The effect of sea urchins as biogenic structures on the local abundance of a temperate reef fish

Transcription

1 Oecologia (2002) 131: DOI /s POPULATION ECOLOGY Kristine B. Hartney Kirsten A. Grorud The effect of sea urchins as biogenic structures on the local abundance of a temperate reef fish Received: 27 November 2000 / Accepted: 12 February 2002 / Published online: 13 April 2002 Springer-Verlag 2002 Abstract Although widely distributed on rocky reefs at Santa Catalina Island (southern California, USA), the small, substrate-oriented fish, Lythrypnus dalli, often reaches its highest densities in areas of extreme vertical relief where the sea urchin, Centrostephanus coronatus, is common. Through surveys and manipulative field experiments, we examined the extent to which the availability of C. coronatus influenced the distribution and abundance of L. dalli. Adult L. dalli were strongly associated with C. coronatus, but not with two shorter-spined urchins, Strongylocentrotus franciscanus and S. purpuratus. Variation in fish abundance among C. coronatus was affected by three features: the amount of space between the body of the urchin and its rocky shelter, the location of occupied shelters, and urchin size. Field manipulations of C. coronatus density indicated that spatial patterns of L. dalli abundance are causally linked to the presence of urchins. Urchins also had a significant effect on recruitment, migration, and survival of L. dalli. These processes were greatly enhanced where urchins were present. Structural substitutions (models) produced about half the effect of live urchins indicating that biological, as well as structural, attributes of urchins may be important to fish. Our results suggest that habitat selection at the time of settlement, small-scale migration of fish after settlement, and differential survival relative to the presence of sea urchins contribute to the formation and maintenance of the association between L. dalli and C. coronatus. By altering the availability of refuges, sea urchins appear to have a direct and positive effect on the local abundance of this temperate reef fish. K.B. Hartney ( ) Department of Biological Sciences, California State Polytechnic University, Pomona, 3801 W Temple Avenue, Pomona, CA 91768, USA kbhartney@csupomona.edu Tel.: K.A. Grorud Department of Marine Biology and Fisheries, Rosenstiel School of Marine and Atmospheric Science, University of Miami, Miami, USA Keywords Centrostephanus coronatus Goby Lythrypnus dalli Positive interactions Shelter Introduction Biotic interactions play a major role in determining the distribution and abundance of species. However, compared to negative forces such as competition and predation (reviewed in Connell 1983; Schoener 1983; Sih et al. 1985), relatively little emphasis has been placed on experimentally evaluating the conditions under which positive interactions are likely to be important (e.g. Witman 1987; Carr 1989; Bertness et al. 1999; Stachowicz and Hay 1999). The significance of these positive forces in affecting the abundance of others may hinge on the presence of organisms that either create or alter habitat (Bell et al. 1991; Jones et al. 1997). Habitat structure is known to influence the abundance of fishes on temperate reefs at various spatial scales (Jones 1988; Holbrook et al. 1990). How these fishhabitat associations are established is poorly understood, but they appear to be related to habitat elements that may attract larvae (Breitburg 1991), promote settlement (Carr 1991, 1994; Jones 1984b; Levin 1991), motivate relocation after settlement (Jones 1984a; Levin 1994; Ault and Johnson 1998), and/or supply resources necessary to grow and evade predators (Ebeling and Laur 1985; Schmitt and Holbrook 1985; Behrents 1987; Connell and Jones 1991). On temperate reefs, macroalgae are a major source of structural complexity, and their presence and composition may have substantial effects on the population dynamics of reef fishes (Choat and Ayling 1987; Carr 1989, 1991; Schmitt and Holbrook 1990). These algal assemblages however, can be strongly affected by sea urchins whose grazing activities can significantly reduce the local abundance of algae and other epibionts (Vance 1979). Yet, as demonstrated on coral reefs, urchins may also serve as physical structures themselves providing a unique but spatially variable habitat for smaller fishes

2 (Randall et al. 1964; Magnus 1967; Teytaud 1971; Sakashita 1992). In addition to macroalgae, many invertebrates such as corals (Kuwamura et al. 1994; Munday et al. 1997), sea anemones (Abel 1960, Elliott 1992), sponges (Bohlke and Robins 1969), shrimps (Grossman 1979; Karplus et al. 1981), and sea urchins (Teytaud 1971) act as biogenic sources of habitat structure to reef fishes. Because several pathways of interaction are possible (including positive and negative effects), it is largely unknown if these other forms of biogenic habitat actually affect the population dynamics and distribution of fishes (but see Karplus et al. 1981; Kuwamura et al. 1994; Munday et al. 1997). This is particularly true for sea urchins whose negative effects as a destructive grazer have typically been emphasized (reviewed in Laurence 1975; Dayton 1985; Schiel and Foster 1986; Jones and Andrew 1990; but see Vance 1978; Coyer et al. 1993). Here we explore the possible positive effects the sea urchin, Centrostephanus coronatus, may have on the local abundance of the bluebanded goby, Lythrypnus dalli. Specifically, we address the following questions: (1) does a causal relationship exist between the distribution and abundance of urchins and fish? (2) what features affect the suitability of sea urchins as habitats for fish? and (3) do urchins affect recruitment, migration, and/or survival of fish? Materials and methods Study system C. coronatus (Diadematidae) inhabit reefs that offer crevices adequate to protect them from diurnal fish predators. Active only at night, C. coronatus do not forage far from their daytime shelters in the rock (<1 m) and return before dawn to the same shelter (Nelson and Vance 1979). Intermediate in body size relative to two other reef dwelling urchins that also occur in southern California, Strongylocentrotus purpuratus and S. franciscanus, the spines of C. coronatus are at least 2 times as long (to 10 cm) as these species. When sheltering, these sharp spines effectively deter the major urchin predator in these waters, the labrid, Semicossyphus pulcher (Nelson and Vance 1979). The small [<47 mm standard length (SL)], locally abundant bluebanded goby typically occupies physically complex, rocky habitats. In areas with low rugosity, gobies are often observed among the spines or within shelters of C. coronatus (Behrents 1983; Steele 1997b). Resting on the rocky substratum during the day, but frequently entering the water column to feed on zooplankton (Hartney 1989), this small, brilliantly colored fish is at particular risk to the piscivorous serranid, Paralabrax clathratus (Steele 1997b, 1998; Forrester and Steele 2000). When threatened, refuge is immediately sought within the nearest shelter, while small cracks in rocks and empty invertebrate shells are used as nest sites and nocturnal resting places. Study areas Field studies were conducted from January 1997 through January 1999 at two sites on the lee (north) side of Santa Catalina Island (33 o 27 N, 118 o 29 W) located approximately 37 km off the coast of southern California. Most studies were conducted on the eastern side of Bird Rock, an islet located 0.6 km from the main island. At this site, a continuous sheet of rock drops precipitously from 3 to 30 m forming a vertical rock wall before giving way to sand. A 507 sparse kelp bed and epibionts (turf) persist along the upper edge of the wall. However, below approximately 4 m, surfaces are generally barren or encrusted with coralline algae, and numerous shallow depressions (<20 cm deep) are typically occupied by the sea urchin, C. coronatus (for map, see Lissner 1980). The second site, Chalk Cliffs, is located approximately 0.8 km south of Bird Rock and lies adjacent to the main island. Here six artificial reefs located on a sand flat at depths of 5 8 m were utilized to examine the influence of urchins on survival of adult fish. Each reef consists of 1 m 3 of concrete into which PVC pipes of varying diameters have been inserted (for map and detailed reef descriptions, see Zahary and Hartman 1985). Patterns of abundance and distribution The abundance of L. dalli is spatially and seasonally variable sometimes reaching densities in excess of 100 fish/m 2 (Zahary and Hartman 1985; Behrents 1983; Steele 1997a). Most individuals do not live for more than 1 year and populations are replenished annually by larval recruitment (Behrents 1983). Settlement of larvae at standard lengths of 9 11 mm begins in late May, peaks in July and August, and ends by January. Growth is rapid and maturity (>20 mm SL) may be reached within 1 month of settlement (Behrents 1983; Steele 1997b). Visual surveys were conducted in January 1997 to quantify the association between L. dalli and C. coronatus. L. dalli and C. coronatus densities were estimated during daylight hours within 100 randomly placed 1 m 2 quadrats at depths of m on the wall at Bird Rock. The number of fish associated with each C. coronatus was recorded separately from fish observed on open rock surfaces (i.e. not associated with an urchin or within the confines of its shelter). All fish observed at this time were adults (>20 mm SL). C. coronatus were recorded as juveniles (test diameter <20 mm) or adults ( 20 mm). The relationship between density of fish and adult urchins was explored using simple linear correlation (Pearson r). To explain variation in fish abundance among C. coronatus, 106 adult C. coronatus were selected at random along the wall at Bird Rock in May 1998 in water depths ranging from 9 20 m. We counted the number of adult L. dalli associated with each urchin and measured test diameter, test volume (assuming a hemispherical shape), shelter volume (assuming a shape described by one half of an ellipse that incorporated measures of the two longest perpendicular axes of the opening to the shelter and greatest depth of the shelter), inclination of the shelter relative to a flat plane, distance to the nearest adult urchin neighbor, number of adult urchins within a 0.5 m radius, proportion of the substratum overgrown by turf (assessed by examining the substrate below 5 points positioned at 10 cm intervals along each of four radii emanating from the focal urchin, n=20), and water depth. Relationships between habitat features and fish abundance were evaluated with linear multiple regression using a forward stepwise procedure. The dependent variable, L. dalli abundance, was square root transformed (Y+0.5) to normalize residuals. Species of sea urchins other than C. coronatus were rare on the vertical wall at Bird Rock; however C. coronatus, Strongylocentrotus franciscanus, and S. purpuratus co-occur in shallower depths (<3 m) where the substrate flattens. To assess patterns of fish occupancy among different species of urchins, we counted the number of fish associated with each species as individuals were randomly encountered in two areas approximating 30 m 2. Effects of urchin availability on fish abundance The co-occurrence of L. dalli and C. coronatus could reflect either a common response by both organisms to specific topographic features of the reef or a causal relationship. If the availability of C. coronatus influences the abundance of L. dalli, then fish abundance should respond directly to changes in C. coronatus abundance.

3 508 Removal experiment In March 1997, permanent quadrats were constructed at least 5 m apart along the wall of Bird Rock at depths of m. Each quadrat measured m (0.5 m 2 ) and contained at least two but usually three C. coronatus. After an initial visual survey of the number of fish and urchins in each quadrat, quadrats were assigned to one of three treatments: (1) unmanipulated control (urchins left undisturbed), (2) disturbance control (urchin removal with immediate replacement), and (3) removal (all urchins permanently removed). Ten quadrats were assigned to each treatment, and quadrats were matched as closely as possible among treatments by water depth and number of urchins initially present. Treatments were interspersed so replicates were not adjacent. After manipulations were complete, all gobies and sea urchins within each quadrat were visually censused on a daily basis for one week and thereafter once a week for 4 weeks. All fish observed were adults (>20 mm SL). Since mean initial densities of L. dalli were not equal among all quadrats, we standardized results for each sampling period by determining the proportional change in density relative to initial density at the start of the experiment prior to any manipulations (i.e. proportional change = [observed density density at the start of the experiment] / density at the start of the experiment). Mean proportional change in density over the sampling periods was then compared among treatments using a repeated measures analysis of variance test followed by post hoc comparisons between treatments using the Tukey Honest Significant Difference (HSD) test. Addition experiment We simulated the addition of urchins to unoccupied space by eliminating all C. coronatus from 20 newly constructed quadrats (0.5 m 2 ) in March These quadrats were kept clear of urchins for the next 2 months, giving sessile invertebrates and algae time to colonize the sites. In May 1997, the number of fish and urchins present were censused prior to assignment of quadrats to treatments. C. coronatus were reintroduced into half of the cleared quadrats (n=10) at a level equal to their pre-removal density (addition treatment) and the remaining quadrats (n=10) were maintained as permanent removal quadrats (removal treatment). All quadrats were initially covered with 1 cm mesh hardware cloth to allow transplanted urchins to adjust to their new shelters (addition treatment) and to prevent movement of urchins into empty quadrats (removal treatment). Caging material was removed from all quadrats at the same time (5 days post-manipulation) and thereafter the number of fish and urchins were censused daily for 5 consecutive days and then once a week for 3 weeks. When settlement began during the final two sampling periods, newly recruited fish were recorded separately from adults. Data for adults and recruits were analyzed separately with a repeated measures ANOVA using mean fish density as the dependent variable. Effects of urchins on recruitment and post-recruitment processes Demographic processes contributing to positive associations between fish and sea urchins were evaluated experimentally. First, we examined the influence of C. coronatus and its physical structure on fish recruitment and migration on natural reefs (Bird Rock) during peak settlement months of July and August At the end of the settlement period (December 1997 and January 1998), we conducted an experiment to test the effects of C. coronatus on survival of adult fish marked and stocked to artificial reefs (Chalk Cliffs). Recruitment and migration To test the effects of structural features specific to sea urchins on recruitment and migration of L. dalli, densities of C. coronatus were manipulated and models were substituted for live urchins at Bird Rock. Models were constructed of 10-cm bamboo skewers and 5-cm toothpicks inserted into a hemispherical body of cement (Quikrete). Depending on test diameter (either 4 or 6 cm), each model possessed either 10 spines (5 skewers and 5 toothpicks) or 15 spines (10 skewers and 5 toothpicks) respectively. All models were spray painted with black enamel (Krylon) to match the color of adult urchins. The flat base of each model was fixed in place with Z-SPAR marine epoxy. Thirty experimental quadrats (0.5 m 2 ) established at Bird Rock were assigned to one of three treatments: (1) live structure (undisturbed quadrats each containing 2 or 3 urchins), (2) artificial structure (substitution of models for live urchins), and (3) no structure (permanent removal of all sea urchins). Quadrats were spaced at least 5 m apart and interspersed so that replicates of the same treatment were not adjacent. At the start of the experiment, all fish were removed from quadrats to minimize the potential effects of prior residents (conspecifics) on recruitment (Steele 1997a). Thereafter, fish were collected from each quadrat using hand nets after application of the anesthetic quinaldine at weekly intervals for 3 weeks and SL measured. Given the short time interval between collections (1 week), fish <14 mm SL were classified as newly recruited fish while those 14 mm were considered to have migrated into quadrats from surrounding areas. Data were first analyzed with MANOVA using the mean density of immigrants and recruits as dependent variables. This was followed by two one-way ANOVAs to test for effects on each variable separately. Data were square root transformed (Y+0.5) prior to all analyses. Adult survival Six artificial reefs located at Chalk Cliffs were used to test the effects of C. coronatus on survival of adult fish. Reefs were 1 m 3 in size and separated by distances >11 m. Prior to the start of the experiment (December 1997), all reefs were cleared of resident fishes and sea urchins. Reefs were then randomly assigned as either an urchin addition (36 C. coronatus added) or control (no urchins added) treatment, with the condition that replicate reefs be nonadjacent. Adult fish (>20 mm) were collected from nearby reefs and marked subcutaneously with a small amount of India ink. After a recovery period of 48 h, marked fish were transplanted in groups of 40 to each reef. After 12 days, all urchins were removed and fish collected. The experiment was repeated twice (14 26 December 1997; 28 December January 1998) using the same protocols each time except that the assignment of reefs to treatments was reversed during the second trial and fish marks were placed on the opposite side of the fish. Because the successful movement of fish among reefs separated by >11 m is negligible (Steele 1998; Forrester and Steele 2000), the loss of marked fish was attributed to mortality. A two-way ANOVA was used to test for effects of urchins and sampling period on the proportion of marked fish that survived to the end of each trial. To assess handling effects and tag loss, 50 marked fish were maintained in an aquarium over the course of the field trials (3.5 weeks). Forty-five (90%) survived and none lost their mark. Thus, we interpret all losses as representing natural mortality. Results Patterns of fish abundance relative to sea urchins Densities of L. dalli and C. coronatus were positively correlated (Pearson r=0.48, P<0.01; Fig. 1). In all, 90% of 2,360 adult L. dalli were either physically associated with C. coronatus or located within its shelter. Large urchins served as hosts more often than small ones: 93% of adult C. coronatus (n=566) harbored fish, whereas, only 6% of juvenile C. coronatus (n=70) were similarly occupied.

4 509 Fig. 1 Relationship between the density of Lythrypnus dalli and density of adult Centrostephanus coronatus based on surveys of m 2 quadrats. The density (mean±1 SE) of L. dalli was 23.6±1.7 fish/m 2 and all fish encountered were adults (>20 mm). The density of adult C. coronatus was 5.7±0.3 urchins/m 2 Table 1 A summary of the stepwise regression model predicting abundance of Lythrypnus dalli based on a variety of habitat parameters. Fish abundance was square root transformed (Y+0.5) for analysis. Variables are listed in order of stepwise inclusion into the model. Independent variables Multiple R 2 Regression coefficient Urchin volume/shelter volume * Shelter location (angle) ** Urchin size (diameter) ** Distance to nearest urchin (cm) Water depth (m) Turf cover (%) *P<0.05, **P<0.001 The number of fish associated with individual C. coronatus on the rock wall of Bird Rock varied widely from 0 to 16 fish per urchin (Fig. 2a) indicating that the suitability of urchins as habitats may vary. Six habitat variables explained 45% of total variation (Table 1). Most of the variation was explained by three factors that described the space available for fish to seek refuge; overall, gobies tended to be least abundant where urchins were tightly bound within shelters, shelters were on top of (rather than under) ledges, and/or urchins were small. In the presence of other sea urchins on the shallow flats at Bird Rock, L. dalli were most commonly observed in association with C. coronatus (Fig. 2b). Of the 72 C. coronatus, 59 S. purpuratus, and 7 S. franciscanus randomly surveyed, 93% of 147 L. dalli counted were associated with the long-spined species, C. coronatus. As observed previously, the number of L. dalli per individual urchin varied widely, but shorter spined urchins (Stronglyocentrotus sp.) rarely harbored more than one fish, and more commonly none. Fig. 2 Variation in the number of L. dalli associated with three species of sea urchins (C. coronatus, S. purpuratus, and S. franciscanus) expressed as a percentage of all urchins observed on a the rock wall at Bird Rock and b shallow flats at Bird Rock Responses of fish to urchin availability Reductions in the density of C. coronatus had a significant and immediate negative effect on the local distribution and abundance of L. dalli at Bird Rock (two-way ANOVA: urchin density F 2,27 =42.7, P<0.001; sampling period F 9,243 =0.51, P=0.86; interaction F 18,243 =1.3, P=0.19; Fig. 3). Overall, the mean proportional change in fish density was significantly greater in quadrats from which all urchins were permanently removed relative to both unmanipulated control (Tukey HSD, P<0.0001) and disturbance control (Tukey HSD, P<0.0001) treatments. There was no significant difference between unmanipulated and disturbance controls (Tukey HSD, P=0.55). Similarly, the reintroduction of C. coronatus to previously empty quadrats resulted in a significant increase in the density of adult fish compared to those quadrats that continued to harbor no urchins (two-way ANOVA: urchin density F 1,18 =11.4; P=0.003; sampling period F 7,126 =1.5, P=0.17; interaction F 7,126 =1.5, P=0.19;

5 510 Fig. 3 Proportional change (mean±1 SE) in density of adult L. dalli (fish/0.5 m 2 ) following the temporary (disturbance control) and permanent removal of C. coronatus from reef quadrats (n=10). Densities (mean±1 SE) of L. dalli at the start of the experiment (prior to manipulations) were 12.1±2.4 (unmanipulated control), 9.5±1.1 (disturbance control), and 4.6±0.8 (removal) Fig. 5 Recruitment and migration of L. dalli to quadrats (0.5 m 2 ) of varying habitat type (substrata with urchins, substrata with models, substrata with neither urchins or models) (n=30). Data are untransformed means±1 SE recruits were more commonly observed (100% versus 40%) and more numerous (2.8±0.5 versus 0.5±0.8 fish/m 2 ) in quadrats with urchins relative to those with no urchins. Thus, the presence of urchins and/or adult fish (as migrants into quadrats with urchins) appeared to influence patterns of recruitment at the time of settlement or shortly thereafter. Effects of urchins on recruitment and post-recruitment processes Fig. 4 Adult density (mean±1 SE) (solid symbols) and recruit density (open symbols) of L. dalli (fish/0.5 m 2 ) following the addition of C. coronatus to previously urchin-free reef quadrats (n=10). Densities (mean±1 SE) of L. dalli at the start of the experiment (prior to manipulations) were 0.4±0.2 (urchin addition) and 0.1±0.1 (urchin removal) Fig. 4). The highest mean density (±1 SE) of adult fish (2.8±0.8 fish/0.5 m 2 ) was reached 23 days after the manipulation occurred (30 May) and was equivalent to nearby undisturbed quadrats (2.8±0.5 fish/0.5 m 2 ) censused as part of another study. The rapid colonization of quadrats occupied by sea urchins to densities normal for this time of the year indicates that adult fish are capable of and do migrate to some degree along the rock wall. Settlement began during the final two survey periods of this experiment, varied significantly between treatments (two-way ANOVA; F 1,18 =9.7, P=0.006), and intensified over the two sampling periods (F 1,18 =12.9, P=0.002) resulting in a significant interaction term (F 1,18 =6.9, P=0.02). On the final day of sampling, newly settled On natural rock reefs, both new recruit and immigrant density were significantly affected by manipulation of structural features (Wilk s Lambda F 4,160 =27.6, P<0.0001; Fig. 5). Colonization, whether by recruitment or migration, was greatest in quadrats inhabited by sea urchins and least in quadrats containing neither urchins or models (two-way ANOVA for recruits: structure F 2,81 =31.7, P<0.001; sampling period F 2,81 =1.4, P=0.24; interaction F 4,81 =1.0, P=0.42; for migrants: structure F 2,81 =69.9, P<0.001; sampling period F 2,81 =1.4, P=0.25; interaction F 4,81 =0.6, P=0.70). Substitution of models for live urchins led to an intermediate effect. Although densities of recruits and migrants associated with models were significantly greater than those of quadrats lacking urchins (Tukey HSD, P<0.001), models were not perfect substitutes for live urchins (Tukey HSD, P<0.001). This suggests that in addition to physical structure, C. coronatus possess other qualities important to maintenance of the fish-urchin association. The density of small and large fish were also significantly and positively correlated (Pearson r=0.69, P<0.001) indicating that settlement and migration were greatest to sites with urchins. Because L. dalli preferentially settle to microhabitats where conspecifics are present (Steele 1997a), the tendency of recruits to occupy sea urchins could have been due to the previous migration of older fish. Survival of adult fish transplanted to artificial reefs was also affected by the presence of sea urchins. Persis-

6 tence of marked adult L. dalli was greater on reefs occupied by urchins (of the 40 marked fish, 12.3±1.5 were recovered during the first trial of the experiment and 19.0±2.6 during the second) compared to those with no urchins (of the 40 marked fish, 10.7±0.9 were recovered during the first trial and 8.0±2.0 during the second). Overall, about 40% of all marked fish transplanted to urchin occupied reefs remained by the end of the experimental period, while 26% persisted on reefs without urchins. The overall effect of urchins on survival of marked L. dalli was significant (ANOVA; F 1,8 =11.6, P=0.009), although effects were inconsistent between the two sampling periods resulting in a significant interaction term (F 1,8 =6.3, P=0.04). The main effect of sampling period was not significant (F 1,8 =1.2, P=0.31). Discussion The strength of fish-habitat associations may vary as a function of a species flexibility in resource use (Holbrook et al. 1990) and biotic stress (Bertness and Calloway 1994). Our findings indicate that the local abundance of a small temperate reef fish, L. dalli, can be affected by the presence of C. coronatus, and our manipulative experiments confirm that the relationship between these species is causal. It is likely, however, that the importance of this fish-urchin association may be site dependent, varying with substratum type, host density, and predation risk (e.g. Elliott 1992). At Bird Rock, where continuous sheets of solid rock limit the opportunity for fish to hide under rocks or in crevices, C. coronatus appear to serve as high quality shelters that elsewhere are supplied by rocky structures. The local abundances of tropical fishes are known to be influenced by a broad range of habitat variables and the factors contributing to microhabitat quality can be quite specific (Ault and Johnson 1998; Holbrook et al. 2000). Similarly, we found that the use of C. coronatus as a habitat for L. dalli is affected by several attributes of the sea urchin relative to the habitat it occupies. Variation in occupancy rates of individual urchins by fish was affected by three features operating together the amount of space between the body of the urchin and its rocky shelter, the location of occupied shelters, and urchin size. The greatest number of fish occurred in association with adult urchins that occupied large/deep shelters flush with the vertical sides of the wall or under slightly protruding ledges. In these situations, urchin spines form a network of interstices within which the fish moved about and sheltered. Fish were rarely found in association with juvenile urchins which tend to fit snuggly within shallow cracks and holes that most closely match their body size (Lissner 1978). The attraction of bluebanded gobies to sea urchins may also vary with the species of urchin present. Elsewhere, L. dalli have been observed to associate with several short-spined species including S. franciscanus (Feder et al. 1974) and Echinometra vanbrundti (Thompson 511 et al. 1979). Whether other urchin species were present at these sites is unknown. However, at Bird Rock where three species co-occur (C. coronatus, S. franciscanus, and S. purpuratus), gobies were more abundant among the spines of C. coronatus relative to others. While more than one factor may be operating, the most obvious difference among these species is spine length, a structural feature that could play a role in attracting and/or retaining fish. Experiments using models to represent the architectural features of biogenic refuges have demonstrated that physical structure is readily utilized by organisms for shelter (Magnus 1967; Woodin 1978). However, our models, while sustaining and attracting fish to a greater extent than featureless habitats, were not perfect substitutes for C. coronatus. This suggests that either our models failed to adequately simulate the physical structure of C. coronatus (e.g. by having fewer spines) or live urchins possess other attributes, in addition to structure, that are important to the association. As inanimate objects, our models were prone to fouling and could not behave as real urchins do, and these behaviors may be linked to the quality of the urchin as an effective refuge for fish. During the day, long spines protrude from and defend the entrance to an urchin s shelter. Normally quiescent, spines are set in motion by mechanical disturbance (Nelson and Vance 1979) and/or changes in light level (Hartney, unpublished data). In response to a disturbance, spines converge in the direction of the stimulus while others spread against the sides of the shelter firmly bracing the animal in its shelter. Highly effective in deterring urchin predators, and presumably potential predators of L. dalli (Paralabrax clathratus) as well, the response of spines to disturbance may be beneficial to both species. As urchins prepare to defend against disturbance, spine movements may cue fish to impending threats and provide time for gobies to seek shelter in more protected sites deep within or behind urchins. At Santa Catalina Island, diurnal fish predators are a significant source of mortality to C. coronatus (Nelson and Vance 1979) and L. dalli (Steele 1997b, 1998) and are suspected to drive species specific foraging and sheltering behaviors. Minimizing temporal overlap with their potential predators, C. coronatus forage in the open only at night and shelter by day in rocky shelters. In contrast, L. dalli are active during the day but spend much of that time either hiding within crevices or perched on the rocky substratum (Steele 1998). This behavior is interrupted by short bursts of agonistic behavior directed at conspecifics and brief foraging bouts into the water column. Although qualitatively these behaviors do not appear to be modified substantially by the presence of urchins, perching behind or in the vicinity of urchin spines may lower overall predation risk. At night, gobies abandon their urchin host (Hartney, unpublished data) and presumably shelter deep within the reef itself. Whether gobies change hosts from one day to the next is not known, but the regularity with which individual urchins return to the same shelters (Nelson and Vance

7 ) is probably sufficient to allow the association to reform on a daily basis. Although settlement of gobies to urchins may contribute initially to the association between the two species, post-settlement processes are likely to be important in reinforcing that relationship. Many temperate reef fishes are capable of selecting specific microhabitats at the time of settlement (e.g. Jones 1984b; Breitburg 1991; Levin 1991; Carr 1994) or may relocate to them later on (Jones 1984a; Levin 1994; Ault and Johnson 1998). At either time, preferences for certain habitat features over others may confer a survival advantage to fishes. By supplying refuges, habitat structure may mediate the effects of predation, a significant determinant of reef fish abundance on local scales (Carr and Hixon 1995; Beets 1997; Steele 1999; Anderson 2001). The colonization of areas occupied by sea urchins by both new recruits and immigrants at Bird Rock suggests that gobies exercise some level of choice relative to habitats occupied. The benefit of these choices may be great, as evidenced by the higher persistence of marked fish on urchin inhabited reefs, although the cost of incorrect choices at the time of settlement and migration success rate are not known. However, fish may minimize risk by traveling from one sea urchin to another while negotiating the distances (several meters) they are known to travel on large continuous reefs (Steele 1997b). Fish also appear to be familiar with their local environment. When urchins were disturbed during the course of our manipulations, fish suddenly left without a host did not scatter haphazardly, but instead headed directionally to the protection of the nearest urchin neighbor. Both the availability of urchins as stepping stones and the general familiarity of a fish with its surroundings are probably important factors affecting predation risk and merit further investigation. Experimental studies such as this provide useful insight into the role positive interactions may play in affecting the population dynamics of fishes on local scales. At Bird Rock, the ecological requirements of one species (C. coronatus) are not only coincidental with another (L. dalli), but the urchin also supplies resources not otherwise available in this habitat, a factor that may assume greater importance in areas of low structural heterogeneity and intense predation pressure (Bertness and Calloway 1994). Understanding the multiple roles (both positive and negative) that species play is essential in understanding and predicting the consequences of their loss on the local abundance of other species and the stability of community structure (Bertness and Leonard 1977). Studies documenting the cascading effects of species loss have usually emphasized the interactive effects of negative processes such as predation and competition (e.g. Paine 1974; Estes et al. 1978; Powers et al. 1996). However, as our study illustrates, even minor changes in the presence of a species can have complex consequences. Typically, sea urchins have been portrayed as destructive grazers in temperate reef communities (Laurence 1975; Dayton 1985; Schiel and Foster 1986; Jones and Andrew 1990). In contrast, our study demonstrates their direct positive effects on the local abundance of a reef fish through provision of habitat structure. Acknowledgements We thank J. Baldelli, D. Robertson, J. Wible and numerous others for their underwater assistance, as well as the staff of the Wrigley Institute for Environmental Studies for logistical support. Comments from R. Bray, S. Sponaugle, M. Steele, C. Osenberg, and an anonymous reviewer on drafts of this manuscript are appreciated. This research was supported by funds from the Ford Foundation (to K.A.G.) and Occidental College. This is Contribution No. 213 of the Wrigley Institute for Environmental Studies. References Abel EF (1960) Liaison facultative d un poisson (Gobius bucchichii Steindachner) et d une anemone (Anemonia sulcata Penn.) en Mediterranee. Vie Milieu 11: Anderson TW (2001) Predator responses, prey refuges, and densitydependent mortality of a marine fish. Ecology 82: Ault TR, Johnson CR (1998) Spatially and temporally predictable fish communities on coral reefs. Ecol Monogr 68:25 50 Beets J (1997) Effects of a predatory fish on the recruitment and abundance of Caribbean coral reef fishes. Mar Ecol Prog Ser 148:11 21 Behrents KC (1983) The comparative ecology and interactions between two sympatric gobies (Lythrypnus dalli and Lythrypnus zebra). PhD Dissertation, University of Southern California, Los Angeles Behrents KC (1987) The influence of shelter availability on recruitment and early juvenile survivorship of Lythrypnus dalli Gilbert (Pisces: Gobiidae). J Exp Mar Biol Ecol 107:45 59 Bell SS, McCoy ED, Mushinsky HR (1991) Habitat structure: the physical arrangement of objects in space. Chapman and Hall, London, UK Bertness MD, Callaway RM (1994) Positive interactions in communities: a post cold war perspective. Trends Ecol Evol 9: Bertness MD, Leonard GH (1997) The role of positive interactions in communities: lessons from intertidal habitats. Ecology 78: Bertness MD, Leonard GH, Levine JM, Schmidt PR, Ingraham AO (1999) Testing the relative contribution of positive and negative interactions in rocky intertidal communities. Ecology 80: Bohlke JC, Robins CR (1969) Western Atlantic sponge-dwelling gobies of the genus Evermannichthys: Their taxonomy, habits, and relationships. Proc Acad Nat Sci Phila 121:1 24 Breitburg DL (1991) Settlement patterns and presettlement behavior of the naked goby, Gobiosoma bosci, a temperate oyster reef fish. Mar Biol 109: Carr MH (1989) Effects of macroalgal assemblages on the recruitment of temperate zone reef fishes. J Exp Mar Biol Ecol 126: Carr MH (1991) Habitat selection and recruitment of an assemblage of temperate zone reef fishes. J Exp Mar Biol Ecol 146: Carr MH (1994) Effects of macroalgal dynamics on recruitment of a temperate reef fish. Ecology 75: Carr MH, Hixon MA (1995) Predation effects on early postsettlement survivorship of coral-reef fishes. Mar Ecol Prog Ser 124:31 42 Choat JH, Ayling AM (1987) The relationship between habitat structure and fish faunas on New Zealand reefs. J Exp Mar Biol Ecol 110: Connell JH (1983) On the prevalence and relative importance of interspecific competition: evidence from field experiments. Am Nat 122:

8 513 Connell SD, Jones GP (1991) The influence of habitat complexity on postrecruitment processes in a temperate reef fish population. J Exp Mar Biol Ecol 151: Coyer JA, Ambrose RE, Engle JM, Carroll JC (1993) Interactions between corals and algae on a temperate zone rocky reef: mediation by sea urchins. J Exp Mar Biol Ecol 167:21 37 Dayton PK (1985) The structure and regulation of some South American kelp communities. Ecol Monogr 55: Ebeling AW, Laur DR (1985) The influence of plant cover on surfperch abundance at an offshore temperate reef. Environ Biol Fish 12: Elliott J (1992) The role of sea anemones as refuges and feeding habitats for the temperate fish Oxylebius pictus. Environ Biol Fish 35: Estes, JA, Smith NS, Palmisano JF (1978) Sea otter predation and community organization in the western Aleutian Islands, Alaska. Ecology 59: Feder HM, Turner CH, Limbaugh C (1974) Observations on fishes associated with kelp beds in southern California. Calif Dept Fish Game Fish Bull 160:1 144 Forrester GE, Steele MA (2000) Variation in the presence and cause of density-dependent mortality in three species of reef fishes. Ecology 81: Grossman GD (1979) Symbiotic burrow-occupying behavior in the bay goby, Lepidogobius lepidus. Calif Fish Game 65: Hartney KB (1989) The foraging ecology of two sympatric gobiid fishes: importance of behavior in prey type selection. Environ Biol Fish 26: Holbrook SJ, Schmitt RJ, Ambrose RF (1990) Biogenic habitat structure and characteristics of temperate reef fish assemblages. Aust J Ecol 15: Holbrook SJ, Forrester GE, Schmitt RJ (2000) Spatial patterns in abundance of a damselfish reflect availability of suitable habitat. Oecologia 122: Jones CG, Lawton JH, Schachak M (1997) Positive and negative effects of organisms as physical ecosystem engineers. Ecology 78: Jones GP (1984a) The influence of habitat and behavioural interactions on the local distribution of the wrasse, Pseudolabrus celidotus. Environ Biol Fish 10:43 58 Jones GP (1984b) Population ecology of the temperate reef fish Pseudolabrus celidotus Bloch and Schneider (Pisces: Labridae). I. Factors influencing recruitment. J Exp Mar Biol Ecol 75: Jones GP (1988) Ecology of rocky reef fish of north-eastern New Zealand: A review. N Z J Mar Freshw Res 22: Jones GP, Andrew NL (1990) Herbivory and patch dynamics on rocky reefs in temperate Australia: the roles of fish and sea urchins. Aust J Ecol 15: Karplus I, Szlep R, Tsurnamal M(1981) Goby-shrimp partner specificity. I. Distribution in the northern Red Sea and partner specificity. J Exp Mar Biol Ecol 51:1 19 Kuwamura T, Yogo Y, Nakashima Y (1994) Population dynamics of goby Paragobiodon echinocephalus and host coral Stylophora pistillata. Mar Ecol Prog Ser 103:17 23 Laurence JM (1975) On the relationship between marine plants and sea urchins. Oceanogr Mar Biol Annu Rev 13: Levin PS (1991) Effects of microhabitat on recruitment variation in a Gulf of Maine reef fish. Mar Ecol Prog Ser 75: Levin PS (1994) Small-scale recruitment variation in a temperate fish: The roles of macrophytes and food supply. Environ Biol Fish 40: Lissner AL (1978) Factors affecting the distribution and abundance of the sea urchin Centrostephanus coronatus Verrill at Santa Catalina Island. PhD Dissertation, University of Southern California, Los Angeles Lissner AL (1980) Some effects of turbulence on the activity of the sea urchin Centrostephanus coronatus Verrill. J Exp Mar Biol Ecol 48: Magnus DBE (1967) Ecological and ethological studies and experiments on the echinoderms of the Red Sea. Stud Trop Oceanogr 5: Munday PL, Jones GP, Caley MJ (1997) Habitat specialisation and the distribution and abundance of coral-dwelling gobies. Mar Ecol Prog Ser 152: Nelson BV, Vance RR (1979) Diel foraging patterns of the sea urchin Centrostephanus coronatus as a predator avoidance strategy. Mar Biol 51: Paine RT (1974) Intertidal community structure: experimental studies on the relationship between a dominant competitor and its principal predator. Oecologia 15: Power ME, Tilman D, Estes JA, Menge BA, Bond WJ, Mills LS, Daily G, Castilla JC, Lubchenco J, Paine RT (1996) Challenges in the quest for keystones. BioScience 46: Randall JE, Schroeder RE, Starck WA (1964) Notes on the biology of the echinoid Diadema antillarum. Caribb J Sci 4: Sakashita H (1992) Sexual dimorphism and food habits of the clingfish, Diademichthys lineatus, and its dependence on host sea urchin. Environ Biol Fish 34: Schiel DR, Foster MS (1986) The structure of subtidal algal stands in temperate waters. Oceanogr Mar Biol Annu Rev 24: Schmitt RJ, Holbrook SJ (1985) Patch selection by juvenile black surfperch (Embiotocidae) under variable risk: Interactive influence of food quality and structural complexity. J Exp Mar Biol Ecol 85: Schmitt RJ, Holbrook SJ (1990) Contrasting effects of giant kelp on dynamics of surfperch populations. Oecologia 84: Schoener TW (1983) Field experiments on interspecific competition. Am Nat 122: Sih A, Crowley P, McPeek M, Petranka J, Strohmeier K (1985) Predation, competition, and prey communities: a review of field experiments. Annu Rev Ecol Syst 16: Stachowicz JJ, Hay ME (1999) Mutualism and coral persistence: the role of herbivore resistance to algal chemical defense. Ecology 80: Steele MA (1997a) The relative importance of processes affecting recruitment of two temperate reef fishes. Ecology 78: Steele MA (1997b) Population regulation by post-settlement mortality in two temperate reef fishes. Oecologia 112:64 74 Steele MA (1998) The relative importance of predation and competition in two reef fishes. Oecologia 115: Steele MA (1999) Effects of shelter and predators on reef fishes. J Exp Mar Biol Ecol 233:65 79 Teytaud AR (1971) Food habits of the goby Ginsburgellus novemlinaetus and the clingfish, Arcos rubiginosis, associated with echinoids in the Virgin Islands. Caribb J Sci 11:41 45 Thompson DA, Findley LT, Kerstitch AN (1979) Reef fishes of the Sea of Cortez: the rocky-shore fishes of the Gulf of California. Wiley, New York Vance RR (1978) A mutualistic interaction between a sessile marine clam and its epibionts. Ecology 59: Vance RR (1979) Effects of grazing by the sea urchin, Centrostephanus coronatus, on prey community composition. Ecology 60: Witman JD (1987) Subtidal coexistence: Storms, grazing, mutualism, and the zonation of kelps and mussels. Ecol Monogr 57: Woodin SA (1978) Refuges, disturbance, and community structure: A marine soft-bottom example. Ecology 59: Zahary RG, Hartman MJ (1985) Artificial marine reefs off Catalina Island: Recruitment, habitat specificity and population dynamics. Bull Mar Sci 37: