Fission–fusion dynamics and spring movements in first-year carrion crows Corvus corone challenge the efficiency of culling strategies

Abstract

This study investigates the spatial behaviour and social dynamics of immature carrion crows (Corvus corone) caught in Paris, with the objective of discussing the efficiency of current culling strategies used to mitigate crop damage. Using GPS tracking data from 80 first-year crows of four different cohorts, we characterized seasonal variations in movement, roosting behaviour, and fission-fusion spatial dynamics. Results show that young crows remain closely associated with their natal sites until March of the year following hatching, after which they disperse to urban, peri-urban and rural areas, particularly in April and May, with peak daily movement distances averaging 2.7 km. This period coincides with the sowing of sensitive crops such as maize and sunflower, during which culling campaigns are often conducted. Furthermore, we document fission-fusion dynamics, with fluid group associations centred around predictable anthropogenic food sources and communal roosts, certainly facilitating repeated interactions and social learning. These findings suggest that current culling practices, largely targeting transient juveniles, are unlikely to achieve long-term reductions in population size or crop damage. Instead, the study highlights the importance of addressing particularly anthropogenic food availability as a potentially more effective and ethical strategy for managing urban, peri-urban and even rural crow populations.

Supplementary Information

The online version contains supplementary material available at 10.1038/s41598-025-17175-y.

Introduction

Corvids exhibit a wide range of social organizations, from territorial breeding pairs to large, dynamic groups of non-breeders. They are capable of social learning, where an individual corvid learns about the dangers and opportunities of the local environment through the observation of its conspecifics¹. Both imitation and signalling enable this horizontal transmission of information, fostering contextual adaptability as well as cooperative strategies^2,3. In a number of species, non-breeders gather in loose flocks, the dynamics of which have been described as a fission-fusion permanent process (for ravens Corvus corax⁴, jackdaws Corvus monedula⁵. Fission-fusion dynamics entail that at a given site, where non-breeders flock, regular emigration and immigration events induce temporal variations in group size and composition, and thus, results in a continuously changing group. The population of non-breeders over a given range spans a network of such interconnected flocking sites, in between which individual non-breeders navigate freely, each with their own collection of favoured locations and varying exposure to other individuals through repeated encounters. As corvids are also capable of differentiated relationships – i.e. they recognize conspecifics individually – this creates for each bird a complex social landscape⁶. Such population dynamics are tightly coupled to foraging and dispersal patterns, which scale varies with time, depending on resource availability and biological cycles throughout the year.

Some native vertebrate predators are considered as pests because of their detrimental damage to agriculture, forestry, livestock and human health⁷. By controlling their numbers, destruction campaigns aim at reducing their impacts on human economic activities and human or livestock health. The efficiency of these controls has been regularly challenged, as reducing the numbers of pest animals does not systematically reduce the damages or risks⁸. Understanding the temporal and spatial dynamics of such animal populations, including individual movements, is therefore crucial to evaluating culling strategies and planning efficient measures intended to reduce damages. Across Europe, over 4 million corvids (1,2 million in France alone)^8,9 are culled annually as a consequence of, and to prevent, their damages to agriculture and other wildlife. Rural corvid populations are often persecuted so that their demographic structure is skewed towards young individuals¹⁰, while urban populations are sheltered from massive regulations and have access to abundant and predictable food wastes¹¹. Carrion Crow Corvus corone displays a flexible sociality. These are generally social birds, with pairs that defend territories year-round and groups of non-breeders that aggregate at roosting and feeding sites. Although large flocks of non-breeding crows are common in urban and rural areas, fission-fusion dynamics are yet to be reported in the species. By analysing GPS-tracking data from 80 immature carrion crows during their first- and second-calendar-years, we aim at reporting on temporal variations in the movement strategy of immature crows hatched in urban environments, and at further documenting the fission-fusion dynamics in the corvid family^4,5. First, we sought to identify the individual dormitories of young crows during their first winter, different from the communal dormitories frequented by numerous immatures. The hypothesis is that these individual fixed and recurrent roosts are located within the territory of hatching, making it possible to evaluate the distances separating the place of birth and the capture site of young tagged crows. Next, we studied the daily dispersal distance of young birds during their first year of life, to identify potential periods of dispersal or prospecting movements, and the spatial extent and temporality of such movements. This was done by measuring the distance between dormitory sites used on two successive days, by considering pairs of consecutive nights for which data was available. Then, we studied the frequency of close locations for young crows from the same hatching cohort, in order to document fission-fusion dynamics in immature carrion crows. Finally, we discuss how these results on movements and fission-fusion dynamics in young carrion crows could impair the efficiency of usual culling strategies deployed to reduce crop and other damages attributed to the species.

Results

Winter roosts and natal sites

During the non-breeding season, from tagging to the end of March, the first-year crows did roost at communal dormitories but most also used individual roosts. Recurrent non-communal roosting sites could be identified for 69 first-calendar-year crows (8 hatched in 2020, 18 in 2021, 21 in 2022, 23 in 2024). The proportion of nights spent at individual roosts averaged 40 ± 30% (range 0-100). Those birds roosting at only communal dormitories probably hatched in a nest close to the roost. For these 69 birds, the average ± s.d. (median; range min – max) distance between the capture site (the cage trap at Jardin des Plantes) and this roosting site was 3 486 ± 5 264 m (2520; 20 – 35 968; see Fig. 1). In at least 10 cases, these roosting sites have been confirmed as the natal territory by direct observations of colour-ringed immatures in company of a pair of adults. We therefore interpret this main non-communal roosting site as the natal site, i.e. the breeding territory of the parents. Plotting the distribution of the distances between the natal site and the capture trap gives indications of the foraging distances of immature crows during their first autumn and winter.

Fig. 1.

(a) Histogram of the distances between individual roosts and the capture trap (localized by the white arrow on panel (b) for 69 first-year carrion crows tracked during winter (up to 31 March following their hatching year), and maps of (c) all non-communal nocturnal roosts (yellow symbols) interpreted as the parental territories, with (b) a zoom on sites located within Paris city. Map data ©2025 Google.

Daily movements and spring exploration

We estimated the distance travelled between nocturnal roosting sites of two successive nights by 80 tagged immature crows, from the 1 of September of their hatching year to the following 31 of August (except for 2024-hatched birds, data until 30 April 2025; data from 23085 pairs of successive nights). Figure 2 synthesizes the information by merging all ‘years’ (1 September to 31 August) and reporting the distances during the calendar year (from January to December). A generalized additive model performed on these data identified two periods of increased movement distances (Fig. 2b; F₄ = 112.4, P < 0.001). This analysis aimed to illustrate movement patterns during dispersal phases, as opposed to the daily foraging distances of individuals consistently roosting within the same area (see model residuals, Fig. 2c). The daily dispersal movement increases in March, rise more sharply in April, and peaks in the last third of April. Thereafter, we measured very low values during the summer, with a moderate increase in November and December. The average daily movement distance reaches 2.66 kilometres in late April, which represents the peak of dispersal movements. We investigated whether body size, as indicated by wing length measured at ringing, could influence the mobility of individual crows. We did not find a significant effect of wing length on the daily distances travelled by a subset of 78 individuals with documented wing length (generalized linear model; wing length, t = -0.08, P = 0.93). However we observed a tendency for larger birds to travel shorter distances during the last third of March, with one significant interaction between wing length and the ten-day period of March 21–30, which had a negative coefficient (t = -2.17, P = 0.030). Figure 3 illustrates movements of tagged crows during their first year.

Fig. 2.

Distance travelled daily by first-year carrion crows, as estimated by the distance between roosting sites used during two consecutive nights, for 80 tagged individuals. Data are grouped by ten-day periods within each month (days 30 and 31 included in the last monthly period). Average + s.d. values are reported. The two graphs below are plots of the generalized additive models performed on the data, reporting the response curve (left) and showing the residuals (right).

Fig. 3.

Heatmap of all locations obtained for 80 carrion crows tagged in Paris during their first year during 2020–2025. Below are two examples of individual tracks (in blue), for birds with different extend of distant spring dispersal movements. Maps have been created using the movebank.org platform.

Fission–fusion dynamics

In the time period of up to 56 months, 80 immature individuals were tracked with GPS-loggers that measured their position with a minimum interval of one hour. We considered four cohorts corresponding to birds hatched in a same year (2020, 2021, 2022 and 2024 – see Table 1). Thus, we could calculate the distance between all dyadic combinations of individuals in a cohort, where the positions of both individuals were obtained (688,551 calculated distance values from tagging to the end of March; 873,129 from April to August). We then filtered for occurrences where individuals stayed within 100 m, which were previously defined as associations. The proportions of dyadic combinations (see Table 2) found to be within 100 m ranged (mean in brackets – for all distance calculations, every individual being included at least once): 0 to 38.6% (mean 3.5%) in 2020, 0 to 40.5% (mean 2.3%) in 2021, 0 to 50.5% (mean 5.8%) in 2022, 0 to 39.6% (mean 1.3%) in 2024. Figure 4 illustrates the co-occurrence rates as obtained for the 2022 cohort from tagging in September 2022 to 31 March 2023. Values obtained for different periods were similar and attest to fission-fusion dynamics too (see Table 2). For the period from 1st April to the end of August of the same year, we obtained quite similar though often just smaller values, probably because this period includes the spring dispersal movements, when some birds wander outside of the usual urban network of feeding and roosting sites. We obtained the following average proportions (maximal individual value in brackets): 4.1% (max. 28.5%) in 2020, 2.8% (max. 38.5%) in 2021, 4.6% (max. 34.6%) in 2022, 1.4% (max. 24.0%) in 2024. For a more global period of immaturity covering 17 months, from 1st April to the end of August of the next year, and for three cohorts, we obtained highly similar average proportions (maximal individual value in brackets): 4.1% (max. 28.5%) in 2020, 2.2% (max. 32.0%) in 2021, and 4.1% (max. 36.4%) in 2022.

Table 1.

Summary statistics of cohorts across the study period (from capture to the end of the summer, Y + 2): mean values ± standard deviation (per individual).

Cohort	Sample size	Started	1 h time steps	Days with time steps
2020	171	2020-09-04	1 440.2 ± 843.0	162.9 ± 96.2
2021	222	2021-09-20	7 861.8 ± 3 872.1	493.2 ± 192.6
2022	152	2022-09-06	9 933.1 ± 3 187.3	509.8 ± 115.9
2024	272	2024-09-03	2 563.4 ± 1 167.6	187.0 ± 53.1

¹3 individuals removed from dyadic combinations due to low-quality data; values do not account for filtered individuals.

²1 individual removed from dyadic combinations due to low-quality data.

Table 2.

Mean association within each cohort, defined for each dyadic combination as the ratio of shared time steps in which the dyadic pair is within 100 m of each other, across three different juvenile time periods. Values are presented as mean association with maximum values in parentheses.

Period	2020	2021	2022	2024
Sep–Mar	3.5% (38.6%)	2.3% (40.5%)	5.8% (50.5%)	1.3% (39.6%)
Mar–Aug	4.1% (28.5%)	2.8% (38.5%)	4.6% (34.6%)	1.4% (24.0%)1
Mar–Aug + 1	4.1% (28.5%)2	2.2% (32.0%)	4.1% (36.4%)	N/A

¹Time period shortened to May 15th, due to data availability at the time of writing.

²Low data after first summer explains the stability across the second and third periods.

Fig. 4.

Association matrix for the percentage of time steps, when two individuals were within 100 m in the study area, as obtained for the 2022 cohort of crows (from tagging to 31st March). Gxxx corresponds to the identity of individuals (green ring with code xxx).

Figure 5 reports the density map of locations recorded for the 80 tagged crows within Paris, as a point density map for 2020–2025 for crows tagged during their first calendar year. The network of gathering sites includes food resources (large parks with picnicking visitors) and communal roosting sites (large parks and cemeteries with tall trees, a closing time before sunset and no nocturnal lights; indicated with white ‘R’s in Fig. 5).

Fig. 5.

A density map of the sites used by tagged carrion crows in Paris, as a point density map during 2020–2025 for 85 crows tagged during their first calendar year. Communal roosts are indicated with a white R, as sites where 2 to 9 tagged crows used to spend the night. Below are two examples of individual tracks, for birds visiting different parts of the site network. Maps have been created using the movebank.org platform.

Discussion

Movements and dynamics

The telemetric monitoring of young crows carried out in Paris has revealed that these birds continue to visit their hatching site, namely their parents’ territory, for many months after fledging until March of the year following their hatching. This maintenance of privileged family relationships over time is certainly evidence of strong social and family ties^12,13, but also of the possibility of learning and transmission for many months after fledging. These family ties further make it possible to identify the hatching sites for a large majority of the tagged young birds, and thus to estimate the distance separating the capture site from these hatching sites, in order to draw the distribution of the distances of origin of the young crows present in Paris. If most individuals originated from Paris or its close suburbs, occasional birds (4 out of 63) hatched at sites located > 10 km from the trap – and up to 30 km. These first-year crows stop visiting their parents’ territory in March, when the adults start a new active breeding season again, building a new nest or reinforcing an old one, and excluding any intruders from their territory - including their young from the previous year^13,14. From then onwards, the young crows will roost at communal dormitories only, such as those marked on Fig. 5.

From April onwards, young birds in their second calendar year start to disperse, without necessarily returning to sleep where they did previously. This seasonal prospecting behaviour results in an increase in the distance separating two successive dormitories. Throughout the year, this distance averages between 400 and 600 m, which corresponds to the use of common dormitories as identified in Fig. 5. In mid-March, the distance reaches one kilometre, peaking at 2.66 km in the last ten days of April. A return to low values was observed from June onwards. In July and August, the distances are minimal, averaging between 190 and 330 m, which corresponds to the period of complete moult for these one-year-old birds. During these months, they replace all their flight feathers (remiges and rectrices¹⁴), and therefore have reduced flight capacity and lift - which is clearly reflected in the average distances moved during the mid-summer.

So, during their first-winter, young crows move into the city to visit feeding sites during the day, and roost at parent’s territory or at communal dormitories during the night. These daily activities and movements are performed individually, though given the number of immature birds frequenting the same feeding and roosting sites, the overall pattern corresponds to fission-fusion dynamics, as described previously in other large corvids^4,5,15,16. Here the study of the dynamics is based on tracking data only. The quantification of the frequency of pairing between two birds should be poorly biased, as being independent of the observation effort that would apply to the visual monitoring of colour-ringed birds⁴, though it could be biased by heterogeneity in battery charge hence location recording frequency throughout the year.

For immature urban carrion crows, roosting with parents during 8–9 months after fledging makes learning and knowledge transmission possible and likely, while the movement patterns within a network of density hotspots favours repeated interactions between individuals at different sites. High fission-fusion dynamics may thus shape the social complexity of carrion crows, as already reported for other large corvids^4,5. Such dynamics are frameworks usually associated with socioecological, communicatory, and cognitive aspects¹⁷.

Implications for the efficiency of regulations

The movements and fission-fusion dynamics, as observed in this tracking-based study, can help to assess the relevance of measures usually undertaken to regulate the species numbers, when crows are held responsible for economic damage, to crops for example. A first important result is that a large part of urban young crows, either hatched in cities or foraging in urban environments thrive during their first year of life, do wander to rural and agricultural areas in spring. It means that crow numbers should inflate in the wider countryside in relation to the numbers and density of crows in regional cities. In cities, the survival of young crows largely depends on food provided by human activities, especially food wastes in public bins¹¹. A similar reliance of immature ravens on anthropogenic food sources was reported during their dispersal¹⁸. The strong dependence of urban crows on anthropogenic food resources has implications for damage management and conservation issues. Reducing the number of crows dwelling in cities or wandering into agricultural lands in the spring, should entail restricting the access of urban crows to food wastes in cities, as a sustainable strategy to regulate their numbers in a more ethical approach than culling.

From 2021 to 2024 (during four full years), 843 of the 879 crows trapped and ringed in Paris were in their first-year of life (F.J. unpubl. data), representing 96% of all captures. All have been captured in a passive cage trap they enter to feed. This type of trap is widely used in France to capture corvids for culling purposes, particularly in spring, when maize and sunflower are sown and their seedlings emerge. These two spring crops are particularly affected by corvids^19–21. After acknowledging the spring movements of young crows, we realized that these movements take place at the end of April and beginning of May, a period that corresponds to the sowing and emergence of these two crops. Trapping organized at this time of year will therefore mainly capture young birds travelling almost daily, i.e. birds which would probably not have remained on the site, while other young birds on the move may visit the site every new day. As young birds account for the vast majority of captures, this culling strategy could be doomed to failure in terms of reducing economic losses on crops, due to the spatial dynamics of the birds in spring. The capacity of these regulations (an average 600,000 crows are culled in France each year) to reduce economic damage to human activities and to regulate the breeding numbers of the species should be studied further^8,22. Because it is mainly the young birds that are culled in spring when they are on the move, we can expect the breeding adults to be little impacted and then to be able to breed each year and replenish the stock of crows, ensuring stable numbers of crows. Indeed, the breeding population of the carrion crow has been stable for over twenty years in France (population growth rate of 1.001, confidence interval 0.999–1.003)²³.

Methods

Study area and data collection

Carrion crows were captured and colour-ringed at Jardin des Plantes in central Paris, France (48,84° N, 2.36° E) since 2015. This urban park located in Paris city, France, near the Seine River, covers 23.5 ha and is mainly frequented by families, tourists and schools for recreational and cultural activities. We trapped crows using a baited cage trap (8 m3), into which crows can enter through a horizontal ladder along the roof, but from which they cannot escape. The trap was visited every day and birds were processed within max. 18 h of capture. From 2015 to 2024, over 1 500 crows have been captured and ringed. We assumed for the purpose of this study that trapped crows are representative of all crows visiting the study site, though we cannot exclude potential biases due to differences in behaviour, food foraging, and propensity to join a captive flock of conspecifics^24,25. In 2020, 2021, 2022 and 2024, 81 immature crows were fitted with a GPS-GSM logger during their first autumn or winter. We used different types of GPS-GSM loggers: Ornitrack tags (OT) from Ornitela, Lithuania (www.ornitela.com), and Druid OMNI tags from Interrex, Poland (www.interrex-tracking.com). OT-10 (10 g), OT-E10 (12 g), OT-15 (15 g), OT-E15 (18 g) and OMNI 3G (10.5 to 15 g) were attached using a Teflon flat ribbon harness (6 mm width), with a neck loop, a breast flat knot (placed just above the top of the sternum) and a metal clip to close the harness at the back of the tag. Later recaptures of tagged individuals, up to 38 months after initial deployment, never revealed any kind of injury due to the tag or harness. Including the tag, harness and rings, the equipment of a bird was always below the limit of 5% of an individual’s body mass, and represented 3.4 ± 0.6% (mean ± s.d.; min. 2.3, max. 5). We controlled for biases in the deployment process that could favour more fit and robust birds. Our analysis showed that the body size of geotagged crows did not differ significantly from that of the other captured juveniles from the same years: 308 ± 14 mm of wing length for the 548 untagged crows for which wing length was documented, 311 ± 12 mm for the 78 documented tagged crows (unpaired t-test, t = 1.78, df = 624, P = 0.07). We are therefore confident that tagged juveniles are a representative sample of the population of young crows visiting the park and likely to enter the trap during their first year of life. All procedures performed in this study involving animals were in accordance with the ethical standards of the French government guidelines. Authorization to capture, ring and tag carrion crows was obtained from CRBPO at Museum National d’Histoire Naturelle, which is the national licensing committee delivering such reglementary authorizations in France, under file reference PP883.

We confirm that the study is reported in accordance with ARRIVE guidelines (https://arriveguidelines.org). Once captured, each crow was ringed with a uniquely coded metal ring from the national ringing scheme (CRBPO), and also with a coded coloured plastic ring on tarsus, displaying a unique individual code, allowing distant individual identification. Age was evaluated using the colour pattern of the palate (pink for 1y crows and blackish for older birds) and moult contrasts between body and flight feathers^14,24.

Since crows are diurnal birds, we intended to get a GPS position every hour, separating locations first obtained from sunrise to sunset, second during nocturnal hours (excluding hours when sunrise and sunset occurred), to identify the position of each night roost. The number of GPS fixes varied between individuals and time of the year, since some loggers did not perform as well as others under poor light conditions, especially during short winter days, when they required larger sampling intervals of several hours. We had GPS data for 81 birds tagged as first-year individuals from September 2020 to the end of April 2025 (see Table S1 in Supplementary Material), and kept the 75 individuals with more than 300 recorded positions. Inconsistencies related to GPS precision were filtered using conservative methods. We defined 65 km/h as the threshold above which a data point was considered non-physiological. Data points with speeds exceeding this threshold before and after the point were filtered out (n = 64). Additionally, to address periods of data scarcity, we removed data points exceeding a distance of 50 km in both directions (n = 1).

Nocturnal roosting sites

To determine where young crows roosted along the year, we filtered tracking data to keep only first-calendar year individuals tagged in the autumn, and their further georeferenced locations obtained at night during the autumn and winter period, from the 1st of September to the 15th of March. This sample contained 67 individuals. The aim was to detect potential aggregates of such nocturnal roosting sites. We considered the nocturnal daily positions to belong to a same site if they were distant of less than 100 m. For each individual, we summarized the data as the number of roosting sites holding 30% and 40%, between 40% and 50%, and more than 50% of all night positions. Some of these sites are communal roosts where numerous immature crows gather at night, other are individual sites suspected to occur in the territory of the parents. To disentangle these two types, we reproduced the same analyses of nocturnal locations for immature crows of one year old or more, i.e. normally fully emancipated from their parents. For young crows in their first year of life (1st September – 31th March), we thereafter evaluated the proportion of nights spent at communal roosting sites, and the main individual roosting sites outside these communal roosts, with the associated proportion of nights spent there.

Daily movements and spring dispersal

We calculated a nocturnal average position for each individual and each calendar day, and estimated a distance moved between two successive days along the individual tracking period. We compared the averaged movement distances by ten-day periods along the year, using a generalized additive model with decadal period as the explanatory variable, and individual as a factor (mgcv package of R 4.3.3 software). This approach allowed us to investigate non-linear trends in the data. Additionally, we investigated the effect of wing length on movement distances, with the hypothesis that larger wingspans could enhance mobility. We employed a generalized linear model to explain movement distances, using wing length at ringing as an explanatory variable, decadal period as a factor, and the interaction between the two to assess how the effect of wing length might vary across different ten-day periods. Wing length data was available for 78 of the 81 studied juveniles.

Fission–fusion in urban environments

We reproduced the methodological approach used by Loretto et al.⁴ to document fission-fusion dynamics. For the immature crows outfitted with GPS loggers, we estimated the distance between all individuals at every possible time step. One time step was defined as one hour, the shortest interval of GPS-fixes during daytime. If a time step of an individual from a combination was missing due to e.g., a low battery level or deviated by more than 20 min, this time step was not used in the subsequent analyses. Thus, we only used the positions of all dyadic combinations during the time when both individuals were outfitted with active GPS-loggers. We defined an association between two individuals to be, when their Euclidean distance at the same time was less than 100 m. The percentage of these occurrences out of all possible time steps was calculated as well as their maximum duration, i.e., number of consecutive time steps within a distance of 100 m. This investigation was performed for three separate time periods in the early life of immature crows: from tagging to the end of March (of the second-calendar year of a crow), from 1st April to 31st August (2nd cy crows), then from 1st April to the end of August of the next year (when the crow is 2-years-old, in its 3rd calendar year). For crows hatched in 2024, which were in their second calendar year at the time of writing, the second period (1st April to 31st August) is shortened, extending only to the 15th of May, compared to the other cohorts.

To identify the areas where individuals met over time, we produced a density map on the movebank.org platform, selecting the data from all immature crows tagged in the city. Subsequently, we visited these sites to describe their environmental characteristics that might have led to aggregations, for example the presence of foraging sites like tourism hotspots or night roosts. The latter have also been determined by GPS fixes between dusk and dawn, as roost sites where at least two individuals spend > 50% of the nights.

Supplementary Information

Below is the link to the electronic supplementary material.

Author contributions

FJ was invested in study conception and design. FJ ringed all birds, managed the database, while CG handled, coded and performed the raw data analyses. FJ performed the statistical models when appropriate. FJ wrote the initial manuscript; CG reviewed and approved the final version.

Data availability

The dataset used and analyzed in this study can be freely viewed and accessed on the movebank.org platform, under the study ID 1266784970, at the link [https://www.movebank.org/cms/webapp?gwt_fragment=page=studiespath=study1266784970] .

Declarations

Competing interests

The authors declare no competing interests.