Journal of Applied Sand in coastal dune constrain Ecology 2006 dynamics landscapes 43, 735 747 diversityand life-historycharacteristics of

DRIES BONTE,* LUC LENS* and JEAN-PIERRE MAELFAIT*t *Ghent University,Department of Biology, Research Unit TerrestrialEcology,KLLedeganckstraat35, B-9000 Ghent, Begium; and tInstitute of Nature Conservation,Kliniekstraat 25, B-1070 Brussels, Belgium

Summary 1. In fragmentedlandscapes where natural disturbanceacts at the local level, community composition can be altered either by 'species-by-species matching', with community- wide character displacement, or by 'constrained species sorting', whereby disturbance favours species with distinct ecological traits. It is necessary to understand such species shifts when aiming to detect bottlenecks associated with changing disturbancedynamics at differentspatial levels. Such an understandingalso provides insight into the underlying ecological mechanisms in operation and allows identification of species of conservation concern. 2. This approach to understanding species' shifts was applied to a study of diversityand life-history variation in relation to sand dynamics in stabilized and dynamic grey dune landscapes located along the coasts of France, Belgium and the Netherlands. 3. Local (oc)diversity appeared to decrease with increasing local sand dynamics in both stable and dynamic landscapes. In dynamic dune landscapes, P-diversity decreased significantly and approached regional (y) diversity under local grey dune stabilization; in stabilized landscapes, P3-diversitywas not affected by local sand dynamics. 4. Comparative analyses of ecological traits revealed shifts in life-history patterns, suggesting that patterns in local diversity resulted from species sorting. Species from fragments characterized by high sand dynamics showed narrowerniche breadths, larger body sizes and longer generation times, while summer-active species tended to become residential after sand stabilization. This provides evidence for constrained species sorting in which natural disturbance (through local sand dynamics) allows only species with distinct ecological traits to persist. 5. Synthesis and applications. Species with larger body sizes, longer generation times and a higher degree of habitat specialization (i.e. related to dispersal ability) would be expected to be more vulnerable at dynamic sites and more prone to extinction than their counterparts from stabilized fragments. In particular, species with a burrowing lifestyle would be expected to go extinct locally as a result of increasing soil development and soil hardness. Shifts in species composition were found to be more pronounced in dynamic landscapes. Hence ensuring conservation of sand dynamics at a landscape level rather than at a local level is of prime importance when aiming to conserve typical psam- mophilous spider species within local assemblages in grey dune habitats. Key-words: body size, comparative analysis, conservation, diurnal activity, generation time, metacommunity, seasonal activity Journal of Applied Ecology (2006) 43, 735-747 doi: 10.1111/j.1365-2664.2006.01175.x

Introduction

? 2006The Authors. Metacommunity theory predicts that the abundance of Journalcompilation Correspondence:Dries Bonte, Ghent University, Department species and community composition in spatially struc- ? 2006British of Biology,Research Unit TerrestrialEcology, KL Ledeganck- tured, heterogeneousenvironments will be altered either EcologicalSociety straat35, B-9000 Ghent, Belgium (e-mail [email protected]). by 'species sorting' or 'mass effects', depending on the

This content downloaded from 157.193.10.229 on Tue, 07 Jul 2015 13:14:03 UTC All use subject to JSTOR Terms and Conditions 736 relativeimportance of dispersalfor the speciesconcerned heterogeneous levels of local disturbance are imposed D. Bonte, L. Lens & (Leibold et al. 2004). In species sorting, i.e. matching of on spatially structuredcommunities, such species shifts J-P Maelfait species with their favoured habitat, the patterns of would be determined by environmental constraints species distribution among patches is expected to relate that limit character displacement. Hence, community to each species' relative ability to exist and successfully wide character shifts (CWCS), rather than character interact with other species in the larger metacommu- displacement, would be expected. nity. Mass effects influence local abundances of species In coastal dunes, aeolic dynamics (e.g. severe sand with high dispersal rates through either net emigration displacement) is a geomorphological process that is (source populations) or net immigration (sink popula- essential in maintaining the landscape in its dynamic tions). If mass effects occur in patches with high local phase (Provoost 2004), i.e. characterized by an extreme disturbance,some species may persist as fugitivesdespite microclimate,high summertemperatures and the absence being poor competitors (Lytle 2001; Leibold & Miller of soil development. Many studies have documented 2004). Trade-offs between both colonization and significant effects on diversity patterns and the assem- competitive abilities can therefore be expected to affect blage structure of plants and of reducing sand assemblage composition and structure. dynamics (Bach 2001; Bonte, Baert & Maelfait 2002; If such trade-offs primarily occurred at a local scale, Kowalchuk, De Souza & Van Veen 2002; Snyder& Boss they may promote species coexistence predominantly 2002; Franks & Peterson 2003; Bonte et al. 2004b; Jun, within patches. Consequently ax-diversity would be Clement & Roze 2004). However, whether shifts in spe- expected to be high (Kneitel & Chase 2004). Alterna- cies composition result from species-by-species match- tively, if such trade-offs existed at a regional scale, a- ing with CWCD or from constrained species sorting, diversity would be expected to be low, as only good i.e. in which disturbance favours species with distinct competitors or good dispersers would be able to persist ecological traits, is poorly documented in animals. In within each locality. Consequently, 3-diversity would plants, aeolic disturbance is known to favour species be expected to be high because species differing in their adapted to sand burial that exhibit fast growth or degree of specialization or dispersal would be expected tolerance to darkness (Maun 1998; Garcia-Mora, to occupy different heterogeneous and homogeneous Gallego-Fernandez & Garcia-Novo 1999; Martinez & patches, respectively.If patch heterogeneity was related Maun 1999;Wilson & Sykes 1999), seasonal growth and to disturbance, good dispersers may, however, fail to flowering (Garcia-Mora, Gallego-Fernandez& Garcia- settle under disturbedconditions. Under such a scenario Novo 1999), below-ground spreading root networks (i) o-diversity would be expected to be low if distur- (Garcia-Mora,Gallego-Femandez & Garcia-Novo 1999), bance is high, (ii) P-diversity would be expected to leaf morphology related to desiccation resistance increase if only a few species were adapted to survive (Garcia-Mora,Gallego-Femandez & Garcia-Novo 1999) under those conditions (Pianka 1994)and (iii) P-diversity and/or enhanced mycorrhizal activity (Maun 1998). In would be expected to decreaseif the assemblagesexperi- animals, spiders from white dunes show higher levels encing disturbance were inhabited by species drawn of thermal tolerance than species preferring thick at random from a regional species pool. Local distur- vegetation in stabilized dune habitats (Almquist 1970). bance is therefore expected to exert a minor effect on The latter species also show a lower resistance to regional diversity (Chase 2003). Landscape history and desiccation (Almquist 1971). In Namib desert spiders, patch connectivity are also assumed to affect local and psammophiliousspecies show adaptationsto desiccation regional patterns of species diversity (Loreau 2000; resistance through larger size (hence reduced surface- Chase 2003; Kneitel & Chase 2004). to-volume ratio), being night active and active only However, whether such shifts in diversity and assem- during the humid season (Henschel 1997). They also blage composition areconstrained by disturbance-related have a burrowinglife-style, a low metabolismand cryptic factors remains open to question. In community-wide coloration. character displacement (CWCD), species sorting (or Earlierstudies (Bonte, Maelfait & Baert2004) revealed species-for-speciesmatching; Schluter 2000b) is expected strong effects of dispersal ability on spider species' dis- to produce groups of ecologically similar species, char- tributions among grey dune fragments. As dispersal is acterized by similar life-history patterns, in different negatively related to the degree of habitat specialization localities. Such patterns are believed to have evolved by (Bonte et al. 2003d) and hence to competitive ability in (i) natural selection promoting divergent phenotypes optimal habitat, patterns of species sorting under that reduce competition or (ii) ecological sorting in community-widecharacter shifts in ecologically relevant which competition induces local extinction (Marchinko, traits are particularlyexpected in populations subjected Nishizaki & Burns 2004). However, from a metacom- to stress and disturbance. Using this conceptual frame- ? 2006 The Authors. munity perspective, species sorting along gradients in work as a basis,we studiedthe patternsof speciesdiversity Journal compilation communities might result in reverse patterns in which and ecological characteristicsin spidersfrom fragmented ? 2006 British locally coexisting species are more similar than random grey dune habitats subject to varying degrees of natural Ecological Society, Journal of Applied selections, as species combinations that are more sim- sand dynamics and anthropogenic stabilization of the Ecology, 43, ilar in resource use appear to be less prone to invasion dunes. To reduce confounding effects of phylogeny, 735-747 by new species(Leibold 1998;Leibold et al. 2004). Where analyses of ecological traits were conducted within the

This content downloaded from 157.193.10.229 on Tue, 07 Jul 2015 13:14:03 UTC All use subject to JSTOR Terms and Conditions 737 most diversespider families and hence functional groups Company North-Holland (PWN). These four dune Sand dynamics that show similar life histories (Foelix 1996) as a result areas differ in biogeography,the level of sand dynamics and spiders of a common evolutionary history (Blondel 2003). and, consequently, spider fauna composition (Bonte Understanding shifts in assemblages et al. 2003a). BOUL and FCD are dynamic dune land- is essential for the detection of bottlenecks within scapes with large areas of mobile dunes, whereas AWD the process of declining sand dynamics. As time and and PWN have been stabilized by human management logistics do not generally permit population studies for during the last century. In each dune area, we sampled prioritizing conservation actions to be conducted on a large grey dune fragments (as the richness of specific species-by-speciesbasis (Bennet & Owens 1997;Jennings xerotherm species declines in small patches; Bonte, et al. 2001; Bessa-Gomez et al. 2003), comparative Baert & Maelfait 2002) covering the full range of local methods may provide insight into the underlying sand dynamics.Seven fragmentswere sampled in BOUL ecological mechanisms in operation and allow prior and FCD, six in AWD and eight in PWN. Within identificationof speciesof conservationconcern (Kolar & each fragment, five pitfall traps (9 cm, filled with a 6% Lodge 2002; Reed & Shine 2002). To this end, addi- formalin-detergentsolution) were installed in a line 5 m tional comparativeanalyses of ecological traits between apart. Trapswere emptied every fortnight and sampling congenerics were used to reduce phylogenetic effects occurred between March-November 1999 to ensure (Gaston & Kunin 1997; Bevill & Louda 1999). As the entire spider assemblage present at each site was decreasing sand dynamics result in deeper soil devel- sampled. As described in Bonte et al. (2003a), PCA opment and higher nutrient availability,we expected an analysis of site-specific environmental parameters increase in eurytopic species with shorter generation revealed a single significant principal component that times, higher local densities and better-developedaerial was stronglycorrelated with aspects of soil development, dispersal capacities. In addition, as sand stabilization sand dynamics and vegetation development (vegeta- results in a more buffered microclimate, we expected tion dimensionality). This principal component was higherfrequencies of summer-active,diurnal and smaller therefore used to indicate and quantify the degree of species because of their lower sensitivity to desiccation. sand dynamics within each landscape fragment. We also considered whether species diversity and ecological patterns were affected by the interaction LOCAL DIVERSITY between local (patch related) sand dynamics and the degree of large scale sand dynamics within the entire a-diversity was assessed by counting the total number dune landscape. Therefore patterns in oc-, P- and y- of species for which both five males and five females diversity and spider ecology were studied in both were recorded. In this way, only resident species were stabilized and dynamic dune landscapes. included and accidental vagrantswere excluded (Bonte, Maelfait & Baert 2004). From a total of 190 captured species (represented by 35975 individuals), 68 species, Methods belonging to five dominant families (Gnaphosidae, Thomisidae, Lycosidae, Erigoninae and Linyphiinae), STUDY AREAS AND SAMPLING STRATEGY were retained (see Appendix SI in the supplementary The studywas conductedin grey dune habitats,known as material). [3-diversitywas estimated by dividing the re- fixed coastal dunes with herbaceous vegetation, varying gional species richnessin grey dune habitats (y-diversity) in sand dynamics.In such habitats,the magnitudeof sand by the number of local (i.e. within patches) species dynamics is related to vegetation and soil development (a-diversity) (Whittaker 1972). as sand stabilizationresults in soil formation, producing a stronger vegetation development ranging from moss ASSESSMENT OF LIFE-HISTORY dunes with sand blowouts towards stable, but low- CHARACTERISTICS productivity,dune grasslands(Provoost et al. 2004). The vegetation consisted of (i) dunes dominated by Atlantic Species-specific body sizes were taken from Roberts moss and (ii) dunes with a distinct soil organic layer,and (1985, 1998). As body size differs between sexes, mean either Cladonio-Koelerietalia(on the lime-richgrey dunes) values of the larger females were used in all subsequent or Trifolio-Festucetaliaovinae (on the decalcified grey analyses. In Lycosidae, two burrowing species [i.e. dunes) grassland (Provoost et al. 2004). From an eco- fabrilis (Clerck 1757) and Arctosa perita logical perspective, the grey dune habitat representsthe (Latreille 1799)] were recorded; all other species live dry component of the 'stresseddune landscape',where exclusively above-ground (Roberts 1998). succession is determinedby top-down regulating stress Data on niche breadth were derived from Hanggi, C 2006 The Authors. factors and bottom-up (xerosere) organization. Stockli & Nentwig (1995), who reviewed the number of Journal compilation Field work was conducted in four dune areas along central habitats in which species occur. ? 2006 British European the North Sea coast of northern France, Belgium and Earlier showed that this index Ecological Society, analysis ranking signi- Journal of Applied the Netherlands:the Boulonnais dunes (BOUL), Flemish ficantly correlated with niche breadth values based on Ecology, 43, coastal dunes (FCD), coastal dunes of the Amsterdam indicator analysis in the Flemish coastal dunes (Bonte 735-747 WaterSupply (AWD) and dunes of the ProvincialWater et al. 2003d). Average activity density was calculated

This content downloaded from 157.193.10.229 on Tue, 07 Jul 2015 13:14:03 UTC All use subject to JSTOR Terms and Conditions 738 by dividing the total number of individuals by the species composition between dynamic and stabilized D. Bonte, L. Lens & number of traps in which the species was captured. dunes, landscape type was included as a fixed factor, J.-P Maelfait While we are aware that pitfall catches reflect a com- while dune area was included as random factor nested bined activity density pattern that shows large differ- within landscape (i.e. BOUL and FCD within dynamic ences as a result of inter- and intraspecies variation landscape, AWD and PWN within stabilized land- (Maelfait & Baert 1975; Bonte, Maelfait & Baert 2004; scapes). Because fragments were distant from each Schmidt et al. 2005), the nature of our analysis (i.e. other, they were treated as independent units and no comparing patterns within families and, especially, additionalcorrection for spatialdependence was applied. between congeneric species) still allowed us to predict Because several species, captured within and between patterns in species density because activity patterns can landscape fragments, cannot be used as independent be assumed to be similar among (very) related species replicates with respect to the influence of sand dynam- occurring in structurally similar habitats. ics on related life-history characteristics, mean values As species-specific data on ballooning propensity (a of these traits within each fragment are the appropriate precursorfor aerial dispersal in spiders;Bell et al. 2005) unit of replication (n = 28). We therefore determined a was lacking, we assigned species to the category of weighted average of continuous life-history character- 'ballooners' if aeronautic dispersal was observed either istics (niche breadth, body size, local activity-density, in the field or in the laboratory, irrespective of its mag- generation time) and analysed these traits by means of nitude (Bonte, Maelfait & Hoffmann 1998; Bonte et al. general linear models, assuming a normal distribution. 2003d; Bell et al. 2005; D. Bonte, unpublished data). Both two-factorinteractions, body size x seasonalactivity Seasonal activity patterns were derived from Roberts and body size x diurnal activity, were modelled. Cate- (1985), catalogues of Belgian spiders (Alderweireldt & gorical data (seasonal activity, aeronautic dispersal Maelfait 1990; Ransy & Baert 1991; Janssens 1993; and diurnal activity) were analysed at fragment level Baert 1996)and personaldata. FollowingSchaefer (1976), with logistic mixed models. Analyses were conducted we assigned spiders to one of four classes: eurychrone with Proc mixed (continuous data) or Proc glimmix species, in which adults are active throughout the year (logistic models) using SAS 9-1 (SAS 2003). Backward (type I); stenochrone species, with their main activity stepwise elimination of landscape x aeolics and land- during March-June (type IIa); stenochrone species, scape was used where these factors did not contribute with an activity peak during June-September(type IIb); significantly to the full model. Residuals (standardized and stenochrone or diplochrone species, with their main residuals in case of logistic regression) were checked for activity during October-March (type IV-V). In combin- normal distribution. Random factors, even in case of ation with data on temporal distribution of immatures low levels of explained variation, were retained in the fromdifferent size classes,this allowedus to assessspecies- final model in order to optimize the level of generality specific generation times expressed as generation time of the distributionmodels (Vaughan & Ormerod 2005). 6 months-'. Specieswith a bienniallife cycle wereassumed Effective degrees of freedom were approximated by the to requireon average1-5 yearsfor complete development. Satterthwaiteprocedure (Verbeke & Molenberghs2000). Diurnal activity patterns were assessed by separate Because all the traits under study are dependent on pitfall trapping during day and night. These data were phylogenetic effects, family effects were also taken into mainlyderived from coastal dunes (R. Bosmans,L. Baert, account. Third-order interactions with family were J.-P.Maelfait & J. Huble, unpublished data; D. Bonte, significant for body size (F4,118= 5-25, P < 0-001), niche unpublished data; Bonte, Hoffmann & Maelfait 2000) breadth (F4,1n7=3-02, P = 0-002), generation time and from the southern part of Belgium (J.-P. Maelfait, (F4120 = 4-11, P = 0-003), local abundance (F4,120= 4-34, L. Baert, D. Bonte, unpublished data). Species were P = 0-003), aerial dispersal (F4,118= 2711, P = 0-034) classified as day active, night active or indifferent, and/or did not meet convergence criteria for seasonal where the species showed similar activities during day activity and day-night activity.Analyses were therefore and night. An overview of all life-history traits is given performed separately for five species-rich (sub)families in Appendix Sl in the supplementary material. showing substantial trait variation (Gnaphosidae, Lycosidae, Thomisidae, Erigoninae and Linyphiinae). Bonferronicorrection was not appliedbecause the chances STATISTICAL ANALYSIS of type I errorsas a result of multiple comparisons were considered to be negligible in comparison with the level Life-history variationsalong assemblages of residual variance resulting from the presence of non- Life-history characteristics were considered for the typical species invading from neighbouring habitats species occurring in the different fragments along the (Bonte, Baert & Maelfait 2002; Moran 2003). ? 2006 The Authors. gradient of sand dynamics. No weight was given Journal compilation to local abundance levels. Mixed models were used to ( 2006 British Comparativeanalysis investigate differences in life-history traits in relation to Ecological Society, Journal of Applied sand dynamics, with 'trait' or 'diversity measure' as the InVal-indicator analysis (Dfifrene & Legendre 1997) Ecology, 43, dependent variable and sand dynamics as a continuous was applied to identify species for dynamic (PCA < -1) 735-747 factor. Because earlier analysis revealed differences in and stabilized (PCA > 1) fragments across all study

This content downloaded from 157.193.10.229 on Tue, 07 Jul 2015 13:14:03 UTC All use subject to JSTOR Terms and Conditions 739 Table 1. Species-couples from five spider (sub) families used 60

Sand dynamics in the comparative analysis of life-history traits S S and spiders 55 0 (Sub) family Species pairs (dynamic-stabilised) . 0 Liocranidae Agroeca lusatica -proxima > 50 Lycosidae Alopecosa barbipes- cuneata a) *0 - . Lycosidae Alopecosafabrilis pulverulenta :5 45 . - Erigoninae Erigonepromiscua dentipalpis c4 Erigoninae Oedothorax apicatus - retusus _Q- < 40 Erigoninae Parapelecopsisnemoralis - Pelecopsisparallela Erigoninae Walckenaeriastylifrons - antica 35 Thomisidae Xysticus kochi- cristatus Thomisidae Xysticus sabulosus - ninnii Gnaphosidae Zelotes longipes - electus 30 Sand stabilisation 4 (b) 0 areas.Within this set, we selected 10 pairs of congeneric 3 species from five different families (Table 1) in which paired species from each couple were indicative of sta- a) bilized vs. dynamic grey dune fragments, respectively. .- 2 ao Comparative analysis within each of the species pairs a) - 03 8,,vCo allowed us to omit phylogenetic effects. Pairwise one- m tailed tests for dependent continuous and count data 1 * Dynamiclandscapes (respectively t-tests, rank sign tests) and categorical o Stabilisedlandscapes data (McNemar X2-tests)were applied. , ,I I 0 -4 -3 -2 -1 0 1 2 3 4 Sand stabilisation Results Fig. 1. Relationship between spider species diversity and the degree of sand dune stabilisation in coastal grey dune SPECIES DIVERSITY fragments. (a) a-diversity; (b) P-diversity. Points: observed a-diversity increased with decreasing sand dynamics data; Full line: modelled relationship for dynamic landscapes; Dashed line: modelled relationship for stabilised landscapes. (r = 036 +?0-12 SE, = 30-84, P < 0-0001; Fig. 1). F1,26 The principal component scores arising from a PCA-analysis Neither = P > nor the landscape type (F1,24 001, 0-05) of site-specific environmental parameters is used to indicate two-factor interaction with local sand dynamics (F1,23.4 an increasing degree of sand stabilisation along the x axis. = 2-04, P > 0-05) was statistically significant, while no variation was attributed to the dune area (s2 < 0.001). In contrast, P-diversity = 14-34, d.f. = 1,23 1, (Fsand_dynaics LIFE-HISTORY CHARACTERISTICS ALONG P = 0-0009; = 0-37, d.f. = 1,2-53, P = 0-591; Flandscape ASSEMBLAGES = Fsand_dynamicsxlandscape 5-15, d.f. =- 1,23-1, P = 0-034) decreased with decreasing sand dynamics (Fig. 1) more Body size was negatively related with sand stabiliz- strongly in dynamic (estimated slope -0-23 ? 0-11 SE) ation for Lycosidae,with a strongerdecrease in dynamic than in stabilized dune landscapes (estimated slope landscapes (r = -055 ? 0-11 SE) than in stabilized -0 07 ? 0 10 SE). As with a-diversity,random effectsdid landscapes (r = -0-20 ? 0-11 SE). For Thomisidae, not significantlyexplain additional variation (s2 < 0-005). Erigoninaeand Linyphiinae,body size remainedconstant irrespectiveof aeolic dynamics (Table 3). For Gnaphosi- dae, an increase in body size was observed in dynamic NICHE BREADTH landscapes (r= 0-45 ? 0-12 SE) but not in stabilized Erigoninae(r = 0 67 ? 0-34) showed a significantincrease landscapes (r = -0-03 + 0-12 SE). No community-wide in generalistspecies with decreasinglocal sand dynamics, interactions occurred between body size patterns and independent of landscape effect (Table 2). However, seasonal activity or diurnal activity. shifts in community-wide niche breadth indicated sig- With increasing sand dynamics, average Lycosidae nificant interactionswith landscape type for the families generationtime decreased(r = -0-03 + 0-01 SE; Table 3). Gnaphosidae and Lycosidae: in dynamic landscapes, For Gnaphosidae and Thomisidae, decreasing genera- decreasing local sand dynamics resulted in a steeper tion times were only observed in dynamics landscapes ( 2006 The Authors. = increase of eurytopic species than in stabilized land- (Gnaphosidae, rdyn= -0-07 ? 0-02 SE, rstab -002 ? 0-02 Journal compilation scapes (Fig. 2a; Gnaphosidae = 2'00 + 0'47 SE, rstab SE; Thomisidae, = -0-06 ? 0-01 SE, rstab= -0-01 + ? 2006 British rdyn rdyn = -0-11 ? 0-47 rdyn = 3-44 ? 0-82 0-02 Ecological Society, SE; Fig. 2b; Lycosidae SE). Journal of Applied SE, rstab= 092 ? 0-84 SE). A similar trend was observed Local activity density patterns (Table 3) decreased Ecology, 43, for Thomisidae (Table 2; rdyn= 1-70 ? 0-71 SE, rstab= for Linyphiinae (r = -0-34 ? 0-16 SE) and Erigoninae 735-747 -0-68 ? 0-72 SE). (r = -0 16 +?008 SE) in both stabilized and dynamic

This content downloaded from 157.193.10.229 on Tue, 07 Jul 2015 13:14:03 UTC All use subject to JSTOR Terms and Conditions 740 Table 2. Influence of local sand dynamicsdynamis (aeolic cs) and landscape type (stabilised versus dynamic) on niche breadth in D. Bonte, L. Lens & grey sand dune spider communities. GLM results for each of the five spider (sub) families studied. Num. d.f: numerator degrees J-P Maelfait of freedom; Den. d.f.: denominator degrees of freedom

(Sub) family Factor Num. d.f. Den. d.f. F p

Gnaphosidae Aeolic dynamics 1 24 8-26 0.008 Landscape 1 24 0-08 0-776 Aeolic dynamics * Landscape 1 24 12-46 0-002 Lycosidae Aeolic dynamics I 24 16-69 0-004 Landscape I 24 1-69 0-206 Aeolic dynamics * Landscape I 24 5-06 0-034 Thomisidae Aeolic dynamics I 25-6 2-28 0-143 Landscape I 2-19 0-30 0-656 Aeolic dynamics * Landscape I 22-4 3-19 0.088 Erigoninae Aeolic dynamics I 25-8 3.73 0-064 Landscape I 2-16 0-06 0-822 Aeolic dynamics * Landscape I 22-5 0-10 0-753 Linyphiinae Aeolic dynamics I 23-7 0-17 0-682 Landscape I 22 3.39 1-195 Aeolic dynamics * Landscape I 22-7 1 11 0-303

28- - (a) / (b) / 26- 40 24- 22 ooo 20 * o ' 30 - 18- * landscapes -- I 0<'J 0 .c 00>- < / ~~~o 16- * ^ 0 z z 14 20 / |^^~ I * Dynamic landscapes / ' o* landscapes ~ o 12 Dynamic Stabilised landscapes o Stabilised landscapes 10 Gnaphosidae 0 Lycosidae 8 I . . I . I . I . I ill I -4 -3 -2 -1 0 1 2 3 4 -4 -3 -2 -1 0 1 2 3 4 Sand stabilisation Sand stabilisation

07- (c)

0-6 0 *0 0-5 0 0 0 S ' 0 * 0-4 S 0 0

0 0 0 * * 0.2 0 0-1

0-0 0 * Lycosidae . . . .I. -4 -3 -2 -1 0 1 2 3 4 -4 -3 -2 -1 0 1 2 3 4 Sand stabilisation Sand stabilisation

0.8- (f) * Erigoninae o Lycosidae v Thnmic.ir4.. 2' 0-6 * * * *. * * a- * * * * --. 1 *> 0-4 * * -

SU 0.0*0 I * .? V v 3: QL-02 vo v v 0-1 Ly0

Lycosidae 0.0 -4o--0 -3 , -2I -1 - 0_-- 1 2 3 4 C 2006 The Authors. -4 -3 -2 -1 0 1 2 3 4 -4 -3 -2 -1 0 1 2 3 4 Sand stabilisation Sand stabilisation Journal compilation ? 2006 British Fig. 2. Relationship between spider life-history characteristicsand the degree of sand stabilisationin coastal grey dune fragments. Ecological Society, (a) Gnaphosidae niche breadth; (b) Lycosidae niche breadth; (c) Lycosidae ballooning ability; (d) Erigoninae ballooning ability; Journal of Applied (e) incidence of burrowing lycosid species; (f) winter activity in Erigoninae, Lycosidae, Thomisidae. Points: observed data; in (b) Ecology, 43, and (d) full line: regression model dynamic landscape, dashed line: regressionmodel stabilised landscape; in (f) full line: regression 735-747 model for Lycosidae, medium dashed line: regression model for Erigoninae, large dashed line: regression model for Thomisidae.

This content downloaded from 157.193.10.229 on Tue, 07 Jul 2015 13:14:03 UTC All use subject to JSTOR Terms and Conditions 741 Table3. Influenceof local sanddynamics (aeolic dynamics) and landscapetype (stabilisedversus dynamic) on body size,local Sand dynamics activity and generation time in grey sand dune spider communities. GLM results for each of the five spider (sub) families studied. and spiders Num. d.f: numerator degrees of freedom; Den. d.f.: denominator degrees of freedom Trait (Sub)family Factor Num. d.f. Den. d.f. F P

Body size Gnaphosidae Aeolic dynamics 1 22-8 6-91 0-015 Landscape 1 2-23 1-73 0-307 Aeolic dynamics * Landscape 1 22-8 6-81 0-016 Lycosidae Aeolic dynamics 1 24 27-29 <0.001 Landscape 1 24 7-82 0-010 Aeolic dynamics * Landscape 1 24 4-36 0-047 Thomisidae Aeolic dynamics 1 24 2-67 0-115 Landscape 1 1'98 5-40 0-148 Aeolic dynamics * Landscape 1 22-8 2-28 0-145 Erigoninae Aeolic dynamics 1 25 0-14 0-713 Landscape 1 26 3-96 0-057 Aeolic dynamics * Landscape 1 24 0-43 0-519 Linyphiinae Aeolic dynamics 1 25 0-29 0-595 Landscape 1 26 20-88 <0-001 Aeolic dynamics * Landscape 1 24 0-86 0-363 Local activity-density Gnaphosidae Aeolic dynamics 1 24 14-19 0.001 Landscape 1 24 2-21 0-499 Aeolic dynamics * Landscape 1 24 9-59 0005 Lycosidae Aeolic dynamics 1 24 6-00 0-022 Landscape 1 24 0.00 0-996 Aeolic dynamics * Landscape 1 24 4-33 0-048 Thomisidae Aeolic dynamics 1 26 0-53 0-482 Landscape 1 25 0.00 0-964 Aeolic dynamics * Landscape 1 24 2-97 0-698 Erigoninae Aeolic dynamics 1 26 3.93 0-058 Landscape 1 25 0-01 0-923 Aeolic dynamics * Landscape 1 24 0-13 0-720 Linyphiinae Aeolic dynamics 1 26 4.47 0-044 Landscape 1 2-29 0-38 0-592 Aeolic dynamics * Landscape 1 22-9 1-17 0-290 Generation time Gnaphosidae Aeolic dynamics 1 23-6 19-14 <0.001 Landscape 1 25 2-33 0-242 Aeolic dynamics * Landscape 1 23-6 6-77 0-016 Lycosiclae Aeolic dynamics 1 26 9-89 0-004 Landscape 1 25 0-50 0-488 Aeolic dynamics * Landscape 1 24 0-46 0-505 Thomisidae Aeolic dynamics 1 24 10-71 0-003 Landscape 1 24 23-36 <0.001 Aeolic dynamics * Landscape 1 24 4-88 0-037 Erigoninae Aeolic dynamics 1 26 0-53 0-471 Landscape 1 25 2-23 0-148 Aeolic dynamics * Landscape 1 24 0-31 0-583 Linyphiinae Aeolic dynamics 1 24-7 0-74 0-399 Landscape 1 2-56 13-38 0-046 Aeolic dynamics * Landscape 1 24 0-73 0-401

landscapes. Gnaphosidae only showed an activity by epigeic, wandering species under declining sand density decrease in dynamic landscapes compared with dynamics (r = -0-439 ? 0-13 SE, F,23 = 11 68, P = 0-002; = stabilizedlandscapes (rdyn -0-29 ? 0-06 SE, rstab= -0-01 Fig. 2e). The relative occurrence of day-active, night- ? 0-06 SE). In contrast, Lycosidae showed an increas- active, type I and type IIa species did not vary in ing activity density pattern in relation to increasingsand relation to sand dynamics (all F< 1 00, P > 0 10), inde- = stabilization, although only in dynamic landscapes (rdyn pendent of landscape type. Lycosidae (r -0-279 ? 0-12 = = 284 ? 0-80 SE, rstab= 0-23 ? 0-83 SE). SE, F,23 3-18, P = 0-087), Thomisidae (r = -0-511 + ? 2006 The Authors. Lycosidae species with ballooning ability increased 0 19, F1,23= 4-09, P = 0-055) and Erigoninae (r = -0 111 Journal compilation with decreasing local sand dynamics (Table 3; r = 0 18 ? 0-06 SE, = 3-36, P = 0-080) tended to show shifts ? 2006 British F1,23 ? 0-09 In towardswinter in localities Ecological Society, SE; Fig. 2c). Erigoninae (Table4), frequencies activity (type IV-V) dynamic Journal of Applied of ballooning species declined only in dynamic dune (Table 4 and Fig. 2f). Only in the latter family was this Ecology, 43, landscapes (rdyn= -0-35 ? 0-08 SE, rstab= 0-05 ? 0-08 SE; shift accompanied by a decrease in type IIb (summer- 735-747 Fig. 2d). In Lycosidae, burrowingspecies were replaced active)species (r = 0 143 ? 0-08 SE, F,23= 459, P=0-043).

This content downloaded from 157.193.10.229 on Tue, 07 Jul 2015 13:14:03 UTC All use subject to JSTOR Terms and Conditions 742 Table 4. Influence of local sand dynamics (aeolic dynamics) and landscape type (stabilised versus dynamic) on the proportion of D. Bonte, L. Lens & species with aerial dispersal propensity and the proportion of winter-active species in grey sand dune spider communities. GLZM J-P Maelfait results for each of the five spider (sub) families studied. Num. d.f.: numeratordegrees of freedom; Den. d.f.: denominator degrees of freedom

Trait (Sub) family Factor Num. d.f. Den. d.f. F P

Aerial dispersal Gnaphosidae Aeolic dynamics Model did not show convergence propensity Landscape Model did not show convergence Aeolic dynamics * Landscape Model did not show convergence Lycosidae Aeolic dynamics 1 23 4-17 0-053 Landscape 1 2 8-07 0-105 Aeolic dynamics * Landscape 1 22 0-57 0-456 Thomisidae Aeolic dynamics 1 23 0-03 0-866 Landscape 1 2 0-26 0-660 Aeolic dynamics * Landscape I 22 0.01 0-912 Erigoninae Aeolic dynamics I 23 1-26 0-273 Landscape I 2 0-11 0-757 Aeolic dynamics * Landscape I 22 4-17 0-053 Linyphiinae Aeolic dynamics I 23 0-25 0-619 Landscape I 2 8-35 0.119 Aeolic dynamics * Landscape I 22 0-33 0-571 Winter activity Gnaphosidae Aeolic dynamics Model did not show convergence Landscape Model did not show convergence Aeolic dynamics * Landscape Model did not show convergence Lycosidae Aeolic dynamics 1 23 3-18 0-087 Landscape 1 2 2-91 0-231 Aeolic dynamics * Landscape I 22 1-47 0-238 Thomisidae Aeolic dynamics I 23 4-09 0-055 Landscape I 2 3-02 0-178 Aeolic dynamics * Landscape I 22 0-17 0-684 Erigoninae Aeolic dynamics I 23 3-36 0-080 Landscape I 2 0-62 0-514 Aeolic dynamics * Landscape I 22 0-02 0-891 Linyphiinae Aeolic dynamics I 22 0-69 0-415 Landscape I 2 7-61 0.110 Aeolic dynamics * Landscape I 21 0-26 0-615

Table 5. Comparative analysis of spider life history traits for congeneric species indicative of dynamic and stabilised coastal grey dune fragments. The species couplets used in the analysis are shown in Table 1

Mean Mean valuelfreq. valuelfreq. dynamic stabilised Trait Test statistic P fragments fragments

Body size Z = 200 0-045 5-932 4-890 Generation time Z = 2-84 0-004 1-455 1-273 Winter activity X2 = 8-25 0-016 70% 0% Night activity X2 = 0-13 0-723 60% 30% Ballooning Z = 123 0-221 60% 80% Niche breadth Z = 2-85 0-004 15-818 38-272 Local activity Z = -0-32 0-752 5-492 5-797

COMPARATIVE ANALYSIS Discussion

Direct comparisons among congeneric spider species ( LOCAL AND REGIONAL EFFECTS ON SPECIES 2006 The Authors. confirmed community-wide shifts in the above reported Journal compilation DIVERSITY life-history traits (Table 5). Indicator species from ? 2006 British dunes were characterized As has also been documented for Ecological Society, dynamic grey by larger agriculturallandscapes Journal of Applied body sizes, longer generation times, a higher frequency (Schmidt et al. 2005), loss of sand dynamics in coastal Ecology, 43, of winter activity and narrower niche breadths grey dunes affected patterns of local spider diversity, 735-747 (Fig. 3). but in a landscape-relatedway. Under decreasing sand

This content downloaded from 157.193.10.229 on Tue, 07 Jul 2015 13:14:03 UTC All use subject to JSTOR Terms and Conditions 743 More complex vegetation promotes coexistence of spe- (a) Sand dynamics ? / cies using different structures for activities such as web and Scheidler Uetz spiders 10 * ./building (Canard 1990; 1990; 1990; Dennis, Young & Bentley 2001). Patterns in assem- blage (dis)similarity showed different relationships, 8 * / with the level of local sand dynamics depending on landscape dynamics (and hence stabilizationhistory): in 6 */ fdynamic dune landscapes, P-diversitydecreased signif- icantly and approached regional (y) diversity under local grey dune stabilization; in stabilized landscapes, 4 * p-diversity was not affected by local sand dynamics. This suggests that all species from a regional pool are 2 able to coexist in stabilized fragments within dynamic landscapes. In both landscape types, patches were Localactivity, P > 0 05 poorer in species and more dissimilar under increasing 2 4 6 8 10 12 sand dynamics. As dynamic dunes consist of open landscapes (Provoost 2004), high levels of landscape (b) connectivity for grey dune (Bonte et al. n) 14 2003c; Maes et al. in press) appear to affect diversity (Chase 2003), allowing specialist species from 12 patterns a) dynamic habitats to coexist with species from stable ,) -o 10 grey dunes under decreasing sand dynamics. Hence a) et al. to be . mass effects (Leibold 2004) appear respon- U0 8 ._, /* sible for the high degree of local coexistence. In con- E e/ *trast, dispersal limitation as a result of reduced patch 2 6 / connectivity is likely to cause higher dissimilarityunder U) a) 0 stabilized local conditions in stabilized landscapes. *' 4 a) Q. According to Leibold et al. (2004), such a pattern is in c) 2 accordance with the patch-dynamic paradigm, where Body size, P > 0.05 local species diversity in identical patches is limited by ' ' '' '' ' dispersal while spatial dynamics are determined by 2 4 6 8 10 12 14 16 local extinction and colonization. t Qo For Thomisidae and Lycosidae, species adapted to (c) D / dynamic conditions disappeared from the regional spe- 70- cies pool in stabilized landscapes, reflected by a lower more 60 - p/-diversityand a homogeneous species composi- tion under high levels of sand dynamics in stabilized 50- coastal dune landscapes. As a result, local (oc)diversity may be constrained by increased sand dynamics at a - 40 local scale, and by decreased sand dynamics at a land- scape scale. 30U-

0 20- PATTERNS IN DIVERSITY AS A RESULT OF ENVIRONMENTAL CONSTRAINTS 10- / Niche breath,P > 005 Changes in species diversity were linked to community- C 10 20 30 40 50 60 70 80 wide character shifts in most of the spider families Species fromdynamic grey dunes studied: species living in highly dynamic habitats showed adaptations to avoid extreme microclimatic Fig. 3. Comparative analysis of spider life history traits for conditions, i.e. reflectedby higher frequencies of winter species indicative of dynamic (x-axis) and stabilised (y-axis) sizes, longer generation times and coastal grey dune fragments. (a) local activity, (b) body size, activity, larger body (c) niche breadth. In each graph the line indicates identical a burrowing life-style for wolf spiders. Only in Gna- values for species from both situations. phosidae, which are dominated by night-active species, ? 2006 The Authors. opposite patternswere observedfor body size, suggesting Journal compilation that body size is only constrained in day-active species ( 2006 British local increased in both stable that are to intense solar radiation. Ecological Society, dynamics, (ac) diversity subject However, Journalof Applied and dynamic landscapes, probably as a result of higher these contrasting patterns disappeared when phylo- Ecology, 43, nutrient availability resulting in a higher and more genetic dependence was taken into account through 735-747 complex vegetation structure (Provoost et al. 2004). comparative analysis, indicating bias as a result of the

This content downloaded from 157.193.10.229 on Tue, 07 Jul 2015 13:14:03 UTC All use subject to JSTOR Terms and Conditions 744 occurrence of species belonging to one genus under species, in which congenerics are indicative of stabil- D. Bonte, L. Lens & stabilized conditions (i.e. Drassodes). Shifts in aerial ized or dynamic sites, showed comparable patterns. As J-P Maelfait dispersal ability showed different patterns among fam- expected,traits prevailing under high local sand dynamics ilies. Assemblages in landscape fragments experiencing were related to desiccation resistance (body size) or high local sand dynamics were dominated by lycosid desiccation avoidance (winter activity, burrowing life species that did not show ballooning, while contrasting style). These shifts resemble those in desert spiders, as patterns were observed for Erigoninae in dynamic documented by Henschel (1997), and adaptations in landscapes. The latter resulted from the presence of psammophilious spiders appear similar in subtropical early pioneer species typical of disturbed conditions, desert and temperate dune ecosystems, indicating com- i.e. Erigone and Oedothoraxspecies, which are frequent parable environmental constraints. ballooners (Thomas, Brain & Jepson 2003). These spe- cies are widely distributed in dynamic habitats, such as ENVIRONMENTAL CONSTRAINTS AND young dune slacks (Bonte, Baert & Maelfait 2002), and SENSITIVITY TO HABITAT FRAGMENTATION invade habitat outside their fundamental niche (mass effects) (Bonte et al. 2004a). In stabilized dunes, source Species from fragments characterized by high levels of populations of the latter species are almost absent sand dynamics showed larger body sizes and longer (Provoost 2004), thereby reducing their presence in generation times, while summer-active species tended (rare) fragments under aeolic dynamics. Erigoninae to become residentialafter sand stabilization.Although species occurring in dynamic fragments tended to show a shift in average activity density was recorded at the narrower niches. Similar patterns were observed for family level in species from four families, comparative Gnaphosidae, Lycosidae and Thomisidae in dynamic analysis did not suggest lower activity density levels in landscapes,because stenotopic species were absent from dynamic sites. Because bias as a result of species-specific stabilized landscapes (e.g. Haplodrassus dalmatensis activity patterns is minimized in the latter kind of ana- (L. Koch 1866), Walckenaeria stylifrons (O.-P. lysis, we cannot conclude that population sizes decline Cambridge 1875), Xerolycosaminiata (C.L. Koch 1834) under increasing aeolic dynamics (under the assump- and Xysticussabulosus (Hahn 1832);Bonte et al. 2003a). tion that activity is related to population density; Comparativeanalyses of ecological traits revealedshifts Maelfait & Baert 1975). However, assemblages experi- in body size, generationtime, seasonal activity and niche encing high sand dynamicswere characterizedby species breadth, suggesting that patterns in local diversity with smaller niche breadths. Consequently, connectiv- resulted from species sorting (Leibold et al. 2004) and ity levels within fragmented landscapes with matrices were constrained by local sand dynamics (i.e. favouring consisting of heterogeneous vegetation can be expected only pre-adapted species). Winter-active species were to be lower (Bonte et al. 2003a). Furthermore, species not entirely replaced by summer-active species under from grey dunes with narrow niches showed a lower stabilized conditions, hence providing evidence for propensity to undertake aerial dispersal (Bonte et al. constrained species sorting (Leibold et al. 2004) in 2003d). In this study, we were only able to distinguish which natural disturbance (sand dynamics) allowed between species with and without (reported) balloon- species with distinct ecological traits to persist only ing dispersal ability, possibly resulting in a failure to under these conditions. Hence complete community- detect community-wide shifts in ballooning propensity wide character displacement appeared restrictedunder in relation to local sand dynamics.Species from dynamic disturbed (non-equilibrium)conditions (Leibold 1998). sites with larger body sizes (Boughton & Malvadkar Consequently, trade-offs between ecological traits at a 2002; Henle et al. 2004), longer generation times and a local scale will only be expressed at low levels of distur- higher degree of habitat specialization (which is related bance, allowing coexistence between good competitors to dispersal rates; Hanski 1999; Malmqvist 2000; and good colonizers.This in turn may resultin high levels Thomas et al. 2001; Bonte et al. 2003d) can be expected of (x-diversity,approaching regional diversity if patch to be more vulnerable and more prone to extinction connectivity is high (dynamic landscapes). Under high than their counterparts from stabilized fragments. In local sand dynamics, communities will show higher particular, species with a burrowing life style can be levels of dissimilarity compared with more stabilized expected to disappear because of increasing soil devel- conditions because of the strict occurrenceof stenotopic opment and soil hardness (Bonte et al. 2003b). In addi- (pre-) adapted species. Under these conditions, local tion, if sand dynamics diminish at the landscape level, trade-offsseem less pronounced (Kneitel & Chase 2004), fragmentationof grey dune fragmentsmay increase and restricting persistence to good competitors (large local sand dynamics may further decrease (Provoost specialists) with low dispersal ability.In Erigoninae, but et al. 2004). Conserving sand dynamics at the land- ? 2006 The Authors. not in other families, good competitors (habitat spe- scape level is therefore of prime importance when aim- Journal compilation cialists) may coexist with good colonizers in dynamic to conserve psammophilous spider species ( 2006 British ing typical landscapes because of mass effects. within local grey sand dune The absence Ecological Society, assemblages. Journal of Applied Analyses at the family level, however, may obscure of specializedGnaphosidae, Lycosidae and Thomisidae Ecology, 43, more detailed shifts in ecological traits. However, species in local assemblages with strong sand dynamics 735-747 comparison of ecological traits within couples of sister in stabilized landscapes (Bonte et al. 2003a) illustrates

This content downloaded from 157.193.10.229 on Tue, 07 Jul 2015 13:14:03 UTC All use subject to JSTOR Terms and Conditions 745 the type of adverse impact that can occur when such Water Company North Holland) for access and help Sand dynamics factors are not taken into consideration. with pitfall sampling. The first author holds a postdoc and spiders fellowship at the Fund for ScientificResearch- Flanders (FWO). MANAGEMENT IMPLICATIONS

Our study provides evidence that spider species inhabit- References ing grey dune fragments subject to high local sand dynamics undergo shifts in life-history characteristics Alderweireldt, M. & Maelfait, J.-P. (1990) Catalogus Vande that render them more vulnerable to extinction than Spinnen VanBelgie. Deel VII. Lycosidae. Studiedocumenten their counterparts from stabilized fragments (Henle van het KBIN 61. Royal Belgian Museum of Natural et al. 2004). Whereas grey dune fragments impose Sciences, Brussels, Belgium. Almquist, S. (1970) Thermal tolerances and preferences of levels of isolation on strong specialist arthropod some dune-living spiders. Oikos, 21, 230-236. populations (Bonte et al. 2003c; Maes et al. in press), Almquist, S. (1971) Resistance to desiccation in some dune community-wide shifts in life-history traits appear less living spiders. Oikos, 22, 225-229. pronounced as a result of the absence of many special- Bach, C.E. (2001) Long-term effects of insect herbivory and sand accretion on succession on sand dunes. ized species in stabilized landscapes. Hence efforts to plant Ecology, 82, 1401-1416. conserve should focus highly specialized species mainly Baert, L. (1996) Catalogus van de spinnen van Belgie. Deel on maximizing population sizes (hence minimizing XIV. Linyphiidae (Erigoninae). Studiedocumenten van het chances of stochastic extinctions) by enlarging the KBIN 82. Royal Belgian Museum of Natural Sciences, remainingfragments and stimulatinglocal sand dynamics. Brussels, Belgium. E.M. & G.S. Because large-scale sand dynamics (at the landscape Bell, J.R., Bohan, D.A., Shaw, Weyman, (2005) Ballooning dispersal using silk: world fauna, phylogenies, level) have an explicit positive effect on both grey dune genetics and models. Bulletin of Entomological Research, connectivity and the magnitude of local sand dynamics 95,69-114. (Provoost et al. 2004), the restoration of local sand Bennet, P.M. & Owens, I.P.F.(1997) Variationin extinction risk dynamics should be integrated with programmes that among birds:chances or evolutionarypredeposition. Proceed- the Series 401-408. aim to conserve or stimulate sand dynamics at a land- ings of Royal Society London, B, 264, Bessa-Gomez, C., Danek-Gontard, M., Cassey, P., Moller, level. The latter could be achieved scape by restoring A.P, Legendre, S. & Clobert, J. (2003) Mating behaviour beach-dune transitions, allowing net sand replacement influences extinction risk: insights from demographic into the dune area and/or by connecting large dynamic modelling and comparativeanalysis of avian extinction risk. dunes through the dismantlement of urban infrastruc- Annales Zoologica Fennici, 40, 231-245. R.L. & S.M. of related tures such as roads. In addition, if sand dynamics were Bevill, Louda, (1999) Comparisons and common speciesin the study of plant rarity. Conservation restored at a local level this would by creating blowouts, Biology, 13, 493-498. most probably be accompanied by additional anthro- Blondel, J. (2003) Guilds or functional groups: does it matter? pogenic disturbance such as trampling (by humans or Oikos,100, 223-231. cattle) or removal of the top soil layer and vegetation. Bonte, D., Baert, L., Lens, L. & Maelfait, J.-P. (2004a) Effects of aerial habitat and Although such disturbance would create similar en- dispersal, specialisation, landscape structureon spiderdistribution across fragmented grey dunes. vironments comparable to those under natural sand Ecography, 27, 343-349. dynamics, it is likely that specialized spider species Bonte, D., Baert, L. & Maelfait, J.-P. (2002) Spider assemblage would be negatively affected by this additional distur- structure and stability in a heterogeneous coastal dune bance, although the empiricalevidence for this is mostly system (Belgium). Journal of Arachnology, 30, 331-343. Van I. & lacking (but see Koehler et al. 1996). In conclusion, Bonte, D., Criel,P., Vanhoutte, L., Thoumout, Maelfait, J.-P.(2004b) The importanceof habitatproductivity, stability ensuring conservation of sand dynamics at a landscape and heterogeneity for spider species richness in coastal grey level rather than at a local level is of prime importance dunes along the North Sea and its implications for conser- when aiming to conserve typical psammophilous spider vation. Biodiversityand Conservation,13, 2119-2134. species within local assemblages in grey dune habitats. Bonte, D., Criel, P., Van Thournout, I. & Maelfait, J.-P. (2003a) Regional and local variation of spider assemblages (Araneae) from coastal grey dunes along the North Sea. Acknowledgements Journal of Biogeography, 30, 901-911. Bonte, D., Dekoninck, W., Provoost, S., Cosyns, E. & The authors would like to thank Angelica Alcantara Hoffmann, M. (2003b) Microgeographical distribution of for separating spiders from other pitfall catches and ants (Hymenoptera: Formicidae) in a coastal dune grass- to the soil P. Criel, N. Vandenbroecke, L. Vanhoutte and I. Van land and their relation structure and vegetation. Biology, 53, 367-377. Thornout for their with identification. Vari- help spider Bonte, D., Hoffmann,M. & Maelfait,J.-P. (2000) Seasonal ous undergraduate students helped with measurement and diurnalmigration patterns of the spider fauna of coastal ? 2006 The Authors. and identification of juvenile spiders during practical grey dunes. Ekologia Bratislava, 19, 5-16. Journal compilation courses. We also would like to thank T. Mougey (Parc Bonte,D., Lens,L., Maelfait, J.-P., Hoffmann, M. &Kuijken, E. ? 2006 British Naturel Regional du Boulonnais, Eden 62), J.-L. (2003c) Patch quality and connectivity influence spatial Ecological Society, dynamics in a dune wolfspider. Oecologia, 135, 227-235. Herrier Administration Nature Journal of Applied (Flemish Community, Bonte, D., Maelfait, J.-P. & Baert, L. (2004) Determining Ecology, 43, AMINAL), M. Hootsman (Amsterdam Water Supply spider species richness in fragmented coastal dune habitats 735-747 dunes) and P. Zumkher and C. De Vries (Provincial by extrapolation and estimation (Araneae). European

This content downloaded from 157.193.10.229 on Tue, 07 Jul 2015 13:14:03 UTC All use subject to JSTOR Terms and Conditions 746 Arachnology 2002 (ed. F. Samu & Cs. Szinetar), pp. 189- Kolar, C.S. & Lodge, D.M. (2002) Ecological predictions and D. Bonte, L. Lens & 197. AKAPRINT kft, Budapest, Hungary. risk assessment for alien fishes in North-America. Science, Bonte, D., Maelfait, J.-P. & Hoffmann, M. Aeronau- 298, 1233-1236. J-P Maelfait (1998) tisch actieve spinnen in een duingebied te Oostduinkerke. Kowalchuk, G.A., De Souza, F.A. & Van Veen, J.A. (2002) Nieuwsbrief Van de Belgische Arachnologische Vereniging, Community analysis of arbuscular mycorrhizal fungi 13, 77-85. associated with Ammophilaarenaria in Dutch coastal sand Bonte, D. Vandenbroecke,N. Lens, L. & Maelfait, J.-P.(2003d) dunes. Molecular Ecology, 11, 571-581. Low propensity for aerial dispersal in specialist spiders from Leibold, M.A. (1998) Similarity and local co-existence of fragmented landscapes. Proceedings of the Royal Society species in regional biotas. EvolutionaryEcology, 12, 95-110. London, Series B, 270, 1601-1607. Leibold, M.A. & Miller, TE. (2004) From metapopulationsto Boughton, D. & Malvadkar, U. (2002) Extinction risk in suc- metacommunities.Ecology, Genetics,and Evolutionof Meta- cessional landscapes subjects to catastrophic disturbances. populations(eds I. Hanski & 0. E. Gaggiotti), pp. 133- Conservation Ecology, 6, 1-24. http://www.consecol.org/ 150. Elsevier/AcademicPress, Amsterdam,the Netherlands. vol6/iss2/art2/, accessed 1 June 2005. Leibold, M.A. Holyoak, M. Mouquet, N. Amarasekare, P. Canard, A. (1990) Heathland spider communities a functional Chase, J.M. Hoopes, M.F. Holt, R.D. Shurin, J.B. Law, R. group study. Acta Zoologica Fennica, 190, 45-50. Tilman, D. Loreau, M. & Gonzalez, A. (2004) The meta- Chase, J.M. (2003) Community assemblage: when should community concept: a frameworkfor multi-scalecommunity history matter? Oecologia, 136, 489-498. ecology. Ecology Letters, 7, 601-613. Dennis, P.,Young, M.R. & Bentley,C. (2001)The effectsof varied Loreau, M. (2000) Are communities saturated?On the relation- grazing management on epigeal spiders, harvestmen and ship between a, Pand y diversity. Ecology Letters, 3,73-76. pseudoscorpionsof Nardusstricta grasslandin upland Scot- Lytle, D.A. (2001) Disturbance regimes and life-history land. Agriculture,Ecosystems and Environment,86, 39-57. evolution. American Naturalist, 157, 525-536. Dufrene, M. & Legendre, P. (1997) Species assemblages and Maelfait,J.-P. & Baert, L. (1975) Contributionto the knowledge indicator species: the need for a flexible asymmetrical of the arachno- and entomofauna of differentwood habitats. approach. Ecological Monographs, 67, 345-366. I. Sampled habitats, theoretical study of the pitfall method, Foelix, R.F. (1996) Biology of Spiders.Oxford University Press, survey of captured taxa. Biologisch JaarboekDodonaea, 46, Oxford, UK. 179-196. Franks, S.J. & Peterson, C.J. (2003) Burial disturbance leads Maes, D. Ghesquiere, A. Logie, M. & Bonte, D. (2006) to facilitation among coastal dune plants. Plant Ecology, Habitat use and mobility of two threatened coastal dune 168, 13-21. insects: implications for conservation. Journal of Insect Garcia-Mora,M.R., Gallego-Fernandez,J.B. & Garcia-Novo,F. Conservation,10, 105-115. (1999) Plant functional types in coastal foredunes in rela- Malmqvist, B. (2000) How does wing length relate to distri- tion to environmental stress and disturbance. Journal of bution patterns of stoneflies (Plecoptera) and mayflies VegetationScience, 10, 27-34. (Ephemeroptera)?Biological Conservation,93, 271-276. Gaston, K.J. & Kunin, WE. (1997) Rare-common differences: Marchinko, K.B. Nishizaki, M.T. & Burns, K.C. (2004) an overview. The Biology of Rarity (eds W.E. Kunin & Community-widecharacter displacement in barnacles:a new K.J. Gaston), pp. 12-29. Chapman & Hall, London, UK. perspectivefor past observations.Ecology Letters, 7,114-120. Hanggi, A., St6ckli, E. & Nentwig, W. (1995) Habitats of Martinez, M.L. & Maun, M.A. (1999) Responses of dune Central EuropeanSpiders. Characterisationof the Habitats mosses to experimental burial by sand under natural and of the most AbundantSpider Species of Central Europeand greenhouse conditions. Plant Ecology, 145, 209-219. Associated Species. Miscellanae Faunistica Helvetica 4. Maun, M.A. (1998) Adaptations of plants to burial in coastal Centre Suisse de cartographie de la faune, Neuchftel, sand dunes. Canadian Journal of Zoology, 76, 713-738. Switzerland. Moran, M.D. (2003) Arguments for rejecting the sequential Hanski, I. (1999) MetapopulationEcology. Oxford University Bonferroni in ecological studies. Oikos, 100, 403-405. Press Inc., New York, Oxford. Pianka, E.R. (1994) Evolutionary Ecology, 5th edn. Harper Henle, K., Davies, K.F., Kleyer, M., Margules, C. & Settele, J. Collins, New York, NY. (2004) Predictors of species sensitivity to fragmentation. Provoost, S. (2004) Het kustecosysteem. LevendeDuinen. Een Biodiversityand Conservation,13, 207-251. Overzicht Vande Biodiversiteit Aan de Vlaamse Kust (eds Henschel, J.R. (1997) Psammophily in Namib Desert spiders. S. Provoost & D. Bonte), pp. 6-54. Mededelingen van het Journal of Arid Environment,37, 695-707. Instituut voor Natuurbehoud 22. Flemish administration Janssens, M. (1993) Catalogus van de spinnen van Belgie. Deel Press, Brussels, Belgium. XIII. Thomisidae. Studiedocumenten van het KBIN 43. Provoost, S. Ampe, C. Bonte, D. Cosyns, E. & Hoffmann, M. Royal Belgian Museum of Natural Sciences, Brussels, (2004) Ecology, management and monitoring of grey dunes Belgium. in Flanders. Journal of Coastal Conservation,10, 33-42. Jennings, S., Dinmore, T.A., Duplisea, D.E., Warr, K.J. & Ransy, M. & Baert, L. (1991) Catalogus Vande Spinnen Van Lancaster, J.E. (2001) Trawling disturbance can modify Belgie. Deel VII. Gnaphosidae Studiedocumenten van het benthic production processes. Journal of Animal Ecology, KBIN 64. Royal Belgian Museum of Natural Sciences, 70, 459-474. Brussels, Belgium. Jun, R., Clement, R.B. & Roze, F. (2004) Primary succession Reed, R.N. & Shine, R. (2002) Lying in wait for extinction: of bryophyte and lichen communities in non-forested ecological correlates of extinction risk among Australian Atlantic coastal dunes: the example of the Pointe d'Arcay snakes. ConservationBiology, 16, 451-461. (France). Nova Hedwigia, 78, 453-468. Roberts, M.J. (1985) The Spiders of GreatBritain and Ireland. Kneitel, J.M. & Chase, J.M. (2004) Trade-offs in community Volume II. Linyphiidae and Check List. Harley Books, ? 2006 The Authors. ecology: linking spatial scales to species coexistence. Ecology Colchester, UK. 69-80. Roberts, M.J. Tirions Tirion, Baarn, The Journal compilation Letters, 7, (1998) Spinnengids. J. & H. Netherlands. ( 2006 British Koehler, H.H., Harder, H., Meyerdirks, Voigt, (1996) The effect of trampling on the microarthropod fauna of SAS (2003) SAS9.1. SAS Institute Inc., Cary, NC. Ecological Society, dune sediments.A case study from Jutland,Denmark. Studies Schaefer, M. (1976) Experimentelle Untersuchungen zum Journal of Applied in European Coastal Management (eds P.S. Jones, M.G. Jahrescyclusund zur Uberwinterungvon spinnen(Araneida). Ecology, 43, Healy & E.T. Williams), pp. 221-231. Samara Publishing Zoologischen Jahrbuch Fur Systematiek, Okologie und 735-747 Limited, Cardigan, UK. Geografieder Tiere, 103, 127-289.

This content downloaded from 157.193.10.229 on Tue, 07 Jul 2015 13:14:03 UTC All use subject to JSTOR Terms and Conditions 747 Scheidler,M. (1990) Influence of habitat structureand vegeta- Vaughan,I.P & Ormerod,S.J. (2005) The continuingchallenges Sand dynamics tion architecture on spiders. Zoologischer Anzeiger, 225, of testing species distribution models. Journal of Applied 333-340. 720-730. and spiders Ecology, 42, Schluter, D. (2000) Ecological character displacement in Verbeke, G. & Molenberghs, G. (2000) Linear Mixed Models adaptive radiation. American Naturalist, 156, 4-16. for LongitudinalData. SpringerSeries in Statistics.Springer, Schmidt, M.H., Roschewith, I., Thies, C. & Tscharntke, T. New York, NY. (2005) Differential effects of landscape and management on Whittaker,R.H. (1972) Evolution and measurementof species diversity and density of ground-dwellingfarmland spiders. diversity. Taxon, 21, 213-251. Journal of Applied Ecology, 42, 281-287. Wilson, J.B. & Sykes, M.T. (1999) Is zonation on coastal sand Snyder, R.A. & Boss, C.L. (2002) Recovery and stability in dunes determined primarily by sand burial or by salt spray? barrierisland plant communities.Journal of CoastalResearch, A test in New Zealand dunes. Ecology Letters, 2, 233- 18, 530-536. 236. Thomas, C.D., Bodsworth, E.J., Wilson, R.J., Simmons, A.D., Davies, Z.G., Musche, M. & Conradt, L. (2001) Ecological Received 7 July 2005; final copy received 24 February 2006 and evolutionary processes at expanding range margins. Editor. Davy McCracken Nature, 411, 577-581. Thomas, C.F.G., Brain, P. & Jepson, P.C. (2003) Aerial activ- ity of linyphiid spiders: modelling dispersal distances from Supplementarymaterial meteorology and behaviour. Journalof AppliedEcology, 29, The following supplementary material is available 526-539. as part of the online article (full text) from http:// Uetz, G.W. (1990) Habitat structure and spider foraging. Habitat Structure. The Physical Arrangementof Objects in w.w.w.blackwell-synergy.com. Space (eds S.S. Bell, E.D. McCoy & H.R. Mushinsky), pp. 325-348. Chapman & Hall, London, UK. Appendix Sl. Species list and trait values.

? 2006 The Authors. Journal compilation ( 2006 British Ecological Society, Journal of Applied Ecology, 43, 735-747

This content downloaded from 157.193.10.229 on Tue, 07 Jul 2015 13:14:03 UTC All use subject to JSTOR Terms and Conditions