Farmland bird diversity requires heterogeneity between and within habitats

Farmland bird diversity requires heterogeneity between and within habitats | Research Square window.SnipcartSettings = { analytics: { enabled: false } }; (function() { var accessVector = localStorage.getItem('access_vector') || ''; window.dataLayer = window.dataLayer || []; if (accessVector) { window.dataLayer.push({ user: { profile: { profileInfo: { snid: accessVector } } } }); } })(); (function(w,d,s,l,i){w[l]=w[l]||[];w[l].push({'gtm.start':new Date().getTime(),event:'gtm.js'});var f=d.getElementsByTagName(s)[0],j=d.createElement(s),dl=l!='dataLayer'?'&l='+l:'';j.async=true;j.src='https://www.googletagmanager.com/gtm.js?id='+i+dl;f.parentNode.insertBefore(j,f);})(window,document,'script','dataLayer','GTM-K279D39R'); Browse Preprints In Review Journals COVID-19 Preprints AJE Video Bytes Research Tools Research Promotion AJE Professional Editing AJE Rubriq About Preprint Platform In Review Editorial Policies Our Team Advisory Board Help Center Sign In Submit a Preprint Cite Share Download PDF Research Article Farmland bird diversity requires heterogeneity between and within habitats Marit Kinga Kasten, Thomas Hiller, Sara Tassoni, Rosalie Böhmer, and 4 more This is a preprint; it has not been peer reviewed by a journal. https://doi.org/ 10.21203/rs.3.rs-9368528/v1 This work is licensed under a CC BY 4.0 License Status: Under Review Version 1 posted 7 You are reading this latest preprint version Abstract Context Birds are declining worldwide, with farmland birds disproportionately affected. Most studies on farmland birds focus on single habitat types, yet agriculturally dominated landscapes are mosaics composed of multiple habitat types like arable land, grassland, forests, and orchards. Objectives We aimed to understand how these habitat types jointly shape farmland bird diversity, particularly regarding local and landscape drivers of alpha and beta diversity. Methods We used passive acoustic monitoring to survey farmland bird communities in 14 mosaic agricultural landscapes (1 km²) in southern Germany that differ in habitat diversity. In total, 224 autonomous recording units were deployed in a grid-based design with sampling intensity proportional to habitat area. Using BirdNET and manual validation, we identified 54 bird species from 2,016 hours of recordings collected over 4.5 months. Results Local species richness (alpha diversity) increased with habitat heterogeneity at both local and landscape scales. Arable sites showed the lowest alpha diversity but comparatively high within-habitat beta diversity, whereas orchards supported high alpha but low within-habitat beta diversity. Beta diversity was highest between habitat types, especially between forests and arable land, reflecting strong contrasts in their structural complexity. Generalized dissimilarity modelling showed that local predictors were more important than landscape-level predictors in explaining bird beta diversity. Habitat associations of bird species were largely consistent with ecological expectations: bird species adapted to dense vegetation occurred mainly in forest-dominated sites, while open-habitat species were associated with arable land. Species with decreasing population trends occurred across all major habitat types. At the landscape scale, gamma diversity increased strongly with landscape diversity. Conclusions Maintaining habitat heterogeneity at multiple spatial scales is critical to conserve farmland bird diversity. passive acoustic monitoring (PAM) habitat heterogeneity biodiversity automated species identification orchard meadow agroecosystems Figures Figure 1 Figure 2 Figure 3 Figure 4 Figure 5 Figure 6 Introduction Across Europe, populations of common bird species have declined by about 25% between 1980 and 2016 (Rigal et al. 2023 ). Farmland birds exhibited even higher losses with an average decline of 56% (Rigal et al. 2023 ). These declines have been attributed to agricultural intensification and changes in the structure of agricultural landscapes (Sirami et al. 2019 ; Rigal et al. 2023 ). As structurally diverse mosaic landscapes were replaced by simplified, agriculturally intensified landscapes, key habitats for many farmland bird species diminished. However, despite extensive research (Sirami et al. 2019 ; Rigal et al. 2023 ), we still lack larger-scale assessments that jointly analyze local (alpha) and compositional (beta) diversity of farmland bird communities (Fig. 1 ) across spatial scales (Cours and Duflot 2025 ). Biodiversity patterns are shaped by ecological processes operating across multiple spatial scales. At the local scale, species richness and composition of bird assemblages is primarily determined by habitat-specific resource availability and structural heterogeneity (Fahrig 2013 ). Habitats rich in woody elements, such as forests or traditional orchards, typically support species associated with semi-open or forested environments. Arable fields, however, favour species adapted to open landscapes. We expect orchards to harbour more species with decreasing population trends than arable land, as in arable land these endangered species could have already disappeared. Notably, local alpha diversity of bird assemblages is also shaped by the broader landscape context. Landscapes with higher compositional heterogeneity can support richer local bird assemblages (Benton et al. 2003 ), because heterogeneous surroundings increase the pool of species which are able to exploit resources across multiple habitat types (habitat complementation; Dunning et al. 1992 ). The species-specific associations with habitat resources and structural features imply that contrasts among habitat types (e.g. forest vs. farmland) strongly influence whether two local sites host similar or distinct bird assemblages. Such habitat-driven differences give rise to beta diversity, i.e. the compositional turnover of species between sites. While alpha diversity of farmland birds across different habitat types is relatively well studied, beta diversity within and between habitat types is rarely studied (Edo et al. 2024 ; Cours and Duflot 2025 ). Sites belonging to the same habitat type can, for example, differ in the quantity and diversity of resources or structural elements they provide due to varying degrees of land-use intensity (Benton et al. 2003 ), creating within-habitat beta diversity (Fig. 1 ). These differences in structure and management should be more pronounced between habitat types, leading to more pronounced values of beta diversity between habitat types (between-habitat beta diversity, Fig. 1 ). Beyond habitat characteristics, spatial processes such as distance decay can further increase beta diversity by limiting dispersal or reducing community similarity with increasing geographic separation (Dambros et al. 2017 ). Overall, however, the relative contributions of between-habitat and within-habitat beta diversity in structurally complex agricultural landscapes remain poorly understood (Fig. 1 ). At the landscape scale, gamma diversity (Fig. 1 ) emerges from the interplay of local alpha diversity and turnover among sites (beta diversity). The habitat diversity hypothesis assumes that more heterogeneous landscapes support higher gamma diversity because they offer a wider range of ecological niches (Benton et al. 2003 ; Duflot et al. 2022 ; Valente et al. 2023 ). Diverse landscapes containing mixtures of forests, grasslands, orchards, and croplands can provide complementary resources that support a broader set of bird species than simplified landscapes. When aggregated regionally, such landscape-level diversity patterns determine how much differently composed landscapes contribute to the overall species pool. In this study, we investigated farmland bird communities across agricultural landscapes of varying structural complexity in southern Germany. To characterize bird communities at the landscape scale, we employed passive acoustic monitoring combined with automatic species recognition using BirdNET. Our aim was to identify the key local and landscape-scale drivers of alpha and beta diversity in farmland bird communities (Fig. 1 ), thereby improving our understanding of how agricultural landscape structure shapes biodiversity across spatial scales. Methods Landscape selection and landscape composition We conducted our study in an agricultural region (20 x 20 km²) south of Stuttgart, Germany (48°46′39″N 09°10′48″E; Fig. 2 a). The landscapes in our study region consist mainly of arable land (predominantly for cereal cultivation), forest (predominantly beech forest), grassland (including meadows and pastures), and orchard (predominantly traditional, extensively managed apple orchards) (Statistisches Landesamt Baden-Württemberg 2024 ; Cullmann 2025 ; Streuobstportal Baden-Württemberg 2025 ). We therefore focused our bird surveys on these four major habitat types (Fig. 1 , Fig. 2 ). In spring 2023, we identified 14 study landscapes, each measuring 1 x 1 km 2 , with the aim of covering a gradient of landscape diversity. Diverse landscapes had a balanced ratio of arable land, grassland, forest and orchards, while simplified landscapes consisted mainly of arable land (Fig. 2 c). On average, 5.1% of the landscapes’ areas were organically managed arable land (minimum: 0.2%, maximum: 21.9%) and 2.1% organically managed grassland (minimum: 0.1%, maximum: 6.6%). Organic management did not correlate with landscape diversity (Pearson’s r = – 0.522). Overall, our study landscapes covered the regional range in terms of land cover. The mean pairwise distance between the centres of the landscapes was 10.5 km ± 4.7 km (SD). The classification of habitat types as grassland (defined as permanent grassland > 5 years) or arable land was based on data from the state of Baden-Württemberg (database for agricultural activities at field level from the Integrated Administration and Control System; Metainformationssystem GDI-BW 2023 ). Traditional orchard meadows were classified based on the state-wide orchard survey (Borngraeber et al. 2020 ). These are characterised by sparsely arranged, often older standard trees with grass cover in the undergrowth and made up most of the orchards in our study region. In contrast, intensively managed orchards were only present in one landscape. Other land cover types, including forests, settlements and roads, were mapped manually using QGIS v.3.36 (Fig. 2 b). Although Baden-Württemberg has densely populated areas, settlements covered only small areas in our rural study landscapes (mean ± SD: 2.5% ± 2%). During fieldwork, we verified the GIS-based habitat classifications and mapped hedges. We calculated landscape diversity as the Shannon diversity index for the final seven land cover types (farmland, forest, grassland, hedges, orchards, roads, settlements) using their proportional coverage as weights (Fig. 2 c). Recording locations and local habitat features We used a grid-based sampling design (Scherber et al. 2019 ) to ensure representative data collection across the four major habitat types (i.e. arable land, grassland, forest, orchard) within the study landscapes. The number of recording locations per habitat type corresponded to its proportional cover in each landscape. To this end, we set up 16 recording locations in each study landscape (224 in total), which were spatially distributed as evenly as possible at intervals of approximately 200 m (Fig. 2 b). We assigned each recording location to the main habitat type which dominated within a 100 m radius (Table S1 ), as this approximates the effective detection range of our acoustic devices (Buchmann and Schurr 2025 ). By using this grid-based sampling design, the resulting number of locations associated to a main habitat type closely matched the habitat proportions in the study landscapes, ensuring that our sampling reliably reflected landscape-wide habitat composition (Figure S1 ). We removed the three recording locations assigned to the habitat type 'settlement' from all analyses due to the small sample size, leaving us with 221 recording locations. However, to keep comparable numbers of recording locations for each landscape, we used all 224 recording locations only in the calculations of gamma diversity. Additional to the main habitat type, we also assessed the local habitat diversity around the recording locations as the Shannon diversity of the seven land cover types in a 100 m radius around the recording locations (analogously to the calculation of landscape diversity, but at a different scale). Furthermore, we counted the number of tree individuals in a 20 m radius around the recording location. c) Bird recording Due to the large number of recording locations, we decided to use passive acoustic monitoring using AudioMoth recorders (Hill et al. 2019 ). This monitoring method has been proven to provide results of equal accuracy and reliability as traditional point counts (Darras et al. 2018 ; Toenies and Rich 2021 ) while requiring much less workforce. We conducted our monitoring across three rounds of recordings in 2023 (23 March to 17 April; 15 May to 14 June; 3 July to 4 August). In each round, every landscape was recorded on two consecutive mornings for three hours each (starting 30 minutes before sunrise). All recorders were set up on poles at 1.5 m above ground facing eastward. We configured the AudioMoths to record sound up to a sampling rate of 192 kHz. Within each three-hour recording window, we recorded bird for 30 seconds, followed by a recording pause of 30 s. In the pooled data set, we consequently had 1,989 recorded hours. In case of bad weather conditions (strong wind or rain), we repeated recordings on the following days. Automated species classification with BirdNET and human validation To classify the bird species, we employed the artificial neural network ‘BirdNET’. Species classification with BirdNET requires consecutive human validation (Wood and Kahl 2024 ), leading to a multi-step process. In the first step, we analysed the recordings in BirdNET Analyzer (version 2.4.0). Each 30-s recording was subdivided into 3-s snippets with 2-s overlap, yielding 28 snippets per recording to maximize detections (Pérez-Granados et al. 2025 ). We then ran the BirdNET algorithm on all snippets using default settings. The output consisted of detected species and associated confidence scores (0.10–0.99) for each snippet. Across all snippets, this resulted in 6,076,185 detected bird calls. Second, we refined the automatic BirdNET species identifications by filtering them against a region-specific species list. Based on the geographic coordinates of the four corner points (NW, NE, SW, SE) and the centre of the study area, we queried BirdNET to generate species lists for each location. These four lists were merged into a comprehensive regional species pool. Three species were removed because they are extremely rare or not yet recorded in the region: Cettia cetti Temminck, Ptyonoprogne rupestris Scopoli, Pyrrhocorax graculus L.. The resulting list contained 168 species. Because water bodies were largely absent from our study landscapes and aquatic birds were not the focus of this study, we further excluded waterbird species. This resulted in a final list of 103 species considered potentially present in our dataset. In a third step, the BirdNET identifications of the 103 species required validation, as BirdNET confidence scores do not represent true probabilities of correct identifications and are not comparable across species (Wood and Kahl 2024 ). For each species, a random subset of 50 segments spanning the full range of BirdNET confidence scores from 0.10 to 0.99 was selected using BirdNET’s segments function (Wood and Kahl 2024 ). These 50 snippets per species were then assessed aurally and visually to verify BirdNET’s identification and assigned a value of 1 for a correct identification or 0 for an incorrect one. Subsequently, a binomial generalized linear model was fitted for each species, with “correctness” as the dependent variable and the logit score as the explanatory variable (see Supplementary Figure for an example model). The logit score represents the original output of the deep neural network (Wood and Kahl 2024 ) and allowed the model to perform better than using the confidence score. From each model, the logit score corresponding to a 90% probability of correct identification was derived and back-transformed to the confidence-score scale provided in the BirdNET outputs. Only BirdNET identifications exceeding the species-specific 90% probability threshold were retained for further analysis, which reduced the dataset to 66 species. A list of all validated species and their species-specific confidence thresholds is provided in Table S1 . Finally, to ensure that only species using the habitat at each location were included, we removed species that were likely only flying over. We retained a species at a given location only if it was validated on two consecutive days in at least one of the three recording rounds. This filter reduced the dataset by nine species, leaving us with 57 species. We then merged Corvus corone , Corvus cornix , and Corvus frugilegus into Corvus spp., and Passer domesticus and Passer montanus into Passer spp., because these species cannot be reliably distinguished by audio recordings alone (Buchmann and Schurr 2025 ). The final dataset comprised 54 species or species groups (see Table S2). Statistical analysis Alpha diversity We quantified alpha diversity as the number of species detected at each recording location (Fig. 1 ). To assess whether local- and landscape-scale factors jointly shaped alpha diversity, we fitted a generalized linear mixed-effects model with a Poisson error distribution using R package lme4 (Bates et al. 2015 ; R Core Team 2024 ). The model included all two-way interactions between local variables (number of trees within 20 m radius, main habitat type, and local habitat diversity) and the landscape-level predictor (landscape diversity). All continuous variables were scaled and centred to account for differences in magnitude. Because recording locations were nested within landscapes, we included landscape identity as a random effect. Significant predictors were identified through stepwise backward model selection using chi-squared tests. The R 2 values for all statistical models were calculated using R package performance (Lüdecke et al. 2021 ). Between-habitat and within-habitat beta diversity To assess differences in the composition of birds, we focused on the 16 recording locations within each landscape. These differences, or beta diversity, can arise both within a single habitat type and between different habitat types (Fig. 1 ). We quantified within-habitat beta diversity (between pairs of locations of the same habitat type, e.g. orchard–orchard) and between-habitat beta diversity (between pairs of locations of different habitat types, e.g. arable–forest) by calculating Sørensen dissimilarity indices for all pairs of recording locations within the same landscape. Each pair was then assigned both to its specific habitat combination (e.g. arable–arable, forest–orchard) and to the corresponding category (within- or between-habitat). Based on the categorization (within- or between-habitat), we could then assess whether between-habitat beta diversity was generally larger or lower than within-habitat beta diversity. To this end, we fitted a linear mixed-effects model from R package lme4 (Bates et al. 2015 ) with Sørensen dissimilarities as response variable and the categorical variable (within- or between-habitat category) as explanatory variable. To avoid pseudoreplication, we also included the additive random effects of the IDs of both recording locations involved in the comparison. F-test and Post-Hoc Tukey test (R package emmeans; Lenth 2024 ) revealed that within-habitat beta diversity values were significantly different from between-habitat beta diversity values. We also examined which habitat combinations exhibited relatively high or low dissimilarity. Due to the significant differences between categories, we fitted two separate models: one for within-habitat and one for between-habitat beta diversity. Habitat combination was used as the explanatory variable in both models. As before, we included the additive random effects of the IDs of both recording locations involved in the comparison. An F-test against the null model assessed whether Sørensen dissimilarities differed significantly among habitat combinations. Significant differences were further evaluated using a post-hoc Tukey test for pairwise comparisons. To visualize patterns, we also performed a non-metric multidimensional scaling (NMDS) analysis based on 221 recording locations assigned to arable land, forest, grassland, and orchard as main habitats. We additionally examined the influence of local and landscape variables on beta diversity at the scale of individual recording locations using a generalized dissimilarity model (GDM). The GDM quantified how environmental predictors explained Sørensen dissimilarities. Predictors included geographic coordinates of recording locations and landscape centres (UTM), local-scale variables (number of trees in 20 m radius; arable, forest, grassland, orchard, hedge, and settlement cover within 100 m; local habitat diversity), and landscape-scale area of all seven land cover types as well as landscape diversity. We fitted the GDM using five I-splines and included geographic information directly in the model. Variable importance was assessed with gdm.varImp from the R package gdm (Fitzpatrick et al. 2025 ). For significant variables, we calculated percentage contributions for representation in a stacked plot. Gamma diversity We quantified gamma diversity as the total number of species detected per study landscape (Fig. 1 ) and examined whether landscape diversity influenced it. For this analysis, we fitted a generalized linear model with a Poisson error distribution and applied stepwise backward model selection based on Chi-squared tests. We also evaluated landscape-level sampling completeness using species accumulation and extrapolation curves (R package iNEXT ). Sampling completeness was calculated as the ratio of observed bird species richness to the estimated asymptotic richness (Chao 1984 ) and subsequently correlated with landscape diversity. Functional traits and population trends of bird species In addition to analyses of overall species richness, we examined how species with different ecological traits were associated with habitat types. To this end, we conducted a non-metric multidimensional scaling (NMDS) analysis based on species identities and trait information from AVONET (Tobias et al. 2022 ). Traits included habitat density preference (open, semi-open, dense; AVONET variable Habitat.Density ) and trophic niche (granivore, invertivore, omnivore, vertivore; AVONET variable Trophic.Niche ). Additionally, long-term population trends (decreasing, stable, increasing) were derived from Annex B of the German National Bird Protection Report (2019), where species classified as decreasing or increasing had lost or gained more than 20% of their population over the preceding 36 years (Table S2). We assessed the influence of the same environmental predictors used in the generalized dissimilarity modelling (GDM; see above) and retained only predictors with significant effects. Results General results We investigated 2,016 h of audio recordings over a period of 4.5 months, in which we detected 54 bird species in total (species richness at a regional level). Fourteen of them occurred at > 50% of the recording locations, while 17 occurred at < 10% of the recording locations (Table S2). The most widespread species were Corvus spp. (at 220 recording locations), Turdus merula L. (at 219), Parus major L. (at 211) and Cyanistes caeruleus L. (at 200). Alpha diversity Alpha diversity represents the local species richness per recording location in a landscape (Fig. 1 ) and ranged between 6 and 30 species (mean ± SD = 18.2 ± 5.1). Alpha diversity was driven by additive effects of the type of the main habitat (X 2 3 = 19.778, P < 0.001), the local habitat diversity (X 2 1 = 11.034, P < 0.001) and landscape diversity (X 2 1 = 10.036, P = 0.002). Orchards had the highest alpha diversity (predicted mean = 20.7 species; 95% confidence interval: 18.9–22.6 species), while alpha diversity was lowest for main habitat type arable land (16.5; 15.5–17.7; Fig. 3 a). Forest and grassland showed intermediate alpha diversity (17.9; 16.3–19.5 and 19.4; 17.7–21.2 respectively). Alpha diversity increased with local habitat diversity and landscape diversity, respectively (Fig. 3 b + c). Local tree number and interactions between local and landscape variables were not significant. The model explained 39.4% of the variance based on fixed effects (marginal R² = 0.394) and 45.4% when including the random landscape effect (conditional R² = 0.454). Within-habitat and between-habitat beta diversity Within-habitat beta diversity showed on average a Sørensen dissimilarity of 0.297 (min = 0.053; max = 0.667; SD = 0.110). The lowest within-habitat dissimilarity values were present for within-orchard and within-forest beta diversity (predicted means ± SE = 0.232 ± 0.019 and 0.243 ± 0.014, respectively; Fig. 4 b; see also smallest ellipses for orchard and forest in NMDS Fig. 4 a). Contrarily, within-arable showed highest dissimilarities (predicted means ± SE = 0.312 ± 0.009; Fig. 4 b; see also largest ellipses for arable land in NMDS Fig. 4 a). Within-grassland beta diversity was not different from both extremes (predicted means ± SE = 0.279 ± 0.020). The fixed effects in this model explained 8.3% of the variance in within-habitat beta diversity (marginal R 2 = 0.083), which increased to 53.4% after accounting for the random effects of the recording locations (conditional R² = 0.534). Between-habitat beta diversity was generally higher than within-habitat beta diversity, with an average Sørensen dissimilarity of 0.333 (min = 0.04; max = 0.697; SD = 0.110; Figs. 4 b + c). Highest between-habitat dissimilarities were found for arable-forest comparisons and forest-orchard comparisons (predicted means ± SE = 0.353 ± 0.013 and 0.382 ± 0.013 respectively; Fig. 4 c; see also least overlapping ellipses in NMDS Fig. 4 a). Here, the fixed effects explained 9.9% of the variance in between-habitat beta diversity (marginal R 2 = 0.099), which increased to 67.1% after accounting for the random effects of the recording locations (conditional R² = 0.671). The generalized dissimilarity modelling (GDM) revealed that all local and landscape variables considered in this study jointly explained 41.2% of the pairwise beta-diversity of bird assemblages of each sampling location. However, there were only four significant variables explaining the beta-diversity patterns. These four predictors account together still for 40.9% of the explained deviance. Most of the explainable deviance was thus driven by local forest cover (60.8%), followed by local arable land cover (19.3%), landscape-wide orchard cover (12.5%) and landscape-wide forest cover (7.4%; Fig. 4 d). By contrast, the geographic distance between sites and all other environmental local- and landscape-scale explanatory variables considered did not significantly explain pairwise beta-diversity. Gamma diversity We measured gamma diversity as the total species richness across all 16 recording locations in a landscape (Fig. 1 ). Gamma diversity ranged from 25 to 45 species (mean ± SD = 37.8 ± 6.0). We sampled on average 87.01% (SD: ± 9.61%, min: 71.53%, max: 99.42%) of the landscapes’ estimated total richness. Sampling completeness of the bird species did not correlate with landscape diversity (Pearson’s r = − 0.30, P = 0.300). However, gamma diversity increased with increasing landscape diversity (estimate ± SE = 0.565 ± 0.179; X 2 1 = 10.326, P = 0.001; Nagelkerke’s R 2 = 0.857; Fig. 5 ). Bird preferences match habitat types The NMDS ordination revealed consistent patterns in species’ trophic niches, habitat-density preferences, and long-term population trends (Fig. 6 ). Omnivorous and invertivorous species showed no clear separation and occurred across all habitat types. In contrast, granivores clustered toward sites associated with open habitats, opposite to forested locations in the ordination space. Species preferring dense habitats were concentrated at forest-dominated sites with high tree cover, whereas open-habitat species grouped with arable and grassland sites on the opposite side of the NMDS. Species associated with semi-open habitats occupied intermediate positions between these extremes. Most species in our dataset exhibited stable long-term population trends, while ten showed increasing trends and seven decreasing trends. Bird species with decreasing population trends (i.e. more than 20% decline in the last 36 years) were scattered broadly across the ordination, indicating that no single habitat type uniquely supports them. Discussion In this study, we aimed to disentangle the roles of local- and landscape-scale drivers of farmland bird diversity in Southern German agricultural landscapes. We used automated acoustic monitoring at 224 recording locations across 14 landscapes and analysed local alpha diversity at each location, within- and between-habitat beta diversity across recording locations and landscape-scale gamma diversity. Local alpha diversity was lowest in arable land (predicted mean = 16.5 species) and highest in orchards (20.7 species). In contrast, within-habitat beta diversity was highest in arable land and grassland and lowest in orchards and forest. Between-habitat beta diversity exceeded within-habitat beta diversity, with stronger differences between structurally contrasting habitats (e.g. arable vs. forest) than between structurally more similar ones (e.g. arable vs. grassland). Consistent with these patterns, species preferring dense habitats occurred mainly in forest-dominated locations, whereas species preferring open habitats were associated with arable-dominated sites. Species with decreasing long-term population trends occurred across all habitat types. Generalised dissimilarity modelling indicated that local-scale variables (e.g. forest cover within a 100 m radius) were more important predictors of beta diversity than landscape-scale variables (e.g. total forest cover per 1 km²). At the landscape scale, overall bird richness (gamma diversity) increased with landscape diversity. Alpha diversity of bird assemblages Across habitat types, we observed contrasts in local species richness that reflect differences in management intensity, structural complexity, and resource availability. For example, orchards had higher bird species richness than arable land (Fig. 3 ) which aligns with previous studies (Wilson et al. 2020 ; Edo et al. 2024 ). These contrasting patterns in species richness between habitat types highlight fundamental differences between intensively managed, structurally simple habitats and more heterogeneous systems with high niche availability. Most of the species detected at more than 90% of the arable locations were widespread generalists, such as the greenfinch ( Chloris chloris (L.)), crows ( Corvus spp.), the great tit ( Parus major L.), and the common blackbird ( Turdus merula L.) (Table S2). This may be because arable land in our study region is predominantly managed conventionally (11% of arable land was farmed organically). Conventional management can reduce the availability of food resources (e.g. insects) and cause frequent disturbance in homogeneous fields which offer few suitable breeding sites (Benton et al. 2003 ). Consequently, the species that are still present in arable land, need to be able to cope with these conditions, favouring omnivorous, granivorous and highly mobile species. In contrast, traditional orchards are extensively managed and provide a wide variety of niches, especially through old trees that are valuable for cavity-nesting species (Edo et al. 2024 ). This high niche availability can also favour insect diversity and abundance (Sattler et al. 2024 ), enabling a higher share of invertivorous bird species in orchards. Consequently, a broad range of species were recorded primarily in orchards, including short-toed treecreeper ( Certhia brachydactyla Brehm), European goldfinch ( Carduelis carduelis (L.)), great spotted woodpecker ( Dendrocopos major (L.)), common nightingale ( Luscinia megarhynchos Brehm), common redstart (Phoenicurus phoenicurus (L.)) and common starling ( Sturnus vulgaris L.) (Table S2). This high overall richness of orchards underlines both their high conservation value and their vulnerability, as traditional orchards are classified as endangered according to the Red List of Biotope Types in Baden-Württemberg (Breunig et al. 2020 ). Grassland and forest sites showed intermediate richness levels (Fig. 3 ) between those of arable land and orchards. Grasslands showing lower bird richness than orchards is consistent with previous research (Hartel et al. 2014 ) and likely reflects the relatively intensive management of grasslands in the region, including frequent mowing and high fertilization. Such management has contributed to declines in formerly common open-land species such as the Northern Lapwing Vanellus vanellus (McKeever 2003 ). Opposite to Edo et al. ( 2024 ), we found that forests showed lower alpha diversity than orchards, as traditional orchards were structurally more diverse than forests in our study landscapes. This structural diversity leads to a higher availability of different niches, profiting both species of open landscapes and light woodlands and thus promoting a higher overall species richness (Edo et al. 2024 ). In our study landscapes, forests are often intensively managed but mostly mixed forests that can support many invertivorous and omnivorous forest-associated species, including the common treecreeper Certhia familiaris L., the wood pigeon Columba palumbus L., black woodpecker Dryocopus martius L., the robin Erithacus rubecula L., and song thrush Turdus philomelos Brehm (Table S2). Beyond the effect of the locally dominant habitat type, both local habitat diversity within a 100-m radius and overall landscape diversity within the 1 × 1 km 2 area increased bird alpha diversity (Fig. 3 ). Some studies support this finding (Barbaro et al. 2005 ; Anderle et al. 2022 ), although most studies could not confirm this relationship (Cours and Duflot 2025 ). Our findings align with the cross-habitat spillover hypothesis in heterogeneous landscapes (Tscharntke et al. 2012 ). At the local scale, birds may benefit from easier access to complementary resources from neighbouring habitat types, such as insects, fruits, or seeds, as well as from a broader range of nesting opportunities (i.e. habitat complementation, Dunning et al. 1992 ). For example, Skylarks ( Alauda arvensis ) may nest in arable fields but forage in adjacent grasslands, while Great Tits ( Parus major ) breed in forest patches yet feed in nearby orchards or hedgerows. Such local habitat heterogeneity can therefore enhance bird richness even in otherwise intensively used farmland, for instance when small woody features or orchard trees mitigate the structural simplicity of arable land. Considering the landscape scale, those landscapes with a higher landscape diversity reflected increasing shares of species-rich habitats such as orchards and forests. This might lead to spillover from these high-richness habitat types to normally low-richness habitats which might increase bird richness in the latter. Finally, landscapes with higher compositional diversity typically have higher landscape connectivity which might facilitate between-habitat movement and local species assembly (Tscharntke et al. 2021 ). Beta diversity of bird assemblages While differences in alpha diversity of birds between habitat types are well established, much less is known about variability in bird assemblages among locations of the same habitat type (Edo et al. 2024 ; Cours and Duflot 2025 ). We could show that beta diversity within and between habitat types can substantially shape bird assemblages in agricultural landscapes. Despite supporting relatively low bird richness, arable land exhibited the highest within-habitat beta diversity (Fig. 4 ), indicating strong turnover in bird assemblages among arable locations. This pattern likely reflects heterogeneity in arable management, including differences in crop types and their phenologies (e.g. winter wheat sown in autumn versus maize sown in spring) and management intensity (e.g. sowing density). Such fine-scale management variation may promote distinct bird assemblages even among arable-dominated locations. For example, the Western Yellow Wagtail (Motacilla flava) persists well in intensively managed cereal fields (Kragten 2011 ), whereas the Eurasian Skylark requires unsown gaps within the cereal fields (Morris et al. 2004 ). In both forest- and orchard-dominated locations, within-habitat beta diversity was significantly lower than in arable land (ca. – 25%; Fig. 4 ). This contrasts with the findings of Edo et al. ( 2024 ), who reported higher within-forest than within-orchard beta diversity. Our results further indicate that orchards, despite supporting high bird alpha diversity, host relatively similar bird assemblages across locations. Even though varying structure and management of traditional orchards can lead to varying habitat quality for bird species (Chaparro et al. 2022 ), our findings suggest that this habitat type can consistently provide suitable conditions for a broad range of species. Like orchards, within-habitat beta diversity in forest-dominated locations was relatively low as both habitat types are more stable over time than arable land. Furthermore, the forests’ limited understory vegetation, little or no deadwood, scarce floral resources, and relatively uniform stand ages might have contributed to the low dissimilarity in bird assemblages across forest sites. Between-habitat beta diversity of bird assemblages exceeded within-habitat beta diversity. The habitats in this study spanned a gradient of structural complexity: from open arable land and grassland, through semi-open orchards, to dense forest stands. As the ecological contrast (Marja et al. 2019 ) between habitats increased along this gradient, the dissimilarity in bird assemblages also increased. For example, arable-forest beta diversity was higher than arable-grassland beta diversity. This pattern is likely driven by trait-specific responses to differences in resource availability and habitat structure. For example, arable-dominated locations were primarily characterized by granivorous species (e.g. Passer spp.) adapted to open habitats, whereas forest-dominated locations were characterized by insectivorous species associated with dense habitats (e.g. common treecreeper Certhia familiaris ; Table S2). Beta diversity between arable land and grassland was rather low, as they are both open habitats with little or no trees and experience both frequent disturbances, such as repeated machinery use for fertilization, pesticide application, or mowing. This smaller ecological contrast likely explains the lower dissimilarity in bird assemblages observed between these two open habitat types. Bird assemblages in arable landscapes were dominated by multi-habitat users, reflecting the need to cope with frequent and often unpredictable disturbances in agricultural systems. In such environments, ecological generalists are favoured that can exploit multiple resources, often across different habitat types. This can result in a low prevalence of specialists. In contrast, forest ecosystems are comparatively stable over time, allowing for narrower ecological niches and supporting a higher proportion of habitat specialists associated with forest conditions. Finally, also the results of a generalized dissimilarity model (GDM) showed that local predictors such as the amount of forest or arable land within a 100 m radius of the sampling location were more important in explaining bird beta diversity than landscape level predictors (Fig. 4 ). Thus, local ecological contrasts, particularly those between arable land and forest, appear to be the principal drivers of beta diversity in farmland bird assemblages. Gamma diversity of bird assemblages Overall, our results empirically prove that different habitat types support distinct bird assemblages. Landscapes with a greater diversity of habitat types sustain a higher gamma diversity than simplified landscapes dominated by only a few, often uniformly managed habitats (Fig. 5 ). This relationship suggests that heterogeneous landscapes provide a broader range of available niches, thereby supporting both habitat specialists and multi-habitat users (Cours and Duflot 2025 ). Increasing evidence further indicates that such landscape-level heterogeneity is essential not only for maintaining species richness but also for preserving the functional resilience of bird communities and the ecosystem services they provide (Weeks et al. 2025 ). Conclusion Our study showed that habitat heterogeneity across local and the landscape scales (i.e. landscape diversity) is essential for maintaining farmland bird diversity at multiple scales, as it enhances bird richness and promotes bird community assembly. Variation in bird community composition (i.e. beta diversity) was driven primarily by local habitat characteristics and ecological contrasts among habitats and is a key component of biodiversity in agricultural landscapes. Therefore, it is important to preserve structurally diverse and heterogeneous landscapes, including a range of habitat types to sustain farmland bird communities. Declarations Competing interests The authors have no relevant financial or non-financial interests to disclose. Dava availability The datasets generated during the current study are available in the supplement. Funding The research was conducted within the cooperative, interdisciplinary doctoral program “Leverage points for Biodiversity Enhancement in Agricultural Landscapes (HABIT)” of the University of Hohenheim and Nürtingen-Geislingen University, funded by the Ministry for Science, Research and Arts of Baden-Württemberg as part of the State Postgraduate Scholarship Program (project number BW6_09). Author Contribution M.K.K. and I.G. conceived the ideas and designed the methodology; M.K.K., S.T., and R.B. collected the data; M.K.K., I.G., and T.H. analyzed the data; M.K.K. and I.G. led the writing of the manuscript. All authors contributed critically to the drafts and approved the final manuscript. Acknowledgement The research was conducted within the cooperative, interdisciplinary doctoral program “Leverage points for Biodiversity Enhancement in Agricultural Landscapes (HABIT)” of the University of Hohenheim and Nürtingen-Geislingen University, funded by the Ministry for Science, Research and Arts of Baden-Württemberg as part of the State Postgraduate Scholarship Program (project number BW6_09). We thank Yasha Auer, Carlos Gonzalez and Pauline Greiner for assistance during field work. Data Availability The datasets generated for this study are available in the supplement. References Anderle M, Paniccia C, Brambilla M, Hilpold A, Volani S, Tasser E, Seeber J, Tappeiner U (2022) The contribution of landscape features, climate and topography in shaping taxonomical and functional diversity of avian communities in a heterogeneous Alpine region. Oecologia, 199(3):499–512 Barbaro L, Pontcharraud L, Vetillard F, Guyon D, Jactel H (2005) Comparative responses of bird, carabid, and spider assemblages to stand and landscape diversity in maritime pine plantation forests. Écoscience, 12(1):110–121 Bates D, Mächler M, Bolker B, Walker S (2015) Fitting Linear Mixed-Effects Models Using lme4. J. Stat. Soft., 67(1):1–48 Benton TG, Vickery JA, Wilson JD (2003) Farmland biodiversity: is habitat heterogeneity the key? Trends in Ecology & Evolution, 18(4):182–188 Borngraeber S, Krismann A, Schmieder K (2020) Ermittlung der Streuobstbestände Baden-Württembergs durch automatisierte Fernerkundungsverfahren. In: LUBW (ed) Naturschutz und Landschaftspflege Baden-Württemberg. Bd. 81: Sammelband mit mehreren Einzelbeiträgen. Naturschutz und Landschaftspflege Baden-Württemberg, vols 81, Karlsruhe, pp 4–22 Breunig T, Demuth S, Cordlandwehr V (2020) Rote Liste der Biotoptypen Baden-Württembergs mit naturschutzfachlicher Beurteilung. Available from https://pd.lubw.de/10247 Buchmann CM, Schurr FM (2025) An automated workflow that combines sound-based bird identification and localization. IBIS Chao A (1984) Nonparametric Estimation of the Number of Classes in a Population. Scandinavian Journal of Statistics, 11:265–270 Chaparro L, Schmieder K, Schurr FM (2022) The value of remotely sensed vs. field-surveyed habitat structure for predicting bird abundance: a case study in traditional orchards. J Ornithol, 163(3):723–733 Cours J, Duflot R (2025) Effects of landscape heterogeneity on bird communities in temperate, boreal, and montane forests – a review. Journal of Avian Biology, 2025(3) Cullmann D (2025) Der Wald in Baden-Württemberg: Ausgewählte Ergebnisse und regionale Auswertungen der Bundeswaldinventur 2022, Freiburg Dambros CS, Morais JW, Azevedo RA, Gotelli NJ (2017) Isolation by distance, not rivers, control the distribution of termite species in the Amazonian rain forest. Ecography, 40(10):1242–1250 Darras K, Batáry P, Furnas B, Celis-Murillo A, van Wilgenburg SL, Mulyani YA, Tscharntke T (2018) Comparing the sampling performance of sound recorders versus point counts in bird surveys: A meta‐analysis. J Appl Ecol, 55(6):2575–2586 Duflot R, San-Cristobal M, Andrieu E, Choisis J-P, Esquerré D, Ladet S, Ouin A, Rivers-Moore J, Sheeren D, Sirami C, Fauvel M, Vialatte A (2022) Farming intensity indirectly reduces crop yield through negative effects on agrobiodiversity and key ecological functions. Agriculture, Ecosystems & Environment, 326:107810 Dunning JB, Danielson BT, Pulliam HR (1992) Ecological Processes That Affect Populations in Complex Landscapes. Oikos, 65(1):169–175 Edo M, Entling MH, Rösch V (2024) Agroforestry supports high bird diversity in European farmland. Agron. Sustain. Dev., 44(1) Fahrig L (2013) Rethinking patch size and isolation effects: the habitat amount hypothesis. Journal of Biogeography, 40(9):1649–1663 Fitzpatrick M, Mokany K, Manion G, Nieto-Lugilde D, Ferrier S (2025) gdm: Generalized Dissimilarity Modeling Hartel T, Hanspach J, Abson DJ, Máthé O, Moga CI, Fischer J (2014) Bird communities in traditional wood-pastures with changing management in Eastern Europe. Basic and Applied Ecology, 15(5):385–395 Hill AP, Prince P, Snaddon JL, Doncaster CP, Rogers A (2019) AudioMoth: A low-cost acoustic device for monitoring biodiversity and the environment. HardwareX, 6:e00073 Kragten S (2011) Shift in crop preference during the breeding season by Yellow Wagtails Motacilla flava flava on arable farms in The Netherlands. J Ornithol, 152(3):751–757 Lenth RV (2024) emmeans: Estimated Marginal Means, aka Least-Squares Means: R package version 1.10.5. Available from https://CRAN.R-project.org/package=emmeans Lüdecke D, Ben-Shachar M, Patil I, Waggoner P, Makowski D (2021) performance: An R Package for Assessment, Comparison and Testing of Statistical Models. JOSS, 6(60):3139 Marja R, Kleijn D, Tscharntke T, Klein A-M, Frank T, Batáry P (2019) Effectiveness of agri-environmental management on pollinators is moderated more by ecological contrast than by landscape structure or land-use intensity. Ecol Letters, 22(9):1493–1500 McKeever CU (2003) Linking grassland management, invertebrates and Northern Lapwing productivity Metainformationssystem GDI-BW (2023) WFS Flächeninformationen und Online-Antrag Baden-Württemberg (FIONA). Available from https://metadaten.geoportal-bw.de/geonetwork/srv/ger/catalog.search#/home (accessed February 2025) Morris AJ, Holland JM, Smith B, Jones NE (2004) Sustainable Arable Farming For an Improved Environment (SAFFIE): managing winter wheat sward structure for Skylarks Alauda arvensis. IBIS, 146(s2):155–162 Pérez-Granados C, Funosas D, Morant J, Marín Gómez OH, Mendoza I, Mohedano‐Munoz MA, Santamaría E, Bastianelli G, Márquez‐Rodríguez A, Budka M, Bota G, La Peña‐Rubio JM de, La García De Morena E, Santa‐Cruz M, La Nava P de, Fernández‐Tizón M, Sánchez‐Mateos H, Barrero A, Traba J, Osiejuk TS, Hart PJ, Navine AK, Montoya Muñoz AF, Araújo CB de, Rosa GLM, Torres IMD, Catalano ALC, Simões CR, Llusia D, Morales MB, Acebes P, Medina JA, Brown N, Astaras C, Karmiris I, Navarrete E, Cauchoix M, Barbaro L, Arend D, Müeller S, González‐García F, González‐Romero A, Mammides C, Pontikis M, Jacuzzi G, Olden JD, Bombaci SP, Marcacci G, Jacot A, Zurano JP, Gangenova E, Varela D, Di Sallo F, Zurita GA, Atemasov A, Tremblay JA, Hutschenreiter A, Monroy‐Ojeda A, Díaz‐Vallejo M, Chaparro‐Herrera S, Briers RA, Sousa‐Lima R, Pinheiro T, Da Silva WC, Calvente A, Dal Molin A, Antonelli A, Gogoleva S, Palko I, Trong HV, Duarte MHL, Saturnino NDS, Silva SR, Rainho A, Lopes P, Schuchmann K-L, Marques MI, Oliverira Tissiani AS de, Littlewood NA, Tuanmu M-N, Cheng Y-R, Chao H, Kepfer‐Rojas S, Aguilera AL, Brotons L, Feldman MJ, Imbeau L, Panwar P, Weed AS, Dehwal A, Attisano A, Theuerkauf J, Oliveira‐Júnior DD, Lima‐Santos CS, Salustio‐Gomes C, Paz RV, Pichorim M, Goodale E, Sebastián‐González E (2025) Optimization of passive acoustic bird surveys: a global assessment of BirdNET settings. IBIS R Core Team (2024) R: A Language and Environment for Statistical Computing. R Foundation for Statistical Computing, Vienna, Austria Rigal S, Dakos V, Alonso H, Auinins A, Benko Z, et al. (2023) Farmland practices are driving bird population decline across Europe. PNAS, 120(21) Sattler C, Schrader J, Hüttner M-L, Henle K (2024) Effects of management, habitat and landscape characteristics on biodiversity of orchard meadows in Central Europe: a brief review. Nature Conservation, 55:103–134 Scherber C, Beduschi T, Tscharntke T (2019) Novel approaches to sampling pollinators in whole landscapes: a lesson for landscape-wide biodiversity monitoring. Landscape Ecol, 34(5):1057–1067 Sirami C, Gross N, Baillod AB, Bertrand C, Carrié R, Hass A, Henckel L, Miguet P, Vuillot C, Alignier A, Girard J, Batáry P, Clough Y, Violle C, Giralt D, Bota G, Badenhausser I, Lefebvre G, Gauffre B, Vialatte A, Calatayud F, Gil-Tena A, Tischendorf L, Mitchell S, Lindsay K, Georges R, Hilaire S, Recasens J, Solé-Senan XO, Robleño I, Bosch J, Barrientos JA, Ricarte A, Marcos-Garcia MÁ, Miñano J, Mathevet R, Gibon A, Baudry J, Balent G, Poulin B, Burel F, Tscharntke T, Bretagnolle V, Siriwardena G, Ouin A, Brotons L, Martin J-L, Fahrig L (2019) Increasing crop heterogeneity enhances multitrophic diversity across agricultural regions. Proceedings of the National Academy of Sciences of the United States of America, 116(33):16442–16447 Statistisches Landesamt Baden-Württemberg (2024) Anbauflächen aller Kultur- und Nutzungsarten seit 2010. Available from https://www.statistik-bw.de/Landwirtschaft/Bodennutzung/LF-NutzngKultFrucht.jsp (accessed January 2025) Streuobstportal Baden-Württemberg (2025) Landesweite Streuobstdatenerhebung. Available from https://streuobst.landwirtschaft-bw.de/,Lde/Startseite/Wissen/Landesweite+Streuobstdatenerhebung (accessed January 2025) Tobias JA, Sheard C, Pigot AL, Devenish AJM, Yang J, Sayol F, Neate-Clegg MHC, Alioravainen N, Weeks TL, Barber RA, Walkden PA, MacGregor HEA, Jones SEI, Vincent C, Phillips AG, Marples NM, Montaño-Centellas FA, Leandro-Silva V, Claramunt S, Darski B, Freeman BG, Bregman TP, Cooney CR, Hughes EC, Capp EJR, Varley ZK, Friedman NR, Korntheuer H, Corrales-Vargas A, Trisos CH, Weeks BC, Hanz DM, Töpfer T, Bravo GA, Remeš V, Nowak L, Carneiro LS, Moncada R AJ, Matysioková B, Baldassarre DT, Martínez-Salinas A, Wolfe JD, Chapman PM, Daly BG, Sorensen MC, Neu A, Ford MA, Mayhew RJ, Fabio Silveira L, Kelly DJ, Annorbah NND, Pollock HS, Grabowska-Zhang AM, McEntee JP, Carlos T Gonzalez J, Meneses CG, Muñoz MC, Powell LL, Jamie GA, Matthews TJ, Johnson O, Brito GRR, Zyskowski K, Crates R, Harvey MG, Jurado Zevallos M, Hosner PA, Bradfer-Lawrence T, Maley JM, Stiles FG, Lima HS, Provost KL, Chibesa M, Mashao M, Howard JT, Mlamba E, Chua MAH, Li B, Gómez MI, García NC, Päckert M, Fuchs J, Ali JR, Derryberry EP, Carlson ML, Urriza RC, Brzeski KE, Prawiradilaga DM, Rayner MJ, Miller ET, Bowie RCK, Lafontaine R-M, Scofield RP, Lou Y, Somarathna L, Lepage D, Illif M, Neuschulz EL, Templin M, Dehling DM, Cooper JC, Pauwels OSG, Analuddin K, Fjeldså J, Seddon N, Sweet PR, DeClerck FAJ, Naka LN, Brawn JD, Aleixo A, Böhning-Gaese K, Rahbek C, Fritz SA, Thomas GH, Schleuning M (2022) AVONET: morphological, ecological and geographical data for all birds. Ecol Letters, 25(3):581–597 Toenies M, Rich LN (2021) Advancing bird survey efforts through novel recorder technology and automated species identification. California Fish and Wildlife, 107(2):56–70 Tscharntke T, Grass I, Wanger TC, Westphal C, Batáry P (2021) Beyond organic farming - harnessing biodiversity-friendly landscapes. Trends in ecology & evolution, 36(10):919–930 Tscharntke T, Tylianakis JM, Rand TA, Didham RK, Fahrig L, Batáry P, Bengtsson J, Clough Y, Crist TO, Dormann CF, Ewers RM, Fründ. Jochen, Holt RD, Holzschuh A, Klein AM, Kleijn D, Kremen C, Landis DA, Laurance W, Lindenmayer D, Scherber C, Sodhi N, Steffan-Dewenter I, Thies C, van der Putten WH, Westphal C (2012) Landscape moderation of biodiversity patterns and processes - eight hypotheses. Biological Reviews, 87:661–685 Valente JJ, Gannon DG, Hightower J, Kim H, Leimberger KG, Macedo R, Rousseau JS, Weldy MJ, Zitomer RA, Fahrig L, Fletcher RJ, Wu J, Betts MG (2023) Toward conciliation in the habitat fragmentation and biodiversity debate. Landsc Ecol, 38(11):2717–2730 Weeks TL, Walkden PA, Edwards DP, Lees AC, Pigot AL, Purvis A, Tobias JA (2025) Land-use change undermines the stability of avian functional diversity. Nature Wilson S, Alavi N, Pouliot D, Mitchell GW (2020) Similarity between agricultural and natural land covers shapes how biodiversity responds to agricultural expansion at landscape scales. Agriculture, Ecosystems & Environment, 301:107052 Wood CM, Kahl S (2024) Guidelines for appropriate use of BirdNET scores and other detector outputs. J Ornithol Additional Declarations No competing interests reported. Supplementary Files KastenetalBirdssupplementarymaterial.docx Cite Share Download PDF Status: Under Review Version 1 posted Reviews received at journal 13 May, 2026 Reviewers agreed at journal 21 Apr, 2026 Reviewers agreed at journal 19 Apr, 2026 Reviewers invited by journal 17 Apr, 2026 Editor assigned by journal 13 Apr, 2026 Submission checks completed at journal 13 Apr, 2026 First submitted to journal 09 Apr, 2026 You are reading this latest preprint version Research Square lets you share your work early, gain feedback from the community, and start making changes to your manuscript prior to peer review in a journal. As a division of Research Square Company, we’re committed to making research communication faster, fairer, and more useful. We do this by developing innovative software and high quality services for the global research community. Our growing team is made up of researchers and industry professionals working together to solve the most critical problems facing scientific publishing. Also discoverable on Platform About Our Team In Review Editorial Policies Advisory Board Help Center Resources Author Services Accessibility API Access RSS feed Manage Cookie Preferences © Research Square 2026 | ISSN 2693-5015 (online) Privacy Policy Terms of Service Do Not Sell My Personal Information {"props":{"pageProps":{"initialData":{"identity":"rs-9368528","acceptedTermsAndConditions":true,"allowDirectSubmit":false,"archivedVersions":[],"articleType":"Research Article","associatedPublications":[],"authors":[{"id":627241782,"identity":"8eac9e13-3383-4368-97d6-c14036ebf216","order_by":0,"name":"Marit Kinga Kasten","email":"data:image/png;base64,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","orcid":"","institution":"University of Hohenheim","correspondingAuthor":true,"prefix":"","firstName":"Marit","middleName":"Kinga","lastName":"Kasten","suffix":""},{"id":627241783,"identity":"78b7763b-ca95-46b2-adc1-b4cb64d0c868","order_by":1,"name":"Thomas Hiller","email":"","orcid":"","institution":"University of Hohenheim","correspondingAuthor":false,"prefix":"","firstName":"Thomas","middleName":"","lastName":"Hiller","suffix":""},{"id":627241784,"identity":"3a965639-c7e6-4c71-9efd-dc56e8766c10","order_by":2,"name":"Sara Tassoni","email":"","orcid":"","institution":"University of Hohenheim","correspondingAuthor":false,"prefix":"","firstName":"Sara","middleName":"","lastName":"Tassoni","suffix":""},{"id":627241785,"identity":"0a954c5c-b603-42dd-92b9-38127eb5763b","order_by":3,"name":"Rosalie Böhmer","email":"","orcid":"","institution":"University of Hohenheim","correspondingAuthor":false,"prefix":"","firstName":"Rosalie","middleName":"","lastName":"Böhmer","suffix":""},{"id":627241786,"identity":"45743abb-69a1-498e-b47e-93d5ce65963d","order_by":4,"name":"Frank M. Schurr","email":"","orcid":"","institution":"University of Hohenheim","correspondingAuthor":false,"prefix":"","firstName":"Frank","middleName":"M.","lastName":"Schurr","suffix":""},{"id":627241787,"identity":"5d89d504-6070-45f5-bdf6-f34221ef7639","order_by":5,"name":"Markus Röhl","email":"","orcid":"","institution":"Nürtingen-Geislingen University of Applied Science","correspondingAuthor":false,"prefix":"","firstName":"Markus","middleName":"","lastName":"Röhl","suffix":""},{"id":627241788,"identity":"21484926-608c-4666-b6ca-72a59a443e41","order_by":6,"name":"Michael Roth","email":"","orcid":"","institution":"Nürtingen-Geislingen University of Applied Science","correspondingAuthor":false,"prefix":"","firstName":"Michael","middleName":"","lastName":"Roth","suffix":""},{"id":627241789,"identity":"c340aa1a-2b8e-48bc-82b2-6a6099980df9","order_by":7,"name":"Ingo Grass","email":"","orcid":"","institution":"University of Hohenheim","correspondingAuthor":false,"prefix":"","firstName":"Ingo","middleName":"","lastName":"Grass","suffix":""}],"badges":[],"createdAt":"2026-04-09 12:23:56","currentVersionCode":1,"declarations":"","doi":"10.21203/rs.3.rs-9368528/v1","doiUrl":"https://doi.org/10.21203/rs.3.rs-9368528/v1","draftVersion":[],"editorialEvents":[],"editorialNote":"","failedWorkflow":false,"files":[{"id":108007268,"identity":"df2e58fa-431f-4d00-8a31-6b3584cd64cc","added_by":"auto","created_at":"2026-04-28 12:59:14","extension":"jpg","order_by":1,"title":"Figure 1","display":"","copyAsset":false,"role":"figure","size":140979,"visible":true,"origin":"","legend":"\u003cp\u003e\u003cem\u003eExplanation of alpha diversity, within-habitat beta diversity, between-habitat beta diversity and gamma diversity. Blue squares represent the 1-km\u003c/em\u003e\u003csup\u003e\u003cem\u003e2\u003c/em\u003e\u003c/sup\u003e\u003cem\u003e landscapes, while coloured dots show the recording locations of the AudioMoths indicating the main habitat type (arable: yellow, forest: green, grassland: brown, orchard: purple). The landscape shown is the same as in Figure 2b.\u003c/em\u003e\u003c/p\u003e\n\u003cp\u003e\u0026nbsp;\u003c/p\u003e","description":"","filename":"1.jpg","url":"https://assets-eu.researchsquare.com/files/rs-9368528/v1/66b9b8b6e85bf7db44f4569d.jpg"},{"id":108181391,"identity":"1a26e98c-638c-4ac9-af53-ab2a89d14bb8","added_by":"auto","created_at":"2026-04-30 08:58:36","extension":"jpg","order_by":2,"title":"Figure 2","display":"","copyAsset":false,"role":"figure","size":156818,"visible":true,"origin":"","legend":"\u003cp\u003e\u003cem\u003eOverview of the grid-based sampling design. a) Map showing the 14 study landscapes south of Stuttgart, Germany; b) grid-based sampling design (Scherber et al. 2019) covering all major habitat types (coloured polygons) in each study landscape in proportion to their relative cover (example for study landscape L7) with white dots depicting the 16 recording locations. c) Proportional cover of the 7 land cover types for all 14 landscapes, arranged by increasing landscape diversity. Colours are consistent in this figure and throughout the whole manuscript with habitats coloured in yellow standing for arable land, green for forest, brown/red for grassland, purple for orchard. Other land cover types are in rose for hedge, light grey for roads and dark grey for settlements.\u003c/em\u003e\u003c/p\u003e","description":"","filename":"2.jpg","url":"https://assets-eu.researchsquare.com/files/rs-9368528/v1/8be173e76e330692c18b8603.jpg"},{"id":107901568,"identity":"46d507d1-5607-4ec7-bbfd-0225ee620a58","added_by":"auto","created_at":"2026-04-27 11:47:44","extension":"jpg","order_by":3,"title":"Figure 3","display":"","copyAsset":false,"role":"figure","size":32157,"visible":true,"origin":"","legend":"\u003cp\u003e\u003cem\u003eAlpha diversity (species richness per recording location) is significantly influenced by additive effects of (a) main habitat type, (b) local habitat diversity and (c) landscape diversity. Local habitat diversity and landscape diversity represent the Shannon diversity indices of habitat heterogeneity in a 100 m radius around the recording location and the corresponding 1-km\u003c/em\u003e\u003csup\u003e\u003cem\u003e2\u003c/em\u003e\u003c/sup\u003e\u003cem\u003e landscape, respectively. The solid black line indicates the model prediction; the whiskers and the grey area indicate the 95% confidence interval of the predictions. Semi-transparent black dots show raw data.\u003c/em\u003e\u003c/p\u003e\n\u003cp\u003e\u0026nbsp;\u003c/p\u003e","description":"","filename":"3.jpg","url":"https://assets-eu.researchsquare.com/files/rs-9368528/v1/b34efefd1004335c6baa2f6b.jpg"},{"id":108006422,"identity":"d5b23361-085e-4d26-ab1b-4acaf48a9c9e","added_by":"auto","created_at":"2026-04-28 12:55:27","extension":"jpg","order_by":4,"title":"Figure 4","display":"","copyAsset":false,"role":"figure","size":61332,"visible":true,"origin":"","legend":"\u003cp\u003e\u003cem\u003eWithin-landscape beta diversity and its influencing variables. a) Non-metric multidimensional scaling (NMDS) showing all 221 \u003c/em\u003erecording \u003cem\u003elocations and their associated main habitat type (arable: yellow, forest: green, grassland: brown, orchard: purple). Ellipses show 95% confidence interval of each habitat type. b) Within-habitat beta diversity (Figure 1): Sørensen dissimilarity values, meaning pairwise comparisons between recording locations of the same habitat type within the same landscape, shown by the type of habitat comparisons. c) Between-habitat beta diversity (Figure 1): Sørensen dissimilarity values, meaning pairwise comparisons between recording locations of different habitat types within the same landscape, shown by the type of habitat comparisons. d) The percentage of the explanatory power of the four significant variables from left to right: landscape orchard cover (light pink), landscape forest cover (light green), local forest cover (dark green) and local arable cover (yellow) from the generalized dissimilarity modelling.\u003c/em\u003e\u003c/p\u003e","description":"","filename":"4.jpg","url":"https://assets-eu.researchsquare.com/files/rs-9368528/v1/08c57625a950285ab4636b49.jpg"},{"id":108011068,"identity":"8809de77-b991-42d4-b04a-b60c55fdc7cd","added_by":"auto","created_at":"2026-04-28 13:14:17","extension":"jpg","order_by":5,"title":"Figure 5","display":"","copyAsset":false,"role":"figure","size":41219,"visible":true,"origin":"","legend":"\u003cp\u003e\u003cem\u003eGamma diversity of 14 differently composed agricultural landscapes (along their Shannon diversity index, from simple to complex). Gamma diversity indicates the species richness across all 16 recording locations in a landscape (see Figure 1). Black dots represent raw data points, the black line the model prediction and the grey zone its 95% confidence interval.\u003c/em\u003e\u003c/p\u003e","description":"","filename":"5.jpg","url":"https://assets-eu.researchsquare.com/files/rs-9368528/v1/2c1174877997638b112b5a1d.jpg"},{"id":107901570,"identity":"21101e66-5b1c-4dcf-9688-7d3457f32531","added_by":"auto","created_at":"2026-04-27 11:47:45","extension":"jpg","order_by":6,"title":"Figure 6","display":"","copyAsset":false,"role":"figure","size":42474,"visible":true,"origin":"","legend":"\u003cp\u003e\u003cem\u003eSpecies information in the four main habitat types. Non-metric multidimensional scaling (NMDS) showing points as species, where the shape of the points indicates the long-term population trend (triangle = increasing, circle = stable, inverse triangle = decreasing), colours indicate their trophic niche (yellow = granivore, brown = invertivore, grey = omnivore, blue = vertivore; AVONET database) and the transparency of the colour indicating the preferred habitat density (fully coloured: dense habitat, semi-transparent: semi-open habitat, no filling: open habitats; AVONET database). Arrows indicate significant environmental variables explaining the distribution of the points in the NMDS. Boxes show exemplary species and further information on these species. Outside of the plot are water pipit (Anthus spinoletta (L.); with traits trophic niche = invertivore, habitat density = open habitats, population trend = stable) and Western jackdaw (Corvus monedula (L.); with traits trophic niche = omnivore, habitat density = open habitats, population trend = stable).\u003c/em\u003e\u003c/p\u003e","description":"","filename":"6.jpg","url":"https://assets-eu.researchsquare.com/files/rs-9368528/v1/87b9c9b1a6fc507be682f246.jpg"},{"id":108494370,"identity":"31bd57e2-89c2-479d-8200-55030df8d95f","added_by":"auto","created_at":"2026-05-05 10:04:18","extension":"pdf","order_by":0,"title":"","display":"","copyAsset":false,"role":"manuscript-pdf","size":787893,"visible":true,"origin":"","legend":"","description":"","filename":"manuscript.pdf","url":"https://assets-eu.researchsquare.com/files/rs-9368528/v1/b874eab3-9f09-4b87-b40e-626c000cd8ae.pdf"},{"id":107901566,"identity":"b435ad0d-2356-4aaa-808b-40fa73799de7","added_by":"auto","created_at":"2026-04-27 11:47:44","extension":"docx","order_by":0,"title":"","display":"","copyAsset":false,"role":"supplement","size":112792,"visible":true,"origin":"","legend":"","description":"","filename":"KastenetalBirdssupplementarymaterial.docx","url":"https://assets-eu.researchsquare.com/files/rs-9368528/v1/11c6ed51e07b45e1bdbd82e7.docx"}],"financialInterests":"No competing interests reported.","formattedTitle":"Farmland bird diversity requires heterogeneity between and within habitats","fulltext":[{"header":"Introduction","content":" \u003cp\u003eAcross Europe, populations of common bird species have declined by about 25% between 1980 and 2016 (Rigal et al. \u003cspan citationid=\"CR30\" class=\"CitationRef\"\u003e2023\u003c/span\u003e). Farmland birds exhibited even higher losses with an average decline of 56% (Rigal et al. \u003cspan citationid=\"CR30\" class=\"CitationRef\"\u003e2023\u003c/span\u003e). These declines have been attributed to agricultural intensification and changes in the structure of agricultural landscapes (Sirami et al. \u003cspan citationid=\"CR33\" class=\"CitationRef\"\u003e2019\u003c/span\u003e; Rigal et al. \u003cspan citationid=\"CR30\" class=\"CitationRef\"\u003e2023\u003c/span\u003e). As structurally diverse mosaic landscapes were replaced by simplified, agriculturally intensified landscapes, key habitats for many farmland bird species diminished. However, despite extensive research (Sirami et al. \u003cspan citationid=\"CR33\" class=\"CitationRef\"\u003e2019\u003c/span\u003e; Rigal et al. \u003cspan citationid=\"CR30\" class=\"CitationRef\"\u003e2023\u003c/span\u003e), we still lack larger-scale assessments that jointly analyze local (alpha) and compositional (beta) diversity of farmland bird communities (Fig.\u0026nbsp;\u003cspan refid=\"Fig1\" class=\"InternalRef\"\u003e1\u003c/span\u003e) across spatial scales (Cours and Duflot \u003cspan citationid=\"CR10\" class=\"CitationRef\"\u003e2025\u003c/span\u003e).\u003c/p\u003e \u003cp\u003eBiodiversity patterns are shaped by ecological processes operating across multiple spatial scales. At the local scale, species richness and composition of bird assemblages is primarily determined by habitat-specific resource availability and structural heterogeneity (Fahrig \u003cspan citationid=\"CR17\" class=\"CitationRef\"\u003e2013\u003c/span\u003e). Habitats rich in woody elements, such as forests or traditional orchards, typically support species associated with semi-open or forested environments. Arable fields, however, favour species adapted to open landscapes. We expect orchards to harbour more species with decreasing population trends than arable land, as in arable land these endangered species could have already disappeared. Notably, local alpha diversity of bird assemblages is also shaped by the broader landscape context. Landscapes with higher compositional heterogeneity can support richer local bird assemblages (Benton et al. \u003cspan citationid=\"CR4\" class=\"CitationRef\"\u003e2003\u003c/span\u003e), because heterogeneous surroundings increase the pool of species which are able to exploit resources across multiple habitat types (habitat complementation; Dunning et al. \u003cspan citationid=\"CR15\" class=\"CitationRef\"\u003e1992\u003c/span\u003e).\u003c/p\u003e \u003cp\u003eThe species-specific associations with habitat resources and structural features imply that contrasts among habitat types (e.g. forest vs. farmland) strongly influence whether two local sites host similar or distinct bird assemblages. Such habitat-driven differences give rise to beta diversity, i.e. the compositional turnover of species between sites. While alpha diversity of farmland birds across different habitat types is relatively well studied, beta diversity within and between habitat types is rarely studied (Edo et al. \u003cspan citationid=\"CR16\" class=\"CitationRef\"\u003e2024\u003c/span\u003e; Cours and Duflot \u003cspan citationid=\"CR10\" class=\"CitationRef\"\u003e2025\u003c/span\u003e). Sites belonging to the same habitat type can, for example, differ in the quantity and diversity of resources or structural elements they provide due to varying degrees of land-use intensity (Benton et al. \u003cspan citationid=\"CR4\" class=\"CitationRef\"\u003e2003\u003c/span\u003e), creating within-habitat beta diversity (Fig.\u0026nbsp;\u003cspan refid=\"Fig1\" class=\"InternalRef\"\u003e1\u003c/span\u003e). These differences in structure and management should be more pronounced between habitat types, leading to more pronounced values of beta diversity between habitat types (between-habitat beta diversity, Fig.\u0026nbsp;\u003cspan refid=\"Fig1\" class=\"InternalRef\"\u003e1\u003c/span\u003e). Beyond habitat characteristics, spatial processes such as distance decay can further increase beta diversity by limiting dispersal or reducing community similarity with increasing geographic separation (Dambros et al. \u003cspan citationid=\"CR12\" class=\"CitationRef\"\u003e2017\u003c/span\u003e). Overall, however, the relative contributions of between-habitat and within-habitat beta diversity in structurally complex agricultural landscapes remain poorly understood (Fig.\u0026nbsp;\u003cspan refid=\"Fig1\" class=\"InternalRef\"\u003e1\u003c/span\u003e).\u003c/p\u003e \u003cp\u003eAt the landscape scale, gamma diversity (Fig.\u0026nbsp;\u003cspan refid=\"Fig1\" class=\"InternalRef\"\u003e1\u003c/span\u003e) emerges from the interplay of local alpha diversity and turnover among sites (beta diversity). The habitat diversity hypothesis assumes that more heterogeneous landscapes support higher gamma diversity because they offer a wider range of ecological niches (Benton et al. \u003cspan citationid=\"CR4\" class=\"CitationRef\"\u003e2003\u003c/span\u003e; Duflot et al. \u003cspan citationid=\"CR14\" class=\"CitationRef\"\u003e2022\u003c/span\u003e; Valente et al. \u003cspan citationid=\"CR40\" class=\"CitationRef\"\u003e2023\u003c/span\u003e). Diverse landscapes containing mixtures of forests, grasslands, orchards, and croplands can provide complementary resources that support a broader set of bird species than simplified landscapes. When aggregated regionally, such landscape-level diversity patterns determine how much differently composed landscapes contribute to the overall species pool.\u003c/p\u003e \u003cp\u003eIn this study, we investigated farmland bird communities across agricultural landscapes of varying structural complexity in southern Germany. To characterize bird communities at the landscape scale, we employed passive acoustic monitoring combined with automatic species recognition using BirdNET. Our aim was to identify the key local and landscape-scale drivers of alpha and beta diversity in farmland bird communities (Fig.\u0026nbsp;\u003cspan refid=\"Fig1\" class=\"InternalRef\"\u003e1\u003c/span\u003e), thereby improving our understanding of how agricultural landscape structure shapes biodiversity across spatial scales.\u003c/p\u003e \u003cp\u003e \u003c/p\u003e"},{"header":"Methods","content":"\u003cdiv id=\"Sec3\" class=\"Section2\"\u003e \u003ch2\u003eLandscape selection and landscape composition\u003c/h2\u003e \u003cp\u003eWe conducted our study in an agricultural region (20 x 20 km\u0026sup2;) south of Stuttgart, Germany (48\u0026deg;46\u0026prime;39\u0026Prime;N 09\u0026deg;10\u0026prime;48\u0026Prime;E; Fig.\u0026nbsp;\u003cspan refid=\"Fig2\" class=\"InternalRef\"\u003e2\u003c/span\u003ea). The landscapes in our study region consist mainly of arable land (predominantly for cereal cultivation), forest (predominantly beech forest), grassland (including meadows and pastures), and orchard (predominantly traditional, extensively managed apple orchards) (Statistisches Landesamt Baden-W\u0026uuml;rttemberg \u003cspan citationid=\"CR34\" class=\"CitationRef\"\u003e2024\u003c/span\u003e; Cullmann \u003cspan citationid=\"CR11\" class=\"CitationRef\"\u003e2025\u003c/span\u003e; Streuobstportal Baden-W\u0026uuml;rttemberg \u003cspan citationid=\"CR35\" class=\"CitationRef\"\u003e2025\u003c/span\u003e). We therefore focused our bird surveys on these four major habitat types (Fig.\u0026nbsp;\u003cspan refid=\"Fig1\" class=\"InternalRef\"\u003e1\u003c/span\u003e, Fig.\u0026nbsp;\u003cspan refid=\"Fig2\" class=\"InternalRef\"\u003e2\u003c/span\u003e). In spring 2023, we identified 14 study landscapes, each measuring 1 x 1 km\u003csup\u003e2\u003c/sup\u003e, with the aim of covering a gradient of landscape diversity. Diverse landscapes had a balanced ratio of arable land, grassland, forest and orchards, while simplified landscapes consisted mainly of arable land (Fig.\u0026nbsp;\u003cspan refid=\"Fig2\" class=\"InternalRef\"\u003e2\u003c/span\u003ec). On average, 5.1% of the landscapes\u0026rsquo; areas were organically managed arable land (minimum: 0.2%, maximum: 21.9%) and 2.1% organically managed grassland (minimum: 0.1%, maximum: 6.6%). Organic management did not correlate with landscape diversity (Pearson\u0026rsquo;s r = \u0026ndash; 0.522). Overall, our study landscapes covered the regional range in terms of land cover. The mean pairwise distance between the centres of the landscapes was 10.5 km\u0026thinsp;\u0026plusmn;\u0026thinsp;4.7 km (SD).\u003c/p\u003e \u003cp\u003eThe classification of habitat types as grassland (defined as permanent grassland\u0026thinsp;\u0026gt;\u0026thinsp;5 years) or arable land was based on data from the state of Baden-W\u0026uuml;rttemberg (database for agricultural activities at field level from the Integrated Administration and Control System; Metainformationssystem GDI-BW \u003cspan citationid=\"CR26\" class=\"CitationRef\"\u003e2023\u003c/span\u003e). Traditional orchard meadows were classified based on the state-wide orchard survey (Borngraeber et al. \u003cspan citationid=\"CR5\" class=\"CitationRef\"\u003e2020\u003c/span\u003e). These are characterised by sparsely arranged, often older standard trees with grass cover in the undergrowth and made up most of the orchards in our study region. In contrast, intensively managed orchards were only present in one landscape. Other land cover types, including forests, settlements and roads, were mapped manually using QGIS v.3.36 (Fig.\u0026nbsp;\u003cspan refid=\"Fig2\" class=\"InternalRef\"\u003e2\u003c/span\u003eb). Although Baden-W\u0026uuml;rttemberg has densely populated areas, settlements covered only small areas in our rural study landscapes (mean\u0026thinsp;\u0026plusmn;\u0026thinsp;SD: 2.5% \u0026plusmn; 2%). During fieldwork, we verified the GIS-based habitat classifications and mapped hedges. We calculated landscape diversity as the Shannon diversity index for the final seven land cover types (farmland, forest, grassland, hedges, orchards, roads, settlements) using their proportional coverage as weights (Fig.\u0026nbsp;\u003cspan refid=\"Fig2\" class=\"InternalRef\"\u003e2\u003c/span\u003ec).\u003c/p\u003e \u003c/div\u003e\n\u003ch3\u003eRecording locations and local habitat features\u003c/h3\u003e\n\u003cp\u003eWe used a grid-based sampling design (Scherber et al. \u003cspan citationid=\"CR32\" class=\"CitationRef\"\u003e2019\u003c/span\u003e) to ensure representative data collection across the four major habitat types (i.e. arable land, grassland, forest, orchard) within the study landscapes. The number of recording locations per habitat type corresponded to its proportional cover in each landscape. To this end, we set up 16 recording locations in each study landscape (224 in total), which were spatially distributed as evenly as possible at intervals of approximately 200 m (Fig.\u0026nbsp;\u003cspan refid=\"Fig2\" class=\"InternalRef\"\u003e2\u003c/span\u003eb).\u003c/p\u003e \u003cp\u003eWe assigned each recording location to the main habitat type which dominated within a 100 m radius (Table \u003cspan refid=\"MOESM1\" class=\"InternalRef\"\u003eS1\u003c/span\u003e), as this approximates the effective detection range of our acoustic devices (Buchmann and Schurr \u003cspan citationid=\"CR7\" class=\"CitationRef\"\u003e2025\u003c/span\u003e). By using this grid-based sampling design, the resulting number of locations associated to a main habitat type closely matched the habitat proportions in the study landscapes, ensuring that our sampling reliably reflected landscape-wide habitat composition (Figure \u003cspan refid=\"MOESM1\" class=\"InternalRef\"\u003eS1\u003c/span\u003e). We removed the three recording locations assigned to the habitat type 'settlement' from all analyses due to the small sample size, leaving us with 221 recording locations. However, to keep comparable numbers of recording locations for each landscape, we used all 224 recording locations only in the calculations of gamma diversity.\u003c/p\u003e \u003cp\u003eAdditional to the main habitat type, we also assessed the local habitat diversity around the recording locations as the Shannon diversity of the seven land cover types in a 100 m radius around the recording locations (analogously to the calculation of landscape diversity, but at a different scale). Furthermore, we counted the number of tree individuals in a 20 m radius around the recording location.\u003c/p\u003e \u003cp\u003e \u003c/p\u003e \u003cp\u003ec)\u003c/p\u003e \u003cp\u003e \u003c/p\u003e\n\u003ch3\u003eBird recording\u003c/h3\u003e\n\u003cp\u003eDue to the large number of recording locations, we decided to use passive acoustic monitoring using AudioMoth recorders (Hill et al. \u003cspan citationid=\"CR20\" class=\"CitationRef\"\u003e2019\u003c/span\u003e). This monitoring method has been proven to provide results of equal accuracy and reliability as traditional point counts (Darras et al. \u003cspan citationid=\"CR13\" class=\"CitationRef\"\u003e2018\u003c/span\u003e; Toenies and Rich \u003cspan citationid=\"CR37\" class=\"CitationRef\"\u003e2021\u003c/span\u003e) while requiring much less workforce. We conducted our monitoring across three rounds of recordings in 2023 (23 March to 17 April; 15 May to 14 June; 3 July to 4 August). In each round, every landscape was recorded on two consecutive mornings for three hours each (starting 30 minutes before sunrise). All recorders were set up on poles at 1.5 m above ground facing eastward. We configured the AudioMoths to record sound up to a sampling rate of 192 kHz. Within each three-hour recording window, we recorded bird for 30 seconds, followed by a recording pause of 30 s. In the pooled data set, we consequently had 1,989 recorded hours. In case of bad weather conditions (strong wind or rain), we repeated recordings on the following days.\u003c/p\u003e\n\u003ch3\u003eAutomated species classification with BirdNET and human validation\u003c/h3\u003e\n\u003cp\u003eTo classify the bird species, we employed the artificial neural network \u0026lsquo;BirdNET\u0026rsquo;. Species classification with BirdNET requires consecutive human validation (Wood and Kahl \u003cspan citationid=\"CR43\" class=\"CitationRef\"\u003e2024\u003c/span\u003e), leading to a multi-step process.\u003c/p\u003e \u003cp\u003eIn the first step, we analysed the recordings in BirdNET Analyzer (version 2.4.0). Each 30-s recording was subdivided into 3-s snippets with 2-s overlap, yielding 28 snippets per recording to maximize detections (P\u0026eacute;rez-Granados et al. \u003cspan citationid=\"CR28\" class=\"CitationRef\"\u003e2025\u003c/span\u003e). We then ran the BirdNET algorithm on all snippets using default settings. The output consisted of detected species and associated confidence scores (0.10\u0026ndash;0.99) for each snippet. Across all snippets, this resulted in 6,076,185 detected bird calls.\u003c/p\u003e \u003cp\u003eSecond, we refined the automatic BirdNET species identifications by filtering them against a region-specific species list. Based on the geographic coordinates of the four corner points (NW, NE, SW, SE) and the centre of the study area, we queried BirdNET to generate species lists for each location. These four lists were merged into a comprehensive regional species pool. Three species were removed because they are extremely rare or not yet recorded in the region: \u003cem\u003eCettia cetti\u003c/em\u003e Temminck, \u003cem\u003ePtyonoprogne rupestris\u003c/em\u003e Scopoli, \u003cem\u003ePyrrhocorax graculus\u003c/em\u003e L.. The resulting list contained 168 species. Because water bodies were largely absent from our study landscapes and aquatic birds were not the focus of this study, we further excluded waterbird species. This resulted in a final list of 103 species considered potentially present in our dataset.\u003c/p\u003e \u003cp\u003eIn a third step, the BirdNET identifications of the 103 species required validation, as BirdNET confidence scores do not represent true probabilities of correct identifications and are not comparable across species (Wood and Kahl \u003cspan citationid=\"CR43\" class=\"CitationRef\"\u003e2024\u003c/span\u003e). For each species, a random subset of 50 segments spanning the full range of BirdNET confidence scores from 0.10 to 0.99 was selected using BirdNET\u0026rsquo;s \u003cem\u003esegments\u003c/em\u003e function (Wood and Kahl \u003cspan citationid=\"CR43\" class=\"CitationRef\"\u003e2024\u003c/span\u003e). These 50 snippets per species were then assessed aurally and visually to verify BirdNET\u0026rsquo;s identification and assigned a value of 1 for a correct identification or 0 for an incorrect one. Subsequently, a binomial generalized linear model was fitted for each species, with \u0026ldquo;correctness\u0026rdquo; as the dependent variable and the logit score as the explanatory variable (see Supplementary Figure for an example model). The logit score represents the original output of the deep neural network (Wood and Kahl \u003cspan citationid=\"CR43\" class=\"CitationRef\"\u003e2024\u003c/span\u003e) and allowed the model to perform better than using the confidence score. From each model, the logit score corresponding to a 90% probability of correct identification was derived and back-transformed to the confidence-score scale provided in the BirdNET outputs. Only BirdNET identifications exceeding the species-specific 90% probability threshold were retained for further analysis, which reduced the dataset to 66 species. A list of all validated species and their species-specific confidence thresholds is provided in Table \u003cspan refid=\"MOESM1\" class=\"InternalRef\"\u003eS1\u003c/span\u003e.\u003c/p\u003e \u003cp\u003eFinally, to ensure that only species using the habitat at each location were included, we removed species that were likely only flying over. We retained a species at a given location only if it was validated on two consecutive days in at least one of the three recording rounds. This filter reduced the dataset by nine species, leaving us with 57 species. We then merged \u003cem\u003eCorvus corone\u003c/em\u003e, \u003cem\u003eCorvus cornix\u003c/em\u003e, and \u003cem\u003eCorvus frugilegus\u003c/em\u003e into \u003cem\u003eCorvus\u003c/em\u003e spp., and \u003cem\u003ePasser domesticus\u003c/em\u003e and \u003cem\u003ePasser montanus\u003c/em\u003e into \u003cem\u003ePasser\u003c/em\u003e spp., because these species cannot be reliably distinguished by audio recordings alone (Buchmann and Schurr \u003cspan citationid=\"CR7\" class=\"CitationRef\"\u003e2025\u003c/span\u003e). The final dataset comprised 54 species or species groups (see Table S2).\u003c/p\u003e \u003cdiv id=\"Sec7\" class=\"Section2\"\u003e \u003ch2\u003eStatistical analysis\u003c/h2\u003e \u003cp\u003eAlpha diversity\u003c/p\u003e \u003cp\u003eWe quantified alpha diversity as the number of species detected at each recording location (Fig.\u0026nbsp;\u003cspan refid=\"Fig1\" class=\"InternalRef\"\u003e1\u003c/span\u003e). To assess whether local- and landscape-scale factors jointly shaped alpha diversity, we fitted a generalized linear mixed-effects model with a Poisson error distribution using R package lme4 (Bates et al. \u003cspan citationid=\"CR3\" class=\"CitationRef\"\u003e2015\u003c/span\u003e; R Core Team \u003cspan citationid=\"CR29\" class=\"CitationRef\"\u003e2024\u003c/span\u003e). The model included all two-way interactions between local variables (number of trees within 20 m radius, main habitat type, and local habitat diversity) and the landscape-level predictor (landscape diversity). All continuous variables were scaled and centred to account for differences in magnitude. Because recording locations were nested within landscapes, we included landscape identity as a random effect. Significant predictors were identified through stepwise backward model selection using chi-squared tests. The R\u003csup\u003e2\u003c/sup\u003e values for all statistical models were calculated using R package performance (L\u0026uuml;decke et al. \u003cspan citationid=\"CR23\" class=\"CitationRef\"\u003e2021\u003c/span\u003e).\u003c/p\u003e \u003cp\u003eBetween-habitat and within-habitat beta diversity\u003c/p\u003e \u003cp\u003eTo assess differences in the composition of birds, we focused on the 16 recording locations within each landscape. These differences, or beta diversity, can arise both within a single habitat type and between different habitat types (Fig.\u0026nbsp;\u003cspan refid=\"Fig1\" class=\"InternalRef\"\u003e1\u003c/span\u003e). We quantified within-habitat beta diversity (between pairs of locations of the same habitat type, e.g. orchard\u0026ndash;orchard) and between-habitat beta diversity (between pairs of locations of different habitat types, e.g. arable\u0026ndash;forest) by calculating S\u0026oslash;rensen dissimilarity indices for all pairs of recording locations within the same landscape. Each pair was then assigned both to its specific habitat combination (e.g. arable\u0026ndash;arable, forest\u0026ndash;orchard) and to the corresponding category (within- or between-habitat).\u003c/p\u003e \u003cp\u003eBased on the categorization (within- or between-habitat), we could then assess whether between-habitat beta diversity was generally larger or lower than within-habitat beta diversity. To this end, we fitted a linear mixed-effects model from R package lme4 (Bates et al. \u003cspan citationid=\"CR3\" class=\"CitationRef\"\u003e2015\u003c/span\u003e) with S\u0026oslash;rensen dissimilarities as response variable and the categorical variable (within- or between-habitat category) as explanatory variable. To avoid pseudoreplication, we also included the additive random effects of the IDs of both recording locations involved in the comparison. F-test and Post-Hoc Tukey test (R package emmeans; Lenth \u003cspan citationid=\"CR22\" class=\"CitationRef\"\u003e2024\u003c/span\u003e) revealed that within-habitat beta diversity values were significantly different from between-habitat beta diversity values.\u003c/p\u003e \u003cp\u003eWe also examined which habitat combinations exhibited relatively high or low dissimilarity. Due to the significant differences between categories, we fitted two separate models: one for within-habitat and one for between-habitat beta diversity. Habitat combination was used as the explanatory variable in both models. As before, we included the additive random effects of the IDs of both recording locations involved in the comparison. An F-test against the null model assessed whether S\u0026oslash;rensen dissimilarities differed significantly among habitat combinations. Significant differences were further evaluated using a post-hoc Tukey test for pairwise comparisons. To visualize patterns, we also performed a non-metric multidimensional scaling (NMDS) analysis based on 221 recording locations assigned to arable land, forest, grassland, and orchard as main habitats.\u003c/p\u003e \u003cp\u003eWe additionally examined the influence of local and landscape variables on beta diversity at the scale of individual recording locations using a generalized dissimilarity model (GDM). The GDM quantified how environmental predictors explained S\u0026oslash;rensen dissimilarities. Predictors included geographic coordinates of recording locations and landscape centres (UTM), local-scale variables (number of trees in 20 m radius; arable, forest, grassland, orchard, hedge, and settlement cover within 100 m; local habitat diversity), and landscape-scale area of all seven land cover types as well as landscape diversity. We fitted the GDM using five I-splines and included geographic information directly in the model. Variable importance was assessed with \u003cem\u003egdm.varImp\u003c/em\u003e from the R package gdm (Fitzpatrick et al. \u003cspan citationid=\"CR18\" class=\"CitationRef\"\u003e2025\u003c/span\u003e). For significant variables, we calculated percentage contributions for representation in a stacked plot.\u003c/p\u003e \u003cp\u003eGamma diversity\u003c/p\u003e \u003cp\u003eWe quantified gamma diversity as the total number of species detected per study landscape (Fig.\u0026nbsp;\u003cspan refid=\"Fig1\" class=\"InternalRef\"\u003e1\u003c/span\u003e) and examined whether landscape diversity influenced it. For this analysis, we fitted a generalized linear model with a Poisson error distribution and applied stepwise backward model selection based on Chi-squared tests. We also evaluated landscape-level sampling completeness using species accumulation and extrapolation curves (R package \u003cem\u003eiNEXT\u003c/em\u003e). Sampling completeness was calculated as the ratio of observed bird species richness to the estimated asymptotic richness (Chao \u003cspan citationid=\"CR8\" class=\"CitationRef\"\u003e1984\u003c/span\u003e) and subsequently correlated with landscape diversity.\u003c/p\u003e \u003cp\u003eFunctional traits and population trends of bird species\u003c/p\u003e \u003cp\u003eIn addition to analyses of overall species richness, we examined how species with different ecological traits were associated with habitat types. To this end, we conducted a non-metric multidimensional scaling (NMDS) analysis based on species identities and trait information from AVONET (Tobias et al. \u003cspan citationid=\"CR36\" class=\"CitationRef\"\u003e2022\u003c/span\u003e). Traits included habitat density preference (open, semi-open, dense; AVONET variable \u003cem\u003eHabitat.Density\u003c/em\u003e) and trophic niche (granivore, invertivore, omnivore, vertivore; AVONET variable \u003cem\u003eTrophic.Niche\u003c/em\u003e). Additionally, long-term population trends (decreasing, stable, increasing) were derived from Annex B of the German National Bird Protection Report (2019), where species classified as decreasing or increasing had lost or gained more than 20% of their population over the preceding 36 years (Table S2). We assessed the influence of the same environmental predictors used in the generalized dissimilarity modelling (GDM; see above) and retained only predictors with significant effects.\u003c/p\u003e \u003c/div\u003e"},{"header":"Results","content":"\u003cdiv id=\"Sec9\" class=\"Section2\"\u003e \u003ch2\u003eGeneral results\u003c/h2\u003e \u003cp\u003eWe investigated 2,016 h of audio recordings over a period of 4.5 months, in which we detected 54 bird species in total (species richness at a regional level). Fourteen of them occurred at \u0026gt;\u0026thinsp;50% of the recording locations, while 17 occurred at \u0026lt;\u0026thinsp;10% of the recording locations (Table S2). The most widespread species were \u003cem\u003eCorvus\u003c/em\u003e spp. (at 220 recording locations), \u003cem\u003eTurdus merula\u003c/em\u003e L. (at 219), \u003cem\u003eParus major\u003c/em\u003e L. (at 211) and \u003cem\u003eCyanistes caeruleus\u003c/em\u003e L. (at 200).\u003c/p\u003e \u003c/div\u003e\n\u003ch3\u003eAlpha diversity\u003c/h3\u003e\n\u003cp\u003eAlpha diversity represents the local species richness per recording location in a landscape (Fig.\u0026nbsp;\u003cspan refid=\"Fig1\" class=\"InternalRef\"\u003e1\u003c/span\u003e) and ranged between 6 and 30 species (mean\u0026thinsp;\u0026plusmn;\u0026thinsp;SD\u0026thinsp;=\u0026thinsp;18.2\u0026thinsp;\u0026plusmn;\u0026thinsp;5.1). Alpha diversity was driven by additive effects of the type of the main habitat (X\u003csup\u003e2\u003c/sup\u003e\u003csub\u003e3\u003c/sub\u003e\u0026thinsp;=\u0026thinsp;19.778, \u003cem\u003eP\u003c/em\u003e\u0026thinsp;\u0026lt;\u0026thinsp;0.001), the local habitat diversity (X\u003csup\u003e2\u003c/sup\u003e\u003csub\u003e1\u003c/sub\u003e\u0026thinsp;=\u0026thinsp;11.034, P\u0026thinsp;\u0026lt;\u0026thinsp;0.001) and landscape diversity (X\u003csup\u003e2\u003c/sup\u003e\u003csub\u003e1\u003c/sub\u003e\u0026thinsp;=\u0026thinsp;10.036, P\u0026thinsp;=\u0026thinsp;0.002). Orchards had the highest alpha diversity (predicted mean\u0026thinsp;=\u0026thinsp;20.7 species; 95% confidence interval: 18.9\u0026ndash;22.6 species), while alpha diversity was lowest for main habitat type arable land (16.5; 15.5\u0026ndash;17.7; Fig.\u0026nbsp;\u003cspan refid=\"Fig3\" class=\"InternalRef\"\u003e3\u003c/span\u003ea). Forest and grassland showed intermediate alpha diversity (17.9; 16.3\u0026ndash;19.5 and 19.4; 17.7\u0026ndash;21.2 respectively). Alpha diversity increased with local habitat diversity and landscape diversity, respectively (Fig.\u0026nbsp;\u003cspan refid=\"Fig3\" class=\"InternalRef\"\u003e3\u003c/span\u003eb\u0026thinsp;+\u0026thinsp;c). Local tree number and interactions between local and landscape variables were not significant. The model explained 39.4% of the variance based on fixed effects (marginal R\u0026sup2; = 0.394) and 45.4% when including the random landscape effect (conditional R\u0026sup2; = 0.454).\u003c/p\u003e \u003cp\u003e \u003c/p\u003e \u003cdiv id=\"Sec11\" class=\"Section2\"\u003e \u003ch2\u003eWithin-habitat and between-habitat beta diversity\u003c/h2\u003e \u003cp\u003eWithin-habitat beta diversity showed on average a S\u0026oslash;rensen dissimilarity of 0.297 (min\u0026thinsp;=\u0026thinsp;0.053; max\u0026thinsp;=\u0026thinsp;0.667; SD\u0026thinsp;=\u0026thinsp;0.110). The lowest within-habitat dissimilarity values were present for within-orchard and within-forest beta diversity (predicted means\u0026thinsp;\u0026plusmn;\u0026thinsp;SE\u0026thinsp;=\u0026thinsp;0.232\u0026thinsp;\u0026plusmn;\u0026thinsp;0.019 and 0.243\u0026thinsp;\u0026plusmn;\u0026thinsp;0.014, respectively; Fig.\u0026nbsp;\u003cspan refid=\"Fig4\" class=\"InternalRef\"\u003e4\u003c/span\u003eb; see also smallest ellipses for orchard and forest in NMDS Fig.\u0026nbsp;\u003cspan refid=\"Fig4\" class=\"InternalRef\"\u003e4\u003c/span\u003ea). Contrarily, within-arable showed highest dissimilarities (predicted means\u0026thinsp;\u0026plusmn;\u0026thinsp;SE\u0026thinsp;=\u0026thinsp;0.312\u0026thinsp;\u0026plusmn;\u0026thinsp;0.009; Fig.\u0026nbsp;\u003cspan refid=\"Fig4\" class=\"InternalRef\"\u003e4\u003c/span\u003eb; see also largest ellipses for arable land in NMDS Fig.\u0026nbsp;\u003cspan refid=\"Fig4\" class=\"InternalRef\"\u003e4\u003c/span\u003ea). Within-grassland beta diversity was not different from both extremes (predicted means\u0026thinsp;\u0026plusmn;\u0026thinsp;SE\u0026thinsp;=\u0026thinsp;0.279\u0026thinsp;\u0026plusmn;\u0026thinsp;0.020). The fixed effects in this model explained 8.3% of the variance in within-habitat beta diversity (marginal R\u003csup\u003e2\u003c/sup\u003e\u0026thinsp;=\u0026thinsp;0.083), which increased to 53.4% after accounting for the random effects of the recording locations (conditional R\u0026sup2; = 0.534).\u003c/p\u003e \u003cp\u003eBetween-habitat beta diversity was generally higher than within-habitat beta diversity, with an average S\u0026oslash;rensen dissimilarity of 0.333 (min\u0026thinsp;=\u0026thinsp;0.04; max\u0026thinsp;=\u0026thinsp;0.697; SD\u0026thinsp;=\u0026thinsp;0.110; Figs.\u0026nbsp;\u003cspan refid=\"Fig4\" class=\"InternalRef\"\u003e4\u003c/span\u003eb\u0026thinsp;+\u0026thinsp;c). Highest between-habitat dissimilarities were found for arable-forest comparisons and forest-orchard comparisons (predicted means\u0026thinsp;\u0026plusmn;\u0026thinsp;SE\u0026thinsp;=\u0026thinsp;0.353\u0026thinsp;\u0026plusmn;\u0026thinsp;0.013 and 0.382\u0026thinsp;\u0026plusmn;\u0026thinsp;0.013 respectively; Fig.\u0026nbsp;\u003cspan refid=\"Fig4\" class=\"InternalRef\"\u003e4\u003c/span\u003ec; see also least overlapping ellipses in NMDS Fig.\u0026nbsp;\u003cspan refid=\"Fig4\" class=\"InternalRef\"\u003e4\u003c/span\u003ea). Here, the fixed effects explained 9.9% of the variance in between-habitat beta diversity (marginal R\u003csup\u003e2\u003c/sup\u003e\u0026thinsp;=\u0026thinsp;0.099), which increased to 67.1% after accounting for the random effects of the recording locations (conditional R\u0026sup2; = 0.671).\u003c/p\u003e \u003cp\u003eThe generalized dissimilarity modelling (GDM) revealed that all local and landscape variables considered in this study jointly explained 41.2% of the pairwise beta-diversity of bird assemblages of each sampling location. However, there were only four significant variables explaining the beta-diversity patterns. These four predictors account together still for 40.9% of the explained deviance. Most of the explainable deviance was thus driven by local forest cover (60.8%), followed by local arable land cover (19.3%), landscape-wide orchard cover (12.5%) and landscape-wide forest cover (7.4%; Fig.\u0026nbsp;\u003cspan refid=\"Fig4\" class=\"InternalRef\"\u003e4\u003c/span\u003ed). By contrast, the geographic distance between sites and all other environmental local- and landscape-scale explanatory variables considered did not significantly explain pairwise beta-diversity.\u003c/p\u003e \u003cp\u003e \u003c/p\u003e \u003c/div\u003e \u003cdiv id=\"Sec12\" class=\"Section2\"\u003e \u003ch2\u003eGamma diversity\u003c/h2\u003e \u003cp\u003e \u003c/p\u003e \u003cp\u003eWe measured gamma diversity as the total species richness across all 16 recording locations in a landscape (Fig.\u0026nbsp;\u003cspan refid=\"Fig1\" class=\"InternalRef\"\u003e1\u003c/span\u003e). Gamma diversity ranged from 25 to 45 species (mean\u0026thinsp;\u0026plusmn;\u0026thinsp;SD\u0026thinsp;=\u0026thinsp;37.8\u0026thinsp;\u0026plusmn;\u0026thinsp;6.0). We sampled on average 87.01% (SD: \u0026plusmn; 9.61%, min: 71.53%, max: 99.42%) of the landscapes\u0026rsquo; estimated total richness. Sampling completeness of the bird species did not correlate with landscape diversity (Pearson\u0026rsquo;s r\u0026thinsp;=\u0026thinsp;\u0026minus;\u0026thinsp;0.30, P\u0026thinsp;=\u0026thinsp;0.300). However, gamma diversity increased with increasing landscape diversity (estimate\u0026thinsp;\u0026plusmn;\u0026thinsp;SE\u0026thinsp;=\u0026thinsp;0.565\u0026thinsp;\u0026plusmn;\u0026thinsp;0.179; X\u003csup\u003e2\u003c/sup\u003e\u003csub\u003e1\u003c/sub\u003e\u0026thinsp;=\u0026thinsp;10.326, \u003cem\u003eP\u003c/em\u003e\u0026thinsp;=\u0026thinsp;0.001; Nagelkerke\u0026rsquo;s R\u003csup\u003e2\u003c/sup\u003e\u0026thinsp;=\u0026thinsp;0.857; Fig.\u0026nbsp;\u003cspan refid=\"Fig5\" class=\"InternalRef\"\u003e5\u003c/span\u003e).\u003c/p\u003e \u003c/div\u003e \u003cdiv id=\"Sec13\" class=\"Section2\"\u003e \u003ch2\u003eBird preferences match habitat types\u003c/h2\u003e \u003cp\u003eThe NMDS ordination revealed consistent patterns in species\u0026rsquo; trophic niches, habitat-density preferences, and long-term population trends (Fig.\u0026nbsp;\u003cspan refid=\"Fig6\" class=\"InternalRef\"\u003e6\u003c/span\u003e). Omnivorous and invertivorous species showed no clear separation and occurred across all habitat types. In contrast, granivores clustered toward sites associated with open habitats, opposite to forested locations in the ordination space. Species preferring dense habitats were concentrated at forest-dominated sites with high tree cover, whereas open-habitat species grouped with arable and grassland sites on the opposite side of the NMDS. Species associated with semi-open habitats occupied intermediate positions between these extremes.\u003c/p\u003e \u003cp\u003eMost species in our dataset exhibited stable long-term population trends, while ten showed increasing trends and seven decreasing trends. Bird species with decreasing population trends (i.e. more than 20% decline in the last 36 years) were scattered broadly across the ordination, indicating that no single habitat type uniquely supports them.\u003c/p\u003e \u003cp\u003e \u003c/p\u003e \u003c/div\u003e"},{"header":"Discussion","content":"\u003cp\u003eIn this study, we aimed to disentangle the roles of local- and landscape-scale drivers of farmland bird diversity in Southern German agricultural landscapes. We used automated acoustic monitoring at 224 recording locations across 14 landscapes and analysed local alpha diversity at each location, within- and between-habitat beta diversity across recording locations and landscape-scale gamma diversity. Local alpha diversity was lowest in arable land (predicted mean\u0026thinsp;=\u0026thinsp;16.5 species) and highest in orchards (20.7 species). In contrast, within-habitat beta diversity was highest in arable land and grassland and lowest in orchards and forest. Between-habitat beta diversity exceeded within-habitat beta diversity, with stronger differences between structurally contrasting habitats (e.g. arable vs. forest) than between structurally more similar ones (e.g. arable vs. grassland). Consistent with these patterns, species preferring dense habitats occurred mainly in forest-dominated locations, whereas species preferring open habitats were associated with arable-dominated sites. Species with decreasing long-term population trends occurred across all habitat types. Generalised dissimilarity modelling indicated that local-scale variables (e.g. forest cover within a 100 m radius) were more important predictors of beta diversity than landscape-scale variables (e.g. total forest cover per 1 km\u0026sup2;). At the landscape scale, overall bird richness (gamma diversity) increased with landscape diversity.\u003c/p\u003e \u003cdiv id=\"Sec15\" class=\"Section2\"\u003e \u003ch2\u003eAlpha diversity of bird assemblages\u003c/h2\u003e \u003cp\u003eAcross habitat types, we observed contrasts in local species richness that reflect differences in management intensity, structural complexity, and resource availability. For example, orchards had higher bird species richness than arable land (Fig.\u0026nbsp;\u003cspan refid=\"Fig3\" class=\"InternalRef\"\u003e3\u003c/span\u003e) which aligns with previous studies (Wilson et al. \u003cspan citationid=\"CR42\" class=\"CitationRef\"\u003e2020\u003c/span\u003e; Edo et al. \u003cspan citationid=\"CR16\" class=\"CitationRef\"\u003e2024\u003c/span\u003e). These contrasting patterns in species richness between habitat types highlight fundamental differences between intensively managed, structurally simple habitats and more heterogeneous systems with high niche availability.\u003c/p\u003e \u003cp\u003eMost of the species detected at more than 90% of the arable locations were widespread generalists, such as the greenfinch (\u003cem\u003eChloris chloris\u003c/em\u003e (L.)), crows (\u003cem\u003eCorvus\u003c/em\u003e spp.), the great tit (\u003cem\u003eParus major\u003c/em\u003e L.), and the common blackbird (\u003cem\u003eTurdus merula\u003c/em\u003e L.) (Table S2). This may be because arable land in our study region is predominantly managed conventionally (11% of arable land was farmed organically). Conventional management can reduce the availability of food resources (e.g. insects) and cause frequent disturbance in homogeneous fields which offer few suitable breeding sites (Benton et al. \u003cspan citationid=\"CR4\" class=\"CitationRef\"\u003e2003\u003c/span\u003e). Consequently, the species that are still present in arable land, need to be able to cope with these conditions, favouring omnivorous, granivorous and highly mobile species. In contrast, traditional orchards are extensively managed and provide a wide variety of niches, especially through old trees that are valuable for cavity-nesting species (Edo et al. \u003cspan citationid=\"CR16\" class=\"CitationRef\"\u003e2024\u003c/span\u003e). This high niche availability can also favour insect diversity and abundance (Sattler et al. \u003cspan citationid=\"CR31\" class=\"CitationRef\"\u003e2024\u003c/span\u003e), enabling a higher share of invertivorous bird species in orchards. Consequently, a broad range of species were recorded primarily in orchards, including short-toed treecreeper (\u003cem\u003eCerthia brachydactyla\u003c/em\u003e Brehm), European goldfinch (\u003cem\u003eCarduelis carduelis\u003c/em\u003e (L.)), great spotted woodpecker (\u003cem\u003eDendrocopos major\u003c/em\u003e (L.)), common nightingale (\u003cem\u003eLuscinia megarhynchos\u003c/em\u003e Brehm), common redstart \u003cem\u003e(Phoenicurus phoenicurus\u003c/em\u003e (L.)) and common starling (\u003cem\u003eSturnus vulgaris\u003c/em\u003e L.) (Table S2). This high overall richness of orchards underlines both their high conservation value and their vulnerability, as traditional orchards are classified as endangered according to the Red List of Biotope Types in Baden-W\u0026uuml;rttemberg (Breunig et al. \u003cspan citationid=\"CR6\" class=\"CitationRef\"\u003e2020\u003c/span\u003e).\u003c/p\u003e \u003cp\u003eGrassland and forest sites showed intermediate richness levels (Fig.\u0026nbsp;\u003cspan refid=\"Fig3\" class=\"InternalRef\"\u003e3\u003c/span\u003e) between those of arable land and orchards. Grasslands showing lower bird richness than orchards is consistent with previous research (Hartel et al. \u003cspan citationid=\"CR19\" class=\"CitationRef\"\u003e2014\u003c/span\u003e) and likely reflects the relatively intensive management of grasslands in the region, including frequent mowing and high fertilization. Such management has contributed to declines in formerly common open-land species such as the Northern Lapwing \u003cem\u003eVanellus vanellus\u003c/em\u003e (McKeever \u003cspan citationid=\"CR25\" class=\"CitationRef\"\u003e2003\u003c/span\u003e). Opposite to Edo et al. (\u003cspan citationid=\"CR16\" class=\"CitationRef\"\u003e2024\u003c/span\u003e), we found that forests showed lower alpha diversity than orchards, as traditional orchards were structurally more diverse than forests in our study landscapes. This structural diversity leads to a higher availability of different niches, profiting both species of open landscapes and light woodlands and thus promoting a higher overall species richness (Edo et al. \u003cspan citationid=\"CR16\" class=\"CitationRef\"\u003e2024\u003c/span\u003e). In our study landscapes, forests are often intensively managed but mostly mixed forests that can support many invertivorous and omnivorous forest-associated species, including the common treecreeper \u003cem\u003eCerthia familiaris\u003c/em\u003e L., the wood pigeon \u003cem\u003eColumba palumbus\u003c/em\u003e L., black woodpecker \u003cem\u003eDryocopus martius\u003c/em\u003e L., the robin \u003cem\u003eErithacus rubecula\u003c/em\u003e L., and song thrush \u003cem\u003eTurdus philomelos\u003c/em\u003e Brehm (Table S2).\u003c/p\u003e \u003cp\u003eBeyond the effect of the locally dominant habitat type, both local habitat diversity within a 100-m radius and overall landscape diversity within the 1 \u0026times; 1 km\u003csup\u003e2\u003c/sup\u003e area increased bird alpha diversity (Fig.\u0026nbsp;\u003cspan refid=\"Fig3\" class=\"InternalRef\"\u003e3\u003c/span\u003e). Some studies support this finding (Barbaro et al. \u003cspan citationid=\"CR2\" class=\"CitationRef\"\u003e2005\u003c/span\u003e; Anderle et al. \u003cspan citationid=\"CR1\" class=\"CitationRef\"\u003e2022\u003c/span\u003e), although most studies could not confirm this relationship (Cours and Duflot \u003cspan citationid=\"CR10\" class=\"CitationRef\"\u003e2025\u003c/span\u003e). Our findings align with the cross-habitat spillover hypothesis in heterogeneous landscapes (Tscharntke et al. \u003cspan citationid=\"CR39\" class=\"CitationRef\"\u003e2012\u003c/span\u003e). At the local scale, birds may benefit from easier access to complementary resources from neighbouring habitat types, such as insects, fruits, or seeds, as well as from a broader range of nesting opportunities (i.e. habitat complementation, Dunning et al. \u003cspan citationid=\"CR15\" class=\"CitationRef\"\u003e1992\u003c/span\u003e). For example, Skylarks (\u003cem\u003eAlauda arvensis\u003c/em\u003e) may nest in arable fields but forage in adjacent grasslands, while Great Tits (\u003cem\u003eParus major\u003c/em\u003e) breed in forest patches yet feed in nearby orchards or hedgerows. Such local habitat heterogeneity can therefore enhance bird richness even in otherwise intensively used farmland, for instance when small woody features or orchard trees mitigate the structural simplicity of arable land. Considering the landscape scale, those landscapes with a higher landscape diversity reflected increasing shares of species-rich habitats such as orchards and forests. This might lead to spillover from these high-richness habitat types to normally low-richness habitats which might increase bird richness in the latter. Finally, landscapes with higher compositional diversity typically have higher landscape connectivity which might facilitate between-habitat movement and local species assembly (Tscharntke et al. \u003cspan citationid=\"CR38\" class=\"CitationRef\"\u003e2021\u003c/span\u003e).\u003c/p\u003e \u003c/div\u003e \u003cdiv id=\"Sec16\" class=\"Section2\"\u003e \u003ch2\u003eBeta diversity of bird assemblages\u003c/h2\u003e \u003cp\u003eWhile differences in alpha diversity of birds between habitat types are well established, much less is known about variability in bird assemblages among locations of the same habitat type (Edo et al. \u003cspan citationid=\"CR16\" class=\"CitationRef\"\u003e2024\u003c/span\u003e; Cours and Duflot \u003cspan citationid=\"CR10\" class=\"CitationRef\"\u003e2025\u003c/span\u003e). We could show that beta diversity within and between habitat types can substantially shape bird assemblages in agricultural landscapes. Despite supporting relatively low bird richness, arable land exhibited the highest within-habitat beta diversity (Fig.\u0026nbsp;\u003cspan refid=\"Fig4\" class=\"InternalRef\"\u003e4\u003c/span\u003e), indicating strong turnover in bird assemblages among arable locations. This pattern likely reflects heterogeneity in arable management, including differences in crop types and their phenologies (e.g. winter wheat sown in autumn versus maize sown in spring) and management intensity (e.g. sowing density). Such fine-scale management variation may promote distinct bird assemblages even among arable-dominated locations. For example, the Western Yellow Wagtail \u003cem\u003e(Motacilla flava)\u003c/em\u003e persists well in intensively managed cereal fields (Kragten \u003cspan citationid=\"CR21\" class=\"CitationRef\"\u003e2011\u003c/span\u003e), whereas the Eurasian Skylark requires unsown gaps within the cereal fields (Morris et al. \u003cspan citationid=\"CR27\" class=\"CitationRef\"\u003e2004\u003c/span\u003e).\u003c/p\u003e \u003cp\u003eIn both forest- and orchard-dominated locations, within-habitat beta diversity was significantly lower than in arable land (ca. \u0026ndash; 25%; Fig.\u0026nbsp;\u003cspan refid=\"Fig4\" class=\"InternalRef\"\u003e4\u003c/span\u003e). This contrasts with the findings of Edo et al. (\u003cspan citationid=\"CR16\" class=\"CitationRef\"\u003e2024\u003c/span\u003e), who reported higher within-forest than within-orchard beta diversity. Our results further indicate that orchards, despite supporting high bird alpha diversity, host relatively similar bird assemblages across locations. Even though varying structure and management of traditional orchards can lead to varying habitat quality for bird species (Chaparro et al. \u003cspan citationid=\"CR9\" class=\"CitationRef\"\u003e2022\u003c/span\u003e), our findings suggest that this habitat type can consistently provide suitable conditions for a broad range of species. Like orchards, within-habitat beta diversity in forest-dominated locations was relatively low as both habitat types are more stable over time than arable land. Furthermore, the forests\u0026rsquo; limited understory vegetation, little or no deadwood, scarce floral resources, and relatively uniform stand ages might have contributed to the low dissimilarity in bird assemblages across forest sites.\u003c/p\u003e \u003cp\u003eBetween-habitat beta diversity of bird assemblages exceeded within-habitat beta diversity. The habitats in this study spanned a gradient of structural complexity: from open arable land and grassland, through semi-open orchards, to dense forest stands. As the ecological contrast (Marja et al. \u003cspan citationid=\"CR24\" class=\"CitationRef\"\u003e2019\u003c/span\u003e) between habitats increased along this gradient, the dissimilarity in bird assemblages also increased. For example, arable-forest beta diversity was higher than arable-grassland beta diversity. This pattern is likely driven by trait-specific responses to differences in resource availability and habitat structure. For example, arable-dominated locations were primarily characterized by granivorous species (e.g. \u003cem\u003ePasser\u003c/em\u003e spp.) adapted to open habitats, whereas forest-dominated locations were characterized by insectivorous species associated with dense habitats (e.g. common treecreeper \u003cem\u003eCerthia familiaris\u003c/em\u003e; Table S2).\u003c/p\u003e \u003cp\u003eBeta diversity between arable land and grassland was rather low, as they are both open habitats with little or no trees and experience both frequent disturbances, such as repeated machinery use for fertilization, pesticide application, or mowing. This smaller ecological contrast likely explains the lower dissimilarity in bird assemblages observed between these two open habitat types. Bird assemblages in arable landscapes were dominated by multi-habitat users, reflecting the need to cope with frequent and often unpredictable disturbances in agricultural systems. In such environments, ecological generalists are favoured that can exploit multiple resources, often across different habitat types. This can result in a low prevalence of specialists. In contrast, forest ecosystems are comparatively stable over time, allowing for narrower ecological niches and supporting a higher proportion of habitat specialists associated with forest conditions.\u003c/p\u003e \u003cp\u003eFinally, also the results of a generalized dissimilarity model (GDM) showed that local predictors such as the amount of forest or arable land within a 100 m radius of the sampling location were more important in explaining bird beta diversity than landscape level predictors (Fig.\u0026nbsp;\u003cspan refid=\"Fig4\" class=\"InternalRef\"\u003e4\u003c/span\u003e). Thus, local ecological contrasts, particularly those between arable land and forest, appear to be the principal drivers of beta diversity in farmland bird assemblages.\u003c/p\u003e \u003c/div\u003e \u003cdiv id=\"Sec17\" class=\"Section2\"\u003e \u003ch2\u003eGamma diversity of bird assemblages\u003c/h2\u003e \u003cp\u003eOverall, our results empirically prove that different habitat types support distinct bird assemblages. Landscapes with a greater diversity of habitat types sustain a higher gamma diversity than simplified landscapes dominated by only a few, often uniformly managed habitats (Fig.\u0026nbsp;\u003cspan refid=\"Fig5\" class=\"InternalRef\"\u003e5\u003c/span\u003e). This relationship suggests that heterogeneous landscapes provide a broader range of available niches, thereby supporting both habitat specialists and multi-habitat users (Cours and Duflot \u003cspan citationid=\"CR10\" class=\"CitationRef\"\u003e2025\u003c/span\u003e). Increasing evidence further indicates that such landscape-level heterogeneity is essential not only for maintaining species richness but also for preserving the functional resilience of bird communities and the ecosystem services they provide (Weeks et al. \u003cspan citationid=\"CR41\" class=\"CitationRef\"\u003e2025\u003c/span\u003e).\u003c/p\u003e \u003c/div\u003e"},{"header":"Conclusion","content":"\u003cp\u003eOur study showed that habitat heterogeneity across local and the landscape scales (i.e. landscape diversity) is essential for maintaining farmland bird diversity at multiple scales, as it enhances bird richness and promotes bird community assembly. Variation in bird community composition (i.e. beta diversity) was driven primarily by local habitat characteristics and ecological contrasts among habitats and is a key component of biodiversity in agricultural landscapes. Therefore, it is important to preserve structurally diverse and heterogeneous landscapes, including a range of habitat types to sustain farmland bird communities.\u003c/p\u003e"},{"header":"Declarations","content":"\u003cp\u003e \u003ch2\u003eCompeting interests\u003c/h2\u003e \u003cp\u003eThe authors have no relevant financial or non-financial interests to disclose.\u003c/p\u003e \u003c/p\u003e\u003cp\u003e \u003ch2\u003eDava availability\u003c/h2\u003e \u003cp\u003eThe datasets generated during the current study are available in the supplement.\u003c/p\u003e \u003c/p\u003e\u003ch2\u003eFunding\u003c/h2\u003e \u003cp\u003eThe research was conducted within the cooperative, interdisciplinary doctoral program \u0026ldquo;Leverage points for Biodiversity Enhancement in Agricultural Landscapes (HABIT)\u0026rdquo; of the University of Hohenheim and N\u0026uuml;rtingen-Geislingen University, funded by the Ministry for Science, Research and Arts of Baden-W\u0026uuml;rttemberg as part of the State Postgraduate Scholarship Program (project number BW6_09).\u003c/p\u003e\u003ch2\u003eAuthor Contribution\u003c/h2\u003e\u003cp\u003eM.K.K. and I.G. conceived the ideas and designed the methodology; M.K.K., S.T., and R.B. collected the data; M.K.K., I.G., and T.H. analyzed the data; M.K.K. and I.G. led the writing of the manuscript. All authors contributed critically to the drafts and approved the final manuscript.\u003c/p\u003e\u003ch2\u003eAcknowledgement\u003c/h2\u003e\u003cp\u003eThe research was conducted within the cooperative, interdisciplinary doctoral program \u0026ldquo;Leverage points for Biodiversity Enhancement in Agricultural Landscapes (HABIT)\u0026rdquo; of the University of Hohenheim and N\u0026uuml;rtingen-Geislingen University, funded by the Ministry for Science, Research and Arts of Baden-W\u0026uuml;rttemberg as part of the State Postgraduate Scholarship Program (project number BW6_09). We thank Yasha Auer, Carlos Gonzalez and Pauline Greiner for assistance during field work.\u003c/p\u003e\u003ch2\u003eData Availability\u003c/h2\u003e\u003cp\u003eThe datasets generated for this study are available in the supplement.\u003c/p\u003e"},{"header":"References","content":"\u003col\u003e\u003cli\u003e\u003cspan\u003eAnderle M, Paniccia C, Brambilla M, Hilpold A, Volani S, Tasser E, Seeber J, Tappeiner U (2022) The contribution of landscape features, climate and topography in shaping taxonomical and functional diversity of avian communities in a heterogeneous Alpine region. Oecologia, 199(3):499\u0026ndash;512\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eBarbaro L, Pontcharraud L, Vetillard F, Guyon D, Jactel H (2005) Comparative responses of bird, carabid, and spider assemblages to stand and landscape diversity in maritime pine plantation forests. \u0026Eacute;coscience, 12(1):110\u0026ndash;121\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eBates D, M\u0026auml;chler M, Bolker B, Walker S (2015) Fitting Linear Mixed-Effects Models Using lme4. J. Stat. Soft., 67(1):1\u0026ndash;48\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eBenton TG, Vickery JA, Wilson JD (2003) Farmland biodiversity: is habitat heterogeneity the key? Trends in Ecology \u0026amp; Evolution, 18(4):182\u0026ndash;188\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eBorngraeber S, Krismann A, Schmieder K (2020) Ermittlung der Streuobstbest\u0026auml;nde Baden-W\u0026uuml;rttembergs durch automatisierte Fernerkundungsverfahren. In: LUBW (ed) Naturschutz und Landschaftspflege Baden-W\u0026uuml;rttemberg. Bd. 81: Sammelband mit mehreren Einzelbeitr\u0026auml;gen. Naturschutz und Landschaftspflege Baden-W\u0026uuml;rttemberg, vols 81, Karlsruhe, pp 4\u0026ndash;22\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eBreunig T, Demuth S, Cordlandwehr V (2020) Rote Liste der Biotoptypen Baden-W\u0026uuml;rttembergs mit naturschutzfachlicher Beurteilung. Available from \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://pd.lubw.de/10247\u003c/span\u003e\u003cspan address=\"https://pd.lubw.de/10247\" targettype=\"URL\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eBuchmann CM, Schurr FM (2025) An automated workflow that combines sound-based bird identification and localization. IBIS\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eChao A (1984) Nonparametric Estimation of the Number of Classes in a Population. Scandinavian Journal of Statistics, 11:265\u0026ndash;270\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eChaparro L, Schmieder K, Schurr FM (2022) The value of remotely sensed vs. field-surveyed habitat structure for predicting bird abundance: a case study in traditional orchards. J Ornithol, 163(3):723\u0026ndash;733\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eCours J, Duflot R (2025) Effects of landscape heterogeneity on bird communities in temperate, boreal, and montane forests \u0026ndash; a review. Journal of Avian Biology, 2025(3)\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eCullmann D (2025) Der Wald in Baden-W\u0026uuml;rttemberg: Ausgew\u0026auml;hlte Ergebnisse und regionale Auswertungen der Bundeswaldinventur 2022, Freiburg\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eDambros CS, Morais JW, Azevedo RA, Gotelli NJ (2017) Isolation by distance, not rivers, control the distribution of termite species in the Amazonian rain forest. Ecography, 40(10):1242\u0026ndash;1250\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eDarras K, Bat\u0026aacute;ry P, Furnas B, Celis-Murillo A, van Wilgenburg SL, Mulyani YA, Tscharntke T (2018) Comparing the sampling performance of sound recorders versus point counts in bird surveys: A meta‐analysis. J Appl Ecol, 55(6):2575\u0026ndash;2586\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eDuflot R, San-Cristobal M, Andrieu E, Choisis J-P, Esquerr\u0026eacute; D, Ladet S, Ouin A, Rivers-Moore J, Sheeren D, Sirami C, Fauvel M, Vialatte A (2022) Farming intensity indirectly reduces crop yield through negative effects on agrobiodiversity and key ecological functions. Agriculture, Ecosystems \u0026amp; Environment, 326:107810\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eDunning JB, Danielson BT, Pulliam HR (1992) Ecological Processes That Affect Populations in Complex Landscapes. Oikos, 65(1):169\u0026ndash;175\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eEdo M, Entling MH, R\u0026ouml;sch V (2024) Agroforestry supports high bird diversity in European farmland. Agron. Sustain. Dev., 44(1)\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eFahrig L (2013) Rethinking patch size and isolation effects: the habitat amount hypothesis. Journal of Biogeography, 40(9):1649\u0026ndash;1663\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eFitzpatrick M, Mokany K, Manion G, Nieto-Lugilde D, Ferrier S (2025) gdm: Generalized Dissimilarity Modeling\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eHartel T, Hanspach J, Abson DJ, M\u0026aacute;th\u0026eacute; O, Moga CI, Fischer J (2014) Bird communities in traditional wood-pastures with changing management in Eastern Europe. Basic and Applied Ecology, 15(5):385\u0026ndash;395\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eHill AP, Prince P, Snaddon JL, Doncaster CP, Rogers A (2019) AudioMoth: A low-cost acoustic device for monitoring biodiversity and the environment. HardwareX, 6:e00073\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eKragten S (2011) Shift in crop preference during the breeding season by Yellow Wagtails Motacilla flava flava on arable farms in The Netherlands. J Ornithol, 152(3):751\u0026ndash;757\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eLenth RV (2024) emmeans: Estimated Marginal Means, aka Least-Squares Means: R package version 1.10.5. Available from \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://CRAN.R-project.org/package=emmeans\u003c/span\u003e\u003cspan address=\"https://CRAN.R-project.org/package=emmeans\" targettype=\"URL\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eL\u0026uuml;decke D, Ben-Shachar M, Patil I, Waggoner P, Makowski D (2021) performance: An R Package for Assessment, Comparison and Testing of Statistical Models. JOSS, 6(60):3139\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eMarja R, Kleijn D, Tscharntke T, Klein A-M, Frank T, Bat\u0026aacute;ry P (2019) Effectiveness of agri-environmental management on pollinators is moderated more by ecological contrast than by landscape structure or land-use intensity. Ecol Letters, 22(9):1493\u0026ndash;1500\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eMcKeever CU (2003) Linking grassland management, invertebrates and Northern Lapwing productivity\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eMetainformationssystem GDI-BW (2023) WFS Fl\u0026auml;cheninformationen und Online-Antrag Baden-W\u0026uuml;rttemberg (FIONA). Available from \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://metadaten.geoportal-bw.de/geonetwork/srv/ger/catalog.search#/home\u003c/span\u003e\u003cspan address=\"https://metadaten.geoportal-bw.de/geonetwork/srv/ger/catalog.search#/home\" targettype=\"URL\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e (accessed February 2025)\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eMorris AJ, Holland JM, Smith B, Jones NE (2004) Sustainable Arable Farming For an Improved Environment (SAFFIE): managing winter wheat sward structure for Skylarks Alauda arvensis. IBIS, 146(s2):155\u0026ndash;162\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eP\u0026eacute;rez-Granados C, Funosas D, Morant J, Mar\u0026iacute;n G\u0026oacute;mez OH, Mendoza I, Mohedano‐Munoz MA, Santamar\u0026iacute;a E, Bastianelli G, M\u0026aacute;rquez‐Rodr\u0026iacute;guez A, Budka M, Bota G, La Pe\u0026ntilde;a‐Rubio JM de, La Garc\u0026iacute;a De Morena E, Santa‐Cruz M, La Nava P de, Fern\u0026aacute;ndez‐Tiz\u0026oacute;n M, S\u0026aacute;nchez‐Mateos H, Barrero A, Traba J, Osiejuk TS, Hart PJ, Navine AK, Montoya Mu\u0026ntilde;oz AF, Ara\u0026uacute;jo CB de, Rosa GLM, Torres IMD, Catalano ALC, Sim\u0026otilde;es CR, Llusia D, Morales MB, Acebes P, Medina JA, Brown N, Astaras C, Karmiris I, Navarrete E, Cauchoix M, Barbaro L, Arend D, M\u0026uuml;eller S, Gonz\u0026aacute;lez‐Garc\u0026iacute;a F, Gonz\u0026aacute;lez‐Romero A, Mammides C, Pontikis M, Jacuzzi G, Olden JD, Bombaci SP, Marcacci G, Jacot A, Zurano JP, Gangenova E, Varela D, Di Sallo F, Zurita GA, Atemasov A, Tremblay JA, Hutschenreiter A, Monroy‐Ojeda A, D\u0026iacute;az‐Vallejo M, Chaparro‐Herrera S, Briers RA, Sousa‐Lima R, Pinheiro T, Da Silva WC, Calvente A, Dal Molin A, Antonelli A, Gogoleva S, Palko I, Trong HV, Duarte MHL, Saturnino NDS, Silva SR, Rainho A, Lopes P, Schuchmann K-L, Marques MI, Oliverira Tissiani AS de, Littlewood NA, Tuanmu M-N, Cheng Y-R, Chao H, Kepfer‐Rojas S, Aguilera AL, Brotons L, Feldman MJ, Imbeau L, Panwar P, Weed AS, Dehwal A, Attisano A, Theuerkauf J, Oliveira‐J\u0026uacute;nior DD, Lima‐Santos CS, Salustio‐Gomes C, Paz RV, Pichorim M, Goodale E, Sebasti\u0026aacute;n‐Gonz\u0026aacute;lez E (2025) Optimization of passive acoustic bird surveys: a global assessment of BirdNET settings. IBIS\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eR Core Team (2024) R: A Language and Environment for Statistical Computing. R Foundation for Statistical Computing, Vienna, Austria\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eRigal S, Dakos V, Alonso H, Auinins A, Benko Z, et al. (2023) Farmland practices are driving bird population decline across Europe. PNAS, 120(21)\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eSattler C, Schrader J, H\u0026uuml;ttner M-L, Henle K (2024) Effects of management, habitat and landscape characteristics on biodiversity of orchard meadows in Central Europe: a brief review. Nature Conservation, 55:103\u0026ndash;134\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eScherber C, Beduschi T, Tscharntke T (2019) Novel approaches to sampling pollinators in whole landscapes: a lesson for landscape-wide biodiversity monitoring. Landscape Ecol, 34(5):1057\u0026ndash;1067\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eSirami C, Gross N, Baillod AB, Bertrand C, Carri\u0026eacute; R, Hass A, Henckel L, Miguet P, Vuillot C, Alignier A, Girard J, Bat\u0026aacute;ry P, Clough Y, Violle C, Giralt D, Bota G, Badenhausser I, Lefebvre G, Gauffre B, Vialatte A, Calatayud F, Gil-Tena A, Tischendorf L, Mitchell S, Lindsay K, Georges R, Hilaire S, Recasens J, Sol\u0026eacute;-Senan XO, Roble\u0026ntilde;o I, Bosch J, Barrientos JA, Ricarte A, Marcos-Garcia M\u0026Aacute;, Mi\u0026ntilde;ano J, Mathevet R, Gibon A, Baudry J, Balent G, Poulin B, Burel F, Tscharntke T, Bretagnolle V, Siriwardena G, Ouin A, Brotons L, Martin J-L, Fahrig L (2019) Increasing crop heterogeneity enhances multitrophic diversity across agricultural regions. Proceedings of the National Academy of Sciences of the United States of America, 116(33):16442\u0026ndash;16447\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eStatistisches Landesamt Baden-W\u0026uuml;rttemberg (2024) Anbaufl\u0026auml;chen aller Kultur- und Nutzungsarten seit 2010. Available from \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://www.statistik-bw.de/Landwirtschaft/Bodennutzung/LF-NutzngKultFrucht.jsp\u003c/span\u003e\u003cspan address=\"https://www.statistik-bw.de/Landwirtschaft/Bodennutzung/LF-NutzngKultFrucht.jsp\" targettype=\"URL\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e (accessed January 2025)\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eStreuobstportal Baden-W\u0026uuml;rttemberg (2025) Landesweite Streuobstdatenerhebung. Available from \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://streuobst.landwirtschaft-bw.de/,Lde/Startseite/Wissen/Landesweite+Streuobstdatenerhebung\u003c/span\u003e\u003cspan address=\"https://streuobst.landwirtschaft-bw.de/,Lde/Startseite/Wissen/Landesweite+Streuobstdatenerhebung\" targettype=\"URL\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e (accessed January 2025)\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eTobias JA, Sheard C, Pigot AL, Devenish AJM, Yang J, Sayol F, Neate-Clegg MHC, Alioravainen N, Weeks TL, Barber RA, Walkden PA, MacGregor HEA, Jones SEI, Vincent C, Phillips AG, Marples NM, Monta\u0026ntilde;o-Centellas FA, Leandro-Silva V, Claramunt S, Darski B, Freeman BG, Bregman TP, Cooney CR, Hughes EC, Capp EJR, Varley ZK, Friedman NR, Korntheuer H, Corrales-Vargas A, Trisos CH, Weeks BC, Hanz DM, T\u0026ouml;pfer T, Bravo GA, Remeš V, Nowak L, Carneiro LS, Moncada R AJ, Matysiokov\u0026aacute; B, Baldassarre DT, Mart\u0026iacute;nez-Salinas A, Wolfe JD, Chapman PM, Daly BG, Sorensen MC, Neu A, Ford MA, Mayhew RJ, Fabio Silveira L, Kelly DJ, Annorbah NND, Pollock HS, Grabowska-Zhang AM, McEntee JP, Carlos T Gonzalez J, Meneses CG, Mu\u0026ntilde;oz MC, Powell LL, Jamie GA, Matthews TJ, Johnson O, Brito GRR, Zyskowski K, Crates R, Harvey MG, Jurado Zevallos M, Hosner PA, Bradfer-Lawrence T, Maley JM, Stiles FG, Lima HS, Provost KL, Chibesa M, Mashao M, Howard JT, Mlamba E, Chua MAH, Li B, G\u0026oacute;mez MI, Garc\u0026iacute;a NC, P\u0026auml;ckert M, Fuchs J, Ali JR, Derryberry EP, Carlson ML, Urriza RC, Brzeski KE, Prawiradilaga DM, Rayner MJ, Miller ET, Bowie RCK, Lafontaine R-M, Scofield RP, Lou Y, Somarathna L, Lepage D, Illif M, Neuschulz EL, Templin M, Dehling DM, Cooper JC, Pauwels OSG, Analuddin K, Fjelds\u0026aring; J, Seddon N, Sweet PR, DeClerck FAJ, Naka LN, Brawn JD, Aleixo A, B\u0026ouml;hning-Gaese K, Rahbek C, Fritz SA, Thomas GH, Schleuning M (2022) AVONET: morphological, ecological and geographical data for all birds. Ecol Letters, 25(3):581\u0026ndash;597\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eToenies M, Rich LN (2021) Advancing bird survey efforts through novel recorder technology and automated species identification. California Fish and Wildlife, 107(2):56\u0026ndash;70\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eTscharntke T, Grass I, Wanger TC, Westphal C, Bat\u0026aacute;ry P (2021) Beyond organic farming - harnessing biodiversity-friendly landscapes. Trends in ecology \u0026amp; evolution, 36(10):919\u0026ndash;930\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eTscharntke T, Tylianakis JM, Rand TA, Didham RK, Fahrig L, Bat\u0026aacute;ry P, Bengtsson J, Clough Y, Crist TO, Dormann CF, Ewers RM, Fr\u0026uuml;nd. Jochen, Holt RD, Holzschuh A, Klein AM, Kleijn D, Kremen C, Landis DA, Laurance W, Lindenmayer D, Scherber C, Sodhi N, Steffan-Dewenter I, Thies C, van der Putten WH, Westphal C (2012) Landscape moderation of biodiversity patterns and processes - eight hypotheses. Biological Reviews, 87:661\u0026ndash;685\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eValente JJ, Gannon DG, Hightower J, Kim H, Leimberger KG, Macedo R, Rousseau JS, Weldy MJ, Zitomer RA, Fahrig L, Fletcher RJ, Wu J, Betts MG (2023) Toward conciliation in the habitat fragmentation and biodiversity debate. Landsc Ecol, 38(11):2717\u0026ndash;2730\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eWeeks TL, Walkden PA, Edwards DP, Lees AC, Pigot AL, Purvis A, Tobias JA (2025) Land-use change undermines the stability of avian functional diversity. Nature\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eWilson S, Alavi N, Pouliot D, Mitchell GW (2020) Similarity between agricultural and natural land covers shapes how biodiversity responds to agricultural expansion at landscape scales. Agriculture, Ecosystems \u0026amp; Environment, 301:107052\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eWood CM, Kahl S (2024) Guidelines for appropriate use of BirdNET scores and other detector outputs. J Ornithol\u003c/span\u003e\u003c/li\u003e\u003c/ol\u003e"}],"fulltextSource":"","fullText":"","funders":[],"hasAdminPriorityOnWorkflow":false,"hasManuscriptDocX":true,"hasOptedInToPreprint":true,"hasPassedJournalQc":"","hasAnyPriority":false,"hideJournal":false,"highlight":"","institution":"","isAcceptedByJournal":false,"isAuthorSuppliedPdf":false,"isDeskRejected":"","isHiddenFromSearch":false,"isInQc":false,"isInWorkflow":false,"isPdf":false,"isPdfUpToDate":true,"isWithdrawnOrRetracted":false,"journal":{"display":true,"email":"[email protected]","identity":"landscape-ecology","isNatureJournal":false,"hasQc":true,"allowDirectSubmit":false,"externalIdentity":"land","sideBox":"Learn more about [Landscape Ecology](https://www.springer.com/journal/10980)","snPcode":"10980","submissionUrl":"https://submission.nature.com/new-submission/10980/3","title":"Landscape Ecology","twitterHandle":"","acdcEnabled":true,"dfaEnabled":true,"editorialSystem":"em","reportingPortfolio":"Springer Hybrid","inReviewEnabled":true,"inReviewRevisionsEnabled":false},"keywords":"passive acoustic monitoring (PAM), habitat heterogeneity, biodiversity, automated species identification, orchard meadow, agroecosystems","lastPublishedDoi":"10.21203/rs.3.rs-9368528/v1","lastPublishedDoiUrl":"https://doi.org/10.21203/rs.3.rs-9368528/v1","license":{"name":"CC BY 4.0","url":"https://creativecommons.org/licenses/by/4.0/"},"manuscriptAbstract":"\u003ch2\u003eContext\u003c/h2\u003e \u003cp\u003eBirds are declining worldwide, with farmland birds disproportionately affected. Most studies on farmland birds focus on single habitat types, yet agriculturally dominated landscapes are mosaics composed of multiple habitat types like arable land, grassland, forests, and orchards.\u003c/p\u003e\u003ch2\u003eObjectives\u003c/h2\u003e \u003cp\u003eWe aimed to understand how these habitat types jointly shape farmland bird diversity, particularly regarding local and landscape drivers of alpha and beta diversity.\u003c/p\u003e\u003ch2\u003eMethods\u003c/h2\u003e \u003cp\u003eWe used passive acoustic monitoring to survey farmland bird communities in 14 mosaic agricultural landscapes (1 km\u0026sup2;) in southern Germany that differ in habitat diversity. In total, 224 autonomous recording units were deployed in a grid-based design with sampling intensity proportional to habitat area. Using BirdNET and manual validation, we identified 54 bird species from 2,016 hours of recordings collected over 4.5 months.\u003c/p\u003e\u003ch2\u003eResults\u003c/h2\u003e \u003cp\u003eLocal species richness (alpha diversity) increased with habitat heterogeneity at both local and landscape scales. Arable sites showed the lowest alpha diversity but comparatively high within-habitat beta diversity, whereas orchards supported high alpha but low within-habitat beta diversity. Beta diversity was highest between habitat types, especially between forests and arable land, reflecting strong contrasts in their structural complexity. Generalized dissimilarity modelling showed that local predictors were more important than landscape-level predictors in explaining bird beta diversity. Habitat associations of bird species were largely consistent with ecological expectations: bird species adapted to dense vegetation occurred mainly in forest-dominated sites, while open-habitat species were associated with arable land. Species with decreasing population trends occurred across all major habitat types. At the landscape scale, gamma diversity increased strongly with landscape diversity.\u003c/p\u003e\u003ch2\u003eConclusions\u003c/h2\u003e \u003cp\u003eMaintaining habitat heterogeneity at multiple spatial scales is critical to conserve farmland bird diversity.\u003c/p\u003e","manuscriptTitle":"Farmland bird diversity requires heterogeneity between and within habitats","msid":"","msnumber":"","nonDraftVersions":[{"code":1,"date":"2026-04-27 11:47:29","doi":"10.21203/rs.3.rs-9368528/v1","editorialEvents":[{"type":"communityComments","content":0},{"type":"editorInvitedReview","content":"","date":"2026-05-13T04:47:13+00:00","index":"hide","fulltext":""},{"type":"reviewerAgreed","content":"286883156807858230795114563354113498448","date":"2026-04-21T18:04:24+00:00","index":"hide","fulltext":""},{"type":"reviewerAgreed","content":"307747834126765167816494673283859444121","date":"2026-04-19T22:18:34+00:00","index":"hide","fulltext":""},{"type":"reviewersInvited","content":"","date":"2026-04-17T16:55:23+00:00","index":"","fulltext":""},{"type":"editorAssigned","content":"","date":"2026-04-13T11:34:40+00:00","index":"","fulltext":""},{"type":"checksComplete","content":"","date":"2026-04-13T11:33:43+00:00","index":"","fulltext":""},{"type":"submitted","content":"Landscape Ecology","date":"2026-04-09T12:17:54+00:00","index":"","fulltext":""}],"status":"published","journal":{"display":true,"email":"[email protected]","identity":"landscape-ecology","isNatureJournal":false,"hasQc":true,"allowDirectSubmit":false,"externalIdentity":"land","sideBox":"Learn more about [Landscape Ecology](https://www.springer.com/journal/10980)","snPcode":"10980","submissionUrl":"https://submission.nature.com/new-submission/10980/3","title":"Landscape Ecology","twitterHandle":"","acdcEnabled":true,"dfaEnabled":true,"editorialSystem":"em","reportingPortfolio":"Springer Hybrid","inReviewEnabled":true,"inReviewRevisionsEnabled":false}}],"origin":"","ownerIdentity":"ca040738-bea7-4b4c-978b-ebae9c066f8c","owner":[],"postedDate":"April 27th, 2026","published":true,"recentEditorialEvents":[{"type":"editorInvitedReview","content":"","date":"2026-05-13T04:47:13+00:00","index":16,"fulltext":""}],"rejectedJournal":[],"revision":"","amendment":"","status":"under-review","subjectAreas":[],"tags":[],"updatedAt":"2026-04-27T11:47:30+00:00","versionOfRecord":[],"versionCreatedAt":"2026-04-27 11:47:29","video":"","vorDoi":"","vorDoiUrl":"","workflowStages":[]},"version":"v1","identity":"rs-9368528","journalConfig":"researchsquare"},"__N_SSP":true},"page":"/article/[identity]/[[...version]]","query":{"redirect":"/article/rs-9368528","identity":"rs-9368528","version":["v1"]},"buildId":"XKTyCvWXoU3ODBz1xrDgd","isFallback":false,"isExperimentalCompile":false,"dynamicIds":[84888],"gssp":true,"scriptLoader":[]}