Studies in Avian Biology No. 13:144-160, 1990.

A CLASSIFICATION SCHEME FOR FORAGING BEHAVIOR OF IN TERRESTRIAL HABITATS

J.V. REMSEN,JR. AND SCOTT KROBINSON

Abstract. Studiesofavian foragingbehavior in terrestrialhabitats sufferfrom a lack of standardization in the kinds of data gathered and in the terminology used to classify different activities. These incon- sistenciespartially reflect the variety of questionsasked about foraging.If a standardterminology were used, then data on foraging behavior could be included among the standard data (e.g., clutch size, body weight, and mating system) routinely recorded for the biology of a species. We propose a system for gathering foraging data for landbirds in which the five basic, sequential componentsofforaging (search,attack, foragingsite, food, and food handling) are quantified separately. Data on searchingbehavior involve measuring continuous variables and are particularly critical for studies of energetics. The “attack” component is most in need of standardized terminology. The system that we propose separatesthe attack perch from the attack maneuver, and further subdivides the maneuvers into near-perch, subsurface,and aerial maneuvers. Each of these general categoriesis further subdivided according to details of attack movements and ways in which substratesare ma- nipulated. Data on attack methods are primarily useful for studies of ecomorphology, but may also be important in bioenergeticand community-level studies. Quantifying the foraging site involves measuring the following variables: general habitat (location in a study area), vertical position, foliage density, and substrate.Although identification and quanti- fication of foods taken in the field is difficult, it can provide valuable information on food size (and taxon for largeritems). Dietary data from stomachsamples are usefulfor studiesof resourcepartitioning when they show dramatic differences,but overlapping diets do not necessarilyindicate that two birds forage in the same way. Food-handling behavior is seldom measured in the field, but is valuable in studiesof optimal foraging behavior and ecomorphology. Intercorrelations between each of these aspectsof foraging can be determined from standard multi- variate analyses.How finely to subdivide categoriesdepends upon the kinds of questionsbeing asked. Key Words: Foraging behavior classification;foraging maneuvers; search;attack; foraging site; diet; food-handling; glean; sally; probe; manipulate.

There are almost as many ways of classifying quantify foraging. For example, the “hawk” cat- and quantifying foraging behavior as there are egory of Sherry (1979) and Holmes et al. (1978, papers on the subject. In part, this variety reflects 1979b) includes attacks on prey that were the fact that no two species or groups of species flying when first seen, attacks on prey flushed forage in exactly the same way, and that no two from foliage by the foraging activities of the bird, habitats present exactly the same foraging op- and chases after prey that the bird attacked and portunities. It is difficult, for example, to quan- missed. Later papers by Robinson and Holmes tify the foraging methods of bark-foragers in the (1982, 1984) and Sherry (1983, 1984) however, same way that one quantifies the foraging of fo- showed that differences among these aerial ma- liage-foragers (Jackson 1979). Another factor neuvers have important implications for diet. contributing to the lack of standardization is that Remsen (1985) showed that the fine details of different kinds of questions often require differ- substrate use (e.g., dead leaves, moss) that are ent kinds of data. Many studies that focus on often ignored in many community studies were resource partitioning record only the details of particularly important in resource partitioning in foraging site selection and omit data on search a tropical bird community. Rosenberg (this vol- and attack movements (e.g., Hertz et al. 1976). ume) further showed that even within a group as In contrast, studies of ecomorphology emphasize specialized as dead-leaf foragers, species differ in prey-attack methods (e.g., Osterhaus 1962; Fitz- the kinds of suspended dead leaves that they patrick 1980, 1985) whereas bioenergetic and search. The classification of foraging behavior, optimal foraging studies emphasize searching therefore, is more than a semantic problem: one movements as well as prey handling (e.g., Sherry can reach different conclusions simply by clas- and McDade 1982, Robinson 1986). Even stud- sifying foraging methods differently. ies addressing the same questions in ecologically In this paper, we propose a system for mea- similar birds do not always measure the same suring and classifying foraging behavior for non- variables. raptorial landbirds. Our goal is to standardize This lack of standardization, however, reflects data-gathering methods and terminology to per- fundamental inconsistencies in the importance mit among-site and among-species comparisons attached to the individual variables used to that are currently handicapped by the absence of

144 CLASSIFICATION OF BEHAVIOR--Remsen and Robinson 145 a common vocabulary. If some standard system volume). We discuss briefly data on bird diets, and terminology were used by those studying and also propose a standard terminology for food- foraging behavior, then we could ask questions handling techniques. The final section of our pa- concerning the frequencies of various behaviors per deals with some of the ways in which data among communities. Such comparisons may can be analyzed to address questions of resource- provide important insights into community or- partitioning, bioenergetics, and ecomorphology. ganization: in a field such as foraging behavior, in which community-wide experimental manip- SEARCHING BEHAVIOR ulations are difficult or limited, a comparative Searching behavior includes those movements used approach among species and communities may to search for food or substrates that hide food; under be the only method available for testing many our definition of searching behavior, “search” ends hypotheses. once food or food-hiding substrates are sighted and The system presented here separates five se- attacked. Searching methods have usually been quan- quential components of foraging behavior, each tified by recording the lengths and rates of movements of which is quantified separately when data are between perches and the time intervals between move- gathered in the field: (1) search (movements lead- ments (e.g., MacArthur 1958, Cody 1968, Williamson ing up to sighting of food or food-concealing sub- 197 1, Fitzpatrick 198 1, Robinson and Holmes 1982). Other variables are: (1) distance covered per unit time; strates); (2) attack (movements directed at food (2) number of stops and time-spent-stopped per unit item or the substrate that conceals it once sight- time; (3) number of attacks (and % successful) per unit ed); (3) foraging site (including general location time; (4) direction ofmovement between stops, in three and specific substrate); (4) food (including type dimensions if appropriate; (5) sequential distribution and size); and (5) food handling (after food item of locations of stops (to calculate return rates to pre- obtained). We recognize that many of these com- vious stops). Between-foraging-site movements can be ponents are not necessarily independent, but we categorized as: (1) walk, (2) hop, (3) jump (leg-powered prefer to quantify each separately and to allow leaps that cover more space than the typical hop), (4) subsequent analyses to show intercorrelations. run, (5) climb (with notations on whether or not the tail is used as a brace), (6) glide, (7) flutter, and (8) fly. At the outset, we recognize that any classifi- Robinson and Holmes (1984) further distinguished be- cation system inflicts typology on what may be tween hops in which American Redstarts (Setophagu gradients of behavior, but we see no other prac- ruticilla) fanned their tails and lowered their wings, tical solution for organizing foraging informa- and hops in which there were no extra movements. tion. Our goal is to distinguish among what we Some birds also use their wings for support when hop- subjectively perceive to be functionally different ping on thin, weak perches, a movement that could be categories. By using standardized terminology, called a “flutter-hop” (Robinson, unpubl. data). Also data on foraging behavior can be included among of interest are postures during searching that are sel- the standard biological data reported for bird dom qualitatively described (for exceptions, see any E. 0. Willis reference) or quantified but that may have species. At present, reference works on birds, subtle morphological correlates. Postures are particu- which typically include detailed, quantitative data larly difficult to categorize because most species move on variables such as clutch size, body weight, and change postures frequently, and because head, wing, molt and migration schedule, and mating sys- and tail orientations are all simultaneously involved. tem, tend to omit, or describe in superficial, qual- Perhaps the advent of telephoto video-cameras will itative ways, all aspects of foraging behavior ex- permit such analyses; in this paper we deal with pos- cept perhaps diet. Foraging data should be tures only peripherally. More amenable to quantifi- included in such reference works, because for- cation are changes of orientation while searching from aging behavior is an integral part of a species ’ a perch, either with head or body movements. For example, many species have characteristic side-to-side biology, and becauseit relates to time-activity bud- movements, whereas others maintain a straight-ahead gets, morphology, habitat selection, and social orientation. Many species have characteristic wing- system. The foraging behavior of many species flicking or tail-wagging movements that accompany may also be as “typical” of that species as any foraging, the significance of which is not often under- other aspect of its biology. A standard vocabu- stood: the frequency of such actions could also be quan- lary will eventually allow a more sophisticated tified.There‘ is alsoa parallel literature on the searching review of the prevalence of various foraging be- behavior of various lizards (Moermond 1979, Huey haviors in birds; such a review will be able to and Pianka 1981). replace the often anecdotal, “who-does-what” In general, searching movements of birds form a approach with quantitative comparisons be- continuum from “active” to “passive” modes (Eck- hardt 1979). Active foragers change perches at a high tween taxa and regions. rate, including many hops (or steps in species that walk), The heart of this paper is the section on the and most flights are short. Passive foragers seldom attack, which is the phase of foraging that is most change perches, but fly long distances when they do in need of a standard terminology based on func- move. The subset of birds that Eckhardt (1979) chose tionally different categories (cf. Moermond, this from the community that he studied fit into active 146 STUDIES IN AVIAN BIOLOGY NO. 13

(primarily wood-warblers) and passive (primarily ty- Many birds steal food from other birds, and some rannids) foragers. A subsequent study of a different speciesrely on this tactic ()for locating forest bird community (Robinson and Holmes 1982) and capturingfood (see Brockman and Barnard [ 19791 however, found that many speciesdid not fit cleanly for review). Finally, some landbirds form mutualistic into either category.Tanagers and , for example, food-searching associationseither with conspecifics(e.g., were intermediate in their rates of hopping and flying Kilham 1979, Mindell and Black 1984) or other bird and in the lengthsof their flights. Another species,the species(Jackson 1985), especiallywithin mixed-species American Redstart appearedto add many movements flocks (e.g., Munn 1986). These sorts of associations of its wings and tail designedto flush prey to its already should always be recorded. active foraging behavior. Furthermore, the searching movements ofthe Black-cappedChickadee (Punts atri- ATTACK BEHAVIOR capillus)depended upon the distribution of whatever We define the “attack” phase as that portion of for- substrate(e.g., dead leaves)an individual of this species aging behavior from the moment when a food item, was searchingat the time. Searchingmovements, there- or food-concealingsubstrate, is sightedto the moment fore, cannot necessarilybe categorizedtypologically in when a capture attempt is made. Thus, we include the same way as attack maneuvers (see below). We within “attack” phase those behaviors aimed at dis- agreewith the approachof most authors who quantify lodging or revealing food before it is sighted, such as and present data on searchingmovements separately various kinds of substratemanipulation (e.g., flaking, (e.g., Robinson and Holmes 1982, Landres and hammering, gaping). We further subdivide the attack MacMahon 1983, Holmes and Recher 1986b). phase into (1) perch and (2) maneuver. Many birds use food-searching methods that are Few studies have quantified parameters concerning similar to attackmethods (see below). Robinson (1986), the characteristicsof the perch from which an attack for example,quantified the rate at which Yellow-rumped is launched.The numerous studiesby E. 0. Willis (see Caciques(Cucicus da) used the following searchtac- references)have shown that a species’ presence in a tics: “probe-searches,”in which an individual searched particular habitat or microhabitat may be determined a dense cluster of leaves; “hover-searches,” in which in part by availability of suitableperches. Certain species an individual searched the tips of foliage while hov- may also specialize on perch types not used by other ering under them in stationary flight; and “hang- species.For example, two small tanagers(Hemispingus searches,” in which an individual searched a nearby xanthophthalmusand H. verticalis)that characteristi- substrate while suspended below it. If any of these cally search the uppersidesof leaves in dense clusters searchingtactics led to a prey capture, then they were do so by perching on the clusters themselves (Parker quantified separately as prey attack maneuvers. The and O’Neill 1980; Parker et al. 1980, 1985). Several only differencebetween a “probe-search” and a “probe- speciesin the vireonid Hylophilus characteris- attack” (see below) was in whether or not prey was tically graspthe margins of leaves for perchesto reach located. This illustrates a problem in distinguishing the undersidesof these leaves (T. A. Parker and JVR, between searchingand attack methods in speciesthat unpubl. data). Furthermore, studies such as those by search for concealed food: any classification system (1976a; cf. Leisler and Winkler 1985) have inevitably includes data on both searchingand attack- revealed important morphological adaptations asso- ing behavior. , for example, both search ciated with particular perch types. for and attack prey by removing an outer layer of bark. Perchtype can be quantified usingthe same variables We have chosen to classify these methods as attacks as thoseused in our schemefor “substrate” (seebelow). (seebelow) on food-concealingsubstrates. Distinguish- In practice,most measurementstaken for the substrate ing between search and attack phases, however, rep- will be the same as those for the perch except for the resentsan unresolved issue in some parts of our clas- details of perch angle and diameter; therefore, the in- sification scheme,especially for those speciesthat peer creasein volume of data to be recorded in the field is closelyand unambiguouslyat particularsubstrates, such minimal. Those speciesthat search while moving do as curled leaves,leaf tops, and crevices.Although these not really have a “perch” per se; for instance, some speciessearch a particular substratein a manner that species search and attack while in continuous flight is analogousto any of the substrate-manipulationma- (“screening,” see below) or while hovering (e.g., Say’s neuvers outlined in the attack maneuver section (see Phoebe [Suyornis suyu], Grinnell and Miller 1944; below), our scheme does not include “peering” or [Sialia spp.], Power 1980; and RestlessFly- “scanning” as an attack maneuver. In the field, we catcher [Myiagru inquieta], Ford et al. 1986). record substratesthat are unambiguouslysearched with Our classification of attack maneuvers categorizes the notation “visual search,” but we are uncertain as them with respectto the complexity and required agil- to how to include suchdata in analyses.Certainly, such ity of each behavior. For example, we assume that visual searchesare important in analysesof substrate aerial maneuvers and substratemanipulation require use. greater agility and more energy than those maneuvers Severalother kinds of searchbehavior are sufficient- in which a food item is removed from a substratenext ly distinct to warrant separatetreatment. Many birds to the bird’s perch. Our classificationalso attempts to follow disturbancesthat exposeprey; suchdisturbances remove where possiblethe influence of substrate;thus, include fires, other animals (particularly other bird foraging motions that appear similar, but are directed species,ungulates, monkeys, and army ants), and hu- at different substrates,are grouped together. Such sim- mans and their machines. Many recent studies place ilarities may be superficial, and foraging motions are disturbance-followers in a separate (e.g., Karr certainly influenced by the types of substratesat which 1980, Terborgh 1980a, Terborgh and Robinson 1986). they are directed. Nevertheless, we prefer to group to- CLASSIFICATION OF BEHAVIOR--Retmen and Robinson 147

TABLE 1. PROFQSEDCLASSIFICATION SCHEME FOR In general, studies of attack behavior in landbirds ATTACK METHODSOF THE FORAGINGBEHAVIOR OF have distinguishedthe conspicuousaerial maneuvers NON-RAPTORIALLANDBIRDS, WITH SYNONYMSOR from other attack behaviors, but the nonaerial attack PRESUMEDSYNONYMS FROM OTHER STUDIES (SEE TEXT) maneuvers have often been lumped in one category, usually“glean.” Such merging of nonaerial maneuvers into one or a few categoriesmight obscureimportant I. Near-uerch maneuvers behavioral differencesamong speciesthat have impli- A. Surface maneuvers cations for adaptive morphology (e.g., Richards and 1. Glean (= pluck, perch-glean, pick) Bock 1973, Partridge 1976a,Leisler and Winkler 1985), 2. Reach (= stretch) searchtactics, niche overlap, and food selection. a. Reach-up (= ) Our classification(Table 1) does not include certain b. Reach-out maneuvers that appear to be rare, such as digging in c. Reach-down (= lean, -under) 3. Hang (= hang-glean) ground by using a strongly curved bill in a hoe-like a. Hang-up (= hang vertical, hang motion (Engels 1940) using spines and twigs as tools for probing (reviewed by Boswall 1977), usingthe head head-up, vertical clinging) b. Hang-down (= hang head-down) as a brace to provide leverage for foot scraping c. Hang-sideways(= hang-side, (DeBenedictis 1966, Kushlan 1983) or as a buttressto vertical clinging) move or dislodgesubstrate (Keast 1968), vibrating feet d. Hang-upsidedown (= hang to startle prey in leaf-litter (Hobbs 1954, Wall 1982 as horizontal) cited by Edington 1983), rustling leaf-litter to startle prey (Potter and Davis 1974), and crushingtwigs with 4. Lunge (= dart, rush) the bill to exposeprey therein (Mountainspring 1987). B. Subsurfacemaneuvers: no substrate manipulation The system in Table 1 can be expanded as needed to include any such rare behaviors. 1. Probe Many studiesof foraging behavior that make inter- C. Subsurfacemaneuvers: substrate manipulation specific comparisons, or intraspecific comparisons among seasonsor habitats, have presented their data 1. Gape in the form of a diversity index and have not included 2. Peck (= tap) the original data with percentage of observations in 3. Hammer (= drill) 4. Chisel each foraging category. Other studies have identified the number of speciesin a community associatedwith 5. Flake (= bill-sweep, toss) various foragingcategories, but have neglectedto iden- 6. Pry tify which speciesbelong in each category. We think 7. Pull that the original data themselves should be presented 8. Scratch to facilitatecomparisons with other studies;they should II. Aerial maneuvers at least be published as appendices. A. Leg-powered maneuvers An outline of the categorieswith definitions of each 1. Leap (= jump-glean, jump). Include leap-distanceand leap-angle. attack behavior follows. Some categoriesare not mu- tually exclusive. For example, a bird that “sally-hov- B. Wing-powered maneuvers ers” might also “probe” while hovering. Therefore, 1. Sally (= hawk, snatch, flycatch, hover-glean, hover). Include many attack maneuvers can have compound names, sallv-distanceand sallv-anale. such as “sally-hover-probe” or “reach-out-gape.” Each maneuver category is accompanied by some a. Sally-strike (= outward-strike, examples from the literature. Our literature survey is upward strike, snatch) intended to be illustrative rather than encyclopedic. b. Sally-glide We tend to cite examples from recent, quantitative C. Sally-stall (= hover, hover- studies, rather than older, more qualitative material. glean) Although the descriptive sectionsof the latter are often d. Sally-hover (= hover, hover- superior,much of the older material is contained with- glean) in more general life-history studies and is therefore e. Sally-pounce (= land-and-glean, more difficult to locate. pounce, dive-glean) In choosing a standard vocabulary, we have at- 2. Flutter-chase tempted to use simple, descriptive terms, which, if 3. Flush-pursue possible,are already frequently used in studies of for- 4. Screen (= hawk) agingbehavior; we have not hesitatedto “synonymize” many favorite terms, including many of our own.

I. Near-perch maneuvers (target food item can be gether similar-appearing behaviors to alert morphol- reached from bird’s perch) ogiststo these potential similarities, rather than to al- A. Surface maneuvers low the substrate category to separate automatically 1. Glean: to pick food items from a nearby suchbehaviors. Becausethe systempresented here also substrate,including the ground, that can be requires that the substratealso be recorded, no infor- reachedwithout full extensionof legsor neck, mation is lost by excluding substratesfrom the behav- no acrobatic movements are involved. Em- ior categories. len’s (1977) and Mountainspring’s (1987) 148 STUDIES IN AVIAN BIOLOGY NO. 13

“pluck,” Fitzpatricks’ (1980) “perch-glean,” branches (Forshaw and Cooper 1979). At and Moermond and Denslows’ (1985) and least one (Metalluru ty- Remsens’ (1985) “pick” are synonyms. Per- rianthina) uses reach-down maneuvers haps the majority of maneuvers performed to reach more than a third of its flowers by most foliage- and ground-searching birds (Remsen 1985). are “gleans.” For example, 5 1% of the forest 3. Hang: to use legs and toes to suspend the species studied in the Andes by Remsen body below the feet to reach food that can- (1985) and 53% of the forest species studied not be reached from any other perched po- in Australia by Ford et al. (1986) used glean sition. “Hang-glean” of Recher et al. (1985) as their principal foraging maneuver. Be- and Robinson (1986) is a synonym. Differ- cause gleaning is presumably the least costly ences in frequency of use of “hang” among maneuver in terms of energy expenditure, similar species may have subtle conse- it is not surprising that it is used so fre- quences for morphology (Partridge 1976a, quently (Remsen 1985; Moermond, this Leisler and Thaler 1982). use “hana” volume). frequently (Forshaw 1973 and references 2. Reach: to extend completely the legs or neck therein). Chickadees and titmice (Paridae), upwards, outwards, or downwards to reach bushtits (Aegithalidae), and some thornbills food (after Moermond and Denslow 1985, (Acanthizu) frequently “hang” to reach un- Remsen 1985). Because most studies sel- dersides of branches and leaf tips (e.g., Gibb dom distinguish “glean” from “reach,” the 1954: Root 1964. 1967: Grant 1966: Stur- frequency with which “reach” maneuvers man 1968; Partridge 1976b; Rabenold 1978; are used is not generally known. Strong in- Moreno 1981; Alatalo 1982; Bell 1985b; terspecific differences among congeners in Recher et al. 1985, 1987; Laurent 1986). ability to reach are associated with mor- The Palm Tanager (Thraupispalmarum) in phological differences (Snow and Snow 197 1, Trinidad “hangs” almost exclusively when Moermond and Denslow 1985). Some fru- searching for insects in foliage (Snow and givores, especially toucans and some tana- Snow 1971). The Blue-backed Conebill gers, obtain their food by reaching (Moer- (Conirostrum sitticolor; Thraupinae) also mond and Denslow 1985). Morse (1967b). uses this maneuver as its primary means of who distinguished “stretching”-which is attack (Remsen 1985). Other insectivores probably equivalent to our “reaching”- that use “hang” regularly include: Rufous- from gleaning, found that it was used sel- browed (Troglodytes rufociliatus; domly (O-5% of all maneuvers) in the six Skutch 1960), some wood-warblers (Root species of wood-warblers studied. Three 1967, Fickenand Ficken 1968, Elliott 1969, further subdivisions may be made with re- Andrle and Andrle 1976, Rabenold 1980), spect to direction: Speckled Tanager (Tanguru guttata; Snow Reach-up:to reach above the bird. This and Snow 197 1); some white-eyes (Zoster- is synonymous with the “crane” of ops;Gill 197 1, Earl& 1983); Ruby-crowned Greenberg (1987b). This maneuver is (Regulussatrupa; Rabenold 1978); used especially frequently to pick prey the fumariid Siptornis striuticollis(Eley et from undersides of leaves. The Pale- al. 1979); and Sharpbill (Oxyruncuscrista- legged Warbler (Basileuterussign&us) tus; De L. Brooke et al. 1983, Stiles and uses this motion, along with the next, Whitney 1983). Some vireos “hang” when more frequently than any other maneu- grasping the margins of leaves to reach food ver (Remsen 1985). that cannot be reached from branches ( Reach-out:to reach lateral to the bird. A griseus,Nolan and Wooldridge 1962; V. maneuver used especially frequently to huttoni and V. gilvus, Root 1967); several pick prey from nearby leaves and tropical vireos (Hylophilus)use this maneu- branches. ver frequently if not predominately (Green- Reach-down:to reach below the plane of berg 1984a; T. A. Parker and JVR, unpubl. the feet. This is synonymous with the data). Many species that extract prey from “lean” of Greenberg (1987b) and prob- hanging dead leaves “hang” (and “reach”) ablv the “duckina-under” of Rabenold to investigate isolated dead leaves (Skutch (1980). This man&ver is used by many 1969 for Automolus ochrolaemus;Green- tanagers, especially Tanguru, when for- berg 1987b; K. V. Rosenberg, unpubl. data; aging on branches (Snow and Snow 197 1; JVR, unpubl. data). Among frugivores that Skutch 1981; Parker and Parker 1982; “hang” to reach fruit are Euphonia violucea Remsen 1984, 1985; Hilty and Brown (Snow and Snow 197 1) and two species of 1986; Isler and Isler 1987); tanagers often woodpeckers (Moermond and Denslow reach-down alternately on opposite sides 1985). Several “hang” to of a branch as they move along the reach flowers (Parker and ONeilll980,’ Par- branch, as does the wren Odontorchilus ker and Parker 1982, Parker et al. 1985, brunt&ii (Parker et al. 1980). A bird-of- Remsen 1985). Four types of “hang” ma- paradise (Purotiucarolue) apparently uses neuvers (Fig. 1) should probably be distin- a similar maneuver when searching guished (modified after Partridge 1976b, CLASSIFICATION OF BEHAVIOR--Remsen and Robinson

Rabenold 1980, Alatalo 1982, Earl& 1983, and Greenberg 1987b): a. Hang-up: to hang, head-up. b. Hang-down: to hang, head-down. This differs from reach-down only in that the bird is clinging to a vertical surface or side of a horizontal surface, rather than perching on the upperside of a surface in reach-down. There is probably a contin- uum between the two maneuvers, and in fact, Moermond and Denslows’ (1985) “reach” would include maneuvers here considered to be “hang-down.” c. Hang-sideways:to hang on the side of a substrate with body axis parallel to the ground and with the birds’ side oriented upwards. d. Hang-upsidedown:to hang, belly-up, on underside of horizontal or diagonal sur- face.

These same four categories may also be applied to the foraging behavior of all specialized bark-foraging birds (e.g., woodpeckers, dendrocolaptids, certhiids) that characteristically hang while searching and attacking; in bark-foraging birds, these maneuvers are probably best considered postures rather than maneuvers.

4. Lunge: those maneuvers in which the food FIGURE 1. “Hang” maneuvers: (a. 1) = “hang-up” item is beyond the range of “reach,” but on vertical perch, (a.2) = “hang-up” on horizontal perch, rapid leg movements rather than flight are (b. 1) = “hang-down” on vertical perch; (b.2) = “hang- used to approach and capture the prey. This down” on horizontal perch; (c) = “hang-sideways”; (d) is synonymous with the “lunge” of Green- = “hang-upsidedown.” Drawing by Donna L. Ditt- berg (1984a), except that Greenbergs’ lunge mann. would include movements that we call “reach-out.” Roots’ (1967) “rush” is a com- bination of our “sally-pounce” (see below) B. Subsurface maneuvers (bird penetrates or ma- followed by our “lunge.” Some studies have nipulates the substrate rather than removing used “dart” for foliage-gleaning birds and food from its surface; the attack is directed at “rush” for ground-foraging birds as pre- food that cannot be seen from the surface with- sumed equivalents. Several ground-foraging out substrate manipulation). birds, particularly thrushes (Heppner 1965; 1. Probe: to insert the bill into cracks or holes Smith 1973; Tye 1981; Willis 1985a, 1986) in firm substrate or directly into softer sub- and ground- (Neomorphus:Willis strates such as moss or mud to capture hid- 1982a), and also some bulbuls (Bleda;Willis den food. This tactic is often associated with 1983a), tyrannids (Muscisaxicola,Smith and specialized morphologies adapted for spe- Vuilleumier 197 1; Corythopis torquata, cific substrates. Most probers have long, Willis 1983b), and (Gymnopithys, slender, decurved bills for reaching deep into Willis 1968; Grallaricula nana, Parker et al. crevices, tubes, holes, and soft substrates 1985) use the lunge maneuver regularly. Ar- such as mud or moss. Those that probe bark boreal birds that also regularly lunge in- often have specialized hindlimb morphol- clude: Red-crowned Ant-Tanager (Habia ogy and tail structure for climbing on and rubica; Willis 1960), Plain-brown Wood- bracing against branches (Richardson 1942, creeper (Dendrocincla fuliginosa; Willis Bock and Miller 1959, Feduccia 1973, Nor- 1972), Chestnut-crowned Gnateater (Cono- berg 1979). Several unrelated groups have pophaga castaneiceps;Hilty 1975); Black- converged on similar morphology associ- headed Grosbeak (Pheucticusmelanoceph- ated with bark probing: the creepers (Cer- alus;Airola and Barrett 1985), the tiny, can- thiidae), some woodcreepers (Dendrocolap- opy antwrens of the Terenura callinota su- tidae), and the Australian perspecies (Remsen et al. 1982; Stiles 1983; (Climacteris spp.). The scythebills (Cam- T. A. Parker, unpubl. data), White-shoul- pylorhamphus spp.) and the Long-billed dered Tanager (Tachyphonus luctuosus; Woodcreeper (Nasica longirostris), with Greenberg 1984a; JVR, unpubl. data), and some of the longest bills relative to body an undescribed species of Cercomacra(For- size of any , use their bills for micariidae; Parker and Remsen 1987). probing deep into holes in tree trunks and 150 STUDIES IN AVIAN BIOLOGY NO. 13

bamboo stems and into large bromeliad feeding birds around the world probe flow- clusters (Pierpont 1986; T. A. Parker, un- ers, especially in the Trochilidae, Nectarini- publ. data; JVR, unpubl. data). Some wood- dae, and Meliphagidae. Woodpeckers and peckers do more probing (and gleaning) than hummingbirds also extend their tongues to the more “typical” maneuvers, probe crevices, holes, and flowers; such such as “peck” and “hammer” (e.g., Bock probing could be labelled “tongue-prob- 1970; Short 1973, 1978; Cruz 1977; Alatalo ing.” 1978; Cruz and Johnston 1979, 1984; Sta- C. Subsurface maneuvers with SubstrateManip- cey 1981; Askins 1983; Pettersson 1983; ulation (maneuvers in which the substrate is .Kattan 1988). Bark-foraging birds that also manipulated beyond insertion of a probe). probe epiphytic vegetation include the 1. Gape: to insert the bill into the substrate as Brown Creeper (Certhia familiaris; Stiles in a probe, but the bill is opened to widen 1978), many woodcreepers (Willis 1983c, the opening. This maneuver is characteristic d), and some woodpeckers (Kilham 1972, of many starlings and American blackbirds Short 1973, Cruz 1977, Askins 1983) and (Icteridae), which have bills and jaw mus- certain fumariids (Skutch 1969, Eley et al. culature adapted for gaping (Beecher 195 1, 1979, Parker et al. 1985, Remsen 1985). 1978; Orians 1985b). Various icterids use Some species without obvious morpholog- their bills to open holes in curled living and ical adaptations for climbing also frequently dead leaves (e.g., orioles [Icterusspp.]), dead probe bark or epiphytes on branches. Al- branches and stems, moss, bromeliad clus- though examples include continental birds, ters, seed clusters, leaf-litter, soil (Sturnella such as the Sharpbill (Stiles and Whitney spp.), clumps of grass, flowers, and large 1983), some (Root 1964, Parker fruits (Cruz 1978; Orians 1985b; Robinson 1986a), and the (Ergaticus 1985, 1986, 1988); they also use “gape” to tuber;Elliott 1969) they are particularly fre- turn over stones, twigs, dung, and other ob- quent on islands or in regions such as New jects that might conceal prey on the ground Guinea and Australia where specialized (Orians 1985b). Several species of wood- bark-searching taxa are rare or absent (Keast warblers, including several Vermivora spp. 1968, Zusi 1969, Cruz 1978). Examples in- (Ficken and Ficken 1968), the Swainsons’ clude: a pachycephalid (Colluricincla har- Warbler (Limnothlypisswainsonii;Meanley monica), a scrub-wren (Sericornismagnus), 1970), and the Worm- Warbler (Hel- and some meliphagid honeyeaters (Keast mitherosvermivorus; Greenberg 1987b), use 1968, Recher et al. 1985); a mimid (Cinclo- the gape maneuver for probing buds, dead certhia rufcauda; Zusi 1969); some Hawai- leaves, and flowers. The Sharpbill “gapes” ian honeycreepers (Richards and Bock 1973); to open tightly rolled young leaves and dead some ictetids (Nesopsarnigerrimus and Ic- leaves (Stiles and Whitney 1983), as does terus leucopteryx;Cruz 1978); and several the woodhoopoe Phoeniculusbollei to open birds-of-paradise (Ptiloris spp., Astrapia crevices in loose bark (Liihrl 1972). In- mayeri, Pteridophora alberti, Diphyllodes stances of gaping are occasionally reported magnijicus; Forshaw and Cooper 1979). in other taxa, such as Meliphagidae (Keast Many speciesthat search hanging dead leaves 1968) and Dendrocolaptidae (Willis 1983~). for hidden probe into these curled 2. Peck: to drive the bill against the substrate leaves (Remsen and Parker 1984 and ref- to remove some of the exterior of the sub- erences therein; K. V. Rosenberg, unpubl. strate. This maneuver is characteristic of data; T. A. Parker, unpubl. data). Similarly, many woodpeckers (Picidae) that excavate some species of small tanagers (Dacnis spp. holes in bark or wood to expose prey. “Peck” and Cyanerpesspp.; Snow and Snow 197 1, is synonymous with the “tap” maneuver of Isler and Isler 1987) use their slender bills some studies of woodpeckers; we recom- to probe inside curled living leaves. Some mend restricting “tap” to those motions that populations of Yellow-throated Warbler are probably exploratory pecks for detecting (Dendroica dominica) probe pine cones wood-borer tunnels or movements, as de- (Ficken et al. 1968, Emlen 1977) or dense scribed by Davis (1965) and Kilham (1972). clusters of pine needles or small leaves (Lack Many parids and at least one icterid (Ne- and Lack 1972). Some ground-foraging birds sopsarnigerrimus; Cruz 1978) also peck to probe in soil, mud, or deep leaf-litter; ex- excavate holes in rotted wood. Ground-for- amples include thrashers (Toxostoma, aging birds use this maneuver in combi- Mimidae; Fischer 1981), Whites’ Thrush nation with “flake” (see below) to dig small (Zoothera dauma; Edington 1983), Rook holes to reach food in the ground (e.g., (Corvusfrugilegus; Waite 1984b), White- thrashers [Toxostoma], Engels 1940, Fi- winged Chough (Corcorax melanorham- scher 198 1; and some thrushes, Tye 198 1). phos;Ford et al. 1986), and the woodcocks Some frugivorous birds use “peck” to break (Scolopaxspp.; Sheldon 197 1). The furna- the outer skin of large fruit (Snow and Snow riid Cinclodesexcelsior probes moss and li- 197 1). The Bananaquit (Coerebaflaveola; chens on rocks and the ground (Fjeldsa et Gross 1958), some hummingbirds (Colwell al. 1987). Hundreds of species of nectar- 1973, Stiles 1985c), some white-eyes (Zos- CLASSIFICATION OF BEHAVIOR--Remsen and Robinson 151

terops; e.g., Gill 1971) and icterids (Rob- and the substrate is grasped briefly between inson, unpubl. data) may use this maneuver, the mandibles (which can be called “toss” usually described as “piercing,” to make a when the distinction can be made). “Flake” hole in the base of flower corollas for “steal- is synonymous with “bill-sweeping” (Clark ing” nectar, but the actual maneuver used 1971) except that it applies to substrates to make the hole is uncertain. The flower- other than leaf-litter. “Flake” is also auuar- piercers (Diglossa) hold the flower with their ently synonymous with R. J. Craig’s (i 984) hooked upper mandible and pierce with their “leaf-pull.” Many bark-foraging woodpeck- sharp, upturned lower mandible (Skutch ers “flake” to dislodge loose sections of bark 1954). (Tanner 1942; Kilham 1965, 1983; Conner 3. Hammer: to deliver a series of pecks with- 1981). The term “scaling” used in many out pausing between pecks. This maneuver studies of woodpeckers to describe removal is mainly restricted to certain woodpeckers of loose bark presumably refers to a com- that use it for excavation of deep holes to bination of our “pecking,” “flaking,” and reach bark- or wood-dwelling insects or sap. “prying.” Some dendrocolaptids (Willis The twig-foraging furnariid Xenopsminutus 1983c, Pierpont 1986) fumariids (JVR, un- also uses this maneuver frequently (Skutch publ. data), and a meliphagid (Melithreptus 1969). Some chickadees and titmice (Pari- brevirostris;Keast 1968) use this maneuver dae) may use this maneuver occasionally to to search through debris clusters and loose open acorns, galls, seeds, and fruits (e.g., bark. Ground-foraging birds that “flake” Parusinornatus and P. rufescens;Root 1964, leaf-litter include some thrushes (Turdus 1967), but the pecks are not delivered as [Skutch 1960, 1981; Clark 1971; Tye 19811; rapidly as in woodpeckers. The distinction kylocichla [Clark 197 1; Holmes and Rob- between hammer and peck, which rests on inson 19881: Alethe [Willis 19861). antbirds whether there is a pause between pecks, may (Formica&, Skutch 1969, Wiihs 1985b, be vague. Counting the number of pecks per Rhopornis,Willis 198 la), leaftossers (Scle- unit time, and thereby eliminating the rurus, Furnariidae; Skutch 1969, Hilty and “hammer” category, is an alternative treat- Brown 1986) thrashers (Toxostoma; Clark ment. 197 1, Fischer 198 l), bulbuls (Bleda; Willis 4. Chisel:like “peck,” but rather than the bill 1983a), the waterthrushes (Seiurus; R. J. being pounded almost perpendicularly into Craig 1984) and homeros (Furnarius; Rob- the substrate, it is aimed more obliquely at inson, unpubl. data). The Dune Lark (Mir- the substrate-usually bark or dead stems- afra erythrochlamys)uses “flake” to dis- and the bill is used as a chisel or lever to lodge sand to excavate small craters to expose dislodge portions of the substrate. The di- hidden seeds (Cox 1983). The fumariid Cin- rection of head movements is forward and clodesexcelsior “flakes” moss and lichens upwards. Slightly to strongly upturned low- from rocks (Fjeldsa et al. 1987). er mandibles that give the bill a somewhat 6. Pry: to insert the bill into a substrate and chisel shape are often associated with species use it as a lever to lift up portions of the specialized on chiseling. Species that seem substrate. This differs from “flake” in that to have converged on this foraging behavior the sides of the bill, rather than the tip, ac- and morphology are some Xenops spp. (Fur- complish the movement of the substrate nariidae: Skutch 1969). the dendrocolautid while the tip remains relatively stationary. Glyphor$nchusspirur& (Skutch 1969),-the Substrates for which “pry” is needed are fumariid Simoxenopsucayali (JVR and T. generally more firmly attached than those A. Parker, unpubl. data), and the dislodged when a bird “flakes.” “Pry” dif- Neoctantesniger (Hilty and Brown 1986); fers from “chisel” in that the tip of the bill and to a lesser degree, (Sit@ spp.) is stationary, instead of moved forward and and sitellas (Sitellu spp.; Holmes and Rech- upward as in chisel. Examples of species er 1986a, b). We invented this category to that use “pry” are: Band-backed Wren match our expectations of how chisel-shaped (Camuvlorhvnchuszonatus: Skutch 1960). bills are used rather than on any data on some- species of dendrocolaptids (Skutch movements used by these species. Although 1945; Willis 1983c, d), a meliphagid (Mel- some brief descriptions (e.g., Glyphoryn- ithreptus validirostris;Keast 1968) many thus; Skutch 1969) fulfill our expectations, woodpeckers (e.g., Short 1973) and a bird- the reality of our “chisel” maneuver re- of-paradise (Astrupia mayeri; Forshaw and mains unclear. Coouer 1979).,_ all of which urv-_ UD_ sections 5. Flake: to brush aside loose substrate with of loose bark, and Sharpbill, which pries sideways, sweeping motions of the bill. Not moss from branches (Stiles and Whitney as much force is required as in chisel or pry 1983). because the substrate dislodged is already 7. Pull: to grasp, pull, or tear, and thereby re- loose or unattached. This category com- move or dislodge sections of the substrate bines two types of motions that are often with the bill. Pullina differs from “flakina” difficult to distinguish in the field: the closed in that the target substrate is grasped in the bill tip is used to brush aside the substrate, bill because extra force is needed to dislodge 152 STUDIES IN AVIAN BIOLOGY NO. 13

more firmly attached potions of substrate. prised 25% of the maneuvers of Chestnut- Birds that pull off loose bark or lichens to sided Warblers (Dendroica pensylvanica) in attack hidden insects include Band-backed winter. Holmes and Robinson (1988) found Wren (Skutch 1960) Plain Titmouse (Purus that about one-fifth of all maneuvers used inornatus; Root 1967), Crested Shrike- by Ovenbirds (Seiurus aurocapillus) and (Falcunculus fiontatus: Recher et al. 1985. Dark-eyed Juncos (Junco hyemalis) were Ford et al. 1986), a bird-of-paradise (Mac: leaps. Greenbergand Gradwohl(l980) con- gregoria pulchra ; Forshaw and Cooper sidered leaping (from ground to foliage) to 1979), some orthonychids (Holmes and be the primary foraging maneuver of Ken- Recher 1986a), some dendrocolaptids(Wil- tucky Warblers (Oporornis formosus) and lis 1983c, d) and Giant Cowbird (Scuphi- Chestnut-backed Antbirds (Myrmeciza duru orvzivoru: Robinson 1988). The Plain exsul). The Chestnut-crowned Gnateater Titmouse also pulls apart leaf galls, flowers, (Conopophaga castaneiceps) leapsto nearby lichens, and curled dead leaves(Root 1967). perches to attack prey (Hilty 1975). Many Thrioadectes rufobrunneus (Skutch 1969) speciesthat follow army-ant swarms prob- and several other fumariids (T. A. Parker; ably “leap-down” from low perches above unpubl. data) pull leaves from bromeliads the ants to capture flushed insects (e.g., to expose prey. Most New World barbets Gymnopithys, Willis 1968;Rhegmatorhina, (Cupito, Eubucco) also pull open large dead Willis 1969; Phfegopsis, Willis 198 1b; Den- leaves, twig galls, and sections of rotting drocincla. Willis 1972. 1979)., Some seed- wood to searchfor prey (Remsen and Parker eating speciesapparently leap onto stems to 1984;T. A. Parker, unpubl. data; SKR, pers. pull seedheads to the ground (Emlen 1977). obs.). The Plush-capped Finch (Catambly- The direction and distanceof the leap should rhynchus) pulls the leaf whorls at the nodes be recorded,just as it is for “sally” (see next on bamboo stems, presumablyto reveal in- account), particularly because a “leap” sects(Hilty et al. 1979, Remsen 1985). The downward (i.e., dropping) probably requires ground-foragingSong Thrush (Turdus phi- only a fraction of the energy than does an lomelos) uses “pull” in its foraging reper- upward or outward leap against gravity. toire (Henty 1976). Many parrots use “pull” B. Wing-powered maneuvers for openingfruits, seeds,flowers, and rotting 1. Sally: to fly from a perch to attack a food wood (Forshaw 1973 and referencesthere- item (and then return to a perch). Most au- in). thors have used separate terms to distin- 8. Scratch: to dislodgesection of substratewith guishsallies directed at aerial prey from those foot movements. This maneuver is used by aimed at nonflying prey. We do not, because many ground-foraging birds around the the foraging site (i.e., air vs. anything else) world; examples include: some orthony- will automatically be recorded more appro- chids (Zusi 1978, Frith 1984), Australian priately in our schemeunder the “substrate” lyrebird (Menuridae; Recher et al. 1985, category (see below); and the maneuver it- Holmes and Recher 1986b), and some self appearsto us to be very similar whether (e.g., Alectura lathami; Frith directed at air or hard substrate.Although 1984). Although most speciesscratch using we acknowledge that the movements di- one foot at a time, many emberizid sparrows rected at flying vs. nonflying food may be (Davis 1957, C. J. 0. Harrison 1967, Hail- different, we prefer to remove the substrate- man 1973, Greenlaw 1976 and references bias from terminology as much as possible. therein) and occasionally some thrushes Another difference between our system and (Turdus; Clark 1983) and icterids (Greenlaw othersis that the term “hawk” hasbeen used 1976) move both feet simultaneouslyto ex- frequently to describe what we here call pose food under leaf-litter or snow. “sally” (e.g., Holmes et al. 1979b). We use II. Aerial maneuvers (bird must leave substrate to “sally” rather than “hawk” because:the dic- reach food) tionary definition of “sally” is closer to this A. Leg-powered maneuvers behavior than is “hawk,” and hawks rarely 1. Leap: to launch into the air to reach a food if ever fly from a lookout perch to attack item too far for a “reach” but too close for flying prey. Similarly, the term “flycatch” a “sallv.” This differs from “sally” in that has been used frequently for sallies after the upward thrust seems to come mostly flying prey, but most “flycatchers,” whether from leg movements rather than wing tyrannids or muscicapids,do not “flycatch” movements (Davies and Green 1976); it is perse, but insteadglean or sally to substrates equivalent to the “jump-glean” of Holmes (e.g., Fitzpatrick 1980). Greenberg (1984a) and Robinson (1988) and presumably the distinguished sallies in which a bird re- “jump” of Hutto (198 1b). Distinguishing turned to the perch from those in which the “leap” from short sallies is often difficult. bird continuesin the same direction by call- Davies and Green (1976) found that “leap” ing the latter “darts.” There is probablymore was the most frequent maneuver used by among-author variability in terms used to Reed Warblers (Acrocephalus scirpaceus), describe aerial maneuvers (e.g., hawk, hov- and Greenberg (1984a) found that it com- er, hover-glean, snatch, sally, flycatch) than CLASSIFICATION OF BEHAVIOR--Remsea and Robinson 153

in any other broad category of foraging be- rick 1980, Willis 1983b) and Catharus havior. thrushes (Paszkowski 1984, Holmes and Robinson 1988). Other ground-foraging Many species have characteristic directions or dis- birds “sally-strike” to catch flying in- tances associated with their sallies that provide an in- sects. Examples include ground-tyrants dex of the average search radius (Fitzpatrick 1981, (Muscisaxicola spp. [Smith and Vuilleu- Robinson and Holmes 1982) and these arc important mier 1971; Fitzpatrick 1980, 1985]), to record (after Fitzpatrick 1980): Rhipidura leucophrys (Ford et al. 1986), and wheatears (Oenanthe spp.; Leisler and a. sally-distance (distance of the sally from Seinbenrock 1983). Some species also use perch to food item). this maneuver to obtain fruit (Skutch b. sally-angle (the qualitative divisions 1969; Fitzpatrick 1980, 1985). Sally- ” “horizontal,” “diagonal- “UP, ” “down, striking species often have wide, scoop- up,” and “diagonal-down” probably rep- like bills and wide gapes that presumably resent maximum possible resolution un- facilitate prey capture in flight (Fitzpat- der most field conditions). Certain species rick 1985). or species groups may have characteristic b. Sally-glide: like sally-strike except the fi- sally angles. Willis (1984), for example, nal approach at the target is a glide (vs. noted that most manakins (Pipridae) continuous flapping in sally-strike). typically sally only at a horizontal angle, Moermond and Denslow (1985) pointed and Holmes and Recher (1986a) found out that many sally-strikers do not use that two species of thornbills differed in continuous, flapping flight in their ap- the angles of their sallies. proach, and they made a convincing case Sally-distance and sally-angle should refer to the ini- for distinguishing those species that used tial attack attempt only; subsequent pursuit of a missed a brief glide from those that did not. It target should be recorded separately. We distinguish is likely that some or many of the ex- five types of sallies based on the birds’ foraging motion amples of sally-strikers above are ac- at the end of the sally: tually sally-gliders. Other than Moer- mond and Denslows’ (1985) data on a. Sally-strike: to attack in a fluid move- frugivores, the prevalence of sally-gliding ment without gliding, hovering, or land- (which they called “sally-scooping”) vs. ing (after the “outward striking” and sally-striking will be revealed only by “upward striking” of Fitzpatrick [ 19801 careful observations. and the “snatch” of Moermond and Sally-stall: to stall in front of the target Denslow [1985]). The “sally-strike,” briefly with fluttering motions at the end whether aimed at flying prey or station- of the sally. Moermond and Denslow ary substrates, is the characteristic attack (1985) noted that many species usually behavior of many Tyrannidae (Hespen- considered to sally-hover (see below) do heide 1971; Fitzpatrick 1980, 1985; not engage in true hovering (flying in Sherry 1984), Muscicapinae and other place), but rather flutter awkwardly in a Old World “flycatchers” (e.g., Croxall stalling motion after a steep attack angle 1977, Davies 1977b, Fraser 1983, More- at the final approach of the sally. Such no 1984), Pipridae (Skutch 1969), Buc- species, mainly and some cotin- conidae (e.g., Skutch 1948; Willis 1982b, gas, use different flight motions and have c), Galbulidae (Hilty and Brown 1986) different morphological adaptations from Meropidae (Fry 1984) Momotidae (e.g., those that hover. We suspect that many Skutch 1947; Willis 198 lc), Alcedinidae of the examples of “sally-hover” noted (Fry 1980), and Conopophagidae (Willis below may actually be “sally-stalling.” 1985b). Numerous species in other fam- As with sally-gliding, only careful obser- ilies use the sally-strike maneuver to vations (or high-speed photography?) will varying degrees, accompanied by mor- reveal its true prevalence among sallying phological adaptations that parallel those birds. seen in more typically sally-striking Sally-hover: like other sallies except that groups (Partridge 1976b, Norberg 1979, the bird hovers at the target substrate at Schulenberg 1983). Most species that use the end of the sally. This is synonymous this maneuver are sit-and-wait predators with Fitzpatricks’ (1980) “hover-glean.” that watch for prey while sitting motion- Most studies do not distinguish between less on an elevated perch, although oth- sally-strike and sally-hover (much less ers search more actively (e.g., tree-climb- sally-glide and sally-stall), and many oth- ingdendrocolaptids [Willis 1972, 1982d, er studies appear to label all sallies to Pierpont 19861 and some vireos [Rob- foliage as “hovering” (e.g., Holmes et al. inson and Holmes 19821). Ground-for- 1979b), even though few of these ma- aging birds that “sally-strike” to capture neuvers actually involve hovering flight. insects on foliage above them include the Unless these maneuvers are distin- tyrannid Corythopis torquata (Fitzpat- guished, the possibility that they require 154 STUDIES IN AVIAN BIOLOGY NO. 13

different morphological adaptations, as 1985, Ford et al. 1986, Holmes and Re- found for frugivores by Moermond and cher 1986b), and Fan-tailed (Cu- Denslow (1985), cannot be addressed. culuspyrrhophanus, Recher et al. 1985), Some tyrannids use the sally-hover ma- some Catharus thrushes (Dilger 1956, neuver regularly (Fitzpatrick 1980, 1985), Paszkowski 1984), some pullbirds (Wil- as do (Regulus;Rabenold 1978, lis 1982b, c), and the Field (Spizella pu- Moreno 198 1, Franzreb 1984) the Blue- silla) and Chipping (S. passerina)spar- gray Gnatcatcher (Polioptila caerulea; rows when foraging for insects (Allaire Root 1967) some sylviid warblers (Phyl- and Fisher 1975). Some vireos (Vireoni- loscopus;Gaston 1974) some wood- dae) use this maneuver when attacking warblers (Ergaticus ruber, Elliott 1969; prey on branches (James 1976, Robinson Dendroica. Rabenold 1978. 1980: and Holmes 1982). Some tropical vireos Greenberg 1984a); an acanthizid (Seri- (Hylophilus) characteristically use this cornis magnirostris; Frith 1984), the maneuver followed immediately by Restless Flycatcher (Myiagra inquieta; hanging on leaf margins when attacking Ford et al. 1986), and some putlbirds undersides of leaves (T. A. Parker and (Sherry and McDade 1982; Willis 1982c, JVR. unwbl. data). A soecial kind of e). Bell (1984) found that at a forest site sally:pounce is used by some seed-eating in New Guinea, 5 of 83 bird species stud- birds that sally to a grass stem, grasp the ied in detail used this maneuver in 18- stem in their feet, and then allow their 24% of his foraging observations: two weight to pull the stem to the ground, monarch-flycatchers (Monarcha and where seeds can be removed more effec- Arses),a cracticid (Peltops),a meliphagid tivelv (Allaire and Fisher 1975). (Melilestes), and a drongo. Similarly, Flutter-ktase: to flush or dislodge prey from Remsen (1985) found that at a forest site a substrate and to then chase the prey. This in the Andes, 4 (all tyrannids) of 33 maneuver is used regularly by foliage-glean- species studied in detail used this ma- ing birds that flutter after a falling or flying neuver in 16-33% of their foraging ob- prey item that has escaped their normal at- servations. In contrast, hovering ac- tack behavior and is often preceded by a counted for only 1% of all prey attacks lunge. Root’s (1967) “tumble” is synony- observed in an Australian eucalypt forest mous (because “tumble” refers to out-of- where 41 species were studied in detail control, sommersaulting movements, we (Recher et al. 1985). Many species use have chosen a new term). Root (1967) found this maneuver when taking fruit. Ex- that Blue-gray Gnatcatchers (Polioptila cae- amples include many tyrannid flycatch- rulea) used this maneuver in 23% of all sal- ers, manakins, and some tanagers (Fitz- lies directed at insects in the air; however, patrick 1980, Willis 1984, Moermond Root suspected that the frequent tail-flash- and Denslow 1985). Some species, in- ing of this species may function to startle cluding kinglets (Leisler and Thaler 1982, insects, therefore making these “flutter- Franzreb 1984) some wood-warblers chases” into “flush-pursuits” (see below) in (Morton 1980a), and the Yellow-rumped our scheme. Morse (1968) found that four Cacique (Robinson 1986), occasionally wood-warblers (Dendroica) used this ma- hover under surfaces to search for food neuver in about 5% of their foraging mo- that cannot be seen from a perch. Hum- tions. We see this maneuver most frequently mingbirds, of course, use this maneuver in foliage-gleaning birds in mixed-species extensively when feeding at flowers or flocks in the canopy of tropical forests; ap- searching foliage and branches; for nec- parently, the escape behavior of many of tar-feeding, however, the parameters their prey involves falling from “sally-distance” and “sally-angle” are the substrate at the approach of a bird pred- usually irrelevant. ator. In particular, the White-shouldered e. Sally-pounce:to land briefly at the end Tanager (Tachyphonusluctuosus) uses the of the sally to take food from substrate. flutter-chase maneuver frequently (Snow and Although the bird is perched when it takes Snow 1971; JVR, unpubl. data). We use this the food item, we classify this maneuver term mainly for species that are not typically as a “sally” because it involves a flight salliers. We recomend recording the dis- after food is spotted at a distance from tance and angle of the chase, just as in the the lookout perch. It is probably syn- sally maneuvers. onymous with Fitzpatrick’s (1980) Flush-pursue:similar to “flutter-chase” ex- “landing-and-gleaning,” Recher et al.‘s cept that species that use this maneuver de- (1985) “pounce,” and Holmes and Rob- liberately (vs. accidentally) flush prey from inson’s (1988) “dive-glean.” Examples hiding places and then pursue the flying or ofbirds that use this maneuver are: many falling prey. This maneuver tends to be open-country tyrannids and muscicapids prominent in the foraging repertoire of (Fitzpatrick 1980, 1985; Fraser 1983), species that use it, most of which have con- bluebirds (Power 1980) Australian rob- spicuous wing or tail spots or stripes that ins (Petroica, Eopsaltria; Recher et al. are flashed to startle hidden prey. Distin- CLASSIFICATION OF BEHAVIOR--Remen and Robinson 155

guishing this maneuver from “flutter-pur- trees at the same height as the small trees. Pro- suit” may be difficult, but because each in- vided that only one observer records the data, a volves fundamentally different tactics, we visual estimate of height (vs. precise measure- believe that to do so where possible is valu- ments) may be the only practical way to obtain able. Among North American species, the such data. Not only does the time required to American Redstart (Setophaga ruticilla; make precise measurements reduce the volume Robinson and Holmes 1982) and, on the of data that can be collected, but it seems unlikely ground, the Northern Mockingbird (Mimus that the birds recognize vertical subdivisions suf- polyglottos;Hailman 1960) most frequently ficiently precisely to warrant such a time invest- use this maneuver. Other examples include: ment. Heterogeneity in canopy height, light pen- Dendrocincla woodcreepers (Willis 1972, etration, and foliage distribution obliterate such 1979) fantails (Rhipidura; Recher et al. precise boundaries. However, differences among 1985, C. J. 0. Harrison 1976, Holmes and observers in the accuracy of such visual estimates Recher 1986a), Monarcha flycatchers (Pear- (Block et al. 1987) reveal the unreliability of such son 1977b), Myiobiustyrannids (Fitzpatrick visual estimates and provide support for use of 1980) Ruddy-tailed Flycatcher (Terenotric- objective measures of height. cuserythrurus, Sherry 1984), and the Myio- III Horizontal position: Many researchers have re- borusredstarts (Parulinae; Remsen 1985). corded the “horizontal” position (e.g., “inner,” 4. Screen:to attack in continous flight (after “middle,” “ outer”) of the bird in the tree or bush. Emlen 1977, Fitzpatrick 1980). (Note that Many species of foliage- and branch-gleaning birds this is a searching behavior as well as an characteristically favor one ofthese foraging zones attack maneuver.) This is synonymous with (e.g., MacArthur 1958 and numerous other stud- “hawk” as used by Remsen (1985) and oth- ies). Whether birds select such zones per se, or are ers for birds that feed in flight. Swallows, keying on differences in foliage density (next cat- swifts, and nighthawks (Chordeiles,Capri- egory) is unknown. It is possible that “horizontal mulgidae) use this maneuver almost exclu- position” and “foliage density” measures are sively. Other birds that may use this ma- largely redundant. However, Greenberg and neuver occasionally include European Gradwohl (1980) and Holmes and Robinson Starling (Sturnusneglectus; Cayonette 1947) (198 1) showed the importance of branch and leaf Golden-naped Woodpecker (Melanerpes arrangement around the bird in determining which chrysauchen;Skutch 1969) Lewis ’ Wood- surfaces can be attacked effectively. Greenberg pecker (M. lewis; Bock 1970), some tyran- (1984a) used a system for “horizontal” position nids (Fitzpatrick 1980) and probably the designed specifically to place the foraging bird in pullbird Chelidoptera tenebrosa (Burton categories with respect to foliage and branch ge- 1976). ometry. IV Foliage density:Foliage density at the point of FORAGING SITE foraging- - observation can be recorded using a qual- itative scale. For example, the system that-we have We suggest recording the following parameters with found to be useful (ea.. Remsen 1985: modified respect to the foraging site used by a foraging bird: (1) . __ from Wiley 197 1) is a scale from “0” to “5” of general habitat, (2) vertical position, (3) horizontal po- increasing foliage density within a one-meter ra- sition, (4) foliage density, and (5) the precise substrate dius around the bird: “0” = no vegetation within from which the food was taken. We discuss each cat- the imaginary l-m sphere; “1” = very low vege- gory briefly. tation density within the sphere (e.g., 95-99% of I. Habitat: Many study areas contain more than one all light passes through sphere); “2” = low density, habitat or microhabitat. Each foraging record 75-95% of light passes; “3” = moderate density, should be assigned to one of the investigators’ 25-75% of all light passes; “4” = high density, general habitat or microhabitat categories to per- only 5-25% of light passes; and “5” = extremely mit examination of the influence of habitat on dense, O-5% of light passes. foraging behavior (e.g., Bilcke et al. 1986). Clas- V. Substrate. We have found the following substrate sification of habitats, a complex and critical topic, categories to be useful: is beyond the scope of this paper. A. Living Foliage II. Vertical position: It has been recognized for de- 1. Plant speciesor “type” (species, genus, or cades that important differences in vertical posi- family when possible; otherwise “broad- tion separate the foraging activities of many closely leaf tree,” “vine,” “palm,” “grass,” “bam- related birds. Furthermore, foraging behavior may boo,” “fern,” “ cactus,” and the like; note change with changes in height above ground. if epiphytic). Many studies (e.g., Hartley Therefore, every foraging record should be as- 1953; Gibb 1954; Willson 1970; Reller signed two values to allow its position to be plot- ted: (1) height-above-ground and (2) distance-to- 1972; Holmes and Robinson 198 1; Woi- canopy (above bird). We have also found a third narski and Rounsevelll983; Robinson and parameter to be of interest: (3) height of the in- Holmes 1984; Franzreb 1984; Bell 1985b; dividual plant in which the bird was foraging. This Morrison et al. 1985, 1987b) have empha- allows us to distinguish species that frequently use sized the importance of distinguishing plant small trees or saplings within the foliage column species. In the tropics, many bird species from those that use the lower foliage of canopy specialize on distinctive plant types such 156 STUDIES IN AVIAN BIOLOGY NO. 13

as bromeliads, bamboo (Parker 1982, perhaps a qualitative scale to indicate de- Remsen 1985) and palms. gree of corrugation); (d) seam between two 2. Leaf size (visual estimate of length and closely growing branches or between vine width of leaf searched). This is probably and supporting trunk (such seams appear necessary mainly in areas where complex- to be particularly favored foraging sites for ity of plant communities prevents quick some dendrocolaptids; e.g., Hylexetustes taxonomic identification of plant species perrotii [Willis 1982f1); (e) lichen- or moss- (and therefore subsequent, more accurate covered (mossy branches are favored sites assessment of leaf size). Leaf buds should for furnarids, dendrocolaptids, several also be distinguished, although these can birds-of-paradise, and tanagers [Skutch be “food” as well as substrate. 1969, 1981; Forshaw and Cooper 1979; 3. Top or Bottom. See Greenberg and Grad- Parker and ONeill’ 1980; Remsen 1984; wohl(1980) and Greenberg (1984a) for the Parker et al. 1985; Remsen 19851); (f) hard, importance ofdistinguishing leaf tops from dead wood with bark removed; (g) soft, leaf bottoms. Greenberg and Gradwohl rotted dead wood (see Alatalo [ 19781, Cruz (1980) also found that a foliage-gleaning and Johnston 119791. Pettersson 119831.and tanager (Ducnis cayana) may inspect brown, Morrison et al. [ 1987b] for examples of the insect-damaged areas on leaves; therefore, importance of distinguishing live from dead observers should be careful to record when branches in bark-foraging birds; the fur- such leaf sections are investigated. nariid Xenops minutus seems to be spe- Dead foliage. See Gradwohl and Greenberg cialized on dead branches, especially those (1982b). Remsen and Parker (1984). and Ro- that have fallen but are caught up in the senberg(this’ volume) for the importance of canopy [Skutch 1969; T. A. Parker and distinguishing live from dead leaves. Size of JVR, unpubl. data]); and (h) holes (favored leaf should also be recorded, as well as con- foraging sites for some dendrocolaptids dition (curled, tattered, or entire; see Rosen- [Willis 1982d, 4). berg, this volume) and general type (e.g., palm, D. Ground broadleaf, bamboo). 1. Surfacetype (e.g., mud, bare soil, leaf-litter, Bark or stem surfaces. Observers should note moss, gravel). that careful observations often reveal that 2. Distanceto nearestcover. many species generally thought to be foliage- 3. Slope(e.g.,flat, moderate slope, steep slope). searchers direct considerable proportions of E. Rock their attacks at branches and stems, such as 1. Size. some species of vireos (Nolan and Wooldridge 2. Surfacetype (e.g., smooth, rough, crevice). 1962; Root 1967; James 1976, 1979; Robin- 3. Surface“angle” (top, bottom, side; vertical son and Holmes 1982; Airola and Barrett or diagonal slope). 1985), tanagers (Snow and Snow 1971, Isler F. Air and Isler 1987) wood-warblers (Morse 1967a, G. Flower (when identification of plant un- b, 1968; Lack and Lack 1972; Emlen 1977; known); as noted by Emlen (1977) it is often Greenberg 1984a), sylviids (Earl& 1983), difficult to distinguish whether some species Hawaiian honeycreepers (Richards and Bock use flowers as sources of food (nectar feeding) 1973) shrikes (Earl& 1983) chats (Frith 1984), or as substrates for searching for arthropods. Old World sallying flycatchers and drongos 1. Corolla length. (Bell 1984), honeyeaters, whistlers, and bab- 2. Color. blers (Keast 1968, Thomas 1980, Wooller and 3. Flower density (estimate no. flowers/unit Calver 198 l), and thornbills (Acunthizu;Bell area; e.g., per 0.5 m*). 1985b, Recher et al. 1987). When recording G. Miscellaneous. Almost every habitat will have use ofthis substrate category, the observer can some substrates that do not fit into the above record: scheme. For example, some species of birds 1. Diameter (visual estimate) search pine cones (Morse 1967a, Ficken and 2. “Angle” of branch (i.e., vertical, horizon- Ficken 1968, Emlen 1977, Moreno 1981), ter- tal, or diagonal). mite nests (Bell 1984), wasp nests (Willis 3. Upper or Lower side (for horizontal or di- 1982fj, spider webs (Young 197 1, Burtt et al. agonal branches). Some species may char- 1977, Douglass 1977, Waide and Hailman acteristically forage on the undersides of 1977. Bell 1984. Brooks 1986. Tiebout 1986. limbs, such as the woodcreeper Xiphorhyn- Parrish 1988, Petit and Petit 1988) dung (An: thus lachrymosus(Willis 1983~). derson and Merritt 1977), and even the skin 4. Plant species,when possible, or plant type of other vertebrates (Rice and Mockford 1954, (see A.l. above). See Jackson (1979) and Orians 1983, Isenhart and DeSante 1985 and Morrison and With (1987) for examples of references therein, Robinson 1988). For fmit- the importance of tree species for wood- eating birds, we do not record a substrate per pecker feeding-site selection. se, but note certain characteristics of the fruit 5. Surfacetype and texture(especially critical under “food taken” (see next section). where identification of plant species is not possible). Examples include: (a) smooth- Although the number of parameters to be recorded green; (b) smooth bark; (c) rough bark (with in this classification of foraging maneuvers and sub- CLASSIFICATION OF BEHAVIOR--Remsen and Robinson 157

FIGURE 2. Sample foraging data transcribed from microcassetteto field notes. Codes: “HT” = height above ground, “DC” = distance-to-canopyabove bird, “FD” = foliage density, “DL” = dead leaf, and “vs” = visual search. Vertical brackets near left margin group consecutive observations on same individual. The thin lines under the “Substrate” column record branch “angles,” and tiny “x” marks record position of bird with respect to branch. (Height variables are in feet, and substratevariables are in inches.) strate characteristics may seem complex and over- For many speciesthat eat small insects,it can even whelming, the advent of microcassettetape-recorders be difficult to determine whether or not a prey item facilitates recording such volumes of data in the field. was capturedat the end of an attack. For thesereasons, Also, transcription of the data can be simplified by most field studies of insectivoresinclude only limited using codes and symbols (Fig. 2). data on prey. Variables measured include prey size (usuallyin relation to bill length, but see Bayer [ 19851 FOOD TAKEN and Goss-Custardet al. [ 19871for cautions) and prey Data on diets are useful for virtually every kind of type (for large prey items such as caterpillars and or- foraging study. Differences in food taken may provide thopterans). Some authors (e.g., Greenberg 1984a) re- information on niches, morphology (principally of the corded each time that a bird wiped its bill after a prey bill), and energetics. Unfortunately, dietary data are attack as an index of success.Reasonably accuratees- usually difficult to obtain in the field, especially for timates of capture rates can be obtained for large prey, insectivores. such as orthopterans that require extensive handling 158 STUDIES IN AVIAN BIOLOGY NO. 13 before they are eaten (Robinson 1986). Many neo- particularly in insectivores, prevents an evaluation of tropical insectivores evidently obtain most of their en- their relative frequencies of use. In addition to quan- ergy from large katydids (Orthoptera: Tettigoniidae) tifying the time taken to manipulate food before swal- and have bills adapted specifically to handle them lowing, we recommend the following terms to describe (Greenberg 198 la). Most temperate-zone insectivores, techniques that we feel are appropriate for field obser- on the other hand, have smaller bills, presumably vations of landbirds: adapted for the smaller arthropods or less agile larvae available during the breeding season. Because large food 1 Engulj?to capture and swallow in one continuous items have more biomass than small items, we think motion, without being held by the bill. that food size should always be recorded where feasible. 2 Gulp (after Moermond and Denslow 1985): to For frugivores, the most important variable is the swallow upon capture without any noticeable ma- plant species. Secondary variables include the color (as nipulation other than being held briefly by the bill. a measure of ripeness), size (especially if the plant species 3 Snap: to pinch momentarily, usually between tips is unknown), and shape of the fruit. For nectarivores, of mandibles and usually to kill prey before further the plant species is again the primary variable of in- handling. terest. If this is unknown, then color, shape, and corolla 4 Mash (after Moermond and Denslow 1985): to length should be recorded. squeeze or move around between the mandibles Data obtained from stomach samples are discussed before swallowing (apparently to kill prey or re- elsewhere in this volume (Rosenberg and Cooper). Here move undesirable portions, such as wings, legs, we wish only to emphasize that stomach samples can shells, and husks); sometimes, juices or pulp are be very useful when they reveal major ordinal levels squeezed out of the food and solid portions dis- of dietary differences among species being compared. carded (Moermond and Denslow 1985, Foster Sherry (1984), for example, showed that species that 1987). This category almost certainly lumps dis- are generally similar in size and foraging behavior can tinct types ofmandibulation that could be revealed differ strikingly in their diets. Dietary analyses of Least by analysis of high-speed photography of food- Flycatchers (Empidonax minimus) and American handling. Redstarts, which are strikingly similar in many aspects 5. Shake: to shake food item violently (to remove of their foraging behavior and foraging-site selection undesirable portions). (Sherry 1979), revealed surprisingly little overlap (Rob- 6. Beat: to beat food item against hard substrate (as inson and Holmes 1982). In this case, knowledge of in above, to kill or remove undesirable portions). diet from stomach samples (redstarts catch manynet- Many small insectivorous birds typically beat in- eropteran leafhoppers) provided information on the sects against branches in a diagonally downward- functional significance of the “flush-chase” attack ma- facing position (e.g., Root 1967). neuver described previously. 7. Rub: to rub food along substrate (usually to re- Data from stomach samples should, however, be move distasteful substances or undesirable por- treated with caution. Because prey items in stomach tions such as hairs and stingers [Sherry and McDade samples can usually only be identified to the level of 19821). order or family, the categories are crude. It is quite 8. Jab: to peck food item with bill tip (to kill it or possible that two species that eat the same orders, fam- open it), usually while clasped with feet. ilies, or even genera of insects could overlap very little 9. Tear: to eviscerate or dissect food item into small- in other aspects of their foraging behavior, particularly er pieces, usually while the food is clasped by one substrate use. Information on diet in the absence of or both feet. data on other components of foraging (e.g., Wiens and 10. Bite: to bite and remove a section of food item Rotenberry 1979) therefore could be misleading. (after Foster 1987). This technique applies as far as we know only to frugivores that take bites from fruit too large to swallow whole. 11. Juggle:to reposition food item, sometimes by toss- FOOD-HANDLING TECHNIQUES ing into air and catching it (to allow or facilitate Once food is “captured,” it may be eaten, delivered swallowing; many species juggle prey to maneuver to offspring or mate, stored (cached), or rejected. We it into a head-first position before swallowing). here consider only the techniques associated with the 12. Clasp:to hold food item with feet. first of these options. The way that food is handled is 13. Anchor: to immobilize food item by fixing it to important because (1) food-handling time must be con- substrate, such as by impaling with sharp objects sidered in the cost : benefit ratio of any food type (e.g., or by wedging food item into crack. Sherry and McDade 1982) (2) it is a factor in studies 14. Drink: intake of liquid food, such as fruit juices ofadaptive morphology (e.g., Sherry and McDade 1982, and nectar. Moermond and Denslow 1985, Foster 1987) and (3) it has important implications for the study of plant- In practice, we have found that in the field, we have frugivore interactions (Howe and Smallwood 1982, time to note only those food-handling behaviors that Moermond and Denslow 1983, Levey 1987b). Food- are not “gulping,” which seems to be the predominant handling techniques, however, have been largely ig- food-handling technique in most insectivorous and nored in studies of arthropod-foraging behavior (for frugivorous birds, with the notation that all “blank” exception, see Sherry and McDade 1982). Fortunately, records refer to gulping. Our scheme leaves out certain the detailed descriptions by some observers (e.g., E. 0. techniques that are presumably very rare, such as Willis and A. F. Skutch) have revealed the distinctive scraping (to remove fruit pulp in snake-like jaw mo- behaviors associated with handling of various food tion; Schaeffer 1953), dropping (to break open), soak- types. The lack of data on food-handling techniques, ing, and drowning. CLASSIFICATION OF BEHAVIOR--Remsen and Robinson 159

ANALYSES OF FORAGING DATA analysis showed that such variables as substrate and tree species were more important in assign- This classification system contains many finely ing species to guilds than attack methods. By subdivided categories. Although too many can contrast, in a similar analysis of an Australian create problems (e.g., small or empty data cells) bird community, which added categories for for statistical analyses, we think that fine sub- flush-chase and manipulative prey-attacks, divisions are preferable during the data-gather- Holmes and Recher (1986a) found that attack ing stage. Their retention allows maximum data methods were also important. The different guild resolution, which in turn, even if sample sizes structures in the two areas may have been influ- are too small for statistical analysis, might gen- enced, therefore, by their differing classification erate insights that can be developed to answer systems. In general, we recommend that manip- specific questions in subsequent studies. Here we ulative attack-methods be distinguished from provide examples of how categories might be methods in which food is simply plucked from combined or subdivided. surfaces or the air in studies of entire commu- I. Ecomorphoiogicalstudies.Fine subdivisions nities. of attack methods, foraging substrate, and III. Single-guild studies (taxonomic guilds, searching behavior are most likely to be useful sense Terborgh and Robinson 1986). Studies that in studies of adaptive morphology. Fitzpatrick focus on ecologically similar species should ben- (1985) for example, showed that many aspects efit from fine subdivisions of substrates and at- of bill and wing shape were strongly associated tack methods. The members of a guild are only with the details of aerial attack methods (see Ta- likely to use a subset of the attack methods shown ble 1) and substrate in tyrannids, whereas leg in Table 1, which should simplify the matrices morphology was more closely related to search- and allow finer subdivisions. Rosenberg (this ing movements and perch types. Fitzpatrick’s volume), for example, included data on the size (1985) classification system of foraging methods, and shape of dead leaves searched, and Green- therefore, combined searching movements, perch berg’s (1987a, b) study of a dead-leaf forager in- types, substrate type, and attack method in an cluded data on searching postures similar to the attempt to include all of the variables that affect subdivisions of near-perch attacks shown in Ta- flycatcher ecomorphology. The bill morpholo- ble 1. Conner’s (1980, 198 1) studies of bark for- gies of bark-foraging birds are also affected by agers showed the importance of different meth- the methods used to manipulate the substrate to ods of manipulating substrates in distinguishing attack concealed food. The finer subdivisions of among species. Fitzpatrick (1980, 198 1) showed near-perch maneuvers (see Table 1) also may be the different ways that syntopic tyrannids differ related to leg and foot morphology (Partridge in the subtle details of how they sally to catch 1976a, Leisler and Winkler 1985). The bill shapes prey. of frugivores and some insectivores may also be IV. Foraging modes (sensuHuey and Pianka associated with particular kinds of food (Green- 198 1) or adaptive syndromes (sensu Eckhardt berg 198 1, Moermond and Denslow 1985, Foster 1979). Studies of foraging modes seek to identify 1987). suites of intercorrelated foraging variables. Many II. Community-level studies.Community-level researchers have shown that the rates and lengths studies probably require the least finely subdi- of searching movements are associated with the vided categories. Communities in wooded hab- lengths and kinds of attack methods (e.g., Wil- itats are likely to include birds that use most of liamson 1971; Eckhardt 1979; Fitzpatrick 1981; the attack methods described in Table 1. If each Robinson and Holmes 1982; Holmes and Recher method were to be broken down by substrate, 1986b; Holmes and Robinson 1988; see also the resulting data matrix would be prohibitively Moermond 1979a and Huey and Pianka 1981 large and would contain many zero values. For for similar analyses of foraging in lizards). In this reason, most studies that seek to identify general, birds that move short distances between guilds use only a few general attack categories perches also obtain food on nearby substrates. (e.g., Holmes et al. 1979b) or use only data on Similarly, species that fly long distances between foraging site (Anderson and Shugart 1974). perches also search and attack over long dis- Holmes et al. (1979b), for example, divided the tances. Studies of adaptive syndromes therefore attack methods of birds in a northern hardwoods include detailed data on searching movements forest into “gleans” (lumping all “near-perch” (including rates and lengths), attack tactics (in- maneuvers in Table l), “hovers” (all sallies to cluding lengths of attacks), and the use of special substrates other than air in Table 1), “probes” foraging tactics such as tail-fanning. Table 2 gives (including all subsurface maneuvers in Table l), examples of adaptive syndromes or foraging and “hawks” (all sallies directed at flying prey modes that have been identified in New World in Table 1). Each of these attack methods was insectivorous birds (modified from Eckhardt then combined with a foraging site. The resulting [ 19791, Fitzpatrick [198 11, and Robinson and 160 STUDIES IN AVIAN BIOLOGY NO. 13

TABLE 2. ADAPTIVESYNDROMES OR FORAGING MODESOF INSECTIVOROUSBIRDS

Associated Foragingmode Searchmovements prey-attackingmaneuvers Open perch or passive searching Infrequent, long flights Long sallies Medium-distance searching Frequent medium-length flights and Sallies and near-perch gleans bouts of hopping Near-surface searching Frequent hops and short flights Near-perch maneuvers, probes Flush-Chasing Conspicuous,frequent flights and Flush-chases hops, wing and tail flicking Manipulative Short periods of movement between Flake, peck, tear, hammer, long periods at the substrate scratch,chisel

Holmes [ 19821). Whether these relationships have lier versions or by incorporating suggestionsfrom global generality remains to be determined. other researchers. We regard this scheme as a V. Energeticsand optimalforaging. Studies of first step towards standardization of the orga- energetics or optimal foraging primarily use data nization and vocabulary of studies of foraging on time intervals between movements and food- behavior of birds. We anticipate that it will be capture rates. Robinson (1986) for example, modified as it is tested and refined by us and, we measured intervals between flights of at least one hope, other researchers. meter as an index of foraging speed and prey- capture rate, and prey size as an index of foraging ACKNOWLEDGMENTS success.Energetic studies therefore require long, We thank C. L. Cummins, R. Greenberg, R. T. timed sequences on individual birds in which Holmes, T. C. Moermond, M. L. Morrison, T. A. Par- the length and kinds of every movement are re- ker, K. V. Rosenberg,and T. S. Schulenbergfor valu- able comments on the manuscript or portions thereof, corded. As already noted, food-handling time is and Donna L. Dittmann for preparation of Figure 1. a critical variable in studies of optimal diet se- The authors’ fieldwork, critical for the formulation of lection (e.g., Sherry and McDade 1982). the ideas and proposed scheme presented here, has been supported generously by (for JVR) John S. CLOSING REMARKS McIlhenny, H. Irving and Laura Schweppe, Babette Although portions of our classification scheme Odom, and the National Geographic Society, and (for have been used by us or other researchers for SKR) the Chapman Fund of the American Museum many years, other portions were novelties gen- of Natural History, the National Science Foundation, erated by rethinking the underlying logic of ear- and the Society of Sigma Xi.