Benefits of increased cover crop diversity for predators and biological pest control depend on the landscape context

Agricultural intensification is the main cause of biodiversity loss in terrestrial ecosystems (IPBES, 2019). By harnessing ecosystem functions supported by biodiversity, agroecology aims to design agricultural landscapes that can maintain commodity production while preserving environment (Bommarco et al., 2013). Biological control crop pests their natural enemies a major service increase sustainability agroecosystems (Lewis 1997; Rusch 2017). It contribute reducing use pesticides and harmful effects on maintaining productivity (Geiger 2010; 2010). Identifying agroecological practices beneficial biological pest services enhancing diversity therefore crucial for future agriculture. Multiple studies have demonstrated positive relationship between functioning across types (Cardinale 2012). In agroecosystems, strengthen delivery key such as (Dainese 2019; Ostandie 2021). Plant particularly multiple plants shape trophic interactions several levels (Andow, 1991; Isbell 2011; Scherber Increasing plant often suggested management option harness biodiversity-based (Aguilera 2020; Kremen & Merenlender, 2018). diversification encompasses non-crop well various be implemented at scales from field-scale (e.g., cover cropping, intercropping) landscape-scale diversifying rotation, semi-natural habitats) (Letourneau While separate local landscape been extensively studied, relative joint spatial remain poorly explored. At field scale, increasing limit density insect damage (Barnes Letourneau Wan 2020). has estimated reduced abundance 23% compared with monocultures 2011). The are mediated direct bottom-up (the resource concentration hypothesis) indirect top-down According hypothesis, more likely find host monospecific communities due high increases immigration decreases emigration (Root, 1973). mechanisms behind these involve disruption localization chemical or physical perturbations, changes quality Castagneyrol Jactel, 2012; Finch Collier, 2000; Thiery Visser, 1986). hypothesis suggests abundant, diverse efficient leading lower densities (Russell, 1989). These higher abundances alternative prey hosts, pollen, nectar, microhabitats than communities. Evidence supporting two non-exclusive hypotheses exists literature (Moreira 2016), but extent which context modulates prevalence remains habitat shapes enemy level they deliver (Bianchi 2006; Chaplin-Kramer 2016). Semi-natural habitats, hedgerows, grasslands forests, support resources habitats preys, pollen overwintering sites (Landis Sarthou 2014; Tena 2015). affected structure, direction magnitude populations damages highly variable (Karp Such context-dependent structure could interactive management. Indeed, according intermediate complexity might non-linearly modulate effect (Tscharntke this options (such flower strip, tillage organic farming) limited very simple complex landscapes, highest expected complexity. (less 20% habitats), species pool supposed too low an response scale. any further aiming improve services. local- focus number recent (Muneret 2019a; Winqvist However, few specifically quantified control, yielded inconsistent findings (Albrecht Tschumi 2015; Moreover, scale differ significantly perennial annual crops, different spillovers organisms (Rand Tscharntke Yet, crops vineyards, understood. study, we explored composition along wide gradient proportion landscape. We asked if enhanced richness independently vineyards. focused vineyards because intensively managed France, total pesticide only 3% land (Sabatier 2014)), many efforts currently undertaken implement practices. Furthermore, local-scale challenging given not option. inter-row however practice gaining attention agroecosystem (Daane 2018; Winter unclear (Wilson hypothesized (1) enhance richness, potential lowly 2011), (2) guilds differently 2019a) (3) simplified Wilson A experiment was conducted conventional located Bordeaux area, France (45°01?28“N, 0°21?13?W) 2018. Nine were selected ranging 22 63% 1 km radius around each site. measure correlated other measures heterogeneity our region mainly composed forests Roschewitz 2005). used radius, line studies, it explains responses predators Petit site, treatment (flower strips) (grassy applied one within same farm soil type. This ensured arthropod treatments subjected exact type experimental resulted nine pairs All planted Merlot variety (Vitis vinifera) cultivated without irrigation, under regular (mostly fungicides), mowing every Papura consisted mixture 20 designed vineyards’ containing range functionally Fabaceae, melliferous plants: see Supplementary Materials full list species). Poaceae (Lolium perenne Festuca rubra). Plants sown manually after 24 April 11 May 2018 out inter-rows, eight inter-rows per plot. Cover distance m plot margin center (Figure S1). Two (soil- vine-dwelling) sampled field. Soil using pitfall traps vine beating nets. samplings three distances treatments: middle (0 m) 15 30 away crops. Three sampling sessions (in June, August September). For session (0, field, filled soap water placed week before being collected individuals pooled identification. Vine arthropods randomly vines session. end, (n = 3), guild 2), 2) 9), sample size n 3 (one session), yielding N 324. Individuals belonging community (spiders, Carabidae, Opiliones, Neuroptera Dermaptera) identified nearest possible taxonomic level, juvenile spiders identification, about 77% all thus genus. calculated groups sample. Rarefied samples where zero. As effort identical experiment, instead rarefied richness. evaluated vineyard pests, European grapevine moth (L. botrana Fam. Tortricidae) (Thiéry 2018), sentinel cards 2019b). L. causes important phytophagy grapes promoting development bunch rot serious qualitative quantitative (Delbac Thiéry, heavily attacked up 10–30 larvae complete destruction depending cultivar (Fermaud Each card 10 eggs laid parchment paper laboratory then glued plastic × 8 cm. Moth obtained own rearing (temperature: 22°C, hygrometry: 60%, photoperiod: 16D/8N). attached stock, close grapes, treatment, left days collected. measured based removed card. measurement July, During last session, 4 egg availability incompatibility spraying programs. Across lost/destroyed final 999 cards. tested (low versus diversity), (proportion 15, abundance, (soil vine) predation rates. proxy sum proportions meadows fallows. regional database PIGMA (http://portail.pigma.org) 1000 site ArcGIS 10.1 (ESRI). generalized linear mixed models (GLMMs) analyze R package lme4 (Bates Core Team, analyzed constrained ordination (partial Redundancy Analysis) vegan (Oksanen model included fixed Treatment (two-level factor: diversity); Guild enemies); Landscape (continuous variable: scaled percentage radius); Distance crop). also random intercept (factor 27 levels) account non-independent observations date first null (Abundance ? + (1|site:session)) verified residuals DHARMa (Hartig Lohse, Based patterns, quadratic improved fit residual plots, biologically relevant autocorrelation inspecting coordinates plot, found no significant pattern. variables Wald Chi-square tests likelihood ratio tests, function drop1 (Zuur 2009), both approaches results. Following Harrison al. (2018) Forstmeier Schielzeth (2011) , report results optimal model. similar approach investigate taxa treatment. above. fitted negative binomial GLMMs provided better LMMs gaussian distribution, even log-transformation. diagnostic followed protocol simplification reporting (Harrison Zuur 2009). modelled predated GLMM trying distributions (binomial, Poisson) poor data plots. following tested: Predation * (1|site:session) Again, plots initial did include non-linear effect. RDA above analyse Treatment, Landscape, predators. summed analysis. matrix Hellinger transformed prior Permutation (using permutations) test significance axes. 5148 198 (Table 1); 73% spiders, Coleoptera, 6% (50 individuals) Dermaptera (13 individuals). 67% 28% Coleoptera 5% Opiliones (with 156 taxa). dominated 86% 10% 82 terms average rate 0.30 (SD: 0.27, 999) experiment. enemies, overall 1(a), 5.82, df 1, p 0.016). predicted 140% modulated (significant interaction: 6.74, 2, 0.034). (i.e., 40% complexities 2(a)). There slight (Chi-square 4.92, 0.027), indicating progressively declined (on 17% less crop; Figure S2)). Finally, differed 121.75, < 0.0001) much arthropods, showed (interaction: 4.91, 0.086) 1.39, 0.239). evidence changed 1.89, 0.169). Taxonomic varied 125.63, 0.0001). (average 8.37 (CI: 7.36–9.52) 5.16 4.46–5.98) guilds, respectively). had 2(b), 4.83, 0.090), nor 2.13, 0.145). 3; F 0.76, 0.676). Taxa considered (F 49.3, 0.001, 3.02, 0.009) interaction 2.79, 0.01) 3). modify 1.08, 0.307). rates depended 10.46, 0.005; 2). Overall, 8.46, 0.015), complexities. shown 2(c), 42% (between 23 45% distance-decay 0.03, 0.86). present study sheds new light agroecosystems. Our reveal locally cascading control. influence enemies. Importantly, highlights clear most complexities, confirming moderator success Russell, 1989), extend previous reports strips adjacent within-field (Sáenz-Romo 2015 functional offered wider refuges additional food sources probably explaining (Balzan Frank Reichhart, 2004; Koricheva Hayes, Langellotto Denno, Wäckers 2007). true sensitive (Bruggisser Daane 2004). Spiders known respond positively vegetation retreats, microclimatic conditions (Entling 2007; Kolb decreased treatments, S2). unsown parts covered spontaneous, sparse treatments. indicate localized area. absence indicates increased outside Contrary change involved general, quality. mixtures benefit would (Tschumi relatively long term providing (Frank Ganser Holland Here, spring during summer early autumn. already sites, enough promote establishment taxa. (pesticide use, tillage) filtered assemblages available pool, may some those resources, allowing increase. Thus, reveals benefits occasionally planting Future should address temporal diversity, (Gardarin ones. pattern opposite overall. below 50% above, inconsistent. landscape-complexity 2012) reflect spillover do (including largely study) against disturbance, suitable (Kolb Mestre Schmidt Tscharntke, More harbour early-season into explain (Hogg Daane, however, benefited limited. Several Haenke 2009; Jönsson Woltz Evidently, depends factors including farming practices, pedoclimatic contrasted Muneret modifying fields farm, cultivar, tends confounding effects, ability detect context. Californian showing flowering parasitism leafhoppers complexity, specific Shannon evenness indices systems aspects accounting arrangement landscape, patches. highlight relationships landscapes. observed associated Instead, (and lesser richness) suggest subset grape predator intra-guild behavioural abundant. responsible eggs, small regulation (Ostandie (2021) spider biomass juveniles families Thomisidae Araneidae Oxyopes sp. represented bulk Larger like harvestmen earwigs acted intraguild feeding disturbed actually (Papura Pinto Da Rocha Nature-based solutions increasingly efficiency real world scarce. Despite scope, insights managing ways conservation production. show diverse, turn damaging Europe, short season). detected (22–50% radius), ranged implications insecticide growers prioritize want maximize provides recommendations farmers policy-makers vineyards: driver diversity. partly funded Europe H2020 program Societal Challenges (H2020-SFS-2014, Grant No. 633999). B. research project SECBIVIT, through 2017–2018 Belmont Forum BiodivERsA call proposals, BiodivScen ERA-Net COFUND programme, funding organisations: AEI/Spain, BMBF/Germany, ANR/France, NWO/Netherlands, UEFISCDI/Romania, FWF/Austria (Grant I 4025-B32) NSF/USA #1850943). thank managers “Les vignerons de Tutiac” hosting Olivier Boissières facilitating. authors declare conflict interest. A. R., D. T. conceived ideas methodology. A., M. G. P. data. performed analyses led writing manuscript. contributed critically drafts gave approval publication. peer review history article https://publons.com/publon/10.1002/2688-8319.12086. codes accessed GitHub (https://github.com/leabeaumelle/EUCLID) archived Zenodo: https://zenodo.org/record/4916881#.YMOivahKg2w (Beaumelle Please note: publisher content functionality information supplied authors. Any queries (other missing content) directed corresponding author article.