The potential of fallow management to promote steppe bird conservation within the next EU Common Agricultural Policy reform

Abstract

1. Agricultural intensification promoted by the European Common Agricultural Policy (CAP) has driven the decline of farmland and steppe bird populations. Policy tools to improve the environmental performance of the CAP—including Agri-Environmental Schemes (AES) and Greening—have often failed, and the new EU agricultural reform (CAP post-2020) offers a new opportunity to integrate effective measures addressing farmland bird declines. Fallow land and its management have proven beneficial for endangered steppe bird species by providing good quality habitat, and therefore has potential to become an effective conservation measure.
2. We used a Hierarchical Distance Sampling community model to evaluate the ability of different conservation regimes to increase the abundance of 37 bird species including endangered steppe birds and other farmland birds in 13,309 ha of fallow land in north-eastern Spain. The conservation regimes were based on different management prescriptions associated with AES, Greening and a local conservation measure promoting extensive fallow management targeting seven steppe bird species (Targeted Fallow Management, TFM).
3. The positive effect of conservation measures increased as their design was more targeted to specific species. TFM increased the abundance of target and other farmland species, while AES and Greening had either no effect or negative effects on bird abundance, respectively. Effects of other Greening conservation measures related to landscape heterogeneity such as crop richness and field size were variable across the community.
4. Policy implications. The success of Targeted Fallow Management as a conservation tool—in contrast to Agri-Environmental Schemes and Greening—highlights the value of applying 1–2 agricultural practices just before the breeding season in fallows situated in optimal locations for target steppe bird species, to increase the abundance of these and other farmland bird species. We translate our findings into specific guidelines that we recommend including within the new eco-schemes and AES present in the CAP post-2020.

1 INTRODUCTION

Farmland habitats cover approximately half of Europe's land surface (Kleijn et al., 2011) and have long experienced biodiversity loss (Pe'er et al., 2014). Agricultural intensification has increased since the inception of the Common Agricultural Policy (CAP) in 1962 and has been the main driver of the steep decline in farmland bird populations in Europe (Voříšek et al., 2010).

Several reforms have attempted to counteract the environmental drawbacks of the CAP, starting in 1992 with the McSharry reform and the implementation of Agri-Environmental Schemes (AES), subsidies for farmers to compensate for the loss of income associated with environmentally friendly practices (Batáry et al., 2015). Subsequent reforms have attempted to complement AES (e.g. decoupling of subsidies from production; Oñate, 2005). However, it was not until the last programming period (2014–2020) that direct payments to farmers were introduced conditional on compliance with three mandatory ‘greening measures’: maintaining permanent grassland, growing a minimum of three different crops and establishing Ecological Focus Areas (EFA)—landscape elements considered important for biodiversity—on 5% of arable land (European Commission, 2013).

In spite of these efforts, European farmland biodiversity remains threatened (Pe’er et al., 2014). Fallow land is a critical EFA for biodiversity (Pe’er et al., 2017), yet its surface decreased by 18% between 2015 and 2018 (European Commission, 2018). AES have complemented EFAs in fostering fallow land, but have not stopped its decrease (Traba & Morales, 2019). Fallow loss is of great concern for farmland birds in Europe (Voříšek et al., 2010), and it has been linked to steppe bird population declines in Spain (Traba & Morales, 2019), as fallows are key for feeding, mating and nesting (de Juana, 2005). The Iberian Peninsula constitutes the European or global stronghold of many steppe bird populations (Burfield, 2005) as it harbours the so-called ‘pseudo-steppes’, extensive areas of cereal fields alternated with fallows as part of a crop-rotation system (Sainz Ollero, 2013). Contrary to the large and continuous extent of the Iberian pseudo-steppe, other European pseudo-steppe areas in France, Italy and the Pannonian region are small and isolated (Burfield, 2005). Fallow land in semi-arid farmlands outside Europe is also increasingly important to buffer the impact of agriculture (e.g. Central Kazakhstan; Kamp et al., 2011).

Steppe birds have narrow micro-habitat requirements, depending on specific vegetation height and cover, and food resources (Robleño et al., 2017). Suitable vegetation structure is species-specific and can be achieved by applying different agricultural practices before the breeding season (Sanz-Pérez et al., 2019). Promoting the presence and management of fallow land linked to conservation goals on the future CAP Agenda is critical considering that several steppe bird species have become endangered (Burfield, 2005).

The new CAP post-2020 will downscale its legislation from the European to the Member State level, which will provide greater flexibility to address environmental needs (European Commission, 2019). Greening will be substituted by both compulsory and voluntary measures (incentives to adopt practices beneficial for the environment called 'eco-schemes'; Pillar I; European Commission, 2019). Agri-Environmental Schemes will be developed by each Member State (Pillar II), enabling further flexibility at the regional scale (European Commission, 2019). The Pillar II will likely receive ~25% from the CAP post-2020 budget (European Council, 2020), and at least 30% of the Pillar II budget will target environmental issues (European Commission, 2019). Using this budget to promote fallow presence and its management could contribute to halting farmland and steppe bird population declines (Tarjuelo et al., 2020). However, further evidence on the ability of fallow management to enhance bird abundance is critical to develop and advocate for science-based policy changes.

Here, we evaluate the effect of fallow management on the abundance of the farmland bird community of an Iberian pseudo-steppe (Lleida Plain, north-eastern Spain). Three types of conservation measures occur in the area, consisting of fallow fields with different management prescriptions: (1) Targeted Fallow Management (TFM), which is a regional conservation measure promoting extensive fallow management to benefit specific specialist steppe bird species (hereafter 'target species'; Sanz-Pérez et al., 2019); (2) AES, which are also aimed at the steppe bird community but adopt more generic management prescriptions (Generalitat de Catalunya, 2020b) and (3) Greening EFAs, which are aimed at biodiversity in general (European Commission, 2013).

We used a Hierarchical Distance Sampling community model to test the effect of fallow surface under the three management regimes on the abundance of 37 farmland bird species of different conservation status. We predicted that the positive effect of conservation measures will increase as their design more explicitly focuses on the requirements of target species. Specifically, we expect that TFM has a positive effect on the abundance of bird species considered as steppe specialists (its target species), as it includes management guidelines to fulfil their ecological requirements (e.g. diverse agricultural practices to ensure optimal vegetation structure and food availability; Figure 1). Although to a lesser extent than TFM, we expect AES to be more efficient than Greening in enhancing steppe and farmland bird abundance, as it includes beneficial management guidelines such as avoiding any form of agricultural management during the breeding season (Figure 1).

FIGURE 1

Main ecological requirements of target steppe bird species in an Iberian cereal steppe of north-eastern Spain (Lleida plain) linked to the management needed to meet their requirements, and consequent hypotheses on the success (green) or uncertain outcomes (orange) of the three conservation tools in enhancing bird abundance: Targeted Fallow Management (TFM), Agri-Environmental Schemes (AES) and Greening (GREEN)

Landscape heterogeneity has shown similar or larger positive effects on biodiversity than farmland management, but with varying effects on the steppe bird community (Concepción & Díaz, 2011; McMahon et al., 2010), as steppe birds are often specialists of homogeneous landscapes (Filippi-Codaccioni et al., 2010). Greening measures have promoted crop richness and the preservation of field borders, aiming at increasing overall biodiversity through landscape heterogeneity. We therefore also investigated the effects of crop richness and field size on the abundance of the farmland bird community.

2 MATERIALS AND METHODS

2.1 Study area and data collection

The study area was located in the Lleida steppe plain (~3,580 km²; Catalonia, NE Spain). This is a semi-arid landscape characterized by an agricultural mosaic with extensive cultivation of winter cereal crops, woody crops (olive and almond), annual fallow fields and sparse natural shrubland. Extensive grazing is present but generally rare in the area.

The study design consisted of 152 transects of 500-m length placed randomly throughout the study area, at a minimum distance of 1,000 m to ensure independence (Buckland et al., 2001). Seventy-five percent of the transects were located within Special Protection Areas mainly designated for steppe bird conservation, belonging to the Natura 2000 European protected areas network. The remaining transects were within steppe-like habitats with similar climatic conditions and landscape characteristics than Special Protection Areas. Transects were sampled annually during the peak breeding season (May) from 2015 to 2019. Bird surveys were performed by seven professional observers between 6 and 10 a.m. in good weather conditions (i.e. no rain, wind speed <20 km/h and temperature between 15 and 30°C). Each survey was conducted by a single observer, who walked along the transect and collected data following a distance sampling protocol (Buckland et al., 2001). Birds were recorded on both sides of the transect when first observed either visually or aurally, and observations were assigned to five distance categories (0–25, 25–50, 50–100, 100–200 and 200–500 m). We limited our analysis to farmland species (i.e. showing general habitat selection patterns for at least one extensive agriculture or dryland habitat according to Estrada et al., 2004) that were not migrating during the survey period, and with more than 15 detections throughout the study. This included 26 common species and 11 species of conservation concern at the European and/or regional scale (Table 1). Seven species of the community—including both common and endangered species—were steppe birds constituting the target species of the TFM conservation measure (Table 1; Section 2.3.1).

TABLE 1. Information on the 37 farmland bird species included in the HDS community model to study the effect of agricultural management on bird abundance in an Iberian cereal steppe of north-eastern Spain (Lleida plain) during 2015–2019. Target species of the Targeted Fallow Management conservation measure are marked by a cross

Common name	Scientific name	Target species^a	Conservation status		Number of detections	Transects occupied (%)
			EU 27^b	Catalonia^c
Eurasian skylark	Alauda arvensis		LC	LC	70	6.17
Red-legged partridge	Alectoris rufa		LC	LC	169	19.01
Little owl	Athene noctua		LC	VU	32	4.01
Eurasian Stone-curlew	Burhinus oedicnemus	X	LC	LC	170	18.8
Greater short-toed lark	Calandrella brachydactyla	X	LC	EN	71	4.56
European goldfinch	Carduelis carduelis		LC	LC	206	17.54
European greenfinch	Carduelis chloris		LC	LC	117	11.36
Lesser short-toed lark	Alaudala rufescens		LC	LC	53	2.81
Common linnet	Linaria cannabina		LC	LC	138	10.56
Montagu's harrier	Circus pygargus		LC	VU	37	4.68
Great spotted cuckoo	Clamator glandarius		LC	VU	46	5.25
European roller	Coracias garrulus	X	LC	LC	130	14.18
Feral pigeon	Columba livia var. domestica		LC	LC	75	8.83
Eurasian jackdaw	Corvus monedula		LC	VU	387	32.52
Stock dove	Columba oenas		LC	LC	121	13.09
Common wood pigeon	Columba palumbus		LC	LC	616	50.07
Common house martin	Delichon urbicum		LC	LC	26	3.35
Eurasian hobby	Falco subbuteo		LC	LC	26	2.95
Common kestrel	Falco tinnunculus		LC	LC	120	15.02
Crested/Thekla's Lark	Galerida sp.		LC	LC	2,931	92.21
Barn swallow	Hirundo rustica		LC	LC	505	44.44
Iberian grey shrike	Lanius meridionalis		VU	EN	60	7.09
Woodchat shrike	Lanius senator		LC	LC	84	8.56
Woodlark	Lullula arborea		LC	LC	157	13.82
European bee-eater	Merops apiaster		LC	LC	425	39.61
Calandra lark	Melanocorypha calandra	X	VU	LC	2,731	68.72
Corn bunting	Emberiza calandra		LC	LC	3,335	85.08
Eurasian tree sparrow	Passer montanus		LC	LC	284	23.31
Rock sparrow	Petronia petronia		LC	LC	31	3.48
Eurasian magpie	Pica pica		LC	LC	775	60.03
Pin-tailed sandgrouse	Pterocles alchata	X	LC	VU	80	6.82
Black-bellied sandgrouse	Pterocles orientalis	X	EN	EN	17	1.07
Red-billed chough	Pyrrhocorax pyrrhocorax		LC	LC	89	8.96
European serin	Serinus serinus		LC	LC	177	13.51
European turtledove	Streptopelia turtur		NT	LC	69	8.02
Little bustard	Tetrax tetrax	X	VU	EN	282	25.38
Eurasian hoopoe	Upupa epops		LC	LC	318	31.92

• ^a Target species of the TFM conservation measure (Mañosa et al., 2020).
• ^b European conservation status according to the IUCN Red List assessment for the 27 EU Member States (LC = Least Concern; NT = Near Threatened; VU = Vulnerable; EN = Endangered; BirdLife International, 2015).
• ^c Regional conservation status according to the Catalogue of Endangered Species in Catalonia (pending approval; Generalitat de Catalunya, 2020a; LC = Least Concern; VU = Vulnerable; EN = Endangered).

2.2 Hierarchical Distance Sampling community model

We fitted a multi-year HDS community model following Sollmann et al. (2016). The process component models local abundance N for a given species s at a transect j and year t following a Poisson distribution (1). Variation in expected abundance is then modelled on the log scale as a function of an intercept and site- and year-specific covariates (2):

(1)

(2)

(3)

(4)

where the species and year-specific intercept is modelled as a random effect with hyperparameters and accounting for the dependence of the data within years for each species (3), and is a random site effect that accounts for the dependence of the data within transects for each species (4). The beta coefficients are related to the site- and year-specific habitat covariates . Under the community model approach, the parameters are species-specific and are modelled with a normally distributed random effect (5):

(5)

where and (the hyperparameters) constitute the community parameters shared by all species.

The observation component of the model links the true abundance of a given species at a site and year to the raw counts through the probability of detection p (6):

(6)

In distance sampling, p is assumed to be perfect at the transect line and to decrease as a function of perpendicular distance x from the transect following a detection function (Buckland et al., 2001). We used a half normal detection function (7):

(7)

where is the scale parameter. With binned distance observations, detection probability for each distance bin k can be calculated as the integral of g(x) over the break points of k:

(8)

where b are the K + 1 breakpoints of the K distance categories and is the width of the kth distance category. In our study, was 25, 25, 50, 100 and 300 m from the first to the fifth distance category, and the strip width was 500 m. Because individuals are assumed to be uniformly distributed around the transects, the individual probability of occurrence in a distance bin is

(9)

Therefore, the cell probability of detection is , and the overall probability of detection (, Equation 6) is the sum over all (Kéry & Royle, 2016).

We modelled from the half normal detection function (7) on the log scale (10). The intercept constituted a species random effect with hyperparameters and (11) and observer was included as random effect following a zero-mean normal distribution with variance (12).

(10)

(11)

(12)

2.3 Predictors of bird abundance

We extracted year-specific variables within 500-m buffers around each transect. For the fallow variables, we extracted the area in hectares (ha) of each fallow type (TFM, AES and Greening) within the buffer of each transect and year (Table 2) and log-transformed these areas. The AES fallow type presented the lowest average area per transect (mean [SD] = 2.16 ha [6.03]), corresponding to half of the average TFM fallow area and one-third of the average Greening fallow area (Table 2). We also extracted landscape heterogeneity variables, related to landscape composition (crop richness) and landscape configuration (field size). We scaled all covariates for analyses, and we modelled all covariate effects as fixed across years.

TABLE 2. Information on fallow field types included in a HDS community model to determine the effect of fallow management on the abundance of a farmland bird community in an Iberian cereal steppe of north-eastern Spain (Lleida plain) during 2015–2019

	TFM^a	AES^b	GREEN^c
Summary statistics
Mean (SD)	4.52 (9.77) ha	2.16 (6.03) ha	5.92 (6.07) ha
Range	0–90.12 ha	0–69.19 ha	0–51.06 ha
Total area	4,771.22 ha	2,282.71 ha	6,255.08 ha
% of transect (buffer) area across years	3.85%	1.63%	4.82%
Main features
Target	Steppe birds (Target species; Table 1)	Farmland bird community	Biodiversity
Who selects fallow fields	Experts conditional on agreement of farmers	Farmers (voluntary measure for extra payment)	Farmers (compulsory measure for basic payment)
Criteria to select fallow fields	Suitable conditions for target species (e.g. location, slope)	Minimum size: 0.5 ha	None
Forbidden management	None (but avoid herbicide when possible)	Herbicide	Herbicide
Most common management applied	- Shredding - Ploughing - Alfalfa sowing - Grazing - No management	Ploughing	Ploughing
Periodicity of management (prescription)	1–3 times per year	Minimum once every two years	None
Most frequent periodicity of management	1–2 times/year	More than 2–3 times/year	More than 2–3 times/year
Criteria to select type and periodicity of management	Suitable vegetation structures for target species requirements (expert's criteria)	Weed control (farmer's criteria)	Weed control (farmer's criteria)
Timing of management	Before breeding season (1st February–15th April)	Wide period (1st September–15th April)	All year
Evaluation and adaptation	Yearly	None	None
References	Giralt et al. (2018) and Sanz-Pérez et al. (2019)	Generalitat de Catalunya (2020b)	Generalitat de Catalunya (2019a)

• ^a Targeted Fallow Management fallow fields.
• ^b Agri-Environmental Schemes fallow fields.
• ^c Greening fallow fields.

2.3.1 Fallow variables

Fallow fields under TFM belonged to a local compensatory conservation measure included in the Environmental Impact Assessment of the Segarra-Garrigues irrigation system. The measure consists in the rental and management of fallow fields by the regional government within Special Protection Areas to promote optimal habitat for seven target species, which where the main steppe bird species found in the study area (Table 1; Mañosa et al., 2020). TFM targets bird species considered as steppe specialists because they are especially vulnerable to agricultural intensification and most of them are endangered (Table 1). TFM has occurred annually since 2014, and consists of specific agricultural practices (Table 2; Sanz-Pérez et al., 2019). The exact timing and type of agricultural practice are adapted to the target species present in each Special Protection Area to meet species-specific requirements during breeding (Giralt et al., 2018; Mañosa et al., 2020).

Fallow fields under AES aim to benefit the whole farmland bird community (Table 2) with special emphasis on steppe birds. Because AES fallow fields were implemented voluntarily by farmers in exchange for subsidies, their location was not always optimal for steppe birds (e.g. next to a road). The AES management prescriptions consist of applying at least one agricultural practice every 2 years between September and April (Table 2). However, farmers often perform intensive management before the breeding season (Table 2; Giralt et al., 2018) resulting in fallow fields mostly cleared from vegetation.

The fallow fields under Greening were acquired by farmers as a type of EFA (chosen among other EFA types) to receive the basic CAP payments. Greening prescriptions for fallow management are targeted towards biodiversity in general, and therefore are very generic, with no timing or periodicity restrictions regarding management (Table 2).

Some fallow fields were under TFM, AES and/or Greening simultaneously. In those cases, we assigned a fallow field to the category with the most targeted management for steppe and farmland bird conservation (i.e. TFM > AES > Greening).

2.3.2 Landscape heterogeneity variables

We quantified crop richness, defined as the number of different crops within each buffer per year, using an annual crop land use map from the regional government (Unique Agrarian Statement/DUN; Generalitat de Catalunya, 2019b). We used the same crop classification as regional farmers do to receive Greening payments (Appendix A; Generalitat de Catalunya, 2019a).

We quantified field size using the regional Geographic Information System of Farming Land (SIGPAC; Generalitat de Catalunya, 2019c). We calculated the yearly average field size for a transect by averaging the total area (ha) of all agricultural fields intersecting with its buffer.

2.4 Model implementation

We implemented the model in a Bayesian framework, using the software JAGS version 4.3.0 (Plummer, 2003), accessed through the jagsUI r package version 1.5.0 (Kellner, 2018). The model code is available in Appendix D. We used normal (0, 10) and uniform (0, 500) priors for the mean and SD hyperparameters of the species random effects, respectively, and uniform (0, 10) priors for the SD of the observer random effect. We ran three parallel Markov chains with 500,000 iterations and a burn-in of 50,000 iterations, thinning chains by 10. We tested for chain convergence using the Gelman–Rubin statistic (values <1.1; Gelman et al., 2013). We assessed model fit by calculating Bayesian p values (Gelman et al., 1996) based on Freeman–Tukey residuals (Appendix C). We report parameter estimates as the posterior means and standard deviations. We considered coefficients as significant when their 95% Bayesian Credible Interval did not overlap zero. We calculated the posterior probability of a positive effect of a predictor variable as the proportion of all posterior samples of the respective coefficient >0.

3 RESULTS

3.1 Community response

The only variable showing a significant positive effect on mean community abundance was TFM fallow , in contrast to the negative, but marginally non-significant, community-level effect of Greening fallows . AES fallow fields had no significant effect on mean community abundance . The community showed a significant negative response to crop richness and field sizes .

3.2 Species-specific abundance

The only variable with significantly positive species-specific effects was TFM fallow area, significantly increasing the abundance of four target species (Pin-tailed sandgrouse, Little bustard, European roller and Eurasian Stone-curlew; see Table 1 for scientific names) and four other species (Figure 2). The posterior probability of a positive effect was >70% for 20 species, including five target species (Figure 2). TFM fallow fields had a significant negative effect on the abundance of two species (Figure 2). Species-specific effects of AES fallow area were non-significant, and the posterior probability of a positive effect was >70% for only one species (Figure 2). Similarly, the posterior probability of a positive effect of Greening fallow fields was >70% for only three species, including one target species (Figure 2). Greening fallow fields did, however, have a significant negative effect on the abundance of three species, including one target species (Greater short-toed lark; Figure 2).

FIGURE 2

Violin plots showing the posterior distributions (black outline: full posterior; white points: mean; inner grey polygon: 95% credible interval) of the species-specific coefficients for the variables Targeted Fallow Management (TFM), Agri-Environmental Schemes (AES) and greening (GREEN), estimated within a HDS community model for 37 farmland species sampled during 2015–2019 in an Iberian cereal steppe (Lleida Plain, Spain). Species are presented by decreasing values of the Fallow TFM coefficient; target species are highlighted in grey. Significant effects are indicated by an asterisk

3.3 Other predictors

Crop richness had a significant positive effect on the abundance of four species (including the target species Little bustard; Figure 3) and a posterior probability of a positive effect >70% for 14 species including four target species but also had significant negative effects on 11 species (including the target species Pin-tailed sandgrouse and Greater short-toed lark; Figure 3). Field size had a significant positive effect on the abundance of two species (including the target species Black-bellied sandgrouse) and the posterior probability of a positive effect was >70% for six species, including three target species (Figure 3). Field size had a significant negative effect on the abundance of nine species (including the target species Little bustard). Bird populations showed only weak fluctuations over time indicating population stability, according to (Table B1, Appendix B).

FIGURE 3

Violin plots showing the posterior distributions (black outline: full posterior; white points: mean; inner grey polygon: 95% credible interval) of the species-specific coefficients for crop richness and field size (i.e. greening measures), estimated with a HDS community model for 37 farmland species sampled during 2015–2019 in an Iberian cereal steppe (Lleida plain, Spain). Species are presented by decreasing values of the crop richness (left panel) and field size (right panel) coefficients; target species are highlighted in grey. Significant effects are indicated by an asterisk

3.4 Model fit

The model presented a good fit (i.e. Bayesian p value >0.1 and <0.9; Table C1, Appendix C) for the abundance component of the community and individual species, and for the detection component of all species except the Calandra lark and Corn bunting, causing a low community-wide Bayesian p value. Lack of fit for the two problem species was due to very few extreme residuals (for 12 and 9 of the 760 transect-year combinations, respectively). Thus, we considered that the low community Bayesian p value did not invalidate overall model results, but that abundance estimates for these species-year-transect combinations may be inaccurate.

4 DISCUSSION

Our findings indicate that the efficiency of the principal CAP conservation tools to enhance the abundance of a farmland bird community, including highly specialized and endangered steppe birds, depended to a great extent on the degree of targeted management for specific species. Targeted Fallow Management increased the abundance of most steppe birds (target species) and other farmland birds as expected, yet AES did not benefit the community. Greening fallows showed no or even negative effects on steppe bird abundance, confirming our expectations. Non-fallow Greening measures promoting landscape heterogeneity showed variable effects across the community, being mostly negative or neutral for specialist steppe bird species.

4.1 Targeted Fallow Management

We expected TFM to benefit steppe birds because it targets these species both in spatial location (Mañosa et al., 2020) and management prescription (Table 2; Sanz-Pérez et al., 2019). Steppe birds can be highly specialized; thus, specific measures shaping vegetation structure of fallow fields are essential to meet their requirements (Robleño et al., 2017). Our results validate the relationship between TFM and habitat suitability for steppe bird occurrence found by Sanz-Pérez et al. (2019), which is further corroborated by the recent increase in the populations of some of the studied species (e.g. the Pin-tailed sandgrouse; Bota et al., 2020; Mañosa et al., 2020).

Our results also demonstrate the potential of TFM to increase the abundance of the entire farmland bird community. Applying different agricultural practices creates a landscape mosaic of different fallow types that allows niche segregation and benefits not only target species but also other farmland birds. Indeed, TFM had a high probability of benefitting populations of other common (e.g. Red-billed chough) and endangered farmland species (e.g. Montagu's harrier). These results suggest that steppe species could be considered an umbrella group for the farmland bird assemblage, likely owing to their high co-occurrence and similar sensitivities to disturbance (Fleishman et al., 2000; see also Moreno et al., 2013). Our results contradict findings by Santana et al. (2014) suggesting that flagship steppe bird species conservation within Special Protection Areas does not benefit the broader bird community, which is likely due to different landscape contexts and/or conservation measures (e.g. fallow land has increased by 17% in our study area, in contrast to its declining trend in the study of Santana et al., 2014).

4.2 Agri-Environmental Schemes

In contrast to our expectations, AES did not benefit the community or species-specific abundances. AES has previously been shown inefficient to enhance endangered species abundance due to poor targeting (Kleijn et al., 2006), and it is likely that AES management prescriptions still allow for excessive management by farmers. Farmers consider that fallows promote harmful weeds and apply intensive weed control (i.e. ploughing >2–3 times/year; Giralt et al., 2018), which likely results in structurally simple and similar fallows that could result unsuitable for most steppe bird species at the start of the breeding season. Alternatively, the presence of TFM (i.e. birds selecting TFM over AES fallow fields) or the low prevalence of AES in the study area may explain its lack of effectiveness (see also Kleijn et al., 2011).

4.3 Greening

Greening EFA fallows did not increase community abundance, which is in accordance with the predicted low success of EFAs in enhancing animal populations (Pe’er et al., 2014, 2017). Greening and AES fallows have common regulations, and therefore some of the reasons behind their lack of success are probably shared. Greening fallows further allow agricultural management during the breeding period, which could cause the negative effects observed for some species. Greening measures promoting landscape heterogeneity had variable results across the community. Habitat heterogeneity does not generally benefit steppe specialists, which are usually ground-nesting species linked to structurally simple habitats (Filippi-Codaccioni et al., 2010; Pickett & Siriwardena, 2011). Our study supports this notion for most target species except the Little bustard, probably because of its need of habitat complementary to fulfil the requirements of its life cycle (Morales et al., 2008). Promoting crop structural diversity (i.e. involving crop management and vegetation structure) rather than general crop diversity has already been advocated for the CAP post-2020 (Josefsson et al., 2017), as it could benefit ground-nesting species such as our target species.

4.4 Conservation implications

The TFM fallow land evaluated within our study represents exceptional conditions for successful target species conservation (e.g. fallow management in optimal areas, expert criteria to choose timing and type of management, exhaustive monitoring). Although the features of TFM are probably too costly and specific to become a general policy prescription, our results provide a basis for developing guidelines towards conservation of farmland and steppe birds. Here, we translate the characteristics of TFM that make them efficient for conservation into specific recommendations for eco-schemes and AES within the CAP post-2020.

Voluntary eco-schemes are considered simple measures attractive to farmers designed by each Member State. We recommend the inclusion of fallow land as an eco-scheme with two simple requirements essential for its success: (a) no agricultural management during the breeding season and (b) guaranteeing the presence of some vegetation cover at the beginning of the breeding season, avoiding bare soil fields.

Agri-Environmental Schemes can be designed and applied at national or regional level (European Commission, 2019), which makes them the perfect policy framework for adjusting conservation measures to local conditions and specialist species (see also Kleijn et al., 2006). We propose adding two requirements to AES regimes: (a) Limiting the number of management actions to 1–3 times/year, outside the breeding season. Vegetation encroachment resulting from fallow land abandonment is as detrimental to steppe bird habitat suitability as excessive management (Sanz-Pérez et al., 2019), and is also despised by farmers, for fear that it will hamper future crop productivity. Promoting moderate fallow management might help changing farmers’ attitudes towards fallows and result in a win–win strategy (Tarjuelo et al., 2020). (b) Aligning the type and timing of the agricultural practices applied with the conservation goals of each Special Protection Area, to promote suitable fallows adapted to the species with priority conservation status in each Special Protection Area.

The European cereal steppe system is a globally significant hotspot for steppe bird diversity and conservation, and our findings are thus of high value for EU Member States harbouring this system. Beyond that, they have the potential to inform bird conservation in cultivated areas of the Eurasian steppe belt, where fallow management could constitute a tool to combat the ongoing land abandonment and benefit steppe birds (Ioffe et al., 2012; Kamp et al., 2011). Moreover, the presence of fallow land and its management has proved positive for sustaining farmland bird populations in central and northern Europe (e.g. Bracken & Bolger, 2006; Doxa et al., 2010), and elsewhere (e.g. Van Buskirk & Willi, 2004). Our findings thus make an important contribution in the global search for efficient pathways to conserve endangered species in agricultural systems where food production and biodiversity need to coexist.

ACKNOWLEDGEMENTS

We acknowledge the field assistance of J. Estrada, S. Sales, J. Castelló, M. Anton, A. Bonan, X. Larruy, A. Petit, F. González, J. Bécares, F. Broto, X. Riera and D. Guixé. We acknowledge Cyril Milleret for providing useful comments. Departament d'Agricultura, Ramaderia, Pesca i Alimentació and Infraestructures de la Generalitat de Catalunya S.A.U. have funded relevant parts of the project. This work was partially supported by the Generalitat de Catalunya through a FI-predoctoral contract to A.S.-P. (2018FI_B1_00196). F.S.-P. and D.G. coordinated fieldwork and G.B. secured funding.

AUTHORS' CONTRIBUTIONS

D.G., F.S.-P., G.B. and A.S.-P. conceived and designed the study; N.P. gathered and organized the data; A.S.-P. and R.S. implemented the analysis. A.S.-P. wrote the manuscript with the help of the rest of co-authors. All the authors contributed to subsequent drafts and gave final approval for publication.

Open Research

DATA AVAILABILITY STATEMENT

Data available via the Dryad Digital Repository https://doi.org/10.5061/dryad.sn02v6x40 (Sanz-Pérez et al., 2021).