Studies in Avian Biology No. 13:6-l 3, 1990.

Quadjication of Resources

ECOLOGICAL AND EVOLUTIONARY IMPACTS OF BIRD PREDATION ON FOREST : AN OVERVIEW

RICHARD T. HOLMES

Abstract. Here I considerthe effectsof bird predation in both ecologicaland evolutionary time on forestinsects: their abundances,morphological and behavioralcharacteristics, life histories,and other traits. Most information concernsthe numerical impact of bird predation on defoliatingLepidoptera, especiallythose exhibiting population irruptions. Data indicate that birds have little effect on prey when prey are at outbreak densities.Although economicallyimportant and conspicuous,such irrup- tions are exhibited by relatively few Lepidopteran species,and even then they are often temporally and geographicallypatchy. I argue that these are unusual events and that the major foods of forest birds are insectsthat are most frequently at low or endemic population levels. Data, particularly those from the more quantitative and experimental studies, suggestthat birds along with other natural enemies help maintain low abundancesof such prey populations. This effect varies seasonally,being greatestduring the birds’ breeding periods when food demand is highest, and may result in frequent food limitation for insectivorousbirds in temperate forests. Another consequenceof the apparentlysustained and often strongnumerical impact of bird predation is evolutionary. Birds, through their selective foraging, are thought to be an important evolutionary force in determining many traits of their prey populations. One such consequencefor larvae is to influence their life styles and feeding scheduleswhich, in turn, determine the extent and patternsoftheir herbivory. In this way, birds through selectiveforaging indirectly affectother ecosystem componentsand processes.Such traits as crypsis,aposematism, restricted choice of feedingsubstrates, rigid feeding schedules,tissue or plant speciespreferences, and other features of life cycle organizationappear often to be a resultof selectivepressures exerted by bird predation. Many of these traits are also influencedby interactionsof the insectswith their host plants, thus forming a diffuse coevolutionarysystem. The implicationsof this view are that birds are not simply frills in ecological systems,but exert through their foraging activities important influencesin communities on both ecologicaland evolutionary time scales. Key Words; Bird foraging;predation; selectiveforaging; evolutionary impact; insectivorousbirds; defoliating Lepidoptera;forest insects.

The role of birds in natural ecosystems has tions showed that a small proportion of total long been discussed. A major issue has been energy and materials flowed through bird com- whether or not birds exert any controlling influ- ponents of natural ecosystems, and largely con- ence on the numbers of their prey. Although this cluded that birds had little direct effect on or possibility has been considered for centuries (see involvement in ecosystem processes(Wiens 1973, Murton 197 l), the first major scientific effort to Sturges et al. 1974, Holmes and Sturges 1975, evaluate such a role began with the studies of the Wiens and Dyer 1975). This led Wiens (1973: U.S. Biological Survey in the early part of this 265) to raise the possibility that birds in grass- century (e.g., McAtee 1932, Martin et al. 195 1) lands “. . . really are frills‘ ’ in the ecosystem, liv- and has continued largely through the efforts of ing and reproducing off its excesseswithout really forest entomologists (e.g., Morris et al. 1958, influencing it in any way.” He predicted, how- Campbell 1973). In general, results indicate that ever, that if birds have an important role, it would although birds consume large numbers of insects, be as controllers of other ecosystem components they rarely seem to exert any controlling or reg- (e.g., prey populations), through which consid- ulating effect, at least not on high populations of erably larger fluxes existed. economically important insects (see reviews by From more recent studies of the interactions McFarlane 1976, Otvos 1979). between birds and their food resources, espe- The possible roles of birds in ecosystem struc- cially manipulative studies, it seems that birds, ture and functioning, particularly in energy flow through their trophic relations, might have a more and biogeochemical cycling, were considered integral role in natural systems than has generally during the International Biological Program era been attributed to them. This has been most ap- in the 1960s and early 1970s. These investiga- parent in studies of bird-plant interactions in

6 IMPACTS OF BIRD PREDATION--Holmes 7

TABLE 1 QUANTITATIVE AND EXPERIMENTAL STUDIES OF THE NUMERICAL IMPACT OF BIRDS ON FOREST INSECTS, MOSTLY LEPIDOPTERA

Organism Population level Method” Reference

Eastern spruce budworm L Lo, Hi Density/consump. Morris et al. (1958) (Choristoneura L M Exclosure Dowden et al. (1953) fumiferana) L Lo, M, Hi Density/consump. Crawford et al. (1983), Crawford and Jennings (1989) Western spruce budworm L-P Lo, Hi Exclosure Torgersen and Campbell (C. occidentalis) (1982) Jack-pine budworm L Lo Density/consump. Mattson et al. (1968) (C. pinus) Larch sawfly L Lo, Hi Density/consump. Buckner and Tumock (Pristiphoruerichsonil] (1965) Ad Lo. Hi DensityIconsump. Buckner and Tumock (1965) Gypsy moth L Hi Density/consump. Inozemtsev et al. (1980) (Lymantria dispar) Codling Moth L-P Lo Exclosure Solomon et al. (1976) (Cydia pomonella) All leaf-dwelling L Lo Exclosure Holmes et al. (1979c, lepidopteran larvae unpubl. data) Peppered moth Ad Lo Experimental Kettlewell (1955) (Biston betularia) release/observ.

L = larvae, P = pupae. Ad = adult stage. /‘See text for descriptmn of methods.

which birds have been shown to be important timating bird densities and insect consumption pollinators (e.g., Regal 1982) and seed predators rates, and then comparing the latter to estimates or dispersers (e.g., Temple 1977, Thompson and of the standing crops of insects in the field. This Willson 1979, Herrera 1984a), influencing the indirect method involves many assumptions and evolution of various traits in their “prey” pop- sources of error that may be compounded at each ulations through selective foraging. Analogous step in the calculations. Yet, for situations in- effects of predators on their prey have recently volving large numbers ofbirds and high densities been explored for terrestrial systems in general of insects, it probably provides reasonable, albeit by Price et al. (1980) and for aquatic systems by order-of-magnitude, estimates (e.g., see Dowden Kerfoot and Sih (1987) and Sih (1987). et al. 1953; Morris et al. 1958; Buckner 1966, In this paper, I review the ecological and evo- 1967; Gage et al. 1970, Crawford and Jennings lutionary impacts of insectivorous birds on their 1989). Experimental approaches provide more prey, with emphasis on their interactions with precise information on the impact of avian pred- caterpillars (Lepidopteran larvae), which are an ators, but also have their problems: they are dif- important food source, especially for birds in ficult to conduct in natural situations and often temperate forests (Royama 1970, Robinson and require great effort and expense to obtain infor- Holmes 1982). I recognize two major interrelat- mation from both control and experimental plots ed ways in which foraging insectivores influence with sufficient replicates to provide statistically prey species: (1) a numerical effect in ecological meaningful results. Early attempts to remove time by reducing prey abundances, and (2) an birds from large forest tracts (e.g., Stewart and evolutionary effect by acting as selective agents Aldrich 195 1, Hensley and Cope 1951) or to that influence the prey’s morphology, behavior, exclude birds from small trees or single branches and life history characteristics, which in turn de- (Mitchell 1952) suffered from these difficulties. termine the activities and ecosystem roles of these Several recent investigations, however, have used insects. more rigorous and extensive experimental tech- niques. THE NUMERICAL IMPACT OF The best data currently available on the effects BIRDS ON INSECT PREY of bird predation on forest insects, mostly Lep- Quantitative assessments of the numerical im- idoptera, come from a relatively few studies that pact of bird foraging have usually involved es- have used either the bird density-prey consump- 8 STUDIES IN AVIAN BIOLOGY NO. 13

0 sprucebudworm n jack-pine budworm

+ Outside exclosure

Moderate High

Prey PopILINicm Level - I FIGURE 1. The impact of predation by forest birds on Lepidopteraas a function of prey density(see Table 5- 1 for referencesand text for further explanation). + Inside exclosure 4- * Outside exclosure tion technique in a detailed and rigorous way, 3- reasonably well controlled exclusion experi- ments, or studies of predation rates on released 2- adult moths (Table 1). Because most such studies were done on prey populations that regularly 1979 undergo periodic irruptions (and often cause eco- 1 -I nomic damage), I have classified the data from 1 each study as being obtained during periods of 00 low, moderate, or high population levels of the prey, based largely on the authors ’ assessments. June July High levels generally represent periods of insect FIGURE 2. Densitiesof Lepidopteran larvae on fo- outbreaks in which defoliation is extensive, mod- liage inside and outside of 10 exclosuresin 1978 and erate levels are those in transition before or after I979 in the Hubbard Brook ExperimentalForest, N.H. peak irruptions, and low levels reflect “normal,” Data from Holmes et al. (1979~) and Holmes and nonoutbreak conditions. Schultz (unpublished). Comparison of results of studies listed in Table 1 reveals (Fig. 1) two major points. First, it seems that birds take only a small percentage of the further illustrated by experiments conducted by available insects when they are present in high me and colleagues at the Hubbard Brook Ex- densities. Although they exhibit both numerical perimental Forest in New Hampshire (e.g., and functional responses to increasing prey den- Holmes et al. 1979~). In 1978 and 1979, we ex- sities (Morris et al. 1958, Sloan and Coppel 1968, cluded birds from patches of understory vege- Mattson et al. 1968, Gage et al. 1970, Holmes tation and measured densities ofall leaf-dwelling and Sturges 1975, Crawford and Jennings 1989), insects inside and outside ofthese exclosures. We birds seem unable to respond sufficiently to in- moved exclosures to different patches of vege- fluence the continued rise in the abundance of tation in 1979. In both years, the numbers of these prey (McFarlane 1976, Otvos 1979). Al- Coleoptera, Hemiptera, and spiders were not sig- though birds cannot keep up with a rapidly ex- nificantly different inside and outside of the ex- panding defoliator population, their relatively closures, probably because these more mobile strong impact at endemic levels (Fig. 1) suggests could readily move through the ap- that such predation could delay the onset of an proximately 2-cm mesh netting. For Lepidop- outbreak, as suggestedpreviously (e.g., Morris et teran larvae, which are more sedentary, the num- al. 1958, McFarlane 1976, Otvos 1979). Indeed, bers outside the exclosures were significantly modeling of spruce budworm populations sug- reduced in several of the sampling periods (Fig. gests that predation by birds may be a significant 2). Because other predators of these larvae, such factor in maintaining endemic population levels as wasps or possibly ants, were not excluded by of this species (Peterman et al. 1979, see also the netting, the reduction can be attributed al- Crawford and Jennings 1989). most entirely to birds. In the two years, birds The second point from Figure 1 is that the reduced larval numbers by 20 to 63%, varying impact of bird predation is proportionately much with the sampling period during the season; the greater when insects are at low densities. This is average reduction in each season was 37%. The IMPACTS OF BIRD PREDATION--Holmes 9

periods ofgreatest impact of bird predation were over the course of the winter. Other examples of in late June and early July in both seasons (Fig. birds reducing local abundances of insects are 2) which were times when birds were feeding given by Stewart (1975) Bendell et al. (198 l), nestlings and fledglings and thus when food de- Loyn et al. (1983) and Takekawa and Garton mand was probably greatest. (1984). and many anecdotal records are cited by These results, along with those in the literature Murton (197 l), McFarlane (1976) and others. (see Table l), suggest that birds can have signif- Finally, numerous studies, some manipulative, icant numerical effects on insect populations at have found significant effects of bird predation endemic levels. This finding is particularly sig- on the abundances of bark beetles and other bark- nificant in view ofthe fact that most forest-dwell- burrowing insects (see review by Otvos 1979). ing Lepidoptera and similar species in temperate Taken together, these findings suggest that birds forests typically occur at low densities and rarely probably have significant numerical effects in a ifever exhibit population irruptions (Morris 1964, wide variety of habitats and ecosystems. Mason 1987b). Even the few species that irrupt Finally, contrary to generalizations by Fretwell become abundant for only short periods and then (1972) and Wiens (1977) that limitation of many decline to low population levels for several years temperate bird populations may occur primarily (Berryman 1987, Wallner 1987). Moreover, when in the winter, evidence is accumulating that food outbreaks occur, they are often geographically may often limit insectivorous bird populations patchy (Campbell 1973, Martinat 1984). The re- in the temperate summer (Martin 1987). Recent sult is that any one forest stand may only occa- studies at Hubbard Brook in New Hampshire, sionally experience an outbreak. For northern for example, indicate that food becomes abun- hardwood forests, this may be once every 1O-20 dant only during insect outbreaks, which occur years (Holmes 1988) much longer than the life- sporadically and infrequently (Holmes et al. 1986, time of most individual birds. Consequently, Holmes 1988). Birds in these deciduous forests birds probably lack highly evolved systems for depend heavily on non-irrupting prey, whose detecting and responding to such temporal and abundances they further depress during the geographic variability, although a few species may breeding period (Holmes et al. 1979~; see above) do so (e.g., MacArthur 1958, Morse 1978b). at a time when the growth and survivorship of Hence, while outbreaks provide a locally abun- newly hatched young are greatly affected (Ro- dant food in some years and places, the endemic denhouse 1986). Birds in this temperate decid- population levels of most Lepidoptera and other uous forest appear to experience prolonged pe- arthropods provide the majority of the food riods of food limitation (Rodenhouse and source for birds most of the time. Holmes, in prep.) partly because of the strong Available data, such as those in Table 1 and numerical effect exerted by the birds themselves. Figure 1, suggest that the low abundances of in- I conclude that birds in temperate forests may sect species may be maintained at least in part exert a strong numerical impact on their arthro- by heavy predation pressure from birds, al- pod prey, and that this may occur most often though wasps (Steward et al. 1988b), ants (Camp- during the height of the breeding period. The bell et al. 1983) small mammals (Smith 1985) effect may be to depress or maintain insect num- as well as viral and other disease organisms, are bers at low levels and, in the case of prey species undoubtedly involved in various combinations. that exhibit population irruptions, to extend the This general importance of natural enemies in periods between such events. This is consistent the regulation of herbivorous insects, while con- with the syntopic population model developed troversial (Hassell 1978, Dempster 1983) is also by Southwood and Comins (1976) in which an supported by studies of the prey organisms em- “endemic ridge” is separated from an “epidemic ploying key factor analysis and other demo- ridge” by a “natural enemy ravine.” More large- graphic techniques (e.g., Varley et al. 1973, Pol- scale experiments on the impact of birds and lard 1979, Mason and Torgersen 1987; also see other enemies on endemic prey populations will Strong et al. 1984). clarify the extent and influence of such interac- It is difficult to generalize about the numerical tions. Extending such studies of the impact of impact of birds on groups other than Lepidop- bird predation on defoliators to tropical or other tera, largely because of the lack of detailed or ecosystems, or to other kinds of prey, experimental studies. However, Gradwohl and should be an important priority. Greenberg (1982b) showed through an exclusion experiment that tropical antwrens (Mymotherulu THE EVOLUTIONARY IMPACT OF fulviventris) reduced arthropods in dead leaf clus- BIRDS ON THEIR INSECT PREY ters by about 44%. Likewise, Askenmo et al. In the long term, the important effect on insect (1977) and Gunnarsson (1983) showed that birds prey of intensive foraging by birds will be evo- removed 17-50% of spiders on spruce foliage lutionary. For example, the 37-57% predation 10 STUDIES IN AVIAN BIOLOGY NO. 13 rates recorded by Kettlewell (1955, 1956, 1973; Nevertheless, passerine birds have been shown see Table 1) on the peppered moth have been to be able to distinguish between shape (Brower generally accepted as evidence of strong selection 1963), color (Jones 1932, Schmidt 1960, Brower by birds for the evolution of morphological and et al. 1964, Bowers et al. 1985), and pattern (Blest behavioral traits in this insect (Cook et al. 1986, 1956, Sargent 1968), which gives them the po- Endler 1986; but see Lees and Creed 1975). Since tential for being discriminate foragers (Curio available evidence indicates that predation at this 1976a). In some early experiments, Ruiter (1952) level by birds may be common (e.g., Table 1, showed that birds could distinguish geometrid Fig. l), it seems likely that birds could have had, caterpillars from similar inanimate objects and continue to exert, a strong selective influence (twigs), although movement of the prey was often on their prey. The possibility that birds and other required for this process to occur. Further, Pie- predators have an evolutionary impact on pat- trewicz and Kamil(1977) showed that Blue Jays terns of crypsis and other supposed predator- (Cyanocitta cristata) could discriminate cryptic avoidance traits in insects has long been recog- Catacola moths on bark, and Mariath (1982) nized (e.g., McAtee 1932, Cott 1940) and seems demonstrated that predation rates by birds var- to be more or less taken for granted by many ied with the proportion and spatial distribution biologists (but see Endler 1986). However, ram- of two morphs of a geometrid caterpillar and ifications of bird predation go beyond the evo- with the color of the plant background. Jeffords lution of crypsis or other antipredator traits that et al. (1979) painted diurnally flying moths to have not, in my opinion, been adequately con- look like swallowtail and monarch butterflies, sidered. These include influences on the life-styles, and showed that predators, mostly birds, distin- feeding patterns, and other characteristics of these guished among the different colors and patterns. insects, which in turn affect their involvement Moreover, Chai (1986) showed that jacamars and role in ecosystem processes, as I discuss be- (Galbula ruficauda) discriminated among tropi- low. cal butterflies on the basis of color and of taste, supporting the hypothesis that birds exert strong BIRDS AS SELECTIVEAGENTS ON INSECT selection pressures influencing the evolution of MORPHOLOGY AND BEHAVIOR mimicry patterns in butterflies. Not all evidence Birds have long been implicated as a major is positive, however. Lawrence (1985), for ex- agent of selection for aposematism (Harvey and ample, found that European Robins (Erithacus Paxton 1981) and mimicry (Wickler 1968, Rob- rubecula) and Great Tits (Parus major) did not inson 1969), as well as for nonmimetic poly- easily learn to detect cryptic prey. morphisms in various prey populations (e.g., Cain The degree to which an insect or other prey and Shepherd 1954, Allen 1974, Wiklund 1975, item is detectable probably depends most strong- Mariath 1982). Differential predation by birds ly on its choice of substrate and on its movement affects the sex ratio of their prey (Bowers et al. patterns. Those that choose an inappropriate 1985, Glen et al. 1981). Baker (1970) proposed substrate or that move at the wrong time should a variety of ways in which predation by birds be more subject to predation. Wourms and Was- may have influenced evolution of the sizes, serman (1985) showed experimentally that prey shapes, colors, and behavior of larval and pupal movement influences birds ’ feeding choices, and stages of Pieris butterflies, and Sherry and Sherry (1984) described how the behavior of cer- McDade (1982) inferred importance of bird pre- tain insects, including their movement patterns, dation on the shapes and sizes of tropical insects. makes them differentially susceptible to bird Also, the evolution of spines, hairiness, and other predators. Since most birds in terrestrial habitats similar features of insects and other prey are usu- are diurnally active predators that hunt by visual ally considered to be anti-predator adaptations means, they will be actively searching for and (Root 1966, Edmunds 1974). Waldbauer and taking prey from a variety of substrates, and any associates (Waldbauer and Sheldon 197 1, Wald- prey organism on the wrong background, moving bauer and LaBerge 1985) proposed that the ear- actively, or being otherwise conspicuous will be ly-season occurrence of certain hymenopteran- quickly removed. With many different bird mimicking Diptera was due primarily to strong species occupying a single habitat, each with dif- selection pressures by inexperienced birds for- ferent searching techniques and methods of prey aging in midsummer. Relevant to all of these capture (Smith 1974b, Robinson and Holmes examples, however, Robinson (1969) pointed out 1982, Gendron and Staddon 1983, Lawrence the paucity of experimental evidence concerning 1985, Holmes and Recher 1986a) and each being the adaptiveness and selective forces influencing fairly opportunistic and catholic in its prey pref- such presumed anti-predator traits. Two decades erence (MacArthur 1958, Rotenberry 1980a, later, this still appears to be the situation. Robinson and Holmes 1982, Sherry 1984), the IMPACTS OF BIRD PREDATION--Holmes I1

risk of predation is potentially high. Among the jor herbivores, namely caterpillars, by their an- bird species in northern hardwoods forests, for tipredator adaptations (i.e., indirectly by bird example, some closely scrutinize nearby sub- predation) and partly by their interactions with strates as they move along branches and twigs, the variable quality of the green leaves on which some examine undersurfaces of branches and they feed (see below). The hypothesis that I want leaves, while others move rapidly and flush prey to develop here is that bird predation, acting in from the foliage and twigs (Robinson and Holmes concert with the host plant and other factors, 1982). Furthermore, some forest birds differen- produces selective forces that act to organize and tially search and take prey from upper versus consequently influence the life history patterns- lower leaf surfaces (Greenberg and Gradwohl particularly feeding schedules-of leaf-chewing 1980, Holmes and Schultz 1988) and from par- forest insects. The arguments are similar to those ticular plant species (e.g., Holmes and Robinson of Price et al. (1980), but focus specifically on 198 1, Holmes and Schultz 1988). They also may bird-insect-plant interactions in forest habitats. use leaf damage caused by chewing insects as Because caterpillars do not mate, defend ter- prey-finding cues (Heinrich and Collins 1983) or ritories, or feed young (Schultz 1983a), their main develop search images (Tinbergen 1960) and “goals” are to accumulate biomass as rapidly as other forms of learning (Orians 198 1) to locate possible and to avoid being killed by natural ene- potential prey. All of these factors make it dif- mies (i.e., parasites, disease, and invertebrate ficult for the prey to go undetected, and likely predators as well as foraging birds; Heinrich have led to the evolution of the observed anti- 1979c, Schultz 1983a). Means of achieving these predator traits. goals may conflict. As argued by Schultz (1983a), The main points are that birds are discriminate maximizing feeding time and food quality should foragers and that they use the appearance and involve feeding throughout the day and night and behavior of their prey as major cues for locating because of variable food quality (see below), the those prey. These findings, coupled with the pos- larva may need to move frequently in search of sibility that birds are often food-limited and that new feeding places. At the same time, to avoid they can depress the numbers of their prey (ex- predation, the insect should minimize exposure cept during insect outbreaks), implicates birds as during feeding, which, if diurnally hunting pred- important and significant selective forces that in- ators are important, might be done by feeding fluence the evolution of many antipredator traits only at night or at least by restricting movement found among insects and other prey organisms. during daylight hours (Schultz 1983a). The situation is complicated because the qual- ECOLOGICALCONSEQUENCES OF THE ity of leaves for herbivorous insects varies sea- EVOLUTION OF ANTIPREDATOR sonally (Feeny 1970, Schultz et al. 1982), from TRAITS BY INSECZTS tree to tree, from one leaf to another (Schultz As reviewed above, most considerations ofthe 1983a, b), and even among different parts of a evolutionary effects of predators on prey have single leaf (Whitham and Slobodchikoff 1981). focussed on the morphological (e.g., size, shape, On sugar maple (Acer saccharum) and yellow color, hairiness) and behavioral (e.g., back- birch (Bet&z allegeniensis) trees at Hubbard ground choice, startle responses) traits of the prey. Brook, for instance, adjacent leaves on a single However, other equally interesting and impor- branch differ in chemical and physical properties tant consequences or ramifications of such traits important to herbivorous insects (Schultz 1983b). affect the life-styles and ecology of these prey Since caterpillars are capable of discriminating organisms. For instance, consider a caterpillar among chemical cues (Dethier 1970) and of mak- that mimics a twig. It must remain motionless ing behavioral “choices” of places to feed (Schultz on its correct substrate for its crypsis to be ef- 1983a), they should be able to respond to such fective, and any movement or change in sub- local variation, although this has not been well strate, at least during the day, is likely to increase documented (see below). Furthermore, short- the probability of its being detected by a foraging term changes in phenolics and other defensive bird. Its feeding may therefore be restricted to compounds can be induced by physical damage night hours when its risk of predation by birds to the leaves, such as that caused by tearing or is lowest. These constraints in turn affect the ways chewing (Haukioja and Niemala 1977, Schultz in which the caterpillar feeds, and hence its pat- and Baldwin 1982, Baldwin and Schultz 1983, tern of herbivory. Herbivorous insects in tem- West 1985, Bergelson et al. 1986, Hunter 1987). perate forests typically consume < 10% of annual Silkstone (1987) found that larvae fed less on leafproduction per year (Mattson and Addy 1975, damaged leaves, while Bergelson et al. (1986) Schowalter et al. 1986); this low level may result showed that simulated damage to single leaves in part from the constraints imposed on the ma- resulted in a significant increase in phenolic com- 12 STUDIES IN AVIAN BIOLOGY NO. 13

pounds within several days and that larvae moved larvae. These observations suggest that some of away from these areas, grew more slowly, and the partial chewing of leaves reported by Hein- took longer to reach the pupal stages. rich (1979~) may in fact have represented food Such short term induction of defensive chem- choice and later rejection by the caterpillar rather icals, if widespread, implies that the longer a cat- than a predator avoidance trait. On predation erpillar stays on a leaf, the higher the probability risk, Dammon (1987) showed that pyralid cat- that it will become less palatable. Thus, to op- erpillars survived better in leaf rolls than when timize feeding and growth, caterpillars may need exposed openly on leaf surfaces, and those on the to move periodically to new leaves in search of undersides of leaves survived better than those higher-quality feeding sites. This results in a trade- on upper surfaces. In addition, the risk factor off situation: if it feeds and moves extensively was apparently so important that the larvae chose during the day, it would be subject to high pre- leaves that were low in food quality. Hairy cat- dation; if it feeds only at night and remains mo- erpillars, which are generally less preferred by tionless through the day, it would probably not avian predators (Root 1966, Whelan et al. 1989) only grow more slowly but also take longer to might be expected to survive better or to have reach the pupal stage. The latter is important different feeding patterns from smooth-skinned, because longer development means the larvae cryptic larvae. However, I am unaware of any will be exposed longer to natural enemies, in- study that has made such a comparison. cluding parasites and disease (Pollard 1979, Caterpillars of some species in the forest at Schultz 1983b, Dammon 1987). Also, in tem- Hubbard Brook differ in feeding schedules and perate zones, night temperatures in spring and patterns of crypsis, which appear to reflect dif- early summer are often cool, which might in- ferent evolutionary responses to predation risk crease the energetic costs of searching at night, (Schultz 1983a). For example, honestaria as well as further slowing metabolic processes (Geometridae) remains motionless all day on and therefore growth. large twigs and branches far from feeding places, If this scenario is correct, one would expect where it closely matches the background; at night some relationship between feeding behavior and it moves long distances from its resting sites to the antipredator traits of the prey. Surprisingly, feeding areas and feeds during the dark hours. A little quantitative or experimental data exist on closely related geometrid species, Anugogu oc- the ecology and behavior of caterpillars with re- ciduaria, feeds during both the day and night, spect to food choice and predation risk, and most but possesses a cryptic pattern that matches the of what does exist is anecdotal. Heinrich (1979~) small twigs and petioles near the leaves where it reported that the feeding strategies and time bud- feeds; it is then able to “lean” over and take bites gets of palatable caterpillars were consistent with out of leaves during the day with only minimal their need to minimize predation. The species body movement (Schultz 1983a). Another geo- he observed either fed only at night or stayed on metrid, Cepphis urmaturiu, matches its own the underside of leaves, and often moved from feeding damage on the leaves and thus remains feeding sites after eating only small amounts of in feeding position throughout the day and night; leaf tissue. They also often clipped off partially it feeds around the clock. Thus, different patterns eaten leaves after feeding on them, which he pro- of crypsis seem to allow insects to exploit their posed was an antipredator trait reducing the food in different ways. This comparison of closely chances that birds would find the larvae by using related species is all the more interesting because leaf-damage cues (Heinrich 1979c, Heinrich and they co-occur on the same host plant, striped Collins 1983). Unpalatable larvae did not cut off maple (Acer pensylvunicum). partially eaten leaves, and were often seen ex- Although many of these ideas need experi- posed while resting and feeding on leaf surfaces mental verification, and more information is during daylight hours (Heinrich 1979~). Bergel- needed on the interactions between bird foraging son and Lawton (1988) found that larvae of two and prey defenses and feeding, the implications Lepidopteran species moved relatively little in from the hypotheses developed here are that the response to foliage damage, but became more evolutionary impact of bird predation, although vulnerable to predation by ants, but not by birds, indirect, has important ramifications on the life when experimentally forced to move. styles of the prey organisms and affects the struc- Schultz (1983a) found that caterpillars are often ture and functioning of other ecosystem com- specific in their choices. He also described ob- ponents. Birds are therefore not simply frills in servations of feeding caterpillars that appeared ecological systems, as suggested by Wiens (1973) to taste (mandibulating leaf edges) and often re- but exert through their foraging activities im- ject feeding sites. Lance et al. (I 987) report sim- portant influences in communities on both eco- ilar behavior by gypsy moth (Lymantria dispar) logical and evolutionary time scales. IMPACTS OF BIRD PREDATION--Holmes 13

ACKNOWLEDGMENTS the interactions between birds and insects, and the many My research at Hubbard Brook, which led to this colleagues and students who have contributed to the review, has been supported by grants from the National Hubbard Brook studies. J. T. Rotenberry, J. C. Schultz, Science Foundation, with the logistic support of the T’. W. Sherry, and B. B. Steele read and constructively Northeast Forest Experiment Station, U.S. Forest Ser- criticized early versions of the manuscript. I greatly vice, Broomall, PA. I thank Dr. Jack C. Schultz for his appreciate their efforts. stimulating discussions and collaborations concerning