Sustainable landscape, soil and crop management practices enhance biodiversity and yield in conventional cereal systems

Functional biodiversity generates essential ecosystem services to agriculture such as biological pest control, pollination and nutrient cycling. Yet, intensive agricultural practices threaten ecological functions thereby posing long-term risks sustainable food production human health (Dainese et al., 2019; Foley 2005). A potential solution is the intensification of farming, which allows farmers enhance sustainability by managing yield-supporting via reduction management (Bommarco 2013; Pywell 2015). But are suitable for this purpose what their relative effect on yield compared conventional, methods? Productivity staple crop winter wheat Triticum aestivum (L.) directly depends agrochemical inputs, well soil characteristics, management, herbivory pests, weed pressure infections with pathogenic fungi (Ray 2012). Degradation fertility (Edmeades, 2003), inappropriate sowing densities (Ozturk 2006) (Dedryver 2010) increase gap between attained in farming systems. Accordingly, healthy soils diversification enhances nutritional balance improves productivity (McLaughlin & Mineau, 1995; Ratnadass Similarly, soil, landscape that favours beneficial (e.g. predators) top-down control herbivorous pests (hereafter ‘pests’) may indirectly improve yield. This was shown fertile (Birkhofer 2008; Tamburini 2016), reduced pesticide applications (Jonsson 2012; Krauss 2011) fields located diverse, well-connected cropping systems landscapes (Fahrig 2015; Martin Redlich, Martin, Steffan-Dewenter, 2018). However, responses heterogeneity often variable or taxon-specific (Karp 2018), both predators not result improved regulation (Tscharntke 2016). Soil, factors also interactively influence predators, (Lichtenberg 2017). For instance, benefits high organic carbon content (SOC) across Europe were mineral fertilizer application (Gagic These trade-offs lower anticipated measures farmers’ trust implementing research real Implementation can further be hampered if drivers difficult manipulate. Even though natural habitat important biodiversity, extent cropland rarely very regions Germany. Other SOC depend type, ploughing frequency rotation. From a practitioner's perspective, it advantageous link adaptable structures field size) actual rotation diversity). In study contrast previous 2017; Garratt we identify reduce while maintaining yields conventional We concurrently consider variation non-adaptable characteristics) commonly used application). Thereby, aim bridge science practice increasing our knowledge local landscape-scale options achieve (Kleijn 2019). furthermore pre-empt criticism practitioners regarding transferability results from small-scale, controlled experiments agroecosystems. Based theory, expected find (a) positive effects usage, characteristics SOC) weeds diseases hypothesized (b) extensive improving structure add intensification. assumed (c) increases spatial temporal resource availability specific arthropods productivity, either negatively (enhanced pressure) positively predator abundance). investigated these links importance using structural equation modelling (SEM). As relationships trophic levels (predators–pests–wheat) negative (top-down processes, i.e. control) (resource-driven, bottom-up host prey availability) assessed (Vidal Murphy, total fourteen pairs (28 fields) selected 2014 near Würzburg/Germany (49°47′N, 9°57′E) along gradient defined % semi-natural cover within 1-km radius around (Figure 1a, detailed design Appendix S1 Supporting Information). Fields paired according contrasting 1b). To avoid correlations other variables, took care pair similar pH, texture margin quality; include different till no-till, vs. input); minimize configurational compositional variables analysis (Appendix S1). Average distance (mean ± SD) 1.07 0.9 11.2 3.8 km. each field, randomized, fully crossed experiment pyrethroid insecticide implemented four adjacent treatment plots (12 × 14 m each, 4 28 = 112 plots, Figure treatments, type amount substance timing followed regional recommendations 1c; Fertilizer (total 190 kg/ha) applied at three growth stages fertilization surrounding SD 215.7 42.8 kg/ha, range 120–339 farmer survey). Insecticide (active beta-cyfluthrin, 300 ml/ha) sprayed once after first surveys 1c) test predators. integrated framework Bavarian Regional Agricultural Institute, required standardized all fields, even thresholds crossed. considered 34 explanatory direct indirect system. separated into biotic sampled scales (Table 1; The likely plant growth, yield, but ground-dwelling providing conditions, Birkhofer 2008) included proxy matter, topsoil pH. derived data experimental set-up (fertilizer treatment) performed autumn 2014. survey covered recent 2013/2014 11.25 0.43 years), focussing known (in)directly affect frequent no-till should less disturbed offers abundant alternative 2008). 1 km (ArcGIS v. 10.4, ESRI) including configuration composition relevant enemies. example, diversity regulation, (Redlich, set (herbivory, disease) (predation) determine comprised abundance activity density plots. Responses aspects (Martin Wende, therefore divided eight functional groups (four groups, Table 1). Within differentiated two main aphid species (Hemiptera: Sternorrhyncha), English grain Sitobion avenae (Fabricius), preferentially occurs feeds reducing rose-grain Metopolophium dirhodum (Walker), stem leaves biomass. Cereal leaf beetles Oulema spp. (Coleoptera: Chrysomelidae) grouped larvae (sessile, cause damage potentially biomass compensation occurs) adults (winged, mobile, responsive factors). Among leaf-dwellers classed ‘active flyers’ (ladybirds [Coccinellidae, larvae], hoverflies [Syrphidae, lacewings [Chrysopidae, larvae]) ‘passive (wind-dispersed web-building spiders [Araneae]). While group benefit large-scale heterogeneity, latter thought respond management. Abundances leaf-dwelling counted 50–100 tillers during transect (two application, 1c). per then 50 summed estimate throughout growing season remaining soil-dwelling (ground [Carabidae] ground-hunting [mainly Lycosidae]), pitfall traps (one sampling round application; additional estimated (% cover) forbs plot infestation rates Fusarium rust (Puccinia spp.). Pathogens recorded visual Compensation response disease common crops identified assessing components. every plot, measured harvest Yield quantity weight (‘grain yield’, t/ha) (t/ha). Both extrapolated 0.25 m−2 subsamples. Thousand kernel (TKW, g) acted quality. components treated final 1), tested among other. quantify developed piecewise SEM. SEM consisted eleven individual path models biomass, TKW, S. avenae, M. dirhodum, adults, ground spiders, predatory carabids, active passive flyers. standardization abundances nature resulted continuous, non-integer data, linear mixed Gaussian distributions. log-transformed normality residuals. Explanatory model pre-selected steps (details Selection due theory 1, coloured squares). Sub-selection based Pearson's correlation coefficients (r|>0.2|, initial grey frames). resulting simplified removing non-significant terms (p > 0.05) backwards elimination (dashed frames, S2). marginal p-values 0.05–0.1) those initially excluded lack only entered inclusion increased fit AICc size influencing abundance), directed separation (D test, Shipley, 2009) revealed missing significant (effects date carabids). (predators–pests–wheat), top-down, enemies blue triangle). positive, resource-driven observed S1), adapted direction accordingly purple preliminary analyses did reveal any interactive fertilisation regime S3), no interaction included. random intercept ‘Field’ nested ‘Pair’ accounted multiple (field pairs) field). Final S2) validated graphically normality, homoscedasticity autocorrelation (Moran's I, p 0.802). Collinearity low (variance inflation factor < 3). final, showed good Fisher's C 356.2, 0.215, conditional (R2) ranged 37% 74% variance explained 2). standardised scaled mean (Schielzeth, 2010). Additional masked pooling surveys. fitted separate (no soil-dwellers conducted). Fixed ‘Survey’ (three levels), (yes/no) interaction. Although survey, rounds affirm effectiveness (i.e. Post hoc Tukey comparisons evaluated differences arthropod treatments. ‘Plot’ repeated landscapes, All R version 3.3.2 (R Development Team, 2016) packages nlme (Pinheiro PiecewiseSEM (Lefcheck, , car (Fox Weisberg, visreg (Breheny Burchett, Our average (±SE) (8.5 0.3 TKW (44.2 0.4 fertilized area (8.1 t/ha, surveys), Germany (grain 8.6 ~ 44 neighbouring France, Europe's largest producer 7.46 47 g). Plant strongly correlated 26% 30% respectively (Figures 2 3a,d). Grain (not biomass) significantly sand loamy (−36%, Figures 3b,e). found pH SOC. had strongest quantity, reductions being greater than 3c,f), yet not, expected, High 2; S1A), quantity. Neither pathogen nor (indirect biomass: −0.06, Contrary expectations, observe biomass). however, larger preparation S1B,C). Of 18 143 individuals surveys, most (89%), (5%), (4%) (2%). application. occurred large S2B,C). Indirectly, enhanced nitrogen effects: 0.17 0.2 respectively). contrast, decreased 78% S2D), follow-on effect: 0.2). No pooled that—despite 1—abundances 60% beforehand unsprayed Still, populations quickly recovered (Tukey comparing non-insecticide Survey 1: 0.22, (after application): 0.001, 3: 0.822; S3A). larvae, lasted longer (Survey 0.99, 2: 0.026; S3B). Unaffected agrochemicals, more prominent sown later year S2E). At scale, influenced numerous S2A), adults—and lesser offspring effect)—were S2F,G). rather weak though, quality respectively. 2,594 consisting (48%), carabids (30%), (14%) flyers (8%, primarily Coccinellidae Syrphidae). those, responded density, particularly flyers) (passive flyers; S4B,C). insecticide-driven affected −0.36). neither (SEM) distinct (additional analyses). Temporal could tested, opened Instead, disturbance, opposed deep ploughing, 32% 49% 4a,d). Furthermore, higher late 4e; S4F). Densities declined addition residues 4b; S4E). Leaf-dwelling 4f). Surprisingly, factor. predictions, scale here played minor role abundances, exceptions. small 4c), edge S4D). Active (but passive) S4A), confirmed heterogeneity. an indication prevalence bottom-up, processes Consequently, found. Input-driven, modern basis linked ongoing environmental problems. Together demands, calls able maintain current productivity. Ecological utilizing services) aims goal. study, pathways investigating multifactorial real-world cereal agroecosystem. inputs relatively indicate without causing pressure. demonstrate biodiversity-enhancing ambiguous highlight under conditions inputs. non-fertilized so (‘plant vigour hypothesis’, (Price, 1991)). indicator seeding milling wheat. yield-damaging effectively controlled. numbers rebounded short-term (Krauss 2011). system avenae. An interesting finding, considering treatments following practices. although typical time, substance, independent recommended frameworks. On farms, unnecessary, precautionary spraying severe death non-target (pyrethroid) rapid resettlement mobile ladybirds tends have augment region case, SE: 1.39 0.12%). condition diversely cropped, provides optimal crops, illustrated matter saprophagous insects springtails, alters physical speeds decomposition unfavourable below-ground prey. Soil-dwelling combined conservation (no-till, diversity, fertilizer, Bai enhancing quality, diverse rotations characterised chemical 1995). explain spiders) fields. addition, timely (as late) earlier maturation (Acreman Dixon, 1985), seen dirhodum. early pronounced counteracted preparation. relate limited coverage land use intensity gradients constraints site selection conditions) choice scale. differs depending life history movement capability taxa (Chaplin-Kramer 2011; multi-scale highlighted advantages landscape-level scales, enemy (<250 >2,000 scale) (<500 scale). studies, 1,000-m (Bosem Baillod Rusch issues testing, restricted priori well-supported intermediate missed some effects, nevertheless relationships, ground-dispersing responding sizes, density. Predators easy access boundary on-field Yet crop-non boundaries hinder dispersal act barriers provide competing resources (Ratnadass Last, monitoring efforts aerial like Vespid wasps readily changes decrease (Bianchi 2006; Bosem 2017) (Schneider amounts smaller habitats favour agriculture. caution against proclaiming panacea Alternatively, heterogeneous experience landscape-wide intensities possible test. shows compensate competition caused (Freeze Bacon, 1990). pathogens (maximum rate 58% 34%, respectively), feeding obvious With exception confirm fears participating would lead losses. reached system, controlling means Soil consequently (Rusch underlying partly, confirmed. related There several explanations finding: dependence (Pywell Vidal way see involved predation exclusion). Rather trait-based been better efficiency varies season. beginning imperative infestations, on, (d) generalist suffer chewing insects. cases, offer possibilities supporting factorial field-scale practicable sustain Strengthening (intensity diversity) restore practices, together tools currently fostered future adaptive Considering threats environment health, losses acceptable price economic turnaround fact, strength legacy fade away, gaps. snapshot time closely climate 2014, persist climatic exaggerated change. strengthen resilience global change new extreme weather events. thank kindly permitting us work land, Matthias Jäger, Verena Seiffert Florian Haydn lab assistance. Crop provided State Ministry Nutrition, Agriculture Forestry Surveying Mapping Authority FP7 Project LIBERATION (‘LInking farmland Biodiversity Ecosystem seRvices effective intensificatION’), funded European Union, grant number: 311781. Open funding enabled organized Projekt DEAL. E.A.M., S.R. I.S.-D. conceived idea designed methodology; collected analysed led writing manuscript. authors interpreted results, contributed critically drafts gave approval publication. Data available Dryad Digital Repository https://doi.org/10.5061/dryad.bk3j9kd9m (Redlich 2020). Please note: publisher responsible functionality information supplied authors. Any queries (other content) corresponding author article.