Looking for a magic island in the sea of agriculture: factors affecting forest patch attractiveness for wild boar in human-dominated landscape

Landscape fragmentation carries the risk of habitat loss for many animal species. This may include wild boars, which are one of the main culprits of conflicts with humans, while isolated patches of forest (forest islands) scattered in the agricultural landscape may affect its ecology and the pressure on the crops. This study aimed to understand the importance of small forest islands located in a human-dominated agricultural landscape in the ecology of wild boar. It explores what features, both of the interior and exterior of the islands, influence their use. The study area covered approximately 220,000 ha of northern Poland, where wild boar use of 200 islands distributed across the agricultural landscape was investigated. For this purpose, the method of counting pellet groups was applied on transects of a total length of 51.6 km. In order to compare the use of islands with compact forest complexes, 84 km of transects were made in the latter, divided into developmental stages and forest type. Among the features of the interior of the islands, increasing cover conditions and the main species of trees (pine, spruce and alder) had a positive influence on the wild boar use. The utilization also depended on the presence of human settlements, other islands, and the type of agricultural crop around the island. When comparing island use to compact forest complexes, islands were used more often than forest regeneration sites and mature stands, but less often than pre-thickets. For the 25% most frequently used by wild boar, islands were more attractive than virtually all stand development stages in compact forest complexes. Due to the cover conditions provided, the occurrence of islands in the agricultural space enables wild boars to use areas more distant from compact forest complexes. This may cause the spread of damage caused by wild boars in the fields, but on the other hand it may reduce the pressure on agricultural areas in the immediate vicinity of compact forest complexes. Intensification of hunting pressure in the vicinity of islands with specific characteristics e.g. good cover conditions, considerable distance from human settlements) will increase the effectiveness of wild boar population control in the agricultural landscape and will have a limiting effect on the damage caused by this species in the fields.

of 200 islands distributed across the agricultural landscape was investigated. For this purpose, the method of counting pellet groups was applied on transects of a total length of 51.6 km. In order to compare the use of islands with compact forest complexes, 84 km of transects were made in the latter, divided into developmental stages and forest type. Results Among the features of the interior of the islands, increasing cover conditions and the main species of trees (pine, spruce and alder) had a positive influence on the wild boar use. The utilization also depended on the presence of human settlements, other islands, and the type of agricultural crop around the island. When comparing island use to compact forest complexes, islands were used more often than forest regeneration sites and mature stands, but less often than pre-thickets. For the 25% most frequently used by wild boar, islands were more attractive than virtually all stand development stages in compact forest complexes. Conclusions Due to the cover conditions provided, the occurrence of islands in the agricultural space enables wild boars to use areas more distant from compact forest complexes. This may cause the spread of damage caused by wild boars in the fields, but on the other hand it may reduce the pressure on agricultural areas in the immediate vicinity of compact forest complexes. Intensification of hunting pressure in the vicinity of islands with specific characteristics e.g. good cover conditions, considerable distance from human settlements) will increase the effectiveness of

Introduction
Populations of many ungulate species in Europe, over the last decades, have significantly increased in size (Mysterud et al. 2002), entailing increasing conflicts with human land management (Trdan and Vidrih 2008). The need for efficient management of ungulate populations is one of priority issues in the human dominated landscapes (Hess et al. 2010;Apollonio et al. 2010;Massei et al. 2015). Modern approaches to game management should take into account a number of different locally occurring elements (eg. knowledge of species' food requirements and competition between them, plasticity in migration patterns, effects of hunting other than just population control etc.) .
One of the most conflicting species, especially in agricultural areas is wild boar (Sus scrofa Linnaeus, 1758), (Massei et al. 2015). It is native to Europe (except Iceland, Ireland and Norway), southern Asia (from Turkey to Japan and Malaysia) and the northwest coast of Africa (Oliver et al. 1993;Barrios-Garcia and Ballari 2012). Its extensive range (Castillo-Contreras et al. 2018;Morais et al. 2019) makes the damage it causes in agricultural landscapes a global issue (Massei et al. 2015). Wild boars cause many conflicts in different spheres of human functioning like farming, home plot gardening, traffic accidents ect. In addition to the aforementioned damage to agricultural crops, they cause traffic accidents , are vectors for zoonotic diseases (Gortázar et al. 2007) and have a negative impact on some plant and animal species (Massei et al. 2011). At the same time, they are very popular as a valuable object of hunting interest (Apollonio et al. 2010). In the face of these multi-level interactions, there has been a growing interest in the management of this species (Brivio et al. 2017). As a result of rapidly increasing population sizes over the past decades (Massei et al. 2015; Morelle et al. 2016), it has become "an native invader" in many places (Carey et al. 2012), that is, being invasive within its own range (Simberloff 2011). Wild boars owe their evolutionary success to their ecological plasticity, which is expressed through diverse spatial and temporal strategies in exploiting resources available in the environment (Mayer and Brisbin 2009). They are able to adapt their behavior in response to structural and functional changes in landscape features, such as altitude, proximity to rivers, forest cover, and atmospheric factors (Virgós 2002;Lemel et al. 2003;Brivio et al. 2017). In view of the above characteristics of wild boar, a better understanding of how populations of this species function in a changing landscape is needed. Habitat fragmentation is cited as one of the major threats to species conservation (Andrén 1994). However, the exact effects of habitat fragmentation on the spatial functioning of species with high reproductive potential, adaptability and migratory capacity (such as wild boar) are poorly documented (Virgós 2002).
Previous studies on the functioning of wild boar in fragmented agricultural landscapes have often focused on the damage caused by the species and the contribution of particular types of crop species to its diet (Ficetola et al. 2014;Gentle et al. 2015). Wild boars require a variety of habitats to meet their needs, providing foraging and shelter and creating suitable breeding sites (Rivrud et al. 2010;Fattebert 2017). Boundaries at the forest-field interface are particularly favored by wild boar as foraging areas (Lemel 1999;Thurfjell et al. 2009). Landscape fragmentation and the associated length of field-forest boundaries are cited as factors that have a significant impact on increasing the intensity of damage to agricultural space, often considered severe (Lemel 1999). At the same time, it has been found that habitat fragmentation is not a barrier to the expansion of new areas by wild boar, which carries the risk of further increasing human-wildlife conflicts, especially in a fragmented agricultural landscape (Rutten et al. 2019a).
One element typical of such a landscape is the occurrence of relatively small patches of forest habitat (forest islands) scattered in open space. To date, there are few studies on how wild boars function in a fragmented landscape, with only one on how the ecological characteristics of islands affect their use by wild boar (Virgos 2002). Studies on the use of compact forest complexes are much more numerous (Fonseca 2008;Keuling et al. 2009). When selecting them, especially in winter, wild boars are guided by the availability of food (Fonseca 2008). Particularly important are deciduous forests with beech and oak providing high-energy seed food (Thurfjell et al. 2009;Ballari and Barrios-García 2014;Brook and van Beest 2014). The soil of deciduous forests contains significant amounts of starchy tubers, bulbs and rhizomes, and many others (Welander 2000), which are eaten by wild boars during rooting activity. It also provides wild boars with large amounts of insects and earthworms (Jokimäki et al. 1998;Baubet et al. 2003). On the other hand, coniferous forests create suitable resting conditions by providing shelter from predators and mitigating adverse weather events (Fonseca 2008).
In this study, we attempted to define factors characterizing the interior and exterior of forest islands influencing wild boar use of dissected forest fragments in human-transformed agricultural landscapes. Out of the former, we predict that island size, species composition, proportion of the main species and stand age will influence the use of islands by wild boar. In addition to the aforementioned characteristics, security cover is also one of the primary elements that determine the quality of the animal's habitat. Since, in general, agricultural landscapes are characterized by their absence, we predict that cover conditions will be an important criterion determining the use of forest islands by wild boars.
In addition to the characteristics of the islands themselves, the environment may also influence their use. Some types of agricultural crops are selected by wild boars particularly readily (e.g. grassland, maize,) while others are chosen significantly less frequently (e.g. rye, peas), (Schley et al. 2008), thus we predict that the type of crop surrounding islands will influence wild boar use of forest islands. In addition, we also expect that the extent to which islands are utilized will depend on the presence of other islands and distance from compact complexes. One of the determinants of space use by ungulates is human disturbance, (Selier et al. 2015). Since the intensity of disturbance is closely related to the intensity of human penetration of the environment (Marchand et al. 2014), we also predict that the distance of islands from buildings and roads and the size of settlements (single buildings, whole villages) will also be a factor influencing their use.
As previously mentioned, the functioning of wild boar in compact forest complexes is better understood than is the case for islands (Keuling et al. 2009;Fonseca 2008). Furthermore, there are studies comparing the functioning of wild boar in different environments (continuous forest, urban area, heathland, field crops e.g. Bruinderink 1995;Podgórski et al. 2013). Despite this, it is not known to what extent forest complexes represent a more attractive environment for wild boar than forest islands scattered in agricultural landscapes. As already mentioned, compact forest complexes are a source of food as well as cover for ungulates (Huynh et al. 2005;Fortin et al. 2005;Merli et al. 2017), while at the same time they tend to have lower levels of human disturbance than agricultural areas. However, because environmental fragmentation is not a barrier for this wild boar to colonize new areas (Rutten et al. 2019a), as discussed above, we predict that forest islands will provide wild boar with habitat of comparable attractiveness to compact forest complexes. The described research also contains an important application thread-isolating the characteristics that determine the use of forest islands by wild boars will allow us to identify patches of forest habitat that should be targeted by hunting pressure to effectively reduce crop damage and wild boar abundance in agricultural environments.
In summary, the aim of our study is to verify the following predictions: (i) the characteristics of the interior of islands significantly affect their use by wild boar, (ii) the intensity of use of islands by wild boar is also influenced by the characteristics of their surroundings, (iii) the attractiveness of forest islands in agricultural space is comparable to the conditions offered by compact forest complexes.

Study area
The study area covered north-eastern Poland of about 220 thousand ha, ranging in latitude from 20.28°E to 21.59°E and longitude from 53.98°N to 54.38°N. Agricultural land predominates in the mentioned area, reaching 60% of the surface. The main agricultural crops are rape, wheat, triticale, rye, barley and oats. Permanent meadows also have a significant share, while maize, beet, millet and buckwheat are grown relatively rarely. The area was under the influence of ASF (African Swine Fiber), which to a different degree, depending on the year, influenced the number of wild boars in particular forest districts. The length of the growing season is about 190 days, while precipitation is 700 mm per year with an average air temperature of about 7 °C. The terrain is predominantly undulating, less frequently flat. The height above sea level ranges from 40 to 100 m. Numerous forms of glacial and fluvialglacial erosion are visible. The soils are moderately fertile as a result of a large share of boulder clays and glacial sands and gravels.
There are three State Forest Districts in the study area-Srokowo in the eastern part, Bartoszyce in the central part and Górowo Iławeckie in the western part (Fig. 1). The forest cover of the study area is just over 22%, of which about 53.4 thousand ha are managed by National State Forests, and only 6.74 thousand ha are privately owned. The State Forests are dominated by deciduous forests (about 66.6%), consisting mainly of oak (about 42%) and birch and black alder. Coniferous forests (ca. 33.4%) are dominated by spruce (ca. 54%) and pine (ca. 39%). Most of the forest islands stay in private hands and occur in great dispersion (similarly to state forests Table 1). In the last decade, the number of forests has significantly increased through afforestation of former agricultural land. There are 898 forest complexes, of which 196 (about 95.82% of the total forest area) are compact complexes with an area over 20 ha. The remaining 702 complexes are small forest islands constituting approx. 4.18% of the forest area, 232 of them do not exceed 1 ha.

Forest characteristics and crop type
The area of the islands ranged from 0.1 to 13.2 ha. The islands appeared in the existing forest inventory description as forests and in the description their characteristics were given. For each island, the following parameters were recorded from the forest inventory: area occupied, main species (tree species with the highest proportion), stand age and proportion  of oak trees over 50 years old in the species composition. For each island, the cover conditions inside the islands were also estimated. For this purpose, a board (150 by 60 cm) painted in black and white squares (30 by 30 cm, 10 in total) was used. It was placed in the central part of the island and from a distance of 50 m from four directions (N, W, S, E) the percentage of visibility of the board was estimated. The average value from the four directions was used for analysis. The surroundings of the islands were also inventoried, giving the type of agricultural crops (grain, rape, meadow, bare soil) located within 200 m of the island (Table 5), the number of objects within a radius of 300 and 1000 m, such as: other islands (0.1-20 ha), buildings (single or whole villages), compact forest complexes (more than 20 ha), the distance to the nearest compact forest complex [m] and the presence of roads (only within 300 m).
The study was also conducted within compact forest complexes. The following groups were distinguished taking into account the developmental stage and stand type (coniferous-deciduous) according to the following scheme: forest regeneration sites (0-10 years), pre-thickets (11-20 years), thickets (21-30 years), pole stage stands (31-50 years), mature stand (over 50 years). Using the map, 42 coniferous (pine, spruce, larch) and 42 deciduous (mainly oak) stands were randomly selected for each of the above-mentioned stand development stages scattered throughout the study area.

Space use of wild boar
Using the map, 150 mid-field forest islands were randomly selected regardless of their ownership. These were 50 islands in each of the following districts: Bartoszyce (2019), Srokowo (2020) and Górowo (2021) and again Bartoszyce (2021). Originally, the research was to be done in two Forest Districts (Bartoszyce and Srokowo) in the following years. However, due to the significant reduction of the wild boar population caused by African Swine Feaver in Srokowo Forest District, Srokowo was replaced with Górowo Iławeckie District. As a result, the research was repeated on the same 50 islands only in Bartoszyce Forest District. The pellet group method (Neff 1968), consists in pellet groups counting in 2 m wide strip transects, was used to estimate the use of individual habitats by wild boar. This is an indirect method that is mainly used for estimating cervid abundance (Härkönen and Heikkilä 1999;deCalesta 2013), but has been gaining popularity in recent years and has been successfully applied to wild boar (Plhal et al. 2014). The advantage of this method is that the relative intensity of animal use of different habitat types can be determined simultaneously (Skarin 2007).
Fecal groups were counted in early spring, just after the snow melted (March-April), but before the beginning of vegetation development in 2019, 2020, and 2021. On each of the 200 islands, groups of wild boar droppings were counted on transects 100 m long and 2 m wide. The number of transects depended on island size according to the following scheme: for islands up to 1 ha-1 transect; 1.01-2.00 ha-2 transects; 2.01-3.00 ha-3 transects; 3.01-4.00 ha-4 transects and for islands over 4.00 ha-5 transects. Two 100 m transects, 2 m wide, were conducted on each of the selected crops, where wild boar droppings groups were counted. In the compact forest stands wild boar droppings were counted separately for stand developmental phases and forest types (deciduous and coniferous). In Srokowo Forest District and in Bartoszyce Forest District in 2019 pellet groups were counted in randomly selected sampling plots-22 (11 for each forest type) for each stand phase. In Górowo and Bartoszyce (2021) Forest Districts there were 20 plots for each stand phase. In every plot, wild boar droppings were counted along 2 transects (100 m long and 2 m wide).

Statistical analyses
Statistical analyses were based on the generalized Poisson log-linear model or analysis of covariance (ANCOVA) for this model. Only in assessing the significance of differences between two independent groups was the non-parametric Mann-Whitney U test used. This was the case for comparisons of the number of groups of wild boar droppings on islands versus forest at different stand development stages and the significance of the difference between coniferous and deciduous stands. All of the analyses were performed using STATISTICA 13.3 software at significant level α = 0.05.

Results
The influence of interior and exterior characteristics of islands on their use by wild boar In total, data was collected along transects of 51.6 km in forest islands and 84 km in compact forest complexes. A total of 623 groups of wild boar droppings were recorded in the islands (Bartoszyce 2019-326; Srokowo 2020-138; Bartoszyce 2021-78; Górowo Iławeckie 2021-81), while in forest complexes 1284 (Bartoszyce 2019-631; Srokowo 2020-216; Bartoszyce 2021-202; Górowo Iławeckie 2021-235). The average density of wild boar droppings (per 100 m of transect) was 1.17 and 1.53 on islands and in compact forest complexes, respectively (the difference between habitats was not significant-p > 0.05).
Out of the characteristics of island interiors examined, only two of them influenced the use of islands by wild boar: cover conditions (p < 0.001) and the main tree species on the island (p = 0.0025) ( Table 2). As the cover conditions (increase in the degree of cover of the board) improved, the use of islands by wild boar clearly increased (p < 0.001) (Table 2; Fig. 2). In terms of the main tree species, the most frequently used islands were those dominated by pine (Pinus sylvestris), spruce (Picea abies) and alder (Alnus glutinosa), less intensively used were those with aspen (Populus tremula) and oak (Quercus robur), and least frequently used were patches of forest dominated by linden (Tilia cordata) and birch (Betula pendula) ( Table 3). Other characteristics of forest islands such as the proportion of the main species, its age, the presence of oak aged over 50 years, and the area and width of the island did not significantly affect their use (in all cases p > 0.05). The type of agricultural crop surrounding the islands had a positive effect on the use of the islands by wild boar (p < 0.001) ( Table 4). Those surrounded by cereals and bare soil were the most frequently used, rape slightly less frequently and meadow the least frequently (Table 5). The distance separating islands from the nearest compact forest complex did not affect the use of islands by wild boar (p > 0.05). Among the objects occurring within 300 m from the island, the buildings had an influence on their use (p < 0.001), and the most  frequented by wild boars were islands in the vicinity of which there were only single buildings or none at all (Table 6). The least frequently used were those located near villages. The presence of other islands within a radius of 300 m was the second most important element (p < 0.007), and their influence was negative (Table 7). The presence of compact complexes and roads within 300 m did not affect the use of islands by wild boars (both p > 0.05). Within a radius of 1000 m from the island, there was also a significant effect of buildings (p < 0.003), but the most frequently used islands were those in the vicinity of which there were no buildings. On the other hand, as the number of buildings/villages increased, island use decreased (Table 8). In addition, the negative influence of other islands (p = 0.061) and the number of compact forest complexes (p = 0.069), whose influence on the use was positive, were at the limit of significance (Table 9).

Use of compact forest complexes by wild boars
The extent to which wild boar utilized compact forest complexes was influenced by stand developmental stage (p < 0.001) and forest type (p < 0.002), (Tables 10 and 11). By far the most common stand developmental stage used by wild boar was prethickets ( Fig. 3), followed, at similar levels, by thickets and pole stage stands. Mature stands were slightly less attractive, and forest regeneration sites were the least frequently used developmental phase. Conifer stands were significantly more frequently used for pre-thickets (p < 0.01) and forest regeneration sites (p < 0.05), while in the other developmental stages the differences were insignificant (p > 0.05). As a result, the interaction developmental stage x stand type was also significant (p < 0.001). Comparing the attractiveness of islands with compact forest complexes depended largely on developmental stage (value averaged across both forest types) (Table 11). Islands were more frequently utilized than forest regeneration sites (p < 0.001) and mature stands (p < 0.03) and at a similar level as thickets and pole stage stands (both p > 0.05). Island utilization was simultaneously lower than pre-thickets (p < 0.001). When comparisons were restricted to the 25% of islands most frequently used by wild boar (Table 11), we found that they were more attractive than almost all stand developmental stages (p < 0.001 in all cases). The only exceptions were pre-thickets-admittedly, even in their case islands were more attractive, but the difference was at the limit of significance (p = 0.07).

Forest islands
Of the six island interior variables analyzed, only two appeared to influence their use by wild boar-cover conditions and major tree species. The former is not surprising, as for ungulates, cover has a fundamental influence on a number of elements of their ecology (Mysterud and Ostbye 1999), such as habitat use (Dechen Quinn et al. 2013), foraging (Borkowski and Ukalski 2012), and size of individual home ranges (Borkowski et al. 2016). The importance of cover in animal ecology is determined primarily by human disturbance (Ohashi et al. 2013) and predation pressure (Courbin et al. 2009;Hins et al. 2009). In the case of agricultural landscapes, which are relatively less frequently used by large carnivores than forested areas (Gervasi et al. 2013), it appears that human disturbance is primarily the determinant of space use by ungulates (Jiang et al. 2008;Ohashi et al. 2013).
The impact of human pressure on the spatial distribution of wild boar in agricultural areas seems to be confirmed by the effect of the presence of human settlements within 300 and 1000 m of the islands that we found. While individual buildings did not play any important role, settlements located within 300 and 1000 m significantly reduced the attractiveness of the islands to wild boar. Similar to our study, Ohashi (2013) observed an increase in space use by wild boars with increasing distance from human settlements, which may also indicate that human factors play an important role in shaping wild boar behavior. A similar effect on space use by wild boars depending on distance from settlements was also shown by Keuling et al. (2008), although this author found a similar effect of roads. In our case, distance from roads did not matter probably because the roads were mostly local and poorly traveled. The human-dominated agricultural landscape is generally characterized by a lack/deficiency of security cover, especially in winter, when most post-harvest fields are either devoid of vegetation or covered by very low winter crops at this time of year. At the same time, the characteristics of such a landscape make the animals that use it vulnerable to a range of human threats associated with agriculture, among other things (e.g., distress from human activity and field work) (Strong and Gates 2006;Stankowich 2008;Larson et al. 2016). As a result, animals modify not only their use of space, but also their diurnal pattern of activity toward a more nocturnal one (Johann et al. 2020). However, hunting pressure appears to be particularly important as a major cause of wild boar mortality (Toigo et al. 2008;Keuling et al. 2013), which can be as high as 40% annually (up to 70% for males) and as high as 97%, according to Merli et al. (2017). Wild boars exposed to hunting pressure may reduce their home range (Keuling et al. 2008), or shift it towards forest areas, which provide shelter (Boitani et al. 1994). It should also be mentioned that in Poland the hunting season for wild boars lasts all year round (Bobek et al. 2021), which may further increase the importance of adequate safety guards, especially in agricultural landscapes. It seems important to emphasize that the presence of cover in the form of islands most likely enables wild boars to use agricultural areas more distant from larger forest complexes, which may modify the distribution and intensity of agricultural damage caused by them (Ficetola et al. 2014). As wild boars use open agricultural areas primarily at night (Johann et al. 2020), forest islands provide them with shelter probably mainly during the day. Thus, by reducing movement (no need to move to more distant compact forest complexes), wild boars are additionally likely to reduce energy expenditure (which seems particularly important during winter) (Massei et al. 1997;Keuling 2008). A second feature of the interior of the islands that influenced their use by wild boar was the main stand species. The greater use of islands dominated by pine, spruce and alder may seem somewhat surprising, as most studies indicate that wild boars prefer deciduous forests, mainly with beech and with oak, because of the abundance of mast (Welander 2000;Fonseca 2008;Thurfjell et al. 2009). In the case of our study, beech occurred singly while the lack of association of island use with the presence of oak can be explained by the surrounding of extensive agricultural areas providing food abundance so that wild boars do not have to rely on mast as their primary source of diet (Rutten et al. 2019b). Furthermore, the relatively small size of the islands, means that they likely provide relatively less mast, further limiting their importance as a nutritional source. In view of this, the most significant reason for the more frequent use of islands with conifer stands is most likely again the cover conditions. When we examined how cover conditions on islands were related to the main tree species (Ferens et al. unpublished data), we found that the availability of cover was highest on islands just dominated by pine, spruce and alder. In the case of mature stands, the main species itself does not provide security cover, but this happens mainly through undercanopy regeneration, especially in the case of pine and spruceevergreen species that provide shelter also in winter. As for patches with alder (a deciduous species), their use was probably due not so much to the presence of undercanopy regeneration, but to the inaccessibility of the areas where alders occurred (wet parts of the site) and the shelter in the form of reeds.
The other island characteristics examined (size, width, and proportion of the main species) were insignificant. The lack of effect of island size, a feature that should theoretically also translate to animal safety inside islands, seems particularly interesting. For example, Laundre (2010) showed that mule deer (Odocoileus hemionus) under pressure from pumas (Puma concolor) used larger patches of forest more often, most likely due in part to better cover conditions and the time it took a predator to detect them. In contrast, small patches of forest (< 0.01 km 2 ) were avoided by mule deer. Moreover, according to Laundre (2010), in addition to shelter, higher food availability influenced the more frequent use of larger patches of forest by mule deer. In our study, this factor, due to the presence of food-abundant fields, was most likely not significant. In addition, the variation in island size in our study was relatively small, with 85% of the islands ranging up to 5 ha. This is most likely an area that is far too small so that if there is a threat within the island (people, predators), wild boars trying to avoid it would only move within the island. In such a situation, they are likely to leave the island, seeking refuge within another forest patch or compact forest complex.
Island use was also influenced by the type of agricultural crop in the surrounding area, with the presence of cereals, bare soil and rape having the strongest effect. Although wild boars are omnivores (Ballari and Barrios-Garcia 2014), plant foods found on agricultural fields, among others, are one of their primary food sources (Schley and Roper 2003). The diet of wild boar usually includes at least one energyrich ingredient such as mast or maize (Mikulka et al. 2018); however, there was no abundant oak or beech seed year at the time of this study, and artificial feeding due to ASF countermeasure strategies (avoiding wild boar concentrations) was severely limited. The more frequent use by wild boars of the islands surrounded by bare soil may therefore on one hand, have been due to the maize grown there in the previous growing season, the residues of which covered by soil after harvest. However, maize fields were rather scarce in the area and therefore, another explanation seems more probable. Namely, the ploughed crops surrounding the forest islands probably influence the wild boar safety, because regardless of the temperature, unfavorable conditions for traverse by man prevail, both on foot and by vehicle (mud, slimy terrain or frozen bumps). It may discourage both hunters (no access to pick up the harvested game) and others to visit such forest islands, which may positively impact the wild boar safety in there. Cereals, on the other hand, are most attractive to wild boars during the milky stage of maturity (Amici et al. 2012), but also in winter the green aboveground parts and roots can be a source of energy, as in the case of rape. Thus, not only safety, but also the type of food provided by agricultural crops around the islands was a factor in their use. It is interesting to note the less frequent use of islands surrounded by grassland by wild boar, as this type of agricultural environment is often reported to be one of the most damaged by wild boar (Schley et al. 2008). Increased foraging of wild boar on grasslands may, among other things, result from the need to supplement the diet with protein (invertebrates) under conditions of artificial feeding with maize, which is a food poor in this component (Schley et al. 2008). However, as mentioned, artificial feeding due to ASF was severely limited in the study area, in effect perhaps not contributing to increased foraging intensity on grasslands. However, the interrelationship in the use of islands and surrounding agricultural crops requires further research.
In our study, the effect of the presence of compact complexes on island utilization was at the limit of significance within a radius of 1000 m while their presence (or not) within a radius of 300 m had no effect on island attractiveness. The lack of clear correlations in island use with the presence of compact forest complexes, as already partially mentioned, was probably due to the fact that the islands themselves acted as an alternative refuge to the complexes. Interestingly, island use was negatively influenced by other islands present in the surrounding area. This was most likely due to the ability of wild boars to use them, at the expense of the frequency of their presence. The importance of other islands in the surrounding landscape is emphasized by the fact that their influence was recorded both within 300 and 1000 m radius (whereas, the mentioned compact forest complexes influenced island use only within a kilometer radius). Thus, forest islands are an important component of the agricultural landscape for the ecology of this species, and their importance seems to be even greater than compact forest complexes and positively affect attractiveness of agricultural landscape for wild boar and perhaps many other ungulate species. The use of compact forest complexes by wild boars differed depending on the developmental stage with a clear indication of the attractiveness of pre-thickets stands and on the type of stand-greater use of coniferous forest regeneration sites and pre-thickets stands than deciduous ones. In both cases this seems to confirm the previously indicated importance of security cover in ungulate ecology-see the earlier sections of the paper for the references. Pre-thickets stands-especially coniferous ones-are characterized by clearly more favorable cover conditions compared to the other developmental phases of the stand (Borkowski 2004). Since the habitat use by wild boars in compact forest complexes was largely determined by the cover conditions, it is no wonder that cover as an essential element of the characteristics of islands was so important in the open agricultural landscape.
Comparison of forest island utilization and individual stand developmental stages in compact forest complexes was another element of the research described herein indicating the importance of shelters to wild boar. The average use of forest islands overall was higher than that of forest regeneration sites and mature stands, at a similar level as thickets and pole stage stands, and lower than that of pre-thickets. However, when the analysis was restricted to the 25% most frequently used islands, they were more attractive to wild boar than all (except pre-thickets stands) developmental stages of compact complexes. Thus, forest islands are an important component of the agricultural landscape for the ecology of this species, and their importance appears to be even greater than compact forest complexes.

Conclusions
The results of the research discussed in the present article enable planning actions aimed at rising effectiveness of the wild boar density regulation in the agricultural landscape. One of these measures is a proper placement of high seats used for individual hunting. They should be placeded near forest islands which are selected by wild boars. Features on the basis of which the islands should be chosen for this purpose are as follows: favourable cover conditions, presence of spruce, pine or alder regeneration, cereal or bare soil surroundings, long distance from other islands and settlements. Since surrounding forest islands with bare soil (which most likely restrict their accessibility to hunters) made them more attractive to wild boars, leaving access lanes would allow hunters to visit such places. An alternative to increasing hunting efficiency could be reducing the attractiveness of the forest island to the wild boar. Such actions may include e.g. removing understorey which providers the cover conditions. Agricultural landscape is characterized of low level of biodiversity while forests islands are the centers of biodiversity (Decocq et al. 2016). Understorey and regeneration thinning would allow more light to reach the forest floor causing an increase in biodiversity in the human dominated landscape (Tsai et al. 2018;Barefoot et al. 2019). It should also be mentioned that the survey was conducted before the beginning of vegetation development thus during other seasons, when deciduous trees wear foliage, cover conditions may change. This can possibly make deciduous forests more attractive for wild boars and other patterns may apply. It is also possible that in mast years, compact forest patches play a larger role in habitat selection of wild boars.
Beside Virgos (2002) our study is the first to address the use of forest islands by ungulates in a human-dominated agricultural landscape. Cover conditions were a determining factor in the use of forest islands by wild boar. Islands with interiors with limited visibility and accessibility (groundwater levels, presence of reeds) were particularly likely to be used. The attractiveness of islands was also influenced by habitat elements in their surroundings. Wild boars were most likely to stay on islands where there were only single buildings (or no buildings) and no other islands causing the "dilusion effect". The attractiveness of islands was also increased by the presence of cereals and bare soil in their surroundings. The occurrence of islands in agricultural space most likely enables wild boars to use areas more distant from compact forest complexes. Thus, forest islands improving the heterogeneity of the agricultural landscape also modify the patterns of wild boar functioning in it (Decocq et al. 2016). On the one hand, this may cause the spread of damage caused by wild boar in fields, but on the other hand, it results in a reduction of pressure in areas in the immediate vicinity of compact forest complexes. Since wild boar is not the only large herbivore species utilizing the agricultural landscape, there is a need for research on other ungulate species in the context of the role of forest islands in agricultural landscape exploration. Funding The project was financially supported by Górowo Iławeckie Forest District, Regional Directorate of State Forests in Olsztyn Data availaibility The datasets generated during and/or analysed during the current study are available from the corresponding author on reasonable request.

Conflict of interest
The authors have no relevant financial or non-financial interests to disclose.