Several populations of cranes, geese, and swans are thriving and increasing in modern agricultural landscapes. Abundant populations are causing conservation conflicts, as they may affect agricultural production and biodiversity negatively. 

Management strategies involving provisioning of attractive diversionary fields where birds are tolerated can be used to reduce negative impact to growing crops. To improve such strategies, knowledge of how the birds interact with the landscape and respond to current management interventions is key.

We used GPS locations from tagged common cranes (Grus grus) and greylag geese (Anser anser) to assess how they use and select differentially managed habitats, such as diversionary fields to decrease impact on agriculture and wetlands protected for biodiversity conservation.

Our findings show a high probability of presence of common cranes and greylag geese in the protected area and in the diversionary field, but also on arable fields, potentially causing negative impact on agricultural production and wetland biodiversity.

We outline recommendations for how to improve the practice of diversionary fields and complementary management to reduce risk of negative impact of large grazing birds in landscapes tailored for both conservation and conventional agriculture.

Study area

The study was conducted in the vicinity of lake Kvismaren (59°10′N,15°23′E), southeast of Örebro in south-central Sweden (Fig.1) from 2017 to 2022. The landscape is flat and dominated by highly productive arable land. The central parts of the study area consist of a mosaic of shallow, eutrophic lakes, reed beds and grazed wet meadows, designated as both a Natura 2000 (i.e., SPA and SAC) and a Ramsar site (Wetlands: a global disappearing act 1970; EC 2016). Kvismaren is a national key area for protection of multi-functional wetland habitats and aimed for staging, breeding, and threatened bird species, e.g., waders, waterfowl, gulls. It is also intended to serve as a refuge area for large grazing birds (common cranes and several goose species; EC 2009). The shallow lakes and wetlands provide good roost sites and the surrounding agricultural landscape favourable foraging opportunities. A diversionary field and various scaring practices are employed in the study area to reduce the risk of crop damage caused by large grazing birds. 

Capturing and tagging birds

This study was based on location data from 17 common cranes and 55 greylag geese equipped with GPS transmitters. We captured pre-fledgling common cranes after a short run from a car or a hide and equipped them with leg or harness mounted GPS transmitters (Ornitela) in July- early August in 2016-2020 within 70 km of Grimsö Wildlife Research Station, Lindesberg (59°43′N,15°28′E; for details about capture procedures, see Månsson, Nilsson, & Hake, 2013). Juvenile common cranes accompany their parents  to the wintering grounds, where parents and juveniles usually part in January (Alonso, Veiga, & Alonso, 1984). They most often visit the study area (Kvimaren the second time as yearlings but sometimes postpone the second visit until they are older. Tagged cranes often return for several seasons (for details see Table S2in Supplemental Information). Greylag geese were captured as adults and equipped with neckband GPS transmitters (OT-N35, OT-N44) in June in 2017-2019. Captures took place in Kvismaren in early mornings by herding moulting, flightless greylag geese by canoes and by foot into net corrals (for capture details see Månsson et al., 2022). The GPS-tagged greylag geese present in the study area were included in the study, often for several consecutive seasons.

Data management

To model relative presence of common cranes and greylag geese in differentially managed habitats during the season of the supplemental feeding in the diversionary field (May 1 to August 15 each year), we compared the GPS locations of the birds (hereafter referred to as ‘used’ locations) with randomly distributed locations (‘available’) (i.e., Resource Selection Functions, RSF´s; Lele & Keim, 2006a).For the studyperiod, we assume that crops are growing and coincide with associated damage risk when the diversionary field was in operation. The analyses were conducted in two steps for both species by assessing; 1) probability of presence in different habitats (e.g., arable land, protected area; model 1, and 2) probability of presence on different crop types in arable fields (e.g., barley, ley, diversionary field; model 2). We defined an uptake area for the diversionary field (i.e., assumed to be within reach for the birds during the daily foraging activities) by using distance from the previous night’s roost location, within our study area, to all daytime locations the following day. The 95% percentile of all these measured distances were then used to define a radius for the uptake area around the diversionary field (5.80 km for common cranes and 3.82 km for greylag geese). W e included the GPS locations and randomly distributed available locations (ratio 1:1) within the uptake area, mirroring the available area for the birds’ daily foraging bouts. The time until the common cranes and greylag geese started using the diversionary field was assessed as either: 1) the number of days until first use after the first supplemental feeding, or if the feeding had started prior to arrival, or 2) as the number of days after arriving to the uptake area.

To ensure highly precise spatial data, we excluded locations with horizontal dilution of precision >7 (D’eon & Delparte 2005) as well as locations fixed by less than three satellites (i.e., 2D). As the study focuses on habitat and field selection on the ground and during daytime foraging activities (i.e., damage risk), locations assumed to be when in flight with speed >10km/h were excluded, as were night locations. Day and night locations were defined by the time of sunset and sunrise, i.e., all locations between sunset and sunrise were defined as night locations and subsequently excluded. The programming of GPS transmitters and the frequency of location fixes varied between 15 and 60 minutes. Thus, to standardize the positioning frequency and to meet the assumption of independence between consecutive individual locations to allow for potential individual movement to another habitat or field, the positioning frequency was standardized to one location/hr (Fieberg et al. 2010). Habitat types were derived from the national land cover data base (0.1 × 0.1 km; Naturvårdsverket, 2020) and the extent of the protected area (Ramsar site; Kvismaren) from the Swedish Environmental Protection Agency. To avoid over-fitting of models, habitats with similar environmental characteristics and management were combined into the following categories: arable land (i.e., conventionally farmed fields with short rotation crops), diversionary field, non-protected wetland and water, other land (e.g., forest, exploited land), pasture (i.e., permanent grassland for grazing), and protected area ((step 1; Table S1). For the locations on arable land (step 2), crop types were derived from the SAM14 database (The Swedish Board of Agriculture), which provides spatially explicit information about cultivated crops at the field level. Crop types were lumped into nine categories based on similar characteristics or sporadic occurrences: barley, beans and peas, diversionary field, ley (i.e., fertilized, productive grasslands for silage production), oat, potatoes, rye and triticale, wheat, and other crop (e.g., rape seed, vegetables; step 2). Data management was done in ArcGIS (version 10.7) and R (version 4.2.1).