Does anthropization affect physiology, behaviour and life‐history traits of Montagu's harrier chicks?

The last century has seen a steep decline in biodiversity, and anthropization is considered one of the major drivers of this decline. Anthropogenic disturbances, due to human presence and/or activities, may be perceived as chronic stressors by wildlife and potentially lead to deleterious effects on traits related to fitness. The main objective of the present study was to highlight the effects of these anthropogenic elements on wild birds on sparsely urbanized farmland, far less studied than in urbanized areas. We investigated during four successive breeding seasons whether the anthropization level, assessed by infrastructure density around nests, and the harvesting conditions around nests may impact physiological, behavioural and life‐history traits of Montagu's harrier Circus pygargus chicks. Higher anthropization levels were associated with higher basal corticosterone levels in nestlings during only one breeding season and a lower body condition at fledging for females, probably because they suffered from higher starvation than males. Nestlings reared in more anthropized areas or in harvested crops before their fledging harboured more fault bars on rectrices than those reared in less anthropized areas or in unharvested crops regardless of year and sex, which is suggestive of higher stress during development. Nestling behaviours were also impacted by anthropization level and harvesting conditions: chicks in harvested crops were more aggressive and in areas with higher anthropization levels more prone to escape than others. Because Montagu's harrier is a protected species, the impacts highlighted in the present study are a matter of concern, especially regarding farmland landscape modifications, and we advise limiting perturbations in areas where Montagu's harriers usually nest.


Introduction
In the last century, the rate of biodiversity loss has run at an unprecedented pace (Pievani, 2014;Ceballos et al., 2015). Birds are no exception, especially in agricultural areas (Inger et al., 2015;Stanton, Morrissey & Clark, 2018;Rosenberg et al., 2019). In Europe, 55% of farmland birds showed population declines between 1980 and 2006 (Voříšek et al., 2010). One of the major causes of this global biodiversity decline is the anthropization process due to the human population increase (Swanson, 1998;McKinney, 2002;Maxwell et al., 2016). Anthropization includes the reduction of natural areas and land use changes, that is the conversion of natural and semi-natural habitats into farmland (Ellis et al., 2010;Lai, Leone & Zoppi, 2017). It also includes disturbances due to human presence, human activities (e.g. cars, tractors), humaninduced noises (e.g. road traffic) or human infrastructures (e.g. high-voltage powerlines), all of which may be perceived as stressors by wildlife (Frid & Dill, 2002;Tarlow & Blumstein, 2007;Almasi et al., 2015), and are thus also causes of biodiversity loss (Maxwell et al., 2016).
Stressful stimulus activates the hypothalamic-pituitaryadrenal axis (HPA) with an increase in the level of plasma glucocorticoids (Ellis, McWhorter & Maron, 2012). In birds, this activation involves the secretion of corticosterone (hereafter CORT) into the blood by the adrenal gland, that is stress-induced CORT (Siegel, 1995;Sapolsky, Romero & Munck, 2000;Romero, 2004). In the case of acute stress, for example an individual facing a predator attack, the CORT level returns to its basal state after a few minutes to several hours, that is homeostasis (Wingfield et al., 1998). Such a pattern is considered adaptive because the short-term rise of CORT secretion allows individuals to reallocate energy and behaviour from normal activities to survival mode (Frid & Dill, 2002;Müller, Jenni-Eiermann & Jenni, 2009;Dantzer et al., 2014), by increasing blood glucose levels, regulating immune system function or suppressing reproductive behaviour (Wingfield et al., 1998;Romero, 2004;Romero & Butler, 2007). Conversely, chronic stress results from persistent or repeated stressors where HPA remains overstimulated, which results in continuous CORT secretion leading to an increased basal CORT level (Wingfield et al., 1998). Desensitization of the receptors of the physiological response may be observed (e.g. dysregulation of negative feedback, downregulation of hormone production) in which case the basal CORT level decreases (Cyr & Romero, 2009). However, chronic stress is likely to incur fitness costs such as physiological, behavioural and/or morphological alterations (Romero & Butler, 2007;Cyr & Romero, 2009).
Studies assessing the effects of direct human disturbances on wild birds are usually conducted in moderate (suburbs) to highly urbanized areas (Halfwerk et al., 2011;Strasser & Heath, 2013;Glądalski et al., 2016) or in areas more prone to recreational activities (González et al., 2006;Ellenberg et al., 2007;Remacha et al., 2016). However, farmland which seems sparsely anthropized compared to cities and suburbs also experiences human activities, such as car traffic, walkers, joggers, cyclists, agricultural activities and anthropogenic noise from villages. Therefore, farmland specialist bird species, which could be less flexible and tolerant than generalists, may be more sensitive to these disturbances, even if they occur at low frequencies (Wretenberg et al., 2006;Filippi-Codaccioni et al., 2008;Heldbjerg, Sunde & Fox, 2018;Giralt et al., 2021). The Montagu's harrier Circus pygargus is a ground-nesting raptor that may breed in seminatural and extensive farmland habitats with low urban density (e.g. urban density <4% in the province of Lleida, Catalunya, north-east of Spain and in the meadows around Chelm, Poland respectively, Kitowski, 2003;Guixé & Arroyo, 2011). However, since the 20th century, this species breeds mainly in intensive farmland habitats, cereal crops, and areas surrounded by human infrastructures (e.g. 9.8% urban density in our study area, Zone Atelier Plaine et Val de Sèvre, France;Bretagnolle et al., 2018). Their chicks are thus exposed to anthropogenic disturbance, either directly or indirectly (Arroyo, García & Bretagnolle, 2004). Moreover, for the majority of nests, harvesting occurs during the rearing period before the chick fledge, though subjected to year-toyear variation. After harvesting, c. 4 m 2 of unharvested crops remain around the nest, and thereby the chicks become more visible from outside the plot and thus more exposed to human disturbance (Rabdeau et al., 2021) and predators such as foxes in our study area (Arroyo, Mougeot & Bretagnolle, 2001;Bravo et al., 2020Bravo et al., , 2022. Our study followed an approach to capture the global effects of direct human disturbances on Montagu's harrier chicks. Assuming that infrastructure density around nests would be positively correlated with the intensity of direct human disturbances (see Benítez-López, Alkemade & Verweij, 2010), we specifically investigated whether this anthropization level, measured by the density of human infrastructure around the nest, affects the physiology, behaviour and body condition of Montagu's harrier fledglings. We also assessed whether harvesting conditions of the crop plot (i.e. crop harvested or not) where the nest was located, impacted the fledglings depending on the density of human infrastructure around the nest. We conducted a multi-trait approach and predicted that chicks from nests located in more anthropized areas would present higher basal CORT and stressor-induced CORT levels associated with a lower body condition and more fault bars than chicks reared in less anthropized areas. According to the results from a previous study in which repeated exposure and handling of chicks by the same experimenters caused a sensitization phenomenon (Rabdeau et al., 2019), we hypothesized that chicks reared in more highly anthropized areas and/or in harvested crops would be more stressed, and thereby more active and aggressive than others.

Ethics statement
All the birds involved in the present study were released at their site of capture (i.e. their nests) after each handling. Bird manipulation was permitted and licensed by the CRBPO (Centre de Recherches sur la Biologie des Populations d'Oiseaux -Museum National d'Histoire Naturelle, licence #1308). The methods used for the capture, handling, banding and blood sampling of the birds comply with French guidelines for the ethical use of animals in research.

Study site and biological model
The study was conducted during four successive breeding seasons (2016)(2017)(2018)(2019) in Western France, within a Long-Term Social-Ecological Research (LTSER) area; specifically, the Zone Atelier Plaine & Val de Sèvre (ZAPVS thereafter, GPS coordinates 46°110 N, 0°280 W, see nest distribution in Supporting Information Figure S1). This 435 km 2 area is managed principally for intensive farming but 9.8% of its surface is urbanized including villages, industrial and agricultural buildings, railways, a motorway, roads and paths (Bretagnolle et al., 2018). Since 1994, the population of Montagu's harriers has been monitored each breeding season (Bretagnolle et al., 2018). This migratory farmland raptor nests on the ground in tall vegetation, especially on cereal crops (Millon et al., 2002;Gillis et al., 2012). Females lay asynchronously up to eight eggs and c. four eggs on average in our studied population, with an average lag of 2 days between successive eggs, depending on food availability (Arroyo et al., 2004;Millon, Arroyo & Bretagnolle, 2008). However, the eggs hatch with an average lag of 1 day. During the rearing period (30-35 days), the chicks depend on their parents for food, thermoregulation and predator protection. Montagu's harriers are local specialist predators; they prey mostly on a small cyclic rodent, the common vole Microtus arvalis in France and shift to insects, such as grasshoppers Tettigonia viridissima, in low-vole years (Arroyo, 1997).

Nest monitoring and measures
Nests were systematically searched for during the whole breeding season . Each nest was recorded by its GPS coordinates. The nests were visited once during incubation to measure the eggs (length and width using a calliper, accuracy AE0.1 mm; mass using Pesola ® electronic scale, accuracy AE0.1 g) and to calculate egg density (i.e. egg mass divided by volume; egg volume = 0.51 × length × width 2 /1000; Hoyt, 1979) in order to estimate the hatching date (egg density decreases linearly from laying to hatching date) and the laying date to plan nest visits during the rearing period (Arroyo et al., 2004;Arroyo, Mougeot & Bretagnolle, 2017). From hatching to fledging, the nests were visited once a week. At 15 AE 2 days old, the chicks were ringed (metal numbered ring from the Muséum National d'Histoire Naturelle) and sexed according to iris colour: brown for females and grey for males (Leroux & Bretagnolle, 1996). Behavioural and physiological data were collected before fledging when the chicks were 26 AE 2 days old. females and 129 males (no sex ratio bias). For four nests, it was not possible to estimate the laying date from egg measures because the nests were discovered after hatching. Therefore, for these four nests, we calculated retroactively the laying date from the date of nest visit with chick measurements: we subtracted the chick age (26 days old) at this nest visit and the incubation duration (30 days; García & Arroyo, 2001) to estimate the laying date. At the end of the breeding period, a final nest visit was conducted to note the brood size at fledging.
Each nest visit when the chicks were 26 AE 2 days old followed the same procedure. After a quiet approach, the chicks were captured as soon as the experimenters arrived at the nest. The first blood sample was collected in the first 3 min of handling (Sample A). Then, the right and left tarsus lengths were measured twice using a digital calliper (accuracy AE0.1 mm) and the nestlings were weighed using a spring scale (Pesola ® 500 g, accuracy: AE5 g) to estimate their body condition. The second blood sample (Sample B) was taken 15 min after the capture. The blood samples were kept refrigerated (0-5°C) for further analyses (see below). Before the release of the chicks at their nest, a photo of their tail was taken to count the number of fault bars on the rectrices (see Supporting Information Figure S2). The behaviour of each chick was recorded throughout the handling (see below). At this nest visit, we also noted whether the crop where the nest was located (harvest thereafter) was unharvested (0) or harvested (1).

CORT analyses
Once back at the laboratory, the blood samples were centrifuged (10 min at 3880 g, Bio Lion XC-LED12K) to separate the plasma from the erythrocytes. The plasma samples were stored at −20°C until titrated by radioimmunoassay to obtain their CORT concentration (Lormée et al., 2003). The intra-assay and inter-assay coefficients of variation were 10.57% and 13.72%, respectively, for the 3 years of the CORT analyses (2017-2019). The basal CORT level was obtained from Sample A. Two stressinduced CORT levels were used: the maximal CORT level, that is the CORT concentration from Sample B, and the change in CORT, that is subtracting the basal CORT level from the CORT level of Sample B (Rabdeau et al., 2019). These two measures (maximal CORT level and change in CORT) provided different information on stress-induced CORT. The maximal CORT level represented the maximum value of CORT experienced by chicks during handling and the change in CORT represented individual plasticity in the stress-induced CORT. In 2016, no CORT analysis was done. This study was conducted under natural conditions within a 435 km 2 area, thus it was not possible to sample all chicks at the same time of the day because of the weather conditions and logistics. We tried to visit the nests in the morning: 106 chicks were sampled before 1 PM, 22 chicks were sampled after 3 PM and for 4 chicks, we did not have the time of the nest visit. The time of the nest visit had no effect (morning vs. afternoon) on the corticosterone levels (Linear Mixed effects Models [LMM], basal CORT: χ 2 = 2.25, d.f. = 1, P = 0.13; maximal CORT: χ 2 = 0.18, d.f. = 1, P = 0.68; change in CORT: χ 2 = 0.01, d.f. = 1, P = 0.91).

Fault bars
The number of fault bars were counted from the chick tail pictures by the same experimenter to avoid potential measurement biases. This study is the first general approach to the global effects of human disturbances on different traits of nestlings, and therefore, we chose to count all kinds of fault bars. After a careful examination, all breaks in barbes and barbules from the finest to the widest (i.e. light, moderate and severe fault bars, see the review by Jovani & Rohwer, 2017) were counted on the most affected feather (Supporting Information Figure S2).

Behaviour
Chick behaviour was assessed by scoring specific behaviours (Rabdeau et al., 2019). First, a general activity score was measured when the experimenter approached the nest and captured the chick (activity capture: act capt ): 0 = no reaction, 1 = it moved backwards or 2 = it ran away. During capture, whether the chick uttered a call (1) or not (0) was noted (vocalization capture: voc capt ). Subsequently, another activity score was tabulated during handling (activity handling: act hand ): 0 = motionless chick, 1 = low reaction chick, 2 = medium reaction chick or 3 = chick was continually in motion. The rate of claw attack (activity claw: act claw ), beak attack (activity beak: act beak ) and calls (vocalization handling: voc hand ) throughout the handling period were recorded (absolute frequency/handling time in min). The behaviours of the chicks were not recorded in 2016. Among and within each year, the chicks were not systematically handled by the same person, but all the experimenters were trained to measure and score the chick behaviours (see Rabdeau et al., 2019).

Proportion of infrastructure around the nest
The GPS coordinates of the nests, as well as the different categories of human infrastructures georeferenced in the BD TOPO v.3 ® (derived from the internal database of the Institut National de l'Information Géographique et Forestière), were incorporated into the Geographic Information System (GIS) database of the LTSER, using QGIS (version 3.4.12). BD TOPO allowed us to distinguish between a highway, onelane road, two-lane road, roundabout, gravel road, path and walkway. All kinds of buildings with a minimum footprint area of 50 m 2 were integrated, including sport fields and cemeteries, as well as transport infrastructures (trainlines, motorways, roads, paths) and high-voltage powerlines. Such linear elements were transformed into polygons with respect to the width of each element. If not referenced in the BD TOPO, an average width was estimated using Google Maps (five measurements were taken per group of elements characterized by the same attributes, e.g. highway, one-lane road, two-lane road, roundabout, gravel road, path, walkway). We chose to quantify the anthropization level around each nest by calculating the cumulated surface area covered by all types of infrastructures instead of considering each infrastructure separately. Indeed, the effects of each infrastructure type cannot be disentangled from each other in a field study without any experimental modifications (which would have been questionable for our protected and declining model species). Our aim was not to identify the major components of disturbance but to realise an integrative assessment of the cumulative effects of all infrastructure types. The infrastructure density around the nest (IDN hereafter) was calculated within a 1000 m radius buffer (see Gormally et al., 2021 for a similar quantification). At this study site, the median distance from the nests to their nearest infrastructure was 530 m (Rabdeau et al., 2021). Thereby, by using twice this distance, we expected to obtain variability in IDN at the local scale. Additionally, a meta-analysis previously evidenced that infrastructures may impact a bird population up to 1000 m away (Benítez-López et al., 2010), supporting our choice of buffer radius.
One nest (two female chicks) had a particular situation and was above 5% of IDN within the 1000 m radius buffer (Supporting Information Figure S3). As this nest (in 2016) would behave as a statistical outlier, it was excluded from all the analyses, but as a quality check, we also conducted the analyses with this nest included and show the results in the Supporting Information (Table S1). Calculated excluding this outlier, IDN ranged from 1.23 to 4.83%, with no statistical variation among the years (Supporting Information Figure S3; Kruskal-Wallis test: χ 2 3 = 2.45, P = 0.48). Although the IDN may seem low, there was IDN variability among the nests for the next analyses: the maximum value of IDN (4.83%) was almost four times the minimum value of IDN (1.23%; see Supporting Information Figure S4). We provided pictures of the habitat within the 1000 m radius buffer around the nests with IDN of 1.23%, 3.0012% and 4.83% using our GIS database and the orthophoto of our study area in Supporting Information ( Figure S5). The nests with 1.23% IDN could have less stressful anthropized surroundings than the nests with 4.83% IDN because of fewer surrounding human activities, such as car traffic, walkers, joggers, cyclists, agricultural activities and anthropogenic noise from the villages (Supporting Information Figure S5). Assessing the effects of human disturbances in sparsely urbanized landscapes was one of our objectives and defined the specificity of this study.

Morphology and behavioural indexes
Body condition was estimated based on mean tarsus length (right and left) and mass by calculating the Scaled Mass Index (SMI), following Peig & Green (2009), using the 'smatr' package (Warton et al., 2012). Briefly, for each chick (i), SMI is defined as where b sma is the slope of the major axis regression of log(mass) on log(mean tarsus length). A principal component analysis (PCA) was performed on all behavioural data (act capt , act hand , act claw , act beak , voc capt and voc hand ) to avoid multiple tests (and also because the behavioural variables were moderately to highly correlated with each other, see Supporting Information Figure S6). Of the six axes, we retained the first three as their standard deviations were above 1 for a total of the cumulative proportion of variance of 74.50% (Table 1). On the first axis of the PCA (PC1), act hand , act beak and act claw were the major contributors and represented the behaviour of the chicks during handling on the second axis (PC2), voc capt and voc hand represented overall vocalization activity and on the third axis (PC3), act capt contributed mostly negatively ( Table 1).

Effects of anthropization
The effect of IDN was tested on basal CORT and the two stress-induced CORT levels (both square root-transformed to meet normality and homoscedasticity assumptions), body condition (log-transformed), PC1, PC2 and PC3 with LMMs (from lmerTest package, Kuznetsova et al., 2017). The two stress-induced CORT levels (maximal CORT level and change in CORT) were highly correlated (Spearman's rank correlation coefficient: ρ = 0.82 and P < 0.0001; see Supporting Information Figure S6); therefore, we chose to present the results on the maximal CORT level in the main text and the change in CORT in Supporting Information (Table S2). Generalised LMM (GLMM) fitted with Poisson distribution was used for the fault bars. In all models (LMMs and GLMM fitted with Poisson distribution), IDN, year, sex, harvest, laying date (Julian calendar), brood size at fledging, IDN × year interaction, IDN × sex interaction and IDN × harvest interaction were included as fixed effects. Nest identity was always included as a random effect to control for the non-independency of chicks from the same nest. Model comparisons were performed using likelihood ratiobased χ 2 statistics to estimate the statistical significance of the explicative variables (from the 'car' package, Fox et al., 2012). We chose to present full models with all fixed effects following Forstmeier & Schielzeth (2011). Significant effects were tested using posthoc tests based on least-squares means if necessary (using the 'emmeans' package, Lenth, Singmann & Love, 2018). The sample sizes varied depending on the traits because some data were unavailable. On the individuals for which all variables were available, we also performed Spearman's rank correlation tests to assess the potential links among the variables (Supporting Information Figure S6). All statistical analyses were run using R software (v 4.0.4, R Core Team, 2021).

Results
The  Fig. 1a). The interaction between IDN and sex, the interaction between IDN and harvest and the main effects of sex, harvest, laying date and brood size at fledging did not influence the basal CORT (Table 2). Maximal CORT level did not vary depending on the interaction between IDN and year, the interaction between IDN and harvest and the main effects of the year, harvest, laying date and brood size but tended to vary depending on IDN interacting with sex (  Fig. 1b). The interaction between IDN and sex influenced the body condition of the chicks (Table 2). Female body condition slightly decreased with IDN (−0.023 AE 0.015 [−0.053; 0.0067]), whereas no effect was detected for males (0.0089 AE 0.016 [−0.023; 0.041]; Fig. 2). The body condition of the chicks was neither influenced by the interactive effect of IDN and year, nor by the interaction between IDN and harvest, or by the main effects of the year, harvest, laying date and brood size (  Fig. 3b). Laying date and brood size did not influence the number of fault bars observed on chicks' rectrices (Table 2). Concerning chicks' behaviour, neither their behaviour during handling (PC1) nor their vocalization (PC2) or escape behaviour at nest approach (PC3) were influenced by the three interactive terms of IDN with year, IDN with sex and IDN with harvest (Table 2). Chicks from nests in harvested crops were more active and attacked more during handling (−0.16 [−0.41; 0.093]) than  Figure S8).

Discussion
Conservation physiology has recently emerged, emphasizing the urgent need to understand the multidimensional processes and effects of human-induced perturbations and animal population declines through processes altering the fitness of individuals (Cooke et al., 2013(Cooke et al., , 2020. In the present study, we investigated whether even low anthropization levels in rural landscapes, estimated by IDN around the nests, may impact the development of Montagu's harrier chicks which also suffer an additional effect of harvesting. Based on different indexes related to physiology, morphology and behaviour that were measured for chicks just before fledging, we found some links between anthropization levels, harvesting conditions and chick development.     Anthropization and associated activities, such as human presence, noises and traffic, are considered chronic stress as chicks reared in more or less perturbated areas are constantly exposed to these unnatural stimuli (Dantzer et al., 2014). Individuals exposed to chronic stress are expected to exhibit higher basal CORT and increased stressinduced CORT although this common thought is challenged, thus questioning this general rule and suggesting potential publication bias (see the review by Dickens & Romero, 2013). In the present study, we found only slight evidence for increased basal CORT levels and only for the year 2019. Other studies also report an increase in basal CORT levels in different species (Ellenberg et al., 2007;Busch & Hayward, 2009;Wingfield & Romero, 2001;Dickens & Romero, 2013;Almasi et al., 2015). This higher basal CORT level in 2019 could be explained by IDN differences among the years with higher IDN found in 2019, but there were no statistical differences in IDN between the years. Therefore, it is difficult to draw conclusions on the relationship between basal CORT and anthropization. On one hand, the favourable and therefore less competitive conditions experienced by this harrier population during 2019  (higher vole abundance) may have allowed all pairs, even lower quality pairs, to raise their chicks to fledge irrespective of their sensitivity to disturbances, resulting in lower nesting failures that year (52% of the nests failed in 2016, 63% in 2017, 40% in 2018, 8% in 2019, data not shown). Moreover, these lower quality pairs could have settled in lower quality areas with higher IDN, by default of space and/or experience, thereby explaining the increased basal CORT of the chicks in areas with higher IDN. These conditions and hypotheses may have contributed to conserving large variability in basal CORT levels and thereby detecting the subtle physiological effects of anthropizationalthough this was also the year with the lowest variability for IDN. This hypothesis is plausible since the probability of nesting failure may increase with increasing anthropogenic stressors perceived by the parents (White & Thurow, 1985;Arroyo & Razin, 2006;Strasser & Heath, 2013). A previous study, conducted on the same population of Montagu's harriers, showed that shy females were less tolerant of disturbances than bold females (Rabdeau et al., 2021). Therefore, in most anthropized areas, the nests of less tolerant parents could fail before fledging during less favourable years in terms of food resources. The chicks measured at fledging in the more anthropized areas may consequently represent chicks raised by tolerant parents and who are themselves also relatively tolerant of human disturbances as personality traits are partly heritable (Cockrem, 2007;Dochtermann, Schwab & Sih, 2015). Accordingly, a reduction in variance associated with low statistical power due to the small sample size may have led to the absence of a detectable relationship observed in 2017 and 2018. On the other hand, finding no increase in basal CORT related to increasing anthropization levels is consistent with the literature (see the meta-analysis by Injaian et al., 2020). Additionally, the variation in basal CORT may depend on the matrix used to titrate glucocorticoid metabolite. Results from meta-analysis showed different trends between faecal and plasma samples: an increase is observed in faecal but not in plasma samples (Dantzer et al., 2014). Therefore, basal CORT from plasma is probably not affected by chronic anthropogenic stress (Dantzer et al., 2014;Injaian et al., 2020) and this finding could explain the small effect of the anthropization level depending on the year. While basal CORT levels were similar between sexes, the maximal CORT level tended to only be influenced by the anthropization level in males. In a previous study on Montagu's harrier nestlings, we did not find any difference in stress-induced CORT between sexes over repetitive manipulations by experimenters during nest visits (Rabdeau et al., 2019). This small anthropization level effect depending on the sex could be explained by a lack of statistical power due to low sample size for basal CORT. These low sex differences may be either suggestive of differential exposure to the stressor and/or of differing physiological constraints between sexes. The first hypothesis is unlikely because all chicks within a nest were exposed to the same perturbations whatever their sex. Interestingly, our results differ from the general trends that show a more important response to human perturbation in males than in females (Dantzer et al., 2014). Contrary to what was expected (see Kleist et al., 2018;Expósito-Granados et al., 2020 for a few recent examples), stress-induced CORT levels slightly decreased with increasing IDN in males. The ontogeny of stress is still poorly understood but it is known that it may, for instance, shape pre-migratory movements (Pakkala et al., 2016) and to a further extent adult physiology, behaviour and fitness (Schoech, Rensel & Heiss, 2011;Crespi et al., 2013). Additionally, CORT level differences between sexes in nestlings are rarely integrated into analyses (Tilgar, Saag & Moks, 2009;Wada et al., 2009;Rensel, Wilcoxen & Schoech, 2010;Tilgar et al., 2010;Kidawa et al., 2014;Pakkala et al., 2016;Newman et al., 2017;Injaian, Taff & Patricelli, 2018;Injaian et al., 2019;Bebus, Jones & Anderson, 2020;Expósito-Granados et al., 2020), or evidenced either no difference (Sockman & Schwabl, 2001;Beaugeard et al., 2019;Grunst et al., 2020) or low sex Figure 5 Effect of IDN on chick behaviour. The dots represent the observed values of PC3 for each chick. The line predicts PC3 depending on IDN, and the shading represents the 95% confidence interval. Note that behaviour at nest approach is the major contributing variable on this axis and is negatively correlated (see Table 1). IDN, infrastructure density around the nest. effects (Rensel, Wilcoxen & Schoech, 2011). This slight decrease in stress-induced response in males may arise from habituation (i.e. repetitive exposures lead to a change in the perception of the stimulus from noxious to innocuous), physiological desensitization of CORT receptors (decrease in the physiological response although the stimulus is still considered noxious) or exhaustion (decrease in energy allocated to the stress physiological system) (reviewed in Cyr & Romero, 2009). However, our data and study design do not allow identifying the processes explaining this slight difference in stress-induced CORT among sexes. Interestingly, female body condition was slightly negatively impacted by anthropization, whereas no effect in males was detected. Body condition at fledging is a fundamental lifehistory trait in migratory species as it determines later survival during migration (Duijns et al., 2017). This small decrease is congruent with the literature on chronic stress consequences (Dickens & Romero, 2013). However, no evident link with stress hormone can be made since basal and stress-induced CORT levels did not seem to be impacted by anthropization (see above), and only a slight correlation can be made between CORT levels and body condition. Alternatively, this result may be explained by a decrease in parental investment as human disturbances can decrease food item delivery by parents to both female and male nestlings (Fernández & Azkona, 1993). The diminution in the body condition of chicks as a result of human disturbances is already known in blue tits Cyanistes caeruleus (Remacha et al., 2016) but without difference between sexes. Contrary to blue tits, Montagu's harriers are sexually dimorphic, males being 15% lighter than females . This implies that females may require more energy to fledge (Teather & Weatherhead, 1988), similar to female American kestrels which consume 7% more food than their brothers (Anderson et al., 1993). It may also imply that females suffer higher starvation when food is scarce than male nestlings (Clutton-Brock, Albon & Guinness, 1985;Teather & Weatherhead, 1988). However, our hypothesis contrasts with the results from a previous study showing that the smaller sex (the male) has a higher probability of starvation in Montagu's harriers; because females are the bigger sex, they could be more competitive than males when consuming food items delivered by the parents (Arroyo, 2002). Taking into account this contrast among studies, the process explaining the lower body condition of female nestlings in areas with higher IDN could be independent of food delivery by the parents but due to other unexplored processes.
In our study area, most of the chicks had fault bars on their rectrices; our results showed that chicks in the most anthropized areas had more fault bars than chicks in less anthropized areas. Moreover, a higher number of fault bars was found on chicks in harvested crop plots than on those in unharvested plots. Fault bars may appear consecutively to chronic or acute stress related to CORT secretion (Jovani & Rohwer, 2017). In European starlings Sturnus vulgaris for instance, individuals experimentally stressed (chronically or punctually) had more fault bars than unstressed individuals (Strochlic & Romero, 2008). Potentially, the increase in fault bars development in our area could be caused by a variety of stressors related to human disturbances. Moreover, after harvesting, only 4 m 2 of crops around the nest remain unharvested. Therefore, nests become more visible and we cannot exclude that neighbouring inhabitants and/or farmers may visit nests. This increase in fault bars could directly result from human disturbances (e.g. nest visits by intruders such as walkers or farmers) but also indirectly due to random punctuated reduction/absence in parental care due to disturbances. For instance, a smaller and less predictable food supply and/or decrease in time spent by the parents around the nest are stressors that can be caused by human disturbances and can trigger fault bar formation in nestlings (Strochlic & Romero, 2008;Rensel et al., 2010;Fokidis et al., 2012). The development of fault bars seems closely related to the stress experienced by the chicks. Our results highlight that their presence could be a good indicative marker of perturbations induced by human activities although further investigations are needed to properly test the hypothesis of the cascading effects of human disturbances on parental care and thus on fault bars of chicks (Jovani & Rohwer, 2017).
Finally, the infrastructure density around the nest and the harvesting conditions also affected chick behaviour. Chicks from nests in harvested crops were more active and aggressive towards the experimenter during handling than those in unharvested crops. Moreover, chicks from nests located in areas with higher anthropization levels displayed more escape behaviour than those in areas with lower infrastructure density around the nest. As for fault bars, higher anthropization levels and harvested crops could increase human disturbances and its perception around the nest and thus increase stress for chicks. Disturbed and stressed chicks could be more aggressive and agitated. These results are consistent with a previous study in which a sensitization phenomenon increases the stress, activity and aggression of chicks over repeated handlings by the same experimenters (Rabdeau et al., 2019). These effects on chick behaviour suggest that higher levels of anthropization and harvesting conditions may be linked to repetitive exposure to the same stimulus (e.g. traffic noise, human visitation). Further studies are needed to properly measure these different stressor stimuli, such as human visitation around the nest, and to test the hypothesis of sensitization.

Conclusion
The present study showed the interest in considering several indexes for establishing how anthropization in the rural landscape may trigger chronic stress for wild fauna. Different physiological, behavioural and life-history traits of Montagu's harrier nestlings were affected by the harvesting conditions and the IDN around the nest and related human activities, although the observed anthropization level was low. Nests with 4.83% IDN seemed to have more stressful anthropized surroundings than nests with 1.23% IDN, because it implied more car traffic, walkers, joggers, cyclists, anthropogenic noise from villages and possibly more nest visits by neighbouring inhabitants. Moreover, nests in harvested crops were more visible Animal Conservation 26 (2023)  from outside the crop plots, that is from paths and roads, and thus could be more exposed to human disturbances. These effects may compromise the recruitment of fledglings in the population through altered survival during migration, due to a decrease in flight ability (fault bars) or low body condition. CORT secretions are known to produce long-term effects on bird quality in adulthood (Blas et al., 2007;Wada & Breuner, 2008;Schoech et al., 2011;Blas, 2015) and may compromise the persistence of some harrier populations. Because Montagu's harrier is a protected farmland raptor, the impacts of the anthropization level and harvesting conditions on nestlings are a matter of concern for the conservation of this species and possibly also for other ground-nesting farmland birds. Further studies are needed to accurately measure the human visitation (type and frequency) around nests and their impacts on chick development: agricultural activities on fields, walkers, joggers, nest visitation by neighbouring inhabitants, by using camera traps, for example. These camera traps would also allow for the measurements of nest visitation by predators. Conservation measures could be thus proposed to farmers and include a delay of the harvest until the fledging as already recommended by Arroyo, García & Bretagnolle (2002), and/or a larger area of unharvested crops around the nest. These measures should be considered a matter of urgency because harvesting occurs earlier and earlier every year due to global warming (see Fatima et al., 2020). Considering that even low levels of anthropization on farmland may have various consequences on birds and could contribute to their decline is vital information and should therefore be considered in the management of all areas where competing uses between groundnesting birds and humans intersect (Cooke et al., 2020).

Supporting information
Additional supporting information may be found online in the Supporting Information section at the end of the article. Figure S1. Spatial distribution of nests in our study area, Zone Atelier Plaine et Val de Sèvre. Figure S2. Fault bars on rectrices of Montagu's harrier chicks: (a) light to moderate fault bars and (b) moderate to severe fault bars (see review by Jovani & Rohwer, 2017). Figure S3. Infrastructure density around the nest by year. Figure S4. Distribution of nests in relation to IDN ranging from 1.23% to 4.83%. Figure S5. Different IDN scenarios within the 1000 m radius buffer (red) around nests (blue): (a) 1.23%, (b) 3.0012% and (c) 4.83%. Figure S6. Spearman's rank correlation coefficients for all behavioural, physiological and life-history traits included in the present study (115 chicks, 58 nests). Figure S7. Number of fault bars by years. Figure S8. Differences in behaviour between years for PC3 are mostly represented by the behaviour at nest approach. Table S1. Summary of models ( a Linear Mixed Models, b Generalised Liner Mixed Models, with Poisson distribution) investigating variation for body conditions and fault bars including the nest excluded from the analyses in 2016 for the 1000 m-radius. Table S2. Summary LMM investigating variation for changes in CORT.