Behavioral Ecology doi:10.1093/beheco/arj053 Advance Access publication 22 February 2006 Fine-scale substrate use by a small sit-and-wait predator

Douglass H. Morse Department of Ecology and Evolutionary Biology, Box G-W, Brown University, Providence, RI 02912, USA Downloaded from https://academic.oup.com/beheco/article/17/3/405/202064 by guest on 28 September 2021

Substrate choice is one of the most important decisions that sit-and-wait predators must make. Not only may it dictate the prey available but also the cover for the predator which may conceal it from prey or its own predators. However, while on a particular substrate the behavior and use of that substrate may vary widely. When naı¨ve, newly emerged crab spiderlings Misumena vatia (Thomisidae) occupied flowering goldenrod Solidago canadensis, their behavior differed markedly on inflorescences with rela- tively sparse and densely packed flower heads as well as on experimentally thinned and unthinned inflorescences. Initially, the spiderlings most often hunted at the thinned sites and hid among the dense flower heads at the unthinned sites, a difference that disappeared in all broods tested after 2–3 h, possibly because of the growing hunger of the initially concealed individuals. Prey capture (dance flies) in the thinned sites initially significantly exceeded that in unthinned sites but subsequently did not differ. However, spiderlings encountered their principal predator, the jumping Pelegrina insignis, significantly more often on unthinned than thinned inflorescences. Even though usage patterns initially differed strikingly, spiderlings did not differ in their rates of quitting the two types of sites. These results suggest a trade-off between foraging and predator avoidance that changes in response to increasing hunger over time. Key words: crab spider, foraging, Misumena vatia, predator avoidance, substrate use. [Behav Ecol 17:405–409 (2006)]

oraging and predator avoidance are two of the most im- patterns on a homogeneous substrate, the flowering inflores- Fportant selective factors confronting most at one cences of goldenrod Solidago canadensis, a favored foraging or more stages in their lives (Dukas, 1998; Morse, 1980). site. Although they emerge from their natal nests with some Although the ideal solution for at-risk individuals is to resources in their yolk sacs, the spiderlings must soon begin to achieve success in both foraging efficiency and predator feed or they will starve (Morse, 1993, 2000; Vogelei and avoidance, the two may seldom if ever be maximized simul- Greissl, 1989). At the same time, allocation to predator avoid- taneously. Usually predator avoidance involves behavior or ance is a particularly important problem for spiderlings be- habitat use that depresses food intake, growth, and poten- cause they are vulnerable to a wide range of predators on tially fitness (Downes, 2001; Lima and Dill, 1990; Sih, 1982). the flowers at this time, especially juvenile jumping Individuals might maximize their success either by seeking (ca. 1.5–10 mg) that frequent these sites (Morse, 1992). Once sites that improve their position in this trade-off or adapt they leave their natal nests, the spiderlings receive no maternal their behavior to the site currently occupied. If alternative protection (Morse, 1992). patches are not available or dangerous to access, they must While carrying out unrelated studies, I noted that spider- adjust their behavior to the site in which they find them- lings recently emerged from their natal nests captured large selves. Although experienced foragers may make behavioral numbers of small dipteran prey on certain goldenrod clones decisions on the basis of earlier encounters with predators at on which I placed them but captured virtually no prey on a site (Hugie, 2003; Jennions et al., 2003), totally naı¨ve for- other clones over a 2-h period. Spaces occurred between the agers, such as just-born young that fend for themselves with- flower heads of inflorescences on which spiderlings captured out parental care, must initially depend entirely on their prey, allowing them to move freely between exposed hunting innate capabilities (Morse, 2000, 2005). Because mortality sites and sheltered areas in the midst of the inflorescences. of young foragers is often particularly high (Curio, 1976; However, flower heads were so dense on other inflorescences Morse, 1992), early behavioral decisions may comprise an that after the spiderlings buried themselves, they did not ac- important factor in their survival and may also even have a cess the surface where their prey were aggregated. This system significant impact at a community level (Lima, 1998; Sinclair thus provided the opportunity to evaluate the effect of fine- and Arcese, 1995) if the vulnerable individuals provide an scale substrate characteristics on foraging behavior and change important resource for the predators. It is thus of particular in this performance over time. interest to investigate the newborns’ behavior on a fre- Therefore, I systematically investigated the role of flower- quented substrate to determine if their performances vary head density in determining the behavior of spiderlings on in response to the conditions experienced and to establish experimentally thinned and unthinned inflorescences of sev- whether their performance changes over time. eral clones, focusing on the following questions: (1) exposed Newly emerged crab spiderlings Misumena vatia (Thomisi- to hunting sites with likely different foraging and predator- dae) provide an excellent opportunity to test exploitation avoidance attributes, do naı¨ve spiderlings routinely use these sites in a way that varies with the fine structure of the inflor- Address correspondence to D.H. Morse. E-mail: d_morse@brown. escences? (2) Does this usage change over time? Naı¨ve spider- edu. lings prefer some flower substrates to others (Morse, 2000, Received 19 October 2005; revised 6 January 2006; accepted 15 2005), but (3) do they exhibit preferences for thinned or January 2006. unthinned goldenrod inflorescences that otherwise resemble each other? The Author 2006. Published by Oxford University Press on behalf of the International Society for Behavioral Ecology. All rights reserved. For permissions, please e-mail: [email protected] 406 Behavioral Ecology

METHODS Pelegrina(¼Metaphidippus) insignis, usually middle-instar juve- niles (88% of the jumping spiders on goldenrod at this time: Study area and subjects Morse DH, unpublished data). Juvenile P. insignis are older I carried out this work at the Darling Marine Center of the and larger than M. vatia spiderlings at this time, though as University of Maine, South Bristol, Lincoln Co., Maine, USA adults they are much smaller than penultimate and adult fe- (43 57# N, 69 33# W), in a 3.5-ha old field surrounded by male M. vatia and subject to predation by them (Morse, 1992). mixed coniferous-deciduous forest. The field, mown yearly in Much smaller numbers of similar-sized juvenile Eris militaris October, contains several grasses (Gramineae) and forbs. The and hoyi (9 and 3% of the jumping spiders, respec- main forbs flowering during the period of this study, August, tively) also occupied goldenrod inflorescences in the study were goldenrods, asters (primarily Aster umbellatus), and wild area and likely also captured M. vatia spiderlings, though carrot Daucus carota. The study area is described in further my data on crab spiderlings killed are currently confined to detail by Morse (2000). P. insignis (Morse DH, unpublished data).

Goldenrods (Solidago spp.) are the numerically dominant Downloaded from https://academic.oup.com/beheco/article/17/3/405/202064 by guest on 28 September 2021 forbs in the old fields and roadsides frequented by M. vatia (Thomisidae) in the vicinity of the study area. The common- Procedures est species in the study area and the one used in this study, S. canadensis, blooms from the latter half of July to early I surveyed the density of goldenrod flower heads on branches September at this site. S. canadensis is clonal and forms large of their inflorescences, randomly selecting fully flowering clusters of up to 75 or more flowering stems, ranging in height branches in the midst of inflorescences for analysis. I counted from 0.75 to 1.5 m. Their pyramidal yellow terminal inflores- the number of flower heads on a branch and measured the cences, approximately 12–25 cm tall, consist of several hori- length of the branch, permitting me to calculate substrate zontal branches, each bearing many (ca. 50–200) flower heads density. Ten branches from each of 10 clones were analyzed of approximately 2.5-mm diam and 4.5-mm height. A flower in this way. head is an urn-shaped structure containing large numbers of Initially, I placed a total of 32 individuals on the inflores- tiny flowers with a ring of ray flowers. Flower heads grow two cences of two goldenrod clones for 2 h. The density of flower to three abreast on the basal parts of the branches, but the heads differed twofold. At that time, large numbers of dance distal parts contain only a single row of heads. flies were visiting the goldenrods. The tiny, second-instar spiderlings (0.4–0.7 mg) emerge To evaluate further the performance of spiderlings on in- from their nests in leaves of the vegetation about 26 days after florescences with different densities of flower heads, I placed egg laying (Morse, 1993) and move rapidly on lines in search 10 individuals per brood on a randomly chosen flowering of satisfactory dwelling sites. Due to the ubiquity of goldenrod branch of a goldenrod clone (unthinned) and 10 others from at this time, many of them recruit to goldenrod, where they the same brood on a paired branch of the same clone from grow rapidly if small insect prey are abundant. If they fail to which I had removed alternate flower heads (thinned). After find a satisfactory site quickly, they usually balloon to an un- placement, I censused these sites at 10, 20, and 30 min to certain fate, which, given the prevalence of forest and water determine whether the spiderlings were hunting partly, or about the study area, will probably be an unsuccessful one fully exposed, or sheltered under the flower heads and (Morse, 1993). whether they had remained at the site on which they were I brought nests into the laboratory shortly before the spider- placed. Thirteen broods were tested in this way. lings emerged and placed these nests in 7-dram vials (2.5-cm To investigate the spiderlings’ behavior over a longer diam and 5-cm tall) at ambient temperature. The spiderlings period, I performed a similar experiment with censuses at emerged naturally from their nests. All spiderlings used in this 30 min, 1, 3, and 24 h, in which I again placed individuals study had emerged during the previous 1–3 days and none on thinned and unthinned flowering branches of goldenrod had previously fed. Neither had they experienced the sub- inflorescences. As before, I used 10 sibs from the same brood strates to which they were exposed in these experiments. How- on both unthinned and thinned branches, repeating this ma- ever, at this stage, they routinely perform their full repertoire nipulation for 12 broods. of substrate choice and dispersal behaviors (Morse, 2000). To evaluate the vulnerability of the M. vatia spiderlings to I selected spiderlings randomly from these broods. predation on different substrates, I used branches of golden- The spiderlings fed primarily on small dance flies (Empi- rod similar to the ones described above for this substrate-use didae sp.), which are periodically available in great numbers, experiment. I ran these branches in pairs, with one branch such that at certain times a spiderling at a ‘‘fly’’ site may feed thinned to approximately two-thirds its normal density and virtually ad libitum. These flies weigh 0.7–0.8 mg, and the spi- the other branch unaltered. I added five newly emerged sec- derlings captured them with little apparent difficulty. In the ond-instar spiderlings to both branches, a density not infre- absence of dance flies, the spiderlings fed primarily on other quently found on sites near a nest. Five minutes later, small, but less common, dipterans and thrips (Thysanoptera). I introduced a middle-instar P. insignis to both branches. Tests In that the spiderlings are sit-and-wait predators, it is impor- were run for 15 min, a period reflecting the time that tant to define my use of ‘‘hunt’’ because they seldom preface a searching of this size would likely remain an attack with searching or stalking tactics. Spiderlings were on or in the vicinity of a goldenrod inflorescence branch. treated as hunting if their large raptorial forelimbs were I recorded contacts or near contacts between jumping spiders spread widely, with the proximal segments roughly at right and crab spiderlings, including predation, hunting method angles to the body, a pose illustrated in Hu and Morse (2004). (roving and sit-and-wait), and evasive behavior of the spider- In contrast, spiderlings noted as ‘‘hiding’’ held their forelimbs lings. I used different individuals and different goldenrod close to their body. Although this position might also charac- branches for each test. Spiderlings used in this experiment terize any spiderling not hunting, when in dense vegetation came from five broods and were randomly assigned, except spiderlings could not spread their forelimbs to strike at prey. that members of the same brood were used for each paired Small jumping spiders are both the commonest and the run on both thinned and unthinned branches. Jumping spi- most important predators of the spiderlings (Morse, 1992). ders were collected from goldenrod inflorescences in the Several species of jumping spiders occupy the goldenrod in- field, using individuals ranging between 2.5 and 5.0 mg, the florescences. The commonest predator on the spiderlings was commonest size range present. Jumping spiders run on Morse • Spider substrate use 407 thinned and unthinned branches were paired for similar size in each replicate. I used JMP Version 3.1 (SAS Institute, Cary, North Carolina, USA) to calculate the repeated measures multivariate analyses of variance (MANOVAs). Null hypotheses were rejected at p ,.05. Means are accompanied by 1 SD (6SD).

RESULTS Lengths of branches, numbers of flower heads on a branch, and resultant density of flower heads differed markedly among representative S. canadensis branches from 10 ran-

domly chosen clones in the study area. More than twofold Downloaded from https://academic.oup.com/beheco/article/17/3/405/202064 by guest on 28 September 2021 range in length of the branches (3.3–7.9 cm range, mean ¼ 5.3 6 1.4 cm) is a consequence of both differences in inflo- rescence size and position within an inflorescence, which I did not explore systematically. Density of flower heads (number/ Figure 1 1 branch length) differed over threefold (10.4–34.8 cm , Percentages of spiderlings placed on natural (unthinned) and mean ¼ 19.4 6 7.38 cm1) and was not correlated with experimentally thinned densities of flowers on goldenrod inflores- branch length (rs ¼ .209, p . .5 in a two-tailed Spearman rank cences that were actively hunting at later points. Thirteen broods, correlation coefficient). 20 individuals per brood. Black bars ¼ experimentally thinned Only spiderlings placed on the low-density clone captured branches of inflorescences, white bars ¼ natural (unthinned) control. prey during the initial observations (G1 ¼ 15.42, n ¼ 12, 20; p , .001, G test). In contrast to burying themselves in the clone with densely packed flower heads, spiderlings on differ (F1,22 ¼ 0.87, p . .3, repeated measures MANOVA with the less dense inflorescences occupied the openings among arcsin transformation). As before, no significant differences the flower heads, where they were in intimate contact with the occurred among the broods (F3,20 ¼ 0.73, p . .5, same test). prey. Similar numbers of dance flies were present on the in- Roughly half of these individuals had left their sites at the end florescences of the two clones (19.3 6 8.2 flies/inflorescence of 24 h (55.0% for the thinned sites and 48.3% for the un- on the clone with no captures, 18.3 6 7.9 on the clone with thinned sites). Neither the cumulative departures (Z ¼ 0.629, captures; six censuses with 10 inflorescences/census; T ¼ 13; p . .5 in two-tailed binomial test) nor the numbers gone at p . .6 in a two-tailed Wilcoxon matched-pairs signed-rank any of the census periods (Z ¼ 0.147–0.641, p . .5–.8, same test), eliminating the likelihood of these differences in cap- tests) differed significantly at the two types of sites. ture rates being driven by prey numbers. In contrast to the short-term experiment, considerable In the short-term experiment, individuals on the thinned numbers of potential prey, mostly dance flies, were present branches of goldenrod inflorescences were significantly more during the longer experiment. The spiderlings caught signif- likely to hunt actively than were those on the unthinned sites icantly more prey over the period that behavior differed in the (Figure 1). This difference appeared clearly at the first census, thinned sites (14) than at the unthinned sites (5) (p ¼ .03 in 10 min after placement and remained relatively constant at a one-tailed binomial test), but no difference in prey capture 20 and 30 min. Numbers of hunting individuals remained stable occurred between these sites after the spiderlings’ behavior at over this period, about twice as high in the thinned sites as the two sites ceased to differ (p . .3, same test; thinned ¼ 24, the unthinned ones (Figure 1). Individuals initially hunting unthinned ¼ 32). tended to remain in hunting position throughout the 30-min Jumping spiders came in direct contact with 1.7 6 0.21 period. Proportions of hunting individuals were significantly spiderlings per run (total ¼ 26 contacts) on the unthinned higher in the thinned sites over this period (F1,24 ¼ 7.70, p ¼ goldenrod before the spiderlings could bury themselves .01, repeated measures MANOVA with an arcsin transforma- among the flower heads, compared to no recorded contacts tion) but did not differ between broods (F2,23 ¼ 0.10, p . .9, on the thinned goldenrod (T ¼ 0, n ¼ 15, p , .001 in a two- same test). In contrast to the original set of observations, few tailed Wilcoxon matched-pairs signed-rank test). Three of potential prey were present on the goldenrod during the these contacts resulted in captures, and in six other instances, short-term experiment. A single small dipteran was caught spiderlings exhibited highly evasive behavior, jumping out of during both the experimental and control runs. the vegetation and running, or hanging on lines below the Less than one-quarter of the individuals had left these sites vegetation for 2 min or more. More than six times as many by the end of the monitoring period (22.3% for the thinned spiderlings temporarily occupied the surface of unthinned sites and 19.2% for the unthinned sites). Neither the cumu- goldenrod as thinned goldenrod when the jumping spiders lative departures (Z ¼ 0.408, p . .6 in two-tailed binomial test) were introduced (3.2 6 0.17 versus 0.5 6 0.16 [T ¼ 0, n ¼ 15, nor the numbers gone at any of the census periods at the two p , .001, same test]), a consequence of the time required for types of sites (Z ¼ 0.042–0.750, p . .4–.9, same tests) were those on the unthinned sites to bury themselves within the significantly different. dense flower heads, providing many more opportunities for Although the second, longer term, test confirmed that in- jumping spiders to contact the spiderlings in the open. Most dividuals on thinned goldenrod branches initially were signif- of these contacts took place over the first 5 min after the icantly more likely to hunt at the surface of the inflorescences introduction of the jumping spiders (21 during 0–5 min than those on the unthinned sites (at both 30 min and 1 h, and 5 during 5–15 min), before many of the spiderlings had Figure 2), this difference disappeared over time, with the re- disappeared amid the flower heads, a significant difference sults at 3 and 24 h revealing no differences in hunting/hiding (Z ¼ 2.942, p ,.01 in two-tailed binomial test). Behavior of behavior between the two types of sites (Figure 2). Increased the jumping spiders differed markedly on the two substrates, numbers of individuals on the unthinned sites emerged to with 10 out of the 15 constantly searching on the unthinned hunting positions in the 3- and 24-h censuses, such that the goldenrod and only one of 15 on the thinned goldenrod overall results from the unthinned and thinned sites did not doing so (G ¼ 12.99, df ¼ 1, p , .01 in a G test). 408 Behavioral Ecology

into unthinned than thinned branches and did not search intensively in the thinned branches, instead usually taking up their own sit-and-wait predatory mode similar to those of the spiderlings occupying these sites. Thus, they exhibited a foraging strategy that differed with the grain of the substrate and resembled that of the spiderlings. The perceived grain of this substrate should vary with the size of these jumping spi- ders, with resultant consequences for the crab spiderlings; however, I did not explore this aspect of their relationship. The similarity of hunting patterns on thinned and un- thinned sites at 3 and 24 h raises the question of whether the initial differences in behavior on thinned and unthinned

inflorescences had a long-term effect because this difference Downloaded from https://academic.oup.com/beheco/article/17/3/405/202064 by guest on 28 September 2021 lasted for only a short time, perhaps as little as 2 h. Clearly, the spiderlings’ response to burrow into the unthinned inflores- Figure 2 cences provided them with more protection than if they oc- Conditions similar to Figure 1, but run for different periods of time. cupied the surface of an inflorescence. At-risk potential prey Twelve broods, 20 individuals per brood. Bars as in Figure 1. suffer higher mortality in unfamiliar circumstances (Altmann, 1998; Morse, 1980), and avoidance behavior should assume the greatest value at this time. However, their subsequent DISCUSSION movement into the open, presumably in response to hunger, I designed these experiments to test differences in substrate suggests a limit to this behavior and a trade-off between for- use associated with subtle differences in substrate structure. aging and predator avoidance. As a result, these individuals Although substrate preferences are clearly demonstrated in should also incur a high predation rate, as strongly suggested patch-choice studies of a wide range of systems (Pyke, 1984; by the jumping spider experiments. Stephens and Krebs, 1986), the behavioral differences in this The experimental manipulations effectively reflected the study demonstrated the ability of the naı¨ve spiderlings to re- significance of fine-scale substrate differences on the deci- spond to differences at a finer scale than that usually investi- sion-making and exploitation patterns of the spiderlings. gated. The innate response of the spiderlings to flower-head The mean density of S. canadensis flower heads along an in- density suggests that members of this population have regularly florescence limb (19.4 flower heads cm1) considerably ex- encountered fine-scale differences in substrate, as does the ceeds that of another common goldenrod in the study area, similar response of the different broods, including the time Solidago juncea (9.4 flower heads cm1: Morse DH, unpub- at which individuals on dense inflorescences shifted to hunt- lished data), that on average blooms slightly earlier, but ing, a risky behavior under the conditions experienced. The whose flowering period overlaps extensively with S. canadensis. spiderlings’ similar rates of leaving the thinned and unthinned A common later flowering, but temporally overlapping, sites in both experiments strongly suggest a lack of preference species, Solidago rugosa, has a flower-head density similar to for one inflorescence type over the other, in spite of their initial S. canadensis (17.3 cm1: Morse DH, unpublished data). Thus, success in capturing prey on the thinned inflorescences and one might predict an initially quantitatively different foraging the danger of hunting on dense inflorescences. This pattern of pattern on S. juncea than on S. canadensis or S. rugosa, which substrate acceptability resembles that of their dance fly prey; would primarily affect the earliest emerging spiderlings. Spi- however, the similar performance of the spiderlings in the derling emergence peaks during the flowering of S. canadensis, short-term experiment, when prey were largely absent, indi- the commonest of the goldenrods, after the peak for S. juncea, cated that the spiderlings’ behavior was independent of their and before the peak for S. rugosa, while the latest spiderlings prey. Given the similar availability of dance fly prey at both emerge simultaneously with S. rugosa. Critical substrates avail- thinned and unthinned sites, one would predict a greater rate able to the spiderlings thus not only vary spatially but also of leaving sites with jumping spiders than sites without them temporally in their physical characteristics. (see Dukas, 1998; Lima, 1998). Thus, the basis for the similar rate of leaving thinned and unthinned sites is unclear. I thank W. Brim-DeForest and T. Jones who assisted in the field. Few efforts have been made to evaluate responses of forag- R. Lutzy and R. Neff offered useful comments on an earlier draft of ers to such fine structural differences in their habitat (Romero the manuscript. G.B. Edwards identified the jumping spiders. I also and Vasconcellos-Neto, 2005). The differences in substrate thank K. Eckelbarger, T.E. Miller, and other staff members of the Ira use involved behavior that facilitated both prey capture and C. Darling Marine Center of the University of Maine for facilitating predator avoidance (see Dukas, 2001a,b). Although the spi- the fieldwork on their premises. Partially supported by the National derlings responded to the subtle substrate differences, their Science Foundation IBN98-16692. dance fly prey apparently did not discriminate between the two sites, at least in ways that affected their abundance. Thus, REFERENCES the spiderling predators exhibited a more discriminating re- Altmann SA, 1998. Foraging for survival. Chicago: University of sponse to subtle differences in their substrate than did their Chicago Press. prey. Such a predator-prey relationship is perhaps unexpected Curio E, 1976. The ethology of predation. Berlin: Springer-Verlag. in a simple food-chain analysis, where specialization of the Downes S, 2001. Trading heat and food for safety: costs of predator prey would be predicted to exceed that of the predator avoidance in a lizard. Ecology 82:2870–2881. (Hairston et al., 1960) but is more likely to occur in sit-and- Dukas R, 1998. Constraints on information processing and their wait predators like M. vatia than in widely ranging predators effects on behavior. In: Cognitive ecology (Dukas R, ed). Chicago: University of Chicago Press; 89–127. (Rosenheim et al., 2004). Dukas R, 2001a. Effects of perceived danger on flower choice by bees. Crab spiderlings were particularly vulnerable when on top Ecol Lett 4:327–333. of the branches, as reflected by the high contact frequency of Dukas R, 2001b. Effects of predation risk on pollinators and plants. just-released spiderlings with jumping spiders on the un- In: Cognitive ecology of pollination (Chittka L, Thomson J, eds). thinned sites. Jumping spiders were much less likely to burrow Cambridge: Cambridge University Press; 214–236. Morse • Spider substrate use 409

Hairston NG, Smith FE, Slobodkin LB, 1960. Community structure, Morse DH, 2000. Flower choice by naı¨ve young crab spiders and the population control, and competition. Am Nat 94:421–425. effect of subsequent experience. Anim Behav 59:943–951. Hu HH, Morse DH, 2004. The effect of age on encounters between Morse DH, 2005. Initial responses to substrates by naı¨ve spiderlings: male crab spiders. Behav Ecol 15:883–888. single and simultaneous choices. Anim Behav 70:319–328. Hugie DM, 2003. The waiting game: a ‘‘battle of waits’’ between pred- Pyke GH, 1984. Optimal foraging theory: a critical review. Annu Rev ator and prey. Behav Ecol 14:807–817. Ecol Syst 15:523–575. Jennions MD, Backwell PRY, Murai M, Christy JH, 2003. Hiding be- Romero GQ, Vasconcellos-Neto J, 2005. The effects of plant structure haviour in fiddler crabs: how long should prey hide in response to on the spatial and microspatial distribution of a bromeliad-living a potential predator? Anim Behav 66:251–257. jumping spider (Salticidae). J Anim Ecol 74:12–21. Lima SL, 1998. Nonlethal effects in the ecology of predator-prey Rosenheim JA, Glik TE, Goeriz RE, Ra¨mert B, 2004. Linking a pred- interactions. Bioscience 48:25–34. ator’s foraging behavior with its effects on herbivore population Lima SL, Dill LM, 1990. Behavioral decisions made under the suppression. Ecology 85:3362–3372. risk of predation: a review and prospectus. Can J Zool 68:619– Sih A, 1982. Foraging strategies and the avoidance of predation by an 640. aquatic insect, Notonecta hoffmanni. Ecology 63:786–796. Morse DH, 1980. Behavioral mechanisms in ecology. Cambridge, Sinclair ARE, Arcese P, 1995. Population consequences of predation- Downloaded from https://academic.oup.com/beheco/article/17/3/405/202064 by guest on 28 September 2021 Massachusetts: Harvard University Press. sensitive foraging: the Serengeti wildebeest. Ecology 76:882–891. Morse DH, 1992. Predation on dispersing Misumena vatia spiderlings Stephens DW, Krebs JR, 1986. Foraging theory. Princeton, New Jersey: and its relationship to maternal foraging decisions. Ecology Princeton University Press. 73:1814–1819. Vogelei A, Greissl R, 1989. Survival strategies of the crab spider Morse DH, 1993. Some determinants of dispersal by crab spiderlings. Thomisus onustus Walckenaer 1806 (, Arachnida, Ecology 74:427–432. Thomisidae). Oecologia 80:513–515.