The role of mosaicity of the post-agriculture area of the Kampinos National Park in determining the diversity of species of spider wasps ( Hymenoptera : Pompilidae )

From 2000 to 2006 a total of 52 CPUE samples of spider wasps (Hymenoptera: Pompilidae) were collected in the mosaic landscape of the Kampinos National Park (Poland), which is a UNESCO Biosphere Reserve. The hypothesis tested was that both pompilid species richness and abundance is positively associated with spatial heterogeneity. The patterns in spider wasp assemblages were identified using a Kohonen artificial neural network (i.e., self-organizing map). The highest numbers and greatest species richness of pompilids were recorded at sites in open habitats, especially those located on dry soils that are the preferred nesting sites of ground nesting (endogeic) spider wasps. However, pompilid distribution depended not only on the character of a sampling site, but also its location in a mosaic of habitats. The highest values of pompilid abundance and species richness were also recorded at sites surrounded by several different habitats. Both parameters were lower at sites in more homogenous areas, where there were fewer habitats within the flight ranges of spider wasps. A group of three “cultural species” (Agenioideus cinctellus, A. sericeus and Auplopus carbonarius) was identified that is significantly associated with wooden buildings. The results of this study are thus consistent with the concept that habitat heterogeneity enhances faunal diversity, as each type of habitat, including anthropogenic ones, potentially contributes to a wider range of available resources. 35 * Corresponding author. MATERIAL AND METHODS


INTRODUCTION
Members of different genera or even subfamilies of Pompilidae look very similar, which is why their identification requires both experience and patience (Wi niowski, 2009).Traditionally the study of this group has been regarded as difficult (Day, 1988).Probably this is why there are only a few ecological papers on pompilids.The latter are studied either as a part of (1) assemblages of predatory Aculeata occurring in certain habitats including raised bogs (Shlyakhtenok, 2007), pine forests (Shlyakhtenok & Agunovich, 2001), linden-oak-hornbeam and thermophilous oak forests and moist meadows (Skibiska, 1989a, b), or (2) a guild of cavity-nesting Aculeata (Tscharntke et al., 1998;Buschini & Woiski, 2008).Moreover, except for the papers of Wi niowski (2005) and Wi niowski & Werstak (2009), there are none on the relationship between the distribution of pompilids and their habitat preferences.
Each species of insect requires a particular set of resources (Speight et al., 2008).The resources used by spider hunting wasps are complex and diverse.They need appropriate nesting sites, nectar or aphid honeydew as a source of energy for adults and a spider-host for larval development (Day, 1988;O'Neill, 2001;Wi niowski, 2009).Nesting behaviour and prey choice varies within the family (Evans & West-Eberhard, 1970;Day, 1988;O'Neill, 2001;Wi niowski, 2009).Pompilids hunt for spiders of many families, choosing spiders of an ecological group rather than a particular species (Finch, 1997;Wi niowski, 2009) and visit plants of many families (Wi niowski, 2009).Nesting strategies of the different species vary with respect to the part of a habitat in which they build their nest, the type of substrate they use and the materials required for nest construction (Evans & West-Eberhard, 1970;Day, 1988;O'Neill, 2001;Winiowski, 2009).This variability in the requirements of the different species accounts for their different responses to particular habitats (Day, 1988;Wi niowski, 2009).
Pompilid species, therefore, are found in environments where they encounter appropriate living conditions.Thus, their species richness and abundance should be positively associated with the diversity of habitats (landscape mosaicity) in an area, which is the hypothesis tested in this paper.The Kampinos National Park (Poland) was chosen for this study because the mosaic landscape of this UNESCO Biosphere Reserve should presumably enhance survival of species and promote diversity (Weibull et al., 2000;Steffan-Dewenter, 2002;Bennett et al., 2006) of this rarely studied group of Aculeata.
digits indicating the site and (2) two letters the habitat sampled (FA -fallow, ME -meadow, PG -psammophilous grassland, SD -sand dune, FR -forest, FT -fruit trees, AF -abandoned farm, and WB -wooden building); codes for samples from fallows additionally contain two digits indicating the age (in years) of the fallow; each code ends with 2 digits in subscript indicating the year sampled (Table 1).For example: the sample 24FA0605 was collected in 2005 at site No 24 located in fallow land abandoned 6 years previously.Because open habitats are preferred by most pompilid species, the samples were collected mainly in open areas.The numbers of samples from the latter reflect to a certain extent the diversity of the landscape outside dense forests, i.e. the number of samples collected differed between habitats in order to avoid habitat under-and overrepresentation in the dataset.Samples from dense forest were collected just in order to provide a point of reference.
More than one sample was collected in successive years at some sites located in fallows (28 samples), meadows (4 samples) and fresh coniferous forest (2 samples) (Table 1).The repetitions could either be deleted in order to base the study on a single sample from each site, or included in the analyses.In an attempt to determine whether other effects are revealed when a richer dataset is used, the latter option was adopted.
The catches of pompilid wasps were standardized by expressing them in terms of catch per unit effort (CPUE) using water pan-traps.Each trap was a plastic bowl, 20 cm in diameter, filled two-third full with a mixture of water (95%), glycol (5%) for preservation and a detergent to break the surface tension.Three traps (two yellow and one white) were placed at each site: (1) on the ground; (2) on poles at the same height as the mean height of the surrounding vegetation; (3) hung on trees; or, (4) placed on the walls of buildings.Each trap was emptied 19 times at ten day intervals during the study.The 19 catches from 3 traps at a site in a given year were treated as one sample.During this study 156 pan traps were used.
Water pan traps, especially yellow and white ones, are effective tools for collecting flower-visiting insects, including pompilids (Duelli et al., 1998;Wi niowski, 2009).This low cost method is not only reliable, but also simple to use and inconspicuous.The latter feature was especially important, because Malaise traps were often stolen or destroyed.Moreover, the pan traps could be used in all the habitats studied unlike Malaise traps or sweeping (both not appropriate for collecting samples on buildings) (Westphal et al., 2008).The effectiveness of pan traps is not dependent on the size, conspicuousness and diurnal activity of insects, weather or researcher's experience, all of which considerably influence results based on sweeping or transect walk (Westphal et al., 2008;Wi niowski, 2009).
Information on the preferred habitats and nesting behaviour of the species was based on the papers by Day (1988) and Winiowski (2009).Each species was classified according to its habitat preferences in one of four groups: (1)  To characterize nesting habits, the species were divided into the following seven guilds (Evans & West-Eberhard, 1970): (i) no nest constructed, prey left in situ; (ii) no nest constructed, clep-toparasites; (iii) nest dug in the ground; (iv) nest in pre-existing cavities in the ground; (v) nest in pre-existing cavities in or above the ground; (vi) nest in pre-existing cavities above the ground; and, (vii) free standing nest above the ground.
The patterns in the abundance of spider wasps were determined using a Kohonen artificial neural network (ANN).Kohonen ANNs are also referred to as self-organizing maps (SOMs).Like all ANNs, they easily deal with non-linearly related variables that are skewly distributed, which is especially useful in analyses of organism counts.The latter are seldom normally distributed (because of many zeroes) and usually it is not possible to normalise them with any transformation (Quinn & Keough, 2002).The unequal number of samples collected from the habitats and/or sites studied is also not a problem in sample ordination performed with Kohonen ANNs.
ANNs are simple structural and functional models of a human brain.They consist of processing units called neurons or nodes.Kohonen ANNs are built of two (input and output) layers of neurons.The number of input neurons is equal to the number of variables (in this study the abundances of 42 taxa).The output neurons are arranged on a two-dimensional lattice (in this study 4 × 4, selected arbitrarily from among other tested options) (Fig. 2).The dataset (42 taxa × 52 samples, log transformed and normalised 0-1) was presented to the input neurons.Each input neuron transmitted information to all the output neurons during the training process.On this basis a model of a spider wasp sample was created in each output neuron.Similarity of sample models is related to the SOM topology, i.e., models in distant neurons differ and those in neighbouring neurons are similar.A given spider wasp sample was finally assigned to the most similar model and the respective output neuron.As a conse-Fig.1. Map of the study area with sites sampled marked with black circles.Names of localities are in bold font.quence, samples in distant neurons are dissimilar while samples in nearby neurons are similar.The latter may not be true if the neighbouring neurons are in different clusters.The clusters of neurons and respective sample models were identified using a hierarchical cluster analysis (Ward linkage, Euclidean distance).More details on Kohonen ANNs are available in many papers on the ecology of invertebrates (Chon et al., 2001;Park et al., 2003;Lek et al., 2005;Penczak et al., 2006;Song et al., 2006;Lek-Ang at al., 2007;Céréghino et al., 2008;Ruggiero et al., 2008;Tszydel et al., 2009;Ba bura et al., 2010).It is worth mentioning that there are a few studies employing SOMs on terrestrial arthropods (Park & Chung, 2006) terans like ants (Groc et al., 2007;Delabie et al., 2009) or social wasps (Corbara et al., 2009).To our knowledge, there are no studies of the patterns in pompilid assemblages using SOMs.
The software used for simulation of the SOM (SOM Toolbox for Matlab, by LADYBIO, Université Paul Sabatier, Toulouse) allows for the visualization of each species relative abundance in models of a spider wasp samples in the output neurons in the form of a gradient of greyness.This visualization may be very helpful in formulating ecological conclusions as species with the same pattern of greyness in the SOM usually have similar habitat preferences.Additionally, the species association with each SOM sub-cluster of neurons (and respective environmental conditions) can be expressed in numeric form using the indicator value (IndVal) of Dufrêne & Legendre (1997).The IndVal for a species in a sub-cluster of neurons was calculated as a product of (1) species frequency in the sub-cluster, (2) average relative species abundance, i.e. average species abundance in the sub-cluster over the sum of average abundances of the species in all sub-clusters, and (3) a constant 100 in order to produce percentages.IndVal has a maximum (100%) when all spider wasp samples with a given species are in a single subcluster of neurons, and when the species was recorded in all samples assigned to that sub-cluster (Dufrêne & Legendre, 1997).The significance level of the maximum observed IndVal for each species was calculated using a Monte Carlo test.Hence, the IndVals and SOM species planes express (numerically and in the form of a gradient greyness, respectively) the importance of each SOM region to a species and thus complement each other.Both allow an identification of the sub-cluster of neurons in which a given species is most abundant and frequent and thus the abiotic conditions it prefers.Nevertheless, species planes produced for sporadically recorded species may be misleading.This is why they were not drawn for species with maximum IndVal 15% (arbitrary criterion).Thus, species planes are not presented for the following 14 species: Arachnospila rufa (Haupt, 1927), A. spissa (Schiödte, 1837), Anoplius caviventris (Aurivillius, 1907), Evagetes proximus (Dahlbom, 1845), E. subglaber (Haupt, 1941), Pompilius cinereus (Fabricius, 1775), Priocnemis agilis (Schuckard, 1837), P. fennica Haupt, 1927, P. gracilis Haupt, 1927, P. hankoi Móczar, 1944, P. minuta (Vander Linden, 1827), P. parvula Dahlbom, 1845, P. pusilla (Schiödte, 1837) and P. vulgaris (Dufour, 1841) (Table 3).
The data on the type of soil at particular sites come from Konecka-Betley ( 2003 2) hydrogenic (humid) soil (moorsh, muckous soil).Habitat heterogeneity in the vicinity of each site was described as the number of habitats adjoining the habitat studied (Table 1).The significance of the differences between the SOM sub-clusters in this variable, as well as in the numbers of specimens of spider wasp and of spider wasp species richness in samples was assessed using a Kruskal-Wallis test (Zar, 1984).

RESULTS
A total of 2318 spider wasps (Pompilidae) belonging to 42 species were captured at the sites studied (Table 3).The highest abundances of pompilids were recorded in meadows (on average 60.5 specimens per sample), fallows (52.5, ranging from 34.9 in pioneer to 59.6 in old fallows), on an abandoned farm (39 specimens), wooden buildings (34.6 specimens), and in forests, on fruit trees and in psammophilous grasslands (25, 23 and 21.5 specimens, respectively).The minimum value was recorded at Fig. 2. The 52 samples of spider wasps assigned to 16 SOM output neurons arranged in a two dimensional grid (4 × 4).Clusters (AB and CD; separated by a dashed line) and sub-clusters (AB1, AB2, CD1, CD2; are shaded) of neurons and respective sample models were identified using a hierarchical cluster analysis.The code for each sample of spider wasps consists of the site number and two letters for the habitat (AF -abandoned farm, FA -fallow, FT -fruit trees, FR -forest, ME -meadow, PG -psammophilous grassland, SD -sand dune, WB -wooden building); codes for samples from fallows additionally include two digits for the age of the fallow (in years); each code ends with 2 digits in subscript indicating the year the sample was collected.
Clear differences between the clusters in the spatial origin of the samples were recorded, as cluster AB contained only samples from open areas, mainly those from (1) fallows and meadows (90%), and, (2) habitats located on dry soil (76%) (Table 2).Cluster CD grouped all samples from wooden buildings, forests and 43% of the samples from fallows (Table 2).Within cluster AB, the samples in sub-cluster AB1 had a mixed composition, but mainly included samples from fallows and meadows, while in sub-cluster AB2 almost all the samples were from fallows on dry soil (Table 2).Within cluster CD, subcluster CD1 contained samples from various sites, but mainly from (1) fallows including almost all samples from 1-2 year old fallows, i.e. in a pioneer successional stage, and the majority of samples (7 out of 9) from fallows on humid soils, and (2) forests (all samples from this habitat).Sub-cluster CD2 was the only one that did not include samples from fallows; it was comprised only of samples collected on wooden buildings (with one exception) (Fig. 2 and Table 2); its homogeneity was also manifest in the fact that all the 13 samples were assigned to only two neurons (D3, D4) (Fig. 2).
Both clusters and sub-clusters significantly (p < 0.001) differed in the environmental heterogeneity in the vicinity of the sites sampled (Fig. 3).Samples assigned to cluster AB were collected at sites surrounded by more types of habitats than those in cluster CD (Fig. 3).
The abundance of spider wasps in the samples assigned to sub-clusters AB1 and AB2 was significantly greater than in sub-clusters CD1 and CD2 (Fig. 3).Also the species richness in sub-clusters AB1 and AB2 was significantly greater than in CD1 and CD2, and additionally in CD1 than in CD2 (Fig. 3).Similarly, a greater total species richness    was recorded for samples in AB2 (33 species) than AB1 and CD1 (in each 24 species) and the fewest species (8) were recorded in CD2 (Table 3).
According to the patterns in the species importance in the SOM and to the IndVals each cluster/sub-cluster can be characterized by species strongly associated with it and thus with relevant types of habitats (Fig. 4 and Table 3).In sub-cluster AB1, the significant highest IndVals were recorded for 3 eurytopic species (group 1): Priocnemis perturbator, P. coriacea Dahlbom, 1843, Anoplius nigerrimus (Scopoli, 1763), and a species preferring woodland areas and forest margins (group 4): Homonotus sanguinolentus (Fabricius, 1793) (Fig. 4 and Table 3).Sub-cluster AB2 was characterized by thermophilous species (group 2): Arachnospila trivialis (Dahlbom, 1843), A. wesmaeli (Thomson, 1870), Episyron albonotatum (Vander Linden, 1827), their obligate brood parasites (group 3): Evagetes crassicornis (Schuckard, 1837), E. dubius (Vander Linden, 1827) and E. littoralis (Wesmael, 1851), an eurytopic species (group 1): Arachnospila anceps (Wesmael, 1851) and species preferring woodlands and forest edges (group 4): Cryptocheilus notatus (Rossius, 1792) and Arachnospila abnormis (Dahlbom, 1842).Thus, in general mainly eurytopic and thermophilous species (groups 1-3) were associated with cluster AB.In sub-cluster CD1 none of the species had a significant maximum IndVal.Nevertheless, Priocnemis cordivalvata Haupt, 1927, P. gracilis and P. vulgaris (all associated with woodlands and forests edges, group 4) were recorded only in samples assigned to this sub-cluster (Table 3).Sub-cluster CD2 (with respective environmental conditions) was preferred by three species, the eurytopic Agenioideus cinctellus (Spinola, 1808) and Auplopus carbonarius (Scopoli, 1763) (group 1) and thermophilous Agenioideus sericeus Vander Linden, 1827) (group 2), all three recorded mainly on wooden buildings.Generally cluster CD (and relevant habitats) was attractive for eurytopic species and those preferring woodland areas and 42 Fig. 4. Importance, greater the darker the shading, of the 28 species of spider wasp with IndVals > 15% (compare with Table 3) in the models of the spider wasp samples.The shading is scaled independently for each species.The three species associated with wooden buildings are enclosed by a dotted line.The symbol of the sub-cluster with the maximum indicator value (IndVal) recorded for a given species, and the respective significance level (in superscript if > 0.05) are presented on the left side of each species plane.Species with the same pattern in the SOM occurred in similar habitats.
forest edges (groups 1 and 4) (Table 3).It should be noted that if a species is not significantly associated with any particular region of the SOM this is because it either has a wide environmental tolerance or low abundance and frequency in samples, which results in the differences between sub-clusters being insignificant.
Species within most ecological groups of spider wasps had very similar patterns on the SOM, i.e., very similar habitat preferences.In groups 1 and 2, species recorded mainly on wooden buildings with significant maximum IndVals were placed in an additional sub-group (Fig. 4 and Table 3).The remaining eurytopic species (group 1) were associated mainly with cluster AB, though they were also present in sub-cluster CD1 (Table 3).Other thermophilous species (group 2) and their cleptoparasites (group 3) were clearly associated with sub-cluster AB2.Moreover, 3 species: Arachnospila ausa (Tournier, 1890), A. minutula (Dahlbom, 1842) and Evagetes dubius, were recorded only in samples assigned to this sub-cluster.Group 4, which includes species preferring woodlands and forest edges, was the least homogenous in terms of SOM patterns.Three species were associated mainly with cluster AB and three others with cluster CD (Fig. 4 and Table 3).
As far as the nesting preferences are concerned (Table 4), one species, Homonotus sanquinolentus (Fabricius, 1793), which does not construct a nest and is assigned to a separate guild (guild i), had a significant highest IndVal in sub-cluster AB1.Among the other guilds, guild v, which includes species nesting in pre-existing cavities in the ground or above it, was the most heterogeneous, as its species attained the maximum observed relative abundances and/or frequencies in the different sub-clusters, except AB2 (Table 3 and 4).The remaining guilds were associated with only one cluster.The endogeic species (guilds iii and iv) without exception were significantly associated with cluster AB (Table 4).Among them the endogeic species nesting in pre-existing cavities in the ground (guild iv) were most common both in sub-clusters AB1 and AB2, while species excavating nests in the ground (guild iii) in sub-cluster AB2 (except for Caliadurgus fasciatellus (Spinola, 1808) and Anoplius viaticus).Similarly, cleptoparasites (guild ii) of species in guild iii were associated mostly with sub-cluster AB2.Species in guilds vi and vii, which nest in pre-existing cavities above ground and construct free standing nests above ground, were most common in cluster CD (Table 4).In summary, endogeic species, excavating nests in the ground, and their cleptoparasites, were mainly associated with sub-cluster AB2 and hypergeic and endogeic/hypergeic species with sub-clusters CD1 and CD2.

DISCUSSION
The number of species caught during this study is approximately half the number of spider wasps recorded in Poland (Wi niowski, 2009).This proportion is similar to that recorded for other Aculeata taxa in the Kampinoski National Park (Szczepko & Kowalczyk, 2001;Szczepko et al., 2002Szczepko et al., , 2009;;Szczepko & Wi niowski, 2009).Moreover, the species richness of spider wasps in the KNP (42 species) is distinctly higher than recorded in other studies carried out in the south of Poland: 28 species in the Ojców National Park and 21 in the Magurski National Park (Wi niowski, 2005;Wi niowski & Werstak, 2009, respectively).
The KNP, together with its buffer zone, is one of the most important faunal refugia in the Polish lowlands.The area offers a wide spectrum of environments for fauna associated with open areas of various types and origins as well as different forests (Andrzejewski, 2003).Although the number of samples assigned to cluster AB was low, the number of species and their abundance and frequency recorded in these samples (collected in open areas often located on dry soil) were higher than those recorded for samples in cluster CD (i.e., samples from forests, buildings and pioneer fallows on humid soil).The majority of pompilid species are inhabitants of relatively open locations and few inhabit woodlands or forests (Day, 1988 3) classified according to the nesting behaviour: (i) no nest constructing -leaving prey in situ, (ii) no nest constructing -cleptoparasites, (iii) digging nest in the ground, (iv) nesting in pre-existing cavities in ground, (v) nesting in preexisting cavities in or above the ground, (vi) nesting in preexisting cavities above the ground, (vii) constructing free standing nest above ground.The subcluster with the maximum observed IndVal for each species is given (and underlined if p < 0.05).Wi niowski, 2009).Moreover, the occurrence of species and the richness and composition of assemblages depend not only on the character of a site but also its position in the land mosaic (Bennet et al., 2006).The diversity of the surrounding landscape influenced the spatial distribution patterns of spider wasp species in the KNP.In cluster AB, which includes samples from open sites surrounded by more diverse landscapes, the number and abundance of species in samples were higher than in cluster CD, which includes samples from more homogenous areas (Fig. 3).More diverse surroundings could increase the amount of available resources or provide additional, new resources (Dunning et al., 1992) like nectar, hosts, nesting sites or over-wintering sites (Weibull et al., 2000;Steffan-Dewenter, 2002;Steffan-Dewenter et al., 2002;Tscharntke et al., 2002Tscharntke et al., , 2005)).Spider wasp species could use different habitats within their flight range (Taylor et al., 1993;Klein et al., 2004) to fulfil their specific requirements (Wi niowski, 2009).A good example is a sample from an open fragment of sand dune assigned to the same neuron (D1) as samples from forests (Fig. 2).At first sight this appears anomalous.Nevertheless, this open fragment of sand dune is small and surrounded on all sides by mixed forest.This explains why species typical of forests, flying over the dune, were caught by the traps located there.Our results are congruent with those of other studies in which more diverse habitats favour more diverse insect communities (Duelli, 1997;Weibull et al., 2000;Steffan-Dewenter, 2002;Tscharntke et al., 2002), by offering a variety of feeding resources, nesting sites and refugia for taxa with different living requirements (Evans & Yoshimoto, 1962).
The different associations of pompilid species with different habitats might also be explained by differences in their nesting biology.According to Evans & Yoshimoto (1962) and Kurczewski et al. (1988) habitat (soil type and cover of vegetation) is a good indicator of the distribution of pompilid taxa.Nesting resources are important for particular species of spider wasp as well as shape their assemblages (Evans & Yoshimoto, 1962;Kurczewski et al., 1988;Wi niowski, 2009).In our study, the presence of species nesting obligatorily in the ground (endogeic, ground-nesters), particularly those that dig nests, is generally limited to sites with exposed areas of friable soil, while those nesting above ground are able to nest in a variety of places, of which some are unavailable to ground-nesters (Evans & West-Eberhard, 1970;Potts et al., 2005).Because most species of spider wasp are associated with open areas and are thermophilous ground nesters (Day, 1988;Wi niowski, 2009) the number of species with significant maximum IndVals (9) was greatest in sub-cluster AB2, which includes samples from old fallows on dry, sandy soils located in a diverse landscape.In contrast, in the sub-cluster CD1 that contained all the samples from forests and almost all samples from fallows on hydrogenic soils, and had the lowest median of mosaicity measure, there were no species with significant IndVals, which indicates that the number of such species may serve as a bio-indicator of environmental quality for a given group of animals.Furthermore, the number of species with significant IndVals varies to some extent independently of species richness as both variables were highest in AB2, but the number of species with significant IndVals was not lowest in the sub-cluster with the lowest species richness (CD2).
Generally, the SOM identified species associated with similar or completely different habitats (i.e., with the same or an opposite pattern on the SOM, respectively).Some of them only occur in particular habitats (stenotopic), while others are common and occur in a wide range of different habitats (eurytopic) (Wi niowski, 2009).Eurytopic species with very different nesting requirements are mainly associated with fallows and meadows (AB1).Thermophilous and endogeic species, especially those that dig nests and their cleptoparasites are mainly associated with fallow abandoned a long time ago on dry, sandy soils (AB2).The species significantly associated with this kind of habitats are Arachnospila trivialis, A. wesmaeli and Episyron albonotatum, which occur at warm sites where they excavate nests in loose sand (Day, 1988;Wi niowski, 2009), and their obligate brood parasites: Evagetes crassicornis, E. dubius E. littoralis and E. pectinipes (Linnaeus, 1758).Moreover, sub-cluster AB2 includes all the samples with the following three species: (1) the thermophilous and endogeic Arachnospila ausa and A. minutula, which are strictly associated and nest in areas with sparse vegetation on sandy ground (Wi niowski, 2009), and (2) Evagetes dubius, which is a cleptoparasite of A. minutula (Day, 1988;Wi niowski, 2009).Species insignificantly associated with forests and fallows on hydrogenic soils (CD1), i.e.Priocnemis cordivalvata, Dipogon subintermedius (Magretti, 1886) and D. bifasciatus (Geoffroy, 1785), belong to hypergeic guilds.This observation is consistent with the results of other studies carried out in Poland, in which they are recorded mainly occurring in sunny edges of forests, where they nest in the burrows of wood borers in dead wood or empty hollow stems (Wi niowski, 2009).In the light of the latter information it is worth noting that of these three species only D. subintermedius was recorded on wooden buildings (CD2).The latter habitat was markedly preferred by three species (Agenioideus cinctellus, A. sericeus and Auplopus carbonarius) with different ecological amplitude, both eurytopic and thermophilous.They belong to three nesting guilds (v, vi and vii) and are species that nest obligatorily or facultatively above the ground.They may be examples of "cultural species" (Duelli & Obrist, 2003a), which are mainly associated with anthropogenic habitats ("cultural" habitats) (Bennet et al., 2006) but sometimes occur (also in the KNP) in non-anthropogenic habitats (Duelli & Obrist 2003b;Tscharntke et al., 2005).Since adult pompilids feed on nectar (Wi niowski, 2009) the presence of garden flowers and honeydew might be an important factor for them.Furthermore, these species hunt for spiders on the walls of buildings and may use cavities in the walls for nesting, which is suggested for the genus Agenioideus (Kurczewski & Spofford, 1986;Wi niowski, 2009) and A. carbonarius (Day, 1988).
Their spectrum of prey is wide and includes both active and web building spiders.In contrast, P. cordivalvata, D. subintermedius and D. bifasciatus only hunt for active spiders, which may be why they were absent or much rarer on wooden buildings though they nest in dead wood (Uetz et al., 1999;Wi niowski, 2009).
In summary, this study highlights not only the importance of open habitats like old fallows and meadows, already known to be good habitats for spider wasps (Day, 1988;Wi niowski, 2009), but also the importance of anthropogenic sites.Our results support the hypothesis that both pompilid species richness and abundance are positively associated with landscape mosaicity, and are consistent with the concept that habitat heterogeneity enhances faunal diversity as each type of habitat (including anthropogenic ones) widens the range of available resources (Benett et al., 2006).

Fig. 3 .
Fig. 3. Number of other habitats adjacent to a sampled habitat, number of pompilid wasps and species richness of a sample.Point indicates the median, whiskers the inter-quartile range.H is the statistics of the Kruskal-Wallis test (df = 3, NAB1 = 10, NAB2 = 11, NCD1 = 18, NCD2 = 13), which was used to test whether the sub-clusters differed significantly.The sub-clusters underlined by the same line were not significantly different in post-hoc comparisons.

TABLE 2 .
Number of spider wasp samples assigned to SOM subclusters in relation to the type of habitats and humidity of soil in open areas.Abbreviations of habitat names: AF -abandoned farm, FA -fallow, FT -fruit trees, FR -forest, ME -meadow, PGpsammophilous grassland, SD -sand dune, WB -wooden building.