The study was conducted in two neighbouring agricultural sites, located in the Municipality of Havelange (Fig. 1A): the Froidefontaine and Emeville farmsteads (Fig. 1B). They are located at 2 km away from each other, in the geological region of Condroz, in Wallonia (Belgium), as defined by Dufrene and Legendre (1991).

Figure 1.  

A. Location of Havelange Municipality in Belgium; B. The location of the two farmsteads in Havelange.

The Froidefontaine farmstead

The Froidefontaine farm (50°23'6''N, 5°8'34.799''E) covers an area of 55 hectares, with a mosaic of varied habitats. One of the management objectives is diversifying the land use by conserving natural areas (mesophilic and wet meadows, limestone slopes, ponds...) and hosting different farming projects in a collaborative way on farming areas. Thus, the farm aims at creating a rich and welcoming landscape for diversity, including biodiversity.

Within the farm, we defined four adjacent habitats (Fig. 2A; Table 1) covering about 10 ha each: a parcel of crops (GC) including a third of the surface with vegetable crops (GC1), a meadow zone (PAT), a young apple orchard (VER) and a wetland (ZH). The parcels were surrounded by hedges principally composed of hornbeam, elderberry, dogwood, hawthorn, maple and European charcoal.

Figure 2.  

A. Froidefontaine farmstead map. GC, PAT, VER and ZH correspond to the sampled parcels, whose details are given in Table 1. Each numbered red dot corresponds to the position of a trio of coloured (white, yellow, blue) pantraps; B. Emeville farmstead map. PAV, FRE, EPI and DIK correspond to the sampled parcels, whose details are given in Table 1. Each numbered red dot corresponds to the position of a trio of coloured (white, yellow, blue) pantraps. BF1, BF2 and BF3 correspond to the sampled flower strips. Each blue or green numbered dot corresponds to the position of a trio of coloured (white, yellow, blue) pantraps for the "feeder" flower patch or the "pollinator" flower patch, respectively.

The Emeville farmstead

The Emeville farm (50°23'2.4''N 5°10'1.199''E) covers an area of just over 40 ha. In 2016, the farm managers and a committee of various partners converted conventionally-managed fields to agroecological farming methods. To allow a complexification of the ecological network and creating an agricultural landscape enriched with biodiversity, the first actions were: laying hedges and grass strips; planting rustic apple trees; breeding Angus cattle (Bos taurus taurus L.) in an orchard; alternating temporary and permanent meadows; arranging of flowered grass strips; using no pesticides and amendments.

The sampling zone covered 15 ha and was divided into seven parcels (Fig. 2B; Table 1), which included six parcels of crops separated by flower strips and one parcel of orchard. Each flower strip (BF1, BF2 and BF3; Fig. 2B) was composed of three plant mix sequences, including a combination of one "feeder" flower patch (BFV) and one "pollinator" flower patch (BFB), separated by the cover crop patch. The cover crop patch was composed of a grass mix of Festuca arundinacea Schreb 1771 and Dactylis glomerata L. 1753 sown at 20 kg/ha. The feeder flower patch was composed of a mix of 40% of clover (Trifolium pratense L. 1753) and 60% of alfalfa (Medicago sativa L. 1753) sown at 25 kg/ha. In order to match Agri-Environmental and Climate Measures (AECM) specifications, the pollinator flower patch was sown at 30 kg/ha and was composed of a mix including 85% of grasses (Poa pratensis L., 1753 Festuca rubra L. 1753 and Agrostis capillaris L. 1753), 2% of leguminous species (Lotus corniculatus L. 1753, Medicago lupulina L. 1753 and T. pratense), 3% of annual flower (Papaver rhoeas L. 1753, Glebionis segetum Fourr. 1869 and Cyanus segetum Hill 1762) and 10% of other flower species (Achillea millefolium L. 1753, Centaurea jacea L. 1753, Daucus carota L. 1753, Leucanthemum vulgare Lam. 1779, Malva moschata L. 1753, Silene latifolia Poir. 1789, Melilotus sp. Mill. 1754, Knautia arvensis Coult. 1828 and Echium vulgare L. 1753).

To assess wild bee and hoverfly diversity, we conducted standardised sampling methods by combining coloured pantraps and netting transects (Földesi and Kovács-Hostyánszki 2014, Westphal et al. 2008, Grundel et al. 2011). Sampling was performed in 2018 and 2019, from April to July. At each collection site (Fig. 2A & B), we positioned a triplet of pantraps (FLORA model with a diameter of 26.5 cm, RINGOT, France) coloured with UV reflecting sprays in white, blue and yellow (ROCOL top tracer model, UK). The pantraps were set-up in line and spaced 3 to 5 metres apart, in order to avoid the attraction coverage bias and to reach the same probabilities of insect capture between the pantraps (Amy et al. 2018, Droege et al. 2010). The pantrap triplets were separated by a minimum of 20 metres, in order to cover each parcel as homogeneously as possible (Carboni and Lebuhn 2003, Eeraerts et al. 2017). Each pantrap was filled with odourless and colourless soapy water every two weeks during one day (from 9:00 AM to 5:00 PM). Every two weeks, we also conducted variable transects with an insect net for one hour in the morning and one hour in the afternoon, for each habitat in Froidefontaine and each flower strip in Emeville (Table 1; Fig. 2). We selected the sampling dates according to the following climatic conditions: temperature higher than 7°C, calm wind (< 12 km/h) and sunny and cloudless day (Westphal et al. 2008). We stocked insects in 70% ethanol for their conservation.

We followed the protocol of Mouret et al. (2007) to prepare, pin and label our collected specimens.

In 2019, we decided to let the yellow pantraps to be continuously activated from mid-May to the end of July with sampling every 10 days to maximise the capture of syrphids and considering that hoverflies have a predilection for the yellow colour (An et al. 2018, Lunau et al. 2018, Wäckers and van Rijn 2012).

Bee specimens were identified at the species level following identification keys of Pauly (2019) for Halictidae, Patiny and Terzo (2010) for Andrenidae and Falk (2015) for the other bee families (Apidae, Colletidae, Megachilidae and Melittidae). All Halicitidae and Andrenidae specimens were confirmed by Alain Pauly (Royal Belgian Institute of Natural Sciences) and Thomas James Wood (University of Mons), respectively. Other bee specimens were confirmed by the reference collections of Gembloux Agro-Bio Tech. Hoverfly specimens were identified at the species level using the identification key of Verlinden (1994). The specimens were then confirmed by Frédéric Francis (University of Liège) and the reference collections of Gembloux Agro-Bio Tech. We applied Belgian Red List of bees for the conservation status of identified species (Drossart et al. 2019).

Thanks to Data Fauna-Flora v.5.1 software (Barbier and Rasmont 2015), we queried the database of Belgian wild bees, on 26 June 2020, for the historical diversity of wild bees in the Havelange Municipality. The selected geographical quadrat was encompassed within latitude from 50°21'14.4''N to 50°24'46.8''N and in longitude from 5°7'12''E to 5°19'26.399''E. The syrphid historical data were not available for Havelange Municipality.

We conducted one-way ANOVA tests to compare species richness and abundance of bee and hoverfly fauna between sampled parcels of Froidefontaine and Emeville farmsteads, separately. We also validated normal distribution of residuals of each ANOVA test. Subsequently, Tukey’s post-hoc tests were used to compare each parcel pair. We separated the flower strips of Emeville farm from the parcel comparisons because they were not sampled with the same effort as those of the sampled parcels. We compared the species richness and abundance of bee and hoverfly fauna between the feeder flower patch (BFV; Fig. 2B) and the pollinator flower patch (BFB; Fig. 2B) using the Student t-test. All statistical analysis were performed using R 4.0.2 (R Development Core Team 2020) and the resulting graphs were built using ggplot2 and ggpubr packages (Kassambara 2020, Wickham 2016).

Over 2 years (2018-2019) of sampling, we collected 4,303 bees and 1,998 syrphids, representing 92 species and morphospecies from 15 genera and six families for the bees and 31 species and morphospecies from 18 genera for the hoverflies (Table 2). Polylectic, oligolectic and cuckoo bee species correspond to 61%, 14% and 25% of bee richness, respectively. However, the relative proportion of specialised bee (0.9%) was low, with polylectic and cuckoo bees corresponding to 94% and 5.1% in abundance of the total sampled bees, respectively (Table 2). All adult hoverfly species were considered as polylectic species (Frank Van de Meutter, pers. comm.), except for Platycheirus immarginatus (Table 2). In the historical database of Belgian wild bees, we retrieved 18 bee species corresponding to 349 specimens between 1918 and 2007. These data are available in Suppl. material 1. With these historical data of the Havelange Municipality, the bee diversity reached 101 different bee species.

For Froidefontaine farmstead, bee richness in VER was significantly higher than in GC (p-value < 0.05; Fig. 3A) and bee abundance in PAT was significantly higher than in GC, VER and ZH (p-values < 0.05; Fig. 3B). Hoverfly diversity in ZH was significantly higher than in VER (p-value < 0.05; Fig. 3C), while hoverfly abundance was homogenous amongst the Froidefontaine parcels (Fig. 3D). For Emeville farmstead, bee and hoverfly richness and bee abundance did not vary amongst parcels (Fig. 4A, B and C), while DIK parcel exhibited significantly greater hoverfly abundance than EPI, FRE and PAV parcels (p-values < 0.05; Fig. 4D). Only for bee richness and hoverfly abundance, the pollinator flower patch BFB showed significantly higher mean values than the feeder flower patch BFV (p-values;Fig. 5A and D).

Figure 3.  

Mean values of species richness and abundance for bee and hoverfly fauna amongst Froidefontaine parcels GC, PAT, VER and ZH (see details given in Table 1). A. Bee richness; B. Bee abundance; C. Hoverfly richness; D. Hoverfly abundance. Letters above the boxplots represent Tukey's post-hoc comparisons.

Figure 4.  

Mean values of species richness and abundance for bee and hoverfly fauna amongst Emeville parcels DIK, EPI, FRE and PAV (see details given in Table 1). A. Bee richness; B. Bee abundance; C. Hoverfly richness; D. Hoverfly abundance. Letters above the boxplots represent Tukey's post-hoc comparisons.

Figure 5.  

Mean values of species richness and abundance for bee and hoverfly fauna amongst flower strips BFB and BFV (see details given in Table 1). A. Bee richness; B. Bee abundance; C. Hoverfly richness; D. Hoverfly abundance. Letters above the boxplots represent Student t-test comparisons.

In our study, we identified 101 different bee species, corresponding to almost one quarter of the Belgian bee fauna (Drossart et al. 2019). Depicting 57.32% of the total bee collected material, the top-five bee species in both farms were Andrena cineraria (19.59%), Apis mellifera (12.94%), A. haemorrhoa (10.71%), A. flavipes (7.25%) and Lasioglossum pauxillum (6.83%).

Both farms presented suitable habitats to these polylectic species, including open wooded spaces, fallow land or lawns. The abundance of Taraxacum spp. (Asteraceae), Salix spp. (Salicaceae), Craetegus spp. (Rosaceae) and fruit trees could explain the dominance of A. cineraria, A. haemorrhoa and A. flavipes populations. Moreover, they usually nest in south-exposed sites, in bare soils or in areas with sparse and short vegetation (Falk 2015). The other common polylectic bees were mainly ground-nesting species belonging to Andrena and Lasioglossum genera, such as A. nitida, A. gravida, L. calceatum or L. lativentre (Table 2).

Uncommon polylectic bee species were also collected. For example, Andrena trimmerana and Halictus maculatus (Fig. 6C) are rarely observed in the Condroz Region and more largely in Belgium. H. maculatus is a little more common in Wallonia and this species is considered as "vulnerable" in Belgium, but "least concern" in Europe (Drossart et al. 2019, Nieto et al. 2014). Moreover, this species forages on Achillea millefolium (Asteraceae), Centaurea spp. (Asteraceae) or Daucus carota (Apiaceae) (Pauly 2019), which were naturally present or cultivated in both farms. In 2019, specimens of A. trimmerana were collected only in the Froidefontaine farmstead, where Rubus spp. (Rosaceae), orchards, umbellifers or Cirsium spp. (Asteraceae) were flowering. Two specimens of Colletes cunicularius were sampled from both farms. This species is specialised on Salix spp. (Salicaceae) or Prunus cerasus L., (Rosaceae) (Falk 2015). While Lasioglossum leucopus was observed in both farms - probably because of the presence of several of its preferred host plants, Ranunculus spp. (Ranunculaceae), Taraxacum spp. (Asteraceae) and D. carota - this species is considered as "near threatened" according to the IUCN Red List Criteria in Belgium (Drossart et al. 2019, Pauly 2019).

Figure 6.  

Dorsal and lateral side of some rare bees observed within the farmsteads. A. Andrena schencki Morawitz 1866; B. Andrena fulvata (Müller 1766); C. Halictus maculatus Smith 1848; D. Melitta leporina (Panzer 1799); E. Hylaeus brevicornis Nylander 1852.

Rarer species were observed within the farmsteads. Collected in the orchard of Froidefontaine, Andrena schencki (Fig. 6A) had not been observed south of the Sambre and Meuse Furrow for more than 30 years (Rasmont and Haubruge 2002). Andrena semilaevis, a very rare species since 1990 in Belgium (Rasmont and Haubruge 2002), was captured in the orchard of Emeville. This polylectic species is mostly observed on the umbellifers (Falk 2015). Forty-six specimens (1.12% of total sampling) of Andrena fulvata (Fig. 6B) were collected in 2019 in all habitats of both farms, while only one observation was encoded in Atlas Hymenoptera repository for Belgium (Rasmont and Haubruge 2002). That probably means a recent installation of the population on the study sites. However, misidentification due to their morphological resemblance to A. angustior could bias its Belgian rarity (T.J. Wood, personal communication). This species nests in calcareous soils and forages principally on Asteraceae flowers, such as Taraxacum spp. (Falk 2015).

The high diversity of wild bees in the two farms could be linked to the presence of semi-natural habitats around the parcels. Indeed, the implantation of hedgerows, flower strips or shrubby strips between the habitats of both farms provides sufficient floral resources during the foraging activity period of polylectic species (Albrecht et al. 2020).

Thirteen bee species were characterised as oligolectic (Drossart et al. 2019), which represented 24 specimens (Table 2).

Two common species, A. praecox and A. vaga and two uncommon species, A. apicata and A. mitis, were collected in different parts of both farms (Table 2). In Belgium, they are considered as Salix spp. specialists. Moreover, these last two species had never been observed in Condroz before and not since 1950 in the south of Wallonia. A. humilis is a specialist of Asteraceae plant species, such as Tragopogon dubius Scopoli 1772, Hieracium pilosella Vaillant 1754 (Scheuchl 2002) or Cichorium spp. and A. labialis is a specialist of leguminous plants (Fabaceae) (Rasmont and Haubruge 2002).

A single specimen of Melittidae family, Melitta leporina (Fig. 6D), was sampled. The female is particurlarly related to the flowers of M. sativa and T. pratense species (Fabaceae) (Dellicour and Michez 2010), which were abundantly present around the wetland of Froidefontaine Farm. One species of Colletidae family, Colletes daviesanus, forages pollen entirely from composite flowers such as tansy, mayweeds or oxeye daisy (Asteraceae) (Falk 2015).

In Froidefontaine habitats, we also sampled a few specimens of Chelostoma rapunculi, Eucera longicornis, Hylaeus signatus, Megachile ericertorum and Osmia leaiana, probably because their preferred flowers were partially present:Trifolium sp., Medicago sp., Cirsium sp., Rubus sp., Centaurea sp. and Stachys sylvatica L. 1753.

We only collected two specimens of cuckoo bumble bees (subgenus Psithyrus Lepeletier), Bombus campestris and B. vestalis, in Froidefontaine wetland and in Froidefontaine orchard (Table 2), respectively. They are considered rare species (Lhomme and Hines 2018) and their presence could be explained by the relative predominance, in the Bombus genus, of their associated host species: B. pascuorum and B. terrestris (Table 2).

Concerning the nomad bees (Nomada spp.), we identified 15 species representing 4.6% of the collected material. They especially parasitise Andrena spp. and their relative abundance is dependent on the proportion of their host bee species (Sheffield et al. 2013). Most of their host species were collected throughout the two years of experiment. For example, we found, in a small proportion, Nomada flavoguttata and N. leucophthalma, which are linked to Micrandrena spp. Ashmead 1899 (Andrena semilaevis, A. subopaca...) and A. apicata, respectively. On the contrary, N. goodeniana and N. ruficornis were largely present due to the strong dominance of A. cineraria and A. haemorrhoa (Rasmont and Haubruge 2002).

All collected Sphecodes spp. are generalist cleptoparasites, except for S. gibbus that parasitises the nests of Halictus species, such as H. maculatus and H. rubicundus. Their relative abundance followed also the abundance of their host species: the most collected S. epphipius is the cuckoo bee of the most collected halictid bee, Lasioglossum pauxillum (Pauly 2019).

Within both farmsteads, Sphaerophoria scripta was, by far, the most abundant hoverfly species, followed by Eristalis tenax and Episyrphus balteatus, corresponding together to almost three quarters of the total number of collected specimens (Table 2). These species are the most common syrphids encountered in Central Europe (Alhmedi et al. 2010, Francuski et al. 2013, Nengel and Drescher 1990). Aphidophagous larvae of S. scripta and E. balteatus are important for pest control in agricultural systems, while E. tenax larvae recycle the organic matter in wet manures, muds or ponds (Sommaggio 1999). We also emphasised the presence of Melanostoma mellinum, which occured in almost each habitat and particularly in flower strips. Adults M. mellinum are specialised in the floral visitation of anemophilous plants (Van der Groot and Grabandt 1970).

Beside these ubiquitous species, rarer species were found in only a few habitats: Xanthogramma pedissequum, Myathropa florea and Ferdinandea cuprea (Fig. 7). Unlike S. scripta and E. tenax, these species do not migrate. The larvae of X. pedissequum feed on aphids reared on the anthills of some Lasius sp. Fabricius 1804 (Hymenoptera: Formicidae) (Speight 2020). The species M. florea and F. cuprea present a microphagous larval stage. In intensified agricultural landscapes, it is conceivable that the environmental requirements of such species are scarcely fulfilled. Notably, microphagous species appear to be particularly sensitive to pesticides (Schweiger et al. 2007). On the contrary, agricultural landscapes of Froidefontaine and Emeville Farms are suitable for these specialist species, because they include semi-natural ecosystems and organic orchards where cattle or sheep are grazing. We also identified two specimens of Platycheirus immarginatus that are specialist foragers on Bolboschoenus maritimus (L.) (Table 2) (Speight 2020). We did not find this plant species in Froidefontaine farmstead, meaning that P. immarginatus might forage on other plant species.

Figure 7.  

Dorsal and lateral side of some rare hoverfly species observed within the farmsteads. A. Ferdinandea cuprea (Scopoli 1763); B. Xanthogramma pedissequum (Harris 1776); C. Myathropa florea (L.).

Continuous sampling represented only 4.33% of the total hoverfly specimens. However, it allowed us to reveal two more hoverfly species, in Emeville flower strips: Xylota sylvarum and X. segnis, whose larvae are saproxylic and live close to roots and dead wood (Speight et al. 2000).

By in-depth sampling, we documented new occurrences of almost 1/4 of Belgian bee fauna in two farms in ecological transition. For the historical region of the Municipality of Havelange, we have almost quintupled the richness of wild bees community despite high quality monitoring of these populations in Belgium (Drossart et al. 2019). There are few studies of this type in a close environment and with comparable methodology. Therefore, comparing our results with other studies seems to be of little relevance. This study leads us to consider that, on small areas undergoing ecological transition, an important richness of pollinators is easily reached. Moreover, it is possible that the conducted survey underestimates the real diversity per plot, even if the pattern of dominance rarity should be maintained. We also lack data at the end of the season, especially for late summer bees, such as Colletes hederae (Schmidt and Westrich 1993). For hoverflies, we still lack inventory data on the scale of the Belgian territory (Frank Van de Meutter, pers. comm.).

The practices on and around the studied farms seemed favourable to pollinators (Fig. 8) and especially to the polylectic species. In Froidefontaine Farm, the land tenure showed strong impact on bee richness and abundance by an alternation of floral bee-feeding parcels, like the Froidefontaine pasture (PAT; Fig. 3B) and bee-nesting parcels, like the Froidefontaine orchard (VER; Fig. 3A & Fig. 8B). On the one hand, late mowing permits the keeping of abundant floral resources throughout the bee activity period (Meyer et al. 2017) and, on the other hand, sheep grazing permits theconservation of some bare soil sites that favour ground-nesting bees (Cane 1991). Landscape micro-habitats, such as ponds, hedgerows or groves, are important to the survival of many pollinator species, especially by providing habitats for hoverfly larvae (Sommaggio 1999). The wetland of Froidefontaine (ZH) (Fig. 8A) harboured higher hoverfly diversity than the other parcels (Fig. 3C), with species like S. scripta, Cheilosia sp. and E. tenax, whose larvae have different diets (i.e. aphidophagous, phytophagous and microphagous, respectively) (Sommaggio 1999, Speight 2020). The cultivated parcel of Froidefontaine (GC) (Table 2) and the pea crop of Emeville (DIK) (Fig. 4D) showed high abundances of aphidophagous hoverflies, likely caused by the high prevalence of aphids on crops. The flower strips separating the parcels of Emeville Farm consisted of a floral mix especially designed to fill the ecological requirements of bees and hoverflies (Fig. 8D). The floral composition of these flower strips attracted more hoverfly specimens than bees, which were mainly represented by A. mellifera (Table 2). Moreover, they were combined with belatedly-mowed hedges that support floral resources for pollinators throughout their activity season. Similarly, the hedgerows bordering the parcels of Froidefontaine (Fig. 8C), coupled with ecological crop management practices (i.e. no-till, no chemical inputs...), promoted the establishment of wild bee populations (Albrecht et al. 2020). Indeed, hedgerows and other semi-natural habitats usually represent superior floral richness and abundance compared to intensive agricultural land use (Hannon and Sisk 2009).

Figure 8.  

Some field pictures in each farm. A. Froidefontaine wetland (ZH); B. Froidefontaine orchard (VER); C. Double hedgerow between Froidefontaine cultivated parcel (GC) and pasture (PAT); D. Emeville flower strip between FRE and EPI parcels (photo credit : I. Van Dorpe); E. Emeville orchard (PAV).

According to the Belgian Red List of bees (Drossart et al. 2019), we have collected several species indexed in threatened categories from diverse habitats of both farms, especially in the orchard and in the wetland of Froidefontaine. These species were represented by one specimen of A. schencki, one specimen of B. campestris, one specimen of E. longicornis and nine specimens of H. maculatus. We also mitigated the data deficiency in Belgium for a few records of bee species, such as A. semilaevis, A. trimmerana and Hylaeus brevicornis (Fig. 6E). Taxonomically recent recognition, split from species complex and morphological similarity with widespread taxa or less studied genera (e.g. Hylaeus sp.) reflect current taxonomic impediments for 9.4% of the Belgian bee richness (Drossart et al. 2019).

Pollinator composition of each farmstead harboured both common and rare species, which indicates that on-farm diversification and organic practices may be an important refuge for rare, Red-Listed or oligolectic pollinator species (Guzman et al. 2019). Restoring or incorporating diverse habitats in agro-ecosystems is therefore a long-term solution for the conservation of pollinating species (Saint Clair et al. 2020).