Differing Definitions of Falcon Hunting Behavior Metrics Prevent Comparisons of Hunting Success

Variation in the level of efficiency with which predators secure prey can help to reveal patterns in the dynamic processes underpinning predator-prey interactions. Consequently, the hunting behavior of raptors has been the focus of numerous and wide-ranging investigations. These studies have used hunting behavior and hunting success to assess relationships between predator and prey (Cresswell and Quinn 2013), and to better understand risk management decisions made by prey species (Dekker and Ydenberg 2004) and among coexisting predators (Buchanan 2012). A complication in using this body of studies collectively is that there are inconsistencies, sometimes substantial, in terminology and definitions of hunting behavior, as was documented for the American Kestrel (Falco sparverius; Bildstein and Collopy 1987). For example, there are differences in some definitions related to when predators target different individuals, when flocks of prey species split apart when attacked, and in some cases, when there are multiple attacks at the same targeted individual. These differences currently preclude meaningful comparisons among studies as they relate to quantifying the number of hunting flights or capture attempts required to make a capture. My objective is to draw attention to these inconsistencies and the need for greater clarity, precision, and some form of standardization in how hunting efforts and success are documented. I have focused specifically on the study of hunting behavior metrics in falcons (including, but not limited, to Merlin [Falco columbarius], Prairie Falcon [F. mexicanus], and Peregrine Falcon [F. peregrinus]) but this may apply to birds of prey in general. Definitions previously used in the study of hunting by falcons, and their differences, are outlined below in the chronological sequence in which they occurred in the published literature. This is followed by recommendations that may be helpful for data presentation and allow commonality among studies.

Rudebeck (1950) appears to have provided the first definition related to falcon hunting behavior. He stated (Rudebeck 1950:75) that “one hunt may consist of one or several stoops,” and then added that “every new stoop … is nearly always directed against the same individual, which has thus never eluded the pursuer in spite of its efforts. If a predatory bird thus strikes many times in succession I have counted the whole course as only one hunt.” Rudebeck then concluded that if the raptor failed in its attempts to capture one quarry and then began to attack another, without first landing, the subsequent attack would be considered another hunt.

Page and Whitacre (1975) slightly modified the definition of Rudebeck (1950). They indicated that “a hunt consists of one or more completed attempts to seize an individual or group of individuals of a species that the raptor is known to eat. Several attempts at a particular individual or group of prey were considered as one hunt unless they were interrupted by the Merlin’s perching or attacking other prey” (Page and Whitacre 1975:75). This definition differs from Rudebeck (1950) in two subtle but important ways. First, it explicitly acknowledged that the attack could involve not only targeted individuals but also a flock, the latter of which was not mentioned as a possibility by Rudebeck (1950). Second, it indicated that a hunt would be considered to have ended when the falcon landed or directed its attack toward new prey. Of significance is the fact that the definition did not specify what constituted “other prey.”

In 1980, Dekker (1980:372) defined the hunt as “consisting of one or more attempts by a peregrine to seize a bird of a size that I considered potential prey.” He added that a hunt “might include only one swoop or more than 20 aimed at the same bird.” Dekker elaborated that “if the falcon turned away and selected a new target, the preceding hunt was classed as unsuccessful.” Dekker also later indicated that the terms hunt and attack were synonymous (Dekker 2009). This definition of the hunt may have differed to an unknown extent from the definition of Page and Whitacre (1975) in that Dekker considered the hunt to have ended if the falcon directed its attention elsewhere, whereas Page and Whitacre (1975:75), who considered that a falcon turning away toward other prey marked the end of a hunt, also indicated that the hunt included attempts to seize individuals or from a group of individuals, which would likely include individuals that were separated from a flock. The relevance of this subtle distinction between “turning away” and “chasing other individuals that are part of a group” in the Page and Whitacre (1975) and Dekker (1980) definitions is not known.

Toland (1987) did not distinctly define hunting behavior in his study of American Kestrels, but he noted that hunts with unknown outcomes were excluded from analyses.

The definition used by Buchanan et al. (1988), adopted to classify observations of hunting activity beginning in 1979, differed from earlier definitions. They defined a hunting flight as “a single flight involving any number of capture attempts at suitable prey in any number of different flocks,” and then added that “a capture attempt is an attempt to seize or knock down a specific prey individual during a hunting flight” (Buchanan et al. 1988:109). The first difference from earlier definitions was the labeling of the capture attempt, which was used rather than attack, because it was more specific to the action involved. For example, a falcon rapidly approaching a flock could be argued to be attacking, even if it made no capture attempt, a distinction pointed out by Ratcliffe (1980). The second and most substantive difference was that the hunting flight could include any number of capture attempts at any number of targets. Although unstated, the hunting flight would therefore either have a known ending, which would be marked by capture of prey or when the falcon landed, or an unknown ending such as when the falcon was no longer in view.

The definition used by Bijlsma (1990:76) was the same as used by Buchanan et al. (1988), including the definitions for both the hunting flight and capture attempt. In addition, Bijlsma (1990) made two points of clarification. First, “a falcon was not considered to be hunting when it followed a straight flight line and ignored the presence of potential prey.” This comment is consistent with the remark, above, attributed to Ratcliffe (1980) regarding attacks that may not constitute actual hunting. Secondly, the capture attempt “was considered successful when the prey was taken or knocked down for a sufficiently long time to be recovered by the falcon.”

Sodhi et al. (1991:128) followed the definition used by Page and Whitacre (1975), describing the hunting attempt as “one or more strikes at potential prey.”

Cresswell and Whitfield (1994:224) defined an attack as “a rapid, directed flight at a clearly identifiable bird or group of birds (multiple stoops or chases at the same bird and were counted as one attack).” This definition appeared to be like definitions of hunting used in other studies (Rudebeck 1950, Page and Whitacre 1975, Buchanan et al. 1988, Dekker 1988), although unlike the other definitions, it did not specify an attempt to seize prey. The parenthetical statement referring only to a single bird as a target, which was confusing, was eliminated from subsequent publications (e.g., Quinn and Cresswell 2004).

The definition used by Leonardi (1999:124) in a study of cooperative hunting by Lanner Falcons (F. biarmicus) was adopted from Cresswell (1994). Leonardi defined the attack as “very rapid flights or stoops toward one or more clearly observed prey (an individual or group of specific prey species).”

Working in South Africa, Jenkins (2000) defined a hunt as “clearly observed attempts by peregrines to … strike prey” (Jenkins 2000:236); the author also considered attempts to flush prey as attempts to capture but those efforts were possibly limited to raids of nesting birds (Jenkins and van Zyl 1994). The wording of this definition suggests that a hunt consists of a flight that includes strikes at prey, but it is also possible that the definition refers only to strikes, like the use of attempt by Page and Whitacre (1975). Although Jenkins recognized that some attacks or strikes may have been done as play, as practice, or even to assess the flight capability of intended quarry, he did not change the definition of hunting but rather classified the observations into categories for analysis purposes. Given that his analyses did not include hunts “which were completed out of view” indicates that he considered them to be consistent with hunts with an unknown outcome.

DeCandido and Allen (2006:54), working at the Empire State Building in New York, defined a hunting attempt as when “a Peregrine Falcon approached to within 1 m of its intended prey.” They added that “repeated stoops at the same prey represented,” although unsuccessful in a capture, “a separate hunting attempt.” The definition of hunting attempt here appears to be the same as used by Sodhi et al. (1991), the use of strike in Rudebeck (1950), completed attempts by Page and Whitacre (1975) and Dekker (1980), and capture attempts as defined by Buchanan et al. (1988). DeCandido and Allen (2006) did not propose a term for the continuation of the hunting effort, as was found in the other definitions described here.

Zoratto et al. (2010:428) defined hunting behavior as “a purposeful flight in an area of potential prey in a manner that led, or could lead to, the capture of a prey.” They classified hunting behavior into two categories: the hunting sequence and the attack. “The term hunting sequence was defined as a single flight involving any number of attacks to the same target, a single bird or a flock.” An attack was defined as “an attempt to seize a specific prey individual during a hunting sequence. One or more attacks on a flock and subsequent pursuit of that flock or part of it were counted as one hunting sequence.” If the hunting effort was discontinued, but “later targeted the same or a different flock, this was counted as a second hunting sequence” (Zoratto et al. 2010:428).

Basso et al. (2021) adopted the definition of Cresswell and Whitfield (1994) for hunting, with one difference. They indicated that an “attack was considered finished when we observed the Peregrine Falcon moving away from shorebirds. If another attack was observed some minutes later (minimum time-lapse between two consecutive attacks considered independent, 5 min), it was considered a new attack” (Basso et al. 2021:13382). The difference from all other definitions, including that of Cresswell and Whitfield (1994), was that Basso et al. included a minimum time threshold of 5 min before a subsequent attack would be considered a separate event. This temporal provision was like the behavioral provision (e.g., the target flock landed) used by Zoratto et al. (2010) in that both used a definable value to imply independence between events.

There are subtle and substantial differences across this group of definitions. The minor differences largely reflect novel terminology for the same or similar actions. An explanation was provided by Zoratto et al. (2010), who pointed out that their use of hunting sequence was consistent with the use of attack by Cresswell and Whitfield (1994) and hunt by others (Rudebeck 1950, Page and Whitacre 1975, Dekker 1980, 2009). Furthermore, their definition of attack was analogous to capture attempt (Buchanan et al. 1988, Buchanan 1996) and to hunting attempt (DeCandido and Allen 2006).

The most substantive difference in definitions involves what constitutes a hunting flight (or a hunting sequence as defined by Zoratto et al. 2010), although some definitions (e.g., DeCandido and Allen 2006) account only for capture attempts and do not define a hunting flight or hunting sequence. In nearly all definitions an act of hunting was classified as a new (separate) episode if the target of the hunting falcon changed (e.g., the falcon redirected its focus to new prey), or if the target landed (Zoratto et al. 2010), or a specific period elapsed between attacks (Basso et al. 2021). In Buchanan et al. (1988), the hunting episode (i.e., hunting flight) continued until the falcon captured prey or landed, or in the case of a hunt with unknown outcome, the falcon was lost from view, or the outcome of the effort could not be determined.

Buchanan et al. (1988) used their definition of the hunting flight because it had an unambiguous end point (i.e., a capture was made or the falcon landed). Merlins and Peregrine Falcons will often follow shorebird flocks for substantial distances, attacking them repeatedly. These falcons, and particularly Prairie Falcons, will also fly away from targeted prey (solitary individuals or flocks) only to gain height, distance, or a new element of surprise in a subsequent attack, sometimes targeting the same prey. Merlins and Peregrine Falcons also break off attacks at one flock and target another flock that may be nearby, and sometimes these other flocks were part of the original flock that had been targeted but then split into two or more flocks when initially attacked by the falcon (Buchanan et al. 1988). In addition, when flocks split apart, single birds would often fail to remain with a flock and for varying amounts of time these solitary individuals would evade an attacking falcon as they attempted to rejoin a flock. Buchanan et al. (1988) saw no logical way to differentiate between these different circumstances, all of which are common, without making judgments that seemed subjective, and for that reason they elected to identify the hunting flight, which was in essence a flight from perch to perch (or perch to capture), unless the falcon was already underway when first observed, and that to be a hunting flight, there must be at least one attempt at prey capture.

The differences among definitions are significant because they have the potential to result in divergent interpretations of hunting activity (i.e., what constitutes a hunt), which therefore influences hunting success estimates. This is particularly problematic because the situation is not resolved merely by using one value or another (e.g., strike or hunt or their seeming equivalents) to compare across studies. For example, in a hypothetical hunting event, suppose a falcon makes three capture attempts in quick succession at the same solitary bird, then diverts to make a capture attempt at a flock, followed by another capture attempt at a second flock, and then culminates the hunting flight with three more capture attempts in quick succession of a solitary bird, the final attempt resulting in capture. That episode would be classified as a single hunting flight under one definition (e.g., Buchanan et al. 1988, Bijlsma et al. 1990), with a resulting hunting success of 100%. The same situation under other definitions could result in recognition of four hunting flights (the three on solitary individuals [hunt #1], the first flock [hunt #2], the second flock [hunt #3], and the attempts at the solitary individual [hunt #4], the last of which was successful), which would amount to 25% success if it was concluded that the falcon had redirected its attention elsewhere as indicated in the parenthetical explanation. To add another hypothetical observation to the one just given, let’s suppose the exact same sequence of behavior was noted on a second occasion, but this time with no capture of prey. The cumulative success according to the Buchanan et al. (1988) definition would be 50% (one capture in two hunts), and under certain other definitions, the cumulative hunting success for both episodes would be 12.5% (one capture in eight hunts). Finally, a third hypothetical example involves the same exact situation as above, except the falcon did not capture prey and was lost from view after the final interaction with the target prey. According to the Buchanan et al. (1988) definition, this third episode would be a hunting flight with an unknown ending and would not be used to calculate success, meaning the cumulative success would remain at 50%, based on hunts with known outcomes. According to some of the other definitions that include a “turning away” or “redirection” clause, this latter episode would be considered three or possibly four hunts, and the cumulative success would then be calculated at 8.3% (one success in 12 hunts). That the success values above do not differ or change in a uniform way across the differing definitions illustrates that there is no logical formula to compare between studies and data sets that use different definitions. Similarly, attacks at flocks or solitary individuals that do not include an actual attempt to capture prey may not be analogous with capture attempts in some of the definitions above.

There is no easy remedy to resolve the differences in definitions, although acknowledging the differences would be helpful. In focused (e.g., Roalkvam 1985, Bildstein and Collopy 1987) or general syntheses of falcon ecology and behavior (e.g., Sale and Watson 2022), especially where some authors may not be familiar with specific details of the definitions mentioned above or elsewhere, understanding these complicating factors would help to explain contrasting results of different studies. In that vein, with respect to Merlins, Prairie Falcons, Peregrine Falcons, and perhaps other raptor species, I offer several recommendations for consideration.

The collective databases developed by authors cited above may involve over 10,000 hunts or hunting flights, however they were defined. Reevaluating the complete record of all hunting flights would be a daunting task and for many records it would not be feasible to reclassify events to conform to another definition because of how the data were initially recorded. Reconstructing only parts of the collective database across studies would likely introduce bias (e.g., by including only records with one or two capture attempts and excluding those with more complex behavior). To maintain consistency with existing study themes through time, definitions used previously by individual researchers or research teams should continue to be used to facilitate direct comparison and interpretation in related studies. In all publications, I recommend that definitions of hunting activity be clearly and thoroughly described, and an attempt made to eliminate unstated assumptions. It is certainly appropriate for researchers to indicate that some ecological comparisons across studies are not possible because of the differing definitions used.

Uncertainty across studies could be partly alleviated in the future if each observation of hunting was documented in sufficient detail to allow for comparisons with other datasets. An approach to this would involve, for any given study, recording data to accommodate all definitions of hunting. To this end, an example might be to record, for all observations: (1) whether or not the attack at quarry (perhaps with a minimum proximal distance used to classify the event as an attack; e.g., an approach within 20 m) involved a capture attempt, (2) the number of capture attempts made at targeted quarry, (3) whether the falcon involved in the hunting effort (following ≥1 capture attempt) veered away to a new target (this would need to be defined consistently across studies; see below), (4) whether the falcon involved in the hunting effort (following ≥1 capture attempt) veered away and did not obviously seek new quarry (i.e., the falcon departed the area), (5) whether a new attack or hunt ensued, which reinitiates the list of options just described, or (6) that the hunting flight, however defined, had ended and was classified according to capture, failure, or an unknown outcome. The time element of Basso et al. (2021) could also be added here. The distinction between item (4) and item (6) is that under some definitions, item (4) would indicate the end of a hunt, which would be included in calculation of hunting success, whereas under the definition of Buchanan et al. (1988) it would indicate a hunting flight with an unknown outcome and would not be used to measure success.

A challenge in finding commonality, given the reasons outlined above, will be to determine what suite of conditions would constitute a falcon that has shifted to a new and distinctly different quarry (e.g., item (3) in the previous paragraph), thus signifying the end of one hunt and the possible beginning of another. To achieve commonality, the fact that different prey species are targeted could be a logical distinction to differentiate between hunts, although with an important caveat. For example, shorebirds often occur in mixed species flocks and a falcon chasing individuals of two different species in succession that came from the same flock would not necessarily represent a changed focus by the falcon, as such a distinction seems ambiguous given that the birds came from the same flock. In contrast, a falcon that hunted shorebirds and then diverted to attack ducks or pigeons would more clearly represent a distinct difference, as these species have lesser associations with one another. This standard, however, would apply only to a small percentage of the ∼2800 hunting flights I have witnessed involving Merlin, Prairie Falcon, and Peregrine Falcon, which suggests that in certain contexts (e.g., certain ecological systems or habitats) this factor may have little value by itself. Potentially, bird-mounted camera technology may enable us to see what the falcon is considering as possible targets for attack and thus illuminate how we define hunting success, although when discontinuing an attack or pursuit and redirecting their attention, falcons commonly return to another target they had obviously noted earlier (J. Buchanan unpubl. data).

Although I have focused on definitions for terminology around hunting behavior (e.g., hunting flights, capture attempts) when quantifying hunting success, broader issues also must be considered depending on the scope of the investigation. Incorporating the concept of energy expenditure and intake (e.g., number of prey items, mass or caloric value per unit time; Warkentin and West 1990, Fargallo et al. 2020) would further the applicability of hunting success and enable comparisons within and between species. Likewise, it may be important to account for factors such as the occurrence of searching for prey from a perch versus hunting from the air, to create a more broadly applicable metric. Lastly, consideration must be given to a potential lack of statistical independence of events associated with calculating hunting success and how a researcher accounts for each prey individual or flock pursued, and each individual predator observed. Robust definitions of the elements of hunting behavior should facilitate meaningful estimation of hunting success and the application of appropriate statistical analysis.

The diversity of hunting strategies across the Falconidae creates challenges for developing a universally applicable set of definitions. Nonetheless, multiple perspectives would likely enhance our collective understanding, so I propose that a group of researchers in this field (e.g., behavioral ecologists and data analysts) be convened to specifically discuss the best means to achieve commonality (i.e., what needs to be measured and how) across all or a subset of the existing or potential studies of falcon predation. That initiative could also identify the most appropriate approaches to analyze data involving not only the hunting flight, but subsets of the hunt such as capture attempts, and thereby accounting for situations where lack of statistical independence should be addressed. It is to our collective benefit that, to enhance understanding of falcon hunting behavior across a broad range of ecological conditions, we use definitions with enough commonality that comprehensive and meaningful syntheses become possible.