Black Stork Back: Species distribution model predictions of potential habitats for Black Stork Ciconia nigra in Sweden

14 Citation: Thulin C-G, Sörhammar M & Bohlin J. 2022. Black Stork Back: Species distribution model predictions of potential habitats for Black Stork Ciconia nigra in Sweden. Ornis Svecica 32: 14–25. https://doi.org/10.34080/os.v32.22081. Copyright: © 2022 the author(s). This is an open access article distributed under the CC BY 4.0 license, which allows unrestricted use and redistribution, provided that the original author(s) and source are credited. R E S E A RCH PA PE R

began to decline rapidly and in the 20 th century the species was lost as a regular breeding bird in Sweden (Ulfstrand 1973).
Loss of habitats and lack of suitable nesting sites adjacent to wetlands and small streams are likely reasons for Black Stork disappearance in Sweden (Artfakta 2020a). Additional reasons may be the use of pesticides (Luthin 1987, Jiguet & Villarubias 2004 and possible threats during the migration, such as illegal hunting and powerline and wind turbine accidents (Tucker & Heath 1994, Smeraldo et al. 2020. Black Storks have been observed sporadically in Sweden since the last documented nesting in 1953 (Svensson et al. 1999). In the 1990s, ornithologists and scientists suggested that it was only a matter of time before Black Stork established again in Sweden (Davner 1993), but establishment has so far failed to appear, even though Black Storks are observed annually.
A reintroduction programme requires a detailed analysis of the ecology and environmental requirements of the target species, i.e., its social behaviour, size of home range and foraging behaviour (Armstrong & Seddon 2008), but also of habitat availability at intended reintroduction sites. Species distribution models based on biological characteristics of a species and previously described habitat preferences have been used to describe the geographical distribution of species and can be a promising tool to guide the planning of species reintroductions by conservation biologists (Meggs et al. 2004, Powell et al. 2005, Poirazidis et al. 2006, Lord et al. 2020. The expected geographical distribution of a species can be predicted by defining a number of features such as vegetation, soil, or climate, which determine the environmental preferences (Powell et al. 2005). The increasing availability of digitised maps and tools in different geographic information systems (GIS) has helped to improve territory analysis and characterisation of habitats (Thatcher et al. 2006). The creation of habitat models has contributed to the development of conservation biology for species in danger of extinction at several spatial levels (Powell et al. 2005). Wintle et al. (2005) claim that, if a habitat model is applied properly, it could be a good and repeatable technique to use in identification of biodiversity values.
In order to facilitate a future reintroduction program for Black Stork in Sweden, we used species distribution models to identify areas suitable for Black Stork breed-

Introduction
The Black Stork Ciconia nigra is a typical forest bird, inhabiting old, sparse forests with limited disturbance frequency (Svensson et al. 1999, Banás et al. 2019) and a high density of watercourses and stagnant water (Augutis & Sinkevičius 2005). Unlike the White Stork C. ciconia, which forages in open habitats, the Black Stork forages mostly in closed, isolated forests ( Jiguet & Villarubias 2004). The distance between the Black Stork nesting site and the foraging area can vary from six to 40 kilometres (Tucker & Heath 1994, Chevallier et al. 2010a, Strazds 2011. Choice of the nesting tree depends on individual ability to build a nest, flyway accessibility to the tree and safety from avian and terrestrial predators, in particular White-tailed Eagle Haliaeetus albicilla and pine marten Martes martes (Strazds 2011). A Black Stork pair tends to return to the same nest tree if the breeding site is beneficial and the nest can ultimately weigh as much as 1,000 kilograms (Strazds 2003). Tree age is of less importance in the choice of nesting site (Lõhmus 2006), but a suitable nest tree must be relatively large in order to support the weight (Lõmus & Sellis 2003, Treinys et al. 2008. The Black Stork prefers to nest in forests with a high proportion of broadleaved trees (~10-20 %), or with a high proportion of aspen Populus tremula (~10-20 %) if the proportion of other broadleaved trees is low (Treinys et al. 2009).
The global Black Stork population has dropped since the mid-1800s, especially in central and western parts of Europe, most likely due to intensified forestry and habitat degradation (Tucker & Heath 1994). However, this trend has since been reversed in many western European countries and the population is currently considered to be stable in a large proportion of the distribution range ( Jiguet et al. 2011). Recent recolonisations have been documented in Denmark and Belgium (Pihl et al. 2003, Tamás 2011. Regionally, however, the trend is still negative. A dramatic population decline persists in Estonia, Latvia and Lithuania (Treinys et al. 2008). In Sweden, the oldest remnants of Black Stork are found at an excavation near Ystad in southernmost Sweden, thought to stem from a nesting female 5,000 years before present (Davner 1993). During the mid-19 th century, the species was found from southern to central Sweden (Svensson et al. 1999, Lindell 2002. However, the distribution and number of Black Storks then number of aggregated cells and thus the accuracy of estimation may be improved (Naesset 2002). Data on the maximum proportion of Norway spruce Picea abies and the minimum proportions of broadleaved trees, aspen and Scots pine Pinus sylvestris were obtained from Treinys et al. (2009), see Table 2. According to Strazds (2011), the minimum diameter of a nesting tree is 28.0 centimetres and the average tree diameter in a stand of nesting Black Storks is 29.3 centimetres. Therefore, we decided to use 29 centimetres as the lower limit for tree diameter in the GIS analysis (Table 2). A distance of 280 metres between a potential nesting site for Black Stork and infrastructure elements is recommended by Treinys et al. (2009). The value used in this study was rounded to 300 metres, to give a larger margin to disturbance objects. The geographical area used in the analysis was limited to the historically known distribution range of Black Stork in the southern part of Sweden, with Dalarna and Gävleborg Counties as northern borders (Figure 1).

MATERIAL
The input data to the mo dels and their spatial resolution are described in Table 3. Stem volume of the different tree species was extracted from the "kNN-Sweden" forest map (Granqvist Pahlén et al. 2004), which is derived from satellite images and field data from the Swedish National Forest Inventory and "k Nearest Neighbour" as described by Franco-Lopez et al. (2001). The information in the kNN-Sweden map is uncertain if the areas analysed are too small. The standard error of the total stem ing in southern and central Sweden, and integrated the model results with practical considerations of reintroduction of Black Stork.

HABITAT ATTRIBUTES
The estimated habitat area for a breeding Black Stork pair is reported to range from 5,000 to 15,000 hectares (Tucker & Heath 1994, Jiguet & Villarubias 2004, Artfakta 2020a. However, in a study in Lithuania the area of 2,500 hectares surrounding each of 81 nests was described in detail, revealing that on average the habitat comprised at least 13 % forest cover, at least 10 km of watercourses longer than 10 km, and less than 5.5 % disturbance objects (Treinys et al. 2008). Thus, in this study, we used the detailed criteria identified by Treinys et al. (2008) and a habitat area of 2,500 hectares around each nest as minimum environmental requirements (Table 1). Watercourses in the analysis were defined as running water including everything from a small brook to a large river (SVAR 2011). Smaller water bodies of stagnant water, such as flooded ditches, were not included in the analysis, due to lack of data. Furthermore, at least 125 hectares (5 %) within 2,500 hectares of suitable habitat for Black Stork had to meet the requirements for suitable nesting sites (Treinys et al. 2009).
In order to model suitable nesting sites, all variables had to be met within an area of one hectare (    crona 2003). The counties used in the GIS analysis were selected from the county map of Sweden and extracted to a new polygon layer. All input data were converted to the size of this polygon layer of counties.

METHOD
Habitat modelling was carried out in ArcGis 10 (Environmental Systems Research Institute, Redlands, California, USA), using a moving window approach with a window size of one hectare. A raster cell was considered a nesting site if, on average, the D BW value was larger than 29 cm, the proportion of Norway spruce was less than 10 %, the combined proportion of European beech Fagus sylvatica, European oak Quercus robur, Scots pine or other deciduous species was at least 30 % and the Euclidian distance to infrastructure elements was more than 300 metres. For modelling the full habitat requirements, i.e., both foraging and nesting, a similar moving window approach was used. A raster cell was considered a possible habitat if, within a 2,500-hectare window, less than 5.5 % of the area was disturbance objects, there was more than 10 km of total waterways, more than 13 % of the area was forest cover, and at least 5 % of the area was suitable nesting areas. Finally, the sum and the proportion of suitable habitat cells were calculated for each county.

Results and discussion
The overall modelling process involved several steps, such as conversion of data, calculations, and merging of the data layers. Based on the selected variables, suitable habitats were found in every county included in the analysis except Gotland (Table 4). Seven counties contained more than 18 % suitable habitat within their total area, while the remaining counties contained less than 10 % suitable habitat (Table 4). The highest proportion of suitable habitat in relation to total area was found in Jönköping County (25.8 %), followed by Blekinge (23.9 %), Västra Götaland (  (20.7 %) Counties, while the lowest proportion (1.8 %) was in Dalarna County. No habitat (0 %) was identified on the island of Gotland (Gotland County) in the Baltic Sea (Figure 1). Altogether, the 17 counties included in the analysis contained 21,705 km2 of suitable Black Stork habitats (Table 4).

GIS ANALYSIS
The results of the GIS analysis indicated extensive availability of suitable breeding habitats for reintroduction of Black Stork in Sweden. In particular, suitable areas for reintroduction were identified in the vicinity of the lakes Vänern, Vättern, and Hjälmaren, and towards the southeast in the counties of Blekinge and Kronoberg (see Figure 1). This is a reasonable finding because of the proximity to water and watersheds and the presence of a large number of restored wetlands, ponds, and dams in these areas. They also provide a mixture of arable land and forests, particularly forest with a relatively larger proportion of deciduous trees. The absence of suitable habitat patches on Gotland and in northern Dalarna seems plausible, since large parts of Dalarna are composed of near-alpine forests and have a harsh climate, unsuitable for the Black Stork. The former distribution limit of the species was near the Dalälven river (Lindell 2002). There has been speculation about whether the Baltic Sea acts as a barrier for Black Stork migration (Davner 1993), meaning that Gotland with its relatively remote location from the mainland may not be appropriate as a nesting site. However, observations of Black Stork have been   (Sackl 1985) to 10.8 per 100 km2 in the Dadia-Lefkimi-Soufli Forest National Park in north-eastern Greece (Alexandrou et al. 2016). Thus, although it is not certain that there is an absolute need for contiguous habitat of at least 2,500 hectares for breeding Black Storks, as postulated in our models, it seems reasonable based on Sackl 1985 andAlexandrou et al. 2016 (i.e., 0.33-2.7 breeding pairs / 2,500 hectares). The size of the habitat likely varies depending on the quality, assuming that the higher the quality, the smaller the area required. We suggest that habitat patches of 2,500 hectares are suitable if they are sufficiently undisturbed and contain enough large trees for nesting, since a number of studies indicate that Black Storks can fly several kilometres to forage and thus remoteness is not necessarily a limiting factor (cf. Strazds 2011).

REINTRODUCTION SUITABILITY
A core issue for a successful reintroduction programme is presence of key prerequisites in the intended reintroduction area, which requires extensive knowledge of the biology of the species in question. The reason behind the recent population declines in Estonia, Latvia and Lithuania (Treinys et al. 2008) is unclear (Zieliński 2006), but it could be due to intensified forestry and habitat degradation (Tucker & Heath 1994), predator avoidance (e.g. Treinys et al. 2016) and shortage of potential mates in margin populations (Konovalov et al. 2019). Rosenvald & Lõhmus (2003) point out that forestry activities are not the only factor causing population decline, but are likely to be strongly linked to a decline. Forest logging escalated in the Baltic countries after 1991 . This resulted in destruction of nesting habitats and contributed to impaired breeding opportunities (Lõhmus et al. 2005).
More than 50 % of the European population is currently distributed in Eastern Europe (Chevallier et al. 2010b), with the highest population density in the Balkan countries, northern Ukraine, Germany, north-eastern Poland, and western Russia (Lõhmus et al. 2005). The current population increase seen in Western Europe could be the result of a migrating population of Black Storks from Eastern Europe searching for new nesting sites (Chevallier et al. 2010c). On the other hand, Treinys et al. (2008) argue that an on going ecological change is occurring in the western and central European populations, allowing Black Storks to establish in fragmented forest areas in agricultural landscapes.
As suggested by our GIS modelling, there are several areas in southern Sweden that may be suitable for Black Stork reintroduction. There are also indications that the Black Stork is favoured by the presence of the European beaver Castor f iber (Tucker & Heath 1994, Svensson et al. 1999. Through the construction of dams and lodges, the beaver frequently causes accumulation of new water bodies, which the stork can utilise in its scavenging for food (Svensson et al. 1999). In Latvia, a positive correlation between the two species has been found, with a high density of Black Stork in areas where the beaver is most frequent (Strazds 2011). The beaver, once extinct in Sweden, was reintroduced in 1922 and the population now exceeds 100,000 individuals, with an increasing trend (Hartman 1994, Hartman 2011). Beaver observations have been reported for many areas in southern and central Sweden (Figure 2). The potential interdependence of beaver and Black Stork (e.g. Tucker & Heath 1994, Svensson et al. 1999, and the current distribution of beaver in Sweden, suggest that it may be favourable to release Black Storks in areas where the two species can coexist, in particular where beaver observations coincide with suitable Black Stork habitats revealed from our study.
Public attitudes can determine whether a conservation effort succeeds or fails (Bremner & Park 2007), particularly if the effort is perceived by the public as controversial, as species reintroductions may be ( Jacobson & Duff 1998). However, public acceptance of captive breeding with subsequent reintroductions has increased and, consequently, the number of reintroduction projects has also increased (Seddon et al. 2007). An example is the reintroduction of the White Stork in Sweden (www.storkprojektet.se), for which the public attitude to capture and release of individuals is almost exclusively positive (Emma Ådahl, pers. comm. 2021). A similar positive attitude could be expected to reintroduction of the Black Stork, as the two species disappeared from Sweden at about the same time (Olsson & Rogers 2009, Svensson et al. 1999. Reintroduction can be implemented using captivity-bred or wild-caught animals (Meltofte 1987, Sarrazin & Barbault 1996. The potential for successful reintroduction is lower when the animals are bred in captivity compared with when they are caught in the wild and transported to new habitats (Griffith et al. 1989). However, the conditions for successful reintroduction of captive-bred animals improve if the animals are well managed, have sufficient amount of genetic variation a broad genetic material, and are prepared for life in the wild through self-contained behaviour in the enclosures (Kleiman 1989). For successful reintroduction in Sweden, several pairs of Black Storks are needed to reduce the risk of inbreeding and increase the gene pool ( Jamieson 2011). In the case of a species which is experiencing a decrease in numbers in several countries, it may be sensible to use specimens from breeding facilities and avoid wild-caught birds. There may also be a risk of wild-caught birds returning to their original location (Oppel & Beaven 2002).
At a White Stork breeding facility, the birds must be ringed and provided with food of good quality, and the enclosures must be cleaned and in good condition. Facility employees must be trained to use techniques to prepare the animals for life in the wild, where they must be able to search for food, avoid predators, and construct nests (Kleiman 1989). The environment in enclosures can be limited and the animals may develop stress and behavioural problems over longer (Young 2003) or shorter periods (Coddington & Cree 1995). Employees in the Swedish White Stork project have not noticed any behavioural change in the birds that has resulted in reduced vitality in the wild (E. Ådahl, pers. comm.). However, the behaviour of the Black Stork is different from that of the White Stork. For instance, when there are numerous adult Black Storks in captivity, they can behave belligerently towards each other FIGURE 2. Locations of recent European beaver Castor fiber observations in Sweden (Artfakta 2020b).
-Observationer av bäver Castor fiber i Sverige (Artfakta 2020b). (Bračko & King 2014). Two Black Stork pairs cannot live in the same enclosure at the same time without a risk of harming each other (Staffan Åkeby, pers. comm.). Thus, a system with geographically separated breeding facilities needs to be developed for Black Storks.
The climate has an effect on the survival of species and certain weather conditions or temperatures may pose obstacles for a species (Olsson 2007). The mortality rate in Black Stork chicks increases with bad weather conditions (Treinys et al. 2007), which has been suggested as an underlying cause of Black Stork disappearance in Sweden. However, over time the weather in Sweden has constantly fluctuated from warmer to colder and from drier to wetter, and vice versa, and Black Storks have been breeding in Sweden since at least 3,000 BC (Davner 1993). Moreover, the Black Stork does not seem very sensitive to climate conditions over its wide distribution from Western Europe to East Asia (Hancock et al. 1992). However, the Black Stork is sensitive to habitat requirements, especially for the choice of nesting site. If a reintroduction programme is launched, Swedish forest management must take into account the habitat requirements of Black Stork during logging and leave groups of thick, old trees of mainly European oak, European beech, aspen and Scots pine. It is also important to avoid drainage of forests, to provide wetlands for Black Stork to forage in, and potentially to regulate beaver hunting to facilitate wetland establishment. Old forests with high humidity may also be of significance for other endangered species (Berg et al. 1995) and thus restoration efforts that aim to create suitable habitats for Black Stork may also benefit other forest-dwelling species.
An additional challenge for Black Stork reintroduction is the migratory behaviour, as the European population of this species spends roughly half the year in its wintering grounds in west or east Africa, heading northwards in April for the breeding season and returning to Africa at the end of August (Lindell 2002), bypassing the eastern or western parts of Europe (Bobek et al. 2008). This may be a delicate matter to overcome, but experiences from the Swedish White Stork project and other reintroduction programmes with migratory species show that this challenge is not insurmountable. The Baltic Sea may serve as a migration barrier that has added to the disappearance of Black Stork from Sweden (Davner 1993 Hancock et al. 1992, p. 71). Thus, a reintroduction programme in Sweden could be reinforced with spontaneous immigration of Black Storks from overseas that intermix with released birds.

Conclusions
We believe that it is worthwhile to launch an effort to restore a breeding Black Stork population in Sweden. Using species distribution modelling, we showed that the necessary habitat requirements are fulfilled in part of southern and central Sweden. A warmer climate, along with the ongoing spread of beavers that spontaneously create wetlands in forested areas, would further facilitate restoration efforts. We also believe that there would be a positive public engagement in Black Stork reintroduction, since this is a charismatic species with a specific mystique that does not impinge on other interests in areal land use. Challenges in reintroduction work include the establishment of breeding infrastructure, finding stock animals, enabling released birds to migrate and, of course, acquiring funding for the work. However, these challenges should not hinder efforts to re-establish the Black Stork in Sweden.