Temporal and spatial heterogeneity of semi?natural habitat, but not crop diversity, is correlated with landscape pollinator richness

Agricultural expansion and loss of (semi-)natural habitats are major drivers pollinator declines (Potts et al., 2010), with associated threats to the pollination services these species provide wild plant populations (Clough 2014; Martins 2015; Pauw & Bond, 2011) crop yields (Fijen 2018; Sritongchuay 2020; Webber 2020). However, reversing trends by converting agricultural fields semi-natural habitat comes high opportunity costs (smaller surface productive land) that may not outweigh benefits increased productivity (Kleijn 2019). Increasing diversity mass-flowering crops is often raised as a promising strategy complement resources habitats, which ultimately could benefit biodiversity levels in landscapes (Fahrig 2011). Whether this approach work likely depends on capacity different sustain complementary diverse communities, or supplement characteristics habitats. we know surprisingly little about potential support rich communities (but see Raderschall 2021; Sirami 2019) make them so relevant for pollinators. For instance, main determinant can (diversity in) flowering boost adding new abundant resources? Does vary over space time? Do types each other partially species? Ecological theory suggests niche heterogeneity key driver coexistence therefore (Benton 2003; Chesson, 2000; Reverté Compared crops, generally much more heterogeneous, both within between landscapes, vary, example, composition plants availability nesting substrates (Williams Kremen, 2007). In addition spatial heterogeneity, same also markedly floral across growing season early senesce replaced later (CaraDonna 2017). This only imply host preferences phenologies (Mallinger 2016), but results continuity time generalist accept wide range (Schellhorn 2015). The relative role temporal resource supply virtually unexplored. A higher explain why than Nevertheless, even though inherently very limited insect-pollinated introduced into an landscape provides niche. mean richer similar without crops. Hence, promoting cultivation might represent insect landscapes. Indeed, single has been found increase abundance richness locally (Diekötter Holzschuh 2013) at scale (Beyer Westphal 2003). More recent studies have examined how one presence responses differing groups (Aguilera 2018). increases landscape-level compares contribution remains untested. Here, address questions using dataset occurring South-Italian two 15 entire season. study, evaluate importance sources promote To aim, sampled (i.e. 4 months). We monitored 26 along gradient increasing cover. specifically asked (a) contribute landscapes? (b) How important compared driving (c) cover moderate similarly adjacent crops? addressed resampling approach, analysed whether sampling scenarios (e.g. pollinators type vs. types; times landscape) would result significant differences cumulative numbers. mainly focused numbers rather densities accurately reflects pool enables better understanding site type. study was conducted Mediterranean located southern Italy, general region Fijen al. (2018). area covered approximately 1,400 km2 dominated wheat cultivation, many cultivated, such olive, faba bean chickpea food feed, onion leek seed production. total, (750 m radius) were selected based (0–8 per (from c. 0.2%–72% area; Table S1). Landscapes separated from 19 ± 18 km (mean 1 SD), except pair where borders slightly overlapped. radii above maximum foraging most solitary bees, 200–300 c.f. (Zurbuchen 2010). Therefore, decided keep analyses. centre 22 focal (18 hybrid production four [hybrid] fields). Four contained no quantified via Google earth aerial imagery initial classifications validated through field visits. Grasslands, woodlands, fallow arable road verges estimated side roads) considered herbaceous mostly annual ruderal plants, while woody usually forest edges hedgerows. abundantly transects consisted Brassicaceae Diplotaxis erucoides, Sinapis arvensis, S. alba), Malvaceae Malva sylvestris), Papaveraceae Papaver rhoeas, Fumaria officinalis), Asteraceae Anthemis tinctoria, A. arvensis) Boraginacea Echium plantagineum, Borago officinalis). Flowering Prunus spinosa, Crataegus monogyna several roses (Rosa sp.). Access kindly granted company Nunhems Netherlands BV (BASF). landscape, (bee syrphid pollinators) surveyed every 2 weeks blooming during (end March end July 2018), resulting eight rounds landscape. Standardized 150 long (150 m2) subdivided three 50 m2 subareas used, ensure distribution whole transect. Transects min pure observation excluding handling time). During surveys, visually recorded all observed interacting flowers. be recognized situ, caught butterfly nets identified morphospecies level laboratory. discarded individuals had (morpho) (c. 6% individuals). ascertained unidentified did influence because they few evenly distributed (Figure Surveys temperatures degrees Celsius, sunny calm days (<5 bft wind), roughly 8 a.m. 5 p.m. (c.f. Kleijn, Days surveys randomized require ethical approval Within areas round vegetation (including pioneer vegetation, grasslands vegetation), varying number shrubs trees). Herbaceous always containing flowers whenever present. achieved shifting, next, exact location up right left avoid sites depended varied 0 round. nine hence vegetation. round, when least 10% open small percentage large If there multiple fields. type, close fixed transect starting 20 edge field. moved, variation stage negligible. Overall, provided data 416 179 122 following (see S2). follows: basil Ocimum basilicum, broccoli Brassica oleracea var. italica, cauliflower oleracea, Cicer arietinum, dill Anethum graveolens, Vicia faba, Allium porrum, flax Linum usitatissimum, lucerne Medicago sativa, cepa, rucola Eruca vesicaria, sulla Hedysarum coronarium, sunflower Helianthus annuus, clover Trifolium sp. vetch All either used vegetable/herb onion), oil-seed flax), feed chickpea), yet harvested animal sulla, vetch). analyse data, approach. necessary our variables interest intensities replication. Not correcting variable largest sample size being strongly related richness, wider environmental sampled. contrast, method with, offset account fact samples probability finding decreases, lead underestimated sites. Our robust replacement allowed us correct still use extensive estimate confidence intervals infer significance. sets compare supported total mix 12 types, bean, frequently leek, late-flowering Mean estimates accumulated 95% obtained randomly 250 combinations (mixture). Twelve highest acceptable estimating means restricted 15. mixture selecting any thus maximizing effects diversity. rounds. Because beans flowered two/three seven/eight, resample analyses only. subsequently test differences, linear models, resampled response explanatory variable. present rarefaction curves visualize rates accumulation increases. influenced grown (n = 26), simple regression analysis. analysis, reasons. Sampling effort completely balanced 97% Furthermore, including bias transects. understand relationship diversity, additionally and/or examine temporally stable individual accumulate it sites). refer resources, respectively. Although stressors competition diseases spatial/temporal turnover, expected limiting factor, differentiating characteristic coexisting first pooled effort/precision. Then, calculated (26 points; landscape). random 10 averaged representative done run replicates dataset. fitted separate variable, source (temporal/spatial) model, tested due heterogeneity. 24) less balance 26). second round). Groups post-hoc Tukey tests. moderates shared nearby (throughout season). models variables, variables. Gaussian error structure chosen model fit performance residuals. checked outliers normal R (R Core Team, package stats (base R) dplyr manage (Wickham 2021), ggplot2 create graphs (Wickham, vegan (Oksanen 2020) spadeR count (Chao 2000). Code freely available corresponding section). 26,123 belonging 49 genera 372 complexes 717 S3 list). Sphaerophoria scripta complex (35% transects), Apis mellifera (32%), Andrena flavipes (25%), Lasioglossum villosulum/medinai (22%), Syritta pipiens (21%) Eristalis tenax (20%). consistently hosted cultivated or, perhaps interesting, combination (Table 1; Figure 1). furthermore species-rich alone combining (non-overlapping intervals; particularly attractive community 42.40 0.37 SE visited relatively 19.97 0.36 SE). Crop 2A; t24 ?0.092, p 0.927). 236 exclusively encountered 13 Crops 136 species, 90% Most visiting (1–3) 2B). honeybee ubiquitous managed farmers poor resources. Temporal (t48 0.946, 0.349; 3A). periods, mid-season (May) lower early/late 3B). five 0%–72% (t105 3.725, < 0.001, ? 0.059; 4A). line density (8.3% 4.9; SD) remained 0.705, 0.482, 0.015). These fairly (Tables S4 S5). Understanding what determine development agroecosystems determining richness. croplands seems depend identity addition, mixtures probably offer set suboptimal conditions disturbed ground pesticides). highlighting value throughout Lastly, suggest landscape-moderated driven supporting larger pools complexity gradient. Pollinator indicates characteristics, specific compose amount habitats) (Hass explanation do permanent cannot complete their life cycle need cavities) visit forage. subset makes Senapathi 2015), readily exploit become show exception constrained remain dependent foraging, reproduction shelter. flower morphology phenology identity). highly blooms June active nectar scarce (Timberlake Leek twice accessible can, therefore, best abundance, demonstrated (Raderschall and, indeed, some shown red oilseed rape (temporarily) abundances Riggi 2021 bumblebee diversity) 2020, 2003, 2003) providing (Holzschuh 2013). Ninety-seven percent highlights effectively exist temporarily attract proportion local pool, illustrated since bloomed. suggesting unique niches (Eeraerts Rivers-Moore include cavity (Rivers-Moore 2020), Semi-natural contain spatially spread out assessed commonly studied (number bee period season) rarely (one repeatedly just underestimate rely around Kleijn Importantly, turnover stronger here, (spatial heterogeneity) includes topographic phenological mismatches (Olliff-Yang Ackerly, conclusion, sites) reason cover, supports widely accepted view simplification jeopardizes (Kennedy well-known pernicious provision scale. Interestingly, nuance, simplified decent 11 patches naturalized landscapes). (stable) intensified conservation, acting refuges concentrate (Boetzl Li Martínez-Núñez Redhead Maintaining help conserving complexity, constant non-crop apparently goes against ecological specialist extinction fragmented (Harrison 2017; equal partly explained opportunistically 2019; species’ tolerance conversely, preference tolerance/preference proportionally concept intensification, shows simultaneously (Bommarco 2013; 2019), subsequent ecosystem (Morandin maintain Instead, heterogeneous Conservation habitat, restoration promotion set-aside maintaining helps gradients. (BASF) access assisted logistics study. thank anonymous reviewers constructive comments. None authors conflict interest. T.P.M.F. D.K. designed led project; T.P.M.F., C.M.-N. conceived ideas manuscript; wrote draft manuscript inputs D.K.; C.G., D.H., W.V. fieldwork; I.R. contributed significantly final version manuscript. Data code Figshare https://doi.org/10.6084/m9.figshare.19153214.v1 (Martinez-Nuñez 2022). Supinfo Please note: publisher responsible content functionality information supplied authors. Any queries (other missing content) should directed author article.