Ecological Applications, 19(1), 2009, pp. 123–132 Ó 2009 by the Ecological Society of America

Grass strip corridors in agricultural landscapes enhance nest-site colonization by solitary wasps

1,3 2 1 ANDREA HOLZSCHUH, INGOLF STEFFAN-DEWENTER, AND TEJA TSCHARNTKE 1Agroecology, Georg-August University, Waldweg 26, D-37073 Go¨ttingen, Germany 2Animal Ecology I, University of Bayreuth, Universita¨tsstr. 30, D-95447 Bayreuth, Germany

Abstract. Corridors that connect otherwise isolated habitats have often been proposed as a management strategy to mitigate negative effects of habitat fragmentation. Non-crop corridors may have the potential to enhance the connectivity for predators in cropland landscapes, especially for species that require multiple habitats, such as cavity- nesting wasps which use wooded habitat for nesting and grassland habitat for foraging. However, the effects of corridors in nonexperimental landscapes have been rarely examined. We studied the species richness and abundance of cavity-nesting wasps and their parasitoids in standardized trap nests located in three habitat types (forest edge, hedge, grass strip) and in three grass-strip types (connected to a forest edge, slightly isolated, highly isolated from a forest edge). Species richness and the abundance of wasps (: Sphecidae, Eumenidae, Pompilidae) were highest at forest edges, which provide natural nesting sites, and lowest in grass strips, with few natural nesting sites. Wasp abundance in grass strips connected to forest edges was 270% higher than in slightly isolated grass strips and 600% higher than in highly isolated grass strips. The abundance of caterpillar-hunting eumenid wasps was 600% higher in connected grass strips than in slightly and highly isolated grass strips. Species richness of wasps was enhanced by 180% in connected grass strips compared to highly isolated grass strips. Parasitism rates were not directly influenced by habitat or grass-strip type, but increased with increasing parasitoid diversity that was higher at forest edges than in grass strips. We conclude that grass-strip corridors enhance the colonization of nesting sites, presumably by facilitating wasp movements. In agricultural landscapes, where nesting sites are limited and food availability changes frequently, rapid colonization of nests may enhance population viability. Higher wasp abundance in connected nesting sites may be directly linked to higher biocontrol of pest caterpillars within the foraging range around nests. Although grass strips can reduce the negative effects of habitat fragmentation, non-crop habitats such as forest habitats and hedges providing nesting sites are required within the home range of wasps to allow reproduction in agricultural landscapes. Key words: cereal fields; connectivity; dispersal; fallow strips; field margins; forest edges; habitat fragmentation; hedges; natural enemies; parasitoids; predators; trap nests.

INTRODUCTION strip corridors that connect remnant non-crop habitats in a crop field matrix may help to conserve Habitat loss and fragmentation are considered as the diversity and associated ecosystem functions in inten- main threats to biodiversity (Saunders et al. 1991, sively used agricultural landscapes. Fahrig 2003). Corridors consisting of small habitat Traditionally, agricultural landscapes consisted of a strips connecting otherwise isolated habitat patches have small-scale patchwork of crop fields, non-crop habitats, been proposed as a management strategy to mitigate and connecting non-crop strips such as hedges and grass negative effects of habitat fragmentation (Simberloff et strips. Many of these non-crop habitats were lost during al. 1992, Rosenberg et al. 1997, Beier and Noss 1998). the last 50 years, when agricultural production was They are expected to facilitate movements between intensified and the area of crop fields expanded in many habitats, to increase population sizes in habitat patches, regions (Tilman et al. 2001, Benton et al. 2003). The loss and to prevent extinction of small populations. The of non-crop area in agricultural landscapes and the purpose of this landscape-scale study was to quantify resulting decline of species diversity have been well corridor effects on nest colonization of cavity-nesting documented (Kremen et al. 2002, Steffan-Dewenter wasps and their interactions with parasitoids. Grass- 2003, Clough et al. 2005, Gabriel et al. 2005, Hendrickx et al. 2007, Holzschuh et al. 2007, O¨ckinger and Smith Manuscript received 22 February 2008; revised 22 May 2008; 2007, Kohler et al. 2008). Even small non-crop strips, accepted 29 May 2008. Corresponding Editor: J. A. Powell. such as field margin strips that are easily established 3 E-mail: [email protected] along the borders of crop fields, have been shown to be 123 124 ANDREA HOLZSCHUH ET AL. Ecological Applications Vol. 19, No. 1 important habitats for many species (e.g., Pywell et al. quently affected by local population extinctions (Kruess 2005, Marshall et al. 2006, Holzschuh et al. 2008). and Tscharntke 1994, Holt 2002). Furthermore, para- However, there are few studies focusing on whether sitoids may have more restrictive dispersal abilities and grass strips are effective in reducing isolation for often perceive isolation at smaller scales than their hosts in agricultural landscapes (but see Joyce et al. 1999, (van Nouhuys and Hanski 2002). Thus, corridors may Collinge et al. 2000, Berggren et al. 2002, Baum et al. alter trophic interactions in favor of the parasitoids. 2004, So¨derstro¨m and Hedblom 2007). The only Further knowledge about corridor effects on trophic replicated study assessing the use of grass strips by interactions may be critical for implementing corridors flying insects in agricultural landscapes found no as successful conservation strategies that aim to promote corridor effects for butterflies (O¨ckinger and Smith one specific trophic level such as wasps as predators of 2008). Most corridor studies on species of open habitats pest insects. were performed in modified landscapes consisting of Using a large-scale approach in real agricultural cleared land surrounded by forest (reviewed in Beier and landscapes, we established standardized trap nests to Noss 1998, Debinski and Holt 2000, O¨ckinger and study how habitat type (forest edge, hedge, grass strip) Smith 2008). However, these results cannot be trans- and grass-strip type (connected, slightly isolated, highly ferred to agricultural landscapes that consist of a isolated) influenced the colonization of these nesting patchwork of habitats differing in habitat quality and sites. First, we hypothesized that species richness and disturbance level. abundances of wasps in trap nests should increase with Organisms in agricultural landscapes are adapted to a increasing availability of natural nesting sites and highly variable environment and are often characterized therefore should be higher in hedges than in grass strips by high mobility and multi-habitat use. Cavity-nesting and higher at forest edges than in hedges. Second, wasps depend on tunnel nests in tree trunks and species richness and abundances of wasps in trap nests in branches in non-crop habitats. Additionally, they visit grass strips should be most similar to forest edges when multiple non-crop habitats and crop fields for provi- a corridor connects trap nests and forest edge. We sioning their brood with arthropod prey and for feeding hypothesized that species richness and abundance of on floral nectar (Klein et al. 2004, Tylianakis et al. 2005). wasps should be higher in connected grass strips than in For multi-habitat users, movements do not only allow slightly isolated grass strips with no connection to a colonization of new habitat patches and exchanges forest edge in 200 m distance, and lowest in highly between populations, but are required for daily foraging isolated grass strips without forest edge in the vicinity in patches separated from the nesting habitat. In this (.600 m). Third, we expected that trap nest colonization way, multi-habitat users are mobile links between areas by higher trophic levels, i.e., parasitoids of wasps, differing in their disturbance levels, and may help to should increase when trap nests are connected to forest maintain ecosystem functions such as biocontrol even in edges through a corridor. sites characterized by high disturbance levels (Lundberg METHODS and Moberg 2003, Tscharntke et al. 2005, Wearing and Harris 2005, Sekercioglu 2006). We can assume that Study sites multi-habitat users do not perceive contrasts between The study was conducted in 2004 around the city of habitat and an unfavorable landscape matrix as Go¨ttingen (518300 N, 98540 E) in southern Lower distinctly as proposed by the island biogeography and Saxony, Germany. The region is characterized by metapopulation theory (Ricketts 2001, Vandermeer and intensively managed agricultural areas dominated by Carvajal 2001, Tscharntke and Brandl 2004). However, cereal fields and patchily distributed fragments of forests corridor benefits can be even greater when matrix- and different seminatural habitats. Twelve spatially habitat contrasts are low than when contrasts are high separated study sites were selected. Study sites were (Baum et al. 2004). Thus, we expect that multi-habitat between 4 and 6 km2 in size, at least 1 km from each users may benefit from movements along non-crop other, and distributed over an area of 22 3 30 km. corridors enabling a rapid colonization of nesting and Within each study site, we established five trap nests, 60 food resources. trap nests in total. As four trap nests were lost during It has been often stressed that corridors can change summer, 56 nests remained for analysis. biotic interactions (Hess 1994, Tewksbury et al. 2002, In order to test our first hypothesis, that wasp species Orrock et al. 2003, Levey et al. 2005, Orrock and richness and abundance increase with increasing avail- Damschen 2005, Townsend and Levey 2005), and may ability of natural nesting sites, trap nests were placed in even have negative effects on focal species by benefiting three habitat types: forest edges, hedges, and grass invasive species or predators (Proches et al. 2005, strips. Forest edges bordered deciduous mixed forests Damschen et al. 2006, Weldon 2006). However, the dominated by the common beech Fagus sylvatica L. effects of corridors on host–parasitoid interactions have Trap nests at forest edges were placed 1 m from the not been evaluated. Higher trophic levels are assumed to outside branches in the grassy strip between the forest have lower abundances and are more dependent on edge and a cereal field (n ¼ 12). Hedges were at least 205 recolonization processes because they are more fre- m long (263 6 29 m; mean 6 SE), bordered farm tracks, January 2009 CORRIDORS AND NEST SITE COLONIZATION 125

FIG. 1. Map of one of 12 study sites, illustrating the spatial arrangement of five wasp trap nests. Trap nests were placed (A) in a forest edge, (B) in a hedge, and (C–E) in grass strips. Grass strips were (C) connected with a forest edge (200 m from the forest edge, with corridor), (D) slightly isolated (200 m, without corridor), or (E) highly isolated (.600 m, without corridor). Trap nests were at least 600 m apart from each other. All farm tracks were accompanied by a grass strip. There were no grass strips between fields. and were not part of a hedge network. Trap nests in and bees is highest between 0 and 300 m distance to a hedges were placed 1 m from the outside branches in the natural forest providing natural nesting sites, whereas grassy strip between the hedge and the farm track (n ¼ species richness at higher distances is constantly 12). Grass strips were situated between a cereal field and declining. Distances between trap nests were at least a farm track, mostly included a narrow ditch, and had a 600 m; thus we considered each trap nest as an naturally developed permanent layer, which was dom- independent replicate. Grass-strip width did not differ inated by grass but also contained flowering herbs. All between connected, slightly isolated, and highly isolated grass strips were part of a grass-strip network with a grass strips (ANOVA, F2,29 ¼ 0.36, P . 0.6). total length of several kilometers. Grass-strip width was Trap nests 2.8 6 0.1 m (mean 6 SE, n ¼ 32 strips), and trap nests were placed in the center of the strip. In order to test the Trap nests enabled us to study species richness, second hypothesis, that the colonization of trap nests abundance, and interactions of aboveground nesting increases in the presence of corridors between source wasps and their natural enemies under standardized habitat and trap nests, we compared three types of grass nest-site conditions (Tscharntke et al. 1998). Trap nests strips per study site (Fig. 1): (1) the ‘‘connected grass consisted of four plastic tubes (20 cm long, 10.5 cm strips’’ ended in a forest edge, and trap nests were placed diameter), each filled with about 200 internodes (20 cm in the grass strip 200 m from the forest edge (n ¼ 11); (2) long)ofcommonreedPhragmites australis. The the ‘‘slightly isolated grass strips’’ paralleled a forest diameters of reed internodes ranged from 2 to 10 mm. edge at a distance of 200 m, and trap nests in these grass The plastic tubes were fixed on a wooden post at a strips were separated from the forest edge by a cereal height of 1.0–1.2 m and shaded by a 41 3 50 cm field (n ¼ 9); (3) the ‘‘highly isolated grass strips’’ were chipboard roof. .600 m from the nearest forest edge (n ¼ 12). Distances The trap nests were left in the field from mid April until between trap nests and forest edges were chosen mid September. In the laboratory, all reed internodes according to the homing distance of trap-nesting bees containing nests were opened. For each reed internode, of similar body size, which varies between 200 and 600 the genus of wasp larvae, the number of brood cells and m, depending on body size (Gathmann and Tscharntke the occurrence of natural enemies were recorded 2002). This is in accordance to data of Klein et al. (2006) (Gathmann and Tscharntke 1999). All reed internodes showing that the species richness of trap-nesting wasps were reared separately to obtain the adults of bees, 126 ANDREA HOLZSCHUH ET AL. Ecological Applications Vol. 19, No. 1 wasps, and their natural enemies for final species total wasps, spider-hunting sphecids, and caterpillar- identification. Most larvae of wasps and natural enemies hunting eumenids. The effects of grass-strip type were were identified to the species level. In some cases, no tested for the parasitism rate of total wasps only, adults emerged or all brood cells were parasitized, so that because the number of isolated grass strips with .10 only the genus (or the family in case of the eumenids) brood cells of these groups was too low (Steffan- could be identified. These reed internodes were included Dewenter and Schiele 2008). The first fixed factor in in the analyses as additional species if no other species of the models was species richness or number of brood cells this genus (or family) were found at the same trap nest. If of hosts, species richness of parasitoids or number of another species of this genus (or family) was found at the parasitized brood cells. Models for these four factors same trap nest, the unidentified species was assumed to were calculated separately because the factors were be the same as the identified species. correlated (Spearman rank correlation, all R . 0.6, P , The parasitism rate was calculated as the number of 0.001). The second fixed factor was ‘‘habitat type’’ or parasitized brood cells divided by the total number of ‘‘grass-strip type.’’ Nonsignificant factors (P . 0.05) brood cells per trap nest. We calculated the parasitism were removed in a manual stepwise backward selection, rate for trap nests with at least 10 brood cells to avoid and models with at least one significant fixed factor are overestimation of parasitism when hosts were rare. presented. Additionally, we performed separate analyses for the Linear mixed-effect models were computed using the two most dominant wasp groups belonging to different ‘‘lme’’ function in the R package ‘‘nlme’’ (version 2.1.1; functional groups of predators (Trypoxylon spp., Sphe- R Development Core Team 2004). Significant effects of cidae, predators of spiders; Eumenidae, predators of the fixed factor revealed by Wald-type F tests were lepidopterous larvae) and their natural enemies. Trap further inspected using the contrasts between mean nests were also colonized by aboveground nesting levels of the trap nest location types. The estimated solitary bees. The generalist red mason bee Osmia rufa contrasts were computed using the ‘‘estimable’’ function was found in 92% of the reed internodes colonized by in the package ‘‘gregmisc.’’ P values of multiple bees. Other bee species were too rare to be analyzed. The comparisons were corrected by the Holm correction study design did not enable us to test corridor effects on (Aickin and Gensler 1996). We transformed the number the red mason bees because their abundance was even of species and the number of brood cells (log10(x þ 1)) significantly lower in trap nests at forest edges than in and the percentage values of the parasitism rate (arcsine- hedges or grass strips (P , 0.05, linear mixed-effects square-root transformation) to meet the assumptions of model with habitat type as fixed factor and site as constant error variance and normality of errors (Sokal random factor). Thus, for the red mason bee, we had to and Rohlf 1995). reject the underlying hypothesis that forest edges serve as starting point for the colonization of trap nests. RESULTS In total, 2470 brood cells of 11 wasp species were Statistical analyses collected from 56 trap nests. We found 1159 spider- We used linear mixed-effects models with one random hunting sphecids (two species; Table 1), 893 caterpillar- factor and one fixed factor (habitat type or grass-strip hunting eumenids (four species), 369 aphid-hunting type) to determine effects of habitat and grass-strip type sphecids (two species), 29 thrips larvae-hunting sphecids on nest colonization (Pinheiro and Bates 2000). Species (one species), and 20 spider-hunting pompilids (two richness and number of brood cells of wasps or species). Twelve species of wasp parasitoids were parasitoids were included as dependent variables. recorded (Table 1). The parasitism rate was 25.5% 6 Additional analyses were performed for the number of 3.0% (mean 6 SE). brood cells of spider-hunting sphecids and caterpillar- hunting eumenids, and for the species richness and Effect of habitat type number of brood cells of parasitoids in nests of spider- In the first step, we tested the effect of habitat type on hunting sphecids and caterpillar-hunting eumenids. nest colonization of wasps (Table 2, Fig. 2). Wasp Trap nests were grouped within a study site by adding species richness increased threefold from 1.4 species in site as random-block factor. First, we analyzed the effect grass strips to 4.2 species at forest edges (Fig. 2). The of habitat type with data from all trap nests. Second, we mean number of brood cells increased by 610%, from analyzed the effect of grass-strip type with data from 10.1 brood cells in grass strips to 72.2 brood cells at grass strips. The number of replicates was 11 for forest edges for total wasps, by 1640% from 1.8 to 31.3 ‘‘connected,’’ 9 for ‘‘slightly isolated,’’ and 12 for ‘‘highly brood cells for spider-hunting sphecids (present in 28% isolated’’ grass strips. Finally, linear mixed-effects of grass strips, 100% of forest edges), and by 480% from models with study site as random block were used to 3.6 to 21.0 brood cells for caterpillar-hunting eumenids test if parasitism rates were influenced by the species (present in 59% of grass strips, 100% of forest edges). richness or the number of brood cells with hosts or Brood cell numbers in hedges were 550% higher than in parasitoids, by habitat type or by grass-strip type. We grass strips for sphecids and 180% higher for total wasps tested the effects of habitat type on the parasitism rate of (Table 2). Forest edges and hedges did not significantly January 2009 CORRIDORS AND NEST SITE COLONIZATION 127

TABLE 1. Trap-nesting species: their prey, abundance, parasitism rates, and parasitoids.

No. brood Parasitized Family, prey Genus Species cells brood cells (%) Parasitoids Sphecidae Spiders Trypoxylon figulus 118 12.7 Chrysis cyanea, C. ignita, Megatoma undata, Braconidae, Ichneumonidae clavicerum 405 9.1 Chrysis spp., M. undata, Melittobia acasta, Ichneumonidae spp. 616 37.0 C. cyanea, C. ignita, assectator, M. undata, Ichneumonidae Passaloecus corniger 75 1.3 Omalus auratus, Ichneumonidae gracilis 57 1.8 Ichneumonidae spp. 237 26.2 G. assectator, M. undata, M. acasta, Ichneumonidae Pemphredon lugens 2 50.0 G. assectator Thrips larvae Spilomena troglodytes 78 0 none Eumenidae Caterpillars Ancistrocerus gazella 85 8.2 C. ignita, M. acasta, Ichneumonidae parietinus 63 1.5 C. ignita, Ichneumonidae trifasciatus 29 20.7 C. ignita, M. undata Symmorphus gracilis 13 23.1 C. ignita, M. undata, M. acasta gen. spp. 702 53.4 C. ignita, M. undata, M. acasta, Braconidae, Ichneumonidae Pompilidae Spiders Dipogon spp. 17 17.6 M. undata, Ichneumonidae Auplopus carbonarius 3 33.3 M. acasta Notes: Data are from 56 trap nests in forest edges (n ¼ 12), hedges (n ¼ 12), and grass strips (n ¼ 32). The abbreviations ‘‘spp.’’ and ‘‘gen.’’ indicate that wasps could not be identified to the species level (or genus level in the case of Eumenidae) if no adults emerged. Braconidae and Ichneumonidae were not further identified. differ in species richness or the number of brood cells for compared to nest colonization in slightly isolated grass any group (Table 2). Parasitoids followed the patterns of strips (paralleling a forest edge in 200 m distance) or their hosts: The species richness of parasitoids and the highly isolated grass strips (without forest edge nearby). number of parasitized brood cells were highest at forest Wasp species richness was 180% higher in connected edges and lowest in grass strips (Table 3). than in highly isolated grass strips, but did not significantly differ between connected and slightly Effects of grass strip corridors isolated grass strips or between slightly and highly In the second step, we tested if nest colonization in isolated grass strips (Table 4, Fig. 3). The total number grass strips connected to a forest edge was enhanced of wasp brood cells was 270% higher in connected than

TABLE 2. Effects of habitat type (forest edge, hedge, grass strip) on species richness and brood cell numbers.

Response variable Habitat type trend F2,42 P Species richness of wasps full model 9.65 ,0.001 forest edge . grass strip ,0.001 hedge vs. grass strip ns forest edge vs. hedge ns Total wasp brood cells full model 9.83 ,0.001 forest edge . grass strip ,0.001 hedge . grass strip 0.067 forest edge . hedge 0.088 Brood cells of caterpillar-hunting eumenids full model 6.08 0.005 forest edge . grass strip 0.004 hedge vs. grass strip ns forest edge vs. hedge ns Brood cells of spider-hunting sphecids full model 16.9 ,0.001 forest edge . grass strip ,0.001 hedge . grass strip 0.003 forest edge . hedge 0.083 Notes: F and P values from linear mixed-effects models including habitat type as the fixed factor and study site as the random-block factor are shown. Trends of differences between habitat types are shown with ‘‘.’’ or ‘‘,’’ indicating that the trend is significant in one direction; ‘‘vs.’’ indicates that there is no significant trend in either direction. P values of multiple comparisons were corrected using the Holm correction; ‘‘ns’’ indicates P . 0.1. 128 ANDREA HOLZSCHUH ET AL. Ecological Applications Vol. 19, No. 1

in slightly isolated grass strips and six times higher in connected than in highly isolated grass strips (Table 4, Fig. 3). For eumenids, the number of brood cells was six times higher in connected grass strips than in slightly and highly isolated grass strips (Table 4). The number of brood cells of spider-hunting sphecids, the species richness of parasitoids, and the number of parasitized brood cells did not significantly differ between grass- strip types (all P . 0.1).

Parasitism rates Parasitism rates were not significantly influenced by habitat type, grass-strip type, number of brood cells of hosts, or species richness of hosts. The parasitism rate increased with increasing species richness of parasitoids

for total wasps (all habitats, F1,36 ¼ 14.47, P , 0.001; grass strips, F1,18 ¼ 8.13, P ¼ 0.011) and for caterpillar- hunting eumenids (all habitats, F1,36 ¼ 24.68, P , 0.001), but not for spider-hunting sphecids. The fixed factor ‘‘species richness of parasitoids’’ could be replaced by the highly correlated factor ‘‘number of brood cells with parasitoids’’ (Spearman rank correlation, R ¼ 0.884, P , 0.001) showing an increase of parasitism rates with increasing number of brood cells with parasitoids in the

models for total wasps (all habitats, F1,36 ¼ 9.52, P ¼ 0.003; grass strips, F1,18 ¼ 8.82, P ¼ 0.008) and for eumenids (all habitats, F1,36 ¼ 11.36, P ¼ 0.002).

FIG. 2. Effects of habitat type on species richness and DISCUSSION number of brood cells of wasps. Results are based on linear mixed-effects models (Table 2). Means and standard errors are The purpose of this study was to quantify habitat and shown. Different letters indicate significant differences (Holm- corridor effects on nest colonization of cavity-nesting corrected P , 0.05). wasps and their interactions with parasitoids. Corridors that connect non-crop nesting habitats in a crop field matrix may help to conserve wasps and associated ecosystem functions in intensively used agricultural landscapes.

TABLE 3. Effect of habitat type (forest edge, hedge, grass strip) on species richness of parasitoids and number of parasitized brood cells.

Species richness No. parasitized of parasitoids brood cells

Response variable Habitat type trend F2,42 P F2,42 P Total wasp nests full model 23.82 ,0.001 15.80 ,0.001 forest edge . grass strip ,0.001 ,0.001 hedge . grass strip 0.005 0.026 forest edge . hedge 0.005 0.026 Nests of caterpillar-hunting full model 14.62 ,0.001 6.88 0.003 eumenids forest edge . grass strip ,0.001 0.002 hedge . grass strip 0.076 ns forest edge . hedge 0.010 0.057 Nests of spider-hunting full model 14.27 ,0.001 14.64 ,0.001 sphecids forest edge . grass strip ,0.001 ,0.001 hedge . grass strip ns 0.061 forest edge . hedge 0.004 0.013 Notes: F and P values from linear mixed-effects models including habitat type as fixed factor and study site as random block factor are shown. Trends of differences between grass-strip types are shown with ‘‘.’’ or ‘‘,’’ indicating that the trend is significant in one direction. P values of multiple comparisons were corrected using the Holm correctionl; ‘‘ns’’ indicates P . 0.1. January 2009 CORRIDORS AND NEST SITE COLONIZATION 129

TABLE 4. Species richness and number of brood cells of wasps in connected, slightly isolated, and highly isolated grass strips.

Response variable Grass-strip type trend F2,18 P Species richness of wasps full model 4.62 0.024 highly isolated , connected 0.021 slightly isolated vs. connected ns highly isolated vs. slightly isolated ns Total wasp brood cells full model 8.41 0.003 highly isolated , connected 0.002 slightly isolated , connected 0.041 highly isolated vs. slightly isolated ns Brood cells of caterpillar-hunting eumenids full model 6.49 0.008 highly isolated , connected 0.024 slightly isolated , connected 0.011 highly isolated vs. slightly isolated ns Brood cells of spider-hunting sphecids full model 0.82 ns Notes: F and P values from linear mixed-effects models including habitat type as the fixed factor and study landscape as the random block factor are shown. Trends of differences between grass- strip types are shown with ‘‘.’’ or ‘‘,’’ indicating that the trend is significant in one direction; ‘‘vs.’’ indicates that there is no significant trend in either direction. P values of multiple comparisons were corrected using the Holm correction; ‘‘ns’’ indicates P . 0.1.

Habitat types The species richness and the number of brood cells of wasps were lowest in grass strips and highest at forest edges. This is consistent with studies from tropical ecosystems showing that trap-nesting wasps were more diverse in forested habitats than in rice fields and pastures (Tylianakis et al. 2005) and more diverse close to the natural forest than in agroforestry cacao systems (Klein et al. 2006). The species richness and the number of brood cells in trap nests can be expected to reflect the number of nesting sites in the surrounding habitat (Tscharntke et al. 1998, 2005, Steffan-Dewenter 2003). However, in our study, all studied wasp groups were also found in trap nests established in grass strips, where natural nesting sites were completely lacking. Generally, natural nesting sites for cavity-nesting wasps are relatively rare in grass strips and restricted to shrubs, trees, tree stumps, and dead wood. Klein et al. (2004) suggested a trade-off between the availability of nesting sites, which may be more abundant in less disturbed areas, and the availability of food resources, which may be more abundant in intensively used fields. The abundances of wasps in Ecuador and of a dominant eumenid wasp in Indonesia were found to increase with increasing land-use intensity, which may have resulted in an increase of pest caterpillars, the main prey of eumenids (Klein et al. 2004, 2006, Tylianakis et al. 2005). High prey abundance in arable fields may cause wasps to search for nesting sites in grass strips adjacent to fields despite the low chance of finding nesting sites away from forest edges and hedges. We assume that FIG. 3. Effects of grass-strip connectivity on the species forest edges play an important role for cavity-nesting richness and number of brood cells of wasps in grass strips. wasps in agricultural landscapes, because large and Trap nests were highly isolated from a forest edge (.600 m, no valuable habitats may serve as starting points for the corridor), slightly isolated (200 m, no corridor), or connected with a forest edge (200 m, grass-strip corridor). Results are colonization of new habitat patches (Debinski and Holt based on linear mixed-effects models (Table 4). Means and 2000). In contrast, nesting sites in grass strips are rare standard errors are shown. Different letters indicate significant but may be important by making new foraging sites differences (Holm-corrected P , 0.05). 130 ANDREA HOLZSCHUH ET AL. Ecological Applications Vol. 19, No. 1 accessible within the foraging distance around the nest. offspring per female could not be recorded, because Grass-strip-nesting wasps can be expected to affect their one female may build brood cells in more than one reed arthropod prey, to change trophic interactions, and to internode. It remains unclear why wasp females prefer enhance biological control in grass strips and adjacent following grass strips instead of crossing cereal fields. fields (Wearing and Harris 2005). We propose three mechanisms: (1) wasps search for nest sites in grass strips rather than in cereal fields because Grass strip connectivity fields are extremely homogeneous; (2) wasps follow Grass strips connected to a forest edge supported 2.7 grass strips because they may provide more floral times more wasp brood cells than slightly isolated grass resources and prey; (3) wasps perceive forest edge and strips that were separated from the forest edge by a grass strip as one habitat type, avoid crossing non-crop– cereal field and six times more brood cells than highly crop boundaries, and prefer following the habitat edges isolated grass strips without forest edge nearby. The (e.g., for butterflies and planthoppers; Ries and Debin- positive effect of the grass-strip corridor was even more ski 2001, Conradt and Roper 2006, Haynes and Cronin distinct for eumenid wasps, with six times more brood 2006). cells in connected than in slightly isolated grass strips. In The loss of grass strips as a result of agricultural contrast to expectations, the vicinity of a forest edge was intensification and the enlargement of arable fields may insufficient to enhance the number of wasp brood cells have contributed to the decline of predators in in grass strips: The nest colonization in slightly and agricultural landscapes by interrupting movements and highly isolated grass strips (200 vs. .600 m distance) did colonization processes. Nest patches connected by not significantly differ. Our results suggest that forest corridors may serve as starting points that benefit edges in intensively used agricultural landscapes only further colonization processes and population viability. have a positive effect as a starting point of colonization High short-term colonization rates may not be always if the forest edge is connected to another nesting habitat related to high population viability because unconnected by a corridor. nest patches may be colonized over the longer term and Positive effects of corridors on flying openland species have similar species richness and abundance as connect- have been well documented from cleared land in a forest ed patches (Beier and Noss 1998). However, in matrix (Sutcliffe and Thomas 1996, Haddad 1999, landscapes with highly limited nesting sites and annually Haddad and Baum 1999, Tewksbury et al. 2002, changing crops and food availability, rapid colonization Haddad et al. 2003, Fried et al. 2005, Haddad and of nest patches may be linked to high overall population Tewksbury 2005, Townsend and Levey 2005). However, viability. compared to agricultural landscapes, the contrast between habitat and matrix was very sharp in these Parasitism rates studies, and insights cannot be directly transferred to Against our expectations that higher trophic levels agricultural mosaic landscapes. As far as we know, there are only two replicated landscape-scale studies (corridor suffer more from isolation than their hosts (Kruess and length . 30 m) on corridor effects on insects in a non- Tscharntke 1994, Tscharntke et al. 1998, Klein et al. forest matrix. Carabids seem to prefer moving along 2006, Albrecht et al. 2007), trophic interactions between hedges, but the exchange between source and receiver predatory wasps and their parasitoids were not directly patches has not been ascertained (Joyce et al. 1999). The affected by habitat type or grass-strip isolation. Para- only landscape-scale study focusing on grass-strip sitism rates increased with increasing species richness corridors and insects in a non-forest matrix showed and abundance of parasitoids only. This is consistent that butterfly dispersal was not influenced by the with results from Tylianakis et al. (2006), who found presence of grass-strip corridors (O¨ckinger and Smith that a higher parasitoid diversity can result in a higher 2008). Evidence for positive corridor effects on species number of attacked host species or a higher number of with lower dispersal ranges exists from studies at smaller attacks per individual host species. Although we found scales (2–30 m) for bush crickets, planthoppers, and no direct relationship between parasitism rate and butterflies (Berggren et al. 2002, Baum et al. 2004, habitat type, forest edges may indirectly enhance So¨derstro¨m and Hedblom 2007), whereas Collinge parastism rates by enhancing the species richness and (2000) found no corridor effects for several insect abundance of parasitoids. Grass-strip types did not groups at this scale. differ in the species richness and the number of brood The higher number of brood cells in connected rather cells with parasitoids, but in the number of brood cells than in slightly isolated grass strips in our study may of hosts, resulting in a trend toward lower parasitism result from a higher number of wasp females colonizing rates in connected trap nests. Corridor effects on the trap nests in connected grass strips. We do not parasitoids were probably lacking because we mainly suspect that the number of offspring per female differed recorded generalist parasitoids (Melittobia acasta, Meg- between grass strip types, because the food resources in atoma undata), which can be expected to be less affected grass strips and adjacent cereal fields can be expected to by isolation than specialist species (Tscharnkte et al. be similar in all grass-strip types. The number of 2005, Albrecht et al. 2007). Further studies will help to January 2009 CORRIDORS AND NEST SITE COLONIZATION 131 elucidate how different species perceive isolation and the Debinski, D. M., and R. D. Holt. 2000. A survey and overview consequences for trophic interactions. of habitat fragmentation experiments. Conservation Biology 14:342–355. CONCLUSIONS Fahrig, L. 2003. Effects of habitat fragmentation on biodiver- sity. Annual Review of Ecology Evolution and Systematics Our results show that forest edges and hedges are 34:487–515. valuable nesting habitats for above-ground nesting Fried, J. H., D. J. Levey, and J. Hogsette. 2005. Habitat wasps in agricultural landscapes. However, the coloni- corridors function as both drift fences and movement conduits for dispersing flies. Oecologia 143:645–651. zation of standardized nests in grass strips suggests that Gabriel, D., C. Thies, and T. Tscharntke. 2005. Local diversity even single old trees or shrubs in open landscapes may of arable weeds increases with landscape complexity. be more rapidly colonized if they are connected to a Perspectives in Plant Ecology, Evolution and Systematics 7: source habitat. Although cavity-nesting wasps are 85–93. adapted to the use of resources in multiple habitats Gathmann, A., and T. Tscharntke. 1999. Landschafts-Bewer- tung mit Bienen und Wespen in Nisthilfen: Artenspektrum, (Klein et al. 2004), crop fields enhanced the isolation of Interaktionen und Bestimmungsschlu¨ssel. Naturschutz und nesting sites. We recommend that landscape-scale Landschaftspflege in Baden-Wu¨rttemberg 73:277–305. studies should focus on the ‘‘effective isolation of Gathmann, A., and T. Tscharntke. 2002. Foraging ranges of habitats’’ (Ricketts 2001) instead of an ‘‘isolation by solitary bees. Journal of Ecology 71:757–764. Haddad, N. M. 1999. Corridor and distance effects on distance.’’ A landscape management strategy that interpatch movements: a landscape experiment with butter- increases connectivity and enhances the rapid coloniza- flies. Ecological Applications 9:612–622. tion of nesting sites may improve the viability of cavity- Haddad, N. M., and K. A. Baum. 1999. An experimental test of nesting wasps in agricultural landscapes and accordingly corridor effects on butterfly densities. Ecological Applica- enhance the predation and biocontrol of lepidopterous tions 9:623–633. Haddad, N. M., D. R. Bowne, A. Cunningham, B. J. larvae and aphids around the nesting habitats. Danielson, D. J. Levey, S. Sargent, and T. Spira. 2003. Corridor use by diverse taxa. Ecology 84:609–615. ACKNOWLEDGMENTS Haddad, N. M., and J. J. Tewksbury. 2005. Low-quality We thank Yann Clough for statistical advice, Reiner habitat corridors as movement conduits for two butterfly Theunert for his help with species identification, and Kristen species. Ecological Applications 15:250–257. Baum, Peter Hamba¨ck, Carsten Dormann, Pe´ter Bata´ry, Haynes, K. J., and J. T. Cronin. 2006. Interpatch movement Jochen Krauss, and Felix Bianchi for helpful comments on and edge effects: the role of behavioral responses to the this manuscript. This research was carried out within the landscape matrix. Oikos 113:43–54. framework of the EU-funded project ‘‘EASY’’ coordinated by Hendrickx, F., et al. 2007. How landscape structure, land-use David Kleijn (QLK5-CT-2002-01495). intensity and habitat diversity affect components of total arthropod diversity in agricultural landscapes. Journal of LITERATURE CITED Applied Ecology 44:340–351. Aickin, M., and H. Gensler. 1996. Adjusting for multiple testing Hess, G. R. 1994. Conservation corridors and contagious when reporting research results: the Bonferroni vs Holm disease—a cautionary note. Conservation Biology 8:256–262. methods. American Journal of Public Health 86:726–728. Holt, R. D. 2002. Food webs in space: on the interplay of Albrecht, M., P. Duelli, B. Schmid, and C. B. Mu¨ller. 2007. dynamic instability and spatial processes. Ecological Re- Interaction diversity within quantified insect food webs in search 17:261–273. restored and adjacent intensively managed meadows. Journal Holzschuh, A., I. Steffan-Dewenter, D. Kleijn, and T. of Animal Ecology 76:1015–1025. Tscharntke. 2007. Diversity of flower-visiting bees in cereal Baum, K. A., K. J. Haynes, F. P. Dillemuth, and J. T. Cronin. fields: effects of farming system, landscape composition and 2004. The matrix enhances the effectiveness of corridors and regional context. Journal of Applied Ecology 44:41–49. stepping stones. Ecology 85:2671–2676. Holzschuh, A., I. Steffan-Dewenter, and T. Tscharntke. 2008. Beier, P., and R. F. Noss. 1998. Do habitat corridors provide Agricultural landscapes with organic crops support higher connectivity? Conservation Biology 12:1241–1252. pollinator diversity. Oikos 117:354–361. Benton, T. G., J. A. Vickery, and J. D. Wilson. 2003. Farmland Joyce, K. A., J. M. Holland, and C. P. Doncaster. 1999. biodiversity: is habitat heterogeneity the key? Trends in Influences of hedgerow intersections and gaps on the Ecology and Evolution 18:182–188. movement of carabid beetles. Bulletin of Entomological Berggren, A., B. Birath, and O. Kindvall. 2002. Effect of Research 89:523–531. corridors and habitat edges on dispersal behavior, movement Klein, A. M., I. Steffan-Dewenter, and T. Tscharntke. 2004. rates, and movement angles in Roesel’s bush-cricket (Metri- Foraging trip duration and density of megachilid bees, optera roeseli). Conservation Biology 16:1562–1569. eumenid wasps and pompilid wasps in tropical agroforestry Clough, Y., A. Kruess, D. Kleijn, and T. Tscharntke. 2005. systems. Journal of Animal Ecology 73:517–525. Spider diversity in cereal fields: comparing factors at local, Klein, A. M., I. Steffan-Dewenter, and T. Tscharntke. 2006. landscape and regional scales. Journal of Biogeography 32: Rain forest promotes trophic interactions and diversity of 2007–2014. trap-nesting hymenoptera in adjacent agroforestry. Journal Collinge, S. K. 2000. Effects of grassland fragmentation on of Animal Ecology 75:315–323. insect species loss, colonization, and movement patterns. Kohler, F., J. Verhulst, R. van Klink, and D. Kleijn. 2008. At Ecology 81:2211–2226. what spatial scale do high-quality habitats enhance the Conradt, L., and T. J. Roper. 2006. Nonrandom movement diversity of forbs and pollinators in intensively farmed behavior at habitat boundaries in two butterfly species: landscapes? Journal of Applied Ecology 45:753–762. implications for dispersal. Ecology 87:125–132. Kremen, C., N. M. Williams, and R. W. Thorp. 2002. Crop Damschen, E. I., N. M. Haddad, J. L. Orrock, J. J. Tewksbury, pollination from native bees at risk from agricultural and D. J. Levey. 2006. Corridors increase plant species intensification. Proceedings of the National Academy of richness at large scales. Science 313:1284–1286. Sciences (USA) 99:16812–16816. 132 ANDREA HOLZSCHUH ET AL. Ecological Applications Vol. 19, No. 1

Kruess, A., and T. Tscharntke. 1994. Habitat fragmentation, Steffan-Dewenter, I. 2003. The importance of habitat area and species loss, and biological control. Science 264:1581–1584. landscape context for species richness of bees and wasps in Levey, D. J., B. M. Bolker, J. J. Tewksbury, S. Sargent, and fragmented orchard meadows. Conservation Biology 17: N. M. Haddad. 2005. Effects of landscape corridors on seed 1036–1044. dispersal by birds. Science 309:46–148. Steffan-Dewenter, I., and S. Schiele. 2008. Do resources or Lundberg, J., and F. Moberg. 2003. Mobile link organisms and natural enemies drive bee population dynamics in fragmented ecosystem functioning: implications for ecosystem resilience habitats? Ecology 89:1375–1387. and management. Ecosystems 6:87–98. Sutcliffe, O. L., and C. D. Thomas. 1996. Open corridors Marshall, E. J. P., T. M. West, and D. Kleijn. 2006. Impacts of appear to facilitate dispersal by ringlet butterflies (Aphanto- an agri-environment field margin prescription on the flora pus hyperantus) between woodland clearings. Conservation and fauna of arable farmland in different landscapes. Biology 10:1359–1365. Agriculture, Ecosystems and Environment 113:36–44. ¨ Tewksbury, J. J., D. J. Levey, N. M. Haddad, S. Sargent, J. L. Ockinger, E., and H. G. Smith. 2007. Semi-natural grasslands Orrock, A. Weldon, B. J. Danielson, J. Brinkerhoff, E. I. as population sources for pollinating insects in agricultural Damschen, and P. Townsend. 2002. Corridors affect plants, landscapes. Journal of Applied Ecology 44:50–59. , and their interactions in fragmented landscapes. O¨ckinger, E., and H. G. Smith. 2008. Do corridors promote Proceedings of the National Academy of Sciences (USA) 99: dispersal in grassland butterflies and other insects? Land- scape Ecology 23:27–40. 12923–12926. Orrock, J. L., and E. I. Damschen. 2005. Corridors cause Tilman, D., J. Fargione, B. Wolff, C. D’Antonio, A. Dobson, differential seed predation. Ecological Applications 15:793– R. Howarth, D. Schindler, W. H. Schlesinger, D. Simberloff, 798. and D. Swackhamer. 2001. Forecasting agriculturally driven Orrock, J. L., B. J. Danielson, M. J. Burns, and D. J. Levey. global environmental change. Science 292:281–284. 2003. Spatial ecology of predator–prey interactions: corridors Townsend, P. A., and D. J. Levey. 2005. An experimental test and patch shape influence seed predation. Ecology 84:2589– of whether habitat corridors affect pollen transfer. Ecology 2599. 86:466–475. Pinheiro, J. B., and D. M. Bates. 2000. Mixed-effects models in Tscharntke, T., and R. Brandl. 2004. Plant–insect interactions S and S-Plus. Springer, New York, New York, USA. in fragmented landscapes. Annual Review of Entomology 49: Proches, S., J. R. U. Wilson, R. Veldtman, J. M. Kalwij, D. M. 405–430. Richardson, and S. L. Chown. 2005. Landscape corridors: Tscharntke, T., A. Gathmann, and I. Steffan-Dewenter. 1998. possible dangers? Science 310:779–783. Bioindication using trap-nesting bees, wasps and parasitoids: Pywell, R. F., E. A. Warman, C. Carvell, T. H. Sparks, L. V. community structure and interactions. Journal of Applied Dicks, D. Bennett, A. Wright, C. N. R. Critchley, and A. Ecology 35:708–719. Sherwood. 2005. Providing foraging resources for bumble- Tscharntke, T., A. M. Klein, A. Kruess, I. Steffan-Dewenter, bees in intensively farmed landscapes. Biological Conserva- and C. Thies. 2005. Landscape perspectives on agricultural tion 121:479–494. intensification and biodiversity—ecosystem service manage- R Development Core Team. 2004. R: a language and ment. Ecology Letters 8:857–874. environment for statistical computing. Foundation for Tylianakis, J. M., A. M. Klein, and T. Tscharntke. 2005. Statistical Computing, Vienna, Austria. Spatiotemporal variation in the diversity of hymenoptera Ricketts, T. H. 2001. The matrix matters: effective isolation in across a tropical habitat gradient. Ecology 12:3296–3302. fragmented landscapes. American Naturalist 158:87–99. Tylianakis, J. M., T. Tscharntke, and A. M. Klein. 2006. Rosenberg, D. K., B. R. Noon, and E. C. Meslow. 1997. Diversity, ecosystem function, and stability of parasitoid host Biological corridors: form, function, and efficacy. BioScience interactions across a tropical habitat gradient. Ecology 87: 47:677–687. 3047–3057. Saunders, D. A., R. J. Hobbs, and C. R. Margules. 1991. Vandermeer, J., and R. Carvajal. 2001. Metapopulation Biological consequences of ecosystem fragmentation—a review. Conservation Biology 5:18–32. dynamics and the quality of the matrix. American Naturalist Sekercioglu, C. H. 2006. Increasing awareness of avian 158:211–220. ecological function. Trends in Ecology and Evolution 21: van Nouhuys, S., and I. Hanski. 2002. Colonization rates and 464–471. distances of a host butterfly and two specific parasitoids in a Simberloff, D., J. A. Farr, J. Cox, and D. W. Mehlman. 1992. fragmented landscape. Journal of Animal Ecology 71:639– Movement corridors—conservation bargains or poor invest- 650. ments. Conservation Biology 6:493–504. Wearing, C. H., and A. C. Harris. 2005. Evaluation of the So¨derstro¨m, B., and M. Hedblom. 2007. Comparing move- predatory wasp, Ancistrocerus gazella, for biological control ments of four butterfly species in experimental grassland of leafrollers in Otago fruit crops. II. Wasp phenology and strips. Journal of Insect Conservation 11:333–342. seasonal changes in prey composition. Biocontrol Science Sokal, R. R., and F. J. Rohlf. 1995. Biometry: the principles and Technology 15:281–298. and practice of statistics in biological research. Third edition. Weldon, A. J. 2006. How corridors reduce Indigo Bunting nest Freeman, New York, New York, USA. success. Conservation Biology 20:300–1305.