Intraguild predation in a structured habitat: distinguishing multiple-predator effects from competitor effects

Transcription

1 Ecology, 90(9), 2009, pp Ó 2009 by the Ecological Society of America Intraguild predation in a structured habitat: distinguishing multiple-predator effects from competitor effects RUSSELL J. SCHMITT, 1,2,3 SALLY J. HOLBROOK, 1,2 ANDREW J. BROOKS, 2 AND JENNIFER C. P. LAPE 1 1 Department of Ecology, Evolution and Marine Biology, University of California, Santa Barbara, California USA 2 Coastal Research Center, Marine Science Institute, University of California, Santa Barbara, California USA Abstract. The ability to forecast community dynamics requires, among other things, an understanding of indirect effects and nonlinearities in webs of interacting species, together with knowledge of how habitat structure mediates interactions. We explored these aspects for a coral reef system in Moorea, French Polynesia, involving intraguild predation where the shared damselfish prey ( juvenile yellowtail dascyllus Dascyllus flavicaudus) and two species of intraguild (IG) prey (the ambush predators arc-eye hawkfish Paracirrhites arcatus and redspotted coral crab Trapezia rufopunctata) shelter together in branching Pocilloporid corals for protection from mobile IG predators. Field experiments revealed that both IG prey had strong adverse effects on survivorship of juvenile damselfish, but that the dominant underlying process was competition for enemy-free space and not direct consumption. In this case, the combined effect of hawkfish and crabs did not result in either risk enhancement or risk reduction for the damselfish. Similarly, the combined influence of IG predators and IG prey on the shared prey also was indistinguishable from that expected from their independent effects. Habitat structure weakened the IG-prey IG-predator interaction; IG prey were much stronger space competitors with damselfish than resource competitors with IG predators. As a consequence, there was not a trophic cascade where damselfish benefited by the adverse effect of IG predators on hawkfish or coral crabs. These results highlight the manner by which habitat structure can mediate species interactions and emphasize the need to understand the underlying mechanisms. Key words: competition; coral reefs; damselfish: enemy-free space; food web; indirect effects; intraguild predation; mortality; multiple predator effects; predator prey interactions; species interaction; trapeziid crabs. INTRODUCTION The ability to forecast community dynamics accurately requires knowledge of how the multiple underlying processes operate, particularly with regard to strong indirect effects and nonlinear responses (Wootton 1994, Agrawal et al. 2007, Schmitz 2007, Stachowicz et al. 2007). These can arise from biotic interactions such as the consumption of a prey species by multiple predator species (Soluk 1993, Sih et al. 1998, Schmitz 2007). It is widely appreciated that the combined influence of multiple predators on a shared prey can be either greater or less than that expected from the independent effects of each consumer (Soluk 1993, Sih et al. 1998, Vonesh and Osenberg 2003, Vance-Chalcraft and Soluk 2005, Griffen and Byers 2006a, b, Rudolf and Armstrong 2008). While risk reduction to prey appears to be the more frequently observed multiple predator effect (MPE) (Sih et al. 1998, Vance-Chalcraft and Soluk Manuscript received 27 June 2008; revised 26 November 2008; accepted 8 December Corresponding Editor: J. J. Stachowicz. 3 schmitt@lifesci.ucsb.edu , Vance-Chalcraft et al. 2007), risk enhancement can occur when one predator elicits a phenotypic (e.g., behavioral) response from a prey that increases its vulnerability to other predators (Soluk and Collins 1988, Vance-Chalcraft and Soluk 2005). Risk reduction can occur either when a predator induces a generalized defense in the prey that reduces its probability of death from several predator species, or when predators interfere with one another and reduce foraging success (Sih et al. 1998, Vance-Chalcraft and Soluk 2005, Griffen and Byers 2006b). One mechanism of predator interference is intraguild predation (Polis et al. 1989, Polis and Holt 1992, Holt and Polis 1997). Intraguild predation (IGP) occurs when species of consumers that compete for the same limiting resource also engage in direct trophic interactions with one another (Polis et al. 1989). In the classic IGP model framework, one predator species feeds on another with which it competes for a shared prey; the top consumer is the intraguild (IG) predator and the resource competitor it consumes is the intraguild prey. The phenomenon has received considerable attention in the theoretical literature (Holt and Polis 1997, Borer et al. 2007, Holt and

2 September 2009 MULTIPLE PREDATOR AND COMPETITOR EFFECTS 2435 Huxel 2007, Amarasekare 2008), which has suggested far-reaching implications of intraguild predation for such fundamental issues as the coexistence of resource competitors, the dynamics of communities, and the distribution and abundance of species. Recent metaanalyses indicate that intraguild predation is prevalent in natural communities (Arim and Marquet 2004, Vance-Chalcraft et al. 2007), but some empirical evidence runs counter to theoretical expectations (Borer et al. 2003, Janssen et al. 2007, Amarasekare 2008). For example, some IGP theory predicts a trophic cascade where the shared prey benefits from the consumption of IG prey by IG predators, although this was not supported by a recent meta-analysis of empirical studies of intraguild predation in unstructured habitats (Janssen et al. 2007). Of course, predators and prey frequently are not well mixed in nature (e.g., Fedriani et al. 2000). Habitat structural complexity can reduce the negative effect of intraguild predation on IG prey but apparently not on shared prey (Janssen et al. 2007). Prey in natural communities use structural refuges to reduce their risk of mortality from predators (Savino and Stein 1982, Holbrook and Schmitt 1988a, b, Hixon and Beets 1993, Steele and Forrester 2005), and IG prey and shared prey can use the same habitat structure for this purpose (Peckarsky and McIntosh 1998). This both reduces the efficacy of the refuge and raises the possibility that shared prey compete with IG prey for enemy-free space. While predicting effects of intraguild predation is challenging even in simple systems, understanding the added complexity of physical refuges requires knowledge of how IG prey and shared prey use habitat structure (Janssen et al. 2007). Furthermore, real interaction webs involving intraguild predation commonly involve more species than the most simplistic three species IGP community module (sensu Holt and Polis 1997), which itself can have at least a dozen direct and indirect interactions (Janssen et al. 2007). Intraguild predation typically occurs in complex food webs, with multiple possible prey items for predators (Rosenheim et al. 1997, Muller and Godfray 1999, Rott and Godfray 2000). Thus multiple species of IG prey can be prevalent in nature, and these intermediate predators may have nonindependent (emergent) effects (MPEs) on prey they share with IG predators. Here we explore a tropical reef network of interacting species with intraguild predation (Fig. 1) to (1) quantify strengths of direct and indirect interactions arising from the mutual use of a structural refuge by a shared prey and two species of intraguild prey, and (2) assess possible emergent effects of multiple predators on the shared prey. The shared prey are juvenile-stage damselfish, the yellowtail dascyllus (Dascyllus flavicaudus), that associate closely with branching coral in the genus Pocillopora (Fig. 2A). The damselfish are fed on by mobile predators that attack from the exterior of the coral (Fig. 2B; Holbrook and Schmitt 2002). Some of these mobile piscivores also feed on two IG prey species FIG. 1. Schematic of predator prey interactions in the branching coral food web studied in Moorea, French Polynesia. The shaded region represents refuge space afforded by branching coral in the genus Pocillopora. The coral is used by two intraguild (IG) prey (arc-eye hawkfish Paracirrhites arcatus; red-spotted coral crab Trapezia rufopunctata) and a shared prey ( juvenile yellowtail dascyllus Dascyllus flavicaudus) as a structural refuge from IG predators (e.g., longface emperor Lethrinus olivaceus). See Fig. 2 for photographs. that, like the damselfish, use Pocillopora colonies for protection (Fig. 1). The IG prey are the arc-eye hawkfish (Paracirrhites arcatus, Fig. 2C), and the red-spotted coral crab (Trapezia rufopunctata, Fig. 2D). Because of how the IG prey forage, we anticipated that the cooccurrence of hawkfish and crabs could enhance risk of young damselfish. We further expected that the joint use of coral by damselfish and these IG prey might negate the theoretical benefit to shared prey from consumption of IG prey by IG predators. We report here results of experiments to explore the nature of effects on juvenile damselfish arising from trophic and competitive interactions in this structured reef habitat. METHODS Study locality and study organisms Research was conducted on the north shore of Moorea, French Polynesia ( S, W), a small, oceanic island surrounded by a barrier reef that forms lagoons that are ;1 to 1.5 km wide and average ;5 m in depth. Field experiments were conducted in the lagoon 300 to 500 m from the barrier reef, where the bottom consisted of a mosaic of patch reefs and sand. Several species of Pocillopora are among the most common branching corals in the mid to outer lagoon habitats of Moorea. Although a number of species of invertebrates and fishes associate with Pocilloporid corals, planktivorous damselfishes (particularly the yellowtail dascyllus Dascyllus flavicaudus) dominate the

3 2436 RUSSELL J. SCHMITT ET AL. Ecology, Vol. 90, No. 9 assemblage (Holbrook and Schmitt 2002, Schmitt and Holbrook 1999a, b, 2000). After a planktonic developmental period, yellowtail dascyllus settle to branching coral where they remain for the rest of their lives. Individuals feed on plankton in the water column near their host coral during the day and retreat among its branches when threatened by predators and at night (Holbrook and Schmitt 2002). The damselfish coral relationship is a mutualism because yellowtail dascyllus (and other planktivorous damselfish) can enhance the growth rate of their coral host, possibly because they excrete ammonia when they shelter (Holbrook et al. 2008), while the host provides refuge space from external piscivores (Holbrook and Schmitt 2002). Mortality rates of juvenile yellowtail dascyllus scale with density because of competition for shelter space (Schmitt and Holbrook 1999a, b); individuals relegated to distal portions of the host by interactions with other residents are more vulnerable to predation by a suite of mobile predators (Holbrook and Schmitt 2002, Schmitt and Holbrook 2007). The symbiotic red-spotted coral crab Trapezia rufopunctata (Castro 1976, 1996) frequently occupies colonies of Pocillopora spp. at Moorea. Some Trapezia species defend their host corals against the corallivorous crown-of-thorns seastar, Acanthaster planci (Glynn 1976, Pratchett 2001), and they can interact aggressively with conspecifics and congeners to maintain territories within a coral (Huber and Coles 1986). In Moorea, T. rufopunctata display their chelae aggressively against fish that inhabit their coral and they can capture recently settled damselfish (R. J. Schmitt and J. C. P. Lape, personal observation). Pocilloporid corals tend to harbor a pair of adult red-spotted coral crabs, although colonies are not always occupied by this crab (J. C. P. Lape, unpublished data). Pocillopora colonies in Moorea also are used as shelter by a small-bodied ambush predator, the arc-eye hawkfish (Paracirrhites arcatus). Large colonies of Pocillopora are their preferred microhabitat (Kane et al. 2009), where they perch to hawk invertebrates and small fishes (Hiatt and Strasburg 1960, Shima et al. 2008) including young damselfishes (S. J. Holbrook and R. J. Schmitt, personal observation). When present, arceye hawkfish typically occur singly on a Pocillopora coral (Kane et al. 2009). Like yellowtail dascyllus, arc-eye hawkfish and trapeziid crabs use the space among the branches of their host Pocillopora for protection against a suite of mobile predators that attack from the exterior of the coral (Holbrook and Schmitt 2002). Some of these piscivores (e.g., emperors, jacks, groupers) functionally are intraguild predators because they consume hawkfish and crabs as well as damselfish that associate with branching corals (Fig. 1); for the purposes of this study we regard this suite of highly mobile consumers as the IG predator. Effects of IG predators and multiple IG prey on shared prey Laboratory feeding trials were conducted to estimate relative consumption rates of juvenile damselfish by both IG prey. Experimental arenas were 40-L Plexiglas aquaria (with flow-through seawater) that each contained a single Pocillopora eydouxi colony (;20 cm diameter) and either one arc-eye hawkfish (6 to 10 cm total length; N ¼ 8) or one red-spotted coral crab (;2cm carapace width; N ¼ 8); controls (N ¼ 2) had a coral colony but no predator. Ten recently settled yellowtail dascyllus (;15 mm total length) that had been captured by divers using hand nets were introduced into each aquarium, and the number that survived was recorded after 24 hours. Two (hawkfish) or three (crab) 24-hour trials were done consecutively for each predator (using naïve dascyllus each time) and the average number of damselfish consumed per 24 hours for each individual predator was calculated. A field experiment quantified the separate and combined effects of arc-eye hawkfish and red-spotted coral crabs on survival of recently settled yellowtail dascyllus. An array of Pocillopora eydouxi, each ;30 cm in diameter and 20 cm tall, was transplanted to a grid of cinder blocks (with 5-m spacing). Each Pocillopora was affixed to a cinder block with Z-Spar Splash Zone Compound (Kop-Coat, Inc., Los Angeles, California, USA) and assigned at random to a treatment. One set of treatments involved an orthogonal manipulation of the presence/absence of one adult arc-eye hawkfish and a pair of red-spotted coral crabs (N ¼ 5 of each of the four treatments). All Pocillopora colonies had a pair of naturally occurring red-spotted coral crabs (;1.5 2 cm carapace diameter), which were removed by divers from all corals except those randomly assigned to the appropriate two crab treatments (i.e., with crabs only, with crabs and hawkfish). A single adult arc-eye hawkfish (;6 7 cm total length), caught elsewhere by divers using hand nets, was released onto each Pocillopora colony assigned to the appropriate two hawkfish treatments (i.e., with hawkfish only, with crabs and hawkfish). All other fishes and invertebrates were removed by hand, and a cylindrical cage was placed over each coral to keep hawkfish from emigrating and other fishes from immigrating. Cages were 1 m in diameter and were constructed from hardware cloth with a mesh size (10 mm) that confined the hawkfish and crabs and excluded external mobile predators but allowed the prey (recently settled yellowtail dascyllus) to pass back and forth through the cage freely to feed in the water column and shelter in the host. Cage tops were adjusted to be ;15 cm above the highest point of the coral; given this and the open space between the coral and sides of the cage, hawkfish appeared able to hunt effectively. After they were added, the yellowtail dascyllus were observed to feed normally in the water column inside and outside of the cage adjacent to the coral host.

4 September 2009 MULTIPLE PREDATOR AND COMPETITOR EFFECTS 2437 FIG. 2. (A) Juvenile yellowtail dascyllus (Dascyllus flavicaudus) hovering near their host Pocillopora coral; (B) a longface emperor (Lethrinus olivaceus) attacking a prey trying to evade capture among the branches of a Pocillopora coral; (C) an arc-eye hawkfish (Paracirrhites arcatus) in ambush position on a branch of a Pocillopora coral; (D) a red-spotted coral crab (Trapezia rufopunctata) in characteristic pose nestled deep between the branches of a Pollicopora coral. Photo credits: (A, B) M. H. Schmitt; (C, D) R. J. Schmitt. The experiment was initiated when 12 recently settled yellowtail dascyllus (;15 mm total length; collected by divers and held in seawater aquaria overnight) were released onto each coral colony by divers, who counted the number remaining on each coral after 24 and 48 hours. Data from the 48-hour census were analyzed by two-way ANOVA (SAS 9.1; SAS Institute 2005), with the presence/absence of hawkfish and of crabs as factors. Because we wanted to test a multiplicative risk model (Soluk and Collins 1988), the variate used in the analysis was the log of the number of dascyllus remaining (Sih et al. 1998). In this model, a significant interaction term in the ANOVA denotes a nonindependent (or emergent) effect of the multiple predators (i.e., either risk enhancement or risk reduction). In addition to this statistical test for an MPE, we calculated the expected number of survivors if the effects of hawkfish and crabs were independent as Ehawkfish;crab ¼ ðshawkfish 3 Scrab Þ=Scontrol where S is the mean number of survivors in each of the treatments (Vonesh and Osenberg 2003, Vance-Chalcraft and Soluk 2005). This was compared with the mean number of survivors from the treatment where

5 2438 RUSSELL J. SCHMITT ET AL. Ecology, Vol. 90, No. 9 both hawkfish and crabs were present to further assess risk enhancement or reduction. Because recently settled yellowtail dascyllus do not move between corals separated by as little as 5 m (Holbrook and Schmitt 2002), we attributed the loss of outplanted damselfish from the coarse -mesh cages described above to mortality and not emigration. However, since juvenile damselfish could and did move through the cage mesh to feed in the water column, some mortality in these treatments could have been caused by external IG predators. To explore the separate and combined effects of external IG predators and internal IG prey, we included additional treatments in the field experiment. These were a second handling control (N ¼ 5) and an IG prey-only treatment (N ¼ 5); in both of these additional treatments the mesh size of the cages was small enough (2 mm) to confine juvenile yellowtail dascyllus within the cage. IG predators had no access to the 12 recently settled yellowtail dascyllus added to each coral in these fine-mesh treatments, which were censused after 24 and 48 hours. One additional treatment (N ¼ 3) consisted of uncaged corals where IG prey were kept removed, but where IG predators had complete access to the 12 outplanted dascyllus on each coral. Data from these three treatments plus the with hawkfish and crab, coarse-mesh treatment, which may have included IG predator effects, were analyzed with a two-way ANOVA (SAS Institute 2005). The variate used in the ANOVA was the log-transformed number of survivors at 48 hours (to test the multiplicative risk model), and the expected number of survivors if IG predators and IG prey had independent effects was calculated as described above and compared with the observed mean from the treatment where IG predators and IG prey all had access to the shared prey. Quantifying direct and indirect interactions The field experiment enabled partitioning of the loss of juvenile damselfish between (1) that caused by direct predation by hawkfish or red-spotted coral crabs, and (2) that due to consumption by IG predators due to sublethal interactions of shared prey with IG prey in the common refuge. To do this, we compared the loss of outplanted dascyllus (after 48 hours) when hawkfish or crabs were present in the coarse-mesh treatment to that in the fine-mesh treatments. For the coarse-mesh IG prey treatments, juvenile damselfish were only susceptible to predation by IG prey when inside the confines of the coral/cage complex, whereas damselfish became vulnerable to predation by IG predators when they swam through the mesh; a presumption is that sublethal interactions with IG prey might cause juveniles to leave the coral/cage complex. Because juvenile damselfish in fine-mesh treatments could not pass out of the cage, they were always confined with the IG prey but were never at risk by IG predators; as such, the fine-mesh treatments provided an estimate of loss due to direct consumption by IG prey. Hence, the difference between coarse-mesh and fine-mesh IG prey treatments is an estimate of the indirect effect of the IG prey on the mortality of the shared prey. A second field experiment explored the separate and joint effects on survival of juvenile yellowtail damselfish of known (older conspecifics) and of suspected (redspotted coral crabs) competitors for enemy-free space to determine whether qualitatively similar patterns emerged as were obtained in the predation experiment. The two factors manipulated in an orthogonal design (yielding four treatments) were the presence/absence of older yellowtail dascyllus and of red-spotted coral crabs. In contrast to the first experiment, this experiment was set up in a lagoon area with numerous patch reefs, primarily formed by Porites rus and Porites lobata, which harbor abundant IG predators (Holbrook and Schmitt 2002, Brooks et al. 2007). Twenty-eight colonies of Pocillopora eydouxi, assigned at random to each treatment (N ¼ 7), were transplanted to cinder blocks arranged ;15 m apart along a grid parallel to the reef crest. Subsequently all fish and invertebrates were removed from the focal corals, and 10 juvenile (4- to 6-month-old) yellowtail dascyllus (30 50 mm standard length [SL]) and /or a pair of adult red-spotted coral crabs (both collected elsewhere in the lagoon) were added in the designated combination for each coral. Manipulations were maintained for several weeks before the experiment was initiated to stabilize group sizes of the dascyllus. Recently settled yellowtail dascyllus were collected using a hand net and housed overnight in laboratory aquaria with flow-through seawater before they were outplanted to the experimental Pocillopora. Just prior to outplanting, experimental corals were cleared by hand of all organisms (except those required for the treatments), and 14 recently settled yellowtail dascyllus were outplanted to each experimental coral. Divers counted survivors (as well as crabs and older conspecifics) each morning for five days. One coral lost its outplanted damselfish immediately after placement and this replicate was eliminated from subsequent analysis. To test a multiplicative risk model, data on the number of survivors at five days were log-transformed and analyzed by two-way ANOVA (SAS Institute 2005) with the presence/absence of older conspecific damselfish and of red-spotted coral crabs as the factors. The distance (m) to the nearest patch reef was included as a covariate in the ANOVA, as was volume (as estimated by the shape and measured dimensions) of each host coral; for presentation, in Fig. 5 effects from proximity to the nearest patch reef and variation in size of the host coral were removed, and the residual mortality associated with the space competitor treatments is plotted. The expected number of survivors was calculated assuming that effects of older conspecifics and redspotted coral crabs on survival of juvenile dascyllus were independent, and this was compared with the observed

6 September 2009 MULTIPLE PREDATOR AND COMPETITOR EFFECTS 2439 mean number of survivors in the treatment where both space competitors were present. RESULTS Effects of IG predators and multiple IG prey on shared prey Laboratory trials confirmed that both IG prey (arc-eye hawkfish and red-spotted coral crabs) consume recently recruited yellowtail dascyllus in the shared refuge. The per capita consumption rate by arc-eye hawkfish ( damselfish consumedpredator 1 d 1 [mean 6 SE]) was roughly seven times greater than for a coral crab ( damselfish consumedpredator 1 d 1 ; t ¼ 2.61, df ¼ 7, P, 0.05). The field experiment to quantify the separate and combined effects of the two IG prey species on survival of young damselfish (where IG predators also had access to the shared prey) revealed that the loss rate of the shared prey was elevated in the presence of either a hawkfish (F 1,16 ¼ 22.1; P, 0.001) or coral crabs (F 1,16 ¼ 5.4; P, 0.05; Fig. 3A). The loss when a hawkfish was present was roughly twice that resulting from the presence of crabs (Fig. 3A). More importantly, however, the combined effect of these two IG prey on the loss rate of young damselfish was not synergistic (nonsignificant hawkfish 3 crab interaction: F 1,16 ¼ 0.001; P. 0.95). The observed mean (695% CI) number of damselfish survivors in the treatment with both IG prey ( survivors) was exceedingly close to and statistically indistinguishable from that expected (3.63) from the null multiplicative risk model (Soluk and Collins 1988) where the effects of hawkfish and crabs on damselfish survival are independent. When the separate and combined effects of IG prey and IG predators were quantified, the loss of young damselfish was fivefold greater from predation by IG predators (Fig. 3B). Moreover, the combined effect of both predator groups did not differ from that expected if their effects were independent (nonsignificant IG prey 3 IG predator interaction: F 1,19 ¼ 0.013; P. 0.35). The expected number of survivors if the effects of IG prey and IG predators were independent (4.36) was not significantly different than that observed ( [mean 6 95% CI]). Quantifying direct and indirect interactions The IG prey could have reduced survivorship of young damselfish in the field experiment directly by consuming them (predation) and/or indirectly by increasing their vulnerability to IG predators. Mortality rates of young damselfish due to direct predation by IG prey (Fig. 4A) were substantially lower than when IG predators also had access to the shared prey (Fig. 3A). Mortality of young damselfish due solely to direct consumption by crabs was exceedingly low and indistinguishable from handling mortality (Fig. 4A). The loss attributable to predation by hawkfish (Fig. 4A), while much greater than handling mortality, was only about a FIG. 3. (A) The separate and combined effects on mortality of juvenile yellowtail dascyllus from two species of intraguild prey (red-spotted coral crabs and arc-eye hawkfish) when IG predators also have access to the damselfish. (B) The separate and combined effects on mortality of juvenile yellowtail dascyllus from intraguild prey and the intraguild predators. Data are the mean (6SE) numbers of yellowtail dascyllus lost from the outplanted cohort in 48 hours. quarter of the mortality rate from hawkfish when IG predators also had access to the shared prey (Fig. 3A). All of the loss from coral crabs and ;75% of that from arc-eye hawkfish occurred because both IG prey greatly increased the vulnerability of young damselfishes to IG predators (Fig. 4B). The second field experiment, involving a known space competitor (older conspecific damselfish) and a putative space competitor (red-spotted coral crabs) of young damselfish, demonstrated that the combined effect of these two groups on the survival of young damselfishes also was not synergistic (Fig. 5). For the group sizes of older conspecifics and crabs used in the experiment, the presence of either older damselfishes or coral crabs had about the same effect on elevating the mortality rate of young damselfish (Fig. 5). Importantly, there was no synergistic effect of older conspecifics and crabs on the mortality (Table 1). The independent effects on mortality from the two groups is further evidenced by the close match between the observed ( [mean 6 95% CI]) and expected (3.2) number of survivors if the effects

7 2440 RUSSELL J. SCHMITT ET AL. Ecology, Vol. 90, No. 9 FIG. 4. (A) Mortality of juvenile yellowtail dascyllus due to direct consumption by red-spotted coral crabs and arc-eye hawkfish. Data are the mean (6SE) numbers of yellowtail dascyllus lost from the outplanted cohort in 48 hours (plotted on same y-axis scale as Fig. 3). (B) The amount of direct mortality of dascyllus due to predation by red-spotted coral crabs or arc-eye hawkfish (solid vertical line between the open square and x-axis) and indirect mortality resulting from competition for refuge space (dashed line between solid circle and open square). The solid circles are the estimated mortalities of juvenile dascyllus when IG predators also have access, and the open squares are the estimated mortalities due solely to consumption by crabs or hawkfish. alone (Sih et al. 1998). While each of the predator groups in our coral reef system caused (directly or indirectly) substantial death rates of damselfish, the independent effects of multiple predators indeed arose from a strong asymmetry among them in their role as predators, coupled with how habitat structure mediated species interactions. The IG prey functionally acted as strong space competitors with the shared prey and weak resource competitors with the IG predators. Empirical studies across a broad range of taxa have revealed that IG predators and IG prey typically show limited spatial overlap (Polis and McCormick 1987, Doncaster 1992, Durant 1998, Fedriani et al. 2000, Raymond et al. 2000, Sergio et al. 2003). In some systems, such as the coral reef interaction web we examined, spatial segregation is achieved by differential use of habitat structure by predators and prey. Prey commonly use habitat structure as a haven from predation (Savino and Stein 1982, Holbrook and Schmitt 1988a, b, Janssen et al. 2007), and IG prey and shared prey can use the same physical shelter (e.g., Hixon and Carr 1997, Peckarsky and McIntosh 1998, Brooks et al. 2007, Shima et al. 2008). Of course, use of the same shelter by shared prey and IG prey can make the refuge unsafe for the shared prey (Janssen et al. 2007, Shima et al. 2008). This perhaps helps explain the finding of a meta-analysis by Janssen et al. (2007), which revealed that the negative effect of IG predators on IG prey was reduced by habitat structure, whereas structure had no influence on the adverse effect of intraguild predators on shared prey. Complementary classes of ecological processes could operate to maintain strong negative effects of intraguild of crabs and older conspecifics were independent. That IG predators were the primary agent of death of young damselfish is supported by the pattern of variation in mortality with proximity of a host Pocillopora colony to patch reefs that harbored multiple IG predators; mortality declined significantly with increasing distance from the host coral to the nearest patch reef (Table 1). DISCUSSION Because nonindependence is the expectation when multiple consumers each strongly affect a common resource (Sih et al. 1998, Vance-Chalcraft and Soluk 2005), it is instructive to understand what attributes of our study system led to the independent joint effects of the multiple predators. In some previous studies, this pattern involved one consumer that caused negligible mortality compared to the other, so the presence of both had effects on the shared prey that did not differ substantively from that caused by the stronger predator FIG. 5. The separate and combined effects of red-spotted coral crabs and older conspecific damselfish on mortality of juvenile yellowtail dascyllus. Data are the mean (6SE) losses of outplanted dascyllus after five days, standardized by removing the effect of proximity to the nearest patch reef and variation in size of host coral (see Table 1).

8 September 2009 MULTIPLE PREDATOR AND COMPETITOR EFFECTS 2441 predators on shared prey that co-inhabit physical refuges with IG prey. One, of course, is predation by IG prey in the shared refuge, and the other involves sublethal effects that could arise, for example, from the perception of threat by the shared prey that alters a phenotypic response (e.g., behavior) and/or from competition with IG prey for enemy-free space (Holt and Lawton 1993, 1994). The type of sublethal response can matter. For example, some types of induced changes in the behavior of a shared prey by an IG prey can result in a multiple predator effect (MPE) where the combined influence of multiple predators is either greater or less than that expected from the independent effects of each predator (Soluk 1993, Sih et al. 1998). By contrast, competition for enemy-free space between IG prey and shared prey should not produce an MPE. While resource competition between IG predators and IG prey is an underlying premise for intraguild predation, to date, the potential for competition for enemy-free space between IG prey and shared prey in structured habitats has not been well addressed theoretically or empirically. In our study system, the strength of direct predation by IG prey on young damselfish was far weaker than that of the mobile IG predators. However, the strength of the indirect interactions of both IG prey on damselfish was quite strong and arose because hawkfish and crabs both increased the vulnerability of the shared prey to IG predators. Nonetheless, the strong sublethal effect of both IG prey did not result in risk enhancement, largely because IG prey and damselfish appear to be mostly space competitors. The weak trophic but strong competitive interactions of the IG prey with young damselfish functionally produced a relatively flat food web compared to the more typical IGP triangle structure; habitat structure weakened the IG-prey IGpredator interaction such that the system essentially behaved as if a mobile predator simply consumed a suite of competing prey at different rates. Of course, IG prey that compete for shelter space with shared prey need not be weak predators. The strength of nonindependent effects from multiple predators in this situation may hinge on the relative strengths of the trophic and competitive interactions by IG prey. Our field observations suggest that behavioral interactions between young yellowtail dascyllus and arc-eye hawkfish or red-spotted coral crabs increased the damselfish s vulnerability to mobile predators in a manner similar to that caused by conspecific damselfish. In a mechanistic study of the biology underlying density dependence in this system, Holbrook and Schmitt (2002) found that damselfish sheltering at or just outside the periphery of the coral s branching structure were most at risk to mobile predators. Red-spotted coral crabs occupy the safest, central region of the host coral where they aggressively defend their space by actively using their chelae to nip at small fishes that encroach (R. J. Schmitt, S. J. Holbrook, A. J. Brooks, and J. C. P. Lape, personal observation); this tends to drive individuals TABLE 1. ANCOVA tables for the loss of outplanted yellowtail dascyllus (Dascyllus flavicaudus) in five days from host corals in Moorea, French Polynesia. Source df MS F P Full model Older conspecifics 1, Crabs 1, ,0.05 Conspecifics 3 crabs 1, Coral volume 1, Nearest patch reef 1, Reduced model Older conspecifics 1, Crabs 1, ,0.05 Coral volume 1, Nearest patch reef 1, ,0.05 Notes: Main effects were the presence/absence of competitors (red-spotted coral crabs and older yellowtail dascyllus). Covariates were distance (m) to the nearest natural patch reef and volume of the host coral; colinearity between the two covariates was low (Pearson r ¼ 0.087; P. 0.65). Both the full (containing the crab 3 conspecific interaction term) and the reduced models are presented. See Fig. 5. toward the risky perimeter of the host. Similarly, arc-eye hawkfish shelter deep among the branches of a host coral when threatened; in doing so, they displace juvenile damselfishes from these safer portions of the refuge. Notably, hawkfish cease all hunting activity and shelter with damselfish upon the approach of an IG predator. Although altered behavior by young damselfish to the threat of predation by hawkfish also may contribute to their heightened vulnerability to IG predators, it did not result in an MPE. Functionally at least, the main source of damselfish mortality from both IG prey appears to involve interference competition for enemy-free space. Given that the intraguild predators in our system were greatly superior resource competitors to both hawkfish and crabs (i.e., the IG prey did not depress shared prey populations as much as the IG predators when each group fed in isolation), it is not surprising we found no evidence of a trophic cascade where young damselfish benefited from IG predators feeding on IG prey. Such a cascading positive effect on shared prey is predicted from three-species models of intraguild predation in well-mixed systems when the IG prey is the superior resource competitor (Holt and Polis 1997). These models further indicate that coexistence, while difficult to achieve, can occur when the competitive advantage of the IG prey is counterbalanced by their consumption by IG predators (Polis and Holt 1992, Holt and Polis 1997, Mylius et al. 2001). While IG prey ought to be excluded when they are the inferior resource competitor, intraguild predation is common in natural communities (Polis et al. 1989, Polis and Holt 1992, Rosenheim et al. 1995, Polis and Winemiller 1996, Arim and Marquet 2004, Briggs and Borer 2005, Borer et al. 2007, Vance- Chalcraft et al. 2007), and IG prey often may not be the superior resource competitor (Janssen et al. 2007). In the system we studied, the IG prey used the same physical

9 2442 RUSSELL J. SCHMITT ET AL. Ecology, Vol. 90, No. 9 structure as the shared prey for protection from IG predators, were inferior resource competitors to the IG predators, but were superior space competitors to the shared prey. Of course the short-term nature of our experiments precludes inferences regarding IGP interactions in relation to coexistence in this system (see Briggs and Borer 2005); nonetheless they clearly revealed how habitat structure modified the IG interaction. Because habitat structure appears to have the general effect of reducing the adverse effect of intraguild predation on IG prey in nature, prey refuges can provide an alternate path to coexistence (Janssen et al. 2007), along with such other complexities as resource subsidies and immigration (Heithaus 2001, Mylius et al. 2001, Finke and Denno 2002, Briggs and Borer 2005). Another aspect of real systems with intraguild predation that has yet to be fully considered is the influence of spatial interactions that arise from differences in the areas over which the participating species forage. For example, in our system, damselfish tend to be highly sedentary and remain closely associated with a single coral, hawkfish are ambush predators that can move among local collections of branching corals, whereas many IG predators such as groupers or emperors forage over much wider seascapes. This potentially general feature of systems can lead to important spatial interactions. While we explored how a structural refuge affected the direct and indirect interactions in a natural system with intraguild predation, the influences of spatial interactions and competition for enemy-free space on coexistence and community dynamics remain to be assessed. ACKNOWLEDGMENTS We thank K. Seydel, S. Beyers, L. Carr, S. Fejtek, R. Fisher, S. Holloway, Y. Ralph, and M. Schmitt for assistance in the field, C. Briggs and C. Osenberg for useful discussion, and B. O Connor for assistance with graphics. We gratefully acknowledge the support of the National Science Foundation (OCE ; OCE ), the W. M. Keck Foundation, and the Gordon and Betty Moore Foundation. This is a contribution of the Moorea Coral Reef (MCR) LTER Site and is Contribution No. 145 of the UC Berkeley Gump Research Station. LITERATURE CITED Agrawal, A. A., et al Filling key gaps in population and community ecology. Frontiers in Ecology and the Environment 5: Amarasekare, P Coexistence of intraguild predators and prey in resource-rich environments. Ecology 89: Arim, M., and P. A. Marquet Intraguild predation: a widespread interaction related to species biology. Ecology Letters 7: Borer, E. T., C. J. Briggs, and R. D. Holt Predators, parasitoids, and pathogens: a cross-cutting examination of intraguild predation theory. Ecology 88: Borer, E. T., C. J. Briggs, W. W. Murdoch, and S. L. Swarbrick Testing intraguild predation theory in a field system: does numerical dominance shift along a gradient of productivity? Ecology Letters 6: Briggs, C. J., and E. T. Borer Why short-term experiments may not allow long-term predictions about intraguild predation. Ecological Applications 15: Brooks, A. J., S. J. Holbrook, and R. J. Schmitt Patterns of microhabitat use by fishes in the patch-forming coral Porites rus. Raffles Bulletin of Zoology Supplement No. 14: Castro, P Brachyuran crabs symbiotic with scleractinian corals: a review of their biology. Micronesica 12: Castro, P Eastern Pacific species of Trapezia (Crustacea, Brachyura: Trapeziidae), sibling species symbiotic with reef corals. Bulletin of Marine Science 58: Doncaster, C. P Testing the role of intraguild predation in regulating hedgehog populations. Proceedings of the Royal Society B 249: Durant, J. M Competition refuges and coexistence: an example from Serengeti carnivores. Journal of Animal Ecology 67: Fedriani, J. M., M. Delibes, P. Ferreras, and J. Roman Local and landscape habitat determinants of water vole distribution in a patchy Mediterranean environment. Ecosciences 9: Finke, D. L., and R. F. Denno Intraguild predation diminished in complex-structured vegetation: implications for prey suppression. Ecology 83: Glynn, P. W Some physical and biological determinants of coral community structure in the Eastern Pacific. Ecological Monographs 46: Griffen, B. D., and J. E. Byers. 2006a. Partitioning mechanisms of predator interference in different habitats. Oecologia 146: Griffen, B. D., and J. E. Byers. 2006b. Intraguild predation reduces redundancy of predator species in multiple predator assemblage. Journal of Animal Ecology 75: Heithaus, M. R Habitat selection by predators and prey in communities with asymmetrical intraguild predation. Oikos 92: Hiatt, R. W., and D. W. Strasburg Ecological relationships of the fish fauna on coral reefs of the Marshall Islands. Ecological Monographs 30: Hixon, M. A., and J. Beets Predation, prey refuges, and the structure of coral-reef fish assemblages. Ecological Monographs 63: Hixon, M. A., and M. H. Carr Synergistic predation, density-dependence, and population regulation in marine fish. Science 277: Holbrook, S. J., A. J. Brooks, R. J. Schmitt, and H. L. Stewart Effects of sheltering fish on growth of their host corals. Marine Biology 155: Holbrook, S. J., and R. J. Schmitt. 1988a. Effects of predation risk on foraging behavior: mechanisms altering patch choice. Journal of Experimental Marine Biology and Ecology 121: Holbrook, S. J., and R. J. Schmitt. 1988b. The combined effects of predation risk and food reward on patch selection. Ecology 69: Holbrook, S. J., and R. J. Schmitt Competition for shelter space causes density-dependent predation mortality in damselfishes. Ecology 83: Holt, R. D., and G. R. Huxel Alternative prey and the dynamics of intraguild predation: theoretical perspectives. Ecology 88: Holt, R. D., and J. H. Lawton Apparent competition and enemy-free space in insect host parasitoid communities. American Naturalist 142: Holt, R. D., and J. H. Lawton The ecological consequences of shared natural enemies. Annual Review of Ecology and Systematics 25: Holt, R. D., and G. A. Polis A theoretical framework for intraguild predation. American Naturalist 149: Huber, M. E., and S. L. Coles Resource utilization and competition among the five Hawaiian species of Trapezia (Brachyura: Xanthidae). Marine Ecology Progress Series 30:

10 September 2009 MULTIPLE PREDATOR AND COMPETITOR EFFECTS 2443 Janssen, A., M. W. Sabelis, S. Magalhaes, M. Montserrat, and T. Van der Hammen Habitat structure affects intraguild predation. Ecology 88: Kane, C. N., A. J. Brooks, S. J. Holbrook, and R. J. Schmitt The role of microhabitat preference and social organization in determining the spatial distribution of a coral reef fish. Environmental Biology of Fishes 84:1 10. Muller, C. B., and H. C. J. Godfray Indirect interactions in aphid parasitoid communities. Researches on Population Ecology 41: Mylius, S. D., K. Klumpers, A. M. deroos, and L. Persson Impact of intraguild predation and stage structure on simple communities along a productivity gradient. American Naturalist 158: Peckarsky, B. L., and A. R. McIntosh Fitness and community consequences of avoiding multiple predators. Oecologia 113: Polis, G. A., and R. D. Holt Intraguild predation: the dynamics of complex trophic interactions. Trends in Ecology and Evolution 7: Polis, G. A., and S. J. McCormick Intraguild predation and competition between desert scorpions. Ecology 68: Polis, G. A., C. A. Myers, and R. D. Holt The ecology and evolution of intraguild predation: potential competitors that eat each other. Annual Review of Ecology and Systematics 20: Polis, G. A., and K. O. Winemiller Food webs: integration of patterns and dynamics. Chapman and Hall, New York, New York, USA. Pratchett, M. S Influence of coral symbionts on feeding preferences of crown-of-thorns starfish Acanthaster planci in the western Pacific. Marine Ecology Progress Series 214: Raymond, B., A. C. Darby, and A. E. Douglas Intraguild predators and the spatial distribution of a parasitoid. Oecologia 124: Rosenheim, J. A., H. K. Kaya, L. E. Ehler, J. J. Marois, and B. A. Jaffe Intraguild predation among biologicalcontrol agents: theory and evidence. Biological Control 5: Rosenheim, J. A., L. R. Wilhoit, P. B. Goodell, E. E. Grafton- Cardwell, and T. F. Leigh Plant compensation, natural biological control, and herbivory by Aphis gossypii on prereproductive cotton: the anatomy of a non-pest. Entomologia Experimentalis et Applicata 85: Rott, A. S., and H. C. J. Godfray The structure of a leafminer parasitoid community. Journal of Animal Ecology 69: Rudolf, V. H. W., and J. Armstrong Emergent impacts of cannibalism and size refuges in prey on intraguild predation systems. Oecologia 157: SAS Institute SAS version 9.1. SAS Institute, Cary, North Carolina, USA. Savino, J. F., and R. A. Stein Predator prey interaction between largemouth bass and bluegill as influenced by simulated, submersed vegetation. Transactions of the American Fisheries Society 111: Schmitt, R. J., and S. J. Holbrook. 1999a. Settlement and recruitment of three damselfish species: larval delivery and competition for shelter space. Oecologia 118: Schmitt, R. J., and S. J. Holbrook. 1999b. Mortality of juvenile damselfish: implications for assessing processes that determine abundance. Ecology 80: Schmitt, R. J., and S. J. Holbrook Habitat-limited recruitment of coral reef damselfish. Ecology 81: Schmitt, R. J., and S. J. Holbrook The scale and cause of spatial heterogeneity in strength of temporal density dependence. Ecology 88: Schmitz, O. J Predator diversity and trophic interactions. Ecology 88: Sergio, F., L. Marchesi, and P. Pedrini Spatial refugia and the coexistence of a diurnal raptor with its intraguild owl predator. Journal of Animal Ecology 72: Shima, J. S., C. W. Osenberg, and C. M. St. Mary Quantifying site quality in a heterogeneous landscape: recruitment of a reef fish. Ecology 89: Sih, A., G. Englund, and D. Wooster Emergent impacts of multiple predators on prey. Trends in Ecology and Evolution 13: Soluk, D. A Multiple predator effects: predicting combined functional responses of stream fish and invertebrate predators. Ecology 74: Soluk, D. A., and N. C. Collins Synergistic interactions between fish and stoneflies: facilitation and interference among stream predators. Oikos 52: Stachowicz, J. J., J. F. Bruno, and J. E. Duffy Understanding the effects of marine biodiversity on communities and ecosystems. Annual Review of Ecology, Evolution, and Systematics 38: Steele, M. A., and G. E. Forrester Small-scale field experiments accurately scale up to predict density dependence in reef fish populations at large scales. Proceedings of the National Academy of Sciences (USA) 102: Vance-Chalcraft, H. D., J. A. Rosenheim, J. R. Vonesh, C. W. Osenberg, and A. Sih The influence of intraguild predation on prey suppression and prey release: a metaanalysis. Ecology 88: Vance-Chalcraft, H. D., and D. A. Soluk Estimating the prevalence and strength of non-independent predator effects. Oecologia 146: Vonesh, J. R., and C. W. Osenberg Multi-predator effects across life-history stages: non-additivity of egg- and larval-stage predation in an African treefrog. Ecology Letters 6: Wootton, J. T The nature and consequences of indirect effects in ecological communities. Annual Review of Ecology and Systematics 25: