2002),T3 (April 2003) and T4 (September 2003) are connected bybroken lines for meadow and by solid lines for oodplain.The starting point is labelled by the corresponding time ofinsertion, and terminated by an open circle, which denotesthe position in summer 2006. Species abbreviations:AcePr, Acetosa pratensis; AgrRe, Agropyron repens;AloPr, Alopecurus pratensis; AngSy, Angelica sylvestris;ArtVu, Artemisia vulgaris; CerHo, Cerastium holosteoides;CheAl, Chenopodium album; CirPa, Cirsium palustre;EchCr, Echinochloa crus-galli; EpiMo, Epilobium mon-tanum; FesPr, Festuca pratensis; FesRu, Festuca rubra;HolLa, Holcus lanatus; HypMa, Hypericum maculatum;PhlPr, Phleum pratense; PoaTr, Poa trivialis; RanAc, Ra-nunculus acris; SteGr, Stellaria graminea; TarOf, Taxacumsect. Ruderalia; UrtDi, Urtica dioica; VerCh, Veronicachamaedrys.
- How does surrounding vegetation affect the course of succession - 689
trivialis and Artemisia vulgaris. Floodplain succes-sional seres (right part of Fig. 1) developed a highcover of grasses and forbs, such as Holcus lanatus,Festuca rubra, Alopecurus pratensis, Cirsium pa-lustre, Angelica sylvestris and Ranunculus acris,which are species of moderately moist grassland ha-bitats.
In CCA 1 (complete dataset), the ve environ-mental variables explained 17.0% of total variabilityin species data (note that in constrained ordinations,the amount of explained variability also reects thereduction in dimensionality, here from 88 speciesto a rather limited number of constrained axes).Forward selection (FS) showed that the best en-vironmental variable explaining plant compositionwas habitat (oodplain versus meadow), followedby year of data recording and season of datarecording (Table 1); the last two were partially con-founded by the successional age and seasonaldifferences in vegetation development, respectively.The year and season of container insertion,although explaining the smallest part of the varia-bility, still exhibited statistically signicant effects.In the 1-year-old containers, the effect of habitat,season of container insertion and year of containerinsertion had roughly equal effects, whereas in thethird year, the effect of habitat was dominant, ex-plaining much more than the year and season of
insertion (the last being totally insignicant). Simi-lar results were also found for the last census, whenthe effect of time of insertion included successionalage, showing increasing importance of habitat withtime and decreasing importance of year and seasonof container insertion. This conrmed the resultsfrom DCA ordination (Fig. 1) the importance ofhabitat increased with successional age, whereaswithin a habitat, species composition converged re-gardless of initial conditions (here signied by timeof insertion).
The temporal changes in species richness wererather idiosyncratic and differed among individualinsertion times (Fig. 2, Table 2). If the main effect ofposition was signicant, then the oodplain wasmore species-rich than the meadow. Interestingly, incontainers inserted in the rst year (T1 and T2), thesecond census revealed a sharp drop in species rich-ness (for both T1 and T2 in the oodplain and T2 inmeadow). This was caused by successful establish-ment of a large number of seedlings of species likeTaraxacum sect. Ruderalia, Poa trivialis or Artemi-sia vulgaris, which grow rapidly and out-competemany annual ruderals. In the oodplain, these rud-erals were then overgrown by plants of moistgrassland. Starting from the third census, the num-ber of species began to increase in the oodplain, butno such decrease was observed in containers insertedin T3 and T4. From the fourth season onwards,species richness started to decrease, and with theexception of containers inserted in T4, all the othercontainers exhibited a sharp decline in species rich-ness during the nal census in summer 2006(Fig. 2).
Our study demonstrated that diaspore avail-ability is an important determinant of the course ofsuccession, and differential seed input causes differ-ent pathways of succession, even when all otherenvironmental conditions are equal. Although wetried to keep exactly the same starting conditions inthe oodplain and in the meadow and constantstarting conditions over all insertion periods, it isdifcult to guarantee complete control in a eld ex-periment. Nevertheless, the containers were lledfrom the same bulk substrate and their assignmentto habitats was random; consequently, the uni-formity of initial conditions between habitats can betrusted. The same bulk substrate was used for allfour initiations so we cannot exclude slight chan-ges in stored substrate, for example in viability of
Table 1. Results of CCA with forward selection for fourdatasets: CCA 1 for the whole dataset, CCA 2 for 1-yearold containers, CCA 3 for the 3-years-old containers andCCA 4 for data from the last census (summer 2006).
CCA 1 (all ve variables explained 17.0%)Position 9.4 22.48 0.002Year of data recording 4.1 10.05 0.002Season of data recording 1.3 3.14 0.002Year of container insertion 1.3 3.07 0.002Season of containerinsertion
0.9 2.12 0.002
CCA 2 (all three variables explained 12.7%)Position 4.4 2.85 0.002Year of container insertion 4.3 2.83 0.002Season of containerinsertion
4.0 2.73 0.002
CCA 3 (all three variables explained 8.3%)Position 5.8 7.10 0.002Year of container insertion 1.6 1.85 0.002Season of containerinsertion
0.9 1.08 0.298
CCA 4 (all three variables explained 11.2%)Position 8.0 6.30 0.002Season of containerinsertion
1.6 1.32 0.026
Year of container insertion 1.6 1.31 0.034
690 Lanta, V. & Leps, J.
fungal spores that might have some effect. Never-theless, we believe that these effects are negligible incomparison with differences in weather and seed in-put at the time of insertion. We are also condentthat climate did not differ between sites a distanceof 500m in a rather at landscape makes climaticdifferences unlikely. The experiment was located onan organic farm, which does not use any chemicals;consequently, it is also unlikely that the containersreceived additional nutrient or pesticide input fromthe nearby arable elds. On the other hand, we didnot want to remove vegetation around the contain-ers (to retain a realistic diaspore input, although afew stolons with a tendency to enter containers wereremoved) so the surrounding vegetation might havea slight effect in succession in containers throughshading. However, the mean height of the vegeta-tion was roughly the same in both habitats (ca. 0.5mat the height of the vegetation season). Therefore,we can be reasonably condent that the differencesbetween habitats are caused by different seed inputto containers, and differences between insertiontimes are caused by differences in weather, seasonand related differences in seed input.
Convergence and divergence
It is generally accepted that diaspore input de-clines with increasing distance from the diasporesource (Salonen 1987; Salonen & Setala 1992;Bakker et al. 1998; Buisson et al. 2006). Final com-position of meadow and oodplain containerswas therefore determined mostly by permanent seedinput from their nearby neighbours. Our ndingsthus conrmed studies focusing on succession overa broad landscape scale (reviewed in Prach &Rehounkova 2006) the importance of nearby pro-pagule availability. The species pool of availablediaspores is indeed one of the basic determinants ofplant community composition (Pakeman et al. 1998;Zobel et al. 2000; Foster et al. 2004; Zobel & Kala-mees 2005).
Seed input, however, changes not only in space(in our case between the two compared habitats),but also in time, among seasons and years (Kala-mees & Zobel 2002; Pakeman & Small 2005; Ejrnaeset al. 2006). Even though the order in which thediaspores of individual species entered the systemdiffered according to the date of container insertion,most of the species from the surrounding area nallyentered the system during the period of the experi-ment, and the same species composition prevailed inall containers in each habitat. This suggested thatthe competitive abilities of individual species, rather
Fig. 2. Number of species per container T1-T4 recordedduring the experiment. Filled circles represent oodplainand open circles represent meadow containers. Circles de-note means and whiskers are standard error of means.
- How does surrounding vegetation affect the course of succession - 691
than order of entry, determined nal communitycomposition. As a consequence, we observed pro-nounced and relatively fast convergence withinhabitats, i.e. in individual habitats the similarity ofcommunities was very high at the end of the studyperiod. In contrast, the effect of time of successioninitiation (both season and year) was very pro-nounced at the initial stages, but decreased withtime. This implies that the initial oristic composi-tion was, in our case, less important than suggestedby Eglers (1952) initial oristic composition con-cept. Whereas the differences between seasons areclearly connected to the predictable course ofweather during a year (and corresponding phenol-ogy of individual species seeWeber 2001; Schwartz2003), the differences between years are purely ran-dom variability (both weather and correspondinglyaffected seed input). Their effects on plant speciescomposition are roughly comparable; in most ana-lyses the effect of a year is slightly more pronouncedthan the effect of a season.
Within the set of containers located in a singlesite, we observed pronounced convergence; how-ever, the similarity of containers initiated at thesame time decreased during succession according totheir location, and we observed a clear divergence.This corresponds to the reasoning of Leps & Re-jmanek (1991) on the effects of differentialsensitivity of early and late successional species onour perception of convergence and divergence insuccession. (However, in our case, the plants shouldbe considered early and mid-successional at best, asthe oldest observed stage was 5 years.) In particular,the early colonist Taraxacum sect. Ruderalia, withits wind-dispersed seeds, was able to colonize thecontainers regardless of location. Later, however,this species was replaced by species with heavierseeds, dispersed from sources typical of eachhabitat.
Vegetation development proceeded from spe-cies composition differentiated at the startaccording to all the studied factors, to communitiesdetermined by habitat only (nal state). Manyplants composing the surrounding vegetation, in-
cluding perennials typical for each habitat, hadsufcient time to establish and gradually out-com-pete the early-germinated plants, represented byruderals requiring high light intensities at the time ofgermination (often gap-colonizers; Fenner 1978).However, all our results are based on a small-scalecontainer experiment, where the surrounding vege-tation is very close; succession on larger plots, wherethe diaspore source is not so close might differ.
Replacement of species during succession
The course of succession was mostly driven bydifferential establishment, selecting the species ac-cording to their propagule availability at a giventime and location, and subsequent competitive in-teractions leading to displacement of some species.In meadow containers, Taraxacum sect. Ruderaliareached the highest cover at very early phases,forming dense cohorts of seedlings; these were laterout-competed by Artemisia vulgaris, a tall herb, ableto shade out the low rosettes of Taraxacum. Bothspecies belong to Asteraceae, are common in neigh-bouring vegetation, and easily recruited from wind-dispersed seeds (Taraxacum has massive seed rain inspring, Artemisia in autumn). Leguminous species,Vicia sp. and Trifolium pratense, both of which haveheavy seeds, appeared later. Species replacementwas also recorded in oodplain containers, wheresuccession started with abundant populations ofthree Ranunculus species (R. acris, R. repens and R.auricomus), and a dense cover of the creeping forbGalium uliginosum and umbellifer Angelica sylves-tris. In later stages these were mainly replaced byrapidly spreading grasses (Deschampsia caespitosa,Festuca pratensis and Holcus lanatus), leguminousLathyrus pratensis (again characterized by heavyseeds), and successfully established seedlings of theshrubs Rubus idaeus and Salix fragilis.
Changes in species richness
Development of the number of species was ra-ther idiosyncratic, differing according to the time
Table 2. Results of repeated measures ANOVA for number of species per container. This was calculated separately for setsof containers inserted on individual dates: T1 (inserted in April 2002), T2 (September 2002), T3 (April 2003) and T4(September 2003).
T1 T2 T3 T4
df F df F df F df F
Position 1,18 17.46 1,18 5.98 1,18 3.93 n.s. 1,18 1.20 n.s.Time 8,144 5.38 7,126 10.54 6,108 30.11 5,90 9.77
Positiontime 8,144 3.38 7,126 0.56 n.s. 6,108 2.25 5,90 1.56 n.s.
692 Lanta, V. & Leps, J.
when the containers were inserted (cf. Crawley1989). Nevertheless, on average, the number of spe-cies was higher in the oodplain. This could havebeen caused by either a higher species pool, or by theabsence of strong dominants in the oodplain.Containers inserted in the rst season experienced astrong decrease in species richness by the time of thesecond census. Since the second census was differentfor containers inserted in spring and summer of therst year, we deduced that the decrease was indeedcaused by strong competition and not by adverseweather conditions. Nevertheless, a decrease wasnot observed in containers inserted in the secondyear, which suggests that the course of successionwas affected by a rather complex interplay of speciesinteraction and weather conditions in individualyears. In all containers, however, the number ofspecies decreased toward the end of the experiment,suggesting the prevailing effect of interspecic com-petition (Connell & Slatyer 1977).
Importance of species identity
The course of succession was signicantly af-fected by some species, establishment of which inindividual containers considerably affected furtherdevelopment. In the oodplain containers, we ob-served a pronounced effect of Cirsium palustre,which created a rosette of leaves, up to 0.5m dia-meter, that was able to suppress all neighbouringplants by shading. This suggests that light avail-ability often controls species abundance throughoutsuccession (Foster & Gross 1998). However, as thisspecies is monocarpic, it required further dis-turbance to establish itself, and consequently onlyrarely did it retain permanent dominance. In mea-dow containers, dominants were often clonal grasses(e.g. Agropyron repens) with vegetative organs al-lowing them to take advantage of nutrientavailability (de Kroon & Groenendael 1997). Thissuggests that the identity of the species, through aunique combination of traits, has a fundamental ef-fect on the course of succession. This, in turn,further stresses the importance of the available dia-spore species composition for the successionaldevelopment.
Our experiment conrmed that seed input (de-termined by the neighbouring vegetation) is thebasic determinant of the course of succession whensoil conditions are kept constant. Weather condi-tions at the time of succession initiation did affectthe course of succession, particularly at the start;however, with successional age, this effect almostdisappeared.
Acknowledgements. We are indebted to Miroslav Srutek
for permission to carry out the experiment on his farm and
Jan Jedlicka for help with inserting containers. This study
was supported by the EU project TLinks (contract num-
ber EVK2-CT-2001-00123) and grants AV0Z60050516,
MSMT 6007665801 and LC 06073.
Bakker, J.P., Elzinga, J.A. & deVries, Y. 1998. Effects of
long-term cutting in a grassland system: perspectives
for restoration of plant communities on nutrient-poor
soils. Applied Vegetation Science 1: 107120.
Barbour, M.G. & Minnich, R.A. 1990. The myth of
chaparral convergence. Israel Journal of Botany 39:
Buisson, E., Dutoit, T., Torre, F., Romermann, C. &
Poschlod, P. 2006. The implications of seed rain and
seed bank patterns for plant succession at the edges of
abandoned elds in Mediterranean landscapes.
Agriculture Ecosystems and Environment 115: 614.
Clements, F.E. 1916. Plant succession: an analysis of the
development of vegetation. Carnegie Institution of
Washington, Washington, DC, US.
Connell, J.H. & Slatyer, R.O. 1977. Mechanisms of
succession in natural communities and their role in
community stability and organization. American
Naturalist 111: 11191144.
Crawley, M.J. 1989. Chance and timing in biological
invasions. In: Drake, J.A., Mooney, H.A., di Castri, F.,
Groves, R.H. & Kruger, F.J. (eds.) Biological invasions:
a global perspective. pp. 407423. John Wiley and Sons,
de Kroon, H. & Groenendael, J. 1997. The ecology and
evolution of clonal plants. Blackhuys, Leiden, NL.
del Moral, R. 2007. Limits to convergence of vegetation
during early primary succession. Journal of Vegetation
Science 18: 479448.
Egler, F. 1952. Vegetation science concepts. I. Initial
oristic composition: a factor in old-eld vegetation
development. Vegetatio 4: 412417.
Ejrnaes, R., Bruun, H.H. & Graae, B.J. 2006. Community
assembly in experimental grassland: suitable
environment or timely arrival? Ecology 87: 12251233.
Fenner, M. 1978. A comparison of the abilities of
colonizers and closed-turf species to establish from
seed in articial swards. Journal of Ecology 66: 953
Fischer, M. & Matthies, D. 1998. Experimental
demography of the rare Gentianella germanica: seed
bank formation and microsite effects on seedling
establishment. Ecography 21: 269278.
Foster, B.L. & Gross, K.L. 1998. Species richness in a
successional grassland: effects of nitrogen enrichment
and plant litter. Ecology 79: 25932602.
Foster, B.L., Dickson, T.L., Murphy, C.A., Karel, I.S. &
Smith, V.H. 2004. Propagule pools mediate
- How does surrounding vegetation affect the course of succession - 693
community assembly and diversity-ecosystem
regulation along a grassland productivity gradient:
the role of propagule availability. Journal of Ecology
Grubb, P.J. 1977. The maintenance of species richness in
plant communities: the importance of the regeneration
niche. Biological Reviews of the Cambridge
Philosophical Society 52: 107145.
Harper, J.L., Williams, J.T. & Sagar, G.R. 1965. The
behavior of seeds in soil. 1. The heterogeneity of soil
surfaces and its role in determining the establishment
of plants from seed. Journal of Ecology 53: 273286.
Inouye, R.S. & Tilman, D. 1995. Convergence and
divergence of old-eld vegetation after 11 years of
nitrogen addition. Ecology 76: 18721887.
Isselstein, J., Tallowin, J.R.B. & Smith, R.E.N. 2002.
Factors affecting seed germination and seedling
establishment of fen-meadow species. Restoration
Ecology 10: 173184.
Kalamees, R. & Zobel, M. 2002. The role of the seed bank
in gap regeneration in a calcareous grassland
community. Ecology 83: 10171025.
Kubat, K., Hrouda, L., Chrtek, J.jun., Kaplan, Z.,
Kirschner, J. & Stepanek, J. (eds.) 2002. Kle` ke
kvetene Ceske republiky [key to the ora of the Czech
Republic]. Academia, Praha, CZ.
Leps, J. & Rejmanek, M. 1991. Convergence or
divergence: what should we expect from vegetation
succession? Oikos 62: 261264.
Matthews, J.A. 1979. Refutation of convergence in a
vegetation succession. Naturwissenschaften 66: 4749.
Nilsson, C. & Wilson, S.D. 1991. Convergence in plant
community structure along disparate gradients: are
lakeshores inverted mountainsides? American
Naturalist 137: 774790.
Pakeman, R.J. & Small, J.L. 2005. The role of the seed
bank, seed rain and the timing of disturbance in gap
regeneration. Journal of Vegetation Science 16: 121130.
Pakeman, R.J., Attwood, J.P. & Engelen, J. 1998. Sources
of plants colonizing experimentally disturbed patches
in an acidic grassland in eastern England. Journal of
Ecology 86: 10321041.
Prach, K. & Rehounkova, K. 2006. Vegetation succession
over broad geographical scales: which factors
determine the patterns? Preslia 78: 469480.
Salonen, V. 1987. Relationship between the seed rain and
the establishment of vegetation in two areas
abandoned after peat harvesting. Holarctic Ecology
Salonen, V. & Setala, H. 1992. Plant colonization of bare
peat surface relative importance of seed availability
and soil. Ecography 15: 199204.
Schwartz, M.D. 2003. Phenology: an integrative
environmental science. Kluwer Academic Publishers,
Sjors, H. 1990. Divergent succession in mires, a
comparative study. Aquilo Series Botanica 28: 67
Spackova, I. & Leps, J. 2004. Variability of seedling
recruitment under dominant, moss, and litter removal
over four years. Folia Geobotanica 29: 4155.
ter Braak, C.J.F. & Smilauer, P. 2002. CANOCO
reference manual and CanoDraw for Windows
users guide: software for canonical community
ordination (version 4.5). Microcomputer Power,
Ithaca, NY, US.
Weber, K.T. 2001. A method to incorporate phenology
into land cover change analysis. Journal of Range
Management 54: A1A7.
Zobel, M. & Kalamees, R. 2005. Diversity and dispersal
can the link be approached experimentally? Folia
Geobotanica 40: 312.
Zobel, M., Otsus, M., Liira, J., Moora, M. & Mols, T.
2000. Is small-scale species-richness limited by seed
availability or microsite availability? Ecology 81:
Received 13 November 2008;
Accepted 13 February 2009.
Co-ordinating Editor: M. Zobel.
694 Lanta, V. & Leps, J.