Allometry of Alarm Calls: Black-Capped Chickadees Encode Information About Predator Size

See allHide authors and affiliations

Science  24 Jun 2005:
Vol. 308, Issue 5730, pp. 1934-1937
DOI: 10.1126/science.1108841


Many animals produce alarm signals when they detect a potential predator, but we still know little about the information contained in these signals. Using presentations of 15 species of live predators, we show that acoustic features of the mobbing calls of black-capped chickadees (Poecile atricapilla) vary with the size of the predator. Companion playback experiments revealed that chickadees detect this information and that the intensity of mobbing behavior is related to the size and threat of the potential predator. This study demonstrates an unsuspected level of complexity and sophistication in avian alarm calls.

Predation is a major cause of mortality for most species of animals, and many produce alarm signals when they perceive a potential predator (1). Alarm calls often differ in acoustic structure, depending on the situation in which they are produced (25). If a species is preyed upon by different predators that use different hunting strategies or vary in the degree of danger they present, selection can favor variation in alarm signals that encode this information (6). Such variation in alarm signals can be used to transfer information about the type of predator [referential alarm call systems (7)], the degree of threat that a predator represents [risk-based systems (8)], or both (9, 10).

In addition to discriminating among broad types of predators (e.g., raptor versus snake), discriminating among morphologically similar predators within a single type (e.g., different species of raptors) could also be adaptive if the predators vary in the degree of threat they pose. One species that is faced with numerous, morphologically similar predators is the black-capped chickadee (Poecile atricapilla). Chickadees are small, common songbirds that are widespread throughout North America. In the nonbreeding season, chickadees form flocks of six to eight birds (11). They use an elaborate system of vocalizations to mediate social interactions in these flocks (12, 13) and to warn conspecifics about predators (14, 15).

Chickadees produce two very different alarm signals in response to predators: When flying raptors are detected, chickadees produce a high-frequency, low-amplitude “seet” alarm call; in response to a perched or stationary predator, they produce a loud, broadband “chick-a-dee” alarm call that is composed of several types of syllables (16) (Fig. 1A). Whereas the “seet” alarm call functions to warn of flying predators, the “chick-a-dee” mobbing alarm call recruits other chickadees [and often many other species (17)] that harass, or mob, a perched predator. This “chick-a-dee” call is a complex vocalization that is also produced in many other situations and encodes information about food and identity (both individual and flock) in addition to information about predators (11).

Fig. 1.

Features of the “chick-a-dee” mobbing vocalization. (A) The call usually contains both “chick” sections (A, B, and C syllable types) and “dee” sections (D syllable types) (11). (B) Acoustic variables measured from power spectrum analyses from the center of a D note. The amplitude has been scaled relative to the highest energy overtone (dB = 0). Acoustic variables were the peak frequency (P), the lower and upper frequencies above –10 dB (L and U, respectively), the frequency of the first and second peaks above –30 dB (F1 and F2, respectively), and the maximum frequency peak above –30 dB (M). The bandwidth at –10 dB was calculated by subtracting L from U; the bandwidth at –30 dB was calculated by subtracting F1 from M. The interval between overtones was determined by subtracting F1 from F2.

We examined variation in the mobbing vocalizations and behavior of black-capped chickadees by conducting predator presentations and playback experiments with chickadees living under semi-natural conditions in large, outdoor experimental aviaries (18). We presented flocks of color-marked chickadees with 13 species of live, perched raptors and two species of live mammalian predators. The predators varied considerably in body size (e.g., factor of 20 difference in body mass between northern pygmy-owl and great horned owl), activity patterns, hunting strategies, and diet (Table 1). The raptors ranged from small, maneuverable predators whose diets include many small birds, to large, less maneuverable predators that eat few small birds. We also used two types of controls: a procedural control with a live bobwhite quail (Colinus virginianus); and no presentation, with observers present as they would be during predator presentations. During each predator presentation, two observers recorded chickadee vocalizations, noting the color band combination of each calling individual. By conducting controlled presentations of live predators to birds living under semi-natural conditions, we could isolate vocal responses to specific species of predators from other features such as the location, behavior, or movement pattern of the predator.

Table 1.

Species presented to chickadee flocks. Length and wingspan were measured from animals used in experiments; mass (26) and diet information (2729) were summarized from published accounts. The sex of each raptor used in the experiments is indicated in brackets.

Predator species Mass (g) Length (cm) Wingspan (cm) Time active Primary diet
Cooper's hawk (Accipiter cooperii) [F] 450 44 81 Day Small birds
Red-tailed hawk (Buteo jamaicensis) [F] 1,080 53 120 Day Small mammals, few birds
Rough-legged hawk (B. lagopus) [M] 990 49 138 Day Small mammals
American kestrel (Falco sparverius) [M] 117 25 58 Day Inverts, small mammals, small birds
Merlin (F. columbarius) [F] 190 28 61 Day Small birds
Peregrine falcon (F. mexicanus) [F] 720 47 120 Day Medium-sized birds
Prairie falcon (F. peregrinus) [F] 720 45 100 Day Small mammals, some birds
Gyrfalcon (F. rusticolus) [M] 1,400 52 115 Day Medium-sized mammals and birds
Northern pygmy-owl (Glaucidium gnoma) [M] 70 15 31 Day Small birds, small mammals
Saw-whet owl (Aegolius acadicus) [M] 80 17 39 Night Small mammals, some small birds
Short-eared owl (Asio flammeus) [M] 350 34 89 Both Small mammals
Great horned owl (Bubo virginianus) [M] 1,400 48 121 Night Small to medium-sized mammals
Great gray owl (Strix nebulosa) [M] 1,080 58 132 Both Small mammals
Domestic cat (Felis domesticus) 15,000 45 Both Birds, small mammals, insects
Ferret (Mustela putorius) 1,000 32 Day Small mammals, eggs, some small birds
Bobwhite quail (Colinus virginianus) 170 22 42 Day Seeds, insects

Spectrographic analyses of the more than 5000 “chick-a-dee” mobbing alarm calls we recorded (Fig. 1) revealed previously unsuspected levels of acoustic variation. The number of mobbing calls produced in response to each predator was highly variable [analysis of variance (ANOVA): F16, 34 = 5.17, P < 0.0001], with the smaller, higher risk, predators eliciting significantly more calls than the larger predators or controls (Tukey's post hoc tests: P < 0.05). The total number of syllables per alarm call differed among predator treatments (F = 3.05, P < 0.0001). In particular, the average number of D syllables, or notes, per call differed significantly across predator treatments (F = 7.771, P < 0.0001). There was a strong inverse relationship between the number of D notes per alarm call and the wingspan of the raptors, with the smallest predators eliciting calls with the most D notes (Fig. 2A; r2 = 0.512, P < 0.0001). There was also a strong inverse relationship between the number of D notes and predator body length when both the mammals and raptors were included in the analysis (Fig. 2B; r2 = 0.361, P < 0.0001).

Fig. 2.

Chickadees vary their mobbing calls in response to variation in predator body size. (A) Mean number of D syllables per call as a function of wingspan for raptors (y = 4.4–0.02x; r2 = 0.512, P < 0.0001). (B) Mean number of D syllables per call as a function of body length for raptors and mammals (y = 4.4–0.4x; r2 = 0.361, P < 0.0001). Each taxonomic group of raptors is represented by a different symbol (⚫ owl; ▲, falcon; ◼, hawk; ×, mammal). A bobwhite quail (♢) was used as the procedural control. The dashed line displays the mean number of D notes per control trial without any stimulus.

Many other acoustic features of these mobbing calls (Fig. 1) also varied in relation to the predator treatment. For example, in comparisons of mobbing calls given in response to a northern pygmy-owl and a great horned owl (small and large predators, respectively), the duration of the “dee” section (all D notes) was significantly longer (ANOVA: F1,14 = 9.984, P = 0.003), the interval between the “chick” and “dee” sections was significantly shorter (F = 11.364, P = 0.001), the first D note of each call was shorter (F = 9.984, P = 0.003), and the interval between the first and second D notes was also shorter in small predator alarm call variants (F = 9.043, P = 0.004). Calls that chickadees produced during the large predator presentations tended to have D notes that contained more high-energy peaks above –10 dB (F = 2.855, P = 0.097) spanning a wider bandwidth (F = 2.719, P = 0.105) than those produced during encounters with small predators. D notes elicited by large predators also tended to have more widely spaced overtones (F = 3.385, P = 0.071). No differences were observed in any of the other measured features (P > 0.2 for all).

Do these acoustic differences in mobbing calls transmit information about the potential predator to other chickadees? We conducted playback experiments (18) to test how chickadees reacted to the mobbing calls that they produced in response to different predators by broadcasting variations of the “chick-a-dee” alarm vocalization associated with different predators. We played mobbing calls that flock mates produced in response to a great horned owl (large predator), a northern pygmy-owl (small predator), and control calls of a pine siskin (Carduelis pinus). Chickadees exhibited longer and more intense mobbing behavior when they heard alarm calls recorded in response to a pygmy-owl than when they heard alarm calls recorded in response to a great horned owl or control vocalizations (Fig. 3). They produced significantly more “chick-a-dee” calls in response to playback of mobbing alarm calls elicited by a small predator than they did when presented with playbacks of mobbing calls elicited by a large predator or control vocalizations (Fig. 3A; Kruskal-Wallis K = 11.50, P = 0.003). Chickadees approached the hidden speaker more closely in response to the small predator mobbing call treatment than in response to the large predator mobbing call or control treatments (K = 14.69, P = 0.001); more individuals approached within 3 m (Fig. 3B; K = 14.40, P = 0.001) and within 1 m (K = 11.34, P = 0.003) of the speaker in response to the small predator alarm calls than in response to the large predator alarm calls or control vocalizations. After playback of small predator alarm calls, chickadees also mobbed for longer periods than they did after playback of large predator alarm calls and control sounds (K = 12.69, P = 0.002).

Fig. 3.

Chickadees respond to predator-specific acoustic variations in their mobbing alarm calls. Two behavioral variables were used to quantify chickadees' responses to the three playback stimuli: control sounds (pine siskin calls), “chick-a-dee” calls produced in response to a large predator (great horned owl), and “chick-a-dee” calls produced in response to a small predator (northern pygmy-owl). (A) Boxplots (showing median, interquartile range (IQR), range, and outliers) of the number of “chick-a-dee” calls produced during the first 90 s after the start of each playback treatment. (B) Boxplots of the number of birds approaching within 3 m of the speaker after each treatment. All pairwise comparisons were significantly different (Mann-Whitney U, P < 0.05).

Previous studies have shown that animals produce different antipredator vocalizations for aerial and terrestrial predators. Most of these studies, however, have presented these two types of predators in different ways (1921), potentially confounding the interpretation that prey distinguish between types of predators and not their location or behavior. Our results show that chickadees do not vocally discriminate between raptors and mammals when they are presented in similar ways, and thus the “chick-a-dee” call does not refer specifically to the type of predator.

Instead, these vocal signals likely contain information about the degree of threat that a predator represents. Maneuverability (e.g., as measured by turning radius, or radial acceleration) is extremely important in determining the outcome of predator-prey interactions and is inversely related to wingspan and body size in birds (22, 23). Body size may be a good predictor of risk for chickadees: Small raptors tend to be much more maneuverable than larger raptors and likely pose a greater threat to chickadees.

In addition to being one of the most subtle and sophisticated signaling systems yet discovered, this system is unusual in that it combines aspects of both referential and risk-based antipredator vocalization systems (10, 24, 25). To denote the presence of a rapidly moving predator (e.g., raptor in flight), chickadees produce a “seet” alarm call. When they encounter a stationary predator (e.g., perched raptor), they use the “chick-a-dee” mobbing call. These two vocalizations appear to be functionally referential to the type of predator encounter (i.e., each denotes a specific type of encounter). In addition, we have shown that subtle variation in the “chick-a-dee” mobbing call reflects the size of a specific predator, a characteristic of a risk-based system. Thus, chickadees convey information about predators at two different levels: A coarse level of encoding (“seet” or “chick-a-dee”) signifies the type of predator encounter, and a fine level of encoding (variants of “chick-a-dee”) signifies the degree of danger presented by that specific predator encounter.

The “chick-a-dee” vocalization is remarkably versatile; it is used in many different contexts and apparently conveys many different types of information. The fact that so much information can be transmitted by subtle variations in one type of vocalization raises some fascinating questions about how finely chickadees can discriminate between similar stimuli, and how they categorize different aspects of their environment.

Supporting Online Material

Materials and Methods

References and Notes

Stay Connected to Science

Navigate This Article