Leadership & Management

A truce with neutral theory: local deterministic factors, species traits and dispersal limitation together determine patterns of diversity in stream invertebrates

A truce with neutral theory: local deterministic factors, species traits and dispersal limitation together determine patterns of diversity in stream invertebrates
of 9
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
   Journal of Animal Ecology  2006  75  , 476–484  © 2006 The Authors.Journal compilation© 2006 British Ecological Society  Blackwell Publishing Ltd  A truce with neutral theory: local deterministic factors, species traits and dispersal limitation together determine patterns of diversity in stream invertebrates  ROSS THOMPSON and COLIN TOWNSEND*  Biodiversity Research Centre, University of British Columbia, 6270 University Blvd., Vancouver, BC, Canada V6T 1Z4; and *  Department of Zoology, University of Otago, P.O.Box 56, Dunedin, New Zealand Summary1.  Studies seeking to explain local patterns of diversity have typically relied on nicheexplanations, reflected in correlations with local environmental conditions, or neutraltheory, invoking dispersal processes and speciation.  2.  We used macroinvertebrate community data from 10 streams that varied independ-ently in local ecological conditions and spatial proximity. Neutral theory predicts thatsimilarity in communities will be negatively associated with distance between sites,while niche theory suggests that community similarity will be positively associated withsimilarity in local ecological conditions.  3.  Similarity in total invertebrate, grazer and predator assemblages showed negativerelationships with distance and, for grazers and predators, positive relationships withlocal ecological conditions. However, the best model predicting community similarity inall three cases included aspects of both local ecological conditions and distance betweensites.  4.  When assemblages were analysed according to dispersal ability, high-dispersal spe-cies were shown to be freely accessing all sites and community similarity was not wellpredicted by either local ecology or spatial separation. Assemblages of species with lowand moderate dispersal ability were best predicted by combined models, including dis-tance between sites and local ecological factors.  5.  The results suggest that the perceived dichotomy between neutral and local environ-mental processes in determining local patterns of diversity may not be useful. Neutraland niche processes structured these communities differentially depending on trophiclevel and species traits.  6.  We emphasize the potential for both dispersal processes and local environmentalconditions to explain local patterns of diversity.  Key-words  :dispersal, local determinism, neutral theory, niche, stream macroinvertebrate.  Journal of Animal Ecology  (2006) 75  , 476–484doi: 10.1111/j.1365-2656.2006.01068.x  Introduction  Stephen Hubbell’s unified neutral theory of biodiver-sity (Hubbell 2001) provoked intense debate amongecologists (e.g. McGill 2003; Volkov et al   . 2003; Gilbert& Lechowicz 2004; Wootton 2005). Hubbell’s theoryconsiders communities to consist of ecologically equi-valent individuals distributed across a fixed number of species derived from a regional species pool. Popula-tion dynamics is modelled at regional and local scales.Replacement individuals at a local scale immigratefrom the regional species pool, while at regional scalesnew individuals result from speciation. The dynamicsof a community can therefore be modelled with aminimum of parameters: regional population size,speciation rate, migration rate and death rate. Such anapproach can successfully generate patterns consistentwith species–area (MacArthur & Wilson 1967) andabundance–frequency relationships (Tokeshi 1999).  Correspondence and present address: Ross Thompson,Department of Biological Sciences, Monash University,Clayton 3800, Victoria, Australia. E-mail: Thompson@zoology.ubc.ca   477  Neutral theory and local determinism  © 2006 The Authors.Journal compilation© 2006 British Ecological Society, Journal of Animal Ecology  , 75  , 476–484  Neutral theory explicitly ignores differences betweenindividuals in response to local ecological conditions.In contrast niche theory suggests that patterns of bio-diversity should closely relate to underlying variabilityin ecological parameters such as physico-chemistry,disturbance regime, productivity and competition withother species (e.g. Tilman 1982; Tokeshi 1999).Neutral theory is difficult to test in practice (Harte2003; Gilbert & Lechowicz 2004; Wootton 2005) par-ticularly as key population parameters have rarely beenmeasured (but see Wootton 2005). Attempts to fit species-abundance curves generated by neutral theory andother models to real data have been made (McGill2003; Volkov et al   . 2003; Adler 2004; Alonso & McKane2004; Chisholm & Burgman 2004), but differences inmodel fit are often negligible (Harte 2003; Hubbell &Borda-de-Agua 2004). Even when the model fits (suchas the log-normal curve fitted to Panamanian treediversity by McGill 2003) it may not inform us aboutthe underlying biological processes. There have beenattempts in recent years to test empirically patternsexpected to emerge from neutral processes (e.g. Condit  et al   . 2002; Gilbert & Lechowicz 2004; Wootton 2005).One of these emergent patterns is that of ‘distancedecay’ (Hubbell 2001). Because dispersal limitationunderlies differences between sites in a neutral world, itis expected that widely separated points will harbourdifferent communities (Harte 2003). Differences inlocal species richness between sites can be explained byrandom extinctions and replacements of species throughtime, a process Hubbell (2001) calls ‘ecological drift’.This process has a direct analogy with ‘neutral alleletheory’ (Kimura 1983), the process whereby changes ingenomes are accumulated passively through time via‘genetic drift’. As in genetics, where the relative impor-tance of genetic drift vs. natural selection has beenwidely debated (e.g. Mayr 1991; Ridley 2002), so toohas Hubbell’s assertion of the dominance of ecologicaldrift over local determinism (Nee & Stone 2003).Neutral theory can be tested by comparing the fit of community data to local ecological conditions, vs. theirfit to distance decay expectations. Under niche theory,similarity between species-abundance matrices will bepositively correlated with similarity in local ecologicalconditions. Neutral theory predicts a negative correla-tion with distance between sites. Such a test is mademore difficult because distance between sites is oftenpositively correlated with differences in local ecologicalfactors (Gilbert & Lechowicz 2004). Such an approachis only valid when local conditions and spatial separa-tion are independent of one another. Stream systemsprovide a useful test system because they generate aspatially constrained set of local conditions with inter-vening inhospitable habitat. Aerial dispersal of adultsallows movement of individuals between patches onrealistic scales.We used macroinvertebrate communities from a setof stream sites in a large river catchment in NewZealand to test neutral and niche predictions. Neutraltheory predicts decreased similarity between inverte-brate communities that are spatially distant. Nichetheory predicts decreased community similarity withlower similarity in ecological conditions. Alternatively,we might expect both neutral and niche processes tocontribute to patterns of local diversity, leading to apattern of decreased similarity in communities as spa-tial distance increases and niche similarity decreases.Because dispersal limitation underlies neutral expecta-tions, we also analyse separately communities withvarying dispersal ability. Community similarity for spe-cies with poor dispersal should be strongly negativelyassociated with spatial separation of sites, while specieswith good dispersal should overcome dispersal limita-tion and be strongly positively associated with localecological conditions.  Materials and methods        Ten streams were sampled as a part of a study of differ-ences in food web structure (see Townsend et al   . 1998).Each sampling location was 30 m long and included atleast one pool and one riffle. All sites were in grasslandcatchments and were sampled once during the australsummer between 5 and 16 January 1995. The studystreams were separate third or fourth order tributariesof the Taieri River in New Zealand (Townsend et al   .1998). The sites were distributed over a total area of 4385 km   2  . Intersite distances were calculated by digi-tizing a map of the area and using image measurementsoftware (    4·01). Direct distances among siteswere calculated by measuring vector lengths betweenall site pairs.        A range of ecological factors were measured at eachsite (see Townsend et al   . 1998 for details). Water chem-istry was measured by taking duplicate water samplesfrom each site and analysing for pH and nitrate plusnitrite (NO   3  + NO   2  ) (American Public Health 1992).These measurements were used because they providegood proxies of variation in land use intensity(Townsend et al   . 1998).Physical parameters were summarized by measuringwidth, maximum depth and flow at the swiftest point(using a Marsh–McBirney ‘Flo-mate’ meter), at 1-mintervals at each location (30 measurements for eachvariable in total). One hundred random substrateparticles were collected in each location and their lengthalong the longest axis measured. Averages and stand-ard deviations for width, depth, flow and substrate sizewere calculated.Disturbance was measured using the methodsreported in Townsend, Scarsbrook & Dolédec (1997a).At each location bed particles were taken corresponding   478  R. Thompson and C. Townsend   © 2006 The Authors.Journal compilation© 2006 British Ecological Society, Journal of Animal Ecology  , 75  , 476–484  to the 50th, 75th and 90th percentiles of the substratumsize distribution. These were painted and arrangedon the stream bed in regular arrays (five rows of threeparticles in each of the three size classes randomlyassigned to transects placed 1 m apart). The movementof these particles was monitored on five occasions fromSeptember 1993 to June 1994. Intensity of bed dis-turbance at a location was calculated as the average of the percentage of painted particles that moved betweenconsecutive sampling occasions (see Townsend et al   .1997a).Organic matter standing crop was measured asdescribed in Townsend et al   . (1998). Ten Surber sam-ples (area 0·06 m   2  , mesh size 250 µ  m) were placed atrandom locations along the stream during a period of base flow. Organic matter derived from these benthicsamples was weighed, ashed (550 °  C for 12 h) andreweighed to ascertain the amount of organic matteron the stream bed. Two 500-mL water samples werefiltered through preweighed Whatman GF-C filters,dried, weighed, then ashed (550 °  C for 3 h) andreweighed to ascertain availability of organic matter asseston (water-borne organic matter).Biofilm at each site was measured as ash free dryweight (AFDW) per m   2  of substrate. Ten cobbleswere gathered at random along the stream reach andscrubbed clean of biofilm into a known volume of dis-tilled water. The sample was homogenized and three15-mL samples were filtered on to preweighed What-man GF-F filters, dried, weighed, ashed (550 °  C for3 h) and reweighed to measure the amount of biofilmon each rock. Each scrubbed rock was wrapped in alu-minium foil, then the foil was weighed and a foil weight/area regression used to estimate the surface area of each rock.Total potential algal production was estimated usinga radioactive carbon isotope [7 mL of 14  C-NaHCO   3  (185 MBq ml   −   1  )] within portable chambers (Fuller &Bucher 1991; Thompson & Townsend 1999). Twentyrock samples (each approximately 10 cm   2  ) were col-lected at random from along the study reach andplaced in two recirculating perspex containers. Thechambers were placed in the study reach and the iso-tope was added. After a 2-h incubation period rockswere removed and the amount of carbon taken up wascalculated by extraction in dimethyl sulphoxide fol-lowed by counting in a scintillation counter. To providean estimate of the amount of carbon fixed per m   2  of substrate, the surface area of each rock was measured.This was calculated by wrapping the rock in aluminiumfoil, weighing the foil, and then applying a weight/arearegression to estimate foil area.       Macroinvertebrate samples were taken from 10 ran-dom locations at each location using a benthic sampler(area 0·06 m   2  , mesh size 250 µ  m, taking the top 5 cm of substratum). Analyses of a subset of these streams haveshown that this effort is sufficient accurately representthe invertebrate community (Thompson & Townsend1999). Samples were preserved in 5% formalin forreturn to the laboratory. Samples were sorted foraquatic invertebrate larvae, excluding those less than1·5 mm. Such a procedure excludes meiofaunal species – but these are rare in these systems (Thompson andTownsend 2004). All invertebrates were identified andcounted. Invertebrate identification was carried out at10–40  ×  magnification, identifying animals to the high-est degree of taxonomic resolution possible (Thomp-son & Townsend 1999).Species were attributed functional feeding groupsbased on similar feeding characteristics. Species weredivided into: grazers (predominantly consume algae),predators (feed only on other invertebrate species), andothers, based on dietary information published fromthese streams (Townsend et al   . 1998). Dispersal abilityof invertebrate species was attributed using an inde-pendent data set gathered for an unrelated study(unpublished data, Ngaire Phillips, NIWA, Hamilton,New Zealand). Dissemination potential was describedas low (10 m), medium (1 km) or high (10 km) based ona combination of published and unpublished literature,and augmented where required with expert knowledge(  sensu  Townsend, Doledec & Scarsbrook 1997b). Alldispersal classes were represented in all trophic groups.     Three potential predictor matrices were extracted fromthe data. Ecological data (Table 1) were standardizedand reduced to a matrix of Euclidean dissimilaritiesbetween sites (    , Plymouth Marine Laborato-ries). A spatial matrix was constructed of pairwise spa-tial distances between sites (Table 2). A combinedecological and spatial matrix was constructed by incor-porating pairwise distances between sites as variablesin the ecological matrix, before standardization andreduction to Euclidean dissimilarities. To ensure thatecological and spatial data were not correlated, thecorrelation between these matrices was tested using aMantel test (Mantel 1967; Bonnet & de Peer 2002).Abundance data for invertebrates were summarizedinto a matrix of pairwise Bray–Curtis similaritiesbetween sites, using the program   (PlymouthMarine Laboratories). To test for an effect of trophicgroup, separate matrices including grazers and preda-tors were extracted. To test for an effect of dispersalability, matrices were extracted of low, intermediateand high dispersal species separately. Associationsbetween the invertebrate matrices and the three predic-tor matrices were assessed using Mantel tests (Bonnet& de Peer 2002).Linear regression was used to identify the bestpredictors of similarities between invertebrate com-munities (SPSS 11.0.12001). Bray–Curtis similaritiesbetween communities were used as the dependentvariable and regressed separately against all ecological   479  Neutral theory and local determinism  © 2006 The Authors.Journal compilation© 2006 British Ecological Society, Journal of Animal Ecology  , 75  , 476–484  factors, spatial distance between sites, and ecologicalfactors and spatial distance combined. Because of non-independence of data, significance was tested usingrandomization tests. The best model was chosen basedon the requirement of minimizing the Akaike Infor-mation Criterion (AIC) (Akaike 1973; Anderson,Burnham & Thompson 2000). AIC values are minimizedfor models with high regression coefficients and incor-porate a minimum of predictors. Plots were generatedbetween Bray–Curtis similarities and each of: the bestecological model; spatial distance; and the best com-bined ecological and spatial distance model.  Results  The 10 streams contained a total of 89 taxa, 67·4% of which we identified to species (Table 3). We identified64·2% of individuals to species level and 98·7% to   Table 1. Local ecological conditions in the 10 streams. Disturbance is an index based on the percentage of bed particles moving in a series of dischargeevents (see Materials and methods)   StreamBlackrockBroadCantonDempstersGermanHealyKye BurnLittle KyeStonySuttonWaterpH6·606·326·686·907·356·946·606·816·817·01chemistryNitrate/nitrite ( µ g L − 1 )15·738·7916·8021·184·117·285·065·304·501·45AlgaePrimary production20·7427·3523·0485·3888·2961·7515·5611·1422·459·26(mg C m − 2  h − 1 )Biomass (g m − 2 )1·451·201·951·322·293·611·242·0611·990·95OrganicSeston (mg L − 1 )7·120·774·590·631·070·340·470·760·880·94matterBenthic (g m − 2 )11·304·8511·453·430·752·560·6213·261·104·14ChannelAverage depth (cm)13·5015·9020·3022·007·3015·806·5021·5014·1011·60SD0·030·110·100·140·020·150·030·150·050·06Average width (m)1·201·181·253·723·283·031·976·652·852·40SD0·710·340·360·690·560·590·381·660·370·26Average flow (cm s − 1 )161·70118·50110·4066·3083·7096·1043·9087·90117·8070·45SD52·76111·9248·1277·1134·83133·5341·8460·2046·4744·61Average substrate (mm)34·7557·4381·19118·9359·6196·75122·0234·19182·63106·22SD35·9970·0696·4499·0362·1383·35113·9021·25120·57112·88Gradient (m m − 1 )0·020·020·010·030·030·030·050·010·020·02Disturbance (%)46·5743·4626·2528·5562·2139·7642·2084·5728·3645·94 Table 2. Direct distances between the 10 sampling locations in kilometres   BlackrockBroadCantonDempstersGermanHealyKye BurnLittle KyeStonyBroad2·56Canton2·893·36Dempsters38·7441·0840·41German92·2694·3291·8571·78Healy96·1198·5795·8873·916·14Kye Burn96·0098·6495·9573·287·421·62Little Kye87·5589·7887·0867·895·2511·1212·00Stony25·0726·5723·0841·4572·5077·4977·8367·40Sutton22·6123·7720·0740·7474·7779·5079·8369·572·85 Table 3. Summary of invertebrate data from the 10 locations showing number of individuals identified, total number of taxa andnumber of taxa in different trophic and dispersal ability groupings   StreamNo. of individualsNo. of taxaTrophic groupingDispersal ability GrazersPredatorsLowModerateHighBlackrock3206401195257Broad2845371297225Canton344748111162611Dempsters29594614135288German304132563225Healy35104510134279Kye Burn277835964227Little Kye330529772166Stony317845111252712Sutton2787359124208   480  R. Thompson and C. Townsend   © 2006 The Authors.Journal compilation© 2006 British Ecological Society, Journal of Animal Ecology  , 75  , 476–484  genus. Seventy-two of the taxa could be attributed adispersal ability value.There was no statistically significant correlationbetween spatial location of sites and their local ecolo-gical conditions (Pearson’s Correlation = 0·108, P  =0·165) (Fig. 1). Therefore our analyses of effects of spatiallocation and ecological conditions were independent.There were high correlation coefficients between thespatial matrix and invertebrate matrices for all taxa,grazers and taxa with low dispersal ability (Table 4).However, these invertebrate groups, as well as preda-tory taxa and taxa with moderate dispersal abilities,also showed moderate to strong correlations with localecological conditions (Table 4). For all invertebratetaxa, grazers, and taxa with low and moderate disper-sal abilities, Mantel test’s found the highest correlationwith the predictor matrix combining spatial separationand local ecological conditions (Table 4).Distance between sites was negatively related to sim-ilarity in patterns of abundance for all invertebrate taxaand for grazers and predators separately (Fig. 2a–c).Grazing community similarity had a negative relation-ship with variability in depth, and a positive relation-ship with disturbance and average current (Fig. 2d). Incontrast, similarity between predator communities wasnegatively associated with current and positively asso-ciated with channel slope and width (Fig. 2e). Similar-ities in the overall invertebrate communities werenot well predicted by any of the ecological variables(Fig. 2f). Linear models incorporating spatial distanceand ecological factors were the best predictors of similarities in invertebrate community structure(Fig. 2g–i), although support for the predator commu-nity model was relatively weak. The models for allgroups included average current speed, with the modelfor grazers also including negative associations withpH and variability in current (Fig. 2g).When the invertebrate community was differentiatedaccording to dispersal ability, a negative relationshipwas found between spatial distance between sites andcommunity similarity for low dispersal and moderatedispersal groups (Fig. 3a,b), but no relationship wasfound for the high dispersal group (Fig. 3c). The slopeof this relationship was higher for the low dispersaltaxa than the moderate dispersal taxa (    ; F    1,86  =4·343, P  = 0·040). A model including disturbance, pri-mary production, seston, algal biofilm biomass, pHand average depth was the best predictor of similaritybetween moderate dispersal groups (Fig. 3e), but lowand high dispersal groups were not well predicted byecological variables (Fig. 3d,f). Models incorporatingspatial distance together with a variety of ecologicalfactors were reasonable predictors of communitysimilarity for low and moderate dispersal groups(Fig. 3g,h), but not for their high dispersal counterpart(Fig. 3i). Models combining spatial distance and eco-logical factors were the best predictors of similarity inall community combinations (Table 4, Figs 3 and 4).  Discussion  Our results are consistent with predictions of neutraltheory and a role for local ecological conditions. Pat-terns of species occurrence and abundance were corre-lated with the spatial arrangement of sites and theconditions that were present at each. Grazer commu-nities (the numerically dominant component of thesestreams) were most strongly correlated with a com-bined model of local ecological conditions and spatialdistance, followed by local ecology alone then spatialdistance alone. In contrast, the predatory componentof the community was most strongly correlated withecological factors alone followed by the combinedmodel. There was also evidence for an interaction withspecies’ dispersal ability. While the strongest correla-tion for taxa with moderate dispersal ability was withthe combined model, taxa with low dispersal abilitywere most strongly correlated with spatial distance,and high dispersal taxa were weakly correlated with allmodels. Perhaps the latter taxa disperse over scales thatexceed those in this study. A previous investigationof communities from a much wider range of sitesthroughout the Taieri catchment showed that geo-graphical location was generally influential in account-ing for community composition but not in the case of invertebrates with strongly flying adults (Townsend  et al   . 2003). Fig. 1. Relationship between local ecological factors and thespatial distance between sites (in kilometers). Table 4. Pearson’s correlation coefficients from Mantel testsbetween invertebrate abundance matrices and the threepredictor matrices   SpatialLocal ecologyLocal ecology + spatialAll invertebrates0·4570·2000·519Trophic groupGrazers0·4150·3750·462Predators0·1480·4070·384Dispersal abilityLow0·3810·5400·594Mid0·0540·3230·369High0·0870·1910·150
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