Alternative flowering crops as potential food sources for beneficial arthropods

Various flowering crops (buckwheat, caraway, carrot, faba bean, flax, lupine, milk thistle, mustard, parsnip, phacelia) were evaluated for their suitability in providing nutritional resources for natural enemies and pollinators in small scale field trials. Flower visitors were documented by visual observation during the full blossom period of these plants. In addition, transparent water traps were installed to record the activity of arthropod groups in the flowering plots. Honey bees and bumble bees were the most frequent flower visitors of crop plants with more or less concealed nectaries and specialized flower structure (mustard, phacelia, lupine and milk thistle). Other wild bees could be observed frequently in flax. Ants and parasitic wasps were mainly active in plants with open flower type and exposed nectaries and stamina (parsnip, buckwheat). Also the extrafloral nectaries of faba bean were frequently visited by ants. Syr-phids, but also various bee species (e.g. Apis , Bombus , Halictus ) were better documented by direct visual observation than by captures in transparent water traps.


Introduction
Biodiversity is decreasing in agricultural landscapes (KREBS et al., 1999;BENTON et al., 2003;GUERRERO et al., 2012). This applies in general, but also regarding elements of functional biodiversity, providing certain ecosystem services in agro-ecosystems (ALTIERI, 1999;ISAACS et al., 2009;GEIGER et al., 2010;MEEHAN et al., 2011). Invertebrate arthropods like pollinator insects and natural pest antagonists are of fundamental importance for crop production. Effective measures for their conservation and promotion are a central requirement which needs to be considered in the common agricultural policy of the EU ("Greening" of the EU agriculture). Consequently, action plans and agri-environment schemes at national scale of European countries are in duty to offer an attractive framework for European farmers for modification of current land use towards a more biodiversity friendly agricultural management (AVIRON et al., 2009);EU-Regulation No 1305BMEL, 2015;BfN, 2016).
Many insects depend on the occurrence of plant-derived food sources (nectar, pollen, leaf exudates, seeds, indirectly honeydew) in one or all stages of their life cycles. There are numerous studies demonstrating the importance of such resources for pest antagonists to reach their full capacity in host-/prey location, egg laying and survival (detailed summary e.g. LUNDGREN, 2009). Important pollinators like wild bee species with various pollen preferences depend on provision of suitable flowering plants throughout the season (PFIFFNER and MÜLLER, 2014). On the other hand, the agricultural landscape gets increasingly poor from diverse and season-long supply of non-crop flowering plants (ISAACS et al., 2009). Strategies to counteract this trend aim for example on measures for the conservation of agricultural wild herbs (which are inherently valuable elements of biological diversity (GRASS et al., 2016), on the establishment of grasslands, field margins, fallows and hedgerows, the targeted sowing of flowering strips and suitable cover crops or intercropping (HOLLAND et al., 2016). Consideration of particular angiosperm crop plants, which have to flower in their cultivation for the agricultural produce and need to be insect-pollinated (TAKI et al., 2010), within the farm management may also increase the availability for nectar and pollen resources in the landscape (HOLZSCHUH et al., 2013;DIEKOTTER et al., 2014).
Furthermore, there is evidence that habitat and landscape heterogeneity have a major effect on the level of biodiversity in the agricultural landscape (BENTON et al., 2003;BIANCHI et al., 2006;HAENKE et al., 2009;CARVAL-HEIRO et al., 2011;HOLZSCHUH et al., 2012;JONSSON et al., 2015). Increasing plant diversity, e.g. by considering polycultures, may improve pest control (ALTIERI, 1999;LETOURNEAU et al., 2011;IVERSON et al., 2014). Such an increase of diversity in crop growing on arable land is also an objective within the Greening measures, suggested by the CAP ("crop diversification", "ecological focus areas", BMEL (2015); BfN (2016) and supporting agri-environment schemes (EU-Regulation No 1305. Suitable, (partially) insect-pollinated angiosperm crop plants can be found as oilseeds, pseudocereals, seed cultures, medicinal and aromatic plants, cover crops, biofuel and protein crops. Beside oilseed rape, such alternative crops are still a niche in most EU-countries and in many cases the management practice for these cultures may lack modern strategies for cultivation, plant protection or integration into crop rotation systems (AGROSCOPE, 2016). Certain cultures like medicinal and aromatic plants are especially interesting for organic growing. However, farmers need here long-term reliability, in particular with regards to the purchase of their products at acceptable prices (MIELKE and SCHÖBER-BUTIN, 2004: keyword: contract farming).
Crops that come into flourishing during their cultivation can be attractive for flower-visiting insects, thus being of importance for pollinators and natural enemies. Their availability in the agricultural landscape at certain periods could appeal to different species or groups of these beneficial insects. Biological determinants such as flower type, flower structure, reward system or the special need for nectar and pollen are critical to the usability of a particular plant by a certain beneficial insect (KUGLER, 1970;CAMPBELL et al., 2012;WÄCKERS and VAN RIJN, 2012). The objective of the work presented here was to examine a range of alternative angiosperm crop plants with respect to their attractiveness during flowering to various beneficial arthropod guilds in a small-scale field trial. The results were used to create a visitor profile for the respective cultivars that can serve as a baseline for further assessment of the contribution of these plants for the potential enhancement of agrobiodiversity.

Study plots
A field choice trial consisting of 2 m × 2 m wide monoculture plots of various plants was established on the experimental field of the Julius Kühn-Institut in South-Hessia (Table 1). The plots were distributed randomly in fourfold repetition over an area of 0.5 ha. Individual plots were separated by 0.7 m-wide strips covered with plastic foil. Annual crops were seeded in spring after mechanical soil tillage whereas two-year crops such as parsnip and carrot were planted as turnip after winter storage. A fertilisation or plant health treatment did not take place. Weeds were removed manually by hacking or digging out until the crops prevailed. The whole field was surrounded by a one meter wide grassy strip where frequent mowing prevented the flowering of naturally occurring weeds. In the year 2013, a second sowing was done for selected plants (yellow mustard, buckwheat) in late summer to represent their growing phase as a green manure plant.

Assessment of flower visitor activity
Visual observation of flower-visiting arthropods was carried out on windless, sunny days at certain times (between 09:00 to 10:00 and 18:00 to 19:00) during the full blos-Originalarbeit som period of the respective crop. Plots were observed for up to ten minutes to register the taxonomic category of each flower visitor of 10 visited inflorescences per plot (= positively evaluated inflorescences). This was done for each crop at four different dates during the flowering period (Table 1). Flower visitors were assigned to the following taxonomic units, considered as beneficial functional groups: (1) honey bees (Apis), (2) bumble bees (Bombus), (3) other wild bees, (4) mainly aphidophagous hoverflies (Syrphinae), (5) hoverflies belonging to the Eristalinae, (6) parasitic Hymenoptera (Ichneumonidae, Braconidae and others), (7) Formicidae, (8) Coccinellidae and (9) all other arthropods. In some cases, the determination to genus or species level was possible (e.g. Apis mellifera, Bombus sp., Halictus sp., Episyrphus balteatus, Eristalis sp., Coccinella septempunctata). These surveys were carried out again during the second phase of late sowing of buckwheat and yellow mustard in the same way.
In addition, the activity of arthropods within the flowering plots was registered using water traps. These consisted of round transparent plastic bowls (diameter 18 cm) with 2 l capacity, which were installed at the level of the florets of each crop. The bowls were filled with saline solution (3% plus a few drops of liquid soap). Water traps were set into the plots at the time of full blossom for 14 days (Table 1). At the same time, similar traps were put in the grass strip around the experimental field and served as control without flowers around. Every two to three days, captured arthropods were removed from the traps and were transferred into ethanol (70%) for further examination in the laboratory. The arthropods were assigned to the same categories as in the visual observation method.

Data evaluation and statistical analysis
Only inflorescences were taken into account when at least one visitor had been observed (= number of posi- Originalarbeit tively evaluated inflorescences) within the observation time. The number of flower visits of a particular taxon was summed up across the observation dates per plot and the relative abundance of this taxon was calculated with reference to the total number of positively evaluated inflorescences observed per plot during the observation period. Captures of the water traps were calculated by taking the sum of trapped specimen of a particular taxon for the duration of the total trapping period (14 days) in each plot.
For graphical analysis, the relative abundance of beneficials (i.e. Apis, Bombus, other wild bees, Syrphidae, parasitic Hymenoptera, Formicidae, Coccinellidae) in relation to the remaining arthropods in the data set of observed flower visitors were presented in bar charts. The relative distribution of these taxa in the late sowing buckwheat/mustard trial was presented in circular diagrams. The appearance of certain beneficial groups in relation to the different flowering crops (termed as "preference") was compared by analysing recorded numbers in appropriate general linear models (GLM) with adequate error distribution (R version 3.2.4). For binomial data (number of flower visitors/positive evaluated inflorescence) binomial or quasibinomial models were used. For count data (number of specimen/water trap), poisson models, quasipoisson models or negative binomial models were chosen. The GLM tested the effect of "crop plant species" on the number or proportion of a particular taxon in water traps and observation data. After determination of significance, multiple comparison between means were performed using the library "multcomp" in order to identify preferences of that taxon for particular crop plants. In some cases, the standard error could not be estimated if only zero values exist for a particular plant -insect combination, which leads to an exclusion of these data from the analysis (indicated by # in Tables  2 to 5).
To estimate the diversity of the visitor profile (only the beneficial taxa) of a particular crop plant, the Berger-Parker-Index D (May 1975) was calculated according to D = N (max) /N, which is the measure of the proportion of the most abundant taxon in the sample (WHEATER et al., 2011).

Observation of flower visitors in the flowering plots
The direct observation method for census of flower visitors in the different blooming crops is certainly biased by the fact that very small arthropods were not detected with a similar probability than arthropods exceeding about 5 mm in size. This should be kept in mind when evaluating the obtained results, especially with respect to the occurrence of e.g. small parasitic wasps, Heteroptera or Coleoptera whose numbers were very probably underestimated.
Considering the suggested beneficial guilds, these contribute to more than 50% of the registered arthropods on most of the crop plants ( Fig. 1 and 2). Plants with flowers of the open disk type (Table 1) were visited by nearly all considered beneficial taxa ( Fig. 1 and 2), but insects with   Originalarbeit phacelia) as well as in 2014 (faba bean, milk thistle). But ants exploited also the extrafloral nectaries of faba bean located at the stipulae of the leaves. Flax was mainly visited by bee species. The composition of the flower visitors on buckwheat and mustard at two different blooming periods (June 2013 versus August 2013) did not differ fundamentally. For mustard, the proportion of Apoidea accounted to 40 to 50% of the registered beneficial guilds (Fig. 4 and 5) during both periods. The proportion of Syrphinae was higher during the late summer period, which was probably a consequence of higher abundance of this taxon later in the year (increased 2 nd generation of bivoltine species). Buckwheat was always attractive to ants which were frequently observed on the nectar-rich flowers of this plant during early as well as late summer.

Preferences of flower visitors for particular flowering crops
There was some overlapping in the blooming period of the various crop plants during the study (Table 1) and flower visiting arthropods could select between different flower resources. Therefore, observation data were also analyzed for plant preferences of particular functional groups by testing the potential effect of plant species on the occurrence of the beneficial guilds in those plots. This was estimated by the relative number of observed visitors per evaluated inflorescence (observation census) as well as by their general activity, measured by the number of individuals caught in water traps. Only those plant species were considered, where four plots/season had been successfully established (buckwheat, parsnip, mustard, flax, phacelia, faba bean, milk thistle). According to this analysis, the recorded number of a particular beneficial group was significantly influenced by plant species in both years of the investigation (Tables 2 to 5).
The honey bee was observed on all plant species in the year 2013 with a preference for mustard, phacelia and buckwheat (Table 2). In 2014, the milk thistle was most attractive for A. mellifera (Table 3). Representatives of Bombus sp. preferred phacelia in 2013 and also milk thistle in 2014. Both Apis and Bombus were rarely caught in the water traps (Table 4 and 5).
Various wild bees occurred in all plots. Members of the genus Halictus were regularly observed as dominant group in flax in both years, collecting pollen there (Table 2 and 3, Fig. 6). Higher numbers of wild bees were caught in the water traps installed in buckwheat (Table 4 and 5); these belong to various genera (Lasioglossum, Andrena, Hylaeus and others).
No clear preference for certain flowering plants could be determined for the hoverflies. In the year 2013, many of them were observed in phacelia, but they occurred also in other plots, indicating that they are attracted to many different plants. In general, they were rarely caught in the water traps.
Parasitic Hymenoptera and ants (Lasius niger, Formica species and others) were highly attracted by inflorescences of parsnip. This plant developed plenty of postfloral nectar droplets on the developing fruits (Fig. 3). Ants were observed on the inflorescences in large numbers (Table 2 and 3) and their prevailing activity was also confirmed by the capture rates in the water traps installed at the height of the inflorescences. Parasitic Hymenoptera were mainly recorded by the water traps and here they were found more frequently in parsnip than in the other crop plots (Table 4 and 5). Parsnip was also the plant most attractive for the Coccinellidae (Table 3).

Visitor profile of the different crop plants
When summarizing the attractiveness of the various crop plants for the beneficial taxa (taking into account obser-  Originalarbeit vational census and water trap records), mustard, buckwheat, parsnip and flax sustained all relevant beneficial taxa in this study ( Table 6). The Berger-Parker-Index D was calculated to prove the evenness in the visitor profile. Thereafter, buckwheat, carrot and flax supported a higher diversity of beneficial taxa, which were attracted to the rich nectar/pollen provision by these plants. The other plants were mainly exploited by one particular beneficial group (D > = 0.5) which was dominant (parsnip, caraway, EFN of faba bean: ants; phacelia and lupine: Bombus; mustard and milk thistle: honey bees). Except of lupine and milk thistle, all tested crop plants were at least four weeks at blossom and flowered -depending on the seeding time -during early to late summer.

Originalarbeit
In addition to semi-natural habitats (flower strips, field margins, hedgerows etc.), also flowering crop plants may contribute to the availability of floral resources in the agricultural landscape. Oilseed rape is an early massflowering crop attractive for many beneficial groups (JAUKER and WOLTERS, 2008;HOLZSCHUH et al., 2013;DIEKOTTER et al., 2014). But other cultivars may provide additional or even better opportunities, e.g. because they flower later in the year or their cultivation does not need the application of insecticides, which may harm flower visitors.
Buckwheat showed the highest diversity of the considered flower visitors in our study (Berger-Parker Index, lowest value D = 0.27), thus supporting a wider range of beneficial taxa (pollinators and pest control agents). Nectaries in buckwheat flowers produce nectar of high viscosity with sugar concentration around 60% sucrose (w/w) (CAWOY et al., 2008;HERZ, unpublished data). Its   (LEE et al., 2004;LEE and HEIMPEL, 2008a, b;WINKLER et al., 2009;LAUBERTIE et al., 2012;PUMARINO et al., 2012;SIGSGAARD et al., 2013;VAN RIJN et al., 2013). Nectar is secreted from nectaries located at the basis of the stamina which can easily be exploited by short-tongue flower visitors, but are also attractive to honey bees and other bee species. It was reported that buckwheat mainly produced nectar in the morning (HEDTKE and PRITSCH, 1993) and we observed a frequent "nectar harvesting" by ants in our experiment in the morning too. How this nectar depletion may affect other taxa with a different temporal foraging pattern needs to be explored under field conditions. Buckwheat has many interesting agricultural uses: it is useful as green manure, fodder and biofuel plant as well as pseudocereal in human food. Furthermore, honey production and use for medicinal purposes is possible (CAWOY et al., 2008). Buckwheat can be easily integrated into crop rotation schemes and the cultivation usually does not require the application of any pesticides or herbicides (HEYLAND et al., 2006). Due to its multifunctional use, several seeding dates are possible (early for seed production, late as green manure or cover crop) and as consequence buckwheat flowering can occur from early June to mid of September in the landscape. Buckwheat is often included as component in annual flower strip plant mixtures (Fig. 7). Also flax was visited by various beneficial groups (D = 0.33). It was highly attractive to wild bees (genus Halictus) as pollen source. Syrphinae were observed too, but it was not clear if they exploited pollen or nectar. Flax opened the flowers mainly during morning and sunshine. The total flowering period was about 3 to 4 weeks. This plant is autophilious, nevertheless nectar and pollen are apparently produced in higher amounts which makes it attractive for many insects. Due to its particular value for beneficial taxa the cultivation of this crop on a broader scale would be desirable. We used a cultivar for oilseed production in our trial. The demand for oilseed is rising (e.g. production of lineseed oil also for human nutrition), especially for the organic sector (HILTBRUNNER et al., 2009). Flax is also recommended as a component of so called "greening" cover crops (ANONYMOUS, 2016), which may increase the cultivation of this plant on a larger scale.
Mustard was found to attract many beneficial groups, whereby honey bees were dominant in both years of the study. Later in the year (second sawing in 2013), also Syrphidae (both Eristalinae and Syrphinae) visited the flowers frequently. The nectar is secreted plentiful at the base of the stamina where insects with prolonged mouthparts can reach it easily. Pollen was collected by many bee species, e.g. the oligolectic Andrena agilissima, indicating its value also for rare endangered species. Also higher numbers of small parasitic Hymenoptera were caught in the water traps in mustard plots, being probably attracted by this flowering plant. They might be able to enter the flower and to crawl to the nectaries for feeding, but this still needs to be confirmed. The different varieties of mustard (Sinapis alba and relatives) can be widely used in agriculture (HEYLAND et al., 2006). They are often used as cover crops and green manure. But they can also deliver oilseed, spices or seeds for technical and medicinal products. Together with buckwheat and phacelia, mustard is included in annual "greening" seed mixtures (ANONYMOUS, 2016, Fig. 7) offering food for a broad range of beneficial taxa.
Due to their more specialized flower type, phacelia, milk thistle, lupine and faba bean provided reward especially for Apoidea, mainly bumble bees and honey bees, which were able to exploit the hidden nectar in the prolonged corolla of these plants. Beneficial taxa with less specialized mouthparts were not observed on the flowers, but ants and parasitic Hymenoptera visited the extrafloral nectaries of faba bean. These EFN already developed on the first leaves of the plants long before flowering. The potential role of these nectar providers for parasitic Hymenoptera requires further research in the field, because laboratory studies showed that parasitoids feed on them and their longevity and fecundity is increased (JAMONT et al., 2013;WALACH and HERZ, 2015). Faba bean and other Leguminosae are suitable cover crops for green manure and within greening regimes and may be even become more important as human food.

Originalarbeit
Many vegetables belong to the Apiaceae. They are mainly cultivated for their vegetative parts (carrot, fennel, parsnip, celery, dill), although some cultivars are grown for seeds (caraway, coriander, fennel, dill) too. For organic seed production, these plants need to be cultivated till flowering and often require successful pollination by insects (HOWLETT, 2012). Apiaceae are known to support many natural enemies due to their exposed nectaries and stamina (KUGLER, 1970; VAN RIJN et al., 2016) and wild forms are often recommended to be part of flower strips along arable crops (e.g. ISAACS et al., 2009;SIVINSKI et al., 2011). We included vegetable cultivars which grew more vigorously than the wild types. Parsnip developed rich nectar droplets of a high sucrose content (around 70% (w/w); HERZ unpublished data), especially on the developing seeds (postfloral nectaries, Fig. 3). These droplets were systematically harvested by ants, but also by other insects with unspecialized mouthparts (beetles, wasps, many different Dipteran groups including Syrphidae). In addition, some wild bee species visited carrot and parsnip indicating the values of these plants for them. Seed cultures of Apiaceae certainly need a particular management, but their cultivation may be suitable for some regions and of particular value for organic growers. They flower mainly in the second half of the summer, thus providing an interesting nectar supply during this time which may be important for migrating and overwintering wasps and flies (e.g. Syrphidae, parasitic Hymenoptera).
To conclude, most of the tested plants in this smallscale study supported many beneficial taxa, although their value for particular groups differed in dependence on biological determinants. But due to this fixed relationship, it can be assumed that the findings here can be also found on a larger scale in the agricultural landscape. Certainly the surroundings of agricultural fields and their provision with available nectar and pollen resources affect the attractiveness of the flowering crop plants as it was demonstrated for the role of flowering strips and other semi-natural habitats (HAENKE et al., 2009;HAENKE et al., 2014;JONSSON et al., 2015). A more diverse array of different crop systems can also have other benefits (less attraction to particular pests), even with economic consequences (LETOURNEAU et al., 2011;IVERSON et al., 2014). Thus, strategies which favour crop diversification by including also alternative crops with new purposes may have the effect of benefitting providers of ecosystem services too. To reach this goal, the elaboration of guidelines is needed how these alternative crops can be integrated into the farm management and into the matrix of the surrounding agricultural landscape.