Wintering Red Kites in Central Spain: Macrohabitat Selection and Population Density Estimate

Simple Summary

The Red Kite is a bird of prey found mainly in Western Europe. Once common in Spain, its numbers have declined sharply over the past two centuries, and by the late 1900s, it had nearly disappeared as a breeding species. While conservation projects have supported the recovery of this species, identifying its wintering locations remains crucial for effective protection. We studied Red Kites in and around the La Mancha Húmeda Biosphere Reserve, a wetland and farmland area in Central Spain. Using over 300 vehicle surveys along both paved roads and dirt tracks during the winters of 2021–2022 and 2022–2023, as well as the 2022 breeding season, we estimated Red Kite density through Distance sampling and compared habitat types. We found no difference in density or detectability from roads versus tracks. In total, 124 birds were seen in winter, but none were observed during the breeding season. This suggests that the reserve hosts a wintering population of around 1400 kites. Spain hosts ~50,000 wintering Red Kites, nearly three-quarters of the European population, with the study area holding 5% of this total. Red Kites favor diverse farmlands over rugged areas, making habitat protection and improvements in heterogeneous agricultural landscapes essential for their recovery in Spain.

Abstract

The Red Kite (Milvus milvus), a Western Palearctic raptor, experienced a severe population decline across its range during the 19th and 20th centuries, nearly disappearing as a breeding species in Spain. Although conservation and reintroduction efforts have facilitated recovery, information on its wintering ecology remains limited. We evaluated the population size and habitat selection of wintering Red Kites in and around the La Mancha Húmeda Biosphere Reserve (MHBR), Central Spain, and assessed whether survey route type affected detectability. Surveys comprised 325 vehicle-based line transects along paved roads and dirt tracks during the winters of 2021–2022 and 2022–2023, and the 2022 breeding season. Detection rates and density did not differ between route types. Across 212 winter transects, we recorded 124 individuals, with none detected during 113 breeding season transects. Density estimates were consistent across winters (1.3–1.4 individuals/km²), yielding a population estimate of approximately 1430 individuals. Habitat analysis indicated clear selection for open agricultural mosaics dominated by fallow land and vineyards, while rugged terrain was avoided. These results identify MHBR as an important wintering area. Yet, super-intensive agriculture is a poor habitat due to its uniformity. Enhancing heterogeneous open landscapes, including farmland, is essential for Red Kite colonization and expansion.

1. Introduction

The Red Kite (Milvus milvus) is a bird of prey endemic to the Western Palearctic [1,2]. Since the 19th century, this species has experienced a significant population decline, disappearing as a breeding species in parts of its original range. Populations declined until the 1970s mainly due to persecution and have experienced renewed declines since the 1990s, driven by the ingestion of rodenticide-treated baits, illegal poisoning, and changes in agricultural practices, particularly within its core breeding range of Spain, France, and Germany [3,4]. In Spain, although not extinct, the Red Kite also suffered a marked decline toward the end of the 20th century [5,6,7]. However, several bird species that experienced declines are now showing signs of recovery in Spain [8,9,10].

For the Red Kite, reintroduction programs have been successful [11], with released populations establishing and increasing in size in the UK since 1989, parts of Italy, and the Republic of Ireland and Northern Ireland since 2007 [12,13,14,15]. Over the past decade, the species has increased by more than 30%, and by over 80% in the last three generations, leading to its reclassification by the IUCN from “Near Threatened” to “Least Concern” [4,11]. The European breeding population is estimated at approximately 30,000 to 35,000 pairs across 17 countries, with core populations in Germany (12,000 to 18,000 pairs), Spain (around 4000 pairs), and France (2300 to 3000 pairs) [11]. Despite the population recovery in Northern Europe, threats such as poisoning and habitat loss remain significant concerns throughout this species’ range [11]. Variation in regional threats, habitat management, and conservation efforts largely explains the contrasting population trend across Europe, with northern and central countries experiencing more stable and recovering populations, whereas southern populations continue to face challenges [4,11,16].

In Europe, the Red Kite exhibits diverse migratory behaviors characterized by partial migration, with individuals adopting migration, sedentarism, or sedentarism combined with post-reproductive movements [17]. Migration routes generally extend from Central Europe to Southern Europe, mainly towards the Iberian Peninsula, Southern Italy, and the Balkans. Juveniles undertake longer and more variable migrations compared to adults, who show more consistent timing and routes [3,17]. In Spain, traditionally considered a sedentary population, recent tracking data reveal that about 10% of individuals migrate within the Iberian Peninsula, 70% are sedentary, and 20% display sedentarism with extensive post-reproductive wandering movements. These movement patterns vary annually between individuals and demonstrate behavioral plasticity influenced by genetic, social, and environmental factors [3,17,18].

While habitat preferences during breeding are well documented in key regions such as Germany [19] or Spain [20], wintering habitats are equally important for conservation [21]. Spain serves as a crucial wintering ground, hosting a significant proportion (~50,000 individuals) of the Western European Red Kite population [16], where the highest wintering concentrations occur in Castilla–León and Extremadura, Northern and Western Spain [22]. The population’s partial migratory behavior, ranging from local residency to short-distance intra-peninsular movements, reflects the species’ flexible use of heterogeneous landscapes under Mediterranean climatic pressures [17]. In addition, the wintering ecology of the species is beginning to be understood, with two spatial strategies recorded: most individuals in Spain remain within a single wintering area throughout the season, while around 20% move between two main areas [23]. Estimating population size and understanding habitat selection are fundamental for effective management, as populations rely on suitable environments and show habitat preferences [6,14,24,25,26,27,28]. Central Spain is considered a region with a relatively low probability of detecting wintering Red Kites [22]. In winter, the species typically selects flat, open landscapes dominated by agricultural mosaics at the macrohabitat level [29,30], while avoiding rugged terrain and woodland areas [31,32].

The La Mancha Húmeda Biosphere Reserve (hereafter MHBR) is a conservation unit encompassing temporary wetlands in Central Spain composed of flat and open landscapes. The inter-lagoon areas are dominated by rainfed crops, increasingly managed through irrigation [33,34]. Conservation efforts in the MHBR have mainly targeted aquatic bird species, which include habitat preservation, legal restrictions on hunting in designated zones, and management interventions addressing emerging threats such as the overabundance of wild boar, which negatively impacts aquatic bird productivity [35]. However, the surrounding agricultural landscapes and road networks may be important for wintering Red Kites, potentially supporting recovery efforts. So, the MHBR and its surrounding areas have the potential to support wintering Red Kites.

The aim of this study was to determine the wintering and breeding population size of the Red Kite in the study area (MHBR and surroundings), assess macrohabitat selection, and evaluate the influence of roads and tracks on detectability, given that previous research suggests roads may influence survey results [14,36]. We predicted that there would be no detections during the breeding season, that individuals would avoid rugged or wooded areas, and that they would primarily occur in flat regions with a high proportion of cultivated land [22,31]. Identifying areas selected by Red Kites during winter will provide the basis for targeted conservation measures within and around the MHBR, allowing management practices to promote the recovery of species afforded legal protection.

2. Materials and Methods

2.1. Study Area

The study was conducted in two distinct regions in Central Spain, both located within the Mediterranean biogeographic Region. The first region is located in the MHBR, which comprises approximately 95 protected wetlands located in the upper basin of the Guadiana River in Central Spain [37,38]. The geological, climatological, and hydrological characteristics of the MHBR are well documented [33,37,39]. From a climatic point of view, the MHBR is conditioned by its geographical location in the center of the Iberian Peninsula, exhibiting features of continentality due to its distance from marine influence (Figure 1;

Table 1. Geographic zones considered for the study of the Red Kite in Central Spain; mean temperature (°C) of 2022; total rainfall (mm) in 2022; number of municipalities per zone; extension area of the zone (km²); number of winter transects (two sampling years); number of breeding season transects (one sampling year, as no breeding population was detected during the first year) and number of total transects. * Notes that climatological data were obtained from the nearest meteorological station due to the absence of data for the specific locality (data obtained from [40]). Zones used by Red Kites are shown in bold.

Geographic Zone	Mean Temperature °C	Total Rainfall (mm)	Number of Municipalities	Area (km2)	Winter 1 + Winter 2 Transects	Breeding Transects	Total Transects
Landete *	14.7	445.8	9	728.24	16 + 16	16	48
Motilla del Palancar	15.2	310.4	8	494.01	12 + 12	12	36
Belmonte	13.5	183.2	8	450.00	22 + 14	14	50
Mota del Cuervo *	13.5	183.2	3	443.40	20 + 11	14	45
Pedro Muñoz *	17.4	274.6	5	1307.04	14 + 8	14	36
Quero *	17.7	168	4	574.96	25 + 14	32	71
Lillo *	18.7	340	6	1025.10	16 + 12	11	39
Total			43	5023.35	125 + 87	113	325

) [33]. The areas between the wetlands include patches of holm oak (Quercus ilex) and maritime pine (Pinus pinaster) woodlands, but the predominant land use is agriculture. Vineyards and cereal crops dominate, with additional patches dedicated to olive and pistachio cultivation [33,34]. The first studied region (MHBR) encompasses several municipalities included in nominated areas of Belmonte, Pedro Muñoz, Quero, Lillo, and Mota del Cuervo (Figure 1; Table 1). The second region (non-reserve region), referred to as Motilla del Palancar and Landete, also includes multiple municipalities (Figure 1; Table 1). Landete and Motilla del Palancar, close but outside of the MHBR, are located in an area transitioning between the Serranía de Cuenca, which is part of the Iberian System, and the La Mancha Plain, which belongs to the Southern Sub-Plateau. Landete remains within the Castilian Branch of the Iberian System, whereas Motilla del Palancar is not part of the Iberian System [33]. This second region presents more extensive forest cover interspersed with agricultural land, primarily dedicated to cereal.

2.2. Red Kite Sampling

Red Kite sampling was conducted by a single observer (JAG-D) during the breeding season (May–June) of 2022 and the winters of 2021–2022 and 2022–2023 in both study regions. Distance sampling surveys followed line transect counts performed from a vehicle along roads and tracks of variable length (Figure 2), according to established protocols [41,42]. Distance sampling relies on four key assumptions: (1) animals are distributed independently of the lines; (2) objects on the line are detected with certainty; (3) distance measurements are exact; and (4) objects are detected at their initial location [41,42]. We acknowledge that conducting distance sampling along roads and tracks may introduce biases because these linear features may not representatively sample the entire study area and could influence animal distribution and detectability [41,42]. These transects were based on paved roads connecting localities, as well as dirt (unpaved) tracks extending perpendicularly from these roads within each geographic zone (Table 1). Censuses were conducted on local, regional, and agricultural roads and tracks within the study area, avoiding highways and national roads to ensure consistent sampling conditions. These roads allowed for maintaining an approximate cruising speed of 30 km/h during the census, providing a standardized survey effort. The “trip” function available in the car was used to control for the distance covered, allowing us to account for variations in road length. Additionally, the road layout permitted safe stopping at the roadside to accurately measure the distance to observed individuals. For each individual observed (unlimited distance survey) during a transect, the perpendicular distance from the vehicle to the bird at the point of first detection was measured using a digital laser rangefinder (Himimi EF01G). This methodology ensured reliable data collection while minimizing potential disturbance and variability caused by traffic conditions. Censuses were not conducted on rainy or windy days to avoid weather conditions that could affect animal activity and detection probability. The starting point of each transect was selected randomly, and surveys were conducted between 9:00 AM and 12:00 PM, when the birds are most active [43]. Vegetation along transect routes was predominantly open, without dense roadside hedges, ensuring good visibility and no limitation to detectability during surveys.

A total of 125 different transects were surveyed. Most were repeated during the following breeding season and the second winter, resulting in 325 samples overall. Of these transects, 52 were conducted along paved roads and 73 along unpaved tracks. Transect length varied depending on the extent of the road or track, ranging from 3.4 to 18.3 km (mean ± SE = 6.2 ± 0.2 km). More than three-quarters of transects were shorter than 7 km (Figure 2).

2.3. Habitat Characterization

Each transect was characterized by its surrounding habitat. The habitat was defined based on the predominant type recorded approximately 1 km on each side of the transect, using a laser rangefinder at 100 m intervals along both sides. For example, a 5 km transect would have 100 reference points (50 on each side). Habitat composition was described as the proportional coverage of different habitat types based on the predominant habitat recorded at each point. This allowed us to estimate the percentage of land cover for each habitat within every transect [44]. Every distinct available habitat was taken into account, totaling 18 habitats (e.g., fallow land, reeds, poplars, trellised vineyards, pinewood). Additionally, in a subsample of MHBR areas (Pedro Muñoz, Quero, Lillo), the presence of rabbit carcasses was recorded during the two winter seasons, differentiating between roads and tracks.

2.4. Statistical Methods

To compare the presence/absence of Red Kites between winters within localities, Chi-squared tests were performed, except for Motilla del Palancar, where Fisher’s exact test was used due to the low expected frequencies [45]. The same procedure was applied to compare their presence/absence between paved roads and tracks. Because the abundance data did not meet assumptions of normality, differences between winters were assessed using Mann–Whitney U tests [45]. The same procedure was used to test if detection distances differed between paved roads and tracks, as the distribution of detection distances did not meet the assumptions of normality. A bivariate linear model was used to test the independence of abundance and transect length [45]. Given the absence of significant differences between paved roads and tracks, we combined samples from both types of environments in our analysis. Finally, the difference in the presence or absence of rabbit carcasses between roads and tracks in the subsample (Pedro Muñoz, Quero, Lillo) was tested using a Chi-squared test. All statistical analyses were conducted in PAST 4.0 [46].

Population density was estimated using Conventional Distance Sampling (CDS) methods implemented in the software Distance 8.0 [47]. This approach estimates detection probabilities based on perpendicular distances alone (Figure 3), allowing robust estimation of population size and density [38]. To define distance intervals for data grouping, several exploratory analyses were conducted (e.g., truncating the dataset at 400 m), after which distances were grouped into 125 m intervals [48], as these provided better model fits and improved the overall performance. Density estimates were generated using different key functions—hazard rate, half-normal, and uniform (

Table 2. Results of the adjusted Distance sampling models are presented for data grouped into 125 m intervals without adjustment terms. Key functions include half-normal, hazard rate, and uniform, with adjustment terms of simple polynomial and cosine. Post-stratifications were applied by transect type (road, tracks) and by winter (2022; 2023). Models are ordered according to Akaike’s Information Criterion (ΔAIC: difference in AIC relative to the best-fitting model). The number of parameters in each model is also shown. Bold marks the best model.

Model (Key Function/Adjustment Term + Pos-Stratification)	# Parameters a	ΔAIC	AIC	X2-p	p
Locality
Hazard rate/simple polynomial	2	0	248.98	0.002	0.16
Half-normal/cosine	2	9.598	258.58
Uniform/cosine	2	21.39	270.38
Transect type
Hazard rate/simple polynomial	4	3.18	252.16
Half-normal/cosine	4	13.58	262.57
Uniform/cosine	4	48.19	297.18
Hazard rate for roads		108.33	108.09	0.16	0.18
Hazard rate for tracks		144.25	144.08	0.02	0.14
Winter b
Half-normal/cosine	3	9.04	258.02
Uniform/cosine	4	60.55	309.53
Half-normal for Winter2022		146.64	146.58	0.11	0.19
Half-normal for Winter2023		111.7	111.44	0.00002	0.22

p probability of detection; X²-p probability for chi-square goodness-of-fit test; ^a AIC did not select adjustment terms; ^b models based on hazard rate did not converge.

)—with cosine and simple polynomial adjustment terms [47,48]. Model selection was based on goodness-of-fit tests and Akaike’s Information Criterion (AIC), with lower AIC values indicating more plausible models [49,50]. Because the same survey routes were used across two winters, post-stratification was applied to obtain seasonal and spatial estimates of population size and density [47].

Differences in habitat features (obtained as explained above) between the two regions (macrohabitats, MHBR vs. non-reserve region) were evaluated using a PERMANOVA test [51]. A SIMPER (Similarity Percentage) analysis was used to determine the contribution of each habitat type in each transect to the overall dissimilarity between macrohabitats (MHBR vs. non-reserve region), expressed as a percentage of total dissimilarity [52]. Note that the presence of rabbit carcasses was not included in these analyses. These two statistical analyses were conducted in RStudio 4.1.0 [53].

3. Results

3.1. Presence of the Red Kite

A total of 124 Red Kites were recorded across 325 transects (48.4% perched). No individuals were observed during the breeding season (n = 113 transects), so these data were excluded from further analyses. Additionally, no Red Kites were detected in the Landete area during both winters, and only a single individual was observed in the Motilla del Palancar area during the first winter. There were no significant differences in detections between the two winters in Motilla del Palancar (Fisher’s exact test, p = 1). Due to the absence of observations in Landete and the very low number recorded in Motilla del Palancar, abundance analyses were conducted for the five remaining areas: Belmonte, Mota del Cuervo, Pedro Muñoz, Quero, and Longar.

In the five localities (n = 156 transects), Red Kites were detected in 47% of itineraries when both winter periods were pooled. First, the presence/absence of Red Kites was compared between paved roads and tracks. Samples from both years showed no significant differences among the five localities (χ² = 2.65, df = 4, p = 0.14, ns). Detection distance between roads (n = 52) and tracks (n = 72) did not differ significantly (Road = 97.3 ± 2.4 m, Track = 109.0 ± 17.7, Mann–Whitney U = 0.74, p = 0.460). Similarly, abundances between the two road types did not differ significantly (Mann–Whitney U = 2632.5, p = 0.296). Comparisons between the two winters based on presence/absence data also revealed no significant differences (χ² = 0.001, p = 0.960), and abundances did not differ significantly between winters (Mann–Whitney U = 2814, p = 0.852). This pattern was consistent across all localities in winter. Additionally, the number of Red Kites per km was independent of itinerary length in both years (first winter: r² = 0.001, p = 0.68; second winter: r² = 0.005, p = 0.058). Comparisons of Red Kite presence/absence between localities showed significant differences, suggesting that some areas had a higher occurrence of the species (χ² = 18.3, df = 4, p = 0.001). No significant differences in presence/absence within localities were identified between winters (Belmonte χ² = 0.83, ns; Mota del Cuervo χ² = 0.53, ns; Pedro Muñoz χ² = 2.81, ns; Quero χ² = 0.05, ns; Lillo χ² = 1.19, ns).

3.2. Estimate Density of the Red Kite

The best-fitting model for estimating population density by locality was the hazard rate function with cosine adjustment (Table 2). Density estimates for each locality ranged from 0.12 to 0.62 individuals per km² in the MHBR (

Table 3. Estimated density of Red Kites per km in each geographic zone, with lower and upper density limits. The population estimate is shown, along with the corresponding lower and upper population estimates of Red Kites. The row “Total” gives the total number of Red Kites across the five areas, calculated by summing the values for each area. Surface area of each locality is provided in Table 1.

Geographic Zone	Estimated Density	Lower Density Limit	Upper Density Limit	Lower Estimate < Population Esimate < Upper Estimate
Belmonte	0.12	0.06	0.23	27 < 54 < 104
Mota del C.	0.35	0.22	0.57	90 < 155 < 253
Pedro Muñoz	0.35	0.19	0.63	249 < 458 < 824
Quero	0.62	0.43	0.91	247 < 357 < 523
Longar	0.40	0.24	0.66	195 < 410 < 677
Total				816 < 1434 < 2381

). Based on these densities and the area with confirmed Red Kite presence (3801.5 km²), the estimated wintering population is approximately 1434 individuals, with a range of 816 to 2381 birds (Table 3). Note that this total value was obtained by adding the estimates for each locality.

3.3. Habitat Characterization

Significant differences in habitat features between both macrohabitat regions (non-reserve region vs. MHBR region) were found in the PERMANOVA test (F = 23.93, df = 124, p < 0.001). According to the SIMPER tests, only five habitats (Ploughed with winter cereal, Trellised vineyard, Fallow land, Holm oak woodland, Goblet-pruned vineyard) were responsible for more than 70% of the variation (cumulative contribution) in habitat types between regions due to the mean abundance differences in % (

Table 4. Results of the SIMPER analysis showing the habitat types contributing most to the dissimilarity between the two macrohabitat regions (non-reserve vs. MHBR). Average—Contribution of each habitat to the average between-group dissimilarity; SD—Standard deviation of the contribution; Ratio—Average-to-standard deviation ratio; avNon-Reserve, avMHBR—Average abundances in each group (non-reserve and MHBR regions, respectively); CuSum—Cumulative contribution to overall dissimilarity (sums to 1); p—Permutation p-value, indicating the probability of obtaining equal or greater average contributions by chance. p < 0.05 (*), p < 0.01 (**), p < 0.001 (***).

Habitat Type	Average	SD	Ratio	avNon-Reserve	avMHBR	CuSum	p
Ploughed with winter cereal	0.163	0.121	1.366	38.290	26.887	0.229	0.001	***
Trellised vineyard	0.110	0.071	1.546	2.820	23.113	0.383	0.001	***
Fallow land	0.092	0.077	1.193	11.110	22.485	0.511	0.626
Holm oak woodland	0.075	0.137	0.552	14.110	1.959	0.617	0.003	**
Goblet-pruned vineyard	0.068	0.086	0.791	0.070	13.557	0.711	0.951
Scrubland	0.056	0.069	0.839	11.540	0.835	0.792	0.001	***
Pinewood	0.054	0.081	0.684	11.04	0.907	0.870	0.001	***
Almond crops	0.032	0.044	0.712	4.96	3.082	0.914	0.059	*
Olive crops	0.026	0.035	0.760	1.750	4.505	0.951	0.589
Juniper	0.014	0.047	0.300	2.790	0.000	0.971	0.001	***
Tamarisk	0.009	0.024	0.367	0.000	1.742	0.983	0.986
Poplar	0.005	0.012	0.402	0.960	0.000	0.990	0.001	***
Holm oak/Pinewood	0.003	0.014	0.220	0.360	0.278	0.994	0.230
Reeds	0.002	0.016	0.102	0.000	0.330	0.996	0.233
Broom shrubland	0.001	0.005	0.207	0.000	0.206	0.998	0.629
Orchards	0.001	0.003	0.297	0.180	0.000	0.999	0.001	***
Isolated buildings	0.000	0.003	0.172	0.000	0.093	1.000	0.742
Other	0.000	0.001	0.241	0.040	0.021	1.000	0.128

). Ploughed with winter cereal areas and holm oak woodland were predominant in the non-reserve macrohabitat, while fallow land, trellised vineyards, and goblet-pruned vineyards in the MHBR region were the most relevant habitats for the differences between macrohabitats. No differences in in the presence/absence of rabbit carcasses were found between roads (n = 58) and tracks (n = 88) in the subsample of localities (Pedro Muñoz, Quero, Lillo) in winter (χ² = 0.005, df = 1, p = 0.95, ns).

4. Discussion

During the breeding season, no Red Kites were detected in the study area, indicating that the MHBR does not host a breeding population. This pattern was also observed in the non-reserve areas surveyed and aligns with previous reports on the species’ breeding distribution in Spain [54]. One possible explanation for the presence of Red Kites in the MHBR during winter, but the absence of breeding is related to habitat suitability. While the MHBR provides adequate resources for wintering, it may lack certain critical factors required for successful breeding, such as optimal nesting sites, sufficient prey availability, or low levels of disturbance. It could be the subject of further studies. In contrast, populations in Germany breed successfully in habitats that appear similar, suggesting regional differences in habitat quality or anthropogenic pressures [55]. Additionally, behavioral and demographic factors may play a role; the wintering individuals in the MHBR could be non-breeding floaters, juveniles, or migratory birds passing through the area [3,17,23]. Moreover, the decline of breeding populations in Spain due to habitat degradation and other threats may further explain the absence of breeding in this region, despite the continued use of the area during winter [16]. These findings highlight the importance of understanding local-scale ecological and conservation contexts when interpreting species distribution patterns.

During the winter season, significant differences in Red Kite presence were observed among the seven geographic areas studied. Notably, the two areas near mountainous terrain in the non-reserve zones (Landete and Motilla del Palancar) had no Red Kite observations during winter, except for one individual, probably due to their high variability in wintering strategies [23], confirming that mountain regions are generally avoided as wintering macrohabitats [22]. In contrast, the remaining five areas in the MHBR region, characterized by steppe-like landscapes, supported the wintering population. These differences appear strongly related to land cover characteristics. The MHBR is dominated by agricultural landscapes with limited forest cover, featuring fallow lands, trellised vineyards, and goblet-pruned vineyards, habitats favorable to Red Kite presence. As aforementioned, this pattern mirrors findings from Germany, where breeding populations are predominantly associated with agricultural land [55]. In Spain, during winter, Red Kites avoid areas with steep topography, dense forests, and tree crops, instead selecting flat, open landscapes composed of crops, agricultural mosaics, pastures, and grasslands [22,31,32]. These preferences are consistent with the habitat selection observed in the MHBR in the present study, as the two steepest areas with the greatest forest cover in the non-reserve zones were avoided. Broad-scale studies across Southwestern Europe have shown that Red Kites inhabiting areas with a higher proportion of open, lowland land cover tend to have smaller home ranges [21]. In raptors, home range size is closely related to food availability, with smaller home ranges typically indicating greater prey abundance [56]. This is especially relevant for the Red Kite, given its reliance on carrion (e.g., rabbit). Consistent with these findings, the wintering population in this study is concentrated in the cultivated areas of the MHBR, suggesting a preference for habitats that provide reliable and abundant food resources.

Based on 3800.5 km² encompassing the five areas with a wintering Red Kite population, our best density estimate is 0.37 individuals/km². This estimate closely matches the density of 0.4 individuals per km² reported for Red Kites in ideal breeding habitats [31]. Comparable densities have also been documented for breeding populations in Southern England [14]. During the breeding season in Wales, Red Kite densities within the occupied territories range widely from 0.2 to 5.6 Red Kites/km² [57]. A higher density of 0.7 individuals per km² was reported in the Pre-Pyrenean region and Western Spain [36]; however, this was observed in preferred wintering habitats of the Iberian Peninsula [22]. Overall, the density we observed for the wintering population in the MHBR is consistent with densities reported for both breeding and wintering Red Kites in other regions.

This study has several limitations. The first limitation relates to the scale, as the study only partially covers the entire territory of the MHBR. However, the results are locally relevant given the importance of the protected area; they should be considered a valuable contribution to the knowledge of a threatened species within Iberia for both specialists and local researchers. The second limitation is the absence of a microhabitat analysis, which represents an important gap for future research, especially if GPS tracking is applied as in recent studies conducted in Spain [20]. Finally, the third limitation concerns the methodology and density estimation. We acknowledge that conducting distance sampling along roads and tracks may introduce bias, since these linear features may not representatively sample the entire study area and could influence animal distribution and detectability. Therefore, density estimates should be interpreted with caution given these methodological constraints [41,42].

5. Conclusions

The MHBR does not host a breeding population of Red Kites, which is consistent with previous reports on the species’ breeding distribution in Spain. Nevertheless, wintering areas can be just as critical for species conservation as breeding areas [19,58]. Our findings highlight that steppe-like agricultural habitats within the MHBR support approximately 5% of the total European Red Kite population during winter. The absence of this species in more forested or mountainous non-reserve areas further reinforces their preference for open landscapes in Central Spain during the wintering period. This study also validates the use of sampling along paved roads, tracks, or a combination of both as effective methods for estimating Red Kite populations in this region. These results carry important conservation implications, emphasizing the need to prioritize the protection of open, heterogenous agricultural landscapes and to guide targeted monitoring efforts aimed at supporting Red Kite wintering populations.