The importance of size, location, and vegetation composition of perennial fallows for farmland birds

Abstract Across Europe, patches of un‐cropped land (field margins, fallows, etc.) have been established and managed as part of agri‐environment schemes (AES) to counteract the decrease in farmland biodiversity. Various studies demonstrate a positive impact of such un‐cropped land on different taxa. However, there is potential to further improve the efficiency of fallow options for farmland birds. In a long‐term monitoring, 12 breeding farmland bird species and sizes of perennial fallows were recorded from 1992 to 2015 in a 6.1 km2 area in Switzerland. Furthermore, habitat composition and fallow characteristics were mapped in 2012. We calculated population trends, analyzed habitat associations and revealed the impact of fallow habitat characteristics on territory density. The proportion of fallows in the study site increased from 1.4% (1992) to 8.5% (2012). Population trends of six of 12 censused species increased significantly over the same time, four species showed no trend and trends of two species decreased. Seven species were analyzed in more detail, for five of them fallows were overrepresented around their territory center points compared to arable fields and grassland. The overall territory density of these five species was higher in small fallows which were not placed next to a wood and which held bramble rubus spp., shrubs and the tall‐growing forb goldenrod (Solidago canadensis and S. gigantea). Our study confirms that perennial fallows are a highly suitable option to support different farmland birds in arable landscapes. Yet, we recommend optimizing fallows through careful site selection and management, such that they are not established on shady locations and are structurally diverse by allowing brambles, shrubs, and tall‐growing forbs to occur. We suggest adapting the Swiss AES in this regard. Biodiversity‐related advisory services available for farmers could increase the probability that fallow options are implemented and managed properly for targeted species.

(Caillet-Bois, Weiss, Benz, & Stäheli, 2018) as well as naturally vegetated field margins as part of a local AES. In this study, we concentrate on the three AES types sown wildflower area, rotational fallow, and naturally vegetated field margin and named them "fallows." In contrast to fallow options in other countries, in our study site the fallows were perennial (lasting for more than one vegetation period).
Being quite divers in spatial configuration, they were established with the aim to create fallow-like habitat by increasing structural and botanical richness in arable landscapes, thereby, for example, provisioning breeding habitat or food sources for birds.
Various studies report a positive influence of field margins and fallows on birds (Burgess et al., 2015;Henderson et al., 2012;Kuiper, 2015a,b;Meichtry-Stier, Jenny, Zellweger-Fischer, & Birrer, 2014;Tryjanowski, 1999) and other species (Haaland, Russell, & Bersier, 2011;Meichtry-Stier et al., 2014;Van Buskirk & Willi, 2004;and references therein). Despite these benefits, farmland bird indices are decreasing in the EU (PECBMS 2016) and Switzerland (see "Swiss Bird Index-Priority Species Agriculture" in Sattler, Knaus, Schmid, & Strebel, 2016). Possible reasons for this are low floristic and structural diversity (ecological quality) of implemented options and the low uptake of the most appropriate options by farmers (Birrer, Spiess, Herzog, Kohli, & Lugrin, 2007;Breeze, Bailey, Balcombe, & Potts, 2014;Hardman et al., 2015). A few studies described specifically what these qualities are for biodiversity and birds in particular. A management that "creates a structurally and floristically diverse sward" thereby providing resources such as insects and seeds as well as the "proximity to a good quality hedgerow" of field margins and fallows were shown to be beneficial for birds (Tscharntke, Batáry, & Dormann, 2011;Vickery et al., 2009). Different results are found about the optimal age and size of un-cropped land for birds, reflecting the diverse local conditions and landscape history (Flade, Plachter, Schmidt, & Werner, 2006;Henderson et al., 2012;Holland, Storkey, Lutman, Henderson, & Orson, 2013;Tscharntke et al., 2011;Zollinger et al., 2013). In terms of structural and floristic diversity, implemented options often look too similar to the crops around instead of creating an ecological contrast. Fallows in Switzerland are designed to increase the floristic and structural diversity ( Figure 1) and thus stand out from the neighboring uniform farmland, providing resources (e.g., food, nesting opportunities, or song posts) that are limited in the surroundings for birds (Batáry, Dicks, Kleijn, & Sutherland, 2015;Hammers, Müskens, van Kats, Teunissen, & Kleijn, 2015). Yet, the influences of the vegetation composition of un-cropped land on birds have been much less investigated (but see Marja & Herzon, 2012). Un-cropped land undergoes succession and especially in older ones, small shrubs, bramble or goldenrod may appear, or mainly grassy vegetation develops.
In Switzerland, fallows are often suboptimally placed on wet, steep, or shady areas (Herzog et al., 2005). There, botanical diversity is impaired because many species of the seed mix do not grow and are outcompeted by a few dominating species. We assume that many fallows could be optimized in location, size, and vegetation composition. To improve fallows for conservation management, more knowledge is needed about the preferred characteristics of such habitats for breeding farmland birds.
The study is embedded in a long-term monitoring of priority farmland birds (birds of conservation concern according to the Swiss legislation) in an ecologically improved landscape. In this project region, perennial fallows were established, providing the possibility to study the relationship between farmland birds and characteristics of the F I G U R E 1 Red-backed Shrike on a Wild Teasel, photo: Markus Jenny fallows. Our questions were threefold: (a) Do abundances of priority bird species increase over the long-term study period? We expect the trends to be more positive than European or Swiss bird indices, because of the comparably high percentage of fallows in the study region. (b) Do territories of priority farmland bird species contain more fallow area than expected according to the available habitat in the study site? Studies from other regions and countries revealed a preference for fallow habitat. (c) Which characteristics of fallows are related to territory density of priority farmland birds and is vegetation composition of the fallows related to territory density? Personal observations revealed that small shrubs, brambles, and goldenrod increase the attractiveness of fallows whereas a high amount of grass is less favored by priority bird species of open farmland.

| Study site
We conducted our study in the very southwest of the Swiss lowland, the Champagne genevoise (420-450 m a.s.l., 46°09′11″N, 6°01′25″E, canton Geneva), which has a warm and dry climate (mean temperature is 11.2°C, mean yearly precipitation is around 700 mm).
The study site (6.1 km 2 ) is dominated by intensively cultivated farmland (mainly winter cereals) and some large gravel pits (covering between 6.1% and 13.2% of the study area in all years) and holds only a few small settlements and woods (Table 1). The depleted gravel pits were filled up and then used again as farmland. Some hedgerows and small coppices structure the wider countryside, the hedges being quite diverse in density and height but mostly lower than 3 m. The landscape composition within the study site is simple (Tscharntke et al., 2011) with less than 20% semi-natural habitats (extensively used meadows and pastures, ponds, hedgerows, woods, ruderal areas). Mean field size is small (1.0 ha) compared to other arable landscapes of Western Europe. Nonetheless, the area is intensively farmed with fields under annual crops and infrequent hedgerows.
The only common semi-natural element is narrow grassy margins (approx. 1 m wide) along the roads.

| Fallows
In this study, fallows were either sown wildflower areas/strips, rotational fallows, or naturally vegetated field margins, showing different characteristics of structural and floristical diversity. Wildflower areas/strips and rotational fallows are options of the Swiss AES (Caillet-Bois et al., 2018), naturally vegetated field margins are part of a local AES. They are designed to halt the overall biodiversity loss in farmed landscapes without a special focus on birds or even single species. Since 1992, perennial fallows have been implemented by farmers in the study site, mostly according to instructions regarding site selection and management by one of the co-authors (BL).
They were either naturally vegetated (left fallow after harrowing) or were sown with species-rich wildflower seed mixes (minimum 17 plant species). In some of them, shrubs and bramble were cleared, in others bramble, shrubs (mainly common dogwood Cornus sanguinea, small willows Salix spp. and small poplars Populus spp.) and goldenrod (mainly Solidago gigantea, but also S. canadensis) occurred in different amounts therein.
Every few years, they were partially mown to reduce the expansion of goldenrod, bramble, and shrubs. Their spatial configuration was quite diverse, most of them being linear strips of 10-25 m width; others were of rectangular (whole parcels) or triangular shape. Size ranged from 0.06 to 3.3 ha (median 0.36 ha). Their age ranged from Category Description

Proportion of study area (%)
Fallow Naturally vegetated perennial un-cropped strips on arable farmland from a local AES, sown wildflower areas or sown rotational fallow from the Swiss AES (either as strip or area); fertilization and treatment with insecticides are not allowed; large-scale chemical or mechanical weed control is not allowed; cut max. once a year on half of the area 8.5 Hedgerow Hedgerows within farmland 0.8

| Data collection
In the long-term monitoring, we censused territory numbers of 12 breeding farmland bird species of conservation concern (Common Whitethroat Monitoring was conducted by walking along paths which allowed audio-visual observation of the entire study site. All observations of each bird species were recorded on maps following the breeding bird census method (Bibby, Burgess, Hill, & Mustoe, 2000;Schmid, Zbinden, & Keller, 2004). To assess breeding territories, species had to be observed during their breeding period in a suitable habitat (Schmid, Luder, Naef-Daenzer, Graf, & Zbinden, 1998)

| Statistics
All analyses were performed in R 3.3.1 (R Development Core Team 2015).
For each species, the population trend was calculated with a generalized linear model (GLM, package "arm") using a Poisson distribution. Territory numbers (response variable) were tested against year and the second and third polynomial of year to account for a To find out which characteristics of fallows were related to territory density, we calculated a habitat model using the data on Because fallows were of different sizes, we included the logarithm of size (in ha) as offset in the model. Thus, we modeled territory densities (number of territories per ha) rather than absolute numbers of territories. The proportions of bramble, goldenrod, and grass were arcsine-square root-transformed and distance to wood was square root-transformed. All explanatory variables were standardized with mean zero and SD = 1. No collinearity was found between the explanatory variables, as variance inflation factors were all <3 (Zuur, Ieno, Walker, Saveliev, & Smith, 2009 (Whittingham, Stephens, Bradbury, & Freckleton, 2006).
Variables were considered significant if their 95% CrI did not include zero. No spatial autocorrelation was found in the final model (checked by variogramm and bubble plot). We graphically assessed if model assumptions were met, as described above.

| Population trends
Territory numbers of six (Whitethroat, Stonechat, Melodious Warbler, Yellowhammer, Red-backed Shrike and Cirl Bunting) of the 12 yearly censused species in the study site increased significantly during our study (Table 3 and Figure 2). Two species (Ortolan and Whinchat) showed no significant trend, yet they were both nonregular breeders in the study site and occurred only in a few years and in very low numbers. The Corn Bunting and the Western Yellow Wagtail showed (the latter on a low level) an increase for a few years followed by a decrease. Only two species (Common Quail and Northern Lapwing) decreased significantly over the years. However, Quail territory numbers were strongly fluctuating throughout the study and Lapwing numbers had been on a very low level from the beginning of the census.

| Compositional analysis
The association of the studied species with habitat categories was not random (randomization test, p-values for all species <0.004). For four of the seven analyzed bird species (Whitethroat, Stonechat, Melodious Warbler and Red-backed Shrike) fallows ranked first before all other habitat categories, that is, fallows were strongly overrepresented in the pseudoterritories (Figure 3, see Supporting Information Appendix S1). The Yellowhammer was associated with fallows and hedgerows. For the Skylark, arable fields, fallows, and hedgerows were ranked highest, whereas the Corn Bunting showed no association with a single habitat category.

| Habitat model
The final model revealed that the density of bird territories was strongly influenced by size, distance to wood, and vegetation composition (proportion of bramble, goldenrod, and shrubs) of the fallows (Table 4).
Territory density was negatively correlated with size but positively with the distance to wood and the proportion of bramble (Figure 4a-c).
Territory density showed a significant relationship with the proportion of the neophyte goldenrod, with an optimum at approximately 37% goldenrod per fallow (Figure 4d). The positive effect of shrubs on territory density was as strong as the effect of bramble, yet not significant.
The least influential (nonsignificant) explanatory variables were age (positive) and the proportion of grasses (negative). Several farmland birds of conservation concern seem to benefit from the ecological improvement of the study region with fallows.

| D ISCUSS I ON
The amount of fallows in the study region increased from 1.4% in 1992 up to 8.5% of study site in 2012 (accordingly 11.0% of the utilized agricultural area). In the same period, we measured a large gain in territory numbers of six of 12 bird species. These are spe- which are records for Switzerland and rare across Western Europe (Bauer et al., 2005;Maumary, Vallotton, & Knaus, 2007). The six species with a positive trend all nest in the herbaceous layer or in shrubs, except the ground breeding Stonechat. In line with former studies showing a positive effect of fallows on birds (Burgess et al., 2015;Henderson et al., 2012;Kuiper, 2015a,b;Meichtry-Stier et al., 2014;Tryjanowski, 1999), we suppose that their populations benefit from the structurally diverse fallows in the study area which provide dense herbaceous vegetation, bramble, and shrubs. In contrast, territory numbers of Corn Bunting, Whinchat, Western Yellow Wagtail, Ortolan, Common Quail, and Lapwing have not significantly increased, a result which follows the general trends in other countries (PECBMS 2016, Sattler et al., 2016. These species were never numerous in the region and Corn Bunting, Whinchat, and Ortolan decrease even in the nearest optimal habitats (own observations). The other species (Western Yellow Wagtail, Common Quail, and Lapwing) may benefit less from fallows as in our region they have other habitat requirements for breeding such as damp soils, extensively used meadows, or crops with sparse and low vegetation (Maumary et al., 2007). A similar result was found for priority farmland birds in the UK, where mainly species nesting in field boundaries (hedgerows and field margins) benefited from the Higher Level Stewardship management .   (Maumary et al., 2007). Fallows seem to meet their demands. They hold a broad diversity of plant species, many of them flowering and thus providing food for birds by attracting insects (Frank & Reichhart, 2004 and references therein).
Thanks to the diverse growth of grass, forbs, and single shrubs, the high structural richness provides shelter and protection against predators for birds and their nests. Also the Skylark, an open field breeder, was associated with fallows, together with arable fields and, surprisingly, hedgerows. This species seems to benefit from fallows, probably using them for foraging while the nest was built in a nearby arable field. Skylarks are known to avoid high structures, but they did not seem to be disturbed by the hedgerows in our study site, most probably because most hedgerows were clearly lower than 3 m. The only species showing no clear association with any habitat category was the Corn Bunting, maybe due to the small sample size.
Yet, in 2012 12 of the 22 territory centers lay in fallows and a previous analysis in the region had shown a significant preference of Corn Bunting for fallows between 1992 and 1996 (Jenny et al., 2002).
Our results highlight the important characteristics of fallows for farmland priority bird species and therefore how to optimize this AES option. Size, location (distance to wood), and vegetation composition (amount of bramble and goldenrod, occurrence of shrubs) had the strongest influence on territory density. The effect of age and the amount of grass were less pronounced.
We found a negative effect of fallow size on territory density per fallow. The same relation was found by Zollinger et al. (2013) in western Switzerland. Strips can hold more territories per ha than larger (rectangular) areas because the birds also use habitats around the strip as foraging habitat and only a part of the territory (and the territory center) lies within the fallow strip. Yet, in strips or small areas, the risk of predation may be higher than in large areas due to the increased proportion of edge area (Donald, Evans, Pain, Muirhead, & Buckingham, 1998;Suvorov, Svobodova, & Albrecht, 2014). Predators prefer to roam along edges and linear structures like field margins (Fernex, Nagel, & Weber, 2011). To avoid such negative predation effects, strips of more than 10 m width were promoted (only 10 of 82 fallows were narrower than 10 m). Further, because territory density is higher in smaller fallows, high abundances of farmland birds could be reached with strips rather than with parcels of fallows (which is usually the case for rotational fallows in Switzerland), thereby also reducing the conflict with food production. Interactions between edge effects, predation, and management costs are complex and need further study.
The nearer a fallow was established to a wood, the lower was its territory density of studied bird species. Fallows placed 200 m from the nearest wood had six times more territories than areas within 10 m of a wood (Figure 4b). One possible reason for this result is the fact that many locations next to woods are shady and wet.
There, the benefit of fallows is impaired due to a reduced botanical F I G U R E 3 Proportion of habitat in the study site (available) and in the pseudoterritories of the seven farmland bird species analyzed with a compositional analysis for habitat association diversity as many forbs need sunny locations. A lower botanical diversity has been related to fewer invertebrates (Baines, Hambler, Johnson, Macdonald, & Smith, 1998;Frank, 2000;Pfiffner & Luka, 2000) and so may impair the food resources for most bird species during the breeding season (Benton et al., 2002). Furthermore, some species may avoid fallows in the proximity to wood to minimize the risk of predation from birds nesting in trees (Michel, Jiménez-Franco, Naef-Daenzer, & Grüebler, 2016  increase structural diversity and habitat heterogeneity of fallows, a key factor driving biodiversity (Benton, Vickery, & Wilson, 2003).
In practice, tall-growing forbs should be included in the seed mixes for fallows, as is partially performed in Switzerland. Thorough information, advisory services, appreciation, and financial support are key factors to motivate farmers to change their fallow management and to select sites -for fallows -that optimally benefit biodiversity (Chevillat et al., 2017;Field, Hill, Carroll, & Morris, 2015;Perkins, Maggs, Watson, & Wilson, 2011).
As mentioned in the introduction, fallows are beneficial for biodiversity in general. We expect structural and floristic richness by brambles and shrubs to positively affect a variety of species (i.e., Orthoptera species, Detzel, 1998). A similar result was found for hedges (Graham, Gaulton, Gerard, & Staley, 2018). The occurrence of goldenrod may have negative impacts specifically on invertebrates because they are not adapted to this neophyte. Optimum values of specific measures (i.e., distance to wood, amount of brambles, or bushes) also vary between different species and taxa.
Therefore, a diversity of habitats should be promoted at the local scale and management prescriptions should aim at structural richness within and between fallows (Graham et al., 2018;Tscharntke et al., 2011).

| CON CLUS ION
Our study reveals that perennial fallows in Switzerland are a suitable option in arable landscapes to benefit different farmland bird species. Yet, location, size, and vegetation composition are important factors. The benefit of fallows can be enhanced through wise site selection and modified management, such that these areas are not established next to woods and are kept structurally diverse, featuring small proportions of brambles, shrubs, and tall-growing forbs. This can be achieved through a seed-mix adapted to the region where fallows are planned to be established. Further, cutting is not recommended or only partially and selectively in order to avoid overgrowing by bramble, shrubs, and invasive plants like goldenrod and to keep shrubs low. Advisory services (ideally by local consultants) are necessary and should be available for all farmers within an agri-environment scheme. Our results are relevant for arable farmland in simple landscapes (less than 20% semi-natural habitats) in Switzerland and may also in wider continental Europe. The abovementioned management measures provide habitat structures which have become very rare in many European countrysides, and which benefit a variety of farmland bird species.

ACK N OWLED G M ENTS
We thank Niklaus Zbinden and Otto Holzgang for project management at the beginning of our study. A "thank you" goes to Pius Korner for statistical advice and to Roman Graf, Lukas Jenni, Markus Jenny, and Judith Zellweger-Fischer for their comments on earlier drafts of the manuscript. We thank everyone who helped to collect data in the field and all involved farmers.

CO N FLI C T O F I NTE R E S T
None declared.

DATA ACCE SS I B I LIT Y
Data on territories numbers and habitat characteristics will be archived on Dryad upon manuscript acceptance.