Introduction
In the EU, farmland covers a substantial 39.1% of the total area (Eurostat 2021), and both agricultural intensification and land abandonment are placing mounting pressure on natural habitats, particularly semi-natural grasslands and traditionally managed meadows (EEA 2020). As human populations grow and consumption patterns shift, intensification has become the leading driver of global biodiversity loss, causing sharp declines in many farmland species. Modern land-use changes, characterised by intensification, mechanisation, and large-scale monocultures, have fragmented and simplified landscapes, leading to severe negative impacts on wildlife populations worldwide (Dudley and Alexander Reference Dudley and Alexander2017; Hughes et al. Reference Hughes, Tougeron, Martin, Menga, Rosado and Villasante2023; Kehoe et al. Reference Kehoe, Romero-Muñoz, Polaina, Estes, Kreft and Kuemmerle2017; Jeanneret et al. Reference Jeanneret, Lüscher, Schneider, Pointereau, Arndorfer and Bailey2021). At the same time, the abandonment of traditional agropastoral practices, once crucial for maintaining habitat diversity and preventing grassland loss, has accelerated shrub and forest encroachment, driving a widespread decline of many grassland species (Kmecl and Denac Reference Kmecl and Denac2018; Scridel et al. Reference Scridel, Stanič, Pacorini, Kravos, Utmar and Olmo2025; Zakkak et al. Reference Zakkak, Radovic, Nikolov, Shumka, Kakalis and Kati2015).
Farmland bird declines have been documented across Europe for decades, with farmland species experiencing the sharpest decline among all European common bird indicators (-60% over the last 43 years; PECBMS 2024). This decline has been more pronounced in Western Europe (-55% in the last 43 years) compared with Central and Eastern Europe (-33% in the last 41 years). Such disparity is largely driven by differences in socio-economic systems, historical land management, and policy transitions, with Western Europe undergoing earlier and more intense agricultural intensification beginning in the mid-twentieth century (Donald et al. Reference Donald, Green and Heath2001, 2006; Rigal et al. Reference Rigal, Dakos, Alonso, Auniņš, Benkő and Brotons2023; Sutcliffe et al. Reference Sutcliffe, Batáry, Kormann, Báldi, Dicks and Herzon2014). This transformation, fuelled by the Green Revolution, EU Common Agricultural Policy (CAP) subsidies, and large-scale mechanisation, led to the widespread loss of small-scale, heterogeneous farmland that traditionally supported diverse bird communities. In contrast, agriculture in Central and Eastern Europe remained largely low-intensity until the 1990s, providing a temporary refuge for farmland birds (Donald et al. Reference Donald, Sanderson, Burfield and Van Bommel2006; Reif and Hanzelka Reference Reif and Hanzelka2020; Sutcliffe et al. Reference Sutcliffe, Batáry, Kormann, Báldi, Dicks and Herzon2014; Tryjanowski et al. Reference Tryjanowski, Hartel, Báldi, Szymański, Tobolka and Herzon2011). However, major changes following EU accession in 2004 triggered rapid intensification and the decline of traditional farming systems, accelerating biodiversity loss. With increasing mechanisation, monoculture expansion, and land abandonment in recent decades, similar declines are now becoming widespread across Central and Eastern Europe, underscoring the urgent need for proactive conservation measures to mitigate further biodiversity loss (Reif and Vermouzek Reference Reif and Vermouzek2018; Reif et al. Reference Reif, Gamero, Hološková, Aunins, Chodkiewicz and Hristov2024). Despite these trends, disparities in research focus and funding opportunities have led to an uneven distribution of farmland bird studies, with the majority conducted in Western Europe. As a result, researchers have called for more pan-European studies on farmland bird species and their conservation strategies to address knowledge gaps and develop effective, regionally tailored conservation actions (Báldi and Batary Reference Báldi and Batáry2011; Benedetti Reference Benedetti2017; Sutcliffe et al. Reference Sutcliffe, Batáry, Kormann, Báldi, Dicks and Herzon2014; Tryjanowski et al. Reference Tryjanowski, Hartel, Báldi, Szymański, Tobolka and Herzon2011; Yosef and Tryjanowski Reference Yosef and Tryjanowski2024).
The Red-backed Shrike Lanius collurio is a migratory passerine of the Palearctic region (Yosef et al. Reference Yosef, Christie, del Hoyo, Elliott, Sargatal, Christie and E2020) and a widely recognised bioindicator of low-intensity farmland management (Flade Reference Flade1994; Latus et al. Reference Latus, Schultz and Kujawa2004; Tryjanowski et al. Reference Tryjanowski, Hartel, Báldi, Szymański, Tobolka and Herzon2011). It is classified as “Least Concern” on the European Red List of Birds (BirdLife International 2021, 2024) but listed as a priority conservation species under Annex I of the EU Birds Directive (Directive Reference Directive2009/147/EC) and Annex II of the Bern Convention (Council of Europe 1979). Its current European population trend is considered stable, although it has shown a negative trend over the past 10 years (-9%), with an even greater long-term decline (-22%; PECBMS 2024). It thrives in a variety of habitats, from upland meadows to lowland farmland, including vineyards and orchards (Lefranc and Worfolk Reference Lefranc and Worfolk2022; Morelli Reference Morelli2012). In addition to agricultural landscapes, the species nests on forest clearings, shrubland, and heathlands (Söderström and Karlsson Reference Söderström and Karlsson2011). Meadows play a crucial role in foraging during the breeding period (May–July), as demonstrated by several studies (Kuźniak and Tryjanowski Reference Kuźniak and Tryjanowski2000; Sfougaris et al. Reference Sfougaris, Plexida and Solomou2014; Vanhinsbergh and Evans Reference Vanhinsbergh and Evans2002), which highlight their support for a rich arthropod community, the primary food source during this time, particularly in grazed areas (Goławski and Goławska Reference Goławski and Goławska2008; Knozowski et al. Reference Knozowski, Nowakowski, Stawicka, Dulisz and Górski2024). However, Red-backed Shrike relies on a mosaic of habitats, using meadows for hunting (Goławski and Goławska Reference Goławski and Goławska2008), and shrubs for nesting, perching, and food storage (Goławski et al. Reference Goławski, Mroz and Goławska2020; Morelli et al. Reference Morelli, Bussière, Goławski, Tryjanowski and Yosef2015, Reference Morelli, Mróz, Pruscini, Santolini, Goławski and Tryjanowski2016; Wozna et al. Reference Wozna, Hromada, Reeve, Szymański, Zolnierowicz and Tobolka2017). The importance of habitat heterogeneity is well documented, as structurally diverse landscapes provide both nesting sites and higher invertebrate abundance, further supporting shrike populations (Latus et al. Reference Latus, Schultz and Kujawa2004; Morelli et al. Reference Morelli, Santolini and Sisti2012; Roilo et al. Reference Roilo, Spake, Bullock and Cord2024; Sfougaris et al. Reference Sfougaris, Plexida and Solomou2014).
In this study, we investigated a 33-year (1992 – 2025) population trend of Red-backed Shrike in an alluvial floodplain in north-east Slovenia. This area, home to meadows of European conservation importance (Štumberger et al. Reference Štumberger, Kaligarič and Geister1993), was historically considered a national stronghold for the species, reaching some of the highest population densities reported in Central Europe, with more than 39 pairs/km2 (Bauer et al. Reference Bauer, Bezzel and Fiedler2005; Denac Reference Denac2003). The Red-backed Shrike has experienced a moderate decline of 25% in Slovenia since 2008 (Kmecl et al. Reference Kmecl, Gamser and Šumrada2023). Its population has also decreased in neighbouring countries: by 17% in Austria since 1998 (Teufelbauer and Seaman Reference Teufelbauer and Seaman2022), by 27% in Croatia since 2015 (Budinski et al. Reference Budinski, Zec, Dender, Kapelj, Čulig and Taylor2022), and by a striking 64% in Italy since 2000 (Rete Rurale Nazionale and Lipu Reference Nazionale and Lipu2024). These trends underscore the need to investigate the factors driving the species’ decline in one of its former strongholds and to identify conservation priorities for its remaining population. Specifically, this study aimed to assess the species’ habitat preferences during the breeding period and investigate the habitat correlates of its population decline from 2004 to 2022. Based on existing literature, we expected the Red-backed Shrike to select ecotonal habitats characterised by grassland–shrub mosaics around its breeding territories (Goławski and Goławska Reference Goławski and Goławska2008; Goławski et al. Reference Goławski, Mroz and Goławska2020; Morelli et al. Reference Morelli, Santolini and Sisti2012; Pedersen et al. Reference Pedersen, Schnedler-Meyer, Ekberg and Tøttrup2018; Sfougaris et al. Reference Sfougaris, Plexida and Solomou2014). We hypothesise that its population decline is primarily driven by the loss of meadows due to the expansion of arable fields in the area over the study period. Additionally, we anticipate that shrub encroachment on formerly managed meadows, resulting from the abandonment of traditional land management practices, has negatively affected this breeding population.
Methods
Study area
The study was conducted in Šturmovci, an alluvial flood-plain in the Sub-Pannonian zoogeographical region of Slovenia (46°22’38.8"N, 15°55’41.0"E; 223 m a.s.l.), at the southern margin of Central Europe. The study site covers an area of 440 ha and it is located between the rivers Drava and Dravinja, Lake Ptuj, and the Haloze hills (Mršić Reference Mršić1997) (Figure 1). The site harbours rich biodiversity, with over 237 bird, 493 plant, and 36 dragonfly species recorded (Štumberger et al. Reference Štumberger, Kaligarič and Geister1993). Protected as a Nature Park since 1979, the area is included in the Natura 2000 site “Drava” under both Birds Directive (SI5000011) and Habitat Directive (SI3000220; Official Gazette RS 2004). Šturmovci has a continental climate, with warm summers (19–21°C), cold winters (-1–1°C), and precipitation ranging from 300 mm to 400 mm in the summer to 150 mm to 200 mm in the winter (NMSS 2016). The sandy-gravel soils were deposited by the Drava and Dravinja rivers, which regularly flooded the area until the construction of a dam on the Drava River in 1979 (Štumberger Reference Štumberger and Polak2000). The flood-plain originally supported extensive alluvial forests, which gradually dried out and were transformed into meadows, predominantly xeric sand calcareous grasslands (N2000 code: 6120) and lowland hay meadows (N2000 code: 6510), through vegetation clearance, regular mowing, and traditional practices such as leaf raking. By 1992, 80–90% of the meadows were still being actively managed, but by the late 1990s, a third had been converted into arable fields. Grazing was practised regularly in the area before the 1979 dam construction (Štumberger et al. Reference Štumberger, Kaligarič and Geister1993), after which farming became focused primarily on arable crop production. These arable fields now cover approximately 40% of the area and are primarily cultivated with maize, wheat, and barley (MAFF 2019). Meanwhile, many of the remaining meadows (15% of the study area) are increasingly affected by shrub encroachment. Current landscape also includes remnants of alluvial and riparian forests, such as Eastern European poplar–willow stands, riverine Salix woodlands, and riparian assemblages of willow, alder, and ash, interspersed with hedgerows and scattered trees (25% cover; IRSNC 2006). Urban areas have remained largely unchanged over time and occupy only a small fraction of the landscape (2%).
Figure 1. The Šturmovci study area, with its boundary indicated by the red line, shown for the years 1997, 2003, 2009, and 2025. Points represent the locations of breeding Red-backed Shrike pairs. The inset map highlights the location of the study area in Europe, shown within the red square. Background maps: 1997: Surveying and Mapping Authority RS (SMARS) (2002), 2003: Google Earth (2006), 2009: Google Earth (2009), 2025: Google Earth (2024).
Bird surveys
Data on the breeding territories of Red-backed Shrike were available from past intermittent surveys conducted in 1992, 1997, and 2003; annually from 2009 to 2015; as well as in 2017, 2019, and 2025 (Denac Reference Denac2003; Štumberger et al. Reference Štumberger, Kaligarič and Geister1993). These were conducted between late June and early July each year, covering the entire study area within a single day. Surveys took place in the morning (07:00–11:00) using the area count method (Bibby et al. Reference Bibby, Burgess, Hill and Mustoe2000), during which all breeding pairs were recorded and mapped with binoculars and a telescope. Experienced ornithologists carefully tracked movement patterns and considered the timing and direction of observations to avoid double counting.
Habitat mapping
Habitat data were obtained from the publicly available Slovenian land-use database (MAFF 2019), which provides a national record of land cover and serves as the basis for implementing measures under the EU CAP. The data on land use are gathered mainly through digitised orthophotos over the whole country, but also through field inspections, satellite imagery, and third-party information (Agricultural Directorate RS 2013). Established in 2002, it is updated per every two to four years. For years without mapping data, habitat areas were estimated by averaging values from the nearest adjacent mapped years (see Supplementary material Table S1). For the purposes of this study, we reclassified the original habitat types into eight major categories, based on their availability in the study area and the ecological requirements of Red-backed Shrike (Brambilla et al. Reference Brambilla, Rubolini and Guidali2007; Casale et al. Reference Casale, Bionda, Falco, Siccardi, Toninelli and Rubolini2012). These were: (1) arable: land ploughed or cultivated for annual or perennial crops, including fallows; (2) meadows: meadows dominated by grasses, clover or other fodder herbs, regularly mown or grazed, not part of crop rotation or ploughed; (3) forest: areas classified as forest under national regulations, with woody vegetation covering more than 75% of the surface; (4) shrub: partially overgrown land (20–75% woody cover) resulting from abandonment or minimal vegetation management; includes hedgerows and riverbanks not classified as forest; (5) orchards: areas with fruit trees, either intensively managed or traditional orchards; (6) marshland: areas frequently flooded or consistently waterlogged; (7) water: surface water-bodies, including rivers, ponds, and lakes; (8) urban: built-up areas, including buildings, roads, paths, and parking lots.
Statistical analysis
To assess the habitat preferences of Red-backed Shrike during the breeding season, we used habitat data only from bird survey years that aligned with years of detailed land-cover mapping, thereby avoiding the use of averaged values from adjacent years and improving the precision of our assessment. This resulted in five years of matched data (2003, 2007, 2010, 2014, and 2017). For each year, we generated random points at 10 times the number of territories as an estimate of habitat availability (Fieberg et al. Reference Fieberg, Signer, Smith and Avgar2021). Random points were distributed across the entire study area, encompassing all the available habitat types described above, including those where Red-backed Shrikes are not typically found (e.g. urban, water), so as not to bias availability estimates. These were created using the dismo package (Hijmans et al. Reference Hijmans, Phillips, Leathwick and Elith2024) with the “randomPoints” function in R (v4.3.2; R Core Team 2023). For both real territories and random points, we created buffers with 80-m radii to represent the average territory size (c.2 ha), following assessments made in previous studies (Brambilla and Ficetola Reference Brambilla and Ficetola2012; Goławski and Meissner Reference Goławski and Meissner2008; Pestka et al. Reference Pestka, Jakubas and Wojczulanis-Jakubas2018). We then extracted the percentage cover and the number of habitat patches for each territory using the landscapemetrics package (Hesselbarth et al. Reference Hesselbarth, Sciaini, With, Wiegand and Nowosad2019). Recognising the importance of habitat heterogeneity for this species (Morelli et al. Reference Morelli, Santolini and Sisti2012; Roilo et al. Reference Roilo, Spake, Bullock and Cord2024), we also calculated the Shannon Diversity Index and an index of configurational heterogeneity (i.e. Aggregation Index metric). The latter takes the number of like adjacencies and divides it by the theoretical maximum number of like adjacencies for a particular class, summed over each class for the given landscape (Hesselbarth et al. Reference Hesselbarth, Sciaini, With, Wiegand and Nowosad2019). It thus, calculates the spatial arrangement of cover types (or patches; Fahrig et al. Reference Fahrig, Baudry, Brotons, Burel, Crist and Fuller2011). Habitat preferences were analysed via generalised linear mixed-effects models (GLMMs) with binomial error distribution, implemented via the glmmTMB package (Brooks et al. Reference Brooks, Kristensen, van Benthem, Magnusson, Berg and Nielsen2017). The binary response variable distinguished true territories (coded as 1) from random points (coded as 0). Year was included as a random intercept effect to account for variations in the number of territories surveyed and the corresponding number of randomly generated locations across years. Following procedures similar to other studies (Scridel et al. Reference Scridel, Tenan, Brambilla, Celva, Forti and Fracasso2022, Reference Scridel, Anderle, Capelli, Forti, Bettega and Alessandrini2024a, Reference Scridel, Stanič, Pacorini, Kravos, Utmar and Olmo2025), to avoid overfitting, we followed a structured variable selection approach where we first tested each predictor independently against the null model. Only predictors that reduced Akaike information criterion corrected for small sample sizes (AICc) (Burnham and Anderson Reference Burnham and Anderson2002, Reference Burnham and Anderson2004) by at least two units relative to the null model were retained for further analysis. To account for potential non-linear responses, given the Red-backed Shrike’s association with ecotonal habitats and possible threshold effects (Casale and Brambilla Reference Casale and Brambilla2009; Lefranc and Worfolk Reference Lefranc and Worfolk2022), we tested both linear and quadratic terms. Supported predictors were subsequently combined into additive models to evaluate their joint influence. These models were then ranked again using AICc via the “dredge” function (Mumin R package; Bartoń Reference Bartoń2023) to identify the best-supported combinations. Collinearity among all predictor variables was first assessed using Pearson’s correlation coefficient (r), applying a threshold of 0.6. When two variables were highly correlated, we retained the one with the highest explanatory power, determined by Nagelkerke’s R 2, calculated using the package performance (Lüdecke et al. Reference Lüdecke, Ben-Shachar, Waggoner and Makowski2021). Collinearity was observed between the Shannon Diversity Index and Aggregation Index (this latter removed). Multicollinearity in the global model was evaluated using the Variance Inflation Factor (VIF), and variables with VIF values ≥5 were removed (Fox and Weisberg Reference Fox and Weisberg2019). To account for model selection uncertainties, we performed a model-averaged procedure of most supported models (i.e. those with ΔAICc <2) and further reported the conditional model average estimates after the removal of uninformative parameters (Arnold Reference Arnold2010). The DHARMa package (Hartig Reference Hartig2022) was used to test for spatial autocorrelation via Moran’s I in the full and averaged model, and no significant autocorrelation was detected (Moran I = 0.002, P = 0.148).
To assess the breeding population trend in the study area from 1992 to 2025, we used a Trends and Indices for Monitoring data model (TRIM) (Pannekoek and van Strien Reference Pannekoek and van Strien1998) fitted in the R environment via the package rtrim (Bogaart et al. Reference Bogaart, van der Loo and Pannekoek2018). TRIM implements a log-linear Poisson regression and generalised estimating equations (GEE) to produce yearly abundance indices and trends. It is a standard tool used to estimate bird population trends in most European countries (Brlík et al. Reference Brlík, Šilarová, Škorpilová, Alonso, Anton and Aunins2021). TRIM allows the estimation of counts for missing sites by estimating missing values from all visited sites with the assumption that changes observed in surveyed sites also apply to non-surveyed sites. In our case, we considered sites as a single unit (i.e. the whole study area), but we had missing data for 21 years (1993–1996, 1998–2002, 2004–2008, 2016, 2018, 2020, 2021–2024), equivalent to 63.64% of the total data set. Various models are available in rtrim and we used model 2 which, in addition to site effects, also assumes all years as possible changing points in the population trend and takes into account overdispersion of the data (overdispersion parameter = 3.462; Scridel et al. Reference Scridel, Utmar, Koce, Kralj, Baccetti and Candotto2024b). Observed counts (in our case the sum of all territories/year) were modelled as a function of site-varying and year-varying effects: log(countsi) = sitei + yeari.
To infer drivers of Red-backed-Shrike breeding territory decline, the log-transformed count of territories across the entire study area for each year (response variables) was modelled using a linear regression in relation to the total cover of the most representative habitat types (i.e. arable, meadow, shrub, and forest) from 2004 to 2022 (when habitat data were available). For years without directly mapped habitat data, we used values averaged from the nearest available years with detailed land-cover information (Table S1), while for missing territory count data, we used predicted values from the TRIM model (Table S5). To explore relationships between habitat types over time, we calculated pairwise Spearman’s rank correlation coefficients (ρ) using the “cor.test” function in R. Due to strong collinearity among habitat variables (Table S4), we fitted separate univariate models for each predictor instead of a multiple regression to avoid multicollinearity and ensure robust inference.
Results
Habitat preference
The habitat preference analysis identified three equally supported models (ΔAICc <2; Table 1), indicating that the probability of territory presence for Red-backed Shrike was strongly influenced by habitat cover, habitat heterogeneity, and the number of habitat patches. Specifically, territory occurrence increased with greater meadow and arable cover, decreased with increasing forest cover, and followed a non-linear (quadratic) relationship with both shrub cover and the Shannon Diversity Index within the 2-ha buffers (Tables 1 and 2 and Figure 2). Year, included as a random intercept effect, explained minimal additional variance, as indicated by the small difference between marginal R 2 (0.425) and conditional R 2 (0.438).
Table 1. Competitive models (ΔAICc <2) for Red-backed Shrike habitat preferences, assessed using a GLMM with a binomial error structure. Predictors are displayed with their respective relationships: (+) linear positive, (-) linear negative, and (∩) non-linear hump-shaped. All models within ΔAICc <2 were averaged to derive model-averaged conditional estimates (Table 2)
Table 2. Model-averaged conditional estimates, z values and lower and upper 2.5% confidence intervals (CIs) for Red-backed Shrike territories habitat preferences
Figure 2. Habitat variables influencing the probability of Red-backed Shrike breeding territory presence, based on the conditional averaged model results (Table 2). Habitat variables were extracted within a 2-ha buffer around both actual and random territories. (Photograph: DOPPS – Birdlife Slovenia, Alen Ploj)
Drivers of Red-backed Shrike decline
According to the TRIM analysis, from 1992 to 2025, the study area experienced a “strong decline” in the number of breeding territories, from 172 to 13, a decline of 92.44% (Wald test 102.60; P <0.001; Table 3; Figure 3a; Table S5). The single most supported predictor describing the variation of total territory number in the area during the period 2004–2022 was the increase in shrub cover (AICc = 2.265, adj.R 2 = 0.797), followed by the increase in arable cover (AICc = 17.951, adj.R 2 = 0.537) and the reduction in meadow cover (AICc = 18.735, adj.R 2 = 0.518; Figure 3b and Table S3). Strong negative correlations between meadow cover and both arable land (Spearman’s ρ = -0.972, P <0.001) and overgrown areas (ρ = –0.874; Table S4) suggest a shift in land use, where meadows may have been either converted to arable fields or abandoned, leading to shrub encroachment through natural succession.
Table 3. Thirty-three-year territory trend coefficients (additive: “add”; multiplicative: “mul”) and corresponding standard errors (se) and Wald test for significance of deviations from a linear trend for the Red-backed Shrike in Šturmovci, Slovenia. Analysis was conducted using TRIM (in R with the rtrim package; Bogaart et al. Reference Bogaart, van der Loo and Pannekoek2018) by fitting a loglinear Poisson regression model that accounted for overdispersion
Figure 3. (a) Red-backed Shrike territory trend based on TRIM analysis (annual imputed abundance estimates), with shaded areas indicating the standard error. The grey rectangle highlights the period for which detailed habitat data were available and subsequently modelled, as shown in panel (b). (b) Territory trend from 2004 to 2022, shown together with habitat cover (ha) for the most representative habitat types in the study area. Both the bird and habitat trends were smoothed using Locally Estimated Scatterplot Smoothing (LOESS) (locally weighted regression) to visualise non-linear changes over time.
Discussion
Our findings provide compelling evidence of a dramatic long-term decline in the breeding population of Red-backed Shrike Lanius collurio within a key Central European stronghold, mirroring broader declines in farmland bird populations across Europe. The 92.44% reduction in breeding territories over 33 years (1992–2025) highlights the species’ vulnerability to the rapid transformation of semi-natural habitats caused by agricultural intensification and land abandonment. This underscores the critical role of permanent meadows in sustaining farmland biodiversity and raises a concern that areas in Central and Eastern Europe, once considered relative refuges for farmland birds, are now facing the same pressures long observed in Western Europe. The sharp decline observed at a site of national importance for biodiversity indicates that similar patterns may also be affecting other species with comparable ecological requirements and could be occurring elsewhere, albeit undetected.
Consistent with previous studies, we found that Red-backed Shrike favours ecotonal landscapes, a mosaic of meadows interspersed with moderate shrub cover (Brambilla et al. Reference Brambilla, Rubolini and Guidali2007; Nijssen et al. Reference Nijssen, Geertsma, Kuper, van den Burg, van Duinen and Versluijs2024; Pedersen et al. Reference Pedersen, Schnedler-Meyer, Ekberg and Tøttrup2018). Permanent meadows, particularly those managed traditionally, provide essential foraging opportunities by supporting abundant invertebrate prey (Goławski and Goławska Reference Goławski and Goławska2008; Knozowski et al. Reference Knozowski, Nowakowski, Stawicka, Dulisz and Górski2024; Svendsen et al. Reference Svendsen, Sell, Bøcher and Svenning2015). A study from the same area showed a preference for recently mown grasslands (Denac Reference Denac2003), where prey like Orthoptera, Coleoptera, and small vertebrates are more accessible. Red-backed Shrike largely avoided forest interiors (Table 2 and Figure 2), which lack suitable nesting shrubs and offer poor foraging opportunities as found in other studies (Bombek and Denac Reference Bombek, Denac, Mihelič, Kmecl, Denac, Koce, Vrezec and Denac2019; Hollander et al. Reference Hollander, Titeux and Van Dyck2012, Reference Hollander, Titeux and Van Dyck2013; Morelli et al. Reference Morelli, Santolini and Sisti2012). Indeed, nesting is usually concentrated in thorny shrubs such as hawthorn Crataegus sp., dog rose Rosa canina, blackthorn Prunus spinosa, and blackberry Rubus spp., which offer protection from predators (Svendsen et al. Reference Svendsen, Sell, Bøcher and Svenning2015; Tryjanowski et al. Reference Tryjanowski, Kuzniak and Diehl2000). However, thornless species like elder Sambucus spp., willow Salix spp., and black cherry Padus serotina, may still be used, especially when growing in patchy formations (Wozna et al. Reference Wozna, Hromada, Reeve, Szymański, Zolnierowicz and Tobolka2017). The quadratic relationship between territory presence and number of shrub patches in our study area reflects an optimal balance, with the highest occurrence observed at around 7–8 large shrubs within a territory of approximately 2 ha (Figure 2). Below this threshold, nesting and perching opportunities may be insufficient, while above it, dense shrub growth likely impedes foraging efficiency. Interestingly, our results also suggest that a limited amount of arable land can positively influence Red-backed Shrike presence, likely due to the high accessibility of widespread prey (including agricultural pests) such as Orthoptera (e.g. Gryllotalpa spp.), Coleoptera, and small mammals (e.g. mice and voles), particularly when embedded within a heterogeneous landscape like that of Šturmovci. This highlights that small, arable patches could still play a complementary role for foraging as found in other studies (e.g. Brambilla et al. Reference Brambilla, Rubolini and Guidali2007). Landscape heterogeneity, measured using the Shannon Diversity Index, was positively associated with breeding presence up to a point. Beyond that, the inclusion of less favourable habitats, such as forests or urban areas, likely reduced habitat quality, supporting the idea that structurally diverse, open landscapes are optimal (Latus et al. Reference Latus, Schultz and Kujawa2004; Sfougaris et al. Reference Sfougaris, Plexida and Solomou2014).
According to Figure 3a, most of the decline in the study area was already in place by the early 2000s, with no signs of recovery afterwards. This timing suggests that earlier waves of agricultural intensification, initiated under Yugoslav land consolidation policies and reinforced by pre-accession support, played a key role in the decline. Between 1945 and 1991, much of the wet meadows in Slovenia were transformed into intensive arable fields (Lisec et al. Reference Lisec, Primožič, Pintar, Bovha, Ferlan and Prosen2013), likely contributing to the local extinction of other farmland species such as the Lesser Grey Shrike Lanius minor, Lesser Kestrel Falco naumanni, and European Roller Coracias garrulus. These pressures continued following Slovenia’s EU accession in 2004, as described by other studies that link the CAP to biodiversity loss (Lovec et al. Reference Lovec, Šumrada and Erjavec2020; Šumrada et al. Reference Šumrada, Kmecl and Erjavec2021). This interpretation is further supported by our model results, which show that shrub encroachment explained more of the 2004–2022 decline than changes in meadow or arable cover. As previously described, this is likely because a substantial portion of meadow loss had already occurred by 2004. Shrub overgrowth, which increased in cover by 57% during 2004–2022, was driven by the abandonment of traditional land management and likely further degraded the remaining habitats for Red-backed Shrike. This reduction in open areas, which are critical for this farmland specialist, may have favoured generalist and predatory species (Tscharntke et al. Reference Tscharntke, Sekercioglu, Dietsch, Sodhi, Hoehn and Tylianakis2008; Zakkak et al. Reference Zakkak, Radovic, Nikolov, Shumka, Kakalis and Kati2015). Increased arable field cover (+40% over the period 2004–2022) was the second most supported predictor in our analysis of population decline in the study area and likely contributed to habitat homogenisation, reducing the availability of suitable breeding and foraging habitats for the species (Table S3). The strong negative correlations between meadow cover and both arable land and shrub cover suggest that meadow loss in the area resulted from a dual process: direct land conversion and unmanaged succession (Table S4). Overall, meadow habitat declined by 62% between 2004 and 2022 and now covers a similar area to shrubland, at approximately 50 ha (Figure 3b). Mechanisms beyond the reduction of breeding territories are closely linked to declines in prey availability and accessibility because of land-cover changes (Hemerik et al. Reference Hemerik, Geertsma, Waasdorp, Middelveld, van Kleef and Klok2015; Pedersen et al. Reference Pedersen, Schnedler-Meyer, Ekberg and Tøttrup2018). Poor body condition in both nestlings and adults has been associated with reduced survival and reproductive success in other European populations (Knozowski et al. Reference Knozowski, Nowakowski, Stawicka, Dulisz and Górski2024). First-year and adult survival, along with sex-specific differences, have been identified as key drivers of the species’ declining population growth rate, although immigration appears to buffer local breeding numbers and help maintain a stable number of breeding pairs (Hemerik et al. Reference Hemerik, Geertsma, Waasdorp, Middelveld, van Kleef and Klok2015; Takács et al. Reference Takacs, Kuźniak and Tryjanowski2004). Notably, the exceptionally high breeding-site infidelity reported by Tryjanowski et al. (Reference Tryjanowski, Goławski, Kuźniak, Mokwa and Antczak2007) may be particularly relevant to understanding local extinction risks, as even marginal declines in habitat quality could trigger increased dispersal and reduced site fidelity. Future studies should incorporate individual marking to explore vital rates and return probabilities in our study area, especially considering the impact of conditions in African wintering grounds (Pasinelli et al. Reference Pasinelli, Schaub, Häfliger, Frey, Jakober and Müller2011).
The population trend of the Red-backed Shrike in our study area is deeply concerning (Table 3 and Figure 3a). The 92.44% decline over the period 1992–2025 indicates that the species is on the verge of local extinction. In western Poland, Takács et al. (Reference Takacs, Kuźniak and Tryjanowski2004) identified a minimum of 45 breeding pairs over a 57-ha area (density of 0.79 pairs/ha) as necessary for population viability. Our 2025 estimate of just 13 breeding pairs in a 440-ha area, reflects an alarmingly low breeding density (of only 0.03 pairs/ha) and limited long-term prospects. A similar scenario has already unfolded on a broader scale, with the species becoming effectively extinct in the UK by the 1990s (Tryjanowski et al. Reference Tryjanowski, Sparks and Crick2006), confined to a single site in the Netherlands (Geertsma et al. Reference Geertsma, van Berkel and Esselink2000), and subject to widespread local extinctions in Italy (Rete Rurale Nazionale and Lipu Reference Nazionale and Lipu2024). Without conservation strategies tailored to local conditions, as successfully applied in other countries (Nijssen et al. Reference Nijssen, Geertsma, Kuper, van den Burg, van Duinen and Versluijs2024), the species risks following a trajectory similar to that of the closely related Lesser Grey Shrike and Woodchat Shrike Lanius senator, both of which have already disappeared from large parts of Europe (Brønskov and Keller Reference Brønskov, Keller, Keller, Herrando, Voříšek, Franch, Kipson and Milanesi2020; Krištín et al. Reference Krištín, Hoi and Kaňuch2024; Kvist et al. Reference Kvist, Giralt, Valera, Hoi, Kristin and Darchiashvili2011; Lefranc and Worfolk Reference Lefranc and Worfolk2022; Schaub and Ullrich Reference Schaub and Ullrich2021). However, it is important to recognise that these declines are shaped by a complex interplay of factors, including not only land-use change but also climatic patterns (Pasinelli et al. 2010; Kvist et al. Reference Kvist, Giralt, Valera, Hoi, Kristin and Darchiashvili2011; Sándor and Domşa Reference Sándor and Domşa2018; Schaub and Ullrich Reference Schaub and Ullrich2021).
To stabilise or reverse the decline of Red-backed Shrike in our study area and beyond, conservation measures should prioritise the maintenance of open mosaic landscapes, as widely recommended across the species’ range. These landscapes should feature extensively managed meadows interspersed with sparse, structurally diverse thorny vegetation ideal for nesting, typically covering 10–30% of the territory, along with hedgerows and elevated perches which are used for hunting (Brambilla et al. Reference Brambilla, Rubolini and Guidali2007; Nijssen et al. Reference Nijssen, Geertsma, Kuper, van den Burg, van Duinen and Versluijs2024; Pedersen et al. Reference Pedersen, Schnedler-Meyer, Ekberg and Tøttrup2018). Given the species’ dependence on large invertebrates and small vertebrates, intensive agricultural practices involving pesticides and fertilizers should be avoided, as they are linked to reductions in prey availability. Instead, encouraging low-intensity grazing and mowing (while leaving unmown patches), as well as maintaining moist habitats such as wet meadows, can enhance arthropod abundance and improve nesting success (Casale et al. Reference Casale, Bionda, Falco, Siccardi, Toninelli and Rubolini2012; Kmecl et al. Reference Kmecl, Figelj, Tout, Bužan and Pallavicini2014; Tryjanowski et al. Reference Tryjanowski, Karg and Karg2003).
Although our study focused on a relatively small and isolated area, historically one of the densest Red-backed Shrike breeding sites reported in the literature, it reflects broader trends affecting farmland birds across Europe. While some gaps in habitat and bird data, particularly regarding temporal coverage, mean the results should be interpreted with caution, the patterns observed align closely with well-documented processes of agricultural intensification and land abandonment. Despite the species currently listed as Least Concern and its overall population trend described as stable (BirdLife International 2021, 2024; PECBMS 2024), recent figures reveal a 9% decline over the past decade and a 22% long-term decrease. Our findings suggest that this apparent stability conceals sharp regional collapses: in one of its former strongholds, the Red-backed Shrike has experienced a 92% decline over 33 years, largely due to the on-going degradation of semi-natural habitats, driven by land-use intensification, abandonment, and insufficient policy action.
Alarmingly, this decline has occurred despite protective designations and regulations intended to safeguard traditional farming landscapes (Official Gazette of Municipalities Ormož and Ptuj 1979), revealing the inadequacy of current agri-environmental schemes. In Slovenia, only around 5% of the agricultural policy budget is allocated to biodiversity conservation (Šumrada et al. Reference Šumrada, Lovec, Juvančič, Rac and Erjavec2020), and existing measures have failed to halt the loss of key habitats such as extensive meadows (Kaligarič et al. Reference Kaligarič, Čuš, Škornik and Ivajnšič2019). An ex-ante evaluation by Lovec et al. (Reference Lovec, Šumrada and Erjavec2020) of CAP implementation during 2014–2020 found that while funding formally aligned with stated objectives, the actual relevance and impact of measures, especially those addressing environmental goals, were weak, fragmented, and poorly targeted. Our findings reinforce these concerns, suggesting that detrimental effects on biodiversity began during the early stages of CAP implementation, and that the structural shortcomings identified by Lovec et al. (Reference Lovec, Šumrada and Erjavec2020) are already becoming obvious in tangible ecological declines.
Addressing these issues will require more than incremental adjustments. A fundamental rethinking of CAP design is urgently needed, one that strengthens the coherence between objectives and instruments, prioritises ecological effectiveness, and embeds rigorous impact evaluation throughout the policy cycle. Most importantly, our study highlights the risk of complacency: even species deemed secure at the continental scale can suffer rapid, unnoticed declines at the regional level. Without targeted, regionally adapted conservation strategies to maintain and restore diverse grassland mosaics, we risk losing not only Red-backed Shrikes, but a broader suite of species tied to Europe’s vanishing meadows. Safeguarding these habitats is no longer just a conservation priority, it is an ecological imperative.
Acknowledgements
We are grateful to volunteers Tilen Basle, Nataša Bavec, Vanesa Bezlaj, Dominik Bombek, Dejan Bordjan, Al Božič, Marko Bunderla, Mitja Denac, Gregor Domanjko, Matej Gamser, Gregor Gros, Jakob Habicht, Eva Horvat, Nikolaj Jelatancev, Rene Karner, Matjaž Kerček, Lana Klemenčič, Marcel Knikar, Ina Knikar, Aleksander Koren, Luka Korošec, Aleks Kotnik, Vanesa Kozina, Janez Leskošek, Simon Marčič, Anamarija Mihovec, Dominika Mihovec, Julija Mihovec, Matija Mlakar Medved, Jure Novak, Tadej Pipan, Borut Pittner, Alen Ploj, Monika Podgorelec, Mojca Podletnik, Luka Poljanec, Nejc Poljanec, Matjaž Premzl, Petra Radolič, Bia Rakar, Simon Sajko, Cene Skrt, Jakob Smole, Željko Šalamun, Borut Štumberger, Aleš Tomažič, Rene Turner, Barbara Vidmar, Mitja Vranetič, and Eva Vukelič for the surveys. Authors contributions: study conception (DD, RL), data collection (DD, RL), data analysis lead (DS), data analysis support (RL, DD), manuscript lead (DS, RL), and manuscript support (DD).
Supplementary material
The supplementary material for this article can be found at http://doi.org/10.1017/S0959270925100191.
Open access
