Changes in the Montagu’s Harrier Circus pygargus diet in Eastern Poland across decades promote insects and reptilians, but not birds and rodents

Abstract We investigated temporal changes in diet composition of the Montagu's Harrier Circus pygargus breeding in natural habitat (calcareous peat bog) in SE Poland. We characterized diet composition in a three‐year period (2007–2009), based on pellet analyses. We investigated whether diet composition was affected by years or stage of breeding. We compared diet of the studied population between 2000s and 1990s and with other populations. We found that the food of the studied population was dominated by insects and mammals (by number) and mammals and birds (by biomass). Biomass and abundance of main prey items differed between studied years because of different air temperatures. We found some interannual differences in contribution of some prey items including higher number of thermophilic prey (insects and amphibians) in warmer years. Comparison of pellet composition in the 1990s and 2000s revealed significant increase in the abundance of thermophilic prey (insects and reptiles) and decrease of mammals including Microtus voles and birds. Those changes may be linked to habitat changes in areas neighboring peat bogs and climate change‐induced changes in prey communities. The studied population was able to respond to changes in foraging habitats and prey composition by opportunistic foraging on easily available prey. The diet of the studied population is the most similar to the geographically closest populations foraging in similar habitats and characterized by high contribution of insects.

abundance is often a limiting factor for breeding density and reproductive performance, and thus, raptors are used as indicators of ecosystem status (Francksen et al., 2017;Kettel et al., 2018;Rodriguez et al., 2020;Sodhi & Ehrlich, 2010). Raptors hunt for their prey either opportunistically, that is, in proportion to their relative abundances, or because of their relative profitability (Tores et al., 2005;Seaton et al., 2008;McKinnon et al., 2013;Rodriguez et al., 2020). Studies on temporal and spatial variation in the diet composition of raptors are important to understand their foraging ecology, breeding performance, population density, or habitat selection (Litvaitis, 2000;Lopez-Lopez et al., 2009;Di Vittorio et al., 2017).
The Montagu's Harrier Circus pygargus is one of the smallest birds of genus Circus sp. (Leroux & Bretagnolle, 1996) (Figure 1). It is a migratory species breeding from North Africa, and West and Central Europe across to India and wintering in Africa and India (Clarke, 1996;Cramp & Simmons, 1980). The Montagu's Harrier is a semicolonial species, at least in some regions, with a flexible pattern of nest distribution, ranging from solitary to colonial (up to 12.7 pairs/10 ha area) (Krogulec & Leroux, 1994). It is a ground-nester traditionally breeding in open landscapes with dense and tall herbaceous plant communities including marshes, lowland heath, rough grasslands, and steppes (Clarke, 1996;Cramp & Simmons, 1980;Terraube et al., 2010). However, in the western Palaearctic, Montagu's Harrier has started to breed in agricultural areas in recent decades, breeding in crops, mainly winter cereal or alfalfa. In the late 20th century, more than 90% of harriers in Iberia, 70%-80% in France, and 40%-50% in other western European countries nested within cereal crops (Arroyo et al., 2003). In the eastern part of Europe, increasing frequency of breeding in crops and decreasing importance of wetlands for nesting have been observed since late 1980s. Before, Montagu's Harriers were nesting entirely in wetlands in this region (Krogulec & Leroux, 1994). In Poland, until the late 1970s, nesting of the Montagu's Harrier was reported exclusively in peat bogs (Tomiałojć & Stawarczyk, 2003). Currently, up to 80%, the nests of this species are recorded in crops (Krupinski, 2017;Mirski et al., 2009).
It is an opportunistic predator preying on various birds and mammals, amphibians, reptiles, invertebrates, and avian eggs (Arroyo, 1997;Terraube & Arroyo, 2011). Diet of populations breeding in higher latitudes is dominated by mammals, especially rodents like voles (Microtus sp.), while southern populations are more focused on birds, reptiles, and insects (Terraube & Arroyo, 2011). Avian prey seems to serve as a very important diet component across the Montagu's Harrier distribution range. In steppe habitats, reptiles contribute considerably to total prey biomass (Terraube et al., 2010).
The relative importance of particular prey items in the diet is significantly explained by their local abundance and foraging habitat characteristics, confirming the opportunistic foraging strategy of this raptor (Terraube & Arroyo, 2011). Raptors may display prey switching strategy, that is, increasing the frequency of hunting on a particular prey species when that species is abundant and decreasing when it is scarce (Murdoch, 1969). This is particularly evident for prey species that undergo seasonal or interannual massive abundance changes (Robinson et al., 2019;Steenhof & Kochert, 1988).
The aim of this study was to investigate temporal (interannual and within-year) variation in diet (based on composition of regurgitated pellets) of Montagu's Harriers breeding in natural habitat (calcareous peat bogs) in SE Poland. We expected higher contribution of ectothermic prey (insects and reptiles) in phases of breeding season with higher air temperatures (i.e., chick-rearing and postfledging periods) compared to the earlier phases (prelaying and incubation). We also expect higher abundance of avian prey in the diet as fledging birds are more available late in the season (Arroyo, 1997).
Considering possible interannual variability in meteorological conditions, we also expected some interannual differences in pellet composition, for example, a higher frequency of ectothermic prey in warmer years. To investigate whether climatic conditions affected the presence/absence of ectothermic prey in the diet at consecutive stages of breeding, we also analyzed meteorological data in particular phases of breeding. Finally, we also compared diet composition in the studied population with historical data from the same area (1985-1995, i.e., a decade before) and with other breeding populations. We expected that due to climate and habitat changes, thermophilic prey (like insects and reptiles) will have a higher contribution to the current diet than in the past. Regarding comparison with other breeding populations, we expected that geographically closest populations (in Poland and Belarus) will be characterized by the similar diet composition.

| Study area
This study was conducted in calcareous peat bogs near Chełm in SE Poland in the nature reserve "Bagno Sereberyjskie" (area of 377 ha) where 7-12 pairs of Montagu's Harriers were breeding in the studied period (2007)(2008)(2009) (Buczek, 2005;Fijalkowski, 1959). The Montagu's Harrier in the studied area builds its nest in clumps of sedges surrounded by water, or in partly paludine areas (Krogulec, 1992).
Direct observations revealed that peat bog, because of dense vegetation dominated by Great Fen-sedge is rarely used by the studied population of Montagu's Harriers as hunting habitat (Krogulec, 1992;Wojtak & Kitowski, 2003;Kitowski, 1994). The study of foraging preferences of the local population revealed that only 19% foraging individuals were observed in dense vegetation dominated by Great Fen-sedge (covering 33% of investigated area), while 48% and 30% individuals foraged in meadows (28% of investigated area) and pastures (15% of area), respectively (Kitowski, 2007).
Thus, birds from local population foraged mainly in meadows, pastures, and arable land neighboring the peat bog. Those areas are characterized by a mosaic of various microhabitats including small waterbodies, groups of trees, and bushes along drainage ditches in meadows (Wojtak & Kitowski, 2003;Kitowski I.-unpublished data).
Home ranges of Montagu's Harrier males in Eastern Poland were estimated to be 67 km 2 , on average, which means that the studied bog area itself cannot sustain the foraging needs and adjacent farmland is crucial for hunting (Krupiński et al., 2020).

| Fieldwork
We reconstructed the Montagu's Harrier diet composition based on prey remains in pellets. We collected them from spring to summer in 2007, 2008, and 2009. Pellets were collected from the railway embankment crossing the "Bagno Sereberyjskie" reserve. Rails, railway bollards, and shrubs at the edge of embankment served as resting, pellets were collected, packed in zip paper bags and transported to the laboratory. In total, we collected 464 pellets.

| Laboratory analyses
We disintegrated collected pellets in order to extract all the solid material, for example, bones, beaks, scales, and insect remains. We determined prey items using keys for the identification of mammals (Pucek, 1984) and birds (Brown et al. 1987;Jenni & Winkler, 1994). We identified mammals by bones and teeth, birds by bones, beaks, and feathers, and lizards by skulls, jaws, skin, and scales. We grouped insects not identified to the species level into the three most frequent taxa (Coleoptera, Hemiptera, and Tettigonideae). To avoid possible biases associated with a secondary predation, we excluded remains of small beetles from the analysis when they were found together with lizard scales or bird remains in the same pellets. As many collected pellets in the field were broken, to avoid multiplication of prey number from fragmented pellets, we pooled all pellet fragments collected during one visit to form one sample.  August as postfledging period (Krogulec, 1992). In total, we per- We analyzed abundance and biomass of prey remains found in pellets. We estimated abundance of particular prey items in all pellets collected during one sampling. We estimated biomass of particular prey item categories based on literature data. For mammals, we used

TA B L E 2 Abundance and biomass of prey items found in pellets of
In statistical analyses, we used only prey categories represented by at least 3 prey items. Due to relatively small sample size and unbalanced sample size for some breeding stages (lack of data from prelaying period in 2007 and from postfledging period in 2009), we firstly compared pellet composition between the breeding stages for all years combined and then between the years with all stages of breeding combined.
To compare the qualitative and quantitative composition of pellets among the stages of breeding and between years, we applied the following multivariate methods: 1. Nonmetric multidimensional scaling (nMDS), an indirect gradient analysis approach that produces an ordination based on a distance matrix explained by the Bray-Curtis similarity measure (Taguchi & Oono, 2005), to visualize dissimilarity of the studied years and stages of breeding; we also visualized environmental variables (meteorological data) in form of biplot; the correlation coefficients between each environmental variable and the NMDS scores are presented as vectors from the origin. The length of the vectors are arbitrarily scaled to make a readable biplot; thus, only their directions and relative lengths should be considered (Hammer et al., 2001); we considered the following guidelines for acceptable stress values: <0.05 = excellent, <0.10 = good, <0.20 = usable, >0.20 = not acceptable (Clarke, 1993).

Analysis of group similarities (ANOSIM), a procedure based on
Bray-Curtis measure of similarity to test differences among groups (Clarke, 1993).
3. The similarity percentage breakdown (SIMPER) procedure to assess the average percentage contribution of individual factors to the dissimilarity between objects in a Bray-Curtis dissimilarity matrix (Clarke, 1993).
We presented nMDS graph and SIMPER results only when we found significant differences between compared categories (phase of breeding or year) in ANOSIM.
After multivariate analyses, we performed unimodal tests (Kruskal-Wallis and Mann-Whitney U tests) to compare particular prey items abundances and biomass and also meteorological conditions between the studied years.
To investigate within seasonal trends in meteorological conditions and presence of thermophilic prey (insects and reptiles) or prey with seasonal peaks of availability (birds), we used Spearman rank correlation. Due to relatively small sample size, we combined data from all years. We transformed consecutive phase of breeding into numeric values ordered in phenological order, that is, with values 1-2 for the prelaying period, 3-4 for the incubation, 5-6 for the chick-rearing period, and 7-9 for the postfledging period.
To assess the similarity of the diet composition in the studied population and other populations of Montagu's Harriers, we used principal component analysis (PCA). We used this technique to reduce the number of variables to a few new ones called factors, representing groups of prey categories. To avoid potential methodspecific biases in diet compositions, we selected only studies based exclusively on pellets analyses. In total, we found and employed in analyses 9 studies (8 from Europe and one from Asia).
We performed nMDS, ANOSIM, SIMPER, and PCA analyses using PAST 3.0 software (Hammer et al., 2001). We expressed results as similarities in nMDS and ANOSIM and as dissimilarities in SIMPER.
We performed unimodal tests and Spearman correlation in R software (R Core Team, 2017) in ggpubr package (Kassambara, 2019).

| Diet composition
In total, we collected 464 pellets (127 in 2007, 179 in 2008, and 158 in 2009). Mean number of prey items per pellet was the highest during postfledging period and the lowest during the chick-rearing period ( Table 2). We identified six main types of prey in the pellets ( Figure 2; Tables 1 and 2). Insects were the most abundant (70.7%) prey category; however, their contribution to the biomass was very small (4.7%) ( Table 2). Only during the postfledging period insects made up 6.8% of total biomass. The most numerous insect, the Field Cricket Gryllus campestris was mainly preyed during the prelaying period.
Amphibians had negligible contribution, both to the total abundance and to the biomass (1.6 and 2.0%, respectively).
Reptiles constituted 5.8% and 5.6% of total abundance and biomass, respectively. During the prelaying and incubation period, their contribution to the diet was higher, making up 10.3% and 8.0% of prey biomass, respectively. Among identified remains of reptiles, those belonging to lizards Lacerta sp. were the most numerous (86.7%) ( Table 1).
Birds were not a very abundant prey group, making up only 4.2% of the total share. However, their contribution to the total biomass was much higher (12.5%) (

| Within-year variation in diet and meteorological conditions
We found significant differences between the phases of breeding (for all years combined) neither in abundance nor in biomass of all prey categories (ANOSIM with Bray-Curtis as distance measure: abundance: R = 0.071, p = 0.233, biomass: R = 0.0193, p = 0.379).
We found significant relationship neither between phase of breeding and meteorological conditions nor between phase of breeding and abundance of various prey items (for all years combined) (Table 3).

| Interannual diet variation
We found significant differences in abundance and biomass of all prey categories between the studied years (for all phases of breed-

We found that other insects and Field Cricket Gryllus campestris
contributed the most (16.5% and 16.1%, respectively) to the pattern of overall average dissimilarity (44%) in prey items abundance between years (SIMPER) (

| Interannual variation in meteorological conditions
Unimodal tests revealed that mean, max, and min air temperatures in

| Comparison to other breeding populations
Principal component analysis performed on data on diet composition in various populations revealed that the first axis (explaining 48.6% of total variance) was the most correlated with reptiles (r = 0.79) while the second axis (explaining 19.1% of total variability) with mammals (r = 0.56) and insects (r = 0.54) (Figure 8).
Our summary of the Montagu's Harrier diet composition ( Figure 8; Table 6) indicates that diet of the Chełm calcareous peat bogs population is the most similar to diet of birds from closely located site in Poland2 (Podlasie) (Figure 8). In both areas, insects constituted the most important part of prey abundance (> 71%) and reptiles were significantly represented. In all location in Poland1 and France1, insects constituted considerable part of the diet (>70% of abundance); in all other sites, their contribution was negligible (<14%). Reptiles serve as an important prey item (54%) only in the study from Kazakhstan, Netherlands, and France2 (>30%); in other sites, their contribution to the total prey abundance was negligible (< 20%). Contribution of mammals into diet of the Chełm calcareous peat bog population was considerably smaller than in majority of other sites except (>45%) from Poland and Kazakhstan with similar or even lower share of this prey item (10%-17%) ( Table 6).

| Factors affecting diet composition in the studied population
Our study revealed that the Montagu's Harrier pellet composition prey. The Field Cricket is known to have high winter mortality of its larval stage (nymphs) (Hochkirch et al., 2006;Remmert, 1992). Thus, In contrast to our expectations, we did not detect withinbreeding season trends in number of ectothermic prey (insects and reptiles) or characterized by seasonal peak of availability (birds). This result may be affected by high availability of an alternative prey providing necessary energy content and nutritional value and/or lack of significant correlation between thermophilic prey and meteorological conditions, especially air temperatures.

| Temporal variation in the Montagu's Harrier diet
We found some changes in diet composition of the studied population compared to studies from the 1990s. They included increase of abundance of insects and reptiles, decrease in number of mammal (especially voles) and avian prey, and presence of new prey species. We interpret increasing role of thermophilic prey (insects and (dominating prey searching technique) (Bildstein, 2017;Clarke, 1996;Rice, 1982). After intensive changes in Polish agriculture after 1990s, main foraging grounds of the studied population of Montagu's Harriers (located in farmlands outside the calcareous peat bogs) were subject to strong transformations including increase in area of habitats covered with high vegetation as a result of wide scale abandonment of herbaceous vegetation mowing and grazing, land fallowing, and strong expansion of rapeseed and maize cultivation (Buczek & Buczek, 2017;Wojtak & Kitowski, 2003). Those transformations resulted from considerable decrease in the number of cows and sheep in E Poland (including the studied area) recorded in early 1990s (Dzun, 2012;Ptasinska-Marcinkiewicz, 2014) and continuing until the years of the study (Główny Urząd Statystyczny, 2020).
After the accession of Poland to the European Union in 2004, rapidly F I G U R E 7 Interannual differences in daily air temperatures (a), mean humidity and total precipitation (b) recorded during the breeding period (all stages combined) of Montagu's Harriers in Włodawa meteorological post (40 km from the studied area) in the studied years. Boxplots show the median (band inside the box), the first (25%) and third (75%) quartile (box), the lowest and the highest values within 1.5 interquartile range (whiskers) and outliers (dots  (Pagorski & Krupinski, 2016). The presence of some new species not reported in 1985-1989(Krogulec, 1992 as European Hamsters or European Water Vole may be explained by more frequent exploitation of habitats not used in the past. Observed processes including decrease in contribution of mammals, especially voles many have some demographic consequences for the local population. Raptors, both diurnal and nocturnal including the Montagu's Harrier, are classified as income breeders gaining nutrients for egg production locally during the prelaying period (Durant et al., 2000;Stephens et al., 2009). Indeed, Montagu's Harriers just arrived from the wintering quarters forage intensively (Krogulec, 1992). One may expect that quality and quantity of food ingested during this period will affect considerably egg and offspring quality. Thus, significant decrease in the number of consumed voles may have affected negatively reproductive performance of the studied population. Microtus vole is considered as a high-energy prey (energy density of one individual is estimated at 52.8 kJ (Masman et al., 1986)) and important source of calcium and other elements (Fritsch et al., 2010;Jankovska et al., 2009;Sawicka-Kapusta et al., 1990). It has been found in Spanish population of the Montagu's Harrier that the proportion of mammalian prey, that is, rabbits in the prelaying diet was significantly related to breeding performance, with higher nest success and productivity in years when diet was dominated by hares (Arroyo & Garcia, 2006). Thus, it cannot be excluded that deterioration of reproductive performance (significant decrease of mean egg volume, mean clutch volume per female, mean dimension of eggs, and number chicks hatched) and increased (up to 75%) brood losses (Wiacek, 2015) leading to total disappearance of the breeding population in 2016 (Buczek & Buczek, 2017) were at least partially affected by changes in diet.    (Kitowski, 2005).

| Comparison to other breeding populations
Parental birds encourage fledglings to leave the fens and fly to neighboring arable lands abundant in grasshoppers. This prey item was the first one self-hunted by the juveniles (Kitowski, 2005).

| Limitations of the study
We are aware of some limitations of our study. We based our study on pellets produced by unknown number of individuals. Comparison of various techniques to study diet of the Hen Harrier (Circus cyaneus) revealed that pellets over-represent mammalian prey and under-represent avian prey; however, they are useful for estimating prey diversity and the frequency of certain prey types in the diet (Redpath et al., 2001).  Note: All studies based on pellet analyses. Values > 50% bolded. Data sources and location: see Figure 10 caption.

TA B L E 6 Summary of Montagu's
Harrier main prey item abundance (%) in various areas study temporal changes in abundance of the most important prey categories. Our study has filled an evident gap in knowledge about within-and between-season diet variation in the Montagu's Harrier.

| CON CLUS IONS
We found that the diet of Montagu's Harrier breeding in calcareous peat bogs (Chełm, Eastern Poland) and foraging in nearby agricul- to respond to changes in foraging habitats and prey composition by opportunistic foraging on easily available prey. However, they probably were not able to fully compensate lower contribution of important diet components (e.g., energy-rich voles) which may have resulted in deterioration of reproductive performance and finally in total collapse of the studied population in 2010s.