EUROPEAN JOURNAL OF ENTOMOLOGY EUROPEAN JOURNAL OF ENTOMOLOGY Seasonal changes in bruchid (Coleoptera: Chrysomelidae: Bruchinae) assemblages along managed highway ecotones*

. Spring and summer composition and species richness of bruchid pre-dispersal seed predator assemblages associated with species of leguminous plants were monitored in a four-year non-experimental survey of 32 service areas along ﬁ ve highways in Hungary. The vegetation bands along highways (delimited by fences) were considered a special type of ecotone where herbaceous plants are exposed to regular mowing and therefore the composition of the vegetation there is very different from the adjacent vegetation. Altogether 57 herbaceous and woody species of leguminous plants were recorded at these sites, harbouring 20 autochthonous, 3 allochthonous, but established, and 4 recently introduced species of bruchid seed predators (Coleoptera: Chrysomelidae: Bruchinae). The species of leguminous plants recorded along highway verges during this project make up approximately one fourth of the Fabaceae in Hungary and of the bruchids ca. 80 % (!) of the species known to occur in Hungary. At half of the service areas, mowing decreased the species richness of leguminous plants compared to that recorded prior to mowing, but not that of their bruchid seed predators. However, the species composition of the bruchid assemblages before and after mowing changed substantially. Null-model analyses indicated a random organization of spring assemblages and a deterministic one of summer assemblages of bruchids; very likely a result of host-speci ﬁ city constraints. Calculations of host speci ﬁ city con-ﬁ rmed the narrow host range recorded for bruchids that emerged from the samples of plants, in spite of new host records, such as three and two Trifolium species for Bruchidius picipes and Bi . sp. prope varius sensu Anto n , respectively, Oxytropis pilosa for Bi. marginalis and Vicia cracca for Bruchus brachialis . Our results show that a surprisingly high number of species of bruchids occur in highway margins, however, the management of the vegetation there prevents a substantial portion of the native bruchid fauna establishing permanent populations. Illyés (1979–2012), a young botanist, who commenced the botanical charac-terisation of the survey sites, however, whose unduly death ended a promising scienti ﬁ c carrier. Deceased (1917–2014).


INTRODUCTION
Community ecologists have long recognized the importance of gradual or abrupt changes in vegetation (ecotones) (Peters et al., 2009) for the distribution, diversity and trophic interactions of animal species. Transitional zones of vegetation such as in edges, borders, boundaries and margins are places where exchange of energy, materials and species take place within landscape mosaics (Risser, 1993;Pullin, 2002). Conservation studies exploring the manifold negative consequences of habitat destruction by fragmentation (Wiens et al., 1985;Cadenasso et al., 2003;Fagan et al., 2003;Strayer et al., 2003) inspired important conceptual and methodological developments. Habitat es. In addition, we sought evidence that transport along highways provides opportunities for non-native species of bruchids to become established. In order to achieve these aims, surveys were carried out with special focus on the presence or absence of leguminous plants and their bruchid seed predators. We did this by (a) creating an inventory of the species of leguminous plants and bruchids occurring at highway service areas; (b) comparing species richness and composition of the assemblages of leguminous plants and their bruchid seed predators before and after mowing at the same sites in subsequent years; and (c) checking for any non-native species of seed predators belonging to the subfamily Bruchinae feeding on leguminous plants.

MATERIAL AND METHODS
32 highway service areas ( Fig. 1) were regularly surveyed for the occurrence of bruchid seed predators, including invasive species, along fi ve highways in Hungary between 2011 and 2014. They included all the major soil and vegetation regions of the country. M0 (5 service areas) is still an incomplete circle around Budapest with great effects of urban areas on highway verges; M1 (6 service areas) is the highway that extends to the westernmost part of the country where the precipitation is higher than average; in contrast M3 (8 service areas) extends to the east where the climate and species are tinted with Eurasian characters; M5 (6 service areas) runs through sandy regions with the lowest precipitation where species from the Balkans occur; the south-western part of the M7 (7 service areas) is affected by adjacent sub-alpine effects and is where species from the Mediterranean occur. In the context of this paper the highways are transects through the country (Fig. 1).
It is emphasized that this study is not experimental. The leguminous plant and bruchid species richness before and after the fi rst mowing were compared in the same areas subsequently; (a) before mowing by sweep-netting, (b) following mowing by collecting pods of leguminous plants, and (c) rearing bruchids from the pods.

Description of sampling sites
The service areas along the M0-M7 highways, with or without a fi lling station, are ca. 300 to 500 m long and 30 to 100 m wide with areas of grass planted with bushes (Berberis, Corylus, Cotoneaster, Crataegus, Ligustrum, Pyracantha, Rosa, Spirea, Symphoricarpus, Viburnum, etc.) and trees (Acer, Chamaecyparis, Fraxinus, Pinus, Populus, Pyrus, Quercus, Sorbus, Tilia, Ulmus, etc.), as well as having naturally growing herbaceous vegetation. landscape fragmentation (van der Ree et al., 2011). Road ecology, a rapidly developing and new discipline (Forman et al., 2003), is concerned with the many kinds of effects roads (linear infrastructures) have on wildlife. Road ecology, however, overwhelmingly focuses on dangers to vertebrate populations and distribution, while data on invertebrates are scarce (Muñoz et al., 2015; but see Knapp et al., 2013).
The vegetation immediately adjacent to controlledaccess highways (thereinafter in the text: highways) is usually different from the surrounding landscape as it is mown at intervals. Highways are also often fenced to prevent larger animals accessing these areas. The area on either side of a highway affected by noise, pollution from vehicles, seed dispersal, etc., however, extends beyond the fence. The areas bordering on to highways are intentionally planted with fast growing species or those that prevent the erosion of slopes. Woody and herbaceous species of plants, not indigenous to Central Europe, make up 40% of the vegetation growing along the sides of roads (Šerá, 2008) and form unusual associations. Therefore, these vegetation bands might provide habitats for introduced species of insects. Road ecotones, fully or in part, are also subject to major disturbances, i.e. regular mowing, which generate variation spatially and temporally (Wiens, 2000), decrease habitat heterogeneity and negatively affect species richness (Kruess & Tscharntke, 2000).
The knowledge of non-native species of herbivorous insects, inadvertently carried and dispersed by vehicles is relatively poor. Along highways there are service areas at regular intervals where vehicles are stationary for long enough for any hitchhiker species to leave them and fi nd a host plant. The fi rst record of the spotted wing drosophila [Drosophila suzukii (Matsumura)] was fi rst recorded in Hungary in 2012 at a service area  and its repeated appearance in subsequent years  is a good example of the phenomenon. Kozár et al. (2013) record 22 species of scale insects new to the Hungarian fauna over a period of 10 years and postulate that a possible source is international transport. Matošević & Živković (2013) list 98 non-native species of herbivorous insects in Croatia, a country bordering on Hungary in the south-west, supposedly introduced by trade, however, members of the coleopterous subfamily Bruchinae are not among them. While Roques & Beenen (2010) report 14 species of seed beetles of 9 genera new to Europe, Yus-Ramos et al. (2014) substantially expand this inventory by describing 42 species in 29 genera, among which there are 27 species belonging to 13 genera of Bruchini. As for Hungary, they list 11 established allochthonous species of Bruchinae. Seed predation might have an important effect on the population dynamics of the plants (Crawley, 2000;.
The aim of this study was to record non-experimental evidence that the management of the vegetation growing alongside highways change the pattern of leguminous vegetation, and as a consequence, cause compositional changes in the associated herbivorous insect assemblag- All parts of the highways are fenced off from the surrounding area by a wire fence with a mesh of 15 × 20 cm. The central part of each service area is heavily affected by human activity, however, there is an additional peripheral larger "buffer zone" close to the fence in each case, where there are usually drainage ditches, or occasionally even a small pond, which is usually 1-3 m below the level of the rest of the area. The buffer zone is isolated from the rest of the area by steel safety barriers and visited only by maintenance and service personnel. Permission was obtained to survey and collect on all parts of the service areas.

Survey of vegetation
In May of 2011 each service area along the M0-M7 highways were surveyed to determine the general status of the vegetation (E. Illyés, unpubl.). The plant associations, species composition, and cover of species in 2 × 2 m quadrates were recorded; the list of the species of plants present along two 20 m long transects, as well as the abundance of each species were recorded; any invasive/ non-native species of plant was noted. Altogether, there were 187 species of vascular plants recorded at the service areas. Cerastium spp., Plantago lanceolata, Poa angustifolia, Silene latifolia ssp. alba and Taraxacum offi cinale were the dominant species. In addition, in the quadrates and transects, there were 18.1 ± 5.5 (SD, N = 36) species of other plants and 27.6 ± 6.7 (SD, N = 34) species of leguminous plants. Characteristic grassland genera were: Bromus, Lolium, Festuca, Poa, Holcus, Potentilla, Medicago, Alopecurus, Plantago and Elymus. The leguminous species were identifi ed by the fi rst author. Several species of plants (e.g., Trifolium incarnatum on M7), not occurring naturally in the area, were extremely abundant at some, especially newly constructed sections of highways due to the sowing of imported seeds on the sloping roadsides in order to stabilize the soil. Likewise, rows of bushes of non-native species (e.g., Eleagnus umbellata, Pyracantha sp.) planted on the slopes provided an important source of nectar for insects.

Management of service areas
The herbaceous vegetation along the highways and especially in the central part of each service area was usually mown to a height of ca. 5 cm twice during May-July. In addition, the buffer zones were also mowed at least once, together with the roadside slopes along the service roads leading in or out of the service areas. Mowing resulted in the removal of fl owers or seed heads of herbaceous plants, especially the tallest species. Species of low stature or prostrate, such as Trifolium repens, T. hybridum, T. fragiferum, T. campestre/aureum, Medicago minima, Vicia lathyroides, etc. thrived under these conditions and individually occupied 0.5-1 m 2 , or even larger patches, mostly along drainage ditches and around tree trunks. Occasionally, mowing resulted in dwarf plants. Plant cuttings were removed from the area. There were always sites within the service areas where the mowing was incomplete. For instance, plants growing close to the fence remained intact and even were able to produce and mature seed, especially those using the fence as a support (e.g., Vicia villosa).

Collection and processing of the samples of insects and plants
The Palearctic bruchids (Coleoptera: Chrysomelidae: Bruchinae) that feed on members of the Fabaceae belong to the pre-dispersal seed predator guild and to the green-pod sub-guild (Szentesi & Jermy, 1995), which include four major taxa: Bruchinae and Curculionidae (Coleoptera), species of microlepidoptera (Lepidoptera) and phytophagous chalcidoid wasps (Hymenoptera). The species in the green-pod sub-guild are narrowly host-specifi c on leguminous plants. Adult bruchids do not consume host tissue, but feed on pollen and nectar of fl owering plants. They oviposit on immature pods containing only developing seeds. Almost all species of bruchids are univoltine and adults are not present in summer, because they die after several egg-laying sessions. The endophagous larval stadia complete their development in seeds. More than one guild member, of either species of bruchids, curculionids or Lepidoptera, might be present in pods of leguminous hosts. The curculionid and lepidopteran seed predators ectophagously consume seeds within pods.
Between 2011-2014 the presence/absence of leguminous species was recorded, bruchid adults were collected in May each year, and mature pods were sampled in July. In order to compare the composition and species richness of bruchid assemblages before mowing (in spring) and after mowing (in summer) two methods of collecting were used: (1) overwintered adults of bruchids were collected at each of the service areas surveyed along all the highways in the fi rst or second, and again in the last week of May in all years. Adults were collected fi rst by sweep-netting and beating nectar producing fl owers of non-hosts (e.g. Lepidium draba, Eleagnus umbellata) and later from fl owering and pod-producing hosts. A sweep-net of 40 cm diameter was used, and one sweepnetting session consisted of 50 strokes of ca. 1 m wide and 25 m long, 10-20 cm above ground level. There were up to a maximum of ten sweep-netting sessions at each service area depending on the size of the parking area and the availability of host plants. Surveys and collections were carried out with the same intensity at the different sites by spending approximately the same time (ca. 30 min) at each service area. (2) Mature pods of herbaceous leguminous hosts were collected in July, and those of woody species in September in 2011-2013. One hundred heads or pods were collected from species of Trifolium and other leguminous species, respectively. All pods were collected if there were less than 100 and only 50 heads of T. pratense due to their large size.
Flower heads and individual pods were picked and placed in paper bags, then transferred to jars the openings of which were closed with linen cloth. The samples were kept outdoors. As adults usually emerge from infested pods in the same year, samples were checked from time to time until November of the same year in order to collect live adults. Samples were quantitatively evaluated after a fi nal inspection the following year. Adults emerging from the samples were placed in 96% EtOH and the species identifi ed by the second author.

Statistical methods
Kruskal-Wallis ANOVAs were used to compare the numbers of species of both leguminous plants and bruchids, respectively, by years and highways. Species numbers on fi gures are medians and percentiles. For comparisons two-tailed tests were used. Leguminous plant species richness recorded in spring and summer along the same highways were compared using the Wilcoxon matched pairs test at p < 0.05 for two dependent samples. This pairwise comparison was necessary as there were at least one mowing session at the same sites between the two surveys. The Jaccard index (Magurran, 2004) was used to compare similarities in the composition of species of leguminous plants and bruchid assemblages, respectively, recorded before and after mowing in each year. Correlations (Pearson's r) between the number of species of leguminous plants and bruchids, respectively, before and after mowing were also calculated. For all computations the Statistica 6.0 program was used (StatSoft, 2003).
A null-model analysis was used to demonstrate signifi cant species co-occurrences. Species co-occurrence refers to possible organizing forces, such as interspecifi c competition, or host plant specifi city that are associated with communities with a particular composition (Gotelli & Graves, 1996). Two presence/absence matrices needed for the null-model analysis were prepared: one for 16 species of bruchids captured at 32 service areas by sweepnetting, and one for 24 species of bruchids reared from samples of leguminous seed collected at 29 service areas. Species of the bruchid genus, Spermophagus, although regularly captured by sweep-netting, were not included in the analysis as they do not feed on leguminous plants. The observed and simulated matrices were compared following 10,000 randomization cycles, the minimum number of iterations suggested by Lehsten & Harmand (2006), at the 0.05 probability level using the EcoSim700 program and sequential swap algorithm with fi xed row and column sums (Entsminger, 2014). Signifi cant co-occurrences of all possible species pairs were calculated using the C-score (Stone & Roberts, 1990), which applied average "checkerboard (CS) units" according to the formula: where S is the number of species, n i and n j the number of occurrences (row totals) of the i th and j th species, and N ij is the number of co-occurrences of the two species. A Bayesian approach was used to detect false signifi cant non-random associations of species using the Pairs program (Ulrich, 2008). If the samples contained mutually exclusive species-pairs one would expect the observed C-score to be signifi cantly higher than the simulated one.
We determined the degree of host specifi city of the bruchid assemblages reared from the samples of leguminous seed. The measure used was the multiple host dissimilarity index (Diserud & Ødegaard 2007) based on May (1990), which indicates level of "effective specialization": where ES is the measure of host specialization, S P is the number of species of bruchids associated with all the host plants, S̅ P is the average number of species of bruchids associated with one host, and P is the number of species of plants. The denominator is in fact the number of all recorded connections between bruchids and host plants in the assemblages. This measure provides values between 0 (all bruchid species are "generalists") and 1 (all are monophagous). We also calculated the specialization of bruchids based on host records in Hungary  and for the Bruchus, Bruchidius and Megabruchidius genera collected, respectively, lumping together results for all years and service areas.

RESULTS
The species of leguminous plants recorded at 32 service areas along highways in Hungary are listed in Table 1. A total of 57 native and naturalized herbaceous and woody species of leguminous plants were recorded, which is approximately 1/4th of the species of the Fabaceae occurring in Hungary. Altogether 24 species of bruchids belonging to four genera were reared from samples collected at the service areas (Table 1). (See the authors' names of plant and insect species in the tables.)

Comparing the numbers of species of leguminous plants
(1) Among service areas Sixteen species of plants were recorded at 50% of the localities (Table 1). These were either species frequently occurring at disturbed sites (Medicago sativa, M. lupulina, Trifolium pratense, T. repens), fugitive annuals (Vicia grandifl ora, V. sativa ssp. nigra, V. villosa, Trifolium aureum/campestre, T. arvense, etc.), or species that tend to grow horizontally and as consequence are little affected by mowing, e.g. Astragalus austriacus, Medicago minima, T. hybridum and Vicia lathyroides. Seventeen species occurred only at 1-2 service areas. Forest edge species, such as Astragalus glycyphyllos, Lathyrus sylvestris, Trifolium alpestre, T. medium, Vicia pisiformis, V. sepium and V. tenuifolia occurred in very low numbers (one individual or patch). Several leguminous plant species (Trifolium repens, Vicia sativa ssp. nigra, V. grandifl ora, V. pannonica ssp. pannonica, V. villosa), attacked by seed eating insects, formed continuous and large (> 100 m 2 ) patches and frequently grew intermingled. Sometimes a single patch (Oxytropis pilosa), or a single individual (Trigonella caerulea) was present, in spite of the regular and heavy disturbance.   (Table 2). Sixteen species of bruchid that feed on leguminous plants were captured by sweep-netting. Based on the samples of (1) Among service areas Bruchids did not inhabit 53% (9 out of the 17) of the most common species of leguminous plants occurring along the highways sampled (Table 1). The remaining eight species either harboured "broadly" oligophagous bruchids, such as Bu. affi nis, or ubiquitous specialists, like Bu. luteicornis, Bu. brachialis, Bi. pusillus and Bi. varius. The most frequent species of bruchids collected by sweep-netting were Bu. brachialis and Bu. luteicornis (present at 62.5% and 53.1% of the service areas, respectively). Whereas the former was usually recorded during the fl owering of its host plant, the latter was present even before fl owers appeared. Several species (Bu. brachialis, Bu. affi nis, Bi. varius and Bi. pusillus) were frequent at 8-10 service areas based on rearing them from seed samples.
(2) Among highways Collection by sweep-netting and rearing from samples of plants. There were no signifi cant differences in the median number of species of bruchids collected either  by sweep-netting or rearing, respectively, recorded for highways in all years (Fig. 3), or for highways alone (collected: H 4;123 = 3.48; p = 0.4817; reared: H 2;96 = 1.76; p = 0.4154). However, considering all highways, signifi cantly more bruchid adults were collected in 2014 than in 2012 or 2013 (H 3;123 = 14.25; p = 0.0026). Multiple comparisons of the highways by years revealed signifi cant heterogeneity (H 4;96 = 11.59; p = 0.0207), however, no pairwise difference was recorded in the number of bruchid adults that emerged from samples. All these differences were computed using Kruskal-Wallis ANOVA and multiple comparisons.  There were no signifi cant co-occurrences of bruchids in the spring assemblage collected by sweep-netting. The observed C-score index was 8.48, whereas the simulated one was 8.42 (p = 0.3965). However, for the assemblage reared from leguminous host plants the observed and simulated C-scores were close to the expected probability level (p = 0.05): 7.025 and 6.868, respectively (p = 0.0760), i.e. there were indications of bruchid co-occurrences. Table 3 provides information about the host affi liation and relative abundance of species of bruchids reared from samples of leguminous plants collected along the highways. The multiple host dissimilarity index (ES) for all the species of bruchids reared was 0.60, indicating that the majority of the species were specialists. The species of Bruchus had altogether 13 insect-host plant connections and an ES of 0.62. For those of Bruchidius it was 24 insect-plant interactions and an ES of 0.54, whereas it was 2 and an ES = 1 for those of Megabruchidius. On the basis of published host records , among the 28 species of bruchids recorded there was only one Bruchus and 10 Bruchidius that were monophagous, and the index was 0.42, indicating a high percentage of "generalists". New host records include Trifolium pratense, T. repens and T. incarnatum for Bi. picipes, T. repens and T. hybridum for Bi. species prope varius sensu Anton, Oxytropis pilosa for Bi. marginalis and Vicia cracca for Bu. brachialis (Table  1), i.e., 4, 3, 2 and 3, more hosts, respectively, than recorded by . This indicates changes in the host specifi city of Bi. picipes, Bi. species prope varius  sensu Anton and Bi. marginalis from mono-to oligophagous, whereas Bu. brachialis remains oligophagous.

Introduced species of bruchids
During the four-year survey, seven non-native species of bruchids were recorded ( Table 2). Three of these (Acanthoscelides pallidipennis, Bruchus pisorum and Bu. rufimanus) became established in the 20 th century and are currently naturalized, four others (Megabruchidius tonkineus, M. dorsalis, Bi. siliquastri and Bi. terrenus) must have been introduced in the early 2000s (see Discussion).

DISCUSSION
Highways are a special type of ecotone. Unlike the borders of natural vegetation that may grade into the adjacent landscape mosaic at different scales (Gosz, 1993), highway ecotones are larger, more homogeneous, and most of all, artifi cial. Ecotones mediate species and effects. There are no natural successional changes due to the almost continuous management (mowing). As a consequence, the species of plants present must be able to tolerate regular disturbance and still be able to produce propagules, or equipped with effective dispersal and colonization mechanisms.
In this study we found that the host plants of bruchids are either fugitive annuals having seed banks, such as Vicia grandifl ora, V. sativa ssp. nigra  or low-stature perennials (Tetragonolobus maritimus, Trifolium fragiferum, T. hybridum and T. repens) ( Table 1). Both groups form patches at service areas and the latter are able to re-sprout in spite of regular mowing. There were no signifi cant differences in the median number of species of leguminous plants in the different years or along different highways either in spring or summer, respectively, with the exception of a lower number of species in the springs of 2012 and 2014. It is assumed that the general lack of a difference is the result of the presence of a dozen ubiquitous species of leguminous plants such as Lathyrus tuberosus, Medicago sativa, M. lupulina, M. minima, Securigera varia, Trifolium campestre/aureum, T. pratense, T. repens, Vicia sativa ssp. nigra, V. grandifl ora and V. villosa along almost all highways and in service areas. These species either grow from seeds, re-sprout after the fi rst mowing (second half of May) or survive in small patches. This is also supported by the high compositional similarity between years as measured by the Jaccard index. In contrast, pairwise comparisons indicate a ca. 50% decrease in the number of species of leguminous plants following mowing (Fig. 2) indicating that, in general, mowing could decrease the species richness of leguminous plants.

Bruchid assemblages
We found supporting evidence for the hypothesis that a very high percentage of the bruchid fauna would be recorded in the service areas along the highways ( Table 2) in spite of the intensive mowing. However, the compositional similarities (measured by the Jaccard index) of the assemblages collected by sweep-netting versus those reared from collected samples of pods were low in certain years and the numbers of species were not correlated (fi rst part of hypothesis b). The median number of species of bruchids did not differ among highways or years (Fig. 3), but signifi cantly more species was collected in the spring of 2014 than during the same period in other years. The lack of differences in the numbers of species could be attributed to the consistent presence of 4-6 species of leguminous plants (listed above) that harboured bruchid species.
Our results confi rmed that for the spring and summer assemblages of bruchids different assembly rules apply (second part of hypothesis b). (1) Adults of actively dispersing species of bruchids congregate on nectar-producing plants (e.g., Eleagnus umbellata, Lepidium draba, etc.) in spring when their host plants are not present or pods are not yet available for oviposition. The very low compositional similarities to the summer assemblages and the lack of signifi cance of co-occurrences revealed by the nullmodel analysis provide further support for the spring assemblages being formed at random by dispersal from the surrounding areas to the highway ecotones.
(2) From the samples of summer pods, however, only those members of the spring assemblages were reared, whose hosts were present at egg-laying and produced mature pods in spite of mowing. The compositional difference between the spring and summer bruchid assemblages is greater if fewer host plants remain following mowing. Thus, it is the availability of leguminous plants at the service areas that determines the composition of bruchid assemblages in summer. This implies that the narrow host-specifi city of the insect seed predators is the ultimate constraint in the structuring the summer assemblages, regulates co-occurrences, determines guild membership Szentesi et al., 2006), and also makes the composition of the summer assemblage more predictable (see more on this below). Indeed, the null-model analysis indicated that this could be the case. The nearly signifi cant C-score probably is both the consequence of the presence of several taxonguilds [complying with close phylogenetic relatedness of members (Schoener, 1986)] feeding on the eight species of leguminous hosts (Table 1) and the occurrence of curculionid and lepidopterous seed predators in the same samples. However, competition, which is generally thought to be a structuring force, might be involved in the formation of the seed predator guilds. There are similar results recorded for phytophagous insects feeding on cruciferous plants, where plant chemistry characteristic of Brassicaceae (an host-associated constraint) is hypothesized as an organizing force by Frenzel & Brandl (1998), irrespective of the feeding guilds.
The lack of signifi cant differences between the species richness of bruchids in spite of the regular removal of above-ground vegetation, and the low similarity of the composition of the species of bruchids in spring and summer assemblages, are possibly both due to the way this community is organized. The long-standing and unsettled dispute over the nature of forces organizing assemblages, guilds and communities still divides ecologists (Weiher & Keddy, 1999). Assembly rules depend on the communities; some are a result of competition (Brown, 1987) others assemble by chance (Simberloff et al., 1999). Moreover, taxon-guilds, with high host specifi city, produce mandatory guilds. Such guilds are recorded for tephritid fl ies (Zwölfer, 1988), hispine beetles (Strong, 1982) and some species of bruchids (Szentesi et al., 2006).

Host range and specifi city
As for the host specifi city of members of the subfamily Bruchinae, we refer to  who provide detailed information on the subject. Their study was based on decades of collecting several thousand samples of leguminous seed and recording the beetles that emerged. Based on this information, the members of the subfamily Bruchinae subfamily recorded in our study are either mono-or oligophagous, and although new host records were added (Tables 1 and 3, and see Results), they did not change the overall status of the host specifi city of the bruchids. A monophagous species feeds only on a single plant species and an oligophagous species feeds on several (but not in all) members of a plant family, in the present case Fabaceae. Although host specifi city is genetically determined, these terms are conventions, because there is no general consensus about the range of host plants that divides "specialist" and "generalist" feeders. Moreover, the spectrum of hosts can vary geographically. For instance, several species of bruchids that are monophagous in Hungary appear to be oligophagous in Western Europe where other hosts, which do not occur in Central Europe, are available (Delobel & Delobel, 2006). Jermy & Szentesi (2003) record a maximum of nine host plants for both Bruchus atomarius and Bruchidius villosus, so even these are narrowly oligophagous based on the above criteria. In contrast, all the species of bruchids recorded in this study fed on one to four species of plants (Table 3). Therefore, and in general, the host spectrum of bruchids is quite narrow and, as a constraint, it underlines the deterministic organization of their assemblages. Phylogenetic analyses strengthened the taxonomical relations and host specifi city of the two bruchid genera (Kergoat et al., 2004. It is generally assumed that it is the bewildering richness of chemical plant constituents (Southon et al., 1994) that plays an important role in host specifi city (Schoonhoven et al., 1998).
The values of the multiple host dissimilarity index (ES) corroborated the above and shed further light on host specifi city. First, it measures the host specialization of the whole bruchid assemblage reared from the samples of leguminous plants (0.60). Its value substantially differed from the one calculated on the basis of the data in the literature (0.42). The reason for the difference was the faunistic survey of Jermy &  that was longer and more thorough, whereas our study was restricted to 4 years and to specifi c habitats. Second, concerning the host specializations of the bruchid genera the ES-values show an opposite trend to that of the published data: species of Bruchus seem to be more specialized than those of Bruchidius. We think that the low incidence of the habitat-specialist and "widely" oligophagous Bu. atomarius in our samples, and the presence of two Trifolium species (T. repens and T. incarnatum) each fed on by 3 species of Bruchidius, but not recorded by , can account for difference in the ES-values.

Non-native species
The detection of non-native species of insects (hypothesis c) in the bruchid seed predator guild highlights the possible role of transport in their distribution (Table  2). With the exception of those with a long colonization history in the Palearctic region, including Hungary, such as: Amorpha fruticosa-A. pallidipennis (Szentesi, 1999), Pisum sativum-Bu. pisorum, and Vicia pannonica-Bu. rufimanus (Bridwell & Bottimer, 1933), all non-native species can be considered recent introductions. The establishment of an allochthonous species does not require the presence of the host plant because (1) there might be alternative or secondary hosts present in the vegetation along highways suffi cient to sustain viable populations of polyphagous herbivorous species, and (2) oligophagous species may colonize hosts with a scattered occurrence along highways. We randomly found such species relations (Albizia julibrissin-Bi. terrenus, Cercis siliquastrum-Bi. siliquastri, or Gleditsia triacanthos-Megabruchidius tonkineus/M. dorsalis) despite not systematically searching for them. Some of these woody species may have spread unaided in the landscape, or are relatively preferred ornamental species in urban surroundings (e.g. Cercis siliquastrum and Albizia julibrissin). Introduced species of bruchids seemingly did not colonize native species of herbaceous leguminous plants because no new bruchid colonists were found during the four-year long survey. There are 17 species of bruchids in Western and Mediterranean countries that could live in Hungary where their 24 species of leguminous host plants occur (Delobel & Delobel, 2006). So far none of them have been found. Although we did not investigate the role of transport in promoting bruchid introductions, the occurrence of new species cannot be ruled out in the future due to the effects of climate change and the ever-increase in road transport. Ecotones are locations at which the effects of climate change can be followed (Risser, 1993). Indeed, the presence of non native species of scale insects in Hungary is associated with such factors (Kozár & Nagy, 1986;Kozár, 2009). The invasive species of bruchids that have been recorded are Asian (Bi. siliquastri, M. tonkineus/M. dorsalis) and North-American (A. pallidipennis) and so far neither their origin (Wendt, 1981;Jermy et al., 2002; or climatic factors have been implicated in their presence in Hungary. Contrary to the above, however, members of local bruchid assemblages, e.g., Bi. varius, which normally feed on common leguminous species like Trifolium repens and T. pratense, readily colonized Trifolium incarnatum that is a component of the seed-mix that was widely used to replant the slopes along the sides of the south-western part of the M7 highway. T. incarnatum is a native of the Hungarian fl ora, which, however, provided new opportunities for autochthonous bruchid colonization due to its recent increase in areas where it was not previously present.