Habitat factors influencing the breeding birds on animal farms: can modern agricultural infrastructure be good for birds?

In large parts of Europe biodiversity in agricultural landscapes has experienced strong declines over the past decades because of land-use change and intensification of cultivation methods, but more recently also due to modernization of villages. One of the aspects of modernization is the arising of large specialized poultry and pig farms. Our study carried out in central and east-central Poland focused on the role of habitats within 101 animal farms in relation to bird species richness and number of pairs. We also compared the density of some bird species on farms and in neighbouring villages to determine the importance of the farm environment for the birds. Species richness increased with the number of buildings, the number of trees and the presence of piles of rubble. The number of pairs increased with the number of trees, and the presence of ponds and piles of rubble. Many-times higher densities on farms than in surrounding villages of several species were noteworthy: Black Redstart, Northern Wheatear, White Wagtail and Crested Lark. Animal farms are an important refuge for several species, including those belonging to drastically declining farmland species. Despite the opinion that the modernization of the villages is having a disastrous effect on the avifauna, the modernization may also have positive aspects for some bird species.


Introduction
In large parts of Europe biodiversity in agricultural landscapes has experienced strong declines over the past decades (Donald et al. 2006;Tryjanowski et al. 2011). This has largely been attributed to land-use change (Newbold et al. 2015;Robillard et al. 2013;Brambilla et al. 2021) and intensification of cultivation methods (Newton 2004;Seppelt et al. 2020). Although traditional settlements (buildings and farmyards) provide a range of microhabitats for farmland biodiversity in general (Hiron et al. 2013;Rosin et al. 2016) the rural settlements are usually not considered as important habitats and a source of farmland biodiversity, including birds. Only recent studies have shown that rural settlements are of key importance for avifauna in the agricultural landscape (Garaffa et al. 2009;Rosin et al. 2016;Šálek et al. 2017), and that the relative contribution of modernization of villages versus agricultural intensification to predicted bird declines was estimated at 88% versus 12% .
Changes in the architecture and structure of rural settlements for several dozens of the last years that co-occurred with agricultural intensification resulted in the significant reduction in village ecological values (Antrop 2004;Jokimäki et al. 2021). The abandonment, modernization and specialization of farms, as well as socio-structural changes, are the main factors which led to bird population declines in rural areas (Hiron et al. 2013;Rosin et al. 2016Rosin et al. , 2021. In particular, populations of species that breed in or on buildings have declined markedly across Europe during the last three decades (Inger et al. 2015;Rosin et al. 2016;Mouldrá et al. 2018). For example, two formerly common species, Starling Sturnus vulgaris and House Sparrow Passer domesticus, have been added to the Red List of Birds of Conservation Concern in the UK because of strong population declines (Gregory et al. 2003).
In Poland and other east European countries, the process of village modernization began in the early 1990s after the fall of communism and continued after 2004 with their accession to the EU (Halamska 2011;Reif and Vermouzek 2019). One of the aspects of changes taking place in the villages and in animal husbandry in Poland was the arising of large specialized farms dealing mainly with poultry and pig breeding (Trajer and Mieczkowski 2018;Mirkowska and Ziętara 2019). In Poland, the most important producer of poultry meat in the EU, the number of large-scale poultry farms (> 40,000 chickens) reached 839 in 2016, and the number of pig farms (> 2000 pigs) was 158 (Augustyńska-Prejsnar et al. 2018). Despite such a significant number of farms, it is not known what bird species are associated with them. However, there are indications of the role of these enterprises on a larger landscape scale and animal farms are usually modern facilities which reduce biodiversity through reductions in food resources and nest site heterogeneity (Møller 2001;Vickery et al. 2001).
Here we present results from a large-scale bird survey covering 101 animal farms in Poland. The study area is within central and east-central Poland, where most of these types of farms in Poland are located (Augustyńska-Prejsnar et al. 2018). We examine the role of habitats within farms in relation to species richness and number of pairs. Our expectation is that some environmental predictors like grassland, fallow, piles of rubble, farm buildings as well as shrubs and trees should increase potential nesting and foraging sites (Tryjanowski et al. 2011). In order to determine the importance of farms for some bird species related to rural settlements, we compare of the density of selected bird species on farms and in neighbouring villages. Knowledge of the avifauna composition of animal farms may also be of practical importance due to the transmission of bacterial and viral pathogens by wild birds (Benskin et al. 2009), including those dangerous for humans (Craven et al. 2000;Tryjanowski et al. 2020).

Study area and design
The study was performed in central and east-central Poland in 2021. This region is dominated by extensive agriculture with arable fields (mainly cereals, corn and potatoes), permanent grasslands and midfield woodlots not exceeding 20% of the area. We selected 101 farms in three regions of this part of Poland located around the towns of Łask, Mława and Siedlce (Fig. 1). Farms isolated from other buildings, i.e. at the edge of a village or alone among fields, were included in the studies. All farms bred animals, mainly chickens (88%). The area of the farm was in most cases surrounded by a fence, or possibly clearly separated from the surroundings due to the way the area was used. The minimum distance between adjacent studied farms was 500 m to ensure our sample units were discrete in terms of breeding bird communities. The farms were rectangular in shape. Trees and shrubs grew along the fences, while buildings and access routes were located in central areas. All farms were highly commercial and focused on industrial livestock breeding, but in some of them there was also arable land sown mainly with grain or potatoes.
We measured several habitats features and other parameters which could be related to species richness and number of pairs (Rosin et al. 2016). The data are presented in Table 1.

Environmental data
The following variables were visually recorded, within the boundaries of the farm: number of buildings, number of trees and shrubs, presence of small ponds, presence of piles of rubble and building materials. Trees and shrubs were defined as all woody vegetation above 0.5 m height. The following habitats were measured with an accuracy of 0.01 ha: area of buildings, ploughed area, grassland area, fallow land area, paved area (paving stones, roads) and total area of farm. These environmental variables were taken from the Google Earth map.

Bird surveys
Bird counts were performed twice in a season on all 101 farms. Five well-trained and experienced observers (> 25 years of birding) performed bird surveys during the breeding season. Observations were carried out along the fences, walking around the entire farm (the farm area was not entered due to exposure to avian flu and African swine fever). The first survey was in the period between 15th April and 15th May; the second survey was between 16th May and 30th June. Counting started from just after dawn (one hour after sunrise) until 11 a.m. local time. When counting birds an observer slowly walked around the farm and noted all birds; only birds present in the farm area were included. Territorial and breeding behaviour and the presence of nests were noted. Each survey at one farm lasted from 10 to 40 min. and it was dependent on the surveyed area. Surveys were done during good weather; no rain and strong wind. As a result, the number of singing males or otherwise observed males or females was produced and used to generate the minimal number of pairs for each species, (Heikkinen et al. 2004). The higher score from each survey for each species was included in analysis (Supplementary Material, Appendix 1).

Statistical analysis
It was assumed that species richness and number of pairs are a function of features that characterize the farms used (see Table 1). We also assumed that both bird species parameters explained by the farm features we measured increased under more optimal farms conditions. Two analyses were employed using the Generalized Linear Mixed Model (GLMM) implemented in lme4 and lmerTest library for R. Species richness and the number of pairs were modelled as response variables by using the Poisson distribution and log-link function. Stochastic processes were also included in the models. Farms on three study area were likely to have unquantifiable characteristics that could not be detected by field study. To control for this effect, we included area identifier as a random intercept in the models.
In the first step, for our response variables, we developed models with all ten predictors, and in the next step, using the VIF procedure implemented in the performance library for R, removed predictors with high and moderate multicollinearity (see Supplementary Material, Appendix 2). This procedure, for each response variable, provided us with a new uncorrelated set of explanatory variables. Next, for each response variable, we employed a stepwise approach, which consisted of adding the next predictor (selected on the basis of VIF) to the modelling frame to find a best model. For all stepwise models (and also for the null model) the random structure (one random factor) was maintained. The best model was selected using the Akaike Information Criterion (Burnham and Anderson 1998). The model with the highest Akaike weight and thus lowest AIC value was considered as the most parsimonious, and for this best model, the slope and statistical significance were calculated for each predictor included in the model.
To check whether the model was distorted or not by overdispersion we divided the deviance (model) by the residual degrees of freedom (model). If the value is lower than two the Poisson-model can be used without any restriction. In both cases (for species richness and number of pairs), we calculated values of 1.19 and 1.94, respectively.
The R-square for the best-supported model was calculated as R-square Nakagawa implemented in the performance library for R (Nakagawa et al. 2017).

Relationship between environmental parameters of farms and bird numbers
For species richness, five models gained support using information-theoretic criteria, showing AIC weights > 0 ( Table 2). Model selection procedures allowed us to identify 5 predictors included in the best-supported model (R-square = 0.255, Table 3), but only three, NBuild, NTree and Pile showed a positive and significant influence on species richness (Table 3, Fig. 2). In the case of number of pairs, only one GLMM showed AIC weights > 0 (Table 2). Model selection procedures showed that the most parsimonious model (R-square = 0.507) included NTree, Ponds, and Pile and all of them had a positive and significant influence on this population parameter (Table 3, Fig. 3).

Importance of animal farms for bird species
The species richness and number of pairs of some bird species described below on farms appears to be high. This phenomenon is shown by comparison of the densities of selected species on farms with those in nearby villages with a similar area, studied a few years earlier (Table 4, authors' data). Many-times higher densities were found on farms than in villages with several noteworthy species: Black Redstart, Northern Wheatear and White Wagtail, as well as nesting Crested Lark only found on animal farms (Supplementary Material, Appendix 3 shows the habitat analyses for these four species). Also, in other parts of Poland, the densities of these species in villages (except Black Redstart in some places) did not reach the values found on the farms described in this study, and Northern Wheatear and Crested Lark were not recorded  In addition to the above-described habitats conducive to the occurrence of some species on farms, it should be remembered that farms were usually fenced with a high, Fig. 2 Dependence between species richness and number of buildings (mean and 95% CIs), number of trees and shrubs (mean and 95% CIs), presence of piles of rubble (mean ± SE, min-max) on animal farms in Poland   Fig. 3 Dependence between number of pairs and number of trees and shrubs (mean and 95% CIs), presence of ponds (mean ± SE, min-max), presence of piles of rubble (mean ± SE, min-max) on animal farms in Poland tight fence that limited the risk of predation by domestic cats, dogs and bigger wild mammals, for both adults and nests, especially of ground nesting species (Söderström et al. 1998;Sims et al. 2008). The occurrence of Crested Lark breeding on farms could be affected by a reduction in predation. This species became extinct in this area 20-30 years ago and its relatively large numbers were only recently discovered on farms (Rzępała and Szczypiński 2020). In villages, the number of predators is much higher, and especially cats can take a huge toll on the numbers of birds (Lepczyk et al. 2003;Krauze-Gryz et al. 2019). A slightly higher number of all noted species on farms as compared to villages (61 vs. 53) may also be the result of the penetration of species from the surrounding habitats, i.e. fields, meadows and forests, because farms were usually very isolated sites, often containing relatively large open habitats as well as clumps of old trees. This situation favoured the nesting of Corn Bunting Emberiza calandra, Little Ringed Plover Charadrius dubius and Tawny Pipit Anthus campestris, as well as Coal Tit Periparus ater and Robin Erithacus rubecula. Because of the lack of large open habitats and clumps of old trees bordering the forest within the village boundaries, these species were not recorded there. On the other hand, the total density of all bird species was twice as high in the villages, and this was partly due to the high density of House Sparrows and Barn Swallows in the villages.

Environmental parameters and bird numbers
Species richness on farms increased with the number of buildings, which in our opinion was most likely caused by increasing the number of nesting places suitable for many species of birds, including several of the most numerous such as swallows, sparrows, White Wagtail, Common Starling and Black Redstart. New farm buildings are well-insulated and as a consequence the availability of potential nesting sites (hollows and cavities) is not large (Rosin et al. 2016), but even such buildings allowed some species to find places to nest, and these species are among the most endangered in villages (Inger et al. 2015;Rosin et al. 2016Rosin et al. , 2020. The number of shrubs and trees increased the species richness and number of pairs as well as providing nesting sites, shelter and foraging areas for many birds such as Sylvia warblers, tits and woodpeckers (Skórka et al. 2018). However, the Common Linnet benefited most from the presence of the bushes, because these provided good places for it to nest. The establishment of farms is conditional on a positive opinion of the Regional Directorate for Environmental Protection, which orders the owners to plant bushes as a natural enrichment of such facilities and to reduce the stench of the farm and the noise of operating fans. Such conditions provide benefits for birds. The most frequently planted hedgerow species are Northern White-Cedar Thuja occidentalis, particularly favoured by the Common Linnet (Green et al. 1994;Mason and Macdonald 2000).
The presence of small water reservoirs within the farms favoured the number of pairs. Probably these reservoirs and their banks were of particular importance for the Northern House Martin for obtaining material for nest construction and for foraging (Murgui 2002) and for the White Wagtail which commonly foraged on the banks of the reservoirs (Pustkowiak et al. 2021); these two species were dominant on farms.
The presence of piles of rubble and building materials had a strong influence on species richness and number of pairs. Such places provided breeding sites for Northern Wheatear and Eurasian Hoopoe Upupa epops, and were characterized by the presence of tall weeds and sometimes also contained edible garbage, which attracted foraging birds (Sandström et al. 2006).
Due to the surveys being carried out from outside the farm itself, we are aware that, especially in larger farms, some of the birds may not have been detected by us, but we believe that due to the extensive ornithological experience of the observers and the fact that the habitats of most birds were located near fences, this number is not likely to be significant.

Conclusions
Animal farms situated in agricultural landscapes seem to be important refuges for some farmland bird species. For the Northern Wheatear and Crested Lark farms may be the only places in the area where these species are breeding. Environmental parameters like the number of buildings and trees on the farm, the presence of small ponds and piles of rubble were important for increasing species richness and the number of pairs. Larger areas of these habitats probably contributed to an increase in food resources and nest site heterogeneity. Therefore, it is important to properly plan usable space and manage animal farms to create a bird-friendly space. Such opportunities are provided by the requirement to obtain a positive opinion, with mitigating habitat measures such as planting trees and shrubs, for the construction of animal farms from the authorities responsible for environmental protection.
Open Access This article is licensed under a Creative Commons Attribution 4.0 International License, which permits use, sharing, adaptation, distribution and reproduction in any medium or format, as long as you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons licence, and indicate if changes were made. The images or other third party material in this article are included in the article's Creative Commons licence, unless indicated otherwise in a credit line to the material. If material is not included in the article's Creative Commons licence and your intended use is not permitted by statutory regulation or exceeds the permitted use, you will need to obtain permission directly from the copyright holder. To view a copy of this licence, visit http:// creat iveco mmons. org/ licen ses/ by/4. 0/.