The influence of habitat and the location of nest-boxes on the occupation of boxes and breeding success of the Kestrel Falco tinnunculus

Abstract Nest-boxes are often used for many bird species, not only passerine hole-nesters but also birds of prey. Increasing the availability of nest-sites, including artificial ones, is one strategy of habitat enhancement with specific conservations goals. One of the species that willingly inhabits boxes is the Kestrel Falco tinnunculus. We studied the influence of habitat in the vicinity of boxes, and nest-box locations, on their occupation and the breeding parameters of Kestrels in Central Europe, in Poland. We used data collected over 17 years for more than 400 nest-boxes. We found that Kestrels prefer nest-boxes hung on power poles and in areas with a lower proportion of forests. Breeding success decreased with the increasing proportion of grasslands and increased with the increasing number of individual trees. We did not find any environmental components with an impact on productivity of Kestrels. Our findings could be useful in future program planning and maintaining stable populations. We recommend locating boxes for Kestrels in arable lands or grassy areas with a mosaic of habitats away from bigger trees. Furthermore, we recommend hanging boxes on power poles and also to avoid hanging boxes on trees, especially near forests.


Introduction
Nest box deployment is a common conservation action for cavity breeding species (Pöysä & Pöysä 2002;Libois et al. 2012) and for some species of birds of prey, especially falcons (Rahman et al. 2014;Gustin et al. 2017;Calabrese et al. 2020). When a population is limited by nest site availability providing artificial nests may increase population size (Catry et al. 2009;Wiebe 2011). Although nest-box provisioning is typically helpful for active bird conservation, it should be applied with caution, because nest-boxes in low-quality habitats may serve as ecological traps and make unsuitable sites attractive for birds with subsequent decreased breeding success (Mänd et al. 2005;Krams et al. 2021). In addition, artificial breeding sites used for many years may be easy to detect by predators or may increase the likelihood of parasitic or similar infestation (Mitrus 2003;Boulahbal et al. 2020). Another downside of nest-boxes is the maintenance required to avoid accidents during the breeding season i.e. nest-box falling from its support during the breeding season (Lindenmayer et al. 2009).
One way to increase the efficiency of conservation efforts is to monitor the effectiveness of ongoing programs and to manage these programs adaptively (Katzner et al. 2005). For habitat enhancement and to increase the population size of birds, several factors of nest-box design and installation at locations where the surrounding habitat is suitable should be taken into consideration, as they may have various effects on degree of occupancy and breeding parameters (Gottschalk et al. 2011;Lambrechts et al. 2012).
*Correspondence: A. Golawski, Siedlce University of Natural Sciences and Humanities, Siedlce, Poland. Email: artur.golawski@uph.edu.pl The Kestrel Falco tinnunculus is one of the bird species for which many programs of active protection through the installation of nest-boxes have been devoted (Valkama & Korpimäki 1999;Fargallo et al. 2001;Kasprzykowski et al. 2021). This falcon is a small raptor widespread in open landscapes throughout the Palearctic (Village 1990). The population in Europe is probably stable, because the observed trend of the population size of this species in the last 10 years is increasing (+ 27%), however the long-term trend is decreasing (−27% in 1980-2019) (PECBMS 2021). Under natural conditions, Kestrels breed mainly on trees in old nests of other bird species e.g. corvids, or in cavities on cliffs and trees. Natural nest-sites can be a scarce resource in many areas, constituting a limiting factor on the Kestrel breeding population (Fargallo et al. 2001). Although it cannot be considered a strict cavitynesting bird, this falcon readily accepts nest boxes and probably prefers them when available (Valkama & Korpimäki 1999;Fay et al. 2019). Also, the success of nesting in boxes was usually higher compared to natural places (Fargallo et al. 2001;Charter et al. 2007).
In studies of the reproduction of Kestrels in boxes, the main focus was on the parameters of the boxes (Valkama & Korpimäki 1999;Charter et al. 2007;Fåsseland 2012), various aspects of the diet (Valkama & Korpimäki 1999;Costantini et al. 2007a;Kaf et al. 2015;Luna et al. 2020) and the impact of weather conditions (Charter et al. 2007;Costantini et al. 2010;Kasprzykowski et al. 2021). On the other hand, detailed data about the influence of habitat in the vicinity of the boxes on their occupation and breeding parameters were described less frequently. In addition, studies were conducted mainly in northern (Scandinavia) and southern Europe (Iberian Peninsula) (Valkama & Korpimäki 1999;Aviles et al. 2001;Fåsseland 2012), characterized by different environmental and climatic conditions compared to the vast lowlands of Central and Western Europe which support a significant proportion of the Kestrel population (Village 1990).
The primary goal of our study was to identify the factors affecting the occupancy and breeding success of Kestrels in newly provided nest-boxes in relation to nest-box location (trees vs. power poles) and the land cover around the nest boxes. In this study, we use data collected over 17 years for more than 400 nest boxes. Essentially, we wanted to identify the factors associated with unfavorable nest-box locations in order to propose new management actions to improve conditions, in this respect, in the agricultural, lowland landscape of Central Europe.

Study area
The study was performed in east-central Poland around Siedlce town (52°12ʹN, 22°17ʹE) in 2004-2020. This region, with an area of about 2,000 km 2 , is dominated by extensive agriculture with arable fields (46.9%), meadows and pastureland with 14.7% coverage, and orchards, mainly apple trees, occupying 2.6% of the land. Woodlands cover 24.8% of the region, urbanized areas 5.7%, and the remaining 5.3% consists of water bodies and wasteland (Statistical Yearbook 2015). The area has a temperate transitional climate (Degirmendzic et al. 2004). The average yearly air temperature in Siedlce ranged from 7.2°C to 9.1°C. July is the warmest month, the average temperature being 19.4°C (Radzka 2020).

Nest boxes and monitoring
Nest boxes for Kestrels started to be installed by the Nature Society "Stork" in the autumn of 2003 and they were replaced if damaged in following years. In 2003-2017, 459 nest boxes were installed (Figure 1), starting from 83 boxes in the first year and then hanging an average of 26.9 (SD = 22.6) new boxes per year. The average distance between the nearest neighboring boxes was 630 m (SD = 631.5 m, n = 459). The same type of nest box (30 cm wide × 40 long × 30 cm high) was used for the entire period of study, constructed from 2 cm thick pine boards and half-open to one side. The roof was secured with a metal sheet or roofing felt, and all the walls were coated with a preservative agent. Nest boxes were hung at the same height on trees (69.7% of the nest boxes) or on poles of power lines (30.3% of the nest boxes), between 6 and 7 m above the ground. (Kasprzykowski et al. 2021). During the breeding season nest boxes were inspected at least twice, with binocular from a vantage point on the ground: once in May (to determine whether or not the nest box was occupied by falcons which should be incubating at this time, Kasprzykowski et al. 2021) and then once or twice again, depending on the age of the fledglings and the possibility of counting them. Occupied nest boxes were those with birds incubating or where alarmed birds flying around the nest box were seen. Successful broods were those where at least one fledgling was observed. If, during the second inspection, the fledglings were too small to be counted, a third inspection was undertaken. The size of the fledged brood was taken to be the number of big feathered nestlings present in boxes in their fourth 2 M. Rzępała et al. week of life (Fargallo et al. 2001;Kasprzykowski et al. 2021) and their number was counted using a telescope and binoculars. In the first years of monitoring, the inspections covered only a proportion of the installed boxes, while since 2012 the majority (80-90%) of them were inspected. Figure 2 shows the proportion of occupied boxes in subsequent years, limited to the number of boxes that were inspected in a given year.

Environmental data
Open-source software QGIS v. 3.22.1. was used to perform habitat analyzes. In the open areas of Europe, Kestrels are known to forage within a few km of the nest, but mostly do so no farther away than 1000 m (Korpimäki et al. 1996;Norrdahl & Korpimäki 2002), thus we measured habitats within 1 km radius of the nest boxes. We made calculations within the assumed range for each box separately. We used data from the Corine Land Cover (CLC) system to analyze the land cover in the vicinity of the nest-boxes. In the analyzes, we distinguished habitats that are commonly used in this type of research (Radović & Tepić 2009;Białas et al. 2020) and were also previously analyzed for Kestrels (Valkama & Korpimäki 1999): urbanized areas, arable fields, grasslands, woodlands and habitat mosaics (areas of small arable fields alternating with small meadows and pastures, as well as wooded areas and small water reservoirs). We downloaded the spatial data of Corine Land Cover from the website of the Polish Office of Inspectorate of Environmental Protection (https://clc.gios.gov.pl/).

Nest-box location and Kestrel reproduction
For the analysis, we also used selected vector data from the Database of Topographic Objects (BDOT10k), the specificity of which corresponds to maps at a scale of 1:10000. Using these resources within a radius of 1 km for each box, we calculated: the area occupied by roads, the length of power lines and the number of single trees. We downloaded topographic data from the website geoportal.gov.pl, which is run by the Head Office of Geodesy and Cartography. The available data on land use showed that in the described area, between 2002 and 2020, changes in use were insignificant: the area of agricultural land decreased by 4.0%, the area of woodlands increased by 1.5% and urbanized areas by 1.5%, while the area of wasteland remained unchanged (UMWM 2020).

Statistical analysis
We calculated the frequency with which each nest box was occupied (using any or all of the following criteria: alarmed birds near boxes, birds in the box in the incubation position, fledglings) during the years of monitoring. We also calculated average breeding success for each box (successful nesting attempts were those in which at least one chick fledged), i.e. the number of years with fledgling(s)/the number of years the box was occupied. For each nest box, we calculated the average number of fledglings for years where the birds bred successfully. Similar to Katzner et al. (2005) we used data from boxes for which we had the above-mentioned nesting parameters from at least 5 years. This was 411 nest boxes in relation to box occupation, 116 nest boxes in relation to breeding success and 35 boxes in relation to the average number of fledglings. We modeled the relationship between the parameters of Kestrel occupancy/productivity mentioned above and environmental variables and nest box location using three GLZ analyses where the dependent variables were: 1) occupancy rate of nest boxes, 2) average breeding success, 3) average number of fledglings. We used GLZ when our data did not match the normal distribution. Because zero values made up a significant portion of the data in relation to occupancy rate of nest boxes we utilized the Tweedie distribution model (Swallow et al. 2016;Carle-Pruneau et al. 2022). In the other two analyses, we assumed a gamma distribution and log link function for the dependent variables (McCullagh & Nelder 1989). Explanatory variables used in all the models were habitat factors and location (tree/pole) of nest boxes (see Table I). A criterion based on the Wald statistic (χ 2 ) was applied to determine which variables had significant influence on occupancy/ productivity parameters in falcons. Explanatory variables were tested for multicollinearity by examining the Variance Inflation Factor (VIF) (Quinn & Keough 2002), when VIF > 5 we discarded the variable from the analysis. However, the variables did not exceed the value of 2.75, so all of them were included in the model. We also selected the models using Akaike information criteria (AICc, Burnham & Anderson 2002) to establish and rank a list of candidate models with Δ i ≤ 2, but it included all analyzed variables, so we did not show this criterion. Using the Spearman correlation, we also checked whether or not there was a relationship between the frequency of occupation of boxes and the number of fledglings. Only those results with a probability of α ≤ 0.05 are assumed to be statistically significant. The analyses were performed in Statistica 12.0 (StatSoft Inc 2014).

Results
Over the period 2004-2020, Kestrels occupied a given box for an average of 33.5% of the seasons (SD 30.7%, Range 0.0-100%, n = 411 nest-boxes). The occupation of boxes depended on two factors (Table II). The frequency of occupancy decreased with increasing area of woodlands and Kestrels occupied boxes placed on poles significantly more often compared to those on trees (Figure 3). The other habitat factors did not affect the occupation of boxes by falcons (Table II).
Breeding success of Kestrels was found for 85.2% of broods, where their fate was determined (SD 17.5%, n = 116), with a range of 20-100% for a given location of nest box. Average breeding success was dependent on two factors (Table II); breeding success decreased with an increasing proportion of grassland, but increased with the increasing number of individual trees. The other habitat factors did not affect the falcon's breeding success (Table II). Neither did the location of the nest-boxes, although the average breeding success in boxes on poles was 90.1% (n = 62), whilst that on trees was only 79.5% (n = 54).
The average number of fledglings in a brood of Kestrels was 4.2 (SD 0.6, n = 35) with the range 2.8-5.5 fledglings for a specific location of the nestbox. There were no habitat factors significantly affecting the number of fledglings (Table II), and the number of fledglings in boxes on trees (n = 8) and on poles (n = 27) was the same, namely 4.2 fledglings/box. The average number of fledglings was not dependent on the frequency of occupation of a given box (Spearman's correlation, r s = −0.27, p = 0.119, n = 35).

Discussion
Our results showed that an increased proportion of woodlands in the area within 1 km of a nest box led to fewer Kestrels occupying boxes. Boxes located on trees were definitely less willingly occupied than those located on power poles. On the other hand, breeding success decreased with the proportion of Table II. Results of GLZ models for: (a) the occupancy rate (number of breeding attempts/number of years the box was available), (b) average breeding success (number of years with fledgling(s)/number of years the box was occupied), (c) average number of fledglings produce. Only nest boxes that were monitored for at least five years are included. Statistically significant effects are emboldened (N number of boxes).

Nest-box location and Kestrel reproduction
grasslands, and increased with an increase in the number of individual trees around a box. Kestrels prefer to breed in open habitats because they can easily hunt in these areas (Garratt et al. 2011;Larsen 2012), and thus woodlands are considered a poor quality habitat for Kestrels (Valkama & Korpimäki 1999;Fåsseland 2012). An increase in woodlands area around the breeding site is not favorable for this species, because birds have to hunt in more distant places, which is associated with greater energy expenditure (Catry et al. 2013). Avoidance of tree-mounted boxes may involve a general avoidance of wooded areas, although a significant number of boxes were mounted on single-growing trees or in small clumps. Perhaps the boxes located on trees are less safe. In this study, we did not find significant statistical differences in breeding success related to the location of the nest box, but on the other hand, the difference of 10.6% between success in both locations is noticeable. Potential Kestrel predators such as Martens Martes sp. (Fargallo et al. 2001) can easily climb to the boxes on trees, and in addition, their occurrence is generally associated with wooded areas (Virgós et al. 2012). Northern Goshawk Accipiter gentilis may also exert pressure on Kestrels settled on forest edges (Petty et al. 2003). The breeding success of falcons decreased with an increase in the area covered by grasslands, and increased with an increase in the number of individual trees around a box. In the meadow and pasture habitats in the study area the highest numbers of Carrion Crows Corvus cornix and Ravens Corvus corax are recorded (Ornoch 2020) and these species are predators of Kestrel broods (Kuznetsov 1998;Csermely et al. 2006), including those nesting in nest boxes (Fargallo et al. 2001). Also, Magpies Pica pica cannot be ruled out as predators (Pererva 1979), their numbers being high in the bushy meadows in this part of Poland (Golawski 2004). Generally, Kestrels prefer grasslands due to the rich food base and the availability of prey (Aschwanden et al. 2005;Garratt et al. 2011), however they avoid settling near corvid's nests (Kuznetsov 1998). On the other hand, the positive impact of single trees on breeding success may be associated with the possibility of foraging opportunities (hunting places) closer to the nest (Aschwanden et al. 2005;Christensen 2012). In addition, individual trees can help falcons to observe the area near the nest and to respond to the threat of approaching predators.
We did not find that any of the examined factors had a significant effect on the number of Kestrel fledglings. No difference in the average number of fledglings between broods located on poles and on trees was also found for a much larger sample of broods of this species (Kasprzykowski et al. 2021). Because nesting near a power line has not been found to have any detrimental effects associated with magnetic field exposure on body condition, serum carotenoids, or oxidative stress in wild Kestrel nestlings (Costantini et al. 2007b), nest boxes on power poles are a better location than trees for falcons. None of the analysed habitat types affected Kestrel reproduction. This could be explained by a large mosaic of habitats in eastern Poland. Both arable lands and meadows appear to provide adequate food conditions for Kestrels, as was found for other species with similar diets (Golawski & Meissner 2008;Morelli et al. 2016).
According to other authors, Kestrels prefer boxes and achieve higher reproductive parameters there than in natural sites (Korpimäki 1983;Fargallo et al. 2009). Probably also in east-central Poland, nesting boxes are also preferred compared to natural sites; as observed in 2019, in an area of 130 km 2 , out of 16 nesting pairs of Kestrels, 12 were using nestboxes (Ornoch 2020). Thus, boxes are a highly preferred breeding place (Valkama & Korpimäki 1999), possibly also because individuals raised in boxes only look for such places (imprinting) for reproduction, as previously shown in the Peregrine Falcon Falco peregrinus (Kirmse 2001).
Therefore, it seems that, continuing the program of active protection of the Kestrel will lead to the maintenance of the stability of the population or even its growth (Kuczyński & Chylarecki 2012). However, protective activity involving replacement/ repair of damaged boxes, cleaning and monitoring are costly. Every year, about 10 boxes "disappear" M. Rzępała et al. with trees cut by the owners of the land, so they should be replaced with new ones, preferably made of waterproof plywood, which extends their life to about 10 years. Boxes placed on power poles should be safer in this respect. We estimate that in the studied area, protective measures may consume approximately EUR 5,000 per year. Therefore, given the low funding of organizations dealing with animal protection, it is a significant amount for these organizations. Active protection of this species is important because other studies imply that nest boxes may not only impact local, but also large-scale population dynamics (Fargallo et al. 2001;Fay et al. 2019).

Conclusions
Our study shows practical suggestions that can be implemented in ongoing conservation programmes in rural populations of Kestrels. We found that breeding pairs prefer nest boxes hung on power poles and in areas with a lower proportion of woodlands. Individual trees also contribute to increasing breeding success, which may possibly allow birds to hunt closer to their nests and to react quickly to predators. Assuming that none of the analyzed environmental components had an impact on the productivity of Kestrels, arable lands or grassy areas with power poles, away from trees where corvids can nest, are proposed as places to locate boxes, preferably within conditions of large mosaics of habitats. However, you should avoid hanging boxes on trees, especially near woodlands. These findings could be useful in future program planning and maintaining stable Kestrel populations.