Landscape Ecol

https://doi.org/10.1007/s10980-020-00991-0 (0123456789().,-volV)( 0123456789().,-volV)

RESEARCH ARTICLE

Arthropod traits and assemblages differ between core patches, transient stepping-stones and landscape corridors

Scott M. Pedley . Paul M. Dolman

Received: 5 August 2019 / Accepted: 6 March 2020 Ó The Author(s) 2020

Abstract habitat arthropods. Species and trait (habitat associa- Context Restoring landscape connectivity can miti- tion, diet, body size, dispersal ability) composition gate fragmentation and improve population resilience, were compared by ordination and fourth corner but functional equivalence of contrasting elements is analyses. poorly understood. Evaluating biodiversity outcomes Results Each network element supported distinct requires examining assemblage-responses across con- arthropod assemblages with differing functional trait trasting taxa. composition. Core patches were dominated by spe- Objectives We compared arthropod species and trait cialist dry-open habitat species while generalist and composition between contrasting open-habitat net- woodland species contributed to assemblages in work elements: core patches, corridors (allowing connectivity elements. Nevertheless, transient patches individual dispersal and population percolation), and (and to a lesser degree, corridors) supported dry-open transient stepping-stones (potentially enhancing meta- species characteristic of the focal grass-heath sites. population dynamics). Trait associations differed markedly among the three Methods Carabids and spiders were sampled from elements. Dispersal mechanisms and their correlates core patches of grass-heath habitat (n = 24 locations differed between taxa, but dry-open species in tran- across eight sites), corridors (trackways, n = 15) and sient patches were characterised by traits favouring recently-replanted clear-fells (transient patches, dispersal (large running hunter spiders and large, n = 19) set in a forest matrix impermeable to open- winged, herbivorous carabids), in contrast to wingless carabids in corridors. Conclusions Core patches, dispersal corridors and Electronic supplementary material The online version of transient stepping-stones are not functionally inter- this article (https://doi.org/10.1007/s10980-020-00991-0) con- tains supplementary material, which is available to authorized changeable within this system. Semi-natural core users. patches supported a filtered subset of the regional fauna. Evidence for enhanced connectivity through & S. M. Pedley ( ) percolation (corridors) or meta-population dynamics Ecology and Environment Research Centre, Department of Natural Sciences, Manchester Metropolitan University, (stepping stones) differed between the two taxa. Chester Street, Manchester M1 5GD, UK e-mail: [email protected] Keywords Dispersal corridors Á Ecological network Á Landscape connectivity Á Movement S. M. Pedley Á P. M. Dolman School of Environmental Sciences, University of East corridors Á Open-habitat network Anglia, Norwich NR4 7TJ, UK 123 Landscape Ecol

Introduction Connectivity is often conceived as movement corridors enabling dispersal between primary habitat Restoring functional connectivity is promoted in patches (Simberloff et al. 1992). Meta-analysis of conservation strategies to facilitate biodiversity resi- movement, considering both direct (movement fre- lience and population survival in the face of anthro- quency and rate) and indirect (abundance and species pogenic landscape fragmentation, land-use richness in connected patches) measures, confirms intensification and climate change (Chetkiewicz corridors increase species’ movement in fragmented et al. 2006; DEFRA 2018; Isaac et al. 2018). At local landscapes (Gilbert-Norton et al. 2010), potentially scales, enhanced connectivity within regional land- recolonising vacant habitat or enhancing persistence scapes may improve local population persistence by through population rescue (Lawson et al. 2012). For enhancing rescue effects and reducing detrimental slowly-dispersing and matrix-sensitive species, corri- impacts of fragmentation (Beier and Noss 1998; dors of suitable quality may allow population perco- Damschen et al. 2006) while at greater scales, lation over multiple generations (Bennett 2003), connectivity is advocated to facilitate population assisting range responses to climate change (Krosby range-response to climatic change (e.g. Heller and et al. 2010). However, corridors, which by their nature Zavaleta 2009; Imbach et al. 2013). While the optimal are narrow with high edge-area ratios (Simberloff level of site aggregation or dispersion may differ et al. 1992), may have assemblages filtered and between the aims of enhancing local persistence or modified by edge impacts (Ewers and Didham 2008; favouring large-scale range-expansion (Hodgson et al. Campbell et al. 2011). Alternatively, stepping stones, 2011), functional connectivity at both scales requires a series of discrete, smaller patches intended to link appropriate interventions within local landscapes otherwise isolated habitat blocks, may provide con- (Jongman et al. 2004, 2011). Within such local nectivity for mobile species capable of occasional networks, although some generalist species may dispersal across the intervening matrix (Schultz 1998). disperse through surrounding matrix habitats, for A reduced edge-area may mitigate edge impacts and specialist species population connectivity often offer better habitat suitability for specialist species, requires connectivity elements of suitable quality although assemblages may be filtered by colonisation and type, with corridors, stepping stones, or a less ability (Kormann et al. 2015). Both of these types of hostile matrix (Bennett 2003) intended to improve connectivity elements are often advocated without dispersal and functional connectivity between ‘core explicit consideration of their relative efficacy or patches’ or ‘ecological refuges’—relict fragments of functional equivalence (e.g. Lawton et al. 2010), with natural or semi-natural habitat that have retained local limited analysis of the trade-offs between differing species-populations over decadal timescales (Davis configurations and taxa. et al. 2013). Despite the abundance of literature For species associated with early-successional or promoting restoration of ecological networks the physically-disturbed habitats, stepping-stone suitabil- relative benefits of corridor linkages or stepping ity may be short-lived, with more frequent dispersal stones are poorly understood (Dolman 2012). Further- required for meta-population persistence (Amarase- more, few connectivity studies report community or kare and Possingham 2001; Johst et al. 2002; Loehle assemblage responses, with research often focused on 2007). Similarly, cyclic or episodic landuse, including a few target species (Haddad et al. 2014). Conserva- much habitat management practiced for conservation, tion strategies seeking to restore ecological connec- the development and building phases of physical tivity are best informed by understanding responses to infrastructure projects, and crop or forestry rotations, connectivity elements of differing configurations may all provide short-lived opportunities for special- across multiple taxa (Pedley et al. 2013b; Kormann ists able to colonise, exploit and move on from et al. 2015; Albert et al. 2017). Additionally, exam- stepping stones. Understanding assemblage responses ining responses in terms of assemblages’ trait- rather to differing configuration of habitat can inform than species-composition supports generalisation to whether biodiversity conservation strategies can take other ecological systems (McGill et al. 2006; Pedley advantage of spatially discontinuous and transient and Dolman 2014; Santini et al. 2016). patches, or instead require permanent and spatially continuous, connectivity elements. 123 Landscape Ecol

Here, we examine arthropod (carabid and spider) Material and methods trait and assemblage responses to open-habitat ele- ments of contrasting configuration, comprising: ‘core Study area and design patches’ (semi-natural grass-heath sites supporting large numbers of scarce and threatened taxa), grass- The study region in eastern England is characterised heath habitat persisting as ‘linear corridors’ (formed by a semi-continental climate, sandy, nutrient-poor by trackways and verges), and transient stepping-stone soil and historically supported extensive semi-natural patches (formed by infrequent systematic physical grass-heath (Dolman and Sutherland 1992). Most disturbance events), with the latter two potential grass-heath was converted to arable or forestry during connectivity elements embedded in a plantation the twentieth century, but remaining grass-heath sites matrix impermeable to open habitat specialists (Ped- support large numbers of scarce and threatened taxa, ley 2012). A fundamental assumption of our study, is characterised by specialist coastal, continental and that assemblage richness and composition in the Mediterranean species dependent on physically dis- recently-created connecting elements has been shaped turbed open habitats (Dolman et al. 2012). Most of the by patterns of individual dispersal and colonisation remaining grass-heaths are designated for their con- (following Gilbert-Norton et al. 2010), particularly for servation value but are fragmented and isolated. open-habitat species for which the matrix is imper- Regional conservation strategy seeks to improve meable. The use of terrestrial arthropods, particularly habitat quality within these remaining grass-heaths, carabids and spiders, provides a useful group with but also to restore functional connectivity between which to examine landscape configuration as they are these ‘core patches’, with a focus on connectivity highly speciose, have high reproductive rates, respond across and within extensive (185 km2) plantation quickly to environmental change and contain many forest (Thetford Forest, 0° 400 E, 52° 270 N) estab- different life history stratagems including dispersal, lished over former open habitats in the early twentieth feeding and body size (Duffey 1968; Robinson 1981; century. Lovei and Sunderland 1996; Marc et al. 1999; Rainio During this study Thetford Forest was dominated and Niemela 2003). by conifers, comprising 80% Corsican (Pinus nigra) We use this system to examine whether assem- and Scots (P. sylvestris) pine, managed by clear- blages in transient patches and corridors within the felling (typically at 60–80 years) and replanting of plantation resembled those of the core patches, or even-aged management patches (hereafter ‘stands’). instead have been filtered in ways that can be predicted The forest landscape also contains open-habitat by life history traits. Three alternative colonisation elements that differ in configuration. Narrow linear hypotheses are considered for these elements: (1) grass-heath ‘corridors’ are retained along trackways assemblages in transient stepping-stone and corridor and their perennial verges that form an extensive elements resemble those of core patches; (2) assem- (1290 km, 18 km2) open-habitat network permeating blages in transient stepping-stone and corridor ele- the forest landscape; following planting of the adja- ments comprise a sub-set of species lacking many cent tree crop each trackway section provides open specialists restricted to core patches; (3) compared to conditions for at least 20 years (see below). Larger assemblages within corridors, transient elements have transient stepping-stone patches (mean area of sam- greater representation of dispersive species. We pled patches: 9.0 ha ± 8.6 SD) are provided by clear- predicted that (P1) assemblages developing in tran- felled and replanted stands that offer short-lived sient patches of open habitat will be represented by (5–7 years) open-habitat that is variably connected small-bodied, aerial-dispersing, habitat generalists, to the corridor network (Fig. 1); these provide an whereas (P2) corridors will have assemblages domi- excellent opportunity to examine colonisation after nated by large-bodied, active-hunting generalists and regular, systematic disturbance events. Vegetation that (P3) edge effects will increase the species number regeneration in these transient patches is derived from and diversity in connectivity elements compared to persistent grass-heath seed-banks, augmented by zoo- core patches, due to an influx of vagrant and generalist chorous and aerial dispersal (Eycott et al. species. 2006a, b, 2007). Both trackways and transient patches

123 Landscape Ecol

Fig. 1 The distribution of core patches of grass-heath, corridor and transient stepping-stone elements sampled within the study landscape provide fine-grained mosaics of modified grass-heath Sites of Special Scientific Interest (SSSI) designated vegetation and bare mineral soil. for biodiversity under UK conservation legislation, of Assemblages of ‘core patches’ were characterised which five were also designated as Special Areas of by sampling three locations within each of eight grass- Conservation (SAC) under the European Habitats heaths (mean area 106 ha ± 130 SD), either abutting Directive. Most were bordered by plantation forest and or located within 2.5 km of the forest or intensive agriculture. Sampled areas comprised (mean = 0.72 km ± 1.11 SD). These included seven physically-disturbed habitats on which characteristic

123 Landscape Ecol regional biota depend (Dolman et al. 2012), including replication for analysis. Although a single sampling areas grazed by sheep and rabbits and recently- year may miss year-on-year differences in composi- ploughed areas associated with heathland tional detail of sampled assemblages, previous management. research on ground invertebrates in the study region Nineteen transient stepping-stones were sampled. has shown relatively stable species compositions (Lin Prior to planting, stands were cleared of coarse woody et al. 2007; Pedley et al. 2013b). A single pitfall debris (residue of clear-felling) and planting lines transect was placed in each of the 15 corridor ploughed exposing extensive bare mineral substrate. elements. Two transects were placed within each of Bare soil is encroached by ground vegetation within the 19 transient stepping-stones, separated by at least 3–4 years (Wright et al. 2009); tree cover then 50 m for independence. In each of the eight heathland progresses with canopy closure after 20 years (He- sites, three independent transects were again separated mami et al. 2005). Extensive sampling shows open- by at least 50 m. We acknowledge that replicate habitat carabid and spider species persist for approx- transects sampled within each grass-heath site, and imately five and seven years following planting (Lin within each transient patch, are not spatially indepen- et al. 2007; Pedley 2012). In the current study dent, and where possible control for this spatial sampling considered stands up to 7 years (n = 5 for pseudo-replication by including site identity as a 1-year-, 3-year- and 5-year-old; n = 4 for 7-year-old). random effect in subsequent analyses (see below). These transient open-habitat patches are fragmented, Arthropod assemblages were sampled in each comprising approximately 6–8% of the 60–80 year transect in May and again in June, the peak periods growth cycle, while the surrounding plantation crop is of arthropod abundance and activity, each time using impermeable to open-habitat species (Bertoncelj and six pitfall traps (each 7.5 cm deep, 6.5 cm diameter, Dolman 2013a, Pedley and Dolman unpublished). filled with 50 ml of 70% ethylene glycol), set 15 m Transient patches available for sampling were clus- apart and opened for seven consecutive days. The two tered in the core plantation area (Fig. 1), but distance catches were pooled to give one composite sample per to the nearest grass-heath site was similar to that of the transect for subsequent analyses. In August 2009, corridor elements (t =-1.472, DF = 30.7, vegetation height in each transect was assessed at 40 P = 0.151; transient mean = 2.1 km ± 1.0 SD, cor- points using a sward stick (diameter 90 mm, weight ridor mean = 1.7 km ± 0.6 SD). 250 g, following Dolman and Sutherland 1992), and Fifteen corridor elements were sampled. Track- percentage of bare substrate visually estimated in ways (mean width 12.7 m ± 3.9 SD) comprised two 20 cm 9 20 cm at each point. At each of these points elements: a central wheeling with sparse vegetation the presence of key structural vegetation features and exposed substrate, flanked by vegetated verges. (herbs, grasses, moss, lichen, shrub, and succulent) Many heathland-associated carabid and spider species was also recorded, providing frequency per transect. have been recorded from this trackway network (Lin Carabid identification followed Luff (2007). Adult et al. 2007; Pedley et al. 2013a) but are excluded by spiders were identified to species following Roberts shading from those trackways within or bordering (1987, 1996), juveniles and sub-adults with undevel- older stands (C 20 years) (Pedley et al. 2013a). oped reproductive structures were not identified. Sampled verges were each adjacent to thicket-aged Analysis of spider assemblages was confined to stands (11–20 year old, following Hemami et al. ground-active species as sedentary invertebrates, such (2005) and a mean of 0.78 km (± 0.25 SD) from the as many web-spinning spider, are poorly represented nearest other sampled trackway. To minimise shading by pitfall catches (Topping and Sunderland 1992). effects, for trackways oriented east–west the northern- Ecological and habitat affinities of identified species most verge was sampled, and for those oriented north– were classified following Luff (1998, 2007) for south the widest verge was sampled. carabids and Harvey et al. (2002) and Roberts (1996) for spiders. Habitat affinities comprised: shaded Arthropod sampling woodland habitats, hereafter ‘woodland’; ‘generalist’ species of multiple or any mesic habitat; and ‘dry- Ground-active arthropods were sampled in 2009 by open’ species associated with dry calcareous or acidic pitfall transects that formed the standardised unit of grassland, dry lowland heathland, dunes, sand or 123 Landscape Ecol gravel pits. Open habitat connectivity elements there- double standardisation) using the vegan package in R fore provide potential habitat for these ‘dry-open’ (Oksanen et al. 2018). Habitat structure variables were species that are excluded from (unable to percolate tested for association with community composition through) the closed-canopy forest matrix, but their (NMDS) using the envfit function in vegan. presence depends on colonisation following dispersal. Arthropod assemblage composition was compared among core patches, corridors, and transient stepping- Life history traits stones by multivariate Generalised Linear Models (GLMs) implemented in the mvabund package (Wang Potential ecological factors that may determine et al. 2012) using likelihood-ratio-tests with a negative assemblage responses to habitat elements were binomial error distribution. selected. We considered the life history traits of Richness and aggregate abundance of dry-open, dispersal ability and foraging type (Pedley and Dol- generalist and woodland species were compared man 2014) for which published information was between landscape elements using Generalised Linear available at the individual species level (Online Mixed Models (GLMM). Site (stand, trackway or Resource 1). To ensure ordinal variables had a useable heath site) was included as a random effect as transects number of species per category, uncommon sub- in corridors may capture greater beta diversity than categories were merged into broader classes. For replicates (separated by C 50 m) within individual spiders, aerial dispersal by ballooning (passively grass-heaths or transient stepping-stones. Models used floating on silk threads) is considered effective for Poisson or negative binomial error as appropriate, with both short- and long-distance dispersal to colonise landscape element means compared by Tukey pair- suitable habitat (Duffey 1998; Bell et al. 2005). wise comparisons. GLMMs were implemented in R Following Lambeets et al. (2008) and Langlands using the glmer and glmmabmb functions from the et al. (2011), we assigned spider species as ballooners packages lme4 (Bates et al. 2015) and glmmADMB. according to the review by Bell et al. (2005). Carabid Linkages within trait data (across species) were dispersal ability was assessed by wing morphology examined by a Principal Coordinates Analysis (PCoA) following Barbaro and van Halder (2009), with applied to a trait-dissimilarity matrix (traits as vari- species exhibiting well-developed wing structures ables, species as samples), using the Gower coefficient (macropterous) considered effective aerial dispersers, (Gower 1971) using the vegan package. The Gower relative to those with rudimentary wing development coefficient was used as it handles mixed data types (brachypterous). Wing-dimorphism (species with both (nominal, ordinal or continuous—here standardised to long- and short-winged forms) was classified sepa- zero mean and unit variance). Associations among rately and is considered to provide dispersal advan- traits were visualised by plotting trait vectors (from tages over brachypterous species, especially in Spearman rank correlation in relation to the first two temporally- and spatially-heterogeneous landscapes PCoA ordination axes) following Langlands et al. (Kotze and O’Hara 2003). (2011); species habitat affinities were examined to visualise correlations with species traits. Species were Analysis also grouped by family (spiders) or tribe (carabids) to allow a qualitative examination of taxonomic influ- Spiders and carabids were analysed separately, con- ence in trait space. Phylogenetic character analysis sidering composite samples from individual transects. was not possible due to unresolved phylogeny of both All analyses were carried out using the R statistical the spider and carabid assemblages. software (R Development Core Team 2018). Sam- Trait-specific responses to habitat types were pling efficiency was compared between landscape analysed using fourth-corner analysis (Dray and elements using sample-size based rarefaction curves Legendre 2008) that tests the link between all calculated using the iNext package (Hsieh et al. 2016). combinations of species traits and environmental Assemblage composition was visualised using non- attributes (here: core patch, corridor and transient Metric Multidimensional Scaling (NMDS) performed landscape elements). By re-sampling three data on a matrix of Bray–Curtis dissimilarities of abun- matrices this approach indirectly relates matrix ‘R’ dance data (square root transformed and Wisconsin (environmental attributes 9 site) to matrix ‘Q’ (trait 9 123 Landscape Ecol species), via matrix ‘L’ (species-abundance 9 site). deviance = 817.8, P \ 0.001 respectively; paired The fourth-corner procedure uses a generalised statis- comparisons all P \ 0.001). For carabids, assem- tic SRLQ that is equivalent to the Pearson correlation blages of core patches were distinct on NMDS axis coefficient r when trait (Q) and environment (R) vari- one, while transient and corridor carabid composition ables are quantitative, to v2 when both sets are differed from each other on axis two (Fig. 2). Simi- qualitative, and to the correlation ratio g2 for mixed larly for spider assemblages, axis one separated core data (Dray and Legendre 2008). patches from both connectivity elements while axis The observed strength of linkage among traits and two again separated transient and corridor elements environmental attributes is assessed against that which (Fig. 2) though to a lesser extent than for carabids. may arise by chance, with null models determined by For both arthropod groups, assemblage composi- the observed structure of data matrices, using ran- tion was significantly associated with all habitat domisation and permutation (Dray and Legendre structure variables (Fig. 2, Online Resource 3). Heath 2008). Permutation Model 1 was used to test the null sites and assemblages were characterised by greater hypothesis that R is not linked to Q, by examining representation of lichens, succulents, bare ground and links between fixed species traits and fixed site short swards; transient patches had taller vegetation, characteristics. This model randomises species’ rela- while corridors had greater cover of herbs, shrubs and tive to site characteristics (permuting within each grasses than transient elements (Fig. 2, Online column of matrix L), but does not re-sample either the Resource 3). species-trait (matrix Q) or environment-site (matrix R) Arthropod species associated with dry-open habi- relations; appropriate (Dray and Legendre 2008)as tats occurred across all three landscape elements. For traits were determined from the literature not empir- dry-open carabids, mean species richness per sample ical sampling, while the environment matrix was was similar between core patches and both types of classified a priori. Prior to analysis, arthropod abun- connectivity element, although their aggregate abun- dance data were square root transformed to reduce the dance was lower in corridors (Fig. 3, Online Resource effect of dominant species. We conducted the analysis 4). Rarefaction showed cumulative richness of dry- first on all recorded species, then separately on only open carabids did not differ between core patches, those associated with dry-open habitats, and finally for corridors or transient patches (Online Resource 2). For the subset of species recorded on grass-heath sites. dry-open spiders, mean richness per sample was Fourth-corner analyses were calculated with 9999 similar between transient elements and core patches permutations, adjusted using false discovery rate but lower in corridors, while abundance was lower in correction procedures that account for repeated test- transient patches than core patches and was again ing, using the ade4 package (Dray and Dufour 2007). lowest in corridors. Cumulative richness of specialist spiders was also significantly lower in corridors (Online Resource 2). Whilst core patches were Results dominated by dry-open species, both types of con- nectivity elements had greater representation of gen- We identified 3660 carabid individuals from 70 eralist and woodland species. For generalists, mean species and 12,483 spiders from 115 species, of which abundance and richness per sample were similar 11,521 spiders from 60 ground-active species were between transient and corridor elements for both taxa, considered in analyses. Pitfall trapping within each with generalists particularly dominant in spider landscape element effectively represented carabids assemblages (Fig. 3). Rarefaction showed similar and ground-spiders with species accumulation curves rates of species accumulation of generalists across approaching asymptotes (Online Resource 2). elements for both taxa (Online Resource 2). Species Carabid and spider assemblages were also success- associated with woodland were particularly scarce in fully represented in ordination space using NMDS the open grass-heath sites (with only two woodland (stress: 0.21 and 0.19 respectively). Both carabid and spider individuals and 33 woodland carabid individ- spider composition (Fig. 2) differed significantly uals recorded), but contributed to assemblages in both between each of the three landscape elements (mv- connectivity elements, with woodland spiders having abund GLMs: deviance = 758.9, P \ 0.001; 123 Landscape Ecol

Fig. 2 Non-metric multi-dimensional scaling (NMDS) ordina- structure variables are displayed as vectors, longer vectors tion of ground-active carabid and spider assemblages represent stronger association with species ordination. Online (stress = 0.21 and 0.19 respectively). Open circles represent Resource 3 gives the vector, R2 and p-value for each sampled transects and shaded ellipses the 95% confidence environmental variable intervals of habitat element centroids. Significant habitat greater mean abundance and richness in corridors than negatively associated with dry-open habitat associa- transient patches (Fig. 3). tion and a running hunting strategy. Spider PCoA axis two was positively associated with strongly inter- Inter-relationship among traits correlated traits of running hunting strategy and larger body size, and with flight dispersal, and was nega- The first two axes of the PCoA represented 57.9% and tively associated with woodland habitat association 60.8% of the variation in traits among carabid and and a stalking hunting strategy. The spider PCoA spider species respectively (Fig. 4, Online Resource showed clustering of the main families (Fig. 4). The 5), indicating inter-correlation among traits in each large, running hunters of the families Lycosidae, case. For carabids, PCoA axis one was positively Gnaphosidae and Clubionidae were grouped towards associated with larger body size, brachypterous wing the upper part of the ordination, but varied along axis morphology and carnivorous feeding, and negatively one between species associated with dry-open habitats associated with inter-correlated traits of herbivory and (in the upper-left of the ordination) and generalist macroptery. Carabids associated with dry-open habitat species (upper right). Thomisidae, Theridiidae and were negatively related with PCoA axis one and Salticidae, were associated with ambush and stalking positively with axis two, habitat generalists were hunting strategies (lower part of the ordination) and negatively related with axis two. Distribution of tended to be smaller-bodied and habitat generalist or carabid tribes within the PCoA suggested some woodland-associated species. phylogenetic pattern, with herbivory only found in the Zabrini (Amara and Curtonotus) and Harpalini Species traits among landscape elements (Harpalus, Bradycellus, Ophonus), which tended to be macropterous and mainly associated with dry-open Fourth corner analysis revealed substantial differences habitats. Wing morphology was not strongly phylo- in the trait composition of assemblages between the genetically conserved in remaining carabid tribes. three landscape elements. Of the seven carabid and For spiders, PCoA axis one was positively associ- five spider traits tested, most correlated significantly ated with ambush hunting and habitat generalists and with at least one landscape element (Table 1). Notably,

123 Landscape Ecol

Carabids Spiders a a 180 40 160 a 140 30 120 a a 100 b a 20 b 80

a 60 c a 10

Mean abundance (+/− se) Mean abundance 40 a b b 20 a b b c 0 0

Transient Corridor Core Transient Corridor Core

a a 7 a 18 Dry−open Generalist 16 a 6 a Woodland 14 5 a 12 a 4 10 a a b 3 8 a a 6 b 2 b 4 a Mean species richness (+/− se) 1 b 2 b c 0 0

Transient Corridor Core Transient Corridor Core

Fig. 3 Mean (± SE) of ground-active carabid and spider composite of two sampling periods. Means that share a abundance and species richness shown separately for generalist, superscript (homogenous sub-sets, a–c) do not differ signifi- woodland-associated and dry-open habitat specialists, across cantly (Tukey pairwise comparisons, P [ 0.05). Online three landscape elements. Means are calculated from standard- Resource 4 gives model statistics, P values and means for each ized pitfall trap transects where each sample represents a comparison all traits that significantly correlated with transient and traits conveying an increasing dispersal ability by corridor elements showed an opposite direction of flight (carabid macroptery) or aerial drift (spider correlation between these connectivity element types. ballooning) were not positively correlated with either Both carabid and spider assemblages in transient type of connectivity element. For carabids, macrop- patches were correlated significantly with larger body tery was negatively correlated with corridors, while size. Transient patches were also positively correlated for spiders ballooning was negatively correlated with with a carnivorous diet for carabids and with a transient elements. Wing-dimorphism for carabids running-hunting strategy for spiders. Importantly, was positively correlated with corridors and core

123 Landscape Ecol

Carabids Spiders

1 Harpalini Lycosidae Zabrini Gnaphosidae Pterostichini Theridiidae Lebiini Thomisidae Run Sphodrini Dry-Open Salticidae 0.6 Flight Carabini Clubionidae Nebriini Woodland Liocranidae Generalist Other Other Size Omni Brachy

Size 0 0 Carn Dimorp Dry-Open Herb Ambush Macro

PCoA 2 (20.0% variance) PCoA 2 (24.1% variance) Woodland −0.6 Stalk

Generalist −1

−11 0 −0.8 0 0.8

PCoA 1 (37.9% variance) PCoA 1 (36.7% variance)

Fig. 4 Principal coordinate analysis (PCoA) showing associ- correlations, with the length and direction indicating the ations among ten traits of 69 carabid species and eight traits of relationship with composite PCoA axes (see Online Resource 59 spider species. Trait vectors represent the Spearman 1 for trait details and Online Resource 5 for trait loadings)

Table 1 Fourth-corner test Carabid—all species Carabid—dry-open results in which different landscape elements are Transient Corridor Core patch Transient Corridor Core patch related to species traits (see Online Resource 1 for trait Body size ?-?- details) Carnivorous ?- ?- Herbivorous -? ? - Omnivorous Macropterous -?-- Brachypterous --?? Wing-dimorphic -?? Spiders—all species Spiders—dry-open Transient Corridor Core patch Transient Corridor Core patch

Body size ?-?- Fourth-corner analysis used Running hunter ?-?- permutation Model 1 and Ambush hunter -? 9999 permutations (see Stalking hunter Online Resource 6 for model statistics) Flight (ballooning) -?-?

patches, but negatively correlated with transient When considering only those species associated elements. Trait correlations tended to be reversed with dry-open habitats, several important trait-habitat between core patches and transient elements; for correlations were apparent. For dry-open carabids, example, small-bodied herbivorous carabids and corridors were correlated with a carnivorous diet and small-bodied ambush-hunting spiders were correlated lack of flight dispersal (brachyptery), while transient with core patches. elements were correlated with large-bodied, herbi- vores with flight dispersal (Table 1). For dry-open 123 Landscape Ecol habitat spiders, transient elements were again corre- core patches of grass heath habitat. While both types lated with large, running hunters lacking ballooning of connectivity element also contained significant dispersal, while corridors were not correlated with numbers of specialist carabids, core patches contained particular traits. Finally, restricting fourth-corner few species that were not specialists of dry-open procedures to only species recorded in core patches habitat. Our results are consistent with others that find did not provide any addition correlations for either generalist species resilient to landscape change and taxa (Online Resource 6) with results largely consis- dominant in disturbed or human-altered landscapes tent to previous analyses of the entire assemblage. (Robinson and Sutherland 2002; Ewers and Didham 2008; Smith et al. 2015). As generalist species tend to be those associated with invasions and also better able Discussion to cope with changing, fragmented and disturbed ecosystems (Marvier et al. 2004), large numbers of Ground-active carabid and spider assemblages in core generalists may be considered a concern for the patches, transient stepping-stones and corridor con- conservation of natural and semi-natural assemblages. nectivity elements differed in both species composi- Compositional edge effects are well known for tion and trait representation, suggesting filtering by forest boundaries (Downie et al. 1996; Muff et al. traits. Consequently connectivity elements differed in 2009; Campbell et al. 2011; Kowal and Cartar 2012); functional composition, both from each other and from Ewers and Didham (2008) found increased beetle the remaining grass heaths, but nevertheless increased diversity at forest/grassland boundaries with compo- beta diversity at the landscape-scale and supported sitional differences extending over 1 km. The greater obligate dry-open habitat species characteristic of core richness of generalist and woodland-associated spe- patches of grass heath. While corridors contained far cies within our corridor and transient elements, than in fewer dry-open individuals than core patches, tran- core patches, may partly be attributable to an influx of sient stepping-stone elements had similar mean dry- ‘vagrants’ from the adjacent matrix. Inflated species open richness to that in core patches. Our data on trait richness in narrow and small habitat patches is well filtering did not resolve to generalised colonisation documented for invertebrates (Halme and Niemela hypotheses that cut across these two taxa, due to 1993; Driscoll and Weir 2005). Webb and Hopkins differences between beetles and spiders in dispersal (1984) found that small heathland fragments sup- mechanisms and also their patterns of associations ported greater invertebrate richness than larger intact among traits. However, patterns regarding habitat sites, while Halme and Niemela (1993) found similar association, generalist domination of connectivity patterns in carabids inhabiting small forest fragments elements, and increased species diversity in connec- in Finland. In the current study it is likely that the tivity elements were consistent with our predictions narrow shape of corridors and the relatively small size (P3), while dispersal, habitat structure (patch quality) of transient elements (and hence large edge ratios), and edge effects (patch configuration) all appeared to favoured the incursion of woodland and generalist influence differences in trait composition between species from the matrix. elements. However, the greater species richness in connec- tivity elements also reflects the greater structural Assemblage composition among landscape heterogeneity of habitats within these elements elements (Bieringer et al. 2013), not just edge-incursion (‘spill-over’) from the matrix. Both types of connec- We predicted that generalist species would dominate tivity element contained a juxtaposition of: open the connectivity elements and that richness and micro-sites with bare disturbed mineral soil (trackway abundance would be greater in these patches, as wheelings and ploughed planting rows), taller grassy expected from edge- and ecotone-effects. We found herbaceous vegetation (trackway verges and baulks that generalist species made up a large proportion of between planting rows) and woody ecotones at their the carabid assemblage and dominated spider assem- margins, providing opportunities for species associ- blages, in both types of connectivity elements, with ated with these individual habitats but also those greater overall species richness of both taxa than in requiring micro-habitat juxtaposition and mosaics 123 Landscape Ecol

(Dolman et al. 2012). Similarly, Driscoll and Weir narrow corridors may have facilitated their colonisa- (2005) found linear strips of Australian mallee habitat tion by aerial-dispersing species. were species-rich due to both ‘strip-specialists’ and matrix species. Assemblages in the core patches of Trait filtering grass heath habitat may represent not so much a pristine reference sample, but a sub-set of the We predicted that assemblages developing in transient landscape-wide biota filtered to exclude generalist patches of open habitat would be represented by small- and woodland associated species. This reflects the bodied, aerial-dispersing species, whereas corridors historic development of heathland from degraded would be dominated by large-bodied, active-hunters. pasture-woodland (Fuller et al. 2017). Trait correlations were in fact opposite to this Importantly, we found that the ground-active prediction. However, while small body size and traits carabid and spider assemblages in connectivity ele- conveying greater dispersal ability were not more ments were significantly different to each other as well prevalent in transient patches overall; when analyses as to assemblages of core patches. Although generalist were confined to specialists of dry-open habitat (that abundance and richness did not differ between tran- are excluded from and do not percolate through the sient stepping-stones and corridor elements for either matrix), flight dispersal significantly correlated with arthropod group, the mean richness and abundance of transient elements for carabids. Where a significant specialist dry-open associated arthropods was greater trait correlation was detected with both corridor and in transient stepping-stone elements than in corridors. transient elements, these were in opposing directions, For spiders, both mean (per sample) and cumulative indicating differential filtering of assemblages in these (measured through rarefaction) specialist richness connectivity elements. were similar between transient elements and core Differences in dispersal trait correlations were patches, but significantly lower in corridors. This is evident between the two arthropod groups and no despite an expectation of efficient dispersal in linear consistent functional response to either connectivity elements with resistant (i.e. internally reflecting) type was found. The ability to disperse aerially is seen borders set in a ‘hard’ matrix (i.e. one that is unlikely as a key trait for insects colonising unstable habitat to provide dispersal or breeding habitat for target (Roff 1990; Ribera et al. 2001) and ballooning spiders species) (Baum et al. 2004; Ockinger and Smith 2008; are often considered early colonists to newly-opened Bertoncelj and Dolman 2013b). This is also contrary to or disturbed habitat, such as after volcanic eruptions findings for a Hempiteran species, where in a soft (Crawford et al. 1995) or in agricultural fields (low-resistance) matrix, both corridors and stepping (Nyffeler and Sunderland 2003; Schmidt and Tscharn- stones significantly improved connectivity, but in a tke 2005). Prevalence of aerial dispersal was only hard matrix stepping-stone elements did not improve significantly greater in core patches for spiders and in dispersal over non-connected controls (Baum et al. transient patches for dry-open carabids. The findings 2004). Despite the expectation of greater ease of for dry-open carabids are consistent with results in colonisation along linear elements in a non-permeable other systems (Gobbi et al. 2007; Moretti and Legg matrix, other mechanisms may contribute to the 2009; Wamser et al. 2012). Moretti and Legg (2009) greater abundance of specialist dry-open associated reported large, highly-mobile insects respond to species in transient stepping-stones. First, although disturbance caused by regular winter fires, while corridors are often conceptualised as conduits for Gobbi et al. (2007) found that younger sites with less individual movement and colonisation between habi- stable soil conditions favouring greater representation tat patches, diffuse resident populations may breed and of winged carbids in an alpine glacial chronosequence. disperse by percolation over several generations. In our long-lived corridors, dry-open carabids con- Second, the larger transient patches, with reduced sisted of flightless carnivorous carabids, consistent shading by adjacent forest and extensive disturbed with results from other stable or linear connectivity mineral soil, may have provided greater habitat elements (Ribera et al. 2001; Wamser et al. 2012). suitability for specialists than provided by corridors. While disturbed and early-successional habitats Third, the larger size of transient patches relative to have greater representation of flight capable species (Gutierrez and Menendez 1997; Ribera et al. 2001; 123 Landscape Ecol

Pedley and Dolman 2014), landscape configuration with poor dispersal to persistent refuges, while those may also influence species distributions at the regional with enhanced dispersal ability can take advantage of scale. Negative correlation of ballooning capable transient and connecting habitat elements. In the spiders with transient stepping-stone elements may plantation landscape studied here, we found that the in part be due to fragmentation of suitable habitat; with assemblages in core patches, corridors and transient passive dispersal representing a high-risk strategy for stepping-stones were dissimilar in terms of both narrow-niche species in fragmented landscapes. Bonte species’ composition and trait representation. Never- et al. (2003b) showed that specialist xerophilic spiders theless, transient elements supported similar richness in fragmented sand dune habitat were less likely to of characteristic dry-open associated specialists for balloon than habitat generalists. We found generalists which remnant heaths are designated. Dry-open habi- were associated with flight dispersal for spiders but not tat species in these transient patches were predomi- for carabids, although brachyptery was associated nately large running spiders and large herbivorous with woodland carabid species, as expected by affinity carabids. However, dispersal strategies for these two of flightless species to more stable habitats (Ribera arthropod groups were likely very different. In con- et al. 2001; Pedley and Dolman 2014). trast, narrow linear corridors were less favourable to For local dispersal, terrestrial movement alone may the open heathland assemblage, with spiders domi- be sufficient for many ground-active species (Samu nated by generalists. To provide connectivity for less- et al. 2003; Brouwers and Newton 2009). In the mobile species and taxa, corridors should reduce edge- current study, large carabids and spiders were corre- related effects, as wider higher-quality corridors are lated with transient stepping-stone elements but not likely to support more specialist species (Lees and corridors, even when analysis was restricted to dry- Peres 2008; Pedley et al. 2013a). Carabids’ ability for open specialists. Results of a systematic review by directional flight, compared to spiders’ passive bal- Brouwers and Newton (2009) showed that larger looning, may explain their greater ability to take carabids covered more ground per day than small advantage of transient connectivity elements within species, likely due to both greater food requirements complex landscapes. Contrasts between carabids and (Lovei and Sunderland 1996) and greater movement spiders in their response to landscape configuration, capability. Kormann et al. (2015) also found larger emphasise the need for assessments to examine arthropods tended to benefit more from increased multiple taxa before making generalisations. Dispersal landscape connectivity, in the form of more closely- corridors and transient stepping-stones provided dif- clustered grassland fragments, than did smaller ferent trait and functional composition and were not species. For spiders, larger body size was associated interchangeable in this system. with running hunters, of which Lycosidae made up the majority of the sampled assemblage. These running Acknowledgements We are grateful to Neal Armour-Chelu hunters have potential terrestrial dispersal distances of for logistical help, the Forestry Commission for study permission and Natural England, Elveden Estate, Forest Heath several hundred metres over a lifetime, with daily District Council, Norfolk Wildlife Trust, and Suffolk Wildlife ranges recorded between 2-50 m (Kiss and Samu Trust for access to heathland sites. Funding was provided by the 2000; Bonte et al. 2003a, 2007). Such lifetime Natural Environment Research Council, Suffolk Biodiversity distances combined with high daily actively rates Partnership and the Suffolk Naturalists’ Society. may explain the positive correlation of running Open Access This article is licensed under a Creative Com- hunters with transient patches, compared to less active mons Attribution 4.0 International License, which permits use, hunting strategies (ambush and stalking) that are sharing, adaptation, distribution and reproduction in any med- associated with smaller body size in spiders. ium or format, as long as you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons licence, and indicate if changes were made. The images or other third party material in this article are included in Conclusions the article’s Creative Commons licence, unless indicated otherwise in a credit line to the material. If material is not included in the article’s Creative Commons licence and your Concepts introduced through the theory of island intended use is not permitted by statutory regulation or exceeds biogeography and metapopulation dynamics suggest the permitted use, you will need to obtain permission directly increasing habitat fragmentation will restrict species 123 Landscape Ecol from the copyright holder. To view a copy of this licence, visit Chetkiewicz CLB, Clair CCS, Boyce MS (2006) Corridors for http://creativecommons.org/licenses/by/4.0/. conservation: Integrating pattern and process. Annu Rev Ecol Evol Syst 37:317–342 Crawford RL, Sugg PM, Edwards JS (1995) Spider arrival and primary establishment on terrain depopulated by volcanic eruption at Mount St-Helens, Washington. Am Midl Nat References 133:60–75 DEFRA (2018) A green future: our 25 year plan to improve the Albert CH, Rayfield B, Dumitru M, Gonzalez A (2017) environment. DEFRA, London Applying network theory to prioritize multispecies habitat Damschen EI, Haddad NM, Orrock JL, Tewksbury JJ, Levey DJ networks that are robust to climate and land-use change. (2006) Corridors increase plant species richness at large Conserv Biol 31:1383–1396 scales. Science 313:1284–1286 Amarasekare P, Possingham H (2001) Patch dynamics and Davis J, Pavlova A, Thompson R, Sunnucks P (2013) Evolu- metapopulation theory: the case of successional species. tionary refugia and ecological refuges: key concepts for J Theor Biol 209:333–344 conserving Australian arid zone freshwater biodiversity Barbaro L, van Halder I (2009) Linking bird, carabid beetle and under climate change. Glob Change Biol 19:1970–1984 butterfly life-history traits to habitat fragmentation in Dolman PM (2012) Mechanisms and processes underlying mosaic landscapes. Ecography 32:321–333 landscape structure effects on bird populations. In: Fuller Bates D, Machler M, Bolker BM, Walker SC (2015) Fitting RJ (ed) Birds and habitat: relationships in changing land- linear mixed-effects models using lme4. J Stat Softw scapes. Cambridge University Press, Cambridge, 67:1–48 pp 93–124 Baum KA, Haynes KJ, Dillemuth FP, Cronin JT (2004) The Dolman PM, Panter CJ, Mossman HL (2012) The biodiversity matrix enhances the effectiveness of corridors and stepping audit approach challenges regional priorities and identifies stones. Ecology 85:2671–2676 a mismatch in conservation. J Appl Ecol 49:986–997 Beier P, Noss RF (1998) Do habitat corridors provide connec- Dolman PM, Sutherland WJ (1992) The ecological changes of tivity? Conserv Biol 12:1241–1252 breckland grass heaths and the consequences of manage- Bell JR, Bohan DA, Shaw EM, Weyman GS (2005) Ballooning ment. J Appl Ecol 29:402–413 dispersal using silk: world fauna, phylogenies, genetics and Downie IS, Coulson JC, Butterfield JEL (1996) Distribution and models. Bull Entomol Res 95:69–114 dynamics of surface-dwelling spiders across a pasture- Bennett AF (2003) Linkages in the landscape: the role of cor- plantation ecotone. Ecography 19:29–40 ridors and connectivity in wildlife conservation, 2nd edn. Dray S, Dufour A (2007) The ade4 package: implementing the IUCN, Gland duality diagram for ecologists. J Stat Softw 22:1–20 Bertoncelj I, Dolman PM (2013a) Conservation potential for Dray S, Legendre P (2008) Testing the species traits-environ- heathland carabid beetle fauna of linear trackways within a ment relationships: the fourth-corner problem revisited. plantation forest. Insect Conserv Divers 6:300–308 Ecology 89:3400–3412 Bertoncelj I, Dolman PM (2013b) The matrix affects trackway Driscoll DA, Weir T (2005) Beetle responses to habitat frag- corridor suitability for an arenicolous specialist beetle. mentation depend on ecological traits, habitat condition, J Insect Conserv 17:503–510 and remnant size. Conserv Biol 19:182–194 Bieringer G, Zulka KP, Milasowszky N, Sauberer N (2013) Duffey E (1968) An ecological analysis of spider fauna of sand Edge effect of a pine plantation reduces dry grassland dunes. J Anim Ecol 37:641–674 invertebrate species richness. Biodivers Conserv Duffey E (1998) Aerial dispersal in spiders. In: Proceedings of 22:2269–2283 the 17th European Colloquium of Arachnology, Edin- Bonte D, Lens L, Maelfait JP, Hoffmann M, Kuijken E (2003a) burgh, pp 187–191 Patch quality and connectivity influence spatial dynamics Ewers RM, Didham RK (2008) Pervasive impact of large-scale in a dune wolfspider. Oecologia 135:227–233 edge effects on a beetle community. Proc Natl Acad Sci Bonte D, Van Belle S, Maelfait JP (2007) Maternal care and USA 105:5426–5429 reproductive state-dependent mobility determine natal Eycott AE, Watkinson AR, Dolman PM (2006a) Ecological dispersal in a wolf spider. Anim Behav 74:63–69 patterns of plant diversity in a plantation forest managed by Bonte D, Vandenbroecke N, Lens L, Maelfait JP (2003b) Low clearfelling. J Appl Ecol 43:1160–1171 propensity for aerial dispersal in specialist spiders from Eycott AE, Watkinson AR, Dolman PM (2006b) The soil fragmented landscapes. Proc R Soc Lond B seedbank of a lowland conifer forest: the impacts of clear- 270:1601–1607 fell management and implications for heathland restora- Brouwers NC, Newton AC (2009) Movement rates of woodland tion. For Ecol Manag 237:280–289 invertebrates: a systematic review of empirical evidence. Eycott AE, Watkinson AR, Hemami MR, Dolman PM (2007) Insect Conserv Divers 2:10–22 The dispersal of vascular plants in a forest mosaic by a Campbell RE, Harding JS, Ewers RM, Thorpe S, Didham RK guild of mammalian herbivores. Oecologia 154:107–118 (2011) Production land use alters edge response functions Fuller RJ, Williamson T, Barnes G, Dolman PM (2017) Human in remnant forest invertebrate communities. Ecol Appl activities and biodiversity opportunities in pre-industrial 21:3147–3161 cultural landscapes: relevance to conservation. J Appl Ecol 54:459–469

123 Landscape Ecol

Gilbert-Norton L, Wilson R, Stevens JR, Beard KH (2010) A Lycosidae) in Hungarian alfalfa fields using mark-recap- meta-analytic review of corridor effectiveness. Conserv ture. Eur J Entomol 97:191–195 Biol 24:660–668 Kormann U, Rosch V, Batary P, Tscharntke T, Orci KM, Samu Gobbi M, Rossaro B, Vater A, De Bernardi F, Pelfini M, F, Scherber C (2015) Local and landscape management Brandmayr P (2007) Environmental features influencing drive trait-mediated biodiversity of nine taxa on small Carabid beetle (Coleoptera) assemblages along a recently grassland fragments. Divers Distrib 21:1204–1217 deglaciated area in the Alpine region. Ecological Ento- Kotze DJ, O’Hara RB (2003) Species decline—but why? mology 32:682–689 Explanations of carabid beetle (Coleoptera, Carabidae) Gower JC (1971) General coefficient of similarity and some of declines in Europe. Oecologia 135:138–148 its properties. Biometrics 27:857 Kowal VA, Cartar RV (2012) Edge effects of three anthro- Gutierrez D, Menendez R (1997) Patterns in the distribution, pogenic disturbances on spider communities in Alberta’s abundance and body size of carabid beetles (Coleoptera: boreal forest. J Insect Conserv 16:613–627 Caraboidea) in relation to dispersal ability. J Biogeogr Krosby M, Tewksbury J, Haddad NM, Hoekstra J (2010) Eco- 24:903–914 logical connectivity for a changing climate. Conserv Biol Haddad NM, Brudvig LA, Damschen EI, Evans DM, Johnson 24:1686–1689 BL, Levey DJ, Orrock JL, Resasco J, Sullivan LL, Lambeets K, Vandegehuchte ML, Maelfait JP, Bonte D (2008) Tewksbury JJ, Wagner SA, Weldon AJ (2014) Potential Understanding the impact of flooding on trait-displace- negative ecological effects of corridors. Conserv Biol ments and shifts in assemblage structure of predatory 28:1178–1187 arthropods on river banks. J Anim Ecol 77:1162–1174 Halme E, Niemela J (1993) Carabid beetles in fragments of Langlands PR, Brennan KEC, Framenau VW, Main BY (2011) coniferous forest. Ann Zool Fenn 30:17–30 Predicting the post-fire responses of animal assemblages: Harvey P, Nellist D, Telfer M (2002) Provisional atlas of british testing a trait-based approach using spiders. J Anim Ecol spiders (Arachnida, Araneae), vols 1, 2. Biological Records 80:558–568 Centre, Abbots Ripton Lawson CR, Bennie JJ, Thomas CD, Hodgson JA, Wilson RJ Heller NE, Zavaleta ES (2009) Biodiversity management in the (2012) Local and landscape management of an expanding face of climate change: a review of 22 years of recom- range margin under climate change. J Appl Ecol mendations. Biol Conserv 142:14–32 49:552–561 Hemami MR, Watkinson AR, Dolman PM (2005) Population Lawton JH, Brotherton PNM, Brown VK, Elphick C, Fitter AH, densities and habitat associations of introduced muntjac Forshaw J, Haddow RW, Hilborne S, Leafe RN, Mace GM, Muntiacus reevesi and native roe deer Capreolus capreolus Southgate MP, Sutherland WJ, Tew TE, Varley J, Wynne in a lowland pine forest. For Ecol Manag 215:224–238 GR (2010) Making space for nature: a review of England’s Hodgson JA, Thomas CD, Cinderby S, Cambridge H, Evans P, wildlife sites and ecological network. DEFRA, London Hill JK (2011) Habitat re-creation strategies for promoting Lees AC, Peres CA (2008) Conservation value of remnant adaptation of species to climate change. Conserv Lett riparian forest corridors of varying quality for Amazonian 4:289–297 birds and mammals. Conserv Biol 22:439–449 Hsieh TC, Ma KH, Chao A (2016) iNEXT: an R package for Lin YC, James R, Dolman PM (2007) Conservation of heathland rarefaction and extrapolation of species diversity (Hill ground beetles (Coleoptera, Carabidae): the value of low- numbers). Methods Ecol Evol 7:1451–1456 land coniferous plantations. Biodivers Conserv Imbach PA, Locatelli B, Molina LG, Ciais P, Leadley PW 16:1337–1358 (2013) Climate change and plant dispersal along corridors Loehle C (2007) Effect of ephemeral stepping stones on in fragmented landscapes of Mesoamerica. Ecol Evol metapopulations on fragmented landscapes. Ecol Complex 3:2917–2932 4:42–47 Isaac NJB, Brotherton PNM, Bullock JM, Gregory RD, Lovei GL, Sunderland KD (1996) Ecology and behavior of Boehning-Gaese K, Connor B, Crick HQP, Freckleton RP, ground beetles (Coleoptera: Carabidae). Annu Rev Ento- Gill JA, Hails RS, Hartikainen M, Hester AJ, Milner- mol 41:231–256 Gulland EJ, Oliver TH, Pearson RG, Sutherland WJ, Luff ML (1998) Provisional atlas of the ground beetles Thomas CD, Travis JMJ, Turnbull LA, Willis K, Wood- (Coleoptera, Carabidae) of Britain. Biological Records ward G, Mace GM (2018) Defining and delivering resilient Centre, Appots Ripton ecological networks: Nature conservation in England. Luff ML (2007) The Carabidae (ground beetles) of Britain and J Appl Ecol 55:2537–2543 Ireland. Royal Entomological Society, St Albans Johst K, Brandl R, Eber S (2002) Metapopulation persistence in Marc P, Canard A, Ysnel F (1999) Spiders (Araneae) useful for dynamic landscapes: the role of dispersal distance. Oikos pest limitation and bioindication. Agric Ecosyst Environ 98:263–270 74:229–273 Jongman RHG, Bouwma IM, Griffioen A, Jones-Walters L, Van Marvier M, Kareiva P, Neubert MG (2004) Habitat destruction, Doorn AM (2011) The pan European ecological network: fragmentation, and disturbance promote invasion by PEEN. Landsc Ecol 26:311–326 habitat generalists in a multispecies metapopulation. Risk Jongman RHG, Kulvik M, Kristiansen I (2004) European eco- Anal 24:869–878 logical networks and greenways. Landsc Urban Plan McGill BJ, Enquist BJ, Weiher E, Westoby M (2006) 68:305–319 Rebuilding community ecology from functional traits. Kiss B, Samu F (2000) Evaluation of population densities of the Trends Ecol Evol 21:178–185 common wolf spider Pardosa agrestis (Araneae: 123 Landscape Ecol

Moretti M, Legg C (2009) Combining plant and animal traits to Robinson RA, Sutherland WJ (2002) Post-war changes in arable assess community functional responses to disturbance. farming and biodiversity in Great Britain. J Appl Ecol Ecography 32:299–309 39:157–176 Muff P, Kropf C, Frick H, Nentwig W, Schmidt-Entling MH Roff DA (1990) The evolution of flightlessness in insects. Ecol (2009) Co-existence of divergent communities at natural Monogr 60:389–421 boundaries: spider (Arachnida: Araneae) diversity across Samu F, Sziranyi A, Kiss B (2003) Foraging in agricultural an alpine timberline. Insect Conserv Divers 2:36–44 fields: local ’sit-and-move’ strategy scales up to risk-averse Nyffeler M, Sunderland KD (2003) Composition, abundance habitat use in a wolf spider. Anim Behav 66:939–947 and pest control potential of spider communities in Santini L, Cornulier T, Bullock JM, Palmer SCF, White SM, agroecosystems: a comparison of European and US stud- Hodgson JA, Bocedi G, Travis JMJ (2016) A trait-based ies. Agric Ecosyst Environ 95:579–612 approach for predicting species responses to environmental Ockinger E, Smith HG (2008) Do corridors promote dispersal in change from sparse data: how well might terrestrial grassland butterflies and other insects? Landsc Ecol mammals track climate change? Glob Change Biol 23:27–40 22:2415–2424 Oksanen J, Blanchet FG, Friendly M, Kindt R, Legendre P, Schmidt MH, Tscharntke T (2005) Landscape context of McGlinn DJ, Minchin PR, O’Hara RB, Simpson GL, sheetweb spider (Araneae: Linyphiidae) abundance in Solymos P, Stevens MHH, Szoecs E, Wagner H (2018) cereal fields. J Biogeogr 32:467–473 Vegan: community ecology package. R package version Schultz CB (1998) Dispersal behavior and its implications for 2.5-3. https://CRAN.R-project.org/package=vegan. reserve design in a rare Oregon butterfly. Conserv Biol Accessed Dec 2018 12:284–292 Pedley SM, Bertoncelj I, Dolman PM (2013a) The value of the Simberloff D, Farr JA, Cox J, Mehlman DW (1992) Movement trackway system within a lowland plantation forest for corridors—conservation bargains or poor investments. ground-active spiders. J Insect Conserv 17:127–137 Conserv Biol 6:493–504 Pedley SM, Dolman PM (2014) Multi-taxa trait and functional Smith YCE, Smith DAE, Seymour CL, Thebault E, van Veen responses to physical disturbance. J Anim Ecol FJF (2015) Response of avian diversity to habitat modifi- 83:1542–1552 cation can be predicted from life-history traits and eco- Pedley SM, Franco AMA, Pankhurst T, Dolman PM (2013b) logical attributes. Landscape Ecol 30:1225–1239 Physical disturbance enhances ecological networks for Topping CJ, Sunderland KD (1992) Limitations to the use of heathland biota: a multiple taxa experiment. Biol Conserv pitfall traps in ecological-studies exemplified by a study of 160:173–182 spiders in a field of winter-wheat. J Appl Ecol 29:485–491 Pedley SM (2012) Effects of experimental disturbance on multi- Wamser S, Diekotter T, Boldt L, Wolters V, Dauber J (2012) taxa assemblages and traits: conservation implication in a Trait-specific effects of habitat isolation on carabid species forest-open landscape mosaic. Doctoral thesis, University richness and community composition in managed grass- of East Anglia, UK lands. Insect Conserv Divers 5:9–18 Pedley SM, Dolman P (unpublished) Forestry clear-fell patches Wang Y, Naumann U, Wright ST, Warton DI (2012) mva- benefit heathland arthropods bund—an R package for model-based analysis of multi- R Development Core Team (2018) R: a language and environ- variate abundance data. Methods Ecol Evol 3:471–474 ment for statistical computing. R Foundation for Statistical Webb NR, Hopkins PJ (1984) Invertebrate diversity on frag- Computing, Vienna mented Calluna heathland. J Appl Ecol 21:921–933 Rainio J, Niemela J (2003) Ground beetles (Coleoptera: Cara- Wright LJ, Hoblyn RA, Green RE, Bowden CGR, Mallord JW, bidae) as bioindicators. Biodivers Conserv 12:487–506 Sutherland WJ, Dolman PM (2009) Importance of climatic Ribera I, Doledec S, Downie IS, Foster GN (2001) Effect of land and environmental change in the demography of a multi- disturbance and stress on species traits of ground beetle brooded passerine, the woodlark Lullula arborea. J Anim assemblages. Ecology 82:1112–1129 Ecol 78:1191–1202 Roberts MJ (1987) The spiders of Great Britain and Ireland. Harley Books, Colchester Publisher’s Note Springer Nature remains neutral with Roberts MJ (1996) Spiders of Britain and Northern Europe. regard to jurisdictional claims in published maps and HarperCollins Publishers Ltd, London institutional affiliations. Robinson JV (1981) The effect of architectural variation in habitat on a spider community—an experimental field- study. Ecology 62:73–80

123