Complex landscapes stabilize farm bird communities and their expected ecosystem services

Birds play many roles within human societies, from voracious consumers of crippling crop pests, to haulers pathogens, symbols peace, love and strength (Karp et al., 2013; Smith 2020; Whelan 2008). are also rapidly declining, largely in response accelerating levels habitat loss land use intensification (Rosenberg 2019). The rapid birds is likely have severe consequences for the functioning coupled human-natural systems by compromising important ecosystem services (Echeverri 2021; Şekercioğlu 2004; (benefits that nature provides humans) disservices (harmful effects well-being) provide people varied. These include a variety supporting (e.g. biodiversity), regulating pathogen spread or disease regulation, insect control pollination), provisioning consumption crops) cultural spiritual enrichment positive negative aesthetics) services/disservices (Figure 1a; Echeverri 2004). This especially true agricultural where functional can either enhance reduce farmer livelihoods (Anderson Gardner 2011; Karp 2013). In turn, management impact bird communities, which could generate feedback loops alter net on broader society as whole (Pejchar 2018; Smith, Kennedy, Taylor, 2021). Management agroecosystems be highly contentious issue, impacting wide range stakeholders beyond individual farmers (Table S1) (Baur, Olimpi A notable example pressure exclude wildlife farms fear they may carry foodborne pathogens Shiga-toxin producing E. coli, Salmonella spp., Campylobacter spp.) versus diversify conservation environmental benefits (Beretti & Stuart, 2008; 2019; Edworthy, 2020). Farmers wish deter due their crops promote suppression vertebrate pests (Smith, Indeed, improve yields via pest but cause large losses through direct Kross 2012). average amount relative balance social implications shift across farming contexts (Olimpi stability avian effective farm management. One reason risk averse operate tight financial margins (Gong 2016; Liu Huang, For example, if occasionally damage crops, inconsistently key arthropods, then hedge bets, increasing pesticide application rates avoid potential an outbreak causing failure (Zhang 2018). Stability service help stabilize yields, allowing more consistent returns realized each year (Bommarco Thus, better understanding factors contributing temporal fluctuations disservice providers decisions ability effectively harness decrease yield fluctuations. Increasing species diversity and/or community evenness provider abundances, thereby delivering provisioning. because fluctuate asynchronously abundance over time, enhanced increases biomass portfolio effect (Blüthgen Ives Carpenter, 2007; Schindler 2015). Additionally, diverse communities contain at least few weather any given disturbance (i.e. diversity), well competitors assume declining density compensation; Hillebrand Blüthgen 2016). Similarly, complexity habitats, natural ecosystems agroecosystems, thought increase (Levin, 1992) reducing dominates unsuitable most (Schindler 2015; Tscharntke local, landscape stability. First, local scale, greater vertical vegetation retain (Heath 2017). Second, continuity resources (Smith That is, shocks caused tillage, harvest, pesticides other actions buffered diversified complex available nearby disturbed area (Tscharntke Third, scale shifting composition even, Here, we conducted point count surveys 106 locations 27 S1); classified using metrics spanning supporting, regulating, services/disservices; created ecosystem-service-and-disservice-weighted indices Table point-level complexity, overall value modifying High Nature Value index, per cent semi-natural cover landscape. We used these data ask (a) Do avian-mediated coincide independent? (b) What importance quantity service- provisioning? While has definitions (Ives 2007), here refer replicates 1b). (c) best joint outcomes dissuade disservices? Our central goal was identify aspects support multi-functional outcomes, considering both service/disservice indices. Across 4 years (2016–2019), surveyed total Oregon (n = 15) Washington 12; Figure 2; states, USA. obtained permission conduct all owners managers. All fell into Northern Pacific Rainforest Bird Conservation Region. Farms were diversified, organic practices (20 certified), grew (no monocultures; mean 46.5 ± 19.8 (SD) grown farm) including cereals, vegetables melons, fruits nuts, oilseed roots, spice beverage medicinal commercial flowers, grasses fodder among others. Livestock integrated operations 1 study 18 forms. spanned contexts, intensified agriculture primarily S1; range: 2.19%–95.7% semi-natural). twice between 20 May 8 August 2016–2019 with peak produce production region. moved along south north transect S1a) two annual survey periods. Survey one roughly corresponded nesting season transect, while fledging flocking periods gregarious species. location (‘point’) 100-m radius every ha farmed maintain density. Points systematically stratified capture uses present 2c,d; Point centres 200 m apart double counting individuals. total, points included this analysis (mean farm: 3.9 3.3 (SD); 1–14). At point, observer recorded number unique individuals seen heard during 10 min period. Surveys sunrise 10:45 a.m., only absence heavy rain, same skilled (OMS) eliminate biases differences. different order visit detection time-of-day If structures interfered visual detectability birds, see around (Šálek Because our traded geographic breadth within-season replication, unable account probability analyses. assumed constant period farms. conciseness, term ‘abundance’ article detected, it should noted missed research approval State University's Institutional Animal Care Use Committee (ACC protocol ASAF #04760). manuscript (which requires passive observation birds) weights estimates prevalence derived mist-netting reported previously press). calculated several provided S1). acknowledge proxies limitations, note proxy S1. (total detected survey) metric services. To estimate delivery (regulating disservice), generated index. do so, observed weighted its species' estimated spp. contacts/survey al. (in Briefly, press) 139 examining 11 traits predictive each. They best-supported models predict understudied analyses common found Snyder, contact score represents fields point. accounted rate will deposit pathogen, enter defecate repeated CCSmax, yielded similar results. Therefore, reader Tables S2–S5 S3 results structural ground herbaceous (0–0.5-m height class), low shrubs/crops (0.5–2 m), tall (2–6 m) trees (>6 10-m location's centre S4a). divided circles four equal quadrants cardinal directions (Kennedy 2010). During survey, vegetative class quadrants. averaged group strata. Vegetation occasion. Finally, ground, shrub, shrub tree eight (4 × 2 repetitions year) giving us values strata complexity. obtain single location, principal components ‘prcomp’ function ‘stats’ package r version 3.6.3 (R Core Team, standardized calculating z-score first (PCs) combined 67.5% variation S4b). PC1 (‘local complexity’; 41.0% variation) associated increased fullness layer. PC2 (‘ground cover’; 26.5% subsequent interested measured intensification/extensification, conservation-friendly practices, index (Pointereau 2010; brief, continuous (lowest value/most intensive) 30 (highest extensive). three sub-component (‘diversity crops’, ‘extensive/intensive practices’ ‘landscape elements’), get weight (10 max). highest ‘diversity crops’ indicator typically small high integrate livestock. inputs, certified organic, stocking densities elements’ incorporate elements farms, such hedges wet grassland. See Taylor , (2021) full details modification. Each had represent years. characterize context, based 2016 National Land Cover Database (Dewitz, 2019) 2.1 km buffer 2e,f) R FRAGSTATS 4.1 (McGarigal Marks, 1995; Semi-natural forest (deciduous, evergreen mixed), scrubland (dwarf scrub shrub/scrub), (grassland/herbaceous, sedge/herbaceous, lichens moss) wetland categories (woody emergent wetlands). Categories not water, ice/snow, developed, barren, pasture/hay cultivated classes. biologically relevant (Jackson Fahrig, 2015) home size label ‘average indices’). Then, coefficient variability. dividing standard deviation (CV σ/µ). inverse (1/CV) 2016), examined Pearson's pairwise correlations (Pearson's r) level. correlations, described scale. single, additive interactive series nested generalized linear mixed fit ‘glmmTMB’ (Brooks 2017) S6). Models binomial distributions (average indices) Gaussian (stability indices). Model assumptions checked ‘DHARMa’ (Hartig, determined optimal random global model set constructing combinations farm, (survey replicate 1–8). selected structure comparing AICc values. effects. ranked examine fixed ‘bbmle’ (Bolker, 2020) identified those supported (∆AICc < 2.0) (Burnham Anderson, 2002). assessed multicollinearity candidate ‘performance’ (Ludecke issue (VIF 5). covariate averaging (within ∆AICc model) ‘MuMIn’ (Barton, Burnham considered variables conditional 95% confidence intervals did overlap zero. suggested context promoting hypothesized away dominant, nomadic identity selection effect; 1c). above influence predictor examined. 15,684 111 (Data S1 Data sites European starling (Sturnus vulgaris 2,093), American robin Turdus migratorius 1,293) barn swallow Hirundo rustica 1,276). starling, violet-green Tachycineta thalassina 27) once, 26 fewer ecosystem-service/disservice-weighted often correlated 3). Within 4) metrics, there mixture 25) 39) Between indices, benefitting positively higher grower appreciation 0.76) 0.79). disapproval 0.70) iconic 0.88), 0.74). damaging generally 0.42). exhibited patterns However, correlation 0.51) compared indices' 0.19). weak relationships (point-level) 4a; 1; Figures S5 S6; S7–S22). Farm-level scores were, however, strong predictors abundances 4b; S7 S8), (Tables S9 S10), S11 S12), S13 S14) S15 S16). strongly S17 S18), S19 S20) S21 S22). Increased decreases indices: food safety risk, 4c; S15–S20). landscapes reduced S12). No 4; Local good none 4d; S8; S23–S38). S25 S26), S27 S28) S29 S30). farm-wide 4e) exception S28). interaction mediating 5; S23 S25). when Value, benefit diminished Value. 4f; S29–S36). S7; S37 S38). S39 S40; S9). contrast, Greater locations, strong, 6; S41). sought how managed stable outcomes. Assessments costs biodiversity done isolation (Peisley limits holistically manage meet multi-stakeholder needs opposition Collectively, suggest diversification without 4b,e). needed suppress level Contention risen recent decades whether attract On hand, concerned illness economic Baur, others advocate conserving decline some want leverage findings themselves responses 1) able disservices. responded similarly examined, (but differently 4). distinct path forward towards holistic agroecosystem impacts contrasts prior work California showed dense field abundance, richness PCA approach distinguish types cover, varying guilds. did, detect stabilizing beneficial suggests wishing biocontrol planting vulnerable arthropod rodent alongside attractive birds. systems, hedges, limiting efficacy ecosystem-service-, disservice-, larger 4b). experience beneficial, harmful, disruption intermediate hypothesis, posits effectiveness structurally simple rather than cleared landscapes, explain finding rotationality livestock 0.48). Crop-livestock decreased nonnative Nonnative native life-history daily movements ephemeral, tracking throughout cycle (Billerman Fischl Caccamise, 1985). destabilize like starlings ephemeral lower ecosystem-disservice-weighted changes align demonstrating embedded strawberry Other system demonstrated risks stably concentrations risky 4c,f). Landscape 4f). impacted variable income fluctuating affecting makes easier reliably accurately plan accentuate birds' mitigate argued (Benton 2003). identity/iconic seemingly counterintuitive. It shifts less flocks blackbirds S10). general public values, differ conservationists possible prefer abundant peoples' familiarity frequent interactions 2017; Gaston, Gaston result limitation ‘celebrity species’ database (Schuetz Johnston, 2019), Google searches fail attitudes (positive negative) 6, There exists relationship (MacArthur, 1955). 2011), asynchrony dynamics selection/identity wherein driven dominant (Hillebrand would strategies policies restoration minimize Although realize 2013), remain poorly adopted world-wide (Pretty 2018), indicating stronger policy incentives needed. Yet, 40% countries globally requirement maintaining habitats working landscapes—despite recognition doing so urgently achieve national commitments UN Convention Biological Diversity Sustainable Development Goals (Garibaldi widely adopted, place certain regions landscape-scale services, landscape-wide implementation agri-environment schemes collaborative Common Agricultural Policy Europe (Dallimer Santos incentivized adopt enrolled private-sector NGO-led eco-certification programs Audubon Certified Grazed BirdFriendly Land) enable commodifies sold price premium (Biggs Several limitations interpreting study. quantifications, (detailed Here highlight shortcomings. vary depending species-specific rates; body size; type vulnerability bird, damage; seasonal Pejchar granivorous provision young insects thus chick rearing consuming later placement diet guilds ignores intraguild predation focused fully extend conventional, large-scale orchard vineyard monocultures), regions. particular, fields, mechanization use) specialized Gonthier specialized, monocultural variance (or invulnerable) single-crop damage. fruit large, passerine coffee producers indirect although scales study, hard them separately. North America, declines concerning compromise vital control) conservation) evidence whole-scale paddock live fencerows disservices, delivery, appear landscape-level features relatively unaffected farm-level diversification. amounts need planning maintenance scales. grateful growers who allowed access took time out incredibly busy schedules facilitate research, esp. V. Alexander. USDA-NIFA-OREI grant 2015-51300-24155 awarded W.E.S., J.P.O., C.M.K. E.E.W-R., USDA NIFA Predoctoral Fellowship 2016-04538 O.M.S., Postdoctoral 2021-67012-35133 Carl H. Elling Endowment School Sciences O.M.S. Any opinions, findings, conclusions recommendations expressed publication authors necessarily reflect view U.S. Department Agriculture. C. Blubaugh H.S. Burns assisted project development. G.H. Ow, A. Tormanen M. Edworthy logistical support. S. Knutie federal master banding permit made assessment possible. thank reviewers (one anonymous self-revealed, Matthias Tschumi) significantly improved quality. declare no conflict interest. conceived ideas; collected count, classification data, analysed figures led writing manuscript; J.M.T. A.E. grower-focused data. contributed critically drafts gave final publication. Dryad Digital Repository https://doi.org/10.5061/dryad.wm37pvmpn Please note: publisher responsible content functionality information supplied authors. queries (other missing content) directed corresponding author article.