changing recruitment in constant species assemblages: implications for predation theory in...

Changing Recruitment in Constant Species Assemblages: Implications for Predation Theory inIntertidal CommunitiesAuthor(s): Carlos D. RoblesSource: Ecology, Vol. 78, No. 5 (Jul., 1997), pp. 1400-1414Published by: Ecological Society of AmericaStable URL: http://www.jstor.org/stable/2266135 .

Accessed: 05/12/2014 15:18

Your use of the JSTOR archive indicates your acceptance of the Terms & Conditions of Use, available at .http://www.jstor.org/page/info/about/policies/terms.jsp

.JSTOR is a not-for-profit service that helps scholars, researchers, and students discover, use, and build upon a wide range ofcontent in a trusted digital archive. We use information technology and tools to increase productivity and facilitate new formsof scholarship. For more information about JSTOR, please contact [email protected].

.

Ecological Society of America is collaborating with JSTOR to digitize, preserve and extend access to Ecology.

http://www.jstor.org

This content downloaded from 128.235.251.160 on Fri, 5 Dec 2014 15:18:44 PMAll use subject to JSTOR Terms and Conditions

http://www.jstor.org/action/showPublisher?publisherCode=esa

http://www.jstor.org/stable/2266135?origin=JSTOR-pdf

http://www.jstor.org/page/info/about/policies/terms.jsp


Ecology, 78(5), 1997, pp. 1400-1414 C 1997 by the Ecological Society of America

CHANGING RECRUITMENT IN CONSTANT SPECIES ASSEMBLAGES: IMPLICATIONS FOR PREDATION THEORY IN INTERTIDAL

COMMUNITIES

CARLOS D. ROBLES

California State University, Los Angeles, California 90032 USA

Abstract. Ecological theory for benthic communities emphasizes intense species interactions that depend on the high productivity of sedentary invertebrates. The keystone predator hypothesis maintains that intense predation by one consumer species is necessary to prevent a prolific, competitively dominant prey species from eliminating other species using the same resource. This study considers the consequences of extreme spatial and temporal variation in the recruitment of a prey species supporting keystone and diffuse predation. Prior experiments on rocky shores of Santa Catalina Island, California, USA, demonstrated that predation by spiny lobsters (Panulirus interruptus) maintained a distinctive red algal turf by killing juvenile mussels (Mytilus californianus and M. galloprovincialis) that otherwise overgrow and replace the algae. In the present study, long-term surveys revealed that high recruitment of the predominant mussel, M. californianus, occurred only on the most wave-exposed sites in certain years; mussel recruitment was slight to nil on relatively protected sites in most years. A predator exclosure experiment consisting of seven replicates placed along the gradient of wave exposure demonstrated that the effects of predation depended upon the spatial differences in recruitment rates. Lobsters on wave-exposed sites functioned as keystone predators; on more sheltered sites, little or no predation, whether by lobsters or the fishes and whelks also foraging on the sheltered sites, was necessary to maintain the algal assemblage. Similar species assemblages can be maintained by markedly different relative levels of crucial ecological rates. In the mid-intertidal zone of Santa Catalina Island, the intense species interactions depicted in the keystone predator hypothesis occurred only at productive, high wave exposure locations; low recruitment of mussels elsewhere preempts both predation and the competition between the mussel and algal assemblages. Thus, red algae dominates rocky shores through different mechanisms over a range of physical conditions. The occurrences of low mussel recruitment do not appear to be anomalies, but rather a consequence of the life history of Mytilus californianus.

Key words: algal turfs; California; diffuse predation; growth; keystone predation; Mytilus; Pan- ulirus; productivity; recruitment; spiny lobsters.

INTRODUCTION

From its beginning, experimental ecology emphasized the effects of intense species interactions driven by the high reproduction of the participating species. Early experimental tests of ecological theory pitted competitors, or predators and their prey, against one another in conditions favoring maximum population growth rates. These were laboratory cultures initially providing wall to wall resources and constant physical conditions (e.g., Gause 1934, Crombie 1946, Huffaker 1958, Park 1962). The usual result, extinctions through intense species interactions, provided empirical support for such contemporary theories as the competitive exclusion principle (Gause 1934, Hardin 1960) and the theory of refugia (Gause 1934, Connell 1970, 1975).

The advent of experimental methods to field studies happened first in rocky intertidal communities, and the theoretical concerns echoed those of the earlier laboratory

Manuscript received 2 July 1996; revised 11 July 1996; ac- cepted 9 August 1996; final version received 19 September 1996.

experiments. With some exceptions (e.g., Hatton 1938, Frank 1965) most field experiments investigated effects of intense competition or predation on cool temperate shores with seasonally high production (reviews in Con- nell 1972, Paine 1994, Menge 1995). Much of the theory founded upon these experiments addresses mechanisms that could disrupt the process of competitive exclusion (discussion in Connell 1978, Paine 1994). The "keystone predator" (Paine 1966, 1974, Menge and Lubchenco 1981), "diffuse predation" (Menge and Lubchenco 1981, discussion in Robles and Robb 1993), "intermediate disturbance" (Connell 1978, Sousa 1979b, 1984), and "in- hibition" hypotheses (Connell and Slatyer 1977, Sousa 1979a) hold one dynamic in common: some agent, whether physical disturbance, a single keystone predator, or perhaps a diverse guild of predators, removes local con- centrations of competitively dominant species, allowing their subordinates to grow.

Without removals, high recruitment of sedentary species should precipitate competition for attachment space. In the contests among species of algae (Sousa

1400



July 1997 CHANGING RECRUITMENT RATES 1401

1979a, Lubchenco 1980, 1986, Schonbeck and Norton 1980, Kastendiek 1982), species of barnacles (Connell 1961a, b, 1970, Dayton 1971), or mussels, algae, and other invertebrates (Paine 1966, 1974, Paine and Levin 1981, Dungan et al. 1982, Underwood et al. 1983, Live- ly and Raimondi 1987, Menge 1992, Robles and Robb 1993), dominant competitors recruit in great numbers and either preempt the colonization of less prolific species or quickly overgrow and displace previously established, but less robust species. High rates of recruitment also figure in the ability of certain dominant species to counter predation, because the "swamping" (Dayton 1971) of the predators assures that significant numbers of the juveniles will survive long enough to grow to sizes that resist the subsequent attacks of predators (Connell 1970, Paine 1976, Robles et al. 1990). Thus, substantial productivity, occurring as recruitment and growth, can drive mechanisms of competition for space and of size-limited predation, and through the latter processes certain dominant species acquire and hold space in the rocky shore environment.

The early, formative experiments in rocky shore communities supported the view that important features of community structure play out in the counterpoint between competition and predation or disturbance (discussion in Connell 1975, 1978, 1983, Paine 1994). Ironically, continuing experimentation has called into question the feasibility of such generalization (discussion in Underwood and Denley 1984, Underwood and Fairweather 1988, Foster 1990, Paine 1991, Estes and Duggins 1995). The doubt was fostered, in part, by the realization that the same manipulations repeated at different times and places often produce markedly different results (e.g., Dayton 1971, Robles and Cubit 1981, Keough 1984b, Fairweather and Underwood 1991, Paine et al. 1985, Dethier and Duggins 1988, Fairweather and Underwood 1991, Robles and Robb 1993, Menge et al. 1994). Reconsideration of experimentation with the predators of the mussel Mytilus californianus suggests that for this example a resolution of apparently contradictory results can be found by considering the variation in rates of the prey's recruitment.

Paine's (1966, 1974) original demonstration of keystone predation in the MytiluslPisaster interaction has since been corroborated by experiments in other intertidal communities (Paine et al. 1985) as well as with different species in other ecosystems (discussion in Menge et al. 1994). But, attempts to reproduce the keystone effect with Pisaster or other predators of M. californianus sometimes failed; that is, later removals of demonstrated keystone predators produced no significant increases in mussel abundances (e.g., Robles and Robb 1993, Menge et al. 1994). In the case of the sea stars, this has been attributed to scarcity of mussel recruitment in unusual years, occurrences that were la- beled anomaly or "artifact" (Paine 1976). Menge et al. (1994) also propose that low recruitment may have

produced some artifact in Pisaster removals at Boiler Bay, Oregon, but they argue that natural mussel recruitment was appreciable and that the effects of several consumers were required to limit mussel abundances, a case of diffuse predation (Robles and Robb 1993). They propose that variation in prey recruitment affects the interaction by regulating whether keystone or diffuse predation occurs at a site.

I repeated removals of spiny lobsters (Panulirus interruptus) and other mussel predators simultaneously at the same shore level but different sites along a wave exposure gradient at Santa Catalina Island, California, USA. The shore level fell along the mid-line of a zone of perennial red algae, into which juvenile M. californianus recruited to varying degrees. I interpret differences among the sites in the outcome of the removals in light of spatial differences in mussel recruitment at the time of the experiment and other, long-term records of mussel recruitment. The result shows that whether predators are necessary to maintain the cover of algae depends on the recruitment of the mussels. Thus, keystone, diffuse or no predation were required to stop re- placement of algae. Moreover, the extreme variation in productivity of this prey species appears to follow naturally from its life history characteristics; low recruitment is neither anomalous nor dependent on extreme physical conditions for its occurrence.

Combining the natural variation of recruitment in the aggregate of theory for benthic communities, leads to a consistent interpretation of diverse experimental out- comes, as other authors suggest (e.g., Underwood et al. 1983, Keough 1984b, Underwood and Denley 1984, Gaines and Roughgarden 1985, Menge et al. 1994).

METHODS

Natural history

The mussels Mytilus californianus and M. galloprovincialis (= M. edulis, McDonald and Koehn 1988, McDonald et al. 1991) recruited to mid-shore levels of Santa Catalina and other Channel Islands in greatest numbers from winter to early spring (Robles 1987; Re- sults: Long-term recruitment surveys). The recruits nes- tled in a turf of coralline and fleshy red algae. If allowed to grow to lengths >1 cm, the juvenile mussels pro- truded from the turf. In most years, this began in spring, at which time they were usually discovered and con- sumed by whelks, Ceratostoma nuttalli and Maxwellia gemma; labrid fishes, Halichoeres semisinctus and Semicossyphus pulcher; and the northern spiny lobster, Panulirus interrupts. In earlier experiments (Robles and Robb 1993) the dominance of algae appeared to be maintained by the concerted effects of all consumers in a sheltered area, or by lobsters alone in more wave- exposed areas.

The algal turf consisted of articulated coralline algae (Corallina officianalis var. chilensis) and the delicate



1402 CARLOS D. ROBLES Ecology, Vol. 78, No. 5

N - BIRD ROCK

_

SANTA CATAUNA_

SAN DIEGO

0 300m

ISTHMUS REEF

BIG FISHERMAN'S COVE

FIG. 1. Location of study sites indicated by their numbers. Prevailing northwesterly swells and tidal obstruction currents first hit the shore at the west end of Bird Rock. Depth contours are given in meters above or below mean lower low water (MLLW).

branching thalli of Gigartina canaliculata, Laurencia pacifica, and other fleshy red algae. The turf formed a dense carpet 1-4 cm thick over rock surfaces from --0.1I to 1.5 m above MLLW (see Murray et al. 1980,

Seapy and Littler 1982, Stewart 1982 for a description of species composition and seasonal variation). This cover dominated mid-shore levels from sheltered back bays to all but the most wave exposed rocky points, where mussel beds extended down from +2.0 m MLLW, restricting the distribution of the turf to the range from "'0.1I to 0.75 m MLLW.

Study site description

Because community composition and dynamics may vary over tidal and wave exposure gradients, the placement the of the sites mattered. One site was placed on a horizontal platform in Big Fisherman Cove, and six sites were placed along a line on a horizontal platform of a nearby rocky islet, Bird Rock (Fig. 1; descriptions in Robles 1987, Robles and Robb 1993). Sites 1, 2, and 7 had been used for the "sheltered," "moderately exposed," and "wave- exposed" experiments, respectively, in Robles and Robb (1993). The experimental plots fell precisely along the horizontal contour of +0.63 m above MLLW, dividing the turf zone vertically roughly in half. Thus, the comparison among sites maintained shore-level, topography, and initial species composition, but varied wave exposure. This differs from other studies with duplicate surveys or experiments that confounded shore-level, topographic, and year

differences among sites (e.g., Fairweather and Un- derwood 1991, Robles and Robb 1993, discussion in Underwood and Petraitis 1993).

Survey methods

Wave exposure.-The study area was protected from wave action by the mainland to the north and Catalina itself, which intercepts much of the wave energy orig- inating in summer storms far to the south (Fig. 1). Consequently, the largest input of wave energy came as westerly swells in fall-winter (United States Army Corps of Engineers Report 1986; W. O'Reilly, Coastal Data Information Program, personal communication). To provide quantitative estimates of the differences in wave action among the sites, I measured bottom flow speeds using electromagnetic current meters (Marsh- McBirney 510) and relative total flow over the bottom using sets of alabaster blocks (Muus 1968).

The current meter records were made as part of a separate study (C. D. Robles et al., unpublished manuscript). The sensors of the meters stood 30 cm above the surface of the turf on metal rods embedded at +0.63 m above MLLW, approximately the mid-line of the turf zone. Mean maximum flow speeds were calculated from individual measurements with a time constant of 0.2 s taken at 1-s intervals within 2-min periods. The 2-min periods began every 15 min from I h before to 1 h after spring high tides (extreme tides above 1.6 m MLLW). Only two current meters were available, so that it was not possible to sample all sites simultaneously. Accordingly, on seven dates from 1989 to 1992 the two meters were run simultaneously at a sheltered location, site 1, and at a relatively more exposed location, between sites 4 and 5. Over the same period of years, asynchronous surveys were made at irregular intervals on sites 1, 2, between sites 4 and 5, and sites between 6 and 7. Unlike the simultaneous sampling of paired sites, the latter assessments could have been biased by the chance selection of sample dates, because wave action may vary from day to day.

The total flow of water should be higher on wave exposed sites because flow speeds are higher and greater wave wash prolongs submergence times. Total flow was estimated as the percentage loss in mass of alabaster blocks (Muus 1968). Blocks with dimensions of 2 X 4 X 6 cm were bolted to the rock within 20 cm of the border of each of the plots used in the exclosure experiment. (The blocks were not placed within the plots to keep from damaging the cover of algae.) They remained in place for two high tides before they were removed and oven dried to constant mass. Two runs of the blocks were made: 28 November 1993 and 26 March 1994.

Long-term surveys of recruitment.-From 1989 to 1995, I estimated recruitment rates at sites 1, 2, and 7, the sites of the prior experiments (Robles and Robb 1993). Two 10 m long transects were laid par- allel to the shoreline on each of the sites, and, by




shaving the sandstone substratum with a chisel, all the attached life was collected in three or four 100- cm2 quadrats spaced randomly along each transect. The samples were examined under a magnifying glass, the mussels were identified, and their lengths measured in 5-mm increments. I used the density of mussels <5 mm long to estimate recruitment. The surveys were repeated at quarterly or more frequent intervals from March 1989 through 1992, and then each March from 1993 to 1995.

The long-term recruitment data were analyzed as a two-way ANOVA (software by SYSTAT) with site and year as the factors. Even resorting to elaborate data transformations (Sokal and Rohlf 1981) the assumption of equal variances could not be met. Acknowledging the violation of the assumption, I present the ANOVA of long-term recruitment with untransformed data because the degrees of bias produced by even extreme heteroscedas.ticity (Underwood 1981) seem unlikely to change the decisions to accept or reject, given the very small P values that resulted.

A "divot survey" for early postmetamorphic recruits.-Analysis of the long-term recruitment surveys raised the question of whether marked spatial differences in mussel recruitment could be attributed to either (1) differences in postsettlement mortality or (2) differences in larval settlement rate among algal species, which themselves differed in abundance among the sites. On 19 April 1996, three 400-cm2 quadrats were dropped haphazardly within sites 3, 5, and 7. These sites spanned a range of wave exposure and recruitment rates. The percent covers of algae were recorded for each quadrat using the point-intercept method (e.g., Cubit 1984), and then divots of turf, each ~'5.0 cm2, were chiseled from the rock. Since the divot was small, it consisted of only one algal species. Corallina officianalis and Gigartina canaliculata were the only algae occurring abundantly on all three sites. Accord- ingly, for each quadrat, five divots were selected at random from the covers of (1) Corallina officianalis, (2) Gigartina canaliculata, and (3) one of several other red algae: Gelidium coulteri, Laurencia pacifica, or Rhodoglossum affine, depending on which were most common at a site.

The numbers of mussels 0.2 to 0.7 mm long were recorded for each divot. This size range included mussels that had settled the day of the sample to approximately the spring high tide 2 wk earlier (A. Martel, personal communication). Mussel species identifica- tions were made using the criteria of Martel et al. (1993). To determine whether postsettlement mortality might have produced the spatial differences in the densities of larger recruits, the densities of the early postmetamorphic recruits were compared among the sites. Mean density at a site was estimated by weighting the mussel counts from divots of a given algal species by that alga's abundances in the quadrats. If postsettlement mortality were solely responsible for the spatial dif-

ferences in densities of larger recruits, then one would expect to find similar densities of early postmetamorphic recruits among the sites.

To determine whether early postmetamorphic recruitment rates depended on the algal species composition, a split-plot ANOVA, with quadrats nested within sites, tested the main effects of site and algal species on the densities of M. californianus. (Densities of M. galloprovincialis were too low to provide tests.) Nesting the quadrats provided an indication of the importance of spatial variation within a site.

Experimental methods

Exclosure design.-In January 1993 I marked three 20 x 40 cm plots at each of the seven sites. Their exact location within a site was haphazard, except for the stipulations that they fall at 0.63 m above MLLW and within a 2-m radius, so that their physical conditions might be similar. Within each site, a treatment (exclosure cage), control for cage effects (open ended cage), and control (open plot) were assigned to -the plots by lottery. The cage itself was wire mesh with 1-cm open- ings, stretched over a welded rectangular frame of steel reinforcing bars 1.3 cm (1/2 inch) in diameter, the legs of which were cemented into holes in the rock. The control for cage effects consisted of the same frame and wire mesh with the end panels removed. Divers observed lobsters and other predators foraging inside these "arch controls."

In the prior experiments (Robles and Robb 1993) at the Cove (site 1) whelks repeatedly penetrated the wire mesh of complete predator removal treatments. Thus, the fact that the effect of treatments was small might have been an artifact of the imperfect exclosure, rather than low recruitment and growth of prey. In the present experiment, I attempted to exclude all whelks by paying careful attention to the fit of the lower margin of the mesh to the irregular stone surface and by placing the exclosures slightly higher (0.63 m as opposed to 0.4 m above MLLW), a height at which whelk densities were lower (Robles 1987). Despite my efforts, some whelks did manage to enter exclosures. I examined all 21 plots at -1-mo intervals over the course of the experiment. A total of seven Ceratostoma nuttalli were observed in the 231 records. Six of the seven were small individuals (spire height <2 cm) that had entered cage exclosures, evidently in response to increased prey densities (see Robles and Robb 1993 for discussion). Four of the six whelks entering enclosures did so at site 1, Big Fish- erman Cove. Whelks were very rare on the turf outside the Cove, and the two whelks found in exclosures on Bird Rock occurred in the final month of the experiment, long after the differences in mussel density had developed. Therefore, the requirements of the experimental design were met with the exception that whelks may have killed some mussels in the predator exclosure at Big Fisherman Cove, site 1.

The changes in percent cover were estimated by the




TABLE 1. Maximum flow speeds (mean ? 1 SE) occurring < 1 h before and after the time of predicted extreme high tides in spring and summer, calculated from asynchronous current meter records. N = number of high tides on which data were logged.

Sites 1 2 3 4 and 5 6 and 7

Speeds (cm/s) 27.7 ? 1.2 60.0 ? 5.1 61.5 ? 5.2 82.9 ? 2.8 86.5 ? 4.9 N 11 2 3 13 13 Group a b b c c

Note: Similar means are grouped using Tukey's hsd test, with the Tukey-Kramer adjustment, and transformed data (Sokal and Rohlf 1981).

point intercept method (e.g., Cubit 1984). The estimates were made at the beginning of the experiment, February 1993, and its conclusion, February 1994. I also made rough visual estimates of percent covers with the wire mesh in place approximately once every 2 mo.

Initial andfinal experimental recruitment. -Samples of the densities of the mussels were also taken from each plot immediately before the exclosures were in- stalled and at the conclusion of the experiment. The samples were 100-cm2 quadrats scraped up and examined as in the other recruitment surveys. The samples were taken from one corner of the plot, selected at random, leaving the center of the plot undamaged for the percent cover surveys. Results of the long-term surveys of mussel recruitment (see below) indicated that peak settlement occurred in winter. The initial sampling in the experiment, therefore, served as a measure of the natural recruitment rates of the sites.

The results of the experiment were evaluated with ANCOVA. Percent cover or density of Mytilus was the dependent variable, treatment group (open, arch, or cage) was the main effect, and the initial density of

50

40

0

CZ)

u)~ 20 C,) 0

10

1 2 3 4 5 6 7

Site

FIG. 2. Percentage loss in mass (means and 1 SE) of alabaster blocks, used to measure total flow over the substratum, plotted vs. the alongshore sequence of the sites for separate runs in November 1993 (hatched bars) and March 1994 (solid bars).

recruits was the covariate. Use of ANCOVA provided a test of the following crucial prediction: if the effect of predator removals depended on prey recruitment, then one would expect to see a significant interaction between the main effect and the covariate, the treatment X recruitment term.

RESULTS

Survey results

Wave exposure.-The estimates of mean maximum flow speeds, measured by current meters on high tides in spring and summer, confirmed a significant difference in wave exposure between the leeward and windward sites. The synchronous measurements yielded values of 25.5 ? 2.4 cm/s and 81.7 ? 25.5 cm/s, respectively [means ? 1 SE, for the leeward (site 1) and windward (between sites 4 and 5) locations]. The flow speeds for the windward location exceeded speeds at the leeward by at least twofold on all seven sample dates (sign test, N = 7, P = 0.0078). Similarly, the asynchronous records indicated a gradient of exposure increasing from Big Fisherman Cove, to the moderately exposed sites in the lee of Bird Rock, to the most exposed sites on the west end of Bird Rock (Table 1).

Paralleling the estimates of current speeds, the mean percentage mass losses of the alabaster blocks increased from site 1 to site 7 (Fig. 2, Table 2). However, mean losses in mass for sites 2 and 7 did not match their position in the alongshore sequence of sites (Fig. 2). Site 2 was located at the leeward end of the islet. Higher energy waves have longer periods, and thereby they refract around obstacles (Denny 1988). During the November 1993 run, I observed large westerly swells bending around Bird Rock, striking the eastern shore in the area of site 2, which ranked above sites 3 and 4 for this run. Mean losses in alabaster mass for site 7 ranked below those of site 6. Evidently, a low rock ledge just to the windward side of site 7 reduced the wave exposure. Whatever the causes of the variation, the subsequent statistical analyses using wave exposure estimates produced the same inferences regardless of whether the alabaster runs were used separately, pooled, or replaced by the alongshore sequence of the sites.

Total flow differed among the sites, but it did not differ among the experimental plots within sites (Table 2).




TABLE 2. Factorial ANOVA of percentage loss in mass of alabaster blocks measuring total flow rates. The factors are treatment group (cage, roof, or control), trial (the November 1993 and March 1994 runs), and site (the seven locations along shore). R2 = 84%.

Source Sum of squares df Mean square F P

Treatment 8.27 2 4.13 0.19 0.826 Trial 636.73 1 636.73 29.85 <<0.001 Site 1056.61 6 176.10 8.25 <<0.001 Three-way interaction 129.77 12 10.81 0.51 0.881 Error 341.36 16 21.34

Notes: Site 6 had fewer records than the others because two and one blocks were lost, during the November 1993 and March 1994 runs, respectively. The limited degrees of freedom prevented inclusion of both three-way and two-way interaction terms. However, none of the possible two-way interactions proved significant in separate models including them but omitting the three-way interaction.

Long-term recruitment surveys. -The surveys revealed consistent differences in mussel recruitment among the three sites (Fig. 3). The two relatively sheltered sites, Big Fisherman Cove (site 1) and the east end of Bird Rock (site 2), received fewer recruits of M. californianus and more M. galloprovincialis than did the wave-exposed west end of Bird Rock, site 7. Since M. californianus was by far the most abundant recruit, total densities of Mytilus spp. recruits were often >2 orders of magnitude higher on the wave-exposed site 7 (Fig. 3).

The alongshore differences were not a consequence of shore-level differences in recruitment. The shore level of the plots was the same for all sites, and surveys of shore-level differences in recruitment (Robles 1987; C. Robles, unpublished data) indicate that the peak recruitment occurs between 0.50 and 0.75 m above MLLW across the wave exposure gradient.

The long-term surveys showed high seasonal and year to year variation. Peak densities of recruits occurred in winter to early spring, but year to year variation appeared to be much greater than seasonal variation (Fig. 3). The temporal variation differed among sites, so that the sheltered sites apparently had no significant recruitment of M. californianus for years at a time. The exposed west end of Bird Rock received high but variable recruitment. These differences are reflect- ed in the significant interaction terms of the ANOVAs run for the annual, winter samples (Table 3a, b).

The divot survey for early postmetamorphic recruits.-The density of early postmetamorphic M. californianus increased from 1.20 + 0.46 inds./100 cm2 (mean + 1 SE) to 39.14 + 9.06 to 200.94 + 33.56 inds./100 cm2, respectively, for sites 3, 5, and 7 (ANO- VA of transformed weighted densities: F = 27.00; df = 2, 6; P = 0.001). M. galloprovincialis increased from 0.34 + 0.34 to 1.60 + 1.60 to 7.78 ? 1.88 inds./100 cm2, respectively, for sites 3, 5, and 7 (ANOVA of weighted densities: F = 7.63; df = 2, 6; P = 0.022). These differences, and the observation that newly metamorphosed mussels (-300 Elm long) were ex- ceedingly rare in the samples from sheltered sites in this and the other surveys, indicate that the spatial dif-

ferences in recruitment were present from the time of settlement.

While very few early recruits were found on any species of algae at the sheltered site 3, they were abundant and about twice as common on coralline as on fleshy red algae at the exposed sites (Table 4). Con- sequently, the split-plot ANOVA yielded a significant site X algae interaction term (Table 5b). Corallina officianalis and Gigartina canaliculata showed similar high percent covers on all three sites; four other species of fleshy red algae were less common (Fig. 4). The variation of algal covers within sites was high, so that only Rhodoglossum affine was found to differ signifi- cantly among the sites (ANOVA; df = 2, 6; P = 0.01). Although some of the among-site difference in mean weighted densities of early postmetamorphic mussels could be attributed differences in percent covers of algae alone, the bulk of the difference stems other factors varying along shore. Spatial variation within sites was not significant (quadrat effect, Table 5a).

Experimental results

Initial experimental recruitment.-At the outset of the experiment, Mytilus californianus were the most abundant recruits in the experimental plots (mean density of M. californianus <5 mm long = 26.3 ? 11 .1 inds./100 cm2 [mean + 1 SE]; N = 21, i.e., 1 sample per plot). M. galloprovincialis was comparatively rare (mean density of M. galloprovincialis <5 mm = 1.3 ? 0.5 inds./100 cm2, N = 21), and it occurred in appreciable numbers only at the most sheltered site, Big Fisherman Cove.

The initial densities of M. californianus recruits, and thus total recruits of Mytilus spp., were positively cor- related with wave exposure (Spearman correlation [rj] between the densities of M. californianus <5 mm and the mean percentage mass losses of alabaster blocks pooled for both runs for each plot: r, = 0.68, N = 21, P = 0.01). Initial recruitment was nil or nearly so for the leeward sites on Bird Rock (sites 2-4) and increased as much as two orders of magnitude towards the wave- exposed west end of Bird Rock (sites 5-7, Fig. 5). M. galloprovincialis densities also differed along the shore-




75 Site 1

60-

45-

30-

15

0 Site 2

N' 60_

0 E

45 0

Cn

230

co 15

0 Site 7 99.7

60-

208.7 45-

30-

1 5

0 0) 0) 0) 0) 0) 0) 0) 0) 0) 0) 0) 0) 0) 0) 0) 0) 0) 0) 0) 0)

m _r 0 > m z a- 0 M z M x z 0 M Mr Mrx

FIG. 3. Densities of recruits of Mytilus californianus (solid bar portions) and M. galloprovincialis (hatched portions) at sites 1, 2, and 7 taken at seasonal to annual intervals from 1989 to 1995. Data show total Mytilus densities (means and 1 SE). (Table 3 ANOVA based on annual, winter samples only).

line, but only because they were comparatively numerous at site 1, Big Fisherman Cove (Fig. 5, Kruskal-Wallis test, mean densities of M. galloprovincialis: H = 13.31, df = 6, P = 0.038). The six sites on Bird Rock spanned a stretch of shoreline 175 m long (Fig. 1). Thus, a

TABLE 3. Factorial ANOVAs of long-term Mytilus recruitment surveys. The factors are year (1989, 1995) and site (1, 2, or 7). Records are from one sample each year in winter, the season that usually showed the greatest mussel recruitment. Data used are inherently heteroscedastic.

Sum of Mean Source squares df square F P

a) Analysis for M. californianus (R2 = 65%) Site 41513.79 2 20756.89 26.37 <<0.001 Year 38439.85 6 6406.64 8.14 <<0.001 Interaction 68420.27 12 5701.69 7.24 <<0.001 Error 73996.67 94 787.20

b) Analysis for M. galloprovincialis (R2 = 42%) Site 439.36 2 219.68 6.62 0.002 Year 720.28 6 120.05 3.62 0.003 Interaction 1066.52 12 88.88 2.68 0.004 Error 3121.63 94 33.21

marked, regular change in recruitment rates occurred over a remarkably short span of apparently uniform habitat.

Results of exclosure.-At the experiments conclusion, the densities of M. californianus > 1 cm long had increased in exclosure plots depending on the level of initial recruitment, while densities of the larger mussels in control plots remained nearly nil (Fig. 6). Thus, the prediction that the outcome of predator enclosures depends on the initial prey recruitment was verified (significant treatment X recruitment term, Table 6). Re- markably, the linear ANCOVA model accounted for nearly all the variation in densities of M. californianus >1 cm (Table 6, R2 = 99.5%). The treatment X recruitment interaction was also significant for M. galloprovincialis (Fig. 7, Table 7) but because this species was comparatively scarce the relationships were not as clear (R2 = 55%).

The survival of masses of Mytilus spp. in the predator enclosures caused changes in percent cover because mussels longer than 1-2 cm overtopped and eventually smothered the woolly algal turf (Fig. 8). The treatment X recruitment interaction was significant, and the linear ANCOVA model accounted for much of the variation in the cover of Mytilus spp. (Table 8, R2 = 90%).

The species composition of the mussel covers matched the among-site differences in recruitment of the two species. M. galloprovincialis covered much more of the exclosure plot in Big Fisherman Cove. The mussel cover at the exposed sites consisted almost com- pletely of M. californianus. I observed the covers in enclosures until early spring 1994. At that time, M. californianus covered almost 100% of the plot at site 6, and M. galloprovincialis alone covered 25% of the exclosure plot in Big Fisherman Cove.

The changes in percent covers and densities cannot reasonably be attributed to artifacts involving effects of the wire mesh itself (discussion in Dayton and Oliver 1980). The coralline and fleshy algae appear to differ in their sensitivity to desiccation and high insolation




TABLE 4. Densities of early postmetamorphic mussels (0.2-0.7 mm long) in the divot survey. Numbers of a mussel species per 5-cm2 divot (means -+ 1 SE) are tabulated by site and algal species.

Algal Site 3 Site 5 Site 7 speciest Mussels (Leeward) (Middle) (Windward)

Corallina sp. M. californianus 0.07 + 0.07 2.13 + 0.53 11.27 + 1.93 M. galloprovincialis 0.07 + 0.07 0.40 ? 0.24 1.33 + 0.47

Gigartina sp. M. californianus 0.13 + 0.09 1.07 ? 0.32 4.94 + 0.57 M. galloprovincialis 0.00 + 0.00 0.00 ? 0.00 0.00 + 0.00

Gelidium sp. M. californianus 0.08 + 0.08 0.40 ? 0.25 -- M. galloprovincialis 0.00 + 0.00 0.00 ? 0.00 ...

Pterocladia sp. M. californianus 0.40 + 0.40 3.33 + 1.23 ... M. galloprovincialis 0.00 + 0.00 0.00 + 0.00 ...

Rhodoglossum sp. M. californianus ... 1.00 + 0.52 5.86 + 1.08 M. galloprovincialis *- 0.00 ? 0.00 0.14 + 0.14

Laurencia sp. M. californian.s * 4.15 + 1.40 M. galloprovincialis 0.17 ? 0.17

t At any given site, some of the species of fleshy red algae were too rare to provide enough samples for estimates.

(Seapy and Littler 1982). Yet, a MANCOVA comparing percent covers of coralline and fleshy algae between open and arch-covered controls at the winter conclusion of the experiment was not significant (MANCO- VA: arcsine transformed percentages blocked by site, Hotelling Trace = 0.899; F = 2.25; df = 2, 5; P = 0.201). Differences between cage plots and controls developed in spring and summer, and at a comparatively low shore level (+0.6 m MLLW), conditions in which desiccation stress appears to be minimal and predation pressure the greatest (Robles and Robb 1993). The initial recruitment rates of M. californianus in plots selected to receive cages (the January 1993 sample) and their final recruitment rates on these plots

TABLE 5. Split plot ANOVA of densities of early postmetamorphic M. californianus in the divot survey.


a) Model I ANOVA for the effects of quadrats within sites (RJ2= 78%)

Site 1117.10 2 558.55 44.05 ... Algaet 133.40 1 133.40 10.52 ... Site X Algae 179.11 2 89.56 7.06 ... Quadrat {Sitel 14.70 6 2.45 0.19 0.978 Algae x Quad-

rat ISite) 11.56 6 1.93 0.15 0.988 Error 925.61 73 12.68 b) Model II ANOVA for effects of sites and algae (R2 = 78%)

Site 1117.10 2 558.55 228.04 0.000 Error 14.70 6 2.45 Algae 133.40 1 133.40 69.22 0.000 Error 11.56 6 1.93 Site X Algae 179.11 2 89.56 46.47 0.000 Error 11.56 6 1.93

t Algae included an articulated coralline alga, Corallina officianalis, and a fleshy red alga, Gigartina canaliculata; other species of fleshy red algae were omitted because they did not occur in abundance at every site.

(the February 1994 sample) were similar (Pearson r = 0.83, P < 0.03), which indicates that the cages did not directly affect recruitment.

DISCUSSION

"Recruitment shadows" and the intensity of species interactions

Over the last 30 yr, benthic ecologists emphasized intense interactions among sedentary species in the postrecruitment phase of their life cycles (reviews in Connell 1972, 1975, Menge and Sutherland 1976, Un- derwood and Denley 1984, Paine 1994, Menge 1995). The keystone predator paradigm, intermediate disturbance hypothesis, and some other theories arising in this period provided explanations for the coexistence of the many species competing for the same, potentially limiting resource. The impetus for these theories may be traced back to the dilemmas presented by the competitive exclusion principle (Gause 1934, Hardin 1960), and thence back to the speculations of Darwin (1859) and Malthus (1798) about the consequences of geometric increase of populations. Laboratory cultures, the first tests of the emerging corpus of ecological theory, confirmed the effects of intense species interaction during rapid population growth (e.g., Gause 1934, Crombie 1946, Huffaker 1958, Park 1962). Thus, the potential for high reproduction of species has served historically as a foil for theoretical discourse, and, necessarily, directed attention to intense species interactions. Whether this potential is manifest in a given natural population is a pivotal ecological question.

For benthic species with planktotrophic larvae, interactions on the shore ensue only after a sequence of crucial events in their early life history. The sequence runs from the spawning of adults, to planktonic de- velopment, to the recruitment of the larvae to shore populations, to the grow-out of these young to suffi-




Site 3 CORALLINA

GIGARTINA

OTHER LAURENCIA

PTEROCLADIA

GELIDIUM

Site 5

CORALLA

_ Rr n~OTHER RHODOGLOSSUM

PTEROCLADIA

GIGARTINA GELIDIUM

Site 7 CORALLINA

LAURENClA

GIGARTINA RHODOGLOSSUM

FIG. 4. Mean percent covers of algal species at study sites in the divot survey.

cient sizes to exert competitive effects. Production may be diminished at any link in the chain preceding recruitment. Adults may find themselves without nearby mates (Pennington 1985) or sufficient resources to reproduce. Zygotes and larvae may be advected away (Parrish et al. 1981, Roughgarden et al. 1988, Bailey 1991, Gaines and Bertness 1992, Hobbs et al. 1992), may starve or die by planktonic predators (Gaines et al. 1985) before reaching the habitats of the adults, and those larvae finally reaching the shore may grow poor- ly, depending on the resources provided by the habitat. Except for predation in the plankton, none of the mechanisms necessarily involve intense species interactions. Yet, working separately or together, these mechanisms appear responsible for the pronounced variation in recruitment rates documented for benthic invertebrates (Caffey 1985, Hughes 1990, Raimondi 1990, Bertness

175

1 M. californianus

M. galloprovinca

105 _ _

70- c-o Cfi| 70_

Z35

1 3 4 2 5 7 6

Sites FIG. 5. Mussel recruitment (inds./100 cm2 plot) at the

outset of the experiment [means and I SE]. Heights of bars represent mean total mussels; shadings represent relative pro- portions of the two species: M. californianus (solid) and M. galloprovincialis (hatch). The bars are arranged on the x-axis by rank of mean total flow (fall and winter means averaged for each site).

et al. 1992; see Gaines and Bertness 1992, Connell 1985 for reviews). Therefore, whether intense postrecruitment competition or predation occurs on the shore may ultimately depend on antecedent factors affecting rates of recruitment.

N 100- 0

To~~~~~~~~~~~R 0 C~~~~

10- Cn

2 1.0 10 100

0

Mean Density of Initial Recruits (no./100 CM2)

FIG. 6. Densities of M. californianus >1 cm at the conclusion of the experiment plotted on log scales as a function of the species' initial recruitment. Circles = open plots, triangles = arch covered plots, and squares = cage covered plots. Site numbers appear inside symbols. Overlapping symbols have been offset to make site numbers legible.




TABLE 6. ANCOVA for densities of M. californianus >1 cm long at the conclusion of the recruitment variation experiment. The main effect is treatment group (cage, roof, open control); the covariate is the initial experimental recruitment of this species. R2 = 99.5%.


Treatment 1.11 2 0.56 0.08 Recruitment 9085.43 1 9085.43 1254.59 ... Interaction 13587.05 2 6793.53 938.11 0.001 Error 108.63 15 7.24

Notes: Spearman correlations (rs) between initial recruitment and densities of M. californianus >1 cm are rs = 0.54, 0.61, and 0.94*, respectively, for open, arch, and cage treatments.

* Significant under procedurewise ax = 0.05, employing the Bonferroni correction.

Recognizing the potential limitations of theory based solely on postrecruitment interactions, benthic ecologists recently focused their attention on the effects of recruitment variation on the structure of benthic populations (e.g., Keough 1984a, b, Caffey 1985, Gaines et al. 1985, Gaines and Roughgarden 1985, Sutherland and Ortega 1986, Sutherland 1987, 1990, Davis 1988, Hughes 1990, Raimondi 1990, Bertness et al. 1992, Menge 1992, Robles et al. 1995). Both theory and empirical evidence argue that variation in recruitment can regulate the intensity of postrecruitment interactions.

The results from Santa Catalina Island suggest that constraints are sometimes imposed early enough in the sequence of events leading to recruitment to preclude

10t- wePV

E

0

E 2~

(I)~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~

0 1.0 1 0 Mean Density of Initial Recruits (no./100 CM2)

FIG. 7. Densities of M. galloprovincialis > I cm at the conclusion of the experiment plottedl on log scales as a function of this species' initial recruitment. Circles = open plots, triangles = arch covered plots, and squares = cage covered plots. Site numbers appear inside symbols. Overlapping symbols have been offset to make site numbers legible, as in the cluster at the origin.

TABLE 7. ANCOVA for densities of M. galloprovincialis > 1 cm long (analysis as in Table 6). R2 55%.


Treatment 0.59 2 0.30 0.124 ... Recruitment 9.71 1 9.71 4.08 ... Interaction 19.61 2 9.81 4.12 0.038 Error 35.71 15 2.38

Mytilus

Control: Open

Coralline 2)(X) 5(Z5 Fleshy Algae Algae

Mytilus

Control: Arch

Coralline 3 Fleshy Algae Algae

Mytilus

Treatment: 6 Cage

Coralline 24 Fleshy Algae Algae

FIG. 8. Triangular coordinate diagrams of the relative cover of Mytilus spp., fleshy red algae, and the coralline alga Corallina officianalis at the end of the experiment (February 1994). The three species groups accounted for ?80% of the total cover of each plot. Symbols near the center of each large triangle represent approximately equal coverages of the three types of cover; symbols near the apex represent a nearly con- tinuous cover of the Mytilus spp. assemblage. Circles = open plots, triangles = arch covered plots, and squares = cage covered plots. The top figure pertains to open plots (control), the middle to arch covered plots (control), and bottom to cage covered plots (treatment).




TABLE 8. ANCOVA for percent covers of Mytilus spp. in the experiment. Sum of squares was calculated for 103 X the arcsine transform (Sokal and Rohlf 1981). The main effect is treatment group, and the covariate is initial recruitment of Mytilus spp. R2 = 90%.


Treatment 512.87 2 256.43 2.13 ... Recruitment 5303.07 1 5303.07 43.98 ... Interaction 8432.52 2 4216.26 34.97 <<0.0001 Error 1808.60 15 120.57

Note: Spearman correlations between initial recruitment and the transformed percent covers are rs = 0.72, 0.16, and 0.92*, respectively, for open, arch, and cage treatments.

* Significant under procedurewise a = 0.05, employing the Bonferroni correction.

intense species interactions. The analysis of the densities of early postmetamorphic recruits suggests that the spatial differences in initial experimental recruitment probably resulted directly from differences in settlement rate, rather than postsettlement mortality. I did not directly observe settlement, and the proximal mechanisms causing the spatial differences in early postmetamorphic recruitment remain a matter of specula- tion. Once established, juvenile mussels either suc- cumb to predation or exert competition.

A finding that species have no impact in habitats in which they do not occur would have been trivial. A noteworthy feature of the present work is that extreme variation in the rates of recruitment of the dominant prey occurred over a small distance within a specific habitat, in which the composition of the characteristic species assemblage remained relatively constant from year to year across a range of physical conditions.

Observations of marked alongshore differences in rates of recruitment of a species are common (e.g., Underwood et al. 1983, Caffey 1985, Gaines et al. 1985, Kendall et al. 1985, Sutherland and Ortega 1986, Raimondi 1990, Sutherland 1990, Bertness et al. 1992). Areas of chronically low recruitment next to consis- tently high recruitment areas within the same habitat I term "recruitment shadows." The existence of recruitment shadows suggests that significant and predictable differences in the intensity of postrecruitment interactions can occur.

Petraitis (1974, 1978) also found a positive correlation between wave exposure and the relative abundances of M. californianus recruits. He found a neg- ative correlation for M. galloprovincialis (=M. edulis). He questions Harger's (1972) hypothesis that M. galloprovincialis competitively excludes M. californianus in sheltered locations, noting that recruitment patterns should preclude the interaction over much of the wave exposure gradient. Analogously, predation on mussels or competition between mussels and algae should be negligible in the recruitment shadow at Bird Rock. But at site 1, unlike Petraitis' (1974, 1978) comparison, I

found that the recruits of different Mytilus species sometimes commingled (Fig. 3; see also Menge et al. 1994, Robles et al. 1995).

The positive relationship between wave exposure and high recruitment may hold some advantage for M. californianus. If recruits grow slowly in sheltered areas, then individual M. californianus should (1) increase size and hence reproductive output relatively slowly and (2) remain vulnerable to predators for pro- longed periods. Recruiting to sites with high wave exposure should increase the likelihood that predator swamping (Dayton 1971) will be successful, because fast growing mussels remain vulnerable for a comparatively short period, and hydrodynamic stresses can interrupt predator foraging more frequently (Menge 1978a, b, Robles and Robb 1993; C. D. Robles et al., unpublished manuscript). This explanation depends on the positive relationship between wave exposure and mussel growth rates (Leigh et al. 1987, Robles and Robb 1993, Sherwood-Stephens 1993) and on size limits to predation for M. californianus (Harger 1970, 1972, Dayton 1971, Paine 1974, Robles et al. 1990). Size-limited predation characterizes the interaction of M. californianus with all of its predators, and the explanation may apply throughout its range, beyond the northern distribution limit of the lobsters at Point Con- ception, California.

Judging by the distribution of adults (see, for example, Stephenson and Stephenson 1972), M. californianus is successful only on rocky shores with high wave exposure. On sites 6 and 7, beds of adult M. californianus do occur at higher shore levels, and on the same shore level 5-10 m to the windward. At high tide, I have observed lobsters roaming freely from the experimental (turf) sites to the extreme vertical and horizontal limits of the mussel beds. In the absence of absolute spatial or age/size refugia (Robles et al. 1990), the dynamics of prey production and consumption evidently tip in favor of the mussels at the wave-beaten extreme of the exposure gradient. Thus it appears that the tendency of M. californianus to recruit most heavily to wave-exposed sites is an advantageous feature of a life history characterized by size-limited predation and indeterminate growth rates (Sebens 1987) that depend on wave exposure.

Changing rates, constant assemblages, and ecological generalities

Depending on one's perspective, field experiments are blessed or damned by variation. The same manipulations often produce different results, and in some cases the differences are extreme (discussion in Dayton 1971, Paine 1976, Underwood and Petraitis 1993; other examples in Robles and Cubit 1981, Keough 1984b, Dethier and Duggins 1988, Fairweather and Under- wood 1991, Robles and Robb 1993, Menge et al. 1994). Arising unexpectedly yet repeatedly, such variation has led to debate over whether generalization from field




studies is possible or even desirable at present (see for example Underwood and Denley 1984, Foster 1990, Paine 1991, Estes and Duggins 1995).

Fairweather and Underwood (1991) describe variation similar to the Catalina findings: the effect of predator removals varied among replicates at different locations. In their case, the predators were whelks (Mor- ula marginalba) preying on barnacles, tube worms, and other sedentary invertebrates on rocky shores of south- eastern Australia. They attribute the variation to pre- sumed small scale differences in wave stress, micro- climate, topography, and historical factors that might influence the intensity of predation.

The composition of the predator fauna can also change from site to site within an intertidal habitat (e.g., Robles and Robb 1993 and Menge et al. 1994). In prior experiments at site 7, exclosure of lobsters with large mesh (2.5-cm2 opening) cages, produced marked increases in mussels, whereas whelks (Ceratostoma nuttalli and Maxwellia gemma) passed through the mesh and prevented similar results in large mesh exclosure experiments at site 1 (Robles and Robb 1993). Whether the core predators are lobsters (Robles and Robb 1993) or sea stars (Menge et al. 1994), predation can shift from "keystone" to "diffuse" over short distances from the wave exposed to sheltered portions of the habitat.

The variation among replicates in the present experiment cannot be attributed to either spatial variation in physical stresses affecting the predators, or to shift- ing composition of the predator fauna. The lobsters were present only at high tide, and wave action might have hindered their foraging at exposed sites (C. D. Robles et al., unpublished manuscript). However, whatever stresses where present did not prevent them from eliminating the mussels even at the comparatively wave exposed sites 5-7. The variation among the replicates occurred within the enclosures, and therefore could not have been caused by differences in the intensity or mechanisms of predation, although such differences may arise along with the shift from diffuse to keystone predation along the wave exposure gradient. Rather, differences in the recruitment of mussels accounted for most of the variation among replicates.

M. galloprovincialis did recruit to site 1, and this species replaced some of the turf in the predator exclosure (see also Robles and Robb 1993). However, this cover was lost immediately upon re-exposure to predators (C. D. Robles, personal observation; Robles and Robb 1993). M. galloprovincialis and closely related congeners (M. trossulus and M. edulis) possess a thin shell, and they remain vulnerable to, and preferred by, mussel predators throughout their lives (Harger 1970, 1972, Elner 1978, Seed 1979). Thus, low intensity diffuse predation, i.e., low densities of lobsters, fishes, and whelks together, could be expected to prevent the establishment of a perennial cover of mussels of this species. In contrast, a prior lobster removal ex-

periment at site 7 (Robles and Robb 1993) produced a dense cover of predominantly large (>5 cm) M. californianus that persisted indefinitely after re-exposure to predators.

I have done four predator removal trials on site 1 in four different years from 1982 to 1994. Whelks penetrated enclosures, and they probably affected the rates at which the mussel replaced the turf. But, considering the outcome of the surveys and the outcome of exclosures where no whelks were present, it seems reason- able to attribute the failure of the removals in very sheltered areas to produce a persistent cover of mussels to the low recruitment of M. californianus.

Two prior studies provide consistent interpretations of spatial differences in the effect a keystone predator by considering variation in prey recruitment. Menge et al. (1994) propose that the occurrence of keystone predation by sea stars (Pisaster ochraceus) on mussels (Mytilus spp.) is associated with high "food input rates," i.e., the recruitment of all sedentary prey species including mussels and barnacles; low food input favors diffuse predation. Presumably, greater prey availability promotes greater biomass of the keystone predator's population by elevating its settlement, growth rates, or densities via alongshore movements (see Robles et al. 1995 for discussion). In the second study, working with observational evidence alone, Es- tes and Duggins (1995) propose that the rates at which kelp beds recover following reintroduction of sea otters (Enhydra lutris) depends, in part, on the recruitment rates of herbivorous sea urchins (Strongylocentrotus spp.). In regions where urchin recruitment is high, the otters, which prefer to consume larger urchins, reduce grazing pressure comparatively slowly, and the recovery of the kelp beds is retarded relative to regions with lower urchin recruitment rates.

Variation in recruitment rates affects experimental assessments of the effects of postrecruitment factors, and the generalities these assessments support (see also Underwood and Petraitis 1993, Menge et al. 1994). Depending on the levels of mussel recruitment, keystone predation, diffuse predation, or no predation may be required to maintain the distinctive assemblage of algae at different times and locations along the shoreline of Santa Catalina Island.

ACKNOWLEDGMENTS

Suggestions by S. Gaines, R. Nakamura, and an anonymous reviewer greatly improved drafts of the manuscript. M. Al- varado, C. Chauteco, C. Hernandez, M. Maiten, N. Nipata- rudi, K. Pan, R. Sherwood-Stephens, and F. Villeda helped with the field work. The expense of the project was borne by the National Science Foundation, nos. R118804679 and OCE9217235001. This is contribution no. 175 of the Publi- cations of the Catalina Marine Science Center.

LITERATURE CITED

Bailey, K. M. 1991. Larval transport and recruitment of Pa- cific hake Meruccius productus. Marine Ecology Progress Series 6:1-9.

Bertness, M. D., S. D. Gaines, E. G. Stephens, and P. 0.




Yund. 1992. Components of recruitment in populations of acorn barnacle Semibalanus balanoides (Linnaeus). Journal of Experimental Marine Biology and Ecology 156:199- 215.

Caffey, H. M. 1985. Spatial and temporal variation in settlement and recruitment of an intertidal barnacle. Ecolog- ical Monographs 55:313-332.

Connell, J. H. 1961a. Effects of competition and predation by Thais lapillus and other factors on natural populations of the barnacle Balanus balanoides. Ecological Mono- graphs 31:61-104.

. 1961b. Influence of interspecific competition and other factors on the distribution of the barnacle Chthamalus stellatus. Ecology 42:710-723.

1970. A predator-prey system in the marine intertidal region. I. Balanus glandula and several predatory species of Thais. Ecological Monographs 40:49-78.

. 1972. Community interactions on marine rocky intertidal shores. Annual Review of Ecology and Systematics 3:169-192.

. 1975. Some mechanisms producing structure in natural communities. Pages 460-490 in M. L. Cody and J. Diamond, editors. Ecology and evolution of communities. Belknap, Cambridge, Massachusetts, USA.

. 1978. Diversity in tropical rain forests and coral reefs. Science 199:1302-1310.

. 1983. On the prevalence and relative importance of interspecific competition: evidence from field experiments. American Naturalist 122:661-696.

. 1985. The consequences of variation in intertidal settlement vs. post settlement mortality in rocky intertidal communities. Journal of Experimental Marine Biology and Ecology 93:11-45.

Connell, J. H., and R. 0. Slatyer. 1977. Mechanisms of Suc- cession in natural communities and their role in community stability and organization. American Naturalist 111: 1119- 1144.

Crombie, A. C. 1946. Further experiments on insect competition. Proceedings of the Royal Society of London (Se- ries B) 133:76-109.

Cubit, J. D. 1984. Herbivory and seasonal abundance of algae on high intertidal rocky shore. Ecology 65:1904- 1917.

Darwin, C. 1859. On the origin of species. John Murray, London, England.

Davis, A. R. 1988. Effect of initial variation in settlement on distribution and abundance of Podoclavella mouccensis Sluiter. Journal of Experimental Marine Biology and Ecol- ogy 117:157-167.

Dayton, P. K. 1971. Competition, disturbance, and community organization: the provision and subsequent utili- zation of space in a rocky intertidal community. Ecological Monographs 41:351-389.

Dayton, P. K., and J. S. Oliver. 1980. An evaluation of experimental analysis of population and community patterns in benthic marine environments. Pages 93-120 in K. R. Tenore and B. P. Coull, editors. Marine benthic dynamics. University of South Carolina Press, Columbia, South Car- olina, USA.

Denny, M. W. 1988. Biology and the mechanisms of the wave-swept environment. Princeton University Press Princeton, New Jersey, USA.

Dethier, M. N., and D. 0. Duggins. 1988. Variation in strong interactions in the intertidal zone along a geographical gradient: a Washington-Alaska comparison. Marine Ecology Progress Series 50:97-105.

Dungan, M. L., T. E. Miller, and D. A. Thompson. 1982. Catastrophic decline of a top carnivore in the Gulf of California intertidal zone. Science 216:989-991.

Elner, R. W. 1978. The mechanics of predation by the shore

crabs, Carcinus maenus on the edible mussel, Mytilus edulis. Oecologia 36:333-344.

Estes, J. A., and D. 0. Duggins. 1995. Sea otters and kelp forests in Alaska: generality and variation in a community ecological paradigm. Ecological Monographs 65:75-100.

Fairweather, P. G., and A. J. Underwood. 1991. Experimental removals of a rocky intertidal predator: variations within two habitats in the effects on prey. Journal of Experimental Marine Biology and Ecology 154:29-75.

Foster, M. S. 1990. Organization of macroalgal assemblages in the Northeast Pacific: the assumption of homogeneity and the illusion of generality. Hydrobiologia 192:21-33.

Frank, P. W. 1965. The biodemography of an intertidal snail population. Ecology 46:831-844.

Gaines, S. D., and M. D. Bertness. 1992. Dispersal of juveniles and variable recruitment in sessile marine species. Nature 360:579-578.

Gaines, S. D., S. Brown, and J. Roughgarden. 1985. Spatial variation in larval concentration as a cause of spatial variation in settlement for the barnacle, Balanus glandula. Oec- ologia 67:267-272.

Gaines, S. D., and J. Roughgarden. 1985. Larval settlement rate: a leading determinant of structure in an ecological community of the marine intertidal zone. Proceedings of the National Academy of Science 82:3707-3711.

Gause, G. E 1934. The struggle for existence. Williams and Wilkins, Baltimore, Maryland, USA.

Hardin, G. 1960. The competitive exclusion principle. Sci- ence 131:1292-1297.

Harger, J. R. E. 1970. Comparisons among growth characteristics of two species of mussels, Mytilus edulis and My- tilus californianus. Veliger 13:44-56.

. 1972. Competitive coexistence among intertidal invertebrates. American Scientist 60:600-607.

Hatton, H. 1938. Essais de bionomie explicative sur quelques especies intercotidales d'algues et d'animaux. Annales De L'Institut Oceanographique De Monaco 17:241-238.

Hobbs, R. C., L. W. Botsford, and A. Thomas. 1992. Influ- ence of hydrographic conditions and wind forcing on the distribution and abundance of Dungeness crab, Cancer ma- gister, larvae. Canadian Journal of Fisheries and Aquatic Science 49:1379-1388.

Huffaker, C. B. 1958. Experimental studies on predation: dispersion factors and predator-prey oscillations. Hilgardia 27:343-388.

Hughes, T. P. 1990. Recruitment limitation, mortality, and population regulation in open systems: a case study. Ecol- ogy 71:12-20.

Kastendiek, J. 1982. Competitor-mediated coexistence: interactions among three species of benthic macroalgae. Jour- nal of Experimental Marine Biology and Ecology 62:201- 210.

Kendall, M. A., R. S. Bowman, P Williamson, and J. R. Lewis. 1985. Annual variation in the recruitment of Sem- ibalanus balanoides on the North Yorkshire Coast 1969- 1981. Journal of the Marine Biology Association of the United Kingdom 65:1009-1030.

Keough, M. J. 1984a. Effects of patch size on the abundance of sessile marine organisms. Ecology 65:423-437.

. 1984b. Dynamics of the epifauna of the bivalve Pinna bicolor: interactions among recruitment, predation, and competition. Ecology 65:677-688.

Leigh, E. G., R. T. Paine, J. R Quinn, and T. H. Suchanek. 1987. Wave energy and intertidal productivity. Proceedings of the National Academy of Sciences (USA) 84:1314- 1318.

Lively, C. M., and P. T. Raimondi. 1987. Desiccation, predation, and mussel-barnacle interactions in the northern Gulf of California. Oecologia 74:304-309.

Lubchenco, J. 1980. Algal zonation in the New England




rocky intertidal community: an experimental analysis. Ecology 61:333-344.

. 1986. Relative importance of competition and predation: early colonization by seaweeds in New England. Page 655 in J. Diamond and T. J. Case, editors. Community ecology. Harper and Row, New York, New York, USA.

Malthus, T. R. 1798. An essay on the principle of population. Reprinted by MacMillan, New York, New York, USA.

Martel, A., K. Beckenback, C. Robles, and M. J. Smith. 1993. Identifying early post-metamorphic stages of marine mussels (Mytilus spp.): combining shell morphology and ge- nomic DNA. Bulletin of the Canadian Society of Zoologists 26:82.

McDonald, J. H., and R. K. Koehn. 1988. The mussels My- tilus galloprovincialis and M. trossulus on the Pacific Coast of North America. Marine Biology 99:111-118.

McDonald, J. H., R. Seed, and R. K. Koehn. 1991. Allo- zymes and morphometric characters of three species of My- tilus in the Northern and Southern hemispheres. Marine Biology 111:323-333.

Menge, B. A. 1978a. Predation intensity in a rocky intertidal community: relation between predator foraging activity and environmental harshness. Oecologia 34:1-16.

. 1978b. Predation intensity in a rocky intertidal community: effect of algal canopy, wave action, and desiccation on predator feeding rates. Oecologia 34:17-35.

. 1992. Community regulation: under what conditions are bottom-up factors important on rocky shores. Ecology 73:756-765.

. 1995. Indirect effects in marine rocky intertidal interaction webs: patterns and importance. Ecological Mono- graphs 65:21-74.

Menge, B. A., E. L. Berlow, C. A. 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 J. Lubchenco. 1981. Community organization in temperate and tropical rocky intertidal habitats: prey refuges in relation to consumer pressure gradients. Ecological Monographs 51:429-450.

Menge, B. A., and J. P Sutherland. 1976. Species diversity gradients: synthesis of the roles of predation, competition, and temporal heterogeneity. American Naturalist 10:351- 369.

Murray, S. N., M. M. Littler, and I. A. Abbot. 1980. Bio- geography of the California marine algae with emphasis on the Southern California Islands. Pages 325-339 in D. M. Power, editor. The California islands: proceedings of a mul- tidisciplinary symposium. Santa Barbara Museum of Nat- ural History, Santa Barbara, California, USA.

Muus, B. J. 1968. A field method for measuring "exposure" by means of plaster balls. Sarsia 34:61-68.

Paine, R. T. 1966. Food web complexity and species diversity. American Naturalist 100:65-76.

. 1974. Intertidal community structure: experimental studies on the relationship between a dominant competitor and its principal predator. Oecologia 15:93-120.

. 1976. Size limited predation: an observational and experimental approach with the Mytilus-Pisaster interaction. Ecology 57:858-873.

. 1991. Between Scylla and Charybdis: do some kinds of criticism merit a response? Oikos 62:90-92.

. 1994. Marine rocky shores and community ecology: an experimentalist's perspective. Excellence in Ecology Se- ries. Volume 4. Ecology Institute, publishers, Olendorf, Germany.

Paine, R. T., J. C. Castillo, and J. Cancino. 1985. Perturbation and recovery patterns of starfish-dominated intertidal assemblages in Chile, New Zealand, and Washington State. American Naturalist 125:679-691.

Paine, R. T., and S. A. Levin. 1981. Intertidal landscapes: disturbance and the dynamics of pattern. Ecological Mono- graphs 51:145-178.

Park, T. 1962. Beetles, competition and populations. Science 138:1369-1375.

Parrish, R. H., C. S. Nelson, and A. Bakun. 1981. Transport mechanisms and reproductive success of fishes in the Cal- ifornia Current. Biological Oceanography 1: 175-203.

Pennington, J. T. 1985. The ecology of fertilization of echi- noid eggs: the consequences of sperm dilution, adult ag- gregation, and synchronous spawning. Biological Bulletin 169:417-430.

Petraitis, P. S. 1974. Settlement patterns of Mytilus edulis and Mytilus californianus and their effects on the distribution of adult populations. Thesis. San Diego State Uni- versity, San Diego, California, USA.

1978. Distributional patterns of juvenile Mytilus edulis and Mytilus californianus. Veliger 21:288-292.

Raimondi, P. T. 1990. Patterns, mechanisms and consequences of variability in settlement and recruitment of an intertidal barnacle. Ecological Monographs 60:283-309.

Robles, C. D. 1987. Predator foraging characteristics and prey population structure on a sheltered shore. Ecology 68: 1502-1514.

Robles, C. D., and J. D. Cubit. 1981. Influence of biotic factors in an upper intertidal community: effects of Diptera larvae grazing on algae. Ecology 62:1536-1547.

Robles, C. D., and J. Robb. 1993. Varied carnivore effects and the prevalence of intertidal algal turfs. Journal of Ex- perimental Marine Biology and Ecology 166:65-91.

Robles, C. D., R. Sherwood-Stephens, and M. Alvarado. 1995. Responses of a key intertidal predator to varying recruitment of its prey. Ecology 76:565-579.

Robles, C. D., D. A. Sweetnam, and J. Eminike. 1990. Lob- ster predation on mussels: shore-level differences in prey vulnerability and predator preference. Ecology 71:1564- 1577.

Roughgarden, J., S. Gaines, and H. Possingham. 1988. Re- cruitment dynamics in complex life cycles. Science 241: 1460-1466.

Schonbeck, M. W., and T. A. Norton. 1980. Factors con- trolling the lower limits of fucoid algae on the shore. Jour- nal of Experimental Marine Biology and Ecology 43:131- 150.

Seapy, R. R., and M. M. Littler. 1982. Population and species diversity fluctuations in a rocky intertidal community relative to severe aerial exposure and sediment burial. Marine Biology 71:87-96.

Sebens, K. P 1987. The ecology of indeterminate growth in animals. Annual Review of Ecology and Systematics 18: 371-408.

Seed, R. 1979. Variation in shell-flesh relationships of My- tilus. The value of sea mussels as prey. Veliger 22:219- 221.

Sherwood-Stephens, R. M. 1993. Environmental correlates of shell growth of Mytilus californianus Conrad 1837. The- sis. California State University at Los Angeles, Los An- geles, California, USA.

Sokal, R. R., and F J. Rohlf. 1981. Biometry. W. H. Freeman. New York, New York, USA.

Sousa, W. P. 1979a. Experimental investigations of disturbance and ecological succession in a rocky intertidal algal community. Ecological Monographs 49:227-254.

1979b. Disturbance in marine intertidal boulder fields: the non-equilibrium maintenance of species diversity. Ecology 60:1225-1239.

. 1984. The role of disturbance in natural communities. Annual Review of Ecology and Systematics 15:353- 391.

Stephenson, T. A., and A. Stephenson. 1972. Life between




tide marks on rocky shores. W. H. Freeman, San Francisco, California, USA.

Stewart, J. G. 1982. Anchor species and epiphytes in intertidal algal turf. Pacific Science 36:45-59.

Sutherland, J. P. 1987. Recruitment limitation in a tropical intertidal barnacle Tetraclita panamensis (Pilsbury) on the Pacific coast of Costa Rica. Journal of Experimental Marine Biology and Ecology 113:267-282.

1990. Recruitment regulates demographic variation of a tropical intertidal barnacle. Ecology 71:955-972.

Sutherland, J. P., and S. Ortega. 1986. Competition condi- tional on recruitment and temporary escape from predators on a tropical rocky shore. Journal of Experimental Marine Biology and Ecology 95: 155-166.

Underwood, A. J. 1981. Techniques of Analysis of Variance in experimental marine biology and ecology. Annual Re- view of Oceanography and Marine Biology 19:513-605.

Underwood, A. J., and E. J. Denley. 1984. Paradigms, explanations, and generalizations in models of the structure of intertidal communities on rocky shores. Pages 151-197

in D. Strong, D. Simberloff, L. G. Abele, and A. B. Thistle, editors. Ecological communities: conceptual issues and the evidence. Princeton University Press, Princeton, New Jer- sey, USA.

Underwood, A. J., E. J. Denley, and M. J. Moran. 1983. Experimental analysis of the structure and dynamics of mid-shore rocky intertidal communities in New South Wales. Oecologia 56:202-219.

Underwood, A. J. and P G. Fairweather. 1988. Supply-side ecology and benthic marine assemblages. Trends in Ecol- ogy and Evolution 4:16-20.

Underwood, A. J., and P. S. Petraitis. 1993. Structure of intertidal assemblages in different locations: how can local processes be compared? Pages 39-51 in R. E. Ricklefs and D. Schluter, editors. Species diversity in ecological communities: historical and geographical perspectives. Uni- versity of Chicago Press, Chicago, Illinois, USA.

United States Army Corps of Engineers. 1986. Southern Cal- ifornia Coastal Data Summary. Report number CCSTWS 86-1. February 1986.



changing recruitment in constant species assemblages: implications for predation theory in...

Documents