Graphic Art

Biology, chance, or history? The predictable reassembly of temperate grassland communities

Biology, chance, or history? The predictable reassembly of temperate grassland communities
of 14
All materials on our website are shared by users. If you have any questions about copyright issues, please report us to resolve them. We are always happy to assist you.
Related Documents
  Ecology , 91(2), 2010, pp. 408–421   2010 by the Ecological Society of America Biology, chance, or history? The predictable reassemblyof temperate grassland communities J ANA  S. P ETERMANN , 1,4 A LEXANDER  J. F. F ERGUS , 1 C HRISTIANE  R OSCHER , 2 L INDSAY  A. T URNBULL , 1 A LEXANDRA  W EIGELT , 3 AND  B ERNHARD  S CHMID 1 1 Institute of Environmental Sciences, University of Zurich, Winterthurerstrasse 190, 8057 Zurich, Switzerland  2 Max Planck Institute for Biogeochemistry, POB 100164, 07701 Jena, Germany 3 Institute of Ecology, University of Jena, Dornburger Strasse 159, 07743 Jena, Germany Abstract.  Many studies have examined invasion resistance in plant communities, but fewhave explored the mechanisms of invasion and how subsequent community reassembly affectscommunity functioning. Using natural dispersal and deliberate seed addition into grasslandcommunities with different compositional and richness histories, we show that invadersestablish in a nonrandom manner due to negative effects of resident functional groups oninvading species from the same functional group. Invaders hence complement communitieswith srcinally low richness levels. Consequently, communities converge toward similar levelsof species richness, high functional richness, and evenness, but not always maximumproductivity. Invasion processes are faster but qualitatively similar when the effect of chance,in the form of dispersal stochasticity, is reduced by seed addition. Thus, dispersal limitationmay influence community assembly, but it does not override functionally predictable assemblymechanisms. Some of the most productive communities prior to invasion are unstable in theface of invasion, leading to decreased productivity following invasion. We suggest thatinvasion into such communities occurs possibly because a pathogen-free niche is availablerather than a resource niche. Thus, pathogens in addition to resource niches may be importantbiological drivers of community assembly. Key words: biodiversity–productivity relationship; community stability; dispersal limitation; ecosystem functioning; invasion resistance; invasiveness; negative feedback; neutral theory; nonrandom invasion; speciesrichness. I NTRODUCTION Biology, chance, and history must all play some rolein community assembly. For example, in order tosuccessfully establish in a new community, a potentialinvader must first arrive, and dispersal is an inherentlystochastic process. However, the relative importance of dispersal limitation and historical contingency vs.deterministic biological interactions is still hotly debated(e.g., Drake 1991, Hubbell 2001, Chase 2003, Fargioneet al. 2003, Turnbull et al. 2005 a ,  b ).The first explanations as to why certain species wereable to successfully invade new communities werecertainly deterministic in nature and focussed mainlyon the biology of the invaders (see, e.g., Elton 1958). Forinstance, some species appeared to be more successfulthan others at dispersing to new sites, at entering newcommunities, or at reaching high population sizes andsuppressing residents (Crawley 1986, Drake et al. 1989).This observation led to a focus on the properties of thesespecies and their associated ‘‘invasiveness’’ (Baker 1967,Sutherland 2004, Richardson and Pysek 2006).Conversely, invasion success might be related to thebiology of the invaded or resident community; forexample, more diverse communities tend to be moreinvasion resistant (Crawley 1987, Burke and Grime1996). This may occur because particular residentspecies or functional groups provide invasion resistance(Crawley et al. 1999, Levine and D’Antonio 1999,Symstad 2000, Hector et al. 2001, Dukes 2002, vanRuijven et al. 2003, Fargione and Tilman 2005) andthese species or functional groups are more likely to befound in higher-diversity communities. The importanceof particular species for community invasion resistanceis therefore analogous to a sampling effect in biodiver-sity–productivity relationships (Hector et al. 2001,Wardle 2001).Finally, interactions between the invader and theinvaded community might be key to understandinginvasion success, analogous to a complementarity effectin biodiversity–productivity relationships (Hector et al.2001, Fargione et al. 2003). In this case, not only theidentity of the invader or the composition of the residentcommunity, but the match between invaders andcommunities plus the respective species abundances Manuscript received 14 December 2008; revised 29 April2009; accepted 15 May 2009. Corresponding Editor: J. M.Levine. 4 Present address: Department of Zoology, University of British Columbia, 6270 University Boulevard, Vancouver,British Columbia V6T1Z4 Canada.E-mail: Peterman@uwinst.uzh.ch408  would be most important in determining the outcome of invasion (e.g., Fargione et al. 2003, Turnbull et al.2005 b , Strauss et al. 2006). Thus, just like speciescoexistence in established communities, invasion andcommunity reassembly would be controlled by density-dependent stabilizing mechanisms (Chesson 2000).These stabilizing mechanisms would be expected tofacilitate invasion by species or functional groups thatare most different from abundant residents (MacArthurand Levins 1967, Abrams 1983, Emery 2007).The most well-known and studied complementaritymechanism within temperate communities is based onresource-use niches (e.g., Harpole and Tilman 2007),which could lead to preferential invasion by species withcomplementary resource requirements compared withthe residents (Fargione et al. 2003, Questad and Foster2008). Increased invasion resistance of species-richcommunities could, according to this hypothesis, beattributed to the lack of unconsumed resources, as someinvasion studies have indicated (e.g., Knops et al. 1999,Hector et al. 2001, Fargione et al. 2003). Anotherstabilizing mechanism potentially underlying invasionpatterns is the presence of pathogens or herbivores—forwhich the invader is a host or resource—in a communitythat contains species closely related to the invader. Thismechanism is similar to the Janzen-Connell effect, inwhich the presence of adult trees reduces the recruitmentsuccess of conspecific juveniles in tropical forests(Janzen 1970, Connell 1971, Augspurger and Kelly1984). We have previously found evidence for thismechanism, operating via negative soil feedbacks, in atemperate grassland community where it was a powerfulpromoter of coexistence between competing functionalgroups (Petermann et al. 2008). Hence, this pathogen-driven feedback could similarly affect invasion patternsand community reassembly after invasion. Becausefunctional groups are based on species traits, taxonomy,or both (for details regarding the functional-groupclassification in this paper, see  Methods: Experimental design , below), we expect species within functionalgroups to share more pests and pathogens (Gilbertand Webb 2007) and to have more similar resourcerequirements and resource-use patterns (Fargione et al.2003). If invasion and community assembly are drivenby one of these two stabilizing mechanisms, between-functional-group effects would be expected to bestronger than within-functional-group effects.In contrast to these deterministic explanations,invasion and community assembly could be independentof the biology of the species and instead be stronglyinfluenced by chance (Hubbell and Foster 1986, Hubbell2001). If invasion into new communities is viewed in thelight of island-biogeographic theory (MacArthur andWilson 1963, 1967) the probability of colonization bynew species inevitably decreases with increasing speciesrichness of the resident community because a largerfraction of the total species pool has already arrived andestablished. Thus, a negative relationship betweencommunity richness and the number of invading specieswould be expected. At the same time, the number of species going extinct is predicted to increase withincreasing resident species richness, as, for the samearea, population sizes are smaller in diverse communi-ties. Equilibrium richness is reached when extinction andcolonization rates become equal. Under this neutralscenario, the compositions of the assembling communi-ties would be random, meaning that they are notpredictable based on the biology of the species, butinstead governed only by demographic and dispersalstochasticity (Hubbell 2001). In the case of establishedcommunities of different initial richness and composi-tion, invasion of new species and subsequent communityreassembly would then lead to the convergence of species richness but not of composition, even underidentical environmental conditions (Fukami et al. 2005).This was indeed found by two recent studies examiningspontaneous invasion via natural dispersal into experi-mental grassland communities of srcinally differentrichness levels and compositions (Pfisterer et al. 2004,Rixen et al. 2008). Species have often been shown to belimited by their dispersal abilities (Turnbull et al. 2000,Clark et al. 2007), and propagule pressure has beenidentified as a major driver of invasion and communityassembly (e.g., Kolar and Lodge 2001). Thus, thecompositional divergence of different communitiesobserved in spontaneous-invasion studies may well bedue to dispersal stochasticity. On the other hand, initialfloristic composition (Egler 1954, Collins et al. 1995) orthe order of species arrivals (Drake 1990, Chase 2003,Zhang and Zhang 2007) may prevent compositionalconvergence. In that case, the communities‘ colonizationand establishment history may override all otherassembly mechanisms and may have a dominantinfluence on the final composition of reassembledcommunities (Drake 1991).The functioning of plant communities, for example interms of primary productivity, has been found to be afunction of species richness (Tilman et al. 1996, 2001,Hector et al. 1999), phylogenetic diversity (Cadotte et al.2008), functional richness (Tilman et al. 1997, Hector etal. 1999), evenness (Wilsey and Potvin 2000, Polley et al.2003, Hillebrand et al. 2008), and composition (Hooperand Vitousek 1997, Tilman et al. 1997, Spehn et al. 2005;for further references see Balvanera et al. [2006]).Therefore, if invasion leads to changes in theseproperties, it is expected to directly or indirectlyinfluence community functioning (Chase 2003, Hooperet al. 2005). However, the consequences of invasion forthe invaded communities, especially with regard to theirfunctioning, are rarely considered (Pfisterer et al. 2004,Rixen et al. 2008).In the present study, we use an established grasslandbiodiversity experiment with a species richness andfunctional-group richness gradient maintained by weed-ing to study the reassembly of communities by invasionand the resulting effects on ecosystem functioning. After February 2010 409REASSEMBLY OF GRASSLAND COMMUNITIES  opening communities with different initial compositionsto spontaneous invasion and to invasion assisted by seedaddition, we examine whether invasion and reassemblyprocesses are dominated by the biological characteristicsof residents or invaders, by the chance effects of dispersal, or by the compositional history of the residentcommunity. Furthermore, we assess the consequence of invasion, not only for richness and composition but alsofor the functioning of reassembled communities in termsof primary productivity. We show that invasion isbiologically predictable on a functional-group basis andonly weakly dependent on dispersal effects. Invasioncomplements species richness and functional composi-tion and thus leads to the decay of positive speciesrichness–productivity relationships. We suggest that theobserved community reassembly processes were drivenby both resource complementarity and pathogen effects.M ETHODS Experimental design The present study was carried out within a largeexperimental platform at Jena, Germany (50 8 55 0 N,11 8 35 0 E). The Jena Experiment is a long-term grasslandbiodiversity–ecosystem functioning experiment (Rosch-er et al. 2004). It is situated in the floodplain of the riverSaale at an altitude of 130 m above sea level and until2001 it was used for agricultural crops. The experimentalgrassland plots were established by sowing in spring2002. The mean annual air temperature is 9.3 8 C; themean annual precipitation is 587 mm.Seventy-eight experimental plots were sown withrandomly assembled species assemblages of 1, 2, 4, 8,or 16 species. The total species pool of the experimentconsisted of 60 native central European plant speciescommon in seminatural grasslands. Four plots contain-ing all 60 species were also sown. Prior to assemblingexperimental communities, the species were groupedinto four functional groups according to a clusteranalysis using ecological and morphological traits (16grasses, 12 legumes, 12 small herbs, 20 tall herbs;Roscher et al. 2004). Each functional group wasrepresented at each richness level. In addition, thenumber of functional groups was varied within species-richness levels as much as possible, including 16 species-richness levels with only one functional group, so thatthe design was almost completely orthogonal withrespect to functional-group composition and speciesrichness (Roscher et al. 2004). There were 16 differentspecies in monoculture; 16 different species composi-tions at richness levels 2, 4, and 8; and 14 differentspecies compositions at richness level 16 (see AppendixC: Table C1). The plots had a size of 20 3 20 m and werearranged in four blocks. In addition, each plot wasassigned  x - and  y -coordinates to account for geograph-ical position in later analyses. All plots were mown twicea year and did not receive fertilizer.Within each plot, we marked four 2 3 2.25 m subplotsfor our invasion experiment (see Plate 1). One pair of subplots was used for the invasion treatment ‘‘cessationof weeding’’ (C) and one pair for the treatment‘‘weeding’’ (W). In each subplot pair, one subplot wasrandomly assigned to the deliberate seed-additiontreatment ( þ ), and the other received only spontane-ous-invader seeds (  ). The seed-addition treatmentincluded seeds of all species from the srcinal experi-mental pool of 60 species and we therefore refer to themas ‘‘internal invaders’’ if they are not part of the sowncommunity of a specific plot. Seeds were added at a rateof 1000 viable (according to standard laboratory tests)seeds/m 2 in April 2005 divided equally among the 60species. Among the spontaneously ( ¼ naturally) invad-ing species there were both ‘‘internal invaders’’ and‘‘external invaders,’’ the latter not belonging to theoriginal pool of 60 species but occurring in thesurroundings of the field site. Thus, our experimentaldesign consisted of the following four subplots: subplot‘‘W  ’’ was weeded twice a year like the remainder of thelarger 20  3  20 m plot to maintain the srcinal set of species (‘‘residents’’) and served as the control (‘‘closed’’community). In subplot ‘‘W þ ’’ internal invader seedswere added and external invader species were removedby weeding, so that only internal invaders couldestablish. In subplot ‘‘C  ’’ weeding was stopped at theend of 2004; hence, internal invaders and externalinvaders could enter the community spontaneously. Insubplot ‘‘C þ ’’ weeding was also stopped at the end of 2004, so that internal and external species could invadespontaneously; additionally, internal-invader seeds wereadded. Generally, soil disturbance caused by weedingwas kept to a minimum by using small knives to cutweed roots and remove them carefully and by allmaintenance being done before the development of aclosed canopy (early April at the start of the growingseason, and July after the first mowing).We harvested aboveground plant biomass (above 3cm) twice a year for three years after the start of theinvasion experiment, i.e., from year 4–6 after the initialestablishment of the plots. Harvests were timed tocoincide with typical grassland harvest times in centralEurope (late May and August). In each subplot werandomly selected an area 20 3 50 cm for harvest. Wesorted the harvested plant material into species, exceptin the first of the two harvests in 2005, when we onlysorted into residents, internal invaders, and externalinvaders, and noted the number of species in eachcategory. Harvested biomass was dried and weighed.Comparative data from weeded monocultures of all 60species and weeded 60-species mixtures were availablefrom another study within the Jena Experiment (Mar-quand et al. 2009). Data analysis We analyzed the biomass and the number of species of residents and internal and external invaders as afunction of the design variables and covariates withordinary mixed-model analyses of variance (Snedecor JANA S. PETERMANN ET AL.410 Ecology, Vol. 91, No. 2  and Cochran 1980). Fixed and random terms were fittedsequentially by multiple regression and results summa-rized in analysis of variance (ANOVA) tables (for moredetails, see Schmid et al. [2002]). Biomass (in g/m 2 ) wasanalyzed as a yearly total, and species richness (perharvest quadrat) as an average of the two harvests peryear. Because sown resident-species richness in the plotswas highly correlated with realized resident-speciesrichness in the harvested area at the start of ourexperiment, we used sown plot richness in all analysesthat investigate the influence of preinvasion communityproperties on invasion. Results did not change whenrealized richness was used. The number of internalinvader species and their biomass was analyzed on afunctional-group basis in a ‘‘home–away’’ contrastanalysis. This allowed a test of the difference in invasionsuccess between communities where each functionalgroup occurred among the residents (‘‘home’’) andwhere it did not (‘‘away’’). In the home–away biomassanalysis we included only data from 2006 and 2007, asthe biomass of individual functional groups was onlyavailable for one of the two harvests in 2005.The first section of this paper focuses on the influenceof community properties and invader-species character-istics on invasion success. Therefore, only data frominvaded subplots were used (C  , C þ , and W þ ) in therespective analyses. The second section of the paperdeals with community changes in response to invasion.Thus, the development of the non-invaded subplot (W  )was compared with invaded subplots that contained thefull invader range (external and internal invaders: C  and C þ ). All analyses that classify invaders byfunctional group exclude external invaders because thegrouping of internal species into functional groups wasbased on an a priori cluster analysis (see  Experimental design , above) and external invaders occurred in verylow species numbers and abundance. Data wereanalyzed using the statistical software R 2.7.2 (RDevelopment Core Team 2008) and GenStat, eleventhedition (VSN International 2008). All error bars anderrors accompanying mean values represent  6 1 stan-dard error of the mean.R ESULTS Community invasibility Following the cessation of the weeding regime,communities of residents accumulated increasing num-bers of invader species with time. However, the numberand biomass of internal invader species (species thatbelonged to the species pool of the experiment) andexternal invader species decreased with increasingresident-species richness, i.e., resistance to invasionincreased with resident-species richness (Fig. 1,  F  1,63  ¼ 80.23,  P  ,  0.001 for the number of internal invaderspecies;  F  1,63  ¼  32.03,  P  ,  0.001 for internal-invaderbiomass;  F  1,67  ¼  22.03,  P  ,  0.001 for the number of external invader species;  F  1,67  ¼  13.61,  P  ,  0.001 forexternal-invader biomass; full ANOVAs can be found inAppendix C: Tables C2–C4). For internal invaders, thiseffect may in part be due to the decrease in the numberof potential internal invader species in more diverseplots (MacArthur and Wilson 1967, Hector et al. 2001).However, this cannot apply to external invaders becausetheir number is not intrinsically related to the number of resident species. Because the biomass of the residentcommunity increased with sown species richness, wetested its direct effect on invader success by includingresident biomass as a covariate in the analysis. Residentbiomass had a strong negative effect on the number andbiomass of internal and external invader species ( F  1,920 ¼ 106.20,  P  , 0.001 for the number of internal invaderspecies;  F  1,920 ¼ 514.27,  P  ,  0.001 for internal-invaderbiomass;  F  1,160  ¼  79.32,  P  ,  0.001 for the number of external invader species;  F  1,160  ¼  10.36,  P  ¼  0.002 forexternal-invader biomass). Nevertheless, the inclusion of resident biomass as a covariate did not affect thesignificance of subsequent terms in the ANOVA,indicating that resident biomass effects were additiveto the other effects. Invasiveness Internal invader species were much more successfulthan external invaders in invading new communities,even if their seeds were not added deliberately. Onaverage, internal invaders made up 85 %  of all invaderspecies and 95 %  of total invader biomass (Fig. 1).Compared with the spontaneous-invasion treatment, thedeliberate addition of seeds of internal invaders furtherincreased the number of successfully invading internalspecies when resident species richness was low (Fig. 1a, F  1,596  ¼  47.44,  P  ,  0.001 for the interaction ‘‘Speciesrichness  3  Seed addition’’) and increased internal-invader biomass at all species-richness levels (Fig. 1c, F  1,595  ¼  8.4,  P  ¼  0.004 for the term ‘‘Seed addition’’).External invaders were neither negatively nor positivelyaffected by the experimental addition of seeds of internalspecies (Fig. 1b, d,  F  1,75 ¼ 2.47,  P ¼ 0.120 for the numberof external invader species; and  F  1,75 ¼ 0.80,  P ¼ 0.375for external-invader biomass). Furthermore, there wasno effect of external invaders on invasion success of internal invaders ( F  1,155  ¼  0.25,  P  ¼  0.620 for thenumber of external invader species;  F  1,155  ¼  0.28,  P  ¼ 0.600 for external-invader biomass).Because of the small biomass contribution of externalinvaders further analyses were carried out only forinternal invaders. Among internal invaders, functionalgroups and species still varied widely in their ability toestablish in new communities. The most successfulinvading functional groups in terms of the number of established species were grasses and small herbs (1.2  6 0.01 and 1.1 6 0.01 invader species per harvest quadrat,respectively, vs. 0.6  6  0.01 legume and 0.6  6  0.01 tall-herb invader species per quadrat [mean  6  SE]). Grassand legume invaders produced the highest biomass (89 6 4 g/m 2 and 87 6 4 g/m 2 , vs. 57 6 4 g/m 2 and 35 6 4g/m 2 for small-herb and tall-herb invaders, respectively). February 2010 411REASSEMBLY OF GRASSLAND COMMUNITIES  When all internal invaders were examined separately atthe species level, we found that the invasiveness of aspecies in terms of biomass production in a newcommunity was weakly positively correlated with itsaboveground biomass in monoculture ( R 2 ¼ 0.15,  F  1,57 ¼ 10.14,  P ¼ 0.002) but strongly positively correlated withits aboveground biomass in 60-species mixtures ( R 2 ¼ 0.51,  F  1,55 ¼ 55.29,  P , 0.001). Thus, the best predictorof invader performance was resident performance of theparticular species in highly diverse resident communities.The success of invader species or functional groupsalso depended on the interaction between the invaderand the resident species in a community. Both thenumber of internal invader species and their biomasswere reduced when the functional group they belongedto was already present among the residents (‘‘home’’),compared to when it was absent (‘‘away,’’ Fig. 2). Weanalyzed this negative interaction (negative home–awayeffect) between the same resident and invading func-tional groups as a separate contrast within all residentand invading functional-group interactions and found itto be significant ( F  1,11  ¼  37.94,  P  ,  0.001 for speciesnumber;  F  1,11 ¼ 6.50,  P ¼ 0.027 for biomass). Additionalinteractions between resident and invader functionalgroups also influenced invader success. However, theseother interactions were less important than the negativehome–away effect, and the latter was even significantwhen tested against these other interactions (i.e., the F IG . 1. (a, b) The number of species and (c, d) the biomass of internal and external invaders as a function of resident-speciesrichness (log scale). The solid lines represent subplots without seed addition, and the dashed lines represent subplots with seedaddition (see  Methods: Experimental design  for details). The data (mean  6  SE) were averaged over the six harvests from years2005–2007. Note the change in the  y -axis scale for the internal and external invaders. For statistical analysis, see Appendix C:Table C2.JANA S. PETERMANN ET AL.412 Ecology, Vol. 91, No. 2
Similar documents
View more...
Related Search
We Need Your Support
Thank you for visiting our website and your interest in our free products and services. We are nonprofit website to share and download documents. To the running of this website, we need your help to support us.

Thanks to everyone for your continued support.

No, Thanks