Field boundary features can stabilise bee populations and the pollination of mass?flowering crops in rotational systems

Pollination is a key ecosystem service to global agriculture, enhancing production in ~75% of crop species (Klein et al., 2007; IPBES 2016). Demand for pollinator-dependent crops has continued rise (Aizen 2019) and there growing international concern over the impact pollinator losses on food (IPBES Within UK, populations declined occupancy by ~25% between 1980 2013 (Powney 2019), driven particularly reduced occurrence rare species. Declining have been linked with disruptions plant-pollinator networks (Redhead 2018) pollination deficits already recorded (Garratt, Breeze, 2014). Since 1930s, UK’s agricultural landscapes moved from diverse mosaics mixed farming semi-natural habitats towards large-scale monocultures (Senapathi 2015). These lower complexity reduce floral resources pollinators (Baude 2016), resulting abundance diversity (Shaw 2020), smaller bumblebee colony size (Bukovinszky 2017) richness (Fijen 2019). The growth mass-flowering crops, such as field beans Vicia faba oilseed rape Brassica napus (hereafter OSR) can benefit pollinators, providing highly abundant those physically able access them (Holzschuh 2013; Westphal 2003). However, their short flowering season rarely covers entire active period local so life history (whether long-lived building, or solitary short-lived) affects whether not corresponding reproductive gains are realised (Riedinger 2015; 2009). Furthermore, practice rotation (whereby arable fields sown different each year prevent disease/pest build up replenish soil nutrients) adds an extra degree spatiotemporality these resources, they predominantly rotated cereal that offer no (Hass 2019; Marja 2018). When absent, must therefore endure (often multiple) intervening years ‘resource drought’. Again, individual species' responses this will be influenced mobility, mobile, (such colony-building Bombus sp.) better travel find disparate while more sedentary (e.g. bees Andrenidae) respond closely in-situ resource fluctuations Fluctuations changes community composition knock-on effects service, which relies both (Garibaldi 2020). Spatial variation abundance, due variable habitat provision may cause level receive when fields, can, turn, generate variability yields (Bartomeus Perrot To combat biodiversity declines, agri-environment schemes support growers increase landscape either taking land out adding along boundaries (Batáry Of boundary measures most commonly supported across Europe, flower-rich margins hedgerows thought provide beneficial (Cole 2020; Garratt 2017). Both absent bloom cannot utilise (Garratt 2017; Grab 2018; They also important nesting/overwintering sites within systems, where much frequently disturbed unsuitable (Ullmann longer-term studies suggest genuinely effective at increasing services if established several successive (Grab Morandin While demonstrated benefits features Albrecht Scheper 2013), less known about influence temporal stability services, partly because monitoring long time-scales challenging. consistency yield crucial farmers, attention being paid role structure functional resilience Redhead Floral serve refuges (Marja help sustain after (Sheffield 2008). Thus, could stabilise vary temporally multiple time-scales. This includes shorter than lifetime weekly, occurs briefly same location) multi-year longer resource-rich only occur given once set number years). latter relevant spatiotemporal dynamics crop-rotated landscapes. In study, we use validated process-based model investigate impacts ground-nesting bee under typical low-diversity cycle OSR bean crops. typically included UK rotations alternating third following two Coston, 2014; Lindström 2016; but see Bishop By using computer simulations, isolate estimate larger areas would ever feasible empirical study. We sample 117 10 × km2 study distributed England, represent realistic feature configurations, simulate 20 rotation. Focusing standard narrow grassy margins, wide examine how amount population (i.e. inter-annual size). compare predicted bumblebees sedentary, short-lived populations. then reliability management approaches maximising stabilising effect features. Poll4pop (Gardner 2020a; developed via Häussler Lonsdorf 2009; Olsson 2015) predicts spatially explicit flower visitation rates central-place-foraging bees) landscape, based nesting foraging preferences distances. It social (accounting time), allows ranges dispersal, preferential rewarding replicate nest site limitation, incorporate fine-scale landscape. operating rasterised landscapes, uneven delivery generated proximity other habitats. accounts seasonal differences cover offered outputs rate per pixel season, time all nests spend pixel. Solitary assumed during one (user-selected) reflecting flight periods majority (Falk, Social bumblebees) three seasons. simulates behaviour assuming queens forage 1 produce workers, 2 additional workers. total gathered workers 3 determine new produced end period. contrast, females solely dependent original nest-founding female. run dispersed reproductives previous starting next reproduce source/sink dynamics. For detailed description model, al. (2017) validation Great Britain (including sensitivity analysis) Gardner (2020c), henceforth G2020. terrestris) Andrenidae). largest wild guilds include many (Hutchinson 2021). take parameters consist estimates density foraging/dispersal distances derived literature (Greenleaf Gathmann & Tscharntke, 2002; Franzén Nilsson, 2010 used Dicks 2017), plus cover, attractiveness expert opinion questionnaire, experts scored experience (maximum n = experts; G2020 details). adopt rather calibrated values, ecological unfeasibility some values. showed parameterisation incorporating produces predictions significantly agree observed abundances transect surveys 239 Britain. capture duration mass-flowering, adjust definitions seasons instead early spring (roughly March–mid-April; assumes foraging), late (mid-April–May; foraging) summer (June–August; half final season. reflect this, scores suburban halved apportioned equally (since represents multiplied duration). Based types, (and linseed/flax, peas, strawberries, raspberries berries) assigned spring, zero spring. opposite approach applied orchards, confirmed prescription maintained significant agreement rerunning procedure described (see Supporting Information). bees, retained subdividing into late), since do show emergence times. twice bees: spring-flying summer-flying species, independent solitaries, is, representing later periods. Tables S3 S4 Information detail expert-derived landcover class feature. (Figure 1), showing area (interpreted proxy intensity production; Figure 2). subset G2020, now select located England contain overlap another Overlapping was permitted represented buffers around central survey interest. compares landscape-level properties, (?25%) overlapping omitted. Generation fully Briefly, CEH Landcover Map 2015 (LCM2015), Ordnance Survey orchard polygons overlaid top location information 2016 rural payments agency databases. m size. Each accompanied maps. first hedgerow locations (‘hedges’) Woody Linear Features Database (Scholefield shown predict insect (Sullivan database does 100% hedgerows, augmented perimeter any parcel claiming options through Countryside Stewardship Scheme 2016. assume (width ‘favourable condition’; Defra, 2007). second map (‘agri-env.’) incorporates buffer strips, fallow ditches whose presence subsidised list scheme options). A 4 width 4–6 recommended strips; e.g. 2020) and, lack features' exact locations, were mapped associated claim. approximation us identify generous well enhanced resources. margin combination grassy, specific fields. (‘narrow’) defined ‘Arable Horticulture’ LCM2015. wide. consists base maps (‘narrow’, ‘hedges’ ‘agri-env.’), approximate real-life configurations added removed will. S1 shows type varies §2.3 snapshot impose 6-year sequence cereal-cereal-OSR-cereal-cereal-field five subsequent note likely undergo reality consecutive cereal), organic low input systems often variable/diverse rotations. choose facilitate simulation reasonable computation sustainably manage health/fertility. constrained stages 6 rotation, respectively. 1, 2, 5. Their progression uniquely predetermined current state. Therefore, randomly stage field. results adjacent stage), alternative realities, random allocation reality. quantify uncertainty introduced calculating mean error realities. years, feeding surviving Section 2.4. enables complete cycles sequence, discarding year. scenarios, times landscape—one state allocations. simulation, calculate pixels landscape) year, reflects magnitude significance fitted coefficient allow fractional change time-averaged (or rms rate) relative no-boundary value depends covered higher (at in-crop) stable variability). Across median reduction types ~3% 5%, large 20%–25% 3). Including hedges generally including suggesting former made contribution fact that, although wider features, small contained S1). range values boundary-feature rotating-crop varied Fitting Equations 8 9 allowed separate unit greater populations) no-boundary-features scenario 4, left-hand plots, green bars). true greatest area, clustered together nearby collectively managed single participating farm. all-boundary-features scenario, intermediate hedge despite common covering far S1) dominating overall response. right-hand provided strongest effect, negative coefficient, indicating ~7% area. mass flowering, statistically (likely respect crops) did significant) (~12% area). in-crop similar trends 4; blue orange Landscapes (>10% peak late-spring flowering) (~8% season) scenario. cases, stronger landscape-level, dilution sub-populations non-crop landcovers remote time. Although (<1%), 80% reductions ~20% ~5% summer, respectively 5, boxes). (median ~5%–10% ‘all features’ scenario), lower, 1%–2% Equation 6, summer. trend average present opposed (compare Figures bars/boxes). An exception increased margins; indicated potential ~900% our particular assumptions. 7 (left panel) strength illustrating decrease amplitude function fits compared 6; scenarios (with confirming decreases cases (right-hand illustrates compensate expected minimum yearly (mean-rms) nonetheless without. explore (hedgerows margins) simulations enabled assess without interference sources weather) times-scales Our findings demonstrate upon buffering against According increases importance seasonally according During high scores, blackthorn Prunus spinosa hawthorn Crataegus monogyna. stabilising, mixes (Byrne delBarco-Trillo, Cole Timberlake promoting bumblebees' requirement sustained throughout prolonged emphasises continuous (which not; least provides likewise populations, interventions lifetimes, building female provisions her own offspring emerge Other complex/bivoltine histories exist Andrena trimmerana; Falk, 2015), simulated Consequently, opportunities offer, productivity still strongly immediate window, As such, extensive experienced longer-lifetime bumblebees. (~100–200 vs. 500–1,000 bumblebees; Carvell 2012; 2002) reliance permanently concurrent patches good advantage effectively shift centre pursuit rotating like mobile boost population, permanent settings maintaining stability, sufficient phenologically self-sustain patch, resorting beyond it. Many low–moderate contrast Tscharntke