Seasonal progression and differences in major floral resource use by bees and hoverflies in a diverse horticultural and agricultural landscape revealed by DNA metabarcoding

The decline in pollinating insects is well documented globally, leading to potentially severe impacts on floristic biodiversity and human health due the loss of pollination ecosystem services (Klein et al., 2007; Lundgren 2016; Smith 2015). Pollinator declines have occurred a combination habitat loss, climate change, pests diseases use pesticides (Potts 2010). As availability floral resources limits pollinators (Goulson 2015), understanding foraging preferences key knowledge need for their effective conservation. Gardens are important, heterogeneous habitats, covering significant areas urban landscapes (Loram 2007). can provide with pollen, nectar nesting sites (Osborne 2008), supporting agricultural (Timberlake 2020) (Potter 2019) settings while increasing connectivity within landscape (Goddard 2009). limited number studies United Kingdom (Wignall elsewhere Northern Europe (Schonfelder & Bogner, 2017) public perception suggests that attitudes towards conservation very positive. However, there wealth information available best plants pollinators, only small recommendation lists based empirical evidence (Garbuzov Ratnieks, 2014), most sold UK garden centres relatively unattractive flower-visiting 2017). Moreover, these broadly target generalisation across wide range functional groups species. Consequently, clear scientific gardens support pollinators. Although vary between pollinator (Bänsch 2020), focus single group (de Vere Honeybees bumblebees frequently studied, however, non-corbiculate bees hoverflies important roles function Additionally, seasonality annual variation influence forage choice (Petanidou highlighting throughout year. There conflicting perspectives as whether native or non-native preferred by but it imperative understand this When surveying visits variety plants, Salisbury al. (2015) found greater abundance near-native taxa than those defined exotic. introduced plant species been shown attract fewer flower visitors natives closely related (Memmott Waser, 2002). DNA metabarcoding has used identify pollen honey 2017; Jones, Brennan, 2021), from bodies (Lucas 2018a; Richardson brood provision nests (Vaudo 2020). advantages approaches include increased taxonomic resolution (Brennan elimination expertise required microscopy (Hawkins overcomes limitations observational methods revealing interactions previously unseen spatial temporal (Arstingstall must be accompanied comprehensive reference library ensure accurate identification. In Kingdom, Barcode project provides 98% coverage all flowering conifers using three barcode markers, rbcL, matK ITS2, allowing reliable identification at genus level majority 2012; Twyford, 2021). Bees were sampled monthly March October during 2018 2019 National Botanic Garden Wales, (51°50′33.4″N 4°08′49.2″W). site diverse (230 ha) set predominately semi-improved (based extent improvement) consists formal organic farmland, designated Nature Reserve,Waun Las NNR (Figure 1). contains over 5,000 world, including many horticultural grown Western Europe. Eight selected sampling broadleaved woodland hedgerows, grassland habitat. abundant both years per transect area season provided Appendix S1. Within each area, 210 m × 2 was established divided into 3 70 sections, walked independently other. Transect walks preferentially undertaken 11:00 15:00 when temperature 10°C. not possible, transects dry days little wind. All seen caught individually stored −20°C prior removal. Further field S2. Permission work ethical approval granted Wales. Pollen washed following modified version protocol described Lucas (2018b). Insects first transferred sterile 1.5 ml collection tube forceps cleaned 70% ethanol insect. catch 1 1% sodium dodecyl sulphate (SDS) 2% polyvinylpyrrolidone (PVP) solution, ensuring any residue sides collected containing Samples shaken TissueLyser II (Qiagen) min 8.5 Hz, stood room 5 min, then again 20 s Hz. Each insect removed placed microcentrifuge ethanol, (see Taxonomic assignment insects, S2). detergent pellet centrifuged 16,200 g supernatant removed. resuspended 400 μl buffer, made up AP1 DNeasy 96 Plant Kit 80 (1 mg/ml) Proteinase K (Qiagen). A extraction. incubated water bath 65°C hr RNase added before disruption 4 30 Hz mm tungsten carbide beads. remaining steps carried out according manufacturer's protocol, excluding QIAshredder second wash stage. negative control included Two regions, rbcL ITS2 amplified via two-step PCR (Table S1, S3). initial final volume μl: template DNA, 10 2× Phusion Hot Start High-Fidelity Mastermix (New England Biolabs UK), 0.4 (2.5 μM) forward reverse primers, 7.2 grade water. repeated twice more pooled purification Illumina 16S 1:0.6 ratio product Agencourt AMPure XP beads (Beckman Coulter). purified further anneal custom unique matched i5 i7 indices sample (Ultramer, Integrated Technologies). This stage 25 first-round product, 12.5 2x Index Primer 6.5 thermal cycling conditions Tag addition confirmed visualisation agarose gel. clean-up followed 1:0.8 Products quantified Qubit 4.0 (Thermo Fisher Scientific) equal concentrations. extraction controls plate sequenced samples an MiSeq (2 300 bp) Liverpool University's Centre Genomic Research (Liverpool, UK). Laboratory contamination Sequence reads processed Ford Jones (2020). Initially, raw sequences trimmed remove low-quality paired merged. Only 450 bp (rbcL) 350 (ITS2) downstream analysis. Identical dereplicated clustered 100% identity singletons (sequence occurring once samples) Sequences compared 5887 (Jones, comprising (Stace, 2019), naturalised alien (Preston 2002) IRIS BG database against blastn, summarising top BLAST hits combining identical results hits. bit scores below percentile excluded. assigned so if bitscore species, sequence If different same genus, genus. belonged multiple genera family, family designation sequence. returning bitscores families orders removed, assuming poor quality sequences. botanical veracity identified assessed considering present study wider landscape. markers sample-by-sample basis verification. Once identifications complete, consensus reached combine differing rule-based, objective, conservative decision process Using taxon summed marker. one marker alone retained. Plants status category Stace (2019). ‘native near native’ comprised also varieties which functionally similar. Naturalised become widespread self-perpetuating wild. classified horticultural. data treated semi-quantitative relative read analyses (Deagle either proportion percentage or, models, sequences, controlling sequencing depth setting total offset, comparable 2021; package mvabund (Wang 2012), multivariate generalised linear model binomial distribution how load composition changed through time. fit strong mean–variance relationship S3), likely distributions rare where mean low, common observation data. To effect time type composition, (coded 1–3, starting spring), year group/order predictor variables, response variable. offset differences 2019; Seasonal changes loads visualised non-metric multidimensional scaling (NMDS) ordination Bray–Curtis dissimilarity returned taxa), vegan (Dixon, 2003). Chi-square contingency test investigate major (constituting 5% sequences) overall), Holm correction testing. split categories ecological trait Functional diversity analysis, S2) tests constituting broader groups. change time, grouped categories. run, variables variable being reads, retaining offset. statistical R v 4.0.2 Analysis separately (Appendix S4). Throughout study, 382 successful 369 individuals No despite surveys out. Pollinators (Syrphidae, n = 195), (Bombus spp., 108), honeybees (Apis mellifera, 44) other (n 22; Table S2, 40,800,709 22,510,682 after stringent (11,305,697 11,204,985 ITS2). regions combined, 191 S1). Six 50% 2): bramble (Rubus spp.), thistles, knapweeds cat's ear (Cirsium/Centaurea/Hypochaeris buttercups lesser celandine (Ranunculus/Ficaria angelica hogweed (Angelica/Heracleum daisy (Asteraceae) meadowsweet (Filipendula ulmaria). An average 17 (SD 9.76) individual 2.55) contributing >1% Overall, we habits four Neither nor order predicted (pollinator group: LR1,363 1753.8, p 0.999, order: LR1,365 953.9, 1.000). ability predict greatest characterising (Diptera, Hymenoptera) rather (bumblebees, honeybees, hoverflies; S3, was, difference Diptera Hymenoptera (x2 46.26, df 5, < 0.001; Figure 3). large (Diptera) Angelica/Heracleum valuable (Hymenoptera). Cirsium/Centaurea/Hypochaeris spp. contributed lower hoverflies, Ranunculus/Ficaria abundantly bees. groups, 4). represented short- long-tongued 50.179, 20, 0.001). attributed short-tongued hypnorum, B. lapidarius, pratorum, lucorum/terrestris agg.) (B. hortorum, pascuorum), utilised Trifolium pratense Rubus Honeybees' similar Impatiens glandulifera bees, significantly body size 433.01, 52, 0.001), extra carrying mostly Heuchera small-sized Rudbeckia/Helenium medium-sized Taraxacum officinale. hoverfly differed various larval requirements 235.4, 48, carnivorous detritivorous utilising herbivorous Season good (LR2,367 2632.8, regardless (LR2,366 816.2, 0.828; Figures S2–S4, 147 170 2019, 71 years. NMDS shows 5). progression visible assessing foraged 2) (Figures S5–S7, Aster Clematis officinale Actaea Geum Bidens/Coreopsis Hoverfly largest 6). Native spring summer autumn (LR1,367 58.10, 0.001) 6), (LR1,366 3.14, 0.369). see array no dominant S9, S8, S10–S12, metabarcoding, reveal visited broad bumblebees, hoverflies. We show shared (Hymenoptera) within. strongly influenced season, utilise end season. Floral overall among (hoverflies, bees), (hoverflies) (bees). comparison thistles umbelliferous less. possible explanation preference accessibility may morphology influencing recorded here generally shorter tongues (King, difficulty fully removing long corollas Cirsium, Centaurea Hypochaeris. While evidently able resource, issue reason prioritising shorter, open flowers demonstrate partitioned traits studying Cullen (2021) had impact local itself, importance relationship. Tongue length widely known affect thought species' vulnerability extinction tend specialise corollae 2005). did find relating bee period interpreted caution required. Non-corbiculate comprise cosmopolitan suite ecologically distinct Kingdom. Halictidae Andrena, along kleptoparasitic Nomada (thoracic width mm) making comparisons difficult. distance (Greenleaf 2007), immediately them sampled. complex habitats occupy (Schirmel 2018) although link studied. resource guilds, highlight mate seeking, therefore additional food, breeding oviposition phytophagous (Rotheray Gilbert, 2011). biggest use, relying S5). phenological patterns result shifting availability, temporally limiting insects. These shifts require alter survive, flight periods short (Ogilvie Forrest, often play role findings supported who showed attracted garden, extended period. high provision. such comes caveat since I. highly invasive, its displace (Chittka Schürkens, 2001). they contribute diets. For example, spring, supplemented lacks essential amino acids needed (Génissel multi-locus approach allows studied fine scale, improving detected discrimination achievable 2021) alternative 2019). information, traditional techniques, example frequent inflorescences (Branquart Hemptinne, 2000). Due potential biases sampling, extraction, amplification (Bell 2016), should semi-quantitative, estimates Analysing approaches, Frequent represented, difficult detect, case Recent developments suggest some cases quantitative (Richardson particularly regarding (Piñol Furthermore, species-level challenging meeting ideal (CBOL Working Group, genus-level designations fine-scale plant–pollinator interactions, identification, universal discriminative available, scale Our conclusions landscapes. allow us evidence-based list growth forms S4, improve previous providing perspective insect, scope observations relevance gardeners, land managers, producers policy makers inform decisions planting greenspace appropriately supported. N.d.V., L.J. A.L. received funding Welsh Government Rural Communities – Development Programme 2014–2020, funded European Agricultural Fund Government. Knowledge Economy Skills Scholarship (KESS2), part-funded Government's Social (ESF). icons contained created Thomas McBride. acknowledge Supercomputing Wales project, Regional (ERDF) None authors conflict interest. conceived A.L., N.d.V. S.C. Data laboratory A.L.; compiled analysed suggestions G.B.; manuscript written contributions authors. gave publication. Raw Read Archive PRJNA763761. Dryad Digital Repository https://doi.org/10.5061/dryad.rjdfn2z9s (Lowe 2022). code https://github.com/colford/nbgw-plant-illumina-pipeline. S1: Supporting S2: Methods S3: Tables S4: Results S5: Discussion Please note: publisher responsible content functionality supplied Any queries (other missing content) directed corresponding author article.