Local and Landscape Drivers of Ground Bird Flocking Behavior in Urban Parks of Buenos Aires City, Argentina

Simple Summary

The flock formation of bird species is a crucial behavioral process that enables them to colonize urban areas. However, the factors influencing the structure and composition of ground-feeding bird flocks have not yet been analyzed. Flock numbers were positively correlated with tree, lawn, and bare ground cover but negatively associated with raptor presence in parks. Flock species richness declined with increased noise and pedestrian traffic but rose in parks where raptors were present. The composition of species in flocks was linked to tree cover, noise, and the presence of raptors. While the Rock Dove (Columba livia) and the Rufous-bellied Thrush (Turdus rufiventris) were more abundant in parks with greater tree cover, the Eared Dove (Zenaida auriculata) and the Monk Parakeet (Myiopsitta monachus) showed increased abundance in more open parks. Zenaida auriculata and Columba livia experienced a decline in abundance in parks where raptors were present. Our findings indicate that resource availability and predation risk are crucial factors shaping flock formation in urban parks.

Abstract

The flock formation of bird species is a crucial behavioral process that enables them to colonize urban areas. However, the factors influencing the structure and composition of ground-feeding bird flocks have not yet been analyzed. This study aimed to relate flock characteristics, including size, number, species richness, and composition, to local and landscape factors in the urban parks of Buenos Aires City, Argentina. Surveys of flocks were conducted in 16 parks during the breeding season, covering both mono-specific and mixed-species flocks. Flock numbers were positively correlated with tree, lawn, and bare ground cover but negatively associated with raptor presence in the parks. Flock species richness declined with increased noise and pedestrian traffic but rose in parks where raptors were present. The composition of species in flocks was linked to tree cover, noise, and the presence of raptors. While the Rock Dove (Columba livia) and the Rufous-bellied Thrush (Turdus rufiventris) were more abundant in parks with greater tree cover, the Eared Dove (Zenaida auriculata) and the Monk Parakeet (Myiopsitta monachus) showed increased abundance in more open parks. Zenaida auriculata and Columba livia experienced a decline in abundance in parks where raptors were present. Our findings indicate that resource availability and predation risk are crucial factors shaping flock formation in urban parks.

1. Introduction

Bird flocks are social groups that can benefit their members, resulting in greater fitness and survival compared to solitary behaviors [1,2]. Foraging birds in groups might enhance predator detection due to more eyes scanning for threats and a dilution effect, as the likelihood of any individual being eaten is inversely related to group size [3,4]. Additionally, birds that form flocks are less vigilant for predators and spend more time foraging [5,6]. Furthermore, aggregations of birds can serve as indicators of food availability for other individuals [7]. These benefits apply to mono-specific and mixed-species flocks [4]. However, mixed-species flocks can offer additional advantages. For instance, individuals in mixed flocks may spend more time feeding than mono-specific flocks [8], or mixed flocks might include sentinel species that provide alarm calls in response to predator presence [9,10].

The assembly of mixed-species flocks has been described as structured or open membership [11]. Structured flocks have been described mostly in understory habitats of lowland evergreen forests, and flocks are characterized by a stable species composition and significant associations between them [12,13,14]. Open-membership flocks have been described for elevation gradients and open habitat types, and flocks have an unstable species composition and random associations between them [15,16,17]. Flock composition varies due to species dynamically joining and leaving the flocks as they enter or exit their territories [11], and flock richness is expected to vary accordingly to the bird richness of communities [16].

Anthropogenic impacts on natural habitats can affect bird flock structure—such as size, numbers, and species richness—and composition both directly by altering food availability and indirectly by reducing the population sizes of flocking species [18,19]. Habitat loss, fragmentation, and decreased habitat diversity have demonstrated negative impacts on flock size and richness, as well as significant changes in species composition in temperate and tropical forests [14,16,20,21,22,23]. Moreover, the changes in composition due to human disturbances exhibit distinct patterns in temperate and tropical regions [19]. In temperate areas, flock composition has displayed a gradual loss of species from the richest to poorer sites or a nested pattern [21,24]. Conversely, in tropical regions, variations in species composition among flocks are more closely related to species replacements or turnover [19,25].

These studies have primarily focused on insectivorous birds living in forests, while research examining terrestrial flocks of granivorous or insectivorous birds is significantly more limited [26,27,28]. This scarcity of studies on terrestrial flocks is even more surprising in urban areas, where flocking behavior appears to be a key trait that facilitates species colonization in cities [29,30,31,32].

Therefore, this study aims to analyze the relationship between ground-feeding flock structure, composition, and environmental variables in urban parks in Buenos Aires City. Food availability has positively influenced flock richness [28], so I anticipated greater flock richness in parks with more lawn and bare ground cover [33]. Larger and less isolated parks would increase flock size, numbers, and richness due to larger populations of flocking birds [34]. In addition, flock size and richness are expected to increase in parks with more raptors, thus increasing flock vigilance. Noise and pedestrian traffic are expected to decrease flock numbers, size, and abundance due to disrupted vocal communication and ongoing disassembly. Since Buenos Aires is a subtropical city, I anticipate a shift in species composition among flocks associated with species turnover.

2. Materials and Methods

2.1. Study Area

The study was conducted in the urban parks of Buenos Aires City (34°35′59″ S, 58°22′55″ W; 25 masl; 3,120,612 inhabitants). The climate type is classified as subtropical humid (Cfa, Köppen classification, [35]), with a mean annual temperature of 18.25 °C and a yearly precipitation of 1236.3 mm (National Meteorological Service). Buenos Aires is a coastal city surrounded by farmland, pastures, tree plantations, and small patches of natural and semi-natural landscapes (Figure 1). Originally, the landscape was composed of grasslands, xerophytic, and deltaic forests [36,37].

A total of 16 parks were selected, separated by at least 500 m (Figure 1). The tree species composition in urban parks is dominated by exotic species native to the Northern Hemisphere, such as the London Plane (Platanus acerifolia), Linden tree (Tilia viridis), and the White Mulberry (Morus alba). However, some species are native to northern Argentina, such as the Tipu tree (Tipuana tipu) and the Silk Floss tree (Ceiba speciose).

2.2. Flock Surveys

Each urban park was visited three times between December 2020 and February 2021, which coincides with the austral summer and breeding season of birds [38]. Bird observations were made during the first four hours of the morning on days without rain or strong winds. Each park was walked through defined pathways, generally following trails, surveying its entire area with the help of binoculars. The time employed in each park varied in direct relation to its area.

I defined a flock as three or more moving individuals nearby (within 10 m of the nearest individual) without indication that the birds were drawn to a concentration of food [15,39], which in parks was primarily provided by humans. I did not track flocks for longer than three minutes to prevent double counting of individuals within the same park and because flocks were often disturbed by humans or dogs [40]. While some flocks were seen in trees, this study focused exclusively on ground-foraging flocks.

In each park, I estimated the following four characteristics of the flocks: (1) flock number, which was the average number of flocks during the three visits; (2) flock richness, the average number of species in each flock during the three visits; (3) flock size, the average number of individual birds in each flock; and (4) mixed flock proportion, the proportion of flocks composed of two or more species compared to the total number of flocks.

2.3. Bird Species Surveys

Bird surveys were conducted in the parks to relate species abundances to flock formation. In each park, one or more fixed points with a 50 m radius were established. In parks smaller than 2 hectares, a point was placed at the center, while in larger parks, two or more points were established, separated by at least 200 m. Point surveys lasted five minutes and were carried out during the first four hours of the day from October to December 2020, coinciding with the bird’s breeding season [38]. Only individual birds foraging or perching in parks were recorded.

2.4. Environmental Variables

A total of nine environmental variables were measured (

Table 1. Environmental variables of urban parks in Buenos Aires City, Argentina.

Scale	Variable	Mean	Range
(a) Local	Area (ha)	4.15	0.36–19.36
	Tree cover (%)	26.32	9.52–38.55
	Lawn cover (%)	30.9	7.46–73.67
	Bare ground cover (%)	11.13	3.72–22.60
	Pedestrians/5 min	12.93	2–39.18
	Noise (decibels)	36.69	15–48.58
	Raptor presence	0.38	0–1
(b) Landscape	Isolation (km)	0.54	0.01–1.05
	Building type	House/high building

). Seven of these variables were collected at the local scale within each park (Table 1(a)). The park area (ha) was assessed using Google Earth Pro. The percent cover of trees, lawns, and bare ground was measured using one or more randomly located points. Each point featured four 20 m lines radiating from the center toward the cardinal directions. For every meter, the presence of each microhabitat cover was recorded. The percent cover was then calculated by dividing the amount of microhabitat presence by the total number of marks and multiplying by 100. Pedestrians and noise were measured simultaneously during bird species surveys conducted from October to December 2020. Pedestrians were counted as the number of people present during five-minute bird point count surveys. Noise levels were recorded in mean decibels 30 s before the bird surveys using the Sound Meter application [41] on a Galaxy S20 FE (Samsung Electronics, Suwon, Republic of Korea). Raptor presence was established from the bird surveys. The recorded species included the Harris’s Hawk (Parabuteo unicinctus) and the Southern Caracara (Caracara plancus).

Landscape variables were those measured outside park boundaries (Table 1(b)). Isolation was assessed using Google Earth Pro and represented the minimum distance (km) to a park of 4 ha or larger. The surrounding building type for each park was a categorical variable indicating whether parks were primarily encircled (>50% covered) by houses or high-rise buildings (three or more stories). Data for this variable were collected through field observations.

2.5. Statistical Analysis

The relationship between the number of flocks occupied by each species and their densities in parks was evaluated using Spearman rank correlation analysis with the cor function in R (version 4.4.2) [42]. The association among species in flocks was examined with the cooccur function from the cooccur package [43]. The cooccur function determines whether the co-occurrence of two species is significantly large and greater than expected (positive association), significantly small and less than expected (negative association), or not significantly different and approximately equal to what was expected (random association) [43]. Species with low occurrence data were excluded from the analysis.

The association between flock characteristics and environmental variables was analyzed with generalized linear models using a Gaussian error structure. Model selection was assessed by backward elimination of non-significant variables (p > 0.05) from the full model using the anova function. Final models were compared with null models using a likelihood ratio test (LRT test) (p < 0.05). Model residuals distribution normality and heteroscedasticity were tested with the DHARMa package [44]. Multicollinearity of environmental variables was assessed with the performance package [45]. Final model results were plotted with the sjPlot package [46].

The differences in species composition among flocks can be attributed to a turnover of species individuals (balanced variation in abundance) or the gradual loss of species individuals between flocks (abundance gradients) [47]. The proportion of these two components was calculated from the total Bray–Curtis (BC) dissimilarity using the beta.multi.abund function from the betapart package [48]. The beta.multi.abund function computes the balanced variation in abundance and the abundance gradient components for multiple flocks. The association between flocking species and the environmental variables was assessed through distance-based Redundance Analysis (dbRDA). This ordination analysis used dissimilarity measures between flocks to relate ordination axes with environmental variables [49]. I used BC dissimilarity between flocks, which takes into account species abundances. dbRDA was run with the capscale function of the vegan package [50]. Models were obtained by backward variable selection and comparisons with null models using a likelihood ratio test (LRT test) (p < 0.05).

3. Results

A total of 1132 individuals of 24 species were detected in 167 flocks. Most of the species belong to the plant-seed diet guild (

Table 2. List of species recorded in flocks, diet guild, number of flocks, mean density in flocks (individuals/flock), and proportion of mixed flocks for each species. Diet guilds are based on the classification by Wilman et al. [51]. VertFishScav refers to species that primarily feed on vertebrates and scavenge. * Exotic species.

English Name	Species	Diet	Flocks	Density	Proportion in Mixed Flocks
Rock Dove	Columba livia *	Plant-Seed	59	4.95	0.75
Picazuro Pigeon	Patagioenas picazuro	Plant-Seed	33	1.61	0.88
Spot-winged Pigeon	Patagioenas maculosa	Plant-Seed	1	1.00	1.00
Eared Dove	Zenaida auriculata	Plant-Seed	63	3.40	0.73
Picui Ground-Dove	Columbina picui	Plant-Seed	6	2.33	0.67
Green-barred Woodpecker	Colaptes melanochloros	Invertebrate	2	1.00	1.00
Southern Caracara	Caracara plancus	VertFishScav	2	1.50	1.00
Yellow-chevroned Parakeet	Brotogeris chiriri *	Frui-Nect	1	3.00	1.00
Nanday Parakeet	Aratinga nenday *	Plant-Seed	1	5.00	0.00
Monk Parakeet	Myiopsitta monachus	Plant-Seed	39	6.54	0.69
Rufous Hornero	Furnarius rufus	Invertebrate	7	1.14	1.00
Cattle Tyrant	Machetornis rixosa	Invertebrate	1	2.00	1.00
Great Kiskadee	Pitangus sulphuratus	Omnivore	1	2.00	1.00
Creamy-bellied Thrush	Turdus amaurochalinus	Frui-Nect	2	1.50	1.00
Rufous-bellied Thrush	Turdus rufiventris	Omnivore	44	3.07	0.64
Chalk-browed Mockingbird	Mimus saturninus	Invertebrate	6	2.33	0.83
Saffron Finch	Sicalis flaveola	Plant-Seed	1	1.00	1.00
Red-crested Cardinal	Paroaria coronata	Invertebrate	1	1.00	1.00
Rufous-collared Sparrow	Zonotrichia capensis	Plant-Seed	3	3.33	0.33
Grayish Baywing	Agelaioides badius	Invertebrate	11	3.36	0.55
Screaming Cowbird	Molothrus rufoaxillaris	Plant-Seed	3	3.33	1.00
Shiny Cowbird	Molothrus bonariensis	Invertebrate	10	2.70	1.00
House Sparrow	Passer domesticus *	Plant-Seed	4	4.25	0.75
European Starling	Sturnus vulgaris *	Omnivore	7	3.71	1.00

). The most common species forming flocks were the Eared Dove (Zenaida auriculata) and the Rock Dove (Columba livia) (Table 2), whereas the most abundant species in flocks were the Monk Parakeet (Myiopsitta monachus) and Columba livia. Columba livia and Zenaida auriculata showed the highest association in flocks (Table S1). Species propensity to flock was strongly positively related to their density in parks (r_s = 0.79, N = 24; Figure 2). The mean percentage of mixed-species flocks was >80% (mean proportion = 0.85; Table 2). However, some species, such as the Grayish Baywing (Agelaiodes badius) and the Rufous-bellied Thrush (Turdus rufiventris), showed the lowest values of mixed flock formation.

The co-occurrence analysis was carried out on 23 species pairs. Only two pairs had a significant negative association (Figure S1), whereas the rest of the species had a random distribution in flocks.

Flock numbers (mean = 11.13, range = 3–30) were highly correlated with flock mixed-species numbers (Pearson correlation, r = 0.93). Therefore, only flock numbers were analyzed. Tree cover, bare cover, lawn cover, and raptor presence significantly explained flock number variation in urban parks (LRT = 34.64, p < 0.001,

Table 3. Final models explaining the relationship between (a) flock numbers, (b) flock richness, (c) flock size, and (d) the proportion of mixed-species flocks with the explanatory variables in urban parks of Buenos Aires City, Argentina.

Response Variable	Parameter	Value	Std. Error	t-Value	p
(a) Flock numbers	Intercept	−18.470	4.762	−3.878	0.003
	Tree cover	0.372	0.091	4.080	0.002
	Bare cover	0.389	0.136	2.867	0.017
	Lawn cover	0.530	0.099	5.382	<0.001
	Raptor presence	−4.607	2.079	−2.216	0.051
(b) Flock richness	Intercept	3.218	0.439	7.334	<0.001
	Noise	−0.036	0.011	−3.201	0.008
	Pedestrians	−0.020	0.007	−2.819	0.017
	Raptor presence	0.406	0.170	2.398	0.035
(c) Flock size	Intercept	6.919	0.610	11.35	<0.001
(d) Mixed flock proportion	Intercept	0.607	0.130	4.660	<0.001

). Flock numbers increased with increasing cover of trees, bare ground, and lawns (Figure 3a–c). On the other hand, flock numbers decreased with raptor presence (Figure 3d).

Flock richness (mean = 1.79, range = 1.00–2.42) varied according to noise, pedestrian traffic, and raptor presence (LRT = 0.89, p = 0.004; Table 3). Flock richness decreased with increasing noise and pedestrian traffic (Figure 4a,b), and increased with raptor presence (Figure 4c).

Flock size (mean = 6.92, range = 3.67–10.50) and the proportion of mixed-species flocks (mean = 0.51, range = 0.00–0.75) did not vary significantly with the explanatory variables (flock size: LRT = 10.23, p = 0.162; mixed flock proportion: LRT = 1.04, p = 0.309; Table 3).

The multiple-flock dissimilarity was 0.99 (BC index), with 0.98 indicating a balanced variation in abundance between flocks. Thus, the primary pattern of species composition change among flocks was the turnover of individuals.

Flock composition varied significantly according to tree cover, noise, and raptor presence (LRT = 3.87, p < 0.001). Flocks composed of Zenaida auriculata and Columba livia were negatively related to raptor presence, whereas Turdus rufiventris was positively associated with raptor presence (Figure 5a). Myiopsitta monachus and the European Starling (Sturnus vulgaris) were positively associated with noisy parks (Figure 5a,b). Zenaida auriculata and Myiopsitta monachus were positively related to open parks, whereas Columba livia and Turdus rufiventris were positively associated with tree cover (Figure 5a).

4. Discussion

Species composition in flocks was linked to the species densities in parks, and species associations were mostly random. Therefore, flock formation supports an open membership pattern, where species gather into flocks based on their local densities. Additionally, the characteristics of flocks and their species composition were related to the local habitat features of parks.

4.1. Flock Structure

The associations between pairs of species were mainly random, with only two pairs exhibiting a negative association. Furthermore, the proportion of flocks occupied by each species was positively correlated with their densities in the parks. Additionally, our results indicated that the turnover of individuals from different species was the main pattern of compositional differences among flocks. These results agree with the idea of open membership [16,25], where species composition in flocks is determined by the species composition in the local community (see also [17,28]). The variable species composition of our flocks contrasts with other studies with more stable flock composition and the presence of nuclear species that contribute to the formation and/or cohesion of flocks [52,53]. For example, Maldonado-Coelho and Marini [14] found that understory flocks of the Atlantic Forest were stable due to the cohesion provided by Habia rubica.

Flock numbers and richness were linked to local environmental factors in urban parks. Flock numbers positively correlated with the percentage of tree cover, lawn area, and bare ground. Tree cover may protect birds, as they can seek refuge in the trees when predators or humans approach [6]. On the other hand, lawn and bare ground cover may be indicators of food availability for flocks [24]. In the study area, many ground species selected these substrates for foraging (LML, unpubl.) [33]. Flocks could help birds to find the most suitable sites for feeding [4]. Additionally, the number of flocks in parks was negatively correlated with the presence of raptors. Raptors, particularly those nesting in parks, could diminish the species densities that typically form flocks, ultimately reducing their numbers.

Flock richness was negatively related to noise and pedestrian traffic. On the one hand, noise can disrupt vocal communication among species in flocks, particularly the vocal alarm calls of vigilant individuals warning of raptor presence [9,54,55]. For instance, in the study area, a mixed-species flock consisting of Mimus saturninus, Passer domesticus, Molothrus bonariensis, and Turdus rufiventris took flight immediately after the vocal call of an individual Mimus saturninus perched in a nearby tree (see also [9]). Additionally, moving vehicles like cars or motorcycles on nearby streets can disturb flocks [56]. On the other hand, pedestrians and their dogs can disrupt the flocks. The energy lost from birds constantly chasing flies can exceed the energetic benefits of feeding in a group, making flocking less advantageous. Furthermore, noise and pedestrian traffic in parks have been linked to decreased species diversity in urban parks [57,58,59]. Consequently, reduced species diversity in parks may lead to fewer species joining flocks.

The presence of raptors in urban parks positively correlated with species richness in flocks. A greater number of species in flocks can improve predator detection strategies, as foraging species may vary in the time they allocate to vigilance and food searching [9,60].

Flock size and the proportion of mixed-species flocks did not vary with the analyzed environmental variables. These results contrast with those obtained by Fernández-Juricic [20], who found larger flock sizes in bigger and better-connected parks. However, Fernández-Juricic [20] focused on forest insectivorous birds, while our study concentrated primarily on granivorous and insectivorous ground foraging birds. Thus, ground-foraging birds may be less sensitive to green area size and isolation.

4.2. Flock Composition

Bird species composition of flocks was related to tree cover, noise, and raptor presence. Turdus rufiventris and Columba livia were more abundant in flocks located in parks with increased tree cover, whereas Myiopsitta monachus and Zenaida auriculata were in more open parks. The open structure of parks could give more visibility to the presence of raptors, pedestrians, or dogs. On the other hand, the positive relationship between Turdus rufiventris and Columba livia with tree cover might be linked to the food resources that trees offer. Turdus rufiventris may also utilize trees to evade predators or human disturbance.

Turdus rufiventris, Myiopsitta monachus, and Sturnus vulgaris were found to be more abundant in flocks in noisy parks. These species might have a high song pitch, which could enhance their vocal communication in such environments. On the other hand, Zenaida auriculata and Columba livia were more abundant in flocks found in parks without raptors. These species are likely significantly influenced by the predation risk posed by raptors, as several observations in the study area showed that Parabuteo unicinctus and Caracara plancus prey on them [61,62] (LML pers. obs.).

4.3. Limitations

Bird surveys and human disturbance were measured in spring, while flock surveys took place in summer. Thus, some temporal variation likely occurred in the explanatory variables between the seasons. For instance, pedestrian traffic and noise may decrease during summer due to less human activity associated with summer holidays. However, I do not believe these temporal dynamics would impact the spatial patterns of human disturbance or the flock structure among parks. Conversely, bird communities likely remain similar between spring and summer, as migratory bird species may stay in the study area.

5. Conclusions

The results showed that urban parks’ most common species forming ground flocks were abundant, suggesting that their local densities mainly drove flock propensity. On the other hand, flock number and richness were influenced by habitat variables and raptor presence, suggesting that resource availability and predation risk are critical factors molding flock formation. The species composition of flocks also varied according to habitat variables and raptor presence, suggesting that some species were more vulnerable to predation risk. In contrast, other species may adapt better to human disturbances, such as noise. Therefore, our results bring novel insights into the behavioral processes of urban bird communities. Flocking behavior in urban birds may improve their resource intake efficiency and avoid bird predation, thus enhancing their fitness.