Dispersal limitation shapes distance‐decay patterns of European spiders at the continental scale

Global Ecol Biogeogr. 2024;33:e13810.   | 1 of 13 https://doi.org/10.1111/geb.13810 wileyonlinelibrary.com/journal/geb Received:20April2023 | Revised:23December2023 | Accepted:18January2024 DOI:10.1111/geb.13810 RESEARCH ARTICLE Dispersal limitation shapes distancedecay patterns of European spiders at the continental scale Ramiro MartínDevasa1,2 | Alberto JiménezValverde3,4 | Fabien Leprieur5 | Andrés Baselga1 | Carola GómezRodríguez6 This is an open access article under the terms of the Creative Commons Attribution License, which permits use, distribution and reproduction in any medium, provided the original work is properly cited. © 2024 The Authors. Global Ecology and BiogeographypublishedbyJohnWiley&SonsLtd. Andrés Baselga and Carola GómezRodríguez contributed equally to this work. 1CRETUS,DepartmentofZoology, Genetics and Physical Anthropology, UniversidadedeSantiagodeCompostela, SantiagodeCompostela,Spain 2DepartmentofGeosciencesand Geography, University of Helsinki, Helsinki, Finland 3DepartamentodeCienciasdelaVida, Facultad de Ciencias, Universidad de Alcalá,AlcaládeHenares,Spain 4DepartmentofBiogeographyandGlobal Change, Museo Nacional de Ciencias Naturales(MNCN-CSIC),Madrid,Spain 5Marine Biodiversity Exploration and Conservation, University of Montpellier, CNRS,Ifremer,IRD,Montpellier,France 6CRETUS,DepartmentofFunctional Biology(AreaofEcology),Universidade deSantiagodeCompostela,Santiagode Compostela,Spain Correspondence RamiroMartín-Devasa,CRETUS, DepartmentofZoology,Geneticsand Physical Anthropology, Universidade de SantiagodeCompostela,Santiagode Compostela,Spain. Email: [email protected] Funding information Ministerio de Educación, Cultura yDeporte,Grant/AwardNumber: FPU17/03016; Ministerio de Ciencia eInnovación,Grant/AwardNumber: PID2020-116587GB-I00 Handling Editor:BrodySandel Abstract Aim: To assess the relative relevance of dispersal limitation and species sorting as drivers of spatial turnover between spider faunas of European territories. Location: Continental Europe. Time period: Present. Major taxa studied: Spiders(OrderAraneae). Methods: Weanalysedhowdistance-decaypatternsdifferbetweennorthernand southern Europe (broadly, territories covered vs. not covered by ice sheets during the lastglacialmaximum,respectively)in15spiderfamilies,usingstandardizeddistances toallowadirectcomparisonbetweenparameters(i.e.slopeandintercept)ofclimatic andspatialdistance-decaymodels.Thus,weassessedNorth–Southdifferencesinparameters of spatial and, independently, climatic distancedecay models, and whether those differences are explained by familyspecific traits related to dispersal ability. Results: Climatic and spatial distancedecay patterns are very similar in northern Europe,whereclimaticandspatialdistancesarehighlycorrelated.Incontrast,slopes are steeper in spatial than in climatic distancedecay curves in southern Europe, where climatic and spatial distances are decoupled. Moreover, family traits related todispersalabilityexplainedtheNorth–Southdifferenceinspatialdistance-decay slopes, as well as the amount of nestednessresultant dissimilarity between southern and northern spider faunas. Main conclusions: Ourresultssuggestthatdifferencesinbetadiversitypatternsbetween northern and southern Europe reflect the strength of dispersal limitation in spiders, which varies across families and leads to different degrees of disequilibrium withcurrentclimaticconditionsdependingonthetaxon.Moreover,intheSouthof Europe, where spatial and climatic distances are uncorrelated, spatial distancedecay models are steeper and have larger explanatory power than climatic distancedecay models, which suggests that dispersal limitation is the main factor shaping current beta diversity patterns of European spiders at the continental scale. 2 of 13 | MARTÍN-DEVASA et al. 1 | INTRODUCTION Understanding the processes behind differences in species compositionbetweensites(i.e.betadiversity)isacentralquestionin ecology and biogeography. For this reason, beta diversity patterns have been widely studied across a wide range of territories, scales, times and taxa (e.g. Ávila et al., 2020; Pavlek&Mammola, 2020; Soininenetal.,2007;Steinbaueretal.,2012).Atlargescales,compositional differences between sites are commonly explained by two majorprocesses,dispersallimitationandspeciessorting(Nekola& White,1999;Soininenetal.,2007),whicharenotmutuallyexclusive (Gravel et al., 2006).Underdispersallimitation,evenifareaswith suitable conditions are available, species may not have the capacity to move across the required distances to colonize those areas, so their spatial distributions would not be in equilibrium with environmental conditions. As a result, the more spatially distant two sites are, the more dissimilar their communities are as well (Hubbell, 2001; Nekola&White,1999).Inturn,underspeciessorting,speciesare present or absent in a site depending on its biotic and abiotic characteristics (Leibold et al., 2004)and,inconsequence,speciesdistributionsareconstrainedbyenvironmentalconditions(e.g.climate)and bioticinteractions(e.g.competitiveexclusion).Inthiscase,themore different the biotic and abiotic characteristics of two sites, the more dissimilartheircommunities(Nekola&White,1999). The interplay between these two processes (dispersal limitation andspeciessorting)canleadtoperdurableeffectsofpastclimatic eventsonpresent-daybiodiversitypatterns(Svenningetal.,2015). Pleistocene glaciations are one of the most important historical climatic events explaining the current distribution of species in Europe (Hewitt, 1996, 2000).Duringthisperiod,northernEuropewascovered by ice sheets, and most of the fauna and flora were restricted to glacial refugia, mainly in the Mediterranean region (Hewitt, 1999). Whentheicesheetsretreated,specieswereabletoexpandtheir ranges and colonize territories in northern latitudes. Consequently, the flora and fauna of northern Europe are the result of a postglacial recolonization process, which may be still ongoing for some taxa(Gómez-Rodríguez&Baselga,2018;Svenning&Skov,2007). Therefore, largescale diversity patterns in Europe are expected to be driven by these two major processes, whose relative relevance would varydependingonthe biologicalgroup:(i)speciessorting, when species have recolonized all the regions where biotic and abiotic conditions are suitable for them, and (ii) dispersal limitation, when species have not been able to reach all the regions where conditions are suitable for them. This is because their distribution ranges are lagging behind the warming climatic conditions since the lastglacialmaximum,LGM(Svenning&Skov,2007). The effects of longterm dispersal limitation and species sorting in the spatial configuration of European biodiversity have been assessed in a wide range of organisms, especially vertebrates and plants (e.g. Astorga et al., 2012; Leprieur et al., 2009; Svenning et al., 2011).Ininvertebrates,theseeffectsarelessstudied,butpreviousanalysessuggestamajorroleofdispersallimitation.Specifically, GómezRodríguez and Baselga (2018) showed that beta diversity patterns are markedly different between northern and southern Europe in beetle groups with low dispersal ability, evidencing the strong imprint of past climatic events in current diversity patterns. WhentheextensionoftheicesheetsduringtheLGMisconsidered, European territories can be classified in northern territories (>48°N) and southern territories (<48°N),broadlyreflectingwhetherthey were covered or not by ice sheets during LGM (GómezRodríguez &Baselga,2018; Hughes et al., 2016).Thus,Gómez-Rodríguezand Baselga (2018)modelledthedecreaseofcommunitysimilaritywith spatial distance, also referred to as ‘distancedecay relationship’ (Nekola&McGill,2014;Nekola&White,1999),insuchbeetlegroups and compared the slopes of distancedecay models between northernandsouthernEurope.TheobservedNorth–Southdifferencesin distancedecay patterns suggested that, in beetle groups with low dispersalability,mostspecieswererestrictedtotheSouthofEurope but a few species, that is, those with good dispersal ability, had been able to reach the North of Europe in a postglacial recolonization process. As a result, GómezRodríguez and Baselga (2018)inferredthat (i)thenorthernEuropeanfaunaisasubsetofthesouthernfauna, composed only of species with high dispersal ability able to colonize distantterritories;and(ii)thespeciescompositionissimilaracross northern territories, resulting in a flat distancedecay curve within the northern region (i.e. high community similarity over long spatial distances,resultinginaweaklongitudinalpattern).Incontrast,the dispersal limited species would show narrower distributions and wouldberestrictedtotheSouthofthecontinent,resultinginsteeper distance-decaycurvesintheSouththanintheNorth.Inotherwords, dispersal limitation would play a major role in determining the beetle fauna of northern Europe but would not shape the longitudinal diversitypatternacrossthisregion(Gómez-Rodríguez&Baselga,2018). Inadditiontospatialdistance-decaypatterns,itisalsopossible to assess the decay of community similarity with climatic distance (GracoRoza et al., 2022; Saito et al., 2015). Analyses of climatic distancedecay relationships aim to assess the effect of species sorting across climatic gradients, in an analogous way as analyses of spatial distancedecay patterns aim to assess the effect of dispersal limitation. Contrasting how well spatial and climatic distances explain community similarity has proven useful to discern the relative relevance of dispersal limitation and species sorting along climatic gradients in multiple biological systems (e.g. Baselga & Leprieur, 2015; Leprieur et al., 2009; Qian et al., 2005; RodriguezArtigas et al., 2016).However,becausethereisusuallysomedegree of correlation between spatial and climatic distances, it is necessary KEYWORDS beta diversity, disequilibrium, dispersal ability, distancedecay, European spiders, standardized distances 14668238, 2024, 4, Downloaded from https://onlinelibrary.wiley.com/doi/10.1111/geb.13810 by Universidade de Santiago de Compostela, Wiley Online Library on [22/04/2024]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License | 3 of 13 MARTÍN-DEVASA et al. to account for their covariation either using variance partitioning (Borcard et al., 1992;Smith&Lundholm,2010)oralternativeapproaches. Here, we standardized climatic and spatial distances in order to directly compare the slopes of spatial and climatic distancedecay models. Because the slopes of distancedecay models inform about the rate at which species are replaced with distance, using standardized distances allows comparing both rates. A complementary approach to assess whether diversity patterns may be the result of incomplete postglacial colonization is to assess whether differences between northern and southern European faunas are related to nestedness (i.e. the northern fauna is a subset ofthesouthernfauna)orspatialturnover(i.e.northernandsouthernfaunasareconstitutedbyuniquesetsofspecies)incommunity composition (Baselga, 2010).Ifnorthernfaunaswerenestedwithin southern faunas, it would point to postglacial recolonization processes primarily constrained by dispersal limitation (Dobrovolski et al., 2012),while,onthecontrary,spatialturnoverwouldsuggest species sorting processes along climatic gradients at continental scales.Inotherwords,ifdispersallimitationwasthemajorfactor shaping largescale beta diversity patterns in Europe, the northern fauna would be a subset of the southern one. Therefore, dissimilarity between southern and northern faunas would be mainly due to nestedness, and groups with poor dispersal ability would show higher differences,asfound in Dobrovolskietal. (2012) forNew World vertebrates. In contrast, if species were not dispersal limited, but their ranges were constrained by climatic tolerances, we would expect a larger contribution of spatial turnover to dissimilaritybetweensouthernandnorthernfaunas,asfoundbySvenning et al. (2011)forEuropeanmammals. Inthispaper,wewillfocusonEuropeanspiders.Spidersareone of the most diverse arthropod groups (with ca. 50,000 species describedtodate,WorldSpiderCatalog[2023] and are present in most terrestrialecosystems[Wise,1993]).Theirlong-distancedispersal ability depends on a method of passive aerial dispersal called ballooning, which is based on being carried by the wind supported by silkthreads(Duffey,1998).Thismethodcanbesoefficientthatsome spiders have been found at altitudes up to 4 km (Freeman, 1946; Glick, 1939)andcantravelmorethan1000 kmduringalifetime(Bell et al., 2005).Spidersarethusoneofthefirsttaxatocolonizeislands (Edwards&Thornton,2001;New&Thornton,1988).Despitethis high dispersal ability, the European spider fauna is expected to be in disequilibrium with climate, as not all species seem to have recolonized the climatically suitable northern territories (Koponen, 1991). Taxa with strong ballooning propensity might have reached the northern latitudes, but it is not clear how different spider groups have lagged behind the retreat of ice sheets (Koponen, 1991)andthusto which degree the legacy of past glaciations imprints the spatial structure of presentday assemblages. Ballooning tendency varies among spider families (Bell et al., 2005; Cardoso et al., 2011; Foelix, 2011; Szymkowiaketal.,2007),whichalsoexhibitdifferentecologicaland morphological characteristics (e.g. preferred habitat/vertical stratum,webbuildingbehaviourorbodysize)thatcanconditionballooning propensity as well (Blandenier, 2009;Dean&Sterling,1985; Larrivée&Buddle,2011).Thisvariationindispersal,morphological and ecological traits must affect the relative relevance of dispersal limitation and species sorting processes across taxa. As for species sorting processes, we focus on the climatic drivers of species composition but not on biotic factors, such as competitive exclusion or priority effects. This choice is based on two considerations. First, spider communities have been shown to be unsaturated. Natural communities usually harbour multiple species that exploit similar niches but are still able to coexist via spatiotemporal niche partitioning (Agnarsson et al., 2016; Mammola et al., 2020;Villanueva- Bonilla et al., 2019)andtheadditionofnon-nativespecieshasbeen shown not to impact the abundances of native species (Burger et al., 2001).Second,thespatialscaleatwhichweareanalysingthe variationinspeciescomposition(Europeancountries)makesiteven more unlikely for species interactions to play a relevant role in the observedpatterns(Araújo&Rozenfeld,2014).Theroleofbothdispersal limitation and species sorting along climatic gradients in driving spider communities has been studied at small spatial scales (e.g. Baldissera et al., 2012; Carvalho et al., 2011;Fernandez-Fournier& Avilés, 2018;Zhangetal.,2018)butnotatacontinentalscale,except for subterranean spiders (Mammola et al., 2019). Here, we aim to study the role of dispersal limitation and climatic constraints in shaping current beta diversity patterns in European spider faunas at large spatial scales (both in terms of grain and extent), by assessing the decrease of faunistic similarity between European countries. To do so, we compared the parameters (interceptandslope)ofspatialandclimaticdistance-decaycurvesof15 spider families in northern and southern Europe (i.e. territories covered and noncovered by ice sheets during the LGM, Figure 1)to assess(i)howtherelationshipbetweencompositionalsimilarityand spatial or climatic distance differs between northern and southern Europe.Wealsoassessed,foreachspiderfamily,(ii)whetherthe differences in species composition between northern and southernEuropearemorerelatedtonestednessorspatialturnover.We expectnestednesstobepredominantiftheNorth–Southfaunistic differences are caused by dispersal limitation, whereas turnover will be predominant if the differences are caused by species sorting.Finally,weanalysed(iii)whetherdispersal-relatedtraitsofspider families explained the differences in species composition and in distance-decayparametersbetweentheNorthandSouthofEurope. We expect a significant relationship between the differences in distancedecay parameters and dispersalrelated traits of spiders if dispersal limitation is the main process driving the composition of European spider assemblages. 2 | MATERIALS AND METHODS 2.1 | Distribution data Presence/absence tables for 15 spider families (3075 species in total) in 39 continental European territories were compiled from arane ae. nmbe. ch (Nentwig et al., 2022)(accession,July2022).We 14668238, 2024, 4, Downloaded from https://onlinelibrary.wiley.com/doi/10.1111/geb.13810 by Universidade de Santiago de Compostela, Wiley Online Library on [22/04/2024]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License 4 of 13 | MARTÍN-DEVASA et al. selected families Agelenidae, Araneidae, Clubionidae, Dictynidae, Gnaphosidae, Hahniidae, Linyphiidae, Liocranidae, Lycosidae, Philodromidae, Pholcidae, Salticidae, Tetragnathidae, Theridiidae and Thomisidae because of their large number of species, which allowed the robust computation of community similarity indices, their wide distribution through continental Europe and their different biological characteristics. The source provides a list of species by country (except for Russia, subdivided into five regions: North, Centre, South, East and West) based on published observations. Thus, most of our territories correspond to European countries, with someremarks.Duetotheirdifferentbiogeographiccharacteristics, islands were excluded from this study. Countries with an area of lessthan2000 km2(Monaco,Liechtenstein,SanMarinoandVatican City)wereexcludedfromourdatasettoavoidextremedifferences in area and, for the same reason, we maintained the subdivision of EuropeanRussiaintofiveterritories(North,Centre,South,Eastand West). Bosnia-Herzegovina and Croatia were considered a single territory because the latter surrounds the former. Kosovo was excluded from the analyses due to the absence of data on some taxa thatmaycausebiasincomputingcommunitysimilarity.Westudied the species–area relationship of the remaining territories to detect ifanyofthemhaveincompleteinventories.Wefittedalinearmodel tothelog(richness) ~ log(area)relationship,andidentifiedinfluential casesbasedonCook'sdistance(cut-off = 0.15,F1,37, 30th percentile[Ayindeetal.,2015]).MoldovaandEuropeanTurkeystoodout (Figure S1),suggestingtheirinventorieswerelargelyincomplete,so both territories were also excluded from subsequent analyses. 2.2 | Faunistic similarity and distancedecay models To assess the effect of dispersal limitation or species sorting along climatic gradients, we compared the decay of faunistic similarity between territories with spatial and, independently, with climatic distance for 15 spider families, and for northern and southern Europe separately. For each family, we constructed two presence/ absence tables, one for northern (territories with centroids at >48°) and one for southern Europe (territories with centroids at <48°).A latitude of 48° has been chosen as cutoff reference because it is the approximate latitude at which the ice sheets covered the continent during the LGM (Hughes et al., 2016).Eachpresence/absence table was used to compute the pairwise similarity between territoriesusingtheSimpson'ssimilarityindex(1-β sim)(Baselga,2010; Simpson,1943),withthefunctionbeta.pair of the R package betapart (Baselga et al., 2023;Baselga&Orme,2012).Spatialdistances between territories were computed as the geodesic distance between its centroids using the geodist function of the R package geodist(Padgham&Sumner,2021).Tocomputeclimaticdistances, we first extracted the mean value of six climatic variables for each territoryfromWorldclim(Hijmansetal.,2005):meanannualtemperature (Bio1), maximum temperature of the warmest month (Bio5),minimumtemperatureofthecoldestquarter(Bio6),annual precipitation (Bio12), precipitation of the wettest quarter (Bio16) andprecipitationofthedriestquarter(Bio17).Thesevariableswere submittedtoaprincipalcomponentanalysis(PCA)withavarimax rotationand thefirsttwo dimensions (explainedvariance = 100%) wereretained.Wecomputedtheclimaticdistancesbetweenterritories as the Euclidean distance between their respective PCA scores (Figure S2).Wealsocomputedthecorrelationbetweenspatialand climatic distances, both in northern and southern Europe, with a Mantel test using 1000 permutations. The degree to which spatial and climatic distances are correlated is relevant because, if correlation is low, the relationship between community similarity and spatial or climatic distance could be attributed to dispersal limitation or species sorting, respectively, and it would be independent of the alternative process. To make the parameters of climatic and spatial distancedecay models directly comparable, we first standardized spatial and climatic distances between 0 and 1. This allows, for example, FIGURE 1 Exampleillustratingthe decay of faunistic similarity with distance for Agelenidae using two territories, oneintheNorthandoneintheSouth, as reference. Thus, territory shades represent their faunistic similarity against areferenceterritory,SpainintheSouth and Norway in the North. The broken line(48°N)indicatedthecut-offusedto classify northern and southern territories. 14668238, 2024, 4, Downloaded from https://onlinelibrary.wiley.com/doi/10.1111/geb.13810 by Universidade de Santiago de Compostela, Wiley Online Library on [22/04/2024]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License | 5 of 13 MARTÍN-DEVASA et al. assessing whether the slopes are steeper in spatial or climatic distancedecay models, pointing to the preponderance of dispersal limitation or species sorting, respectively. Each standardized distance value (ds)wascomputedasds = (dimin(d))/(max(d)-min(d)) where di is the original spatial or climatic distance value, and d is thevectorofalloriginalspatialorclimaticdistances.Standardized spatial and climatic distancedecay models were built fitting powerlaw, negative exponential and Gompertz functions (MartínDevasaetal.,2022a)totherelationshipbetweensimilarityand standardized spatial/climatic distances with the decay.model function of the R package betapart. The negative exponential model wasselectedforsubsequentanalysesasitshowedthelowestAIC values in most cases (Tables S1 and S2inSupplementaryMaterial). The differences between spatial and climatic distancedecay parameters in each spider family were independently assessed for theNorthandtheSouthwiththe zdep statistic, a test specifically designed to compare parameters of models fitted with pairwisedependentdata(Martín-Devasaetal.,2022b),usingthezdep function(with1000resamples)oftheRpackagebetapart. Additionally, to control for the evolutionary relationships of spider families, we also tested for the difference in the average of intercepts and slopes between standardized spatial and climatic distancedecay models with a phylogenetic paired ttest using the phyl.pairedttest of the R package phytools (Revell, 2012).Tothisend,aphylogenetic tree for the 15 spider families included in our analyses was obtained by pruning the phylogenetic tree in MacíasHernández et al. (2020a),keepingonlyonespeciesperfamily.Wechosecommon species with a wide distribution across Europe (Table S3). To compute the dissimilarity between northern and southern faunas as a whole, the territorylevel presence/absence tables were collapsed into a new table coding the presence/absence of each species in two larger regions, northern (>48° N)andsouthernEurope (<48° N).Thiscollapsedtablewasusedtocomputethepairwisedissimilarity between northern and southern faunas (β sor),andthefractions of dissimilarity related to species turnover (β sim)andnestedness (β sne),followingthebetadiversitypartitioningframeworkintroduced by Baselga (2010).Whetherspatialturnoverornestedness-resultant dissimilarity was the dominant fraction was determined using the β sim/β sor ratio (βratio).Thus,thehighertheβratio, the higher the relative importance of turnover over nestedness. 2.3 | Relationship between beta diversity patterns and dispersalrelated traits of spiders To study whether dispersalrelated traits explain beta diversity patterns of spiders in Europe, we assessed the relationship of these traits with (i) the North–South difference in intercepts of spatial andclimaticdistance-decaymodels,(ii)theNorth–Southdifference inslopesofspatialandclimaticdistance-decaymodelsand(iii)the values of turnover (β sim)andnestedness-resultantdissimilarity(β sne) between the North and South of Europe. Ballooning is the main longdistance dispersal method of spiders, so spider families were classified according to their ballooning tendency in frequent and infrequent ballooners. The assignation of each spider family to these categories was based on studies of ballooning behaviour and the proportions of different families observed in aerial samples (Bishop & Riechert, 1990; Blandenier, 2009; Blandenier & Fürst, 1998; Dean&Sterling,1985; Greenstone et al., 1987; Pearce et al., 2005). Following Carvalho and Cardoso (2014),familiesthatrepresentless than1%ofthetotalnumberofindividualsinaerialsampleswere classified as infrequent ballooners, whereas families representing morethan1%ofindividualswereclassifiedasfrequentballooners. Wealsoincorporatedintotheanalysissomeadditionaltraitsthat can influence the ballooning tendency and efficiency, and have also been used as proxies of spiders dispersal ability in previous papers (Carvalho&Cardoso,2014;Jiménez-Valverdeetal.,2010):maximum female body size, as bigger species tend to fall at shortest distances (Dean&Sterling,1985);vegetationstrata,ashighstratainhabitants are easily carried by the wind (Blandenier, 2009; Platnick, 1976)and silk production, as not producing silk hampers ballooning (Bonte et al., 2004;Larrivée&Buddle,2011).Weestimatedthemaximum female body size of each family by obtaining a random sample of 50 species and averaging the maximum female body size across species. For families with less than 50 species, all species were considered. Female body size data were obtained from Nentwig et al. (2022), Esyunin and Sozontov (2016), Koch (1879), Tyschchenko (1965), Marusik and Koponen (1998),Ponomarev(2009),Ledoux(2014)and MacíasHernández, Ramos, et al. (2020).Weassignedeachfamilya vertical strata category—soil, vegetation or both—based on the classification of Cardoso et al. (2011).Finally,wealsoclassifiedspider families as web builders or nonweb builders. The spider families' traits can be found in Table 1. We acknowledge that using trait data at the family level is a broad approximation. However, spider lifehistory traits (e.g. ballooning propensity, web building or web type, etc.) are generally well conserved within families (Cardoso et al., 2011;Carvalho&Cardoso,2014),whichhasallowedmany studies to successfully use familylevel trait data in biogeographical studies(e.g.Jiménez-Valverdeetal.,2010;Suárezetal.,2023;Wu et al., 2017).Thus,theuseoffamily-leveldispersaltraitsasproxies of dispersal ability is a reasonable approximation to assess crossfamily differences in biogeographic patterns likely affected by dispersal ability, such as distancedecay patterns. To assess whether the differences in spatial and climatic distancedecay parameters between northern and southern Europe are related to dispersal traits of spider families, we used phylogeneticgeneralizedleastsquare(PGLS)analyseswiththepgls function of the R package caper(Ormeetal.,2018),consideringasdependent variable(i)thedifferenceinintercepts(northernintercepts—southernintercepts),(ii)thedifferenceinslopes(northernslopes—southernslopes)and(iii)thedissimilarityinspeciescompositionbetween northernandsouthernfaunasdueto (iii.a)turnover(βsim)or(iii.b) nestedness (β sne).Traitsinevolutionarilyrelatedgroupsmaypresent phylogeneticautocorrelationthatcaninflatetypeI(falsepositive) andtypeII(falsenegative)errors(Gittleman&Kot,1990),soweused PGLSmodelstoaccountforphylogeneticautocorrelationasinferred 14668238, 2024, 4, Downloaded from https://onlinelibrary.wiley.com/doi/10.1111/geb.13810 by Universidade de Santiago de Compostela, Wiley Online Library on [22/04/2024]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License 6 of 13 | MARTÍN-DEVASA et al. from the pruned phylogenetic tree based on MacíasHernández et al. (2020a)(Figure S3).Wecomputed16differentPGLSmodels, considering all possible trait combinations, and selected the model withthelowestAICasbest. 3 | RESULTS Wefirstmeasuredthecorrelationbetweenspatialandclimaticdistances to know whether any relationship between community similarity and spatial or climatic distance could be attributed to dispersal limitation or climatic sorting, respectively, independently of the alternative process. Correlation between spatial and climatic distance was high in the North (Mantel test, Pearson r = 0.72,p < 0.001),but lowintheSouth(Manteltest,Pearsonr = 0.14,p = 0.175)(Figure S4). In southern Europe, faunistic similarity markedly decreased with spatial distance (Figure 1 shows the decay of faunistic similarityagainstareferenceterritory,Spain,forillustrativepurposes).As a result, spatial distance explained a relevant fraction of variation in community similarity in most families (pseudoR2 >0.3 in nine families, see Figure 2),whileclimaticdistanceexplainedanegligible fraction (pseudoR2 ≤0.07inallfamilies,Figure 2).Onaverage, the intercept was significantly lower (phylogenetic paired ttest, t12 = 3.41,p = 0.005) for climatic (0.78 ± 0.08[SD])thanforspatial models(0.86 ± 0.06[SD]),andmeanslopewassignificantlysteeper (t12 = −6.76,p < 0.001)forspatial(−0.53 ± 0.29[SD])thanforclimatic (−0.08 ± 0.07[SD])models(Figure 3).Whencomparisonsweremade within each family using the zdep statistic, the intercepts were also significantly lower for climatic curves in 10 of 15 families, and slopes were significantly steeper for spatial than for climatic distancedecay curves in 12 of 15 families (Table S4).Allparametersareshownin Table S5. In northern Europe, faunistic similarity hardly decreased with spatial distance (Figure 1 shows the decay of faunistic similarity againstareferenceterritory,Norway,forillustrativepurposes).As a result, spatial distance explained a small fraction of variation in community similarity (pseudoR2 >0.3 in only three families, see Figure 4),andclimaticandspatialdistanceexplainedasimilarfraction of variation in most families (difference in pseudoR2 between climaticandspatialmodels≤0.1inallfamiliesbuttwo,Figure 4).In the North, the average intercept was significantly lower (phylogenetic paired ttest t12 = 3.25,p = 0.007)forclimatic(0.94 ± 0.04[SD]) thanforspatialmodels(0.96 ± 0.03[SD]),andmeanslopewassignificantly flatter (phylogenetic paired ttest t12 = −4.17,p = 0.001)for climatic(−0.18 ± 0.12[SD])thanforspatial(−0.26 ± 0.11[SD])models (Figure 3),althoughthedifferencewasmuchlessmarkedthaninthe South(differencebetweenmeanslopeof0.08intheNorthvs.0.45 in the South). When comparisons were made within each family, spatial and climatic intercepts and slopes were significantly different only for one of the 15 families (Table S4).Takenaltogether,these results evidenced that spatial and climatic distancedecay models arecoupledintheNorthofEurope,butnotintheSouth,whereonly spatial distance, but not climatic distance, is a relevant predictor of faunistic similarity. When the relationship between family traits and the North– Southdifferenceinparametersofspatialdistance-decaymodelswas assessed, we found that southern slopes were steeper than northern slopes (Figure 3a)andtheNorth–Southdifferenceinslopewas explained by all dispersal traits, as the best model included ballooningtendency(smallerdifferenceinfrequentballooners),vegetation strata(smallerdifferenceinhigherstrata)andmaximumfemalesize (largerdifferenceinbiggerfamilies)ofthespiderfamilies(R2 = 0.57, F4,10 = 3.29,p = 0.06)(Tables S6 and S7).Inturn,thebestmodelofthe North–Southdifferenceininterceptincludedonlyvegetationstrata Ballooning Web builders Strata Max size females (mm) Agelenidae Infrequent Yes Both 9.57 Araneidae Frequent Yes Vegetation 11.70 Clubionidae Infrequent No Vegetation 7.66 Dictynidae Infrequent Yes Both 3.14 Gnaphosidae Infrequent No Soil 7.83 Hahniidae Infrequent Yes Soil 2.66 Linyphiidae Frequent Yes Both 2.80 Liocranidae Infrequent No Soil 5.95 Lycosidae Frequent No Soil 10.03 Philodromidae Frequent No Both 6.62 Pholcidae Infrequent Yes Both 5.50 Salticidae Frequent No Both 6.03 Tetragnathidae Frequent Yes Vegetation 9.86 Theridiidae Frequent Yes Both 3.83 Thomisidae Frequent No Both 7.10 TABLE 1 Classificationofspider families according to dispersal traits. 14668238, 2024, 4, Downloaded from https://onlinelibrary.wiley.com/doi/10.1111/geb.13810 by Universidade de Santiago de Compostela, Wiley Online Library on [22/04/2024]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License | 7 of 13 MARTÍN-DEVASA et al. and was not significant (R2 = 0.25,F2,12 = 2.00,p = 0.18)(Tables S6 and S8).Regardingclimaticdistance-decaymodels,wefoundthatinterceptswerehigherintheNorththanintheSouth(Figure 3b),and thatthebestsupportedmodelfortheNorth–Southdifferencein intercept explained a large fraction of variance and, as predictors, included ballooning tendency (smaller difference in frequent ballooners),vegetationstrata(smallestdifferenceobservedinhigherstrata, see Table S9)andmaximumfemalesize(positiveeffect)ofthespider families (R2 = 0.64,F4,10 = 4.45,p = 0.02)(Tables S6 and S9).However, thebestmodeloftheNorth–Southdifferenceinclimaticdistance- decay slope was not significant (R2 = 0.28, F2,12 = 2.32, p = 0.14) (Tables S6 and S10).Itshouldbenotedthatalternativemodelswere also equally plausible (ΔAIC<2)andtheyarereportedinTable S6. When the dissimilarity between the northern and southern European spider faunas was assessed, nestednessresultant dissimilarity (β sne)wasthelargestcomponentofcompositionaldifferences between northern and southern Europe for 10 of the 15 families studied (βratio <0.5). In contrast, species turnover was the largest component in the families Clubionidae, Dictynidae, Linyphiidae, Lycosidae and Thomisidae (βratio >0.5) (Table 2). The amount of nestednessresultant dissimilarity between southern and northern Europe was well explained by dispersal traits of spider families. The best model included ballooning tendency (more nestedness in infrequentballooners),maximumfemalebodysize(positiveeffect)and vegetationstrata(lessnestednessinhigherstrata)ofspiderfamilies (R2 = 0.68,F4,10 = 5.28,p = 0.01)(Tables S6 and S11).Inturn,thebest modeloftherelationshipbetweendispersaltraitsandNorth–South species turnover only included female size and was not significant (R2 = 0.14,F1,13 = 2.19,p = 0.16)(Tables S6 and S12).Equallyplausible models (ΔAIC<2)arereportedinTable S6. 4 | DISCUSSION Ourresultsshowthatdispersallimitationisthemainfactorshaping presentday beta diversity patterns of European spiders at the continentalscale.Wecandrawthisconclusionbecause,insouthernEurope,(i)spatialandclimaticdistancesarenotcorrelated,(ii) community similarity is well explained by spatial distance but not by climaticdistanceand(iii)slopesaremuchsteeperinspatialthanclimaticdistance-decays.Inturn,innorthernEurope,theexplanatory power of spatial and climatic distances is similar, as expected from the high correlation of spatial and climatic distances. Therefore, in northern Europe, we cannot easily infer the causal processes behindtheobserveddistance-decaypatterns.Wecouldspeculatethat either(i)climaticconstraintsaremorerelevantinnorthernEurope due to the prevalence of harsher climatic conditions or, alternatively, (ii)thatthesamedispersallimitationobservedinsouthernEurope is behind the observed distancedecay patterns in the North, and climatic distance explains the pattern just because it is correlated to spatial distance. Finally, dispersalrelated traits of spider families explain (i) the North–South difference between slopes of spatial FIGURE 2 Distance-decaymodels(fittedwithanegativeexponentialfunction)usingstandardizedspatialandclimaticdistancesfor 15spiderfamiliesinsouthernEurope.Spatialandclimaticdistanceswerescaledto[0,1]rangetoallowadirectcomparisonofmodel parameters.SandCindicatethepseudo-R2ofthespatialandclimaticdistance-decaymodels,respectively.Int:*indicatessignificant differencesbetweeninterceptsandSlp:*indicatessignificantdifferencesbetweenslopesaccordingtothezdep statistic. 14668238, 2024, 4, Downloaded from https://onlinelibrary.wiley.com/doi/10.1111/geb.13810 by Universidade de Santiago de Compostela, Wiley Online Library on [22/04/2024]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License 8 of 13 | MARTÍN-DEVASA et al. distance-decaycurves,and(ii)theamountofnestedness-resultant dissimilarity between northern and southern faunas. Moreover, the relationships between differences in slope and the morphological and ecological traits of spiders were in agreement with the predictions derived from their hypothesized link with dispersal. Taking all together, our results point to a major role of dispersal limitation and postglacial recolonization lags in biogeographic patterns of spiders at a continental scale, as previously found for beetles (GómezRodríguez&Baselga,2018). Previous studies have revealed that community composition in spiders generally depends on the combination of dispersal and environmental(climateandvegetationstructure)constraints(Ávila et al., 2020; Barton et al., 2017; Carvalho et al., 2011; Mammola et al., 2019; RodriguezArtigas et al., 2016; Tonkin et al., 2016; Zhang et al., 2018), but the importance of each process seems to change with scale. At small spatial scales, species sorting has been frequently suggested as the most important process driving spider community composition due to the strong relationship between environmental variables and spider communities (Bowden &Buddle,2010; Finch et al., 2008;Jiménez-Valverdeetal.,2010; Jiménez-Valverde&Lobo,2007).Dispersallimitationhasalsobeen shown to affect the spatial structure of spider communities, for example, in dune and cave systems (Bonte et al., 2004; Carvalho et al., 2011; Mammola et al., 2019).Atthecontinentalscale,the preponderance of dispersal limitation as a driver of differences in species composition is here revealed by multiple results, as argued above. These lines of evidence not only include macroecological patterns based purely on the distribution of species (as the contrastbetweenclimaticandspatialdistance-decaymodels)butalso clear relationships between these distributional patterns and organismal traits related to dispersal (ballooning, vegetation stratum orbodysize).Forexample,ballooningwasnegativelyassociated withNorth–Southdifferencesindistance-decayslopes,suggesting that smaller differences in slope were linked to families with higherdispersalability.MarkedNorth–Southdifferencesinthe slopes of spatial distancedecay curves have been previously interpreted as the result of dispersal limitation in a previous paper onEuropeanbeetles(Gómez-Rodríguez&Baselga,2018).Therationale behind GómezRodríguez and Baselga (2018)studyisthat, under strong dispersal limitation, most species have not been able to recolonize the North of Europe, so these species are restricted to southern territories. This leads to steeper distancedecay curves insouthernEurope.Incontrast,thefewspeciesabletocolonize the northern regions are good dispersers which tend to occupy larger regions, leading to flatter distancedecay slopes in northern Europe. For European spiders, here we show the same contrast between northern and southern slopes and, as for beetles, the difference between slopes is well explained by dispersalrelated traits. The relationship between these traits and the North– Southdifferenceinclimaticdistance-decayinterceptsmightjust FIGURE 3 Differencebetweennorthern(bluedots)andsouthernEurope(redtriangles)ininterceptsandslopesof(a)spatialand(b) climatic distancedecay models. 14668238, 2024, 4, Downloaded from https://onlinelibrary.wiley.com/doi/10.1111/geb.13810 by Universidade de Santiago de Compostela, Wiley Online Library on [22/04/2024]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License | 9 of 13 MARTÍN-DEVASA et al. reflect the tendency to smaller biotic similarities in southern than in northern Europe. This inference can be reached because, in southern Europe, similarity depends on spatial distance but is unrelated to climatic distance, and spatial and climatic distances are mostly decoupled. This suggests that smaller intercepts in climatic distance-decaymodelsintheSoutharejusttheresultofspatially distantpairsofcountries(withlowsimilarity)thatareclimatically similar(smallclimaticdistances).Theseresultsdonotruleoutthe relevance of species sorting and biotic processes (e.g. priority or competitiveexclusion)atsmallerspatialscales,butsuggestthat dispersal limitation is the major process behind compositional differences at large spatial scales. BesidesleadingtomarkedNorth–Southdifferencesindistance- decay patterns, dispersal limitation and incomplete postglacial colonization also resulted in northern faunas being a subset of southern ones, as has also been observed in a variety of taxa (e.g. Fattorini &Ulrich,2012; Griffiths, 2017; Hortal et al., 2011),includingspidersinnorthernCanada(Loboda&Buddle,2018)andsubterranean spiders in Europe (Mammola et al., 2019).Itcouldbearguedthat environmental tolerances in a climatic gradient may also produce a nested structure if only organisms with high tolerance could colonize the extreme climates of northern Europe (Ulrich et al., 2009). Previous work has already pointed out that temperature is an important variable driving spider alpha and beta diversity (Finch FIGURE 4 Distance-decaymodels(fittedwithanegativeexponentialfunction)usingstandardizedspatialandclimaticdistancesfor 15spiderfamiliesinnorthernEurope.Spatialandclimaticdistanceswerescaledto[0,1]rangetoallowadirectcomparisonofmodel parameters.SandCindicatethepseudo-R2ofthespatialandclimaticdistance-decaymodels,respectively.Int:*indicatessignificant differencesbetweeninterceptsandSlp:*indicatessignificantdifferencesbetweenslopes. TABLE 2 Dissimilaritybetweennorthernandsouthern spider faunas derived from species turnover (β sim),derivedfrom nestednessresultant dissimilarity (β sne)andproportionoftotal observed dissimilarity (β sor)derivedfromturnover(theratio between β sim and β sor, βratio)foreachspiderfamily. βsim βsne βratio Agelenidae 0.118 0.571 0.171 Araneidae 0.063 0.132 0.321 Clubionidae 0.114 0.080 0.587 Dictynidae 0.214 0.105 0.670 Gnaphosidae 0.143 0.345 0.292 Hahniidae 0.063 0.223 0.219 Linyphiidae 0.294 0.152 0.659 Liocranidae 0.100 0.329 0.233 Lycosidae 0.220 0.142 0.607 Philodromidae 0.111 0.233 0.323 Pholcidae 0.083 0.393 0.175 Saliticidae 0.114 0.292 0.281 Tetragnathidae 0.053 0.197 0.208 Theridiidae 0.093 0.268 0.258 Thomisidae 0.207 0.190 0.521 Note: The higher the βratio, the higher the relative importance of turnover over nestedness. 14668238, 2024, 4, Downloaded from https://onlinelibrary.wiley.com/doi/10.1111/geb.13810 by Universidade de Santiago de Compostela, Wiley Online Library on [22/04/2024]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License