SPECIES INTERACTIONS IN INTERTIDAL FOOD WEBS: PREY OR PREDATION REGULATION OF INTERMEDIATE PREDATORS?
BRUCE A. MENGE 
BRYON A. DALEY [2,4]
Abstract. Most natural food webs have more than one predator species, and many have trophic interactions among these predators. When a top predator feeds on an intermediate predator and they both feed on a shared basal resource, a phenomenon labeled intraguild predation (IGP), the potential exists for complex food web dynamics due to predation and competitive effects. Here we investigate the relative importance of direct predation vs. competition by a top predator on an intermediate predator.
The study system is the rocky intertidal interaction web formed by the predatory seastar Pisaster ochraceus, the predatory whelks Nucella emarginata and N. canaliculata, and a shared resource species, the mussel Mytilus trossulus. Previous experiments documented strong negative effects of Pisaster on mussels and whelks, but the mechanisms responsible for the effects on whelks, whether competition or predation, were not identified. Here we report results of a field experiment that manipulated both Mytilus and Pisaster to determine the short- and longer-term changes in whelk populations. Using a simplified dynamic model for changes in abundance over the initial stages of the experiment, we separated and quantified the top-down effect of direct predation by seastars vs. the bottom-up effects of competition and food limitation. Short-term results were in agreement with longer-term responses. Results suggest that direct and indirect bottom-up influences of mussels were far stronger than predation, and thus, whelk increases in the absence of seastars were due to reduced competition with Pisaster. Large differences between the body sizes of seastars and whelks make it difficult to determine the ultimate nature of the resource under competition between predators. Small mussels may constitute only a food resource for seastars, but to some extent, they also represent a microhabitat for whelks. Differences in the magnitude of the response to mussel manipulations between Nucella species might be due to slight differences in the way the species utilize mussel beds. Some of the predictions of theoretical IGP models regarding coexistence and stability of species may not apply to this interaction web because it includes species with both "open" and "closed" populations, rather than just closed populations as assumed by the models.
Key words: bottom-up effects; effect sizes; exploitation competition; intraguild predation, mussels; Mytilus; Nucella; Pisaster; rocky intertidal; seastars; top-down effects; whelks.
Most food webs have more than one predator, and many also have a complex pattern of competitive and trophic interactions among these predators. In the simplest food webs, predators feed on the trophic level immediately below: a "top" predator feeds on an "intermediate" predator, which in turn feeds on herbivores or basal species. When this pattern of trophic interactions dominates food webs, the removal of species at the top trophic level results in a chain of effects producing a classic trophic cascade with alternation of positive and negative biomass responses between trophic levels (e.g., Hairston, Smith, and Slobodkin 1960, Oksanen et al. 1981, Carpenter et al. 1985, Bronmark et al. 1992, Hunter and Price 1992, Moran et al. 1996). Specific predictions about the forces that regulate different trophic levels can then be made and some of them have been experimentally tested (e.g., Bronmark et al. 1992, Wootton and Power 1993, Wootton et al. 1996). However, this food web scenario can be substantially complic ated by the existence of intraguild predation (IGP), in which a top predator feeds on an intermediate predator and both feed on a shared prey (Polis et al. 1989). In this case, predictions about the effects of predation will depend on the relative strengths of exploitation competition for prey, vs. direct predation of one predator on another (Spiller and Schoener 1990, Polis and Holt 1992, Holt and Polis 1997).
The widespread occurrence of intraguild predation and other forms of omnivory in natural food webs (e.g., Bradley 1983, ter Braak 1986, Menge and Sutherland 1987, Polis 1991, 1994, Diehi 1992, 1993) has led Polis and Strong (1996) to question the validity of the trophic level concept and the generality of trophic cascades. If intraguild predation among predators is strong, predation should keep predator densities low and prey abundance is predicted to be high (Fig. 1, model A). Alternatively, if intraguild predation is weak, but predator effects on shared prey are strong, exploitation competition may structure the higher trophic levels and prey should be subject to severe reductions (Fig. 1, model B; see Spiller and Schoener , Diehl ). A plethora of indirect effects, including resource competition, apparent competition, and mutual enhancement are possible in food webs with IGP (Polis et al. 1989). A simple theoretical analysis of a three species community with IGP, leads to the prediction that fo r an intermediate IG predator to coexist with a top IG predator (see Fig. 1), the former must be superior at exploiting shared prey resource (Polis and Holt 1992). Building on these simple models, Holt and Polis (1997) have recently formulated a basic theoretical framework for intraguild predation, leading to the identification of a series of conditions necessary for the stable coexistence of IG predators, IG prey and a basal resource species. Predictions from these models have not been experimentally tested.
In most marine intertidal food webs, several co-occurring predator species interact trophically (e.g., Dayton 1971, Menge 1983, Robles and Robb 1993, Navarrete and Menge 1996). The focus of most work in these systems has been on interactions affecting the basal trophic level (filter feeders and macroalgae), however, and relatively little experimental work has documented the effects of predation among trophically high species (but see Garrity and Levings 1981, Menge et al. 1986, Barkai and McQuaid 1988, Robles and Robb 1993).
On the Oregon coast, recent studies have suggested an inverse relation between predation by seastars, the "top" keystone Pisaster ochraceus and the size and local abundance of "intermediate" predators, the whelks Nucella canaliculata and N. emarginata (Menge et al. 1994, Navarrete and Menge 1996). The mechanism(s) underlying the increase in whelk density and, in some cases, average individual size following the removal of seastars was (were) not revealed by these experiments. Although seastars prey on whelks, frequencies of whelks in Pisaster diet are always low (Landenberger 1968, Menge 1972, Menge and Menge 1974, Sanford 1999). While this might suggest that predation by seastars on whelks is weak, the prevalence of prey items in the diet of predators is often not a good indicator of the actual impact of predators on prey populations (Paine 1980, Fairweather and Underwood 1983, Polis 1994). Steady predation on whelks, which are not abundant in the low intertidal zone where these consumers coexist, could hav e an important effect on whelk abundance or size. Alternatively, the changes could have been due to reduced exploitation competition. While Pisaster has a very broad diet that includes barnacles, limpets, snails, and chitons, mussels are its preferred prey (Paine 1974). Whelks, Nucella spp., feed almost exclusively on mussels and barnacles (Dayton 1971, Palmer 1984, Wieters and Navarrete 1998). Thus, whelks share a major prey resource with seastars, and both seastars and whelks have been documented to severely depress mussel abundance (Dayton 1971, Paine 1974, Menge et al. 1994, Navarrete and Menge 1996). Previous experiments found no detectable effects of whelks on seastars (Navarrete and Menge 1996), and thus if competition occurs, it is expected to be highly asymmetric in favor of seastars. Since both seastars and whelks feed most intensively on barnacles and mussels, they clearly co-occupy the predator guild in this community.
In this study we experimentally investigated the relative effects of intraguild predation vs. exploitation competition on the abundance of two species of whelks. We tested the effect of predation by manipulating seastar density, as in our earlier experiments. We evaluated the effect of exploitation competition by manipulating the abundance of small mussels that are available to both seastars and whelks. The relative sizes of seastars and whelks with respect to mussel prey makes it difficult to determine the ultimate nature of the resource under competition, however. The small mussels consumed by whelks and seastars are similar in size to adult whelks, and both are much smaller than adult seastars, In addition to food resources and unlike adult seastars, whelks could find refuge from other predators (e.g., birds) or from environmental conditions by hiding amongst the mussels, although by doing so they are probably more exposed to incidental predation by seastars. Although our experimental areas provided abund ant refuges for whelks under all treatment combinations, the experiments were not specifically designed to determine the relative importance of mussel beds as food or microhabitat. Thus, our evaluation of "exploitative competition" includes both the indirect effect of seastars on whelks through reducing food and, to some extent, through removing microhabitat.
By determining the responses of whelks to all combinations of seastar and mussel presence and absence and using a simple dynamic model that incorporates IGP, we were able to quantify and separate the effects of direct predation by and competition with seastars, as well as the direct bottom-up effect of prey resource. Our results suggest that the direct and indirect bottom-up influence of mussels was far stronger than predation, and thus that whelk increases in the absence of seastars were primarily due to reduced exploitation competition with Pisaster.
This study was carried out from April 1994 to May 1995 at Strawberry Hill (44[degrees]l5' N, l24[degrees]07' W; hereafter SH), located [sim]50 km south of Newport on the central Oregon coast, and described in Menge et al. (1994) and Navarrete and Menge (1996). Briefly, this site consists of a broad rocky bench whose numerous channels, tidepools, and outcrops break into a complex topography. Wave-exposed, low zone rocky surfaces are dominated at this site by sessile and mobile invertebrates; while seaweeds and surfgrass are present, they are a relatively minor component of community structure as compared to other sites along the coast of Oregon. Further, abundances of certain invertebrates vary annually in a predictable pattern (S. Navarrete, B. Menge, and B. Daley, personal observation). In late summer, M. trossulus are usually nearly absent; the barnacle Chthamalus dalli is typically the most abundant sessile invertebrate, while much rock surface is bare. Mussel settlement peaks in early fall (Menge et al. 1994), usually leading to the formation of large patches (from tens to hundreds of square meters in size) of M. trossulus in the low zone by mid- to late-winter. Another barnacle, Balanus glandula, typically settles from spring through autumn (Menge 2000), briefly reaching moderately high densities in the low zone. The gooseneck barnacle, Pollicipes polymerus, sometimes also settles heavily in the low zone, and can form extensive patches persisting for several years (B. Menge, unpublished data). Seastar activity increases from late winter into summer, depleting the mussel patches by mid to late summer.
Seastar predator and mussel prey manipulations
The mussel patches characteristic of the wave-exposed low zone at SH offered a satisfactory compromise between the conflicting needs for a relatively large area, with high predator densities, but an area small enough to make mussel removal feasible. The occurrence of mussels in relatively large patches permitted us to test the effects of predation and food on whelks using relatively (potential) artifact-free manual manipulative methods (as opposed to exclusion cages or fences).
Sixteen mussel patches (experimental plots) between 1.4 and 15.6 [m.sup.2], relatively isolated from each other by surge channels and spread along a 500-m stretch of shore were located and marked with 10 X 10 cm Plexiglas plates engraved with treatment codes and plot number in March and April, 1994. Plot size varied according to the presence of surge channels and crevices and was not manipulated. The design included four replicates for each of the four combinations of seastar predator (present: +, or absent: -) and mussel prey (present, absent) treatments: +Pisaster+Mytilus, +Pisaster-Mytilus, - Pisaster+Mytilus, and -Pisaster-Mytilus. Each treatment combination was assigned randomly to plots, although in two cases we had to reassign treatments (nonrandomly). These reassignments were made because seastar-removal plots were so close to seastar-present plots that removal would have influenced densities in the neighboring nonremoval plots. Immediately above the experimental plots there were beds of the large mus sel Mytilus californianus, which guaranteed the provision of refuge for whelks even in the absence of M. trossulus.
On 11 April 1994, abundance of mussels, other sessile invertebrates and whelks was estimated using 0.25-[m.sup.2] quadrats subdivided into a grid of 25 10 x 10 cm squares. Percent cover of sessile animals was estimated visually with the aid of the 10 X 10 cm grid. Density of whelks was estimated by counts of individuals in the entire quadrat. Since quadrats represented subsamples in each experimental plot, analyses used means of all quadrats for each plot. Seastar density was simply the number of individuals occurring in each experimental plot; divided by plot area. Plot area was estimated by measuring distances from the center of each plot to plot edge, estimated at 36[degrees] intervals, for a total of 10 measurements. These data were plotted on polar coordinate paper and area was estimated using a planimeter.
After these premanipulation estimates were made, seastars and mussels were removed from the corresponding removal plots. Seastars were pried from each removal plot and placed in appropriate habitat away from the plots. Mussels were removed using scrapers, a process that necessarily also removed most acorn barnacles intermingled among the mussels. These were "press" perturbations (sensu Bender et al. 1984), so re-invading seastars and recruiting mussels were removed repeatedly. For seastars, removal frequency varied with seastar activity and season. In spring and summer, removals were done on most visits to this site, so ranged from daily to approximately every two weeks. In fall and winter, weather often prevented access to the site, but we were able to detach seastars at two to four week intervals. Due to time constraints, we were not always able to record seastar density in the seastar control plots during these visits.
Because we were unable to prevent re-invasion of seastars from removal plots during fall 1994 and because our design allowed whelks (and other predators) to prey on mussels in the patches, the mussel cover in the +Mytilus plots (unmanipulated mussel beds) decreased to the levels created in the --Mytilus plots. By late winter 1995, the mussel cohort from fall 1994 had again reached covers [greater than]50% in the +Mytilus plots, while rescraping the -Mytilus plots, after the fall mussel recruitment pulse, maintained mussel cover close to zero. Thus, the "mussel treatment" was effective during spring and summer 1994; it was not effective from late fall 1994 to late winter 1995, and then it was effective again in spring and summer 1995.
After initiation of the experiment, abundances of whelks and sessile invertebrates were estimated eight times between April 1994 and May 1995. In addition, we twice estimated average shell lengths per plot of each whelk species, once at initiation and once again in June 1994. This was done to ensure that size structures of the populations in each patch were initially comparable among the plots in the different treatments.
Because the effectiveness of the mussel treatment varied during the course of the experiment, data were divided into four time intervals for purposes of analysis: (a) early spring 1994, before manipulations were performed, (b) late spring and summer 1994, when the mussel treatment was effective, (c) fall and winter 1994-1995, when the mussel treatment was not effective but the plots continued to be monitored, and (d) spring and summer 1995, when the mussel treatment was again effective.
Whelk densities before manipulations were compared using two-way ANOVA with Mytilus and Pisaster densities as orthogonal treatments. Log-transformed data exhibited visually normal distribution and variance homogeneity according to Cochran's C test (Underwood 1997). Log transformation seemed appropriate not only to improve variance homogeneity, but because the mussels and seastars are expected to have multiplicative effects on whelk population size (Wilbur and Fauth 1990, Wootton 1994, Osenberg and Mittelbach 1995). Since it can be argued that Nucella species might affect each other's densities within experimental plots, making separate ANOVAs nonindependent, we also compared densities before manipulations using a two-way MANOVA (Huberty and Morris 1989, Scheiner 1993) with the two Nucella species as dependent variables. Because the same experimental plots were followed over time, changes in whelk densities after application of treatments were analyzed using repeated-measures multivariate analysis of variance (RM-MANOVA; Crowder and Hand 1990, von Ende 1993), separately for both Nucella species.
Analyses were based on mean whelk densities per replicate during spring--summer 1994, fall--winter 1994-1995, and spring--summer 1995. Again, a multivariate analysis of variance considering both Nucella species simultaneously was conducted to compare the overall treatment means. Low degrees of freedom did not allow us to conduct repeated-measures MANOVA. Percent cover values of acorn barnacles (Balanus and Chthamalus) were similarly analyzed with RM-ANOVA, but only for the last three sample dates (January, April, and May 1995), because their abundances were near zero in all treatments until fall 1994. Arcsine transformation of percent cover data was necessary to meet the assumption of variance homogeneity (Underwood 1997). Since seastars might also respond to the mussel manipulations in the + Pisaster plots, we compared seastar abundance between + Mytilus and --Mytilus treatments using one-way RM-ANOVA. A one-way ANOVA was used to compare premanipulation seastar abundances.
Since whelks of the genus Nucella have no free swimming larval stage, their populations are "closed"; that is, local processes impinging on the reproductive potential of adults will directly affect local population abundance (Connell 1970, Spight 1974, Spight and Emlen 1976). Therefore, we used a simple difference equation model to fit the observed changes in local abundance of whelks. The model was
log ([N.sub.t]/[N.sub.0]) = ([f.sub.1]M - [f.sub.2]P - [m.sub.N])t
where [N.sub.t] is the abundance of whelks at time t and [N.sub.0] the initial abundance of whelks in each experimental plot; [f.sub.1] is a per-whelk scaling function that expresses the rate of acquisition and transformation of mussel prey into new whelk individuals; [f.sub.2] is the seastar per capita consumption of whelks; [m.sub.N] is the natural mortality and emigration rate of whelks; and M and P are the mussel and seastar abundances, respectively. An equation for the simultaneous changes in abundance of mussels allowed us to estimate the effect of seastars on mussel abundance and thus indirectly on whelks, Following Gaines and Lafferty's (1995, see also Gurney and Nisbet ) modeling approach to completely open populations, mussel changes were expressed as follows:
log([M.sub.t]/[M.sub.0]) = ([lambda] - [f.sub.3]N - [f.sub.4]P - [m.sub.M])t
where [M.sub.t] is the mussel abundance at time t and [M.sub.0] that at the beginning of the experiment; [f.sub.3] and [f.sub.4] are the per capita consumption rates of mussels by whelks and seastars, respectively (equivalent to the alpha terms in Navarrete and Menge 1996), [m.sub.M] is the mussel mortality rate in the absence of predators, and [lambda] is the mussel recruitment rate. Since experiments were started in spring, several months after the fall recruitment of mussels, [lambda] was assumed to be negligible during the course of the experiments. Because estimates of predation effects on mussels were done by comparing +Mytilus plots with and without seastars, relaxing the assumption about lack of mussel recruitment would not significantly alter our results, as long as recruitment was not biased by the removal of seastars from the -Pisaster plots. As a first approximation, we considered that at the beginning of the experiment Pisaste r densities would remain relatively unchanged by application of the Mytilus treatment. This assumption proved to be partially wrong because of fast aggregative and emigration responses of seastars to Mytilus manipulations.
Using only data from the first 35 d of the experiment to minimize potential indirect effects due to species not considered in the experimental manipulations (Billick and Case 1994, Wootton 1994), yet allowing enough time to measure exploitation competition on the target whelk species, we regressed the natural logarithm of whelk abundance data ([N.sub.t]/[N.sub.0]) over time in days. Estimates of the slopes of this relationship under each treatment combination were obtained using ordinary least squares. These slopes will be estimates of the following components of the dynamic equation for whelks under the different treatment combinations: (a) the slope in treatment -Mytilus-Pisaster will be an estimate of m, (b) the slope in treatment -Mytilus +Pisaster will estimate [f.sub.2]P - m, (c) the slope in treatment +Mytilus-Pisaster will estimate [f.sub.1]M - m; and (d) the slope in + Mytilus + Pisaster will estimate [f.sub.1]M - [f.sub.2]P - m. Subtracting the natural mortality term, it was possible to estimate th e per-population effect size of seastar predation, [f.sub.2]P, and the direct bottom-up effect of prey, [f.sub.1]M. Confidence intervals (95%) around the slope estimates were calculated from the least squares regression. Confidence intervals for the resulting parameter estimates were calculated using error propagation rules for random independent errors (Taylor 1982).
Similarly, to obtain an estimate of the, effect of Pisaster on mussel abundance ([f.sub.4]P) during the same first 35 d of the experiment, we regressed the log-transformed proportional abundance of mussels ([M.sub.t]/[M.sub.0]) in the +Mytilus plots against time. The least squares slope of this relationship in the +Pisaster treatment provided an estimate of the [f.sub.3]N - [f.sub.4]P - [m.sub.M] term in the equation for mussel changes (assuming negligible mussel recruitment during the period). The slope under the -Pisaster treatment estimated [f.sub.3]N - [m.sub.M]. Since we have previously shown that predation by Pisaster on mussels is unaffected by the presence of whelks (Navarrete and Menge 1996), subtraction of ([f.sub.3]N - [m.sub.M]) from ([f.sub.3]N - [f.sub.4]P - [m.sub.M]) provided an estimate of [f.sub.4]P, which multiplied by the effect of mussels on whelks ([f.sub.1]M) provided an estimate of the indirect effect of exploitative competition of Pisaster on whelks (see Fig. 1). It is important to n ote that all our estimates are dependent on the existing levels of predators and prey and that per capita. coefficients of interaction strength (see Laska and Wootton 1998, Berlow et al. 1999) cannot be estimated because densities of prey and predators are expected to change during the course of the experiment.
Premanipulation patterns of community structure in the plots did not vary with treatment. Mussels (Mytilus trossulus) initially covered 40-60% of the substratum and were by far the most abundant sessile species, with average covers of the barnacles Balanus glandula and Chthamalus dalli [sim]15% and 10%, respectively (Fig. 2). Covers of the larger barnacle Semibalanus cariosus, the lepidomorph Pollicipes polymerus, and anemones Anthopleura spp. all averaged [less than]5%. Similarly, no significant premanipulation differences in whelk densities were detected (Fig. 3, "Before manipulations"; Table 1), nor there were differences in densities of seastars between plots assigned to +Pisaster and -Pisaster treatments (Fig. 4A, F 1.02, P = 0.3507, df = 1, 6). Seastar densities did not differ among treatments before manipulations. Hence, experimental results were unlikely to be affected by any premanipulation differences in abundances of sessile invertebrates or the whelks.
As in most such experiments, seastar "removal" did not result in the complete absence of Pisaster. Seastars commonly reinvaded the removal plots in a matter of days following the manual removals, particularly in spring and summer when seastars are most active (see also Menge et al. ). Despite these reinvasions and considering that seastar densities were reduced to zero after each monitoring, the. overall density of seastars in the +Pisaster treatment (2.12 [+ or -] 0.24 individuals/[m.sup.2] [mean [+ or -] 1 SE]) was [greater than]2.5X the density in the -Pisaster treatment (0.84 [+ or -] 0.31 individuals/[m.sup.2], Fig. 4). However, between September 1994 and February 1995, densities of seastars in the unmanipulated +Pisaster declined to levels similar to those in the seastar removal plots (Fig. 4), rendering the Pisaster treatment ineffective during that period.
Mussel removal was highly effective through September 1994, and only slightly less effective from then until the end of the experiment (Fig. 2). The modest increase in mussel cover in -Mytilus treatments was due to recruitment and growth, and our efforts to avoid removing barnacles while maintaining mussel removals. The effectiveness of the mussel treatment varied greatly over time, however. Mussel cover in +Mytilus plots decreased to the low levels maintained in the -Mytilus plots between late summer 1994 and early winter 1994-1995 (Fig. 2). During this period, seastar reinvasions evaded our control, resulting in near-elimination of mussels in these treatments for [sim]3 mo (Fig. 2). We continued to maintain the experiments, however, and by February 1995, mussel cover had largely rebounded, through recruitment and growth, to the levels seen in summer 1994 (Fig. 2).
Predation and prey effects
The manipulations led to strong responses by the whelks by spring 1994 (Fig. 3). Overall, whelk abundance increased in spring 1994, stayed high through the summer, decreased in winter 1994-1995, and increased again in spring 1995 (Fig. 3), following the typical seasonal changes reported previously for the Oregon coast (Berlow 1994, 1997, Navarrete 1996, Navarrete and Menge 1996). Changes in whelk abundance differed markedly with treatment and, to a lesser extent, with species of Nucella. In spring-summer 1994, densities of both Nucella species increased [greater than]3X in the presence of mussels (+Mytilus) while they remained relatively unchanged when mussel beds were removed (Fig. 3). The proportional change was slightly lower and more variable for N. emarginata.
In winter 1994-1995, overall densities of both whelk species decreased, but for N. canaliculata, the differences between the +Mytilus and -Mytilus treatments were still evident and proportionally the same as those observed in spring-summer 1994 (Fig. 3). In contrast, differences between mussel treatments tended to disappear for N. emarginata toward the beginning of winter. In spring-summer 1995, densities of N. canaliculata again increased in the +Mytilus treatment, reaching the levels observed the previous year. Overall densities of N. emarginata also increased in spring-summer 1995, but they did not reach the values observed the previous spring-summer, nor were there the same proportional differences between +Mytilus and -Mytilus treatments (Fig. 3). No effect of Pisaster deletion was detected throughout the experiment for either Nucella species.
Statistical analyses indicated that high densities of N. canaliculata depended on Mytilus but not on Pisaster (Table 2, between-subjects main effects; Mytilus was significant, Pisaster was not). The effect of Mytilus did not vary with either Pisaster density (Table 2, no significant effect of Pisaster X Mytilus interaction) or over time (Table 2, no significant Mytilus X Time interaction). In partial contrast, densities of N. emarginata evidently varied independently of Pisaster density, and depended on Mytilus density but only at certain times (Table 2, significant within-subjects effect of Mytilus X Time). Temporal variability in the response of. N. emarginata to the treatments was so great (Fig. 3) that, despite trends similar to those for N. canaliculata, the overall between-subjects effect of Mytilus on density of this whelk was not significant (Table 2).
Similar results were obtained when the overall temporal means of both Nucella species were considered as dependent variables and compared using MANOVA (Table 3). Presence of Mytilus but not of Pisaster affected whelk abundance. Analysis of standardized canonical coefficients (see Schemer 1993) suggests that the response was mostly due to N. canaliculata.
In summary, the Mytilus-Nucella link is a strong effect, the direct Pisaster-Nucella link is a very weak effect, and N. canaliculata was more affected by mussel manipulations than N. emarginata. Nevertheless, long-term treatment effects could have been affected, to some extent, by our failure to maintain complete Pisaster removals throughout the experiment and by the natural decline in seastar abundance after the first 4 mo of the experiment (see Fig. 4).
Calculation of rates of change in whelk and mussel abundances during the first 35 d of the experiment, when seastar densities were high, were used to further evaluate prey and predation effects on whelks. These analyses also provide a way to estimate the magnitude of predation by, and competition with, Pisaster as well as food limitation by Mytilus and in general lend support to the long-term results. Log-transformed whelk abundance data over the initial stages of the experiment followed reasonably linear trends, indicating that slope estimates are good descriptors of the changes in whelk abundance (Fig. 5, Table 4). Based on these slopes, the size of total effects per whelk indicate that prey (resource) limitation was the main factor driving changes in local abundance in both Nucella species (Fig. 6). The effect of direct predation by seastars was not significantly different from zero. The effect of competition with Pisaster, obtained by multiplying the negative effect of seastars on mussels, [f.sub.4]P = - 0.0346 (95% confidence interval 0.0039) by the positive effect that mussels had on whelks ([f.sub.1]M, Table 4), was small but significant (nonzero) on both whelk species. The direct positive effect of mussels on Nucella was slightly different for each species. For Nucella emarginata, the positive bottom-up effect of mussel prey was slightly smaller than for N. canaliculata (Fig. 6).
Although our mussel manipulations were performed to observe the response of whelks to prey availability, seastars in the +Pisaster plots also responded to mussel removal (Fig. 4, Table 5). In spring 1994, seastar density increased in +Mytilus plots (Fig. 4), following the spring-summer increase in activity and abundance typical for the Oregon coast (Menge et al. 1994, Navarrete and Menge 1996). Seastar density in these plots remained high until late summer 1995, when it sharply decreased to winter levels. In contrast, seastars tended to abandon --mussel plots, and never reached the densities observed in +mussel plots (Fig. 4). By late winter, after seastars had eliminated mussels in the +Mytilus plots, seastar densities in +mussel and --mussel plots became indistinguishable. In spring 1995, seastar density began to increase again, but no clear differences between +mussel and --mussel plots was apparent when the experiment was terminated (Fig. 4). Thus, the initial removal (spring and summer 1994) of mussels from the --Mytilus plots prevented the typical spring increase in seastar abundance, probably because Pisaster concentrated on +mussel beds.
Of the alternative interaction web models (Fig. 1), this experiment supports an indirect-competition interpretation of the interactions between seastars and whelks at Strawberry Hill. Removal of mussels suppressed the spring increase in whelk abundance (Fig. 3). Further, when mussel abundances declined due to natural causes, either from reductions by seastars in +Pisaster plots or as a response to direct predation by whelks, whelk densities dropped to near zero. No significant direct response to Pisaster density was observed, suggesting that consumption of Nucella by seastars does not have an effect on the local abundance of whelks (Fig. 6). Our previous demonstration of strong negative effects of Pisaster on Nucella abundance and size (Menge et al. 1994, Navarrete and Menge 1996) was evidently caused by indirect effects mediated by mussels, and not by intraguild predation.
There are at least two important consequences of these results for our view of marine intertidal communities. First, our results reinforce the idea that "food chain" omnivory (consumers feeding on more than one trophic level, like seastars feeding on mussels and whelks; Fig. 1b) is the most important trophic interaction in this intertidal community and one that dominates its dynamics. Our work, with that of Paine (1969) suggests that three potentially important food chains in this community all involve weak predation (Pisaster [right arrow] Nucella [right arrow] Mytilus; Pisaster [right arrow] Nucella [right arrow] Balanus; Pisaster [right arrow] Tegula [right arrow]macroalgae). Paine (1969), for example, showed that the trophic interaction between Pisaster ochraceus and the herbivorous gastropod Tegula funebralis did not seem to have important consequences for the rest of the community. In our study, the direct trophic interaction between Pisaster and the principal predatory species in their diet, Nucella s pp., also appears of minor consequence for the rest of the community. Thus, although Pisaster is the top predator of three important intertidal food chains, it preys on species in three trophic groups (whelks, herbivorous gastropods, sessile animals) and two trophic levels (primary carnivores, herbivores) and is therefore an omnivore. The direct links to herbivores and to primary carnivores are weak, however, so the direct omnivorous link to sessile animals is the central dynamic of this web. The indirect effect of Pisaster on Nucella mediated through mussels, however, does have important effects on populations of Nucella and on the local structure of low-intertidal communities (Menge et al. 1994, Navarrete and Menge 1996). Others have argued that the existence and prevalence of omnivory (sensu Pimm and Lawton 1977) in marine communities may effectively prevent the manifestation of trophic cascades under natural conditions (see Strong 1992, Polis and Strong 1996). In our system, while predation by top predato rs on both intermediate predators and herbivores does occur, the intensity of these links is too weak to modify the overall direct negative effect of top predators on basal prey species or to significantly affect population abundance at intermediate trophic levels.
Diehl (1993) has shown that in most food webs with omnivorous links experimentally examined, the direct negative effect of the top predator in the system offsets any indirect positive effects mediated by the intermediate predator. He argues that this is at least partly a consequence of the relative body sizes of the top and intermediate predators with respect to the shared prey resource. The pattern of body sizes in our rocky intertidal subweb lends support to Diehl's argument. A much larger seastar top predator feeds on similarly sized whelks and mussels, and the release from predation pressure accrued by mussels when seastars include whelks in their diet does not offset its direct negative effect. We suggest that the empirical pattern of body sizes found in most food webs, including the intertidal zone, may often introduce a bias when comparing the impact of prey resource between top and intermediate predators. Besides providing food/energy for both predators, the basal resource may also provide important microhabitat for the intermediate predator. Our evaluation of "exploitative competition" between predators may thus be influenced by intrinsic differences in the way seastars and whelks perceive mussel beds. Experimental separation of these types of prey effects may be very challenging in most systems.
Secondly, these results offer further insights into discussions about the relative importance of the ecological processes that regulate trophic levels in natural communities (e.g., Hairston et al. 1960, Menge and Sutherland 1976, 1987, Oksanen et al. 1981, Menge and Olson 1990, Strong 1992). While many studies have quantified the top-down effect of predators on prey populations (see reviews by Sih et al. 1985, Kerfoot and Sih 1987, Menge and Farrell 1989, Holt and Lawton 1994, Menge 1995), there are few examples of studies in which both top-down and bottom-up factors (e.g., prey effects) have been simultaneously quantified at intermediate and higher trophic levels (Diehl 1993; but see Power 1992, Wootton and Power 1993, Osenberg and Mittelbach 1995). Most studies that experimentally quantified bottom-up factors have concentrated on the influence of nutrients on algal or plant growth. Experimental quantification of the effect of prey on herbivore or carnivore predators is generally more rare (Sih et al. 1985, Kerfoot and Sih 1987) and almost nonexistent in the marine realm (but see Fairweather [19881, and the elegant study of Robles et al. ). The scarcity of studies quantifying prey effects on predators is probably due to the logistical difficulties of incorporating the usually larger temporal and spatial scales over which prey affect predator population dynamics. For instance, while predation can be viewed as a quasi-instantaneous reduction in per capita growth rate of the prey, gains from predation by the predator population need to be assimilated and then converted into new predators through reproduction.
At least two approaches have been used to get around this problem to enable comparison of prey and predation effects on intermediate trophic levels. First, population changes induced by predators and changes in individual body mass induced by the prey can be transformed into biomass as a common currency (e.g., Osenberg and Mittelbach 1995). The assumption made here is that individual body mass of predators is correlated positively and monotonically with fitness (Osenberg and Mittelbach 1995). A second approach is to simply quantify the short-term aggregative responses induced by prey manipulations and assume that population level responses would follow a similar trend (e.g., Robles et al. 1995, Diehl 1995).
This latter approach, in combination with observations of longer-term ([greater than]14 mo) population responses, is the one followed here. Overall there was good agreement between the long term results and the analysis of rates of change at the beginning of the experiment, and this led us to the conclusion that the intermediate predators of the genus Nucella are primarily limited by bottom-up (prey) factors (Fig. 6). The two Nucella species responded with slightly different intensities to mussel manipulation. Differences in whelk responses are not likely explained by differences in food preferences. Both species show remarkably similar feeding preferences when subjected to a choice of mussel and barnacle species (Palmer 1984, Wieters and Navarrete 1998). Yet, there might be differences in terms of their ability to switch to less preferred prey species as the abundance of the preferred prey is reduced ("predator switching", Murdoch and Oaten 1975). If Nucella emarginata is able to more rapidly switch to the barnacle Balanus glandula after the removal of Mytilus (see the increase in B. glandula in - Mytilus treatments on the second summer, Fig. 2), it could maintain higher densities than N. canaliculata in the face of a drastic increase of the most abundant prey. We have, however, no data on diet of Nucella species through the experiment. Another possibility is related to the way both species use mussel beds. While both whelks are found in patches in mussel beds and exhibit similar patterns of spatial and temporal variation (Navarrete 1996), field observations suggest that individuals of N. canaliculata are more often found within and underneath the Mytilus trossulus mat than are N. emarginata, which more often occur within open gaps. The apparently differential use of alternative mussel-associated microhabitats could also cause differences in whelk responses to the mussel bed removals and could explain the slightly larger mussel effect on N. canaliculata as compared to N. emarginata (Fig. 6). Further studies des igned to quantify the relative effect of mussels as food and microhabitat for whelk species are warranted.
Meaningful quantification of the strength of interactions among species is one of the major goals of community ecologists, yet agreement on how this should be done is elusive (Laska and Wootton 1998, Berlow et al. 1999). Here, we used a simple model that includes terms for changes in whelk and mussel abundances over relatively short periods of time to quantify the total population strength of direct and indirect links in the subweb. Because in food webs with omnivory the abundances of all species are expected to be coupled, our absolute estimates of interaction strength are dependent on the existing levels of prey and predators in the system. However, the relative sizes of the different factors (prey effect, predation, and competition) should not be greatly affected. The observed rapid changes in seastar abundance following the removal of mussels could lead to an underestimation of the total (population) predation effects on whelks. However, comparisons of predation effects in the presence of mussels (bottom two curves in Fig. 5) suggest that this was not the case. The effect of seastars in presence of mussels, when they have both a direct negative effect due to predation and an indirect negative effect due to competition, was rather small and can be accounted for by competition alone.
Using a series of simple models, Holt and Polis (1997; see also Polis and Holt 1992) formulated a basic theoretical framework for the study of intraguild predation. Assuming that prey and predator species have population dynamics dominated by constraints on local reproduction of adults (i.e., are "closed"), they reached the general conclusion that for an IG predator to coexist with an IG prey, the latter must be a superior competitor at exploiting the basal resource. In the present case the IG predator, not the IG prey, was a superior competitor. However, these models maybe inappropriate for understanding the conditions for coexistence among Pisaster, Nucella spp., and Mytilus. Nucella spp. have closed populations due to their deposition of benthic egg cases that produce crawl-away juveniles, and therefore their populations should respond directly to local conditions (Spight 1974, Spight and Emlen 1976). In contrast, both Pisaster and Mytilus have free-swimming, pelagic, widely dispersing larval stages that spend several weeks in the water column, and their populations are thus "open" (Gaines and Lafferty 1995). This means, for instance, that the population growth rate of Pisaster is not strictly dependent on the local gains accrued by exploiting either Nucella or Mytilus, but depends on settlement rate and available space (Gaines and Lafferty 1995). Thus, Holt and Polis' (1997) criterion for coexistence--at resource equilibrium ([R.sup.*]), per capita resource exploitation of IG predator (fP([R.sup.*]) [less than] per capita resource exploitation of IG prey (fN([R.sup.*])--may not strictly hold. The large empirical differences in relative body sizes of IG predators and IG prey, typical of many food webs, are another reason why theoretical predictions may not hold in real systems (Diehl 1993). Much more theoretical work is needed to incorporate the life history features of most marine organisms (complex life cycles, Roughgarden et al. ) into models and predictions of food webs with intraguild predation. We are just starting to unveil its consequences in natural communities.
We thank Gary Allison, Eric Berlow, Jennifer Burnaford, Elizabeth Dahlhoff, Zahna Neskoui, Laura Ryan, Eric Sanford, and Evie Wieters for assistance in the field. We are grateful for the support of grants from NSF (OCE92-17459; to B. Menge, P. Wheeler, and T. Strub), the Andrew W. Mellon Foundation (to B. Menge and J. Lubchenco), and of a John Simon Guggenheim Memorial Foundation Fellowship (to B. Menge). SAN acknowledges support from Fundacion AndesChile (Programa Reinsercion) and FONDAP OYBM # 3 while writing this manuscript. This is contribution number 4 from PISCO, the Partnership for Interdisciplinary Studies of Coastal Oceans: A Long-Term Ecological Consortium funded by the David and Lucile Packard Foundation.
(1.) Departamento de Ecologfa and Esracion Costera de Investigaciones Marinas, Pontificia Universidad Catolica de Chile, Casilla 114-D, Santiago, Chile
(2.) Department of Zoology, Oregon State University, Corvallis, Oregon 97331-2914 USA
(3.) E-mail: email@example.com
(4.) Present address: Section of Ecology and Systematics, Corson Hall, Cornell University, Ithaca, New York 14853 USA.
Barkai, A., and C. McQuaid. 1988. Predator-prey role reversal in a marine benthic ecosystem. Science 242:62-64.
Bender, E. A., T. J. Case, and M. E. Gilpin. 1984. Perturbation experiments in community ecology: theory and practice. Ecology 65:1-13.
Berlow, E. L. 1994. Temporal and spatial variation in indirect interactions and nonlinear effects of predators. Bulletin Ecological Society of America 75:15.
Berlow, E. L. 1997. From canalization to contingency: historical effects in a successional rocky intertidal community. Ecological Monographs 67:435-460.
Berlow, E. L., S. A. Navarrete, C. J. Briggs, B. A. Menge, and M. E. Powers. 1999. Quantifying variation in strengths of species interactions. Ecology, 80:2206-2224.
Billick, I., and T. J. Case. 1994. Higher order interactions in ecological communities: what are they and how can they be detected? Ecology 75:1529-1543.
Bradley, R. A. 1983. Complex food webs and manipulative experiments in ecology. Oikos 41:150-152.
Bronmark, C., S. P. Klosiewski, and R. A. Stein. 1992. Indirect effects of predation in a freshwater, benthic food chain. Ecology 73:1662-1674.
Carpenter, S. R., J. F. Kitchell, and J. R. Hodgson. 1985. Cascading trophic interactions and lake productivity. BioScience 35:634-639.
Connell, J. H. 1970. A predator-prey system in the marine intertidal region. I. Balanus glandulo and several predatory species of Thais. Ecological Monographs 40:49-78.
Crowder, M. J., and D. J. Hand. 1990. Analysis of repeated measures. Chapman and Hall, London, UK.
Dayton, P. K. 1971. Competition, disturbance, and community organization: the provision and subsequent utilization of space in a rocky intertidal community. Ecological Monographs 41:351-389.
Diehl, S. 1992. Fish predation and benthic community structure: the role of omnivory and habitat complexity. Ecology 73:1646-1661.
Diehl, S. 1993. Relative consumer sizes and the strengths of direct and indirect interactions in omnivorous feeding relationships. Oikos 68:151-157.
Diehl, S. 1995. Direct and indirect effects of omnivory in a littoral lake community. Ecology 76:1727-1740.
Fairweather, P. G. 1988. Consequences of supply-side ecology: manipulating the recruitment of intertidal barnacles affects the intensity of predation upon them. Biological Bulletin 175:349-354.
Fairweather, P. G., and A. J. Underwood. 1983. The apparent diet of predators and biases due to different handling times. Oecologia 56:169-179.
Gaines, S. D., and K. D. Lafferty. 1995. Modeling the dynamics of marine species: The importance of incorporating larval dispersal. Pages 389-412 in L. McEdward, editor. Ecology of marine invertebrate larvae. CRC Press, New York, New York, USA.
Garrity, S. D., and S. C. Levings. 1981. A predator-prey interaction between two physically and biologically constrained tropical rocky shores gastropods: direct, indirect and community effects. Ecological Monographs 51:267-286.
Gurney, w. S. C., and R. M. Nisbet. 1998. Ecological dynamics. Oxford University Press, New York, New York, USA.
Hairston, N. G., F. E. Smith, and L. B. Slobodkin. 1960. Community structure, population control, and competition. American Naturalist 94:421-425.
Holt, R. D., and J. H. Lawton. 1994. The ecological consequences of shared natural enemies. Annual Review of Ecology and Systematics 25:495-520.
Holt, R. D., and G. A. Polis. 1997. A theoretical framework for intraguild predation. American Naturalist 149:745-764.
Huberty, C. J., and J. D. Morris. 1989. Multivariate analysis vs. multiple univariate analyses. Psychological Bulletin 105:302-308.
Hunter, M. D., and P. W. Price. 1992. Playing chutes and ladders: heterogeneity and the relative roles of bottom-up and top-down forces in natural communities. Ecology 73:724-732.
Kerfoot, W. C., and A. Sib, editors. 1987. Predation. Direct and indirect impacts on aquatic communities. University Press of New England, Hanover, New Hampshire, USA.
Landenberger, D. E. 1968. Studies on selective feeding in the Pacific starfish Pisaster in southern California. Ecology 49:1062-1075.
Laska, M. S., and T. J. Wootton. 1998. Theoretical concepts and empirical approaches to measuring interaction strength. Ecology 79:461-476.
Menge, B. A. 1972. Foraging strategy of a starfish in relation to actual prey availability and environmental predictability. Ecological Monographs 42:25-50.
Menge, B. A. 1983. Components of predation intensity in the low zone of the New England rocky intertidal region. Oecologia 58:141-155.
Menge, B. A. 1995. Indirect effects in marine rocky intertidal interaction webs: patterns and importance. Ecological Monographs 65:21-74.
Menge, B. A. 2000. Recruitment vs. post-recruitment processes as determinants of barnacle population abundance on Oregon rocky shores. Ecological Monograph, in press.
Menge, B. A., E. L. Berlow, C. Blanchette, S. A. Navarrete, and S. B. Yamada. 1994. The keystone species concept: variation in interaction strength in a rocky intertidal habitat. Ecological Monographs 64:249-286.
Menge, B. A., and T. M. Farrell. 1989. Community structure and interaction webs in shallow marine hard-bottom communities: tests of an environmental stress model. Advances in Ecological Research 19:189-262.
Menge, B. A., J. Lubchenco, L. R. Ashkenas, and F Ramsey. 1986. Experimental separation of the effects of consumers on sessile prey in the low zone of a rocky shore in the Bay of Panama: direct and indirect consequences of food web complexity. Journal of Experimental Marine Biology and Ecology 100:225-269.
Menge, B. A., and A. M. Olson. 1990. Role of scale and environmental factors in regulation of community structure. Trends in Ecology and Evolution 5:52-57.
Menge, B. A., and J. P. Sutherland. 1976. Species diversity gradients: synthesis of the roles of predation, competition, and temporal heterogeneity. American Naturalist 110:51-369.
Menge, B. A., and J. P. Sutherland. 1987. Community regulation: variation in disturbance, competition, and predation in relation to environmental stress and recruitment. American Naturalist 130:730-757.
Menge, J., and B. A. Menge. 1974. Role of resource allocation, aggression and spatial heterogeneity in coexistence of two competing intertidal starfish. Ecological Monographs 44:189-209.
Moran, M. D., T. P. Rooney, and L. E. Hurd. 1996. Top-down cascade from a bitrophic predator in an old-field community. Ecology 77:2219-2227.
Murdoch, W. W., and A. Oaten. 1975. Predation and population stability. Advances in Ecological Research 9:1-131.
Navarrete, S. A. 1996. Variable predation: Effects of whelks on a mid intertidal successional community. Ecological Monographs 66:301-322.
Navarrete, S. A., and B. A. Menge. 1996. Keystone predation: interactive effects of two predators on their main prey. Ecological Monographs 66:409-429.
Oksanen, L., S. D. Fretwell, J. Arruda, and P. Niemela. 1981. Exploitation ecosystems in gradients of primary productivity. American Naturalist 118:240-261.
Osenberg, C. W., and G. G. Mittelbach. 1995. The relative importance of resource limitation and predator limitation in food chains. Pages 134-148 in G. Polls and K. O. Winemiller, editors. Food webs: integration of patterns and dynamics. Chapman and Hall, New York, New York, USA.
Paine, R. T. 1969. The Pisaster-Tegula interaction: prey patches, predator food preference, and individual community structure. Ecology 50:950-961.
Paine, R. T. 1974. Intertidal community structure: experimental studies on the relationship between a dominant competitor and its principal predator. Oecologia 15:93-120.
Paine, R. T. 1980. Food webs, linkage interaction strength, and community infrastructure. Journal of Animal Ecology 49:667-685.
Palmer, A. R. 1984. Prey selection by thaidid gastropods: some observational and experimental field test of foraging models. Oecologia 62:162-172.
Pimm, S. L., and J. H. Lawton. 1977. Number of trophic levels in ecological communities. Nature 268:329-331.
Polis, G. A. 1991. Complex trophic interactions in deserts: an empirical critique of food-web theory. American Naturalist 138:123-155.
Polis, G. A. 1994. Food webs, trophic cascades and community structure. Australian Journal of Ecology 19:121-136.
Polis, G. A., and R. D. Holt. 1992. Intraguild predation: the dynamics of complex trophic interactions. Trends in Ecology and Evolution 7:151-154.
Polis, G. A., C. A. Myers, and R. Holt. 1989. The evolution and ecology of intraguild predation: competitors that eat each other. Annual Review of Ecology and Systematics 20:297-330.
Polis, G. A., and D. R. Strong. 1996. Food web complexity and community dynamics. American Naturalist 147:813-846.
Power, M. B. 1992. Top-down and bottom-up forces in food webs: do plants have primacy? Ecology 73:733-746.
Robles, C., and J. Rohb. 1993. Varied carnivore effects and the prevalence of intertidal algal turfs. Journal of Experimental Marine Biology and Ecology 166:65-91.
Robles, C., R. Sherwood-Stephens, and M. Alvarado. 1995. Responses of a key intertidal predator to varying recruitment of its prey. Ecology 76:565-579.
Roughgarden, J., S. D. Gaines, and H. Possingham. 1988. Recruitment dynamics in complex life cycles. Science 241:1460-1466.
Sanford, E. 1999. Regulation of keystone predation by small changes in ocean temperature. Science 283:2095-2097.
Scheinmer, S. M. 1993. MANOVA: multiple response variables and multispecies interactions. Pages 94-112 in S. M. Scheiner and J. Gurevitch, editors. Design and analysis of ecological experiments. Chapman and Hall, London, UK.
Sih, A., P. Crowley, M. McPeek, J. Petranka, and K. Strobmeier. 1985. Predation, competition, and prey communities: a review of field experiments. Annual Review of Ecology and Systematics 16:269-311.
Spight, T. M. 1974. Sizes of populations of a marine snail. Ecology 55:712-729.
Spight, T. M., and J. M. Emlen. 1976. Clutch sizes of two marine snails with a changing food supply. Ecology 57:1162-1178.
Spiller, D. A., and T. W. Schoener. 1990. A terrestrial field experiment showing the impact of eliminating top predators on foliage damage. Nature 347:469-472.
Strong, D. R. 1992. Are trophic cascades all wet? differentiation and donor-control in speciose ecosystems. Ecology 73:747-754.
Taylor, J. R. 1982. An introduction to error analysis.the study of uncertainties in physical measurements. University Science Books, Mill Valley, California, USA.
ter Braak, C. J. E 1986. Canonical correspondence analysis: a new eigenvector technique for multivariate direct gradient analysis. Ecology 67:1167-1 179.
Underwood, A. J. 1997. Experiments in ecology. Cambridge University Press, Melbourne, Australia.
von Ende, C. N. 1993. Repeated-measures analysis: growth and other time-dependent measures. Pages 113-137 in S. M. Schemer and J. Gurevitch, editors. Design and analysis of ecological experiments. Chapman and Hall, London, UK.
Wieters, E. A., and S. A. Navarrete. 1998. Spatial variation in feeding preferences of the rocky intertidal whelks Nucella canaliculata and N. emarginata. Journal of Experimental Marine Biology and Ecology, in press.
Wilbur, H. M., and J. E. Fauth. 1990. Experimental aquatic food webs: interactions between two predators and two prey. American Naturalist 135:176-204.
Wootton, J. T. 1994. Putting the pieces together: testing the independence of interactions among organisms. Ecology 75:1544-1551.
Wootton, J. T., M. S. Parker, and M. E. Power. 1996. Effects of disturbance on river food webs. Trends in Ecology and Evolution 273:1558-1561.
Wootton, J. T., and M. E. Power. 1993. Productivity, consumers, and the structure of a river food chain. Proceedings of the National Academy of Sciences USA 90:1384-1387.
Two-way ANOVA comparing Nucella densities among experimental plots before density manipulations on a rocky intertidal shore of the central Oregon coast. (A) Separate ANOVAs for each Nucella species. (B) MANOVA considering both Nucella species as dependent variables. A) ANOVA Nucella canaliculata Source df MS F P Pisaster 1 0.440 2.08 0.1747 Mytilus 1 0.010 0.05 0.8284 Pisaster X Mytilus 1 0.109 0.51 0.4870 Residual error 12 A) ANOVA Nucella emarginata Source MS F P Pisaster 0.055 0.12 0.7345 Mytilus 0.139 0.31 0.5903 Pisaster X Mytilus 0.001 0.01 0.9615 Residual error B) MANOVA Numerator df Denominator df Wilks' [lambda] P Pisaster 2 11 0.7506 0.2065 Mytilus 2 11 0.9441 0.7290 Pisaster X Mytilus 2 11 0.9372 0.7001 Repeated-measures ANOVA comparing the responses of Nucella species to predators (Pisaster) and prey (Mytilus) treatments. Nucella canaliculata Source of variation df MS F P Between subjects Pisaster 1 0.949 3.43 0.0886 Mytilus 1 14.37 51.97 0.0001 Pisaster x Mytilus 1 0.17 0.65 0.4373 Error 12 0.27 Within subjects Time 2 2.800 24.59 0.0001 Time x Pisaster 2 0.086 0.75 0.4813 Time x Mytilus 2 0.158 1.39 0.2687 Time x Pisaster x Mytilus 2 0.059 0.52 0.5996 Error (Time) 24 0.114 Nucella emarginata Source of variation MS F P Between subjects Pisaster 0.010 0.01 0.9345 Mytilus 2.610 1.67 0.2202 Pisaster x Mytilus 0.007 0.01 0.9495 Error 1.56 Within subjects Time 3.810 44.59 0.0001 Time x Pisaster 0.095 1.11 0.3467 Time x Mytilus 0.337 3.95 0.0328 Time x Pisaster x Mytilus 0.017 0.20 0.8229 Error (Time) 0.085 Notes: The sphericity test showed no significant deviations from the assumption of circularity of the treatment differences matrix for both Nucella species (see von Ende  for an explanation of the test and assumption; chi-square = 1.12, P = 0.5702 for N. canaliculata, and chi-square = 0.678, P = 0.7123 for N. emarginata). Thus, the uncorrected univariate probabilities are presented here. Multivariate analysis of variance to compare the effects of the Pisaster (predator) and Mytilus (prey) treatments on both Nucella species simultaneously (dependent variables). Source of Numerator Denominator variation df df Wilks' [lambda] P Pisaster 2 11 0.846 0.3985 Mytilus 2 11 0.227 0.0003 Pisaster X Mytilus 2 11 0.982 0.9054 Standardized canonical coefficients Source of N. N. variation canaliculata emarginata Pisaster 2.257 -0.536 Mytilus 2.286 -0.587 Pisaster X Mytilus 2.295 -0.527 Notes: Standardized canonical coefficients are a measure of the relative importance of each whelk species to the significance of that effect. Results of least-squares linear regressions analysis for (A) log-transformed whelk abundance ([N.sub.t]/[N.sub.0]) and (B) mussel abundance ([M.sub.t]/ [M.sub.0] over time in days. A. Nucella canaliculata Treatment Slope (CI) [r.sup.2] P - Pisaster - Mytilus -0.0194 0.91 0.0009 [- m] (0.0048) - Pisaster + Mytilus +0.0219 0.85 0.0012 [[f.sub.1]M - m] (0.0073) + Pisaster - Mytilus -0.0204 0.72 0.0157 [[f.sub.2]P - m] (0.0134) + Pisaster + Mytilus +0.0197 0.49 0.0576 [[f.sub.1]M - [f.sub.2]P - m] (0.0075) A. Nucella emarginata Treatment Slope (CI) [r.sup.2] P - Pisaster - Mytilus -0.0076 0.68 0.0226 [- m] (0.0046) - Pisaster + Mytilus +0.0244 0.61 0.0224 [[f.sub.1]M - m] (0.0195) + Pisaster - Mytilus -0.0003 0.09 0.9394 [[f.sub.2]P - m] (0.0087) + Pisaster + Mytilus +0.0149 0.49 [[f.sub.1]M - [f.sub.2]P - m] (0.0089) 0.0540 B. Mytilus trossulus Treatment Slope [r.sup.3] P -Pisaster +Mytilus -0.0033 0.85 0.0255 [[f.sub.3]N - m] (0.0007) +Pisaster +Mytilus -0.0313 0.91 0.0008 [[f.sub.3]N - [f.sub.4]P - m] (0.0039) Notes: Estimated terms of dynamic equations are indicated in' square brackets under each treatment. A model without intercept was used to force the regression through the origin. Numbers in parentheses show the 95% confidence interval centered around each slope estimate. Repeated-measures ANOVA comparing the density of seastars Pisaster ochraceus in plots with and without beds of Mytilus trossulus. Source of variation df MS F P Between subjects Mytilus 1 0.461 6.23 0.0468 Error 6 0.074 Within subjects Time 2 0.206 6.61 0.0116 Time X Mytilus 2 0.016 0.50 0.6164 Error(Time) 12 0.031 Notes: The sphericity test showed no significant deviations from the assumption of circularity of the treatment differences variance matrix for both Nucella species (see von Ende 1993; chi-square = 1.95, P = 0.3761). Thus, the uncorrected univariate probabilities are presented here.
|Printer friendly Cite/link Email Feedback|
|Author:||NAVARRETE, SERGIO A.; MENGE, BRUCE A.; DALEY, BRYON A.|
|Article Type:||Statistical Data Included|
|Date:||Aug 1, 2000|
|Previous Article:||RECRUITMENT FAILURE, LIFE HISTORIES, AND LONG-TERM DECLINE OF CARIBBEAN CORALS.|
|Next Article:||TRAIT-MEDIATED INDIRECT EFFECTS IN LARVAL ANURANS: REVERSING COMPETITION WITH THE THREAT OF PREDATION.|