Abstract. This study examined the diet of the generalized predatory sea star Pisaster giganteus, comparing proportions of various species in the diet to abundance in the habitat. Comparisons to assess temporal variation were made to similar data collected in the same site in 1981 (Harrold 1981). Relative frequencies of prey species in the diet and in the habitat were tabulated. The past and present diets of sea stars overlap considerably, with a Morista overlap index of 0.667 and nearly identical Shannon-Weiner diversity indices (2.035 in 1981 compared to 2.029 now). However, a Chi-square analysis reveals several major differences: (1) the absence of the vermetid gastropod Petaloconchus montereyensis, and (2) an increase in the frequency of two other sessile prey species, the barnacle Balanus spp. and another vermetid, Serpulorbis squamigerus. The absence of Petaloconchus in the diet was due to its low abundance in the habitat. The increased feeding on Balanus and Serpulorbis cannot be explained by changes in abundance, which were not statistically significant. Rather. I hypothesize that the absence of Petaloconchus, previously a reliable, easily obtainable source of energy, led to increased feeding on these other sessile prey species. Further study is needed to explore the impact of other factors, such as individual sea star variation. INTRODUCTION The diets of various asteroids have been well studied (e.g. Mauzey et. al 1968, Menge 1972, Nakashima 1974, Harrold 1981, Herrlinger 1983). Observed prey choices and proportions in the diets have been examined for several generalist sea stars. Pisaster giganteus is a subtidal sea star which feeds by everting its cardiac stomach on prey and digesting externally. Harrold (1981) analyzed various aspects of its feeding ecology by comparing the abundance of prey in the diet to availability (factoring in numerical abundance, ease of capture and presence in the same microhabitat), caloric yield and predator choice. However, since then, little work has been done to follow up this study and record or explain any temporal variation. Menge (1972) reported temporal variation in prey choice by L. hexactis prey, both daily (with the tides) and seasonally, as the availability or ease of capture of different prey species varied. Seasonal changes followed abundance patterns of calorie-rich or calorie-poor species. However, studies have not been done to explore variation in feeding ecology from year to year and make comparisons to environmental changes such as prey abundance. This study compared the diet of Pisaster giganteus to Harrold (1981), with two main goals: (1) to record differences in diet composition and prey abundance, and (2) to infer an explanation from these results. The results showed several major dietary changes correlated with the disappearance of a frequently eaten species in the habitat. Analysis of these changes confirms previous speculation on Pisaster foraging strategy. MATERIALS AND METHODS This study was conducted in the kelp forest of the Hopkins Marine Life Refuge (HMLR), Pacific Grove, California, within an approximately 40 mx 40 m patch of kelp forest, centered on a black cable that runs north-south through the kelp (Fig. 1). The bottom consists mostly of rocky substrate, with some sandy areas, I limited all data collection to rocky substrate, the preferred habitat of Pisaster giganteus. Studies were conducted using SCUBA, in 32 dives between April 23 and May 23, 1996. The mean density of the sea stars was calculated from 47 10 m' circular quadrats. I placed these quadrats at random locations by choosing random compass bearings and numbers of kicks (to a limit of 15) from the center of the study site. I moved quadrats to the next random location if more than half of the area fell on sandy substrate. For each sea star I encountered in the quadrat, I recorded size (length from the tip of the longest arm to the opposite interradius) and overturned it to determine whether it was feeding and the number and identity of any prey species. From this, the proportion of feeding sea stars and the percent of each prey species in the diet were calculated. Sea stars were returned to approximately their original position after measurements were taken. The main criterion for a feeding observation was an everted cardiac stomach. Some sea stars had prey items pressed against the stomach, and others simply had an everted stomach, e.g. if the prey was attached to the substrate. Several sea stars were in crevices and inaccessible, and could not be overturned, tabulated these sea stars in a separate category since their feeding behavior could not be determined. I collected additional feeding data by overturning and recording any encountered sea stars in 20 m swaths perpendicular to the cable. These swaths were examined as closely as the quadrats so that the data would not be skewed to favor more visible sea stars. The same sea stars may have been tabulated more than once during the study, but repetition was most likely uncommon and thus had little effect. I compared the observed diet to Harrold (1981) by using the Shannon-Weiner diversity index, the Morista overlap index and the electivity index of Ivlev (1961). The Shannon-Weiner index was calculated as H' - -Ep, ln p,, where p, is the proportion of the it species. The Morista overlap index was calculated as 22xy,/(Exj'+Ly2), where x, and y, are the proportions of the i species in the diet in 1996 and 1981 respectively. The electivity index of Ivlev (1961) was calculated as E' - (D-H)/(D4H), where D = relative abundance of a given prey species in the diet, and H = relative abundance of that species in the habitat. I calculated the mean density of various prey items by plaging quadrats at random compass bearings and kicks from the center of the study site. More abundant snails (Tegula spp. and Calliostoma ligatum) were counted within 0.25 m2 quadrats, 20 for Tegula and 19 for Calliostoma. Tegula data were supplied by Hunt (1996). I counted 21 10 m’ circular quadrats for less abundant species (Astraea gibberosa, Mitra idae, Serpulorbis squamigerus, Ceratostoma foliatum and Conus californica). The abundance of sessile species that aggregate on rocky substrate (Petaloconchus montereyensis, Balanus spp., and polychaete worms) were measured as percent cover, using 15 0.25 m quadrats divided into grids of 25 10 cm x 10 cm squares. The proportion of grid intersections falling on a given species was used to estimate percent cover. grouped polychaetes together since I initially did not distinguish them in the diet. Balanus spp. were rare and thus also counted individually within 1 m quadrats. Percent cover measurements were used as estimates of relative abundance for Petaloconchus and polychaetes, and relative abundance of individually counted prey species was calculated separately as the density of a prey species divided by the total density of all prey species in 1 m'. To eliminate the effects of seasonal variations, only data from Harrold (1981) taken during spring (late March through June) were used. Calliostoma abundance numbers were compared to measurements by Sellers (1977), taken during the same period and in the same area as Harrold (1981). RESULTS During the study, I sampled 130 sea stars, of which 61 (46.93) were feeding. Eleven sea stars (8.53) were lodged in crevices and inaccessible. The density of Pisaster giganteus was 1.15 + 1.47 per 10 m’ (mean + S.D.), not significantly different from Watanabe (1984), t.=1.159. Based on size measurements of 124 sea stars, the mean size (+ S.D.) was 13.4 + 3.1 cm. The most frequently found prey items were the barnacle Balanus and the vermetid gastropod Serpulorbis, with relative frequencies of 28.793 and 22.723 respectively (Fig. 2). No sea stars were observed feeding on Petaloconchus. The species found in the diet, in decreasing order of frequency, were Balanus spp., Serpulorbis, polychaetes, Tegula pulligo, Calliostoma, Tegula montereyi, Astraea and Ceratostoma (tie), and Mitra, Nassarius mendicus and an unidentified mussel (tie). The diet composition was very similar to Harrold (1981). The Shannon-Weiner diversity index was 2.029 compared to 2.035 (Harrold, unpublished data). These diets, in terms of actual species and proportions, also overlapped quite substantially (Morista index - 0.667). A Chi-square analysis, however, showed a statistically significant difference between the diets (y=28.3, po0.005). Qualitatively, the largest differences in the diet between Harrold (1981) and my study were in the frequencies of Petaloconchus, Serpulorbis, Balanus and polychaete worms (Fig. 3). The most abundant individually counted prey species was Calliostoma, followed by T. pulligo, T. montereyi, T. brunnea, Serpulorbis, Balanus spp., Mitra, Ceratostoma, Astraea and Conus (Table 1). No Petaloconchus was encountered while recording abundance data. Polychaetes had a 6.253 cover. Only Petaloconchus had a significantly different abundance from previous measurements (p.0.01, Table 1). No Petaloconchus was observed in the quadrats, although it was the most abundant prey species (18.83 cover) in Harrold (1981). A qualitatively large increase in Serpulorbis abundance was statistically insignificant. Prey species with the greatest changes in the diet were compared to changes in relative abundance (Fig. 4). Balanus and Serpulorbis both increased in the diet more than in the habitat (Balanus abundance actually decreased). The increase of polychaetes in the diet was approximately the same as the increase in abundance, and the disappearance of Petaloconchus was likewise reflected in both the diet and the habitat. The relative proportion of prey species in the diet and abundance in the habitat were used to calculate the electivity indices for each of the species now (Table 2) and in Harrold (1981) (Table 3). For the species analyzed in Fig. 4, the electivity indices were compared (Fig. 5). These indices increased for both Balanus and Serpulorbis and decreased for polychaetes. DISCUSSION The relative frequency of polychaetes in the diet increased by approximately the same amount as their relative abundance. The dietary increase of polychaetes can thus be explained by the sea stars' taking more worms in proportion to their higher availability, rather than making a choice or adjustment to their diet. However, the increased relative frequency of Balanus and Serpulorbis in the diet was larger than any abundance changes, which suggests another cause. According to foraging theory, a predator must make choices to maximize energy input while minimizing effort and difficulty to gain that energy (Stephens and Krebs 1986). Emlen (1966) demonstrated this with two species of carnivorous gastropods, which optimized foraging with respect to time and energy conservation. Menge (1974) showed how a predaceous snail made foraging choices to maximize biomass intake per time and decrease search time. Unpredictability of prey species has also been linked to generalization of diets and prey choice (Menge 1972). Harrold (1981) found that regardless of the relative availability of T. pulligo (high caloric yield) and Petaloconchus (low caloric yield), some Petaloconchus was always taken. He hypothesized that this was due to the uncertainty of capturing T. pulligo, which exhibits an escape response and is often found in a different microhabitat than Pisaster. The strategy was thus to always take some abundant but low-energy prey since even abundant high-energy prey might not be reliably available. Both species that increased the most in the diet, Balanus and Serpulorbis, are sessile species, and thus cannot escape a foraging sea star. Algae and other encrusting organisms found on the shell of Serpulorbis render it less recognizable (Harrold 1981), but this defensive response is not as effective as that of mobile snails (such as Calliostoma or T. pulligo). Although these sessile prey are not now as abundant as Petaloconchus was in 1981, they are easily captured upon encounter. Thus, the increased frequency of Balanus and Serpulorbis may reflect an adjustment to replace Petaloconchus as a predictable energy source. These results call for further study and more data. Since the sample size of this study was small, the resolution of abundance and dietary changes is fairly low. Small frequency changes could not be detected, and some apparent changes may have been partly due to sampling error. More feeding data could correct any such effects and yield more conclusive results. The role of the sessile polychaetes in Pisaster's diet should be further explored. Although their increase in the sea star's diet may be explained by an increase in abundance, more data may indicate a more precise change which may also stem from the absence of Petaloconchus. Another unresolved question is the role of individual variation in the observed differences, as has been displayed in other predators (West 1986, 1988). Other methods of characterizing prey availability could be explored, to see whether the results are comparable. Harrold (1981) used two distinct methods of measuring abundance, numerical abundance, counting the number of each species, and encounter abundance, which factored in the length of each encounter with a prey species. Encounter abundance would theoretically account for effects of prey avoiding predation by living in different microhabitats or other adaptations, which would render these prey less available than their numerical abundance might suggest. This method should differentiate between a choice made by a predator and a simple inability to find and capture prey. However, this method seems to undersample rare events, since the limitations of SCUBA prevent long-term observation of sea stars. Harrold (1981) notes that for many sessile species which do not exhibit escape behaviors, numerical abundance is an adequate measure of availability. He noted a small difference in electivity indices of Serpulorbis based on encounter abundance rather than numerical abundance, but a smaller difference than for mobile snails such as T. pulligo and Calliostoma. Thus, for the species highlighted in my study (Balanus, Serpulorbis, and Petaloconchus), numerical abundance should suffice. However, using encounter abundance may show whether mobile species also played a role in replacing Petaloconchus in the diet. Menge (1972) defined prey availability as the ratio of calories consumed from a certain prey to calories encountered. However, since this uses the actual diet numbers, it includes predator choice, making prey availability a function of the predator, not simply the prey and environment, unless one assumes that the diet follows actual availability. My study focused on response to abundance changes, and caloric data were not estimated, so this method was not used. Given the time and methodological limitations of this study, though, preliminary conclusions suggest that Pisaster giganteus has compensated for the absence of Petaloconchus by feeding more frequently on other sessile prey, keeping with the strategy of maintaining a dependable energy source to offset the risk of not capturing calorically more attractive prey. Further study should support these findings. LITERATURE CITED Emlen, J.M. 1966. Time, energy, and risk in two species of carnivorous gastropods. Doctoral thesis, University of Washington, Seattle, WA. Harrold, C. 1981. Feeding ecology of the asteroid Pisaster giganteus in a kelp forest system: prey selection, predator-prey interaction, and energetics. Doctoral thesis, University of California, Santa Cruz, CA. Herrlinger, T. 1983. The diet and predator-prey relationships of the sea star Pycnopodia helianthoides from a central California kelp forest. Doctoral thesis, San Jose State University, San Jose, CA. B. 1996. Bathymetric distribution of three species of kelp Hunt, forest gastropods (Trochidae: Tegula) in 1978 and 1996. Unpublished student paper, Hopkins Marine Station, Pacific Grove, CA. V.S. 1961. Experimental ecology of the feeding of fishes, Ivlev, Yale University Press, New Haven, Connecticut, 302 pp. Mauzey, K.P., C. Birkeland and P.K. Dayton. 1968. Feeding behavior of the asteroids and escape responses of their prey in the Puget Sound region. Ecology, 49:603-619. Menge, B. 1972. Foraging strategy of a starfish in relation to actual prey availability and environmental predictability. Ecological Monographs, 42:25-50. Menge, J. 1974. Prey selection and foraging period of the predaceous rocky intertidal snail, Acanthina punctulata. Oecologia (Berl.) 17, 293-316. Nakashima, R. 1974. Asteroid predation in Monterey Bay. Senior thesis, University of California, Santa Cruz, CA. Sellers, R.G. 1977. The diets of four species of Calliostoma (Gastropoda, Trochidae) and some aspects of their distribution within a kelp bed. Master's thesis, Stanford University, Stanford, CA. Stephens, D.W. and J.R. Krebs. 1986. Foraging Theory. Princeton University Press, Princeton, New Jersey, 247 pp. Watanabe, J.M. 1984. The influence of recruitment, competition, and benthic predation on spatial distributions of three species of kelp forest gastropods (Trochidae: Tegula). Ecology, 65(3):920-936. West, L. 1986. Interindividual variation in prey selection by the snail Nucella (Thais) emarginata. Ecology, 67(3):798-809. 1988. Prey selection by the tropical snail Thais melones; a study of interindividual variation. Ecology, 69 (6):1839- 1854. Table 1: Abundance of prey species compared to Harrold (1981). Numbers given as mean i standard error. Calliostoma numbers from Sellers (1977). All numbers normalized to 1 m?. Previous numbers based on N=21 for Tegula, N=ll for Serpulorbis, N=13 for Balanus, N=100 for Calliostoma, N=14 for polychaetes and Petaloconchus, and N=15 for others. NS-not significant. See Materials and Methods for 1996 N values. 1996 (per m) Prey Species 1981 (per m t-statistic 8.8 + 2.6 T. pulligo 9.2 1 5.4 U.0658 NS T. montereyi 3.4 + 1.O 2.84 1 1. U.327 NS T. brunnea 2.6 + 1.6 1.52 + U.8 O.614 NS O.057 + O.0I3 Astraea 39 NS 0.09 + 0.05 0. Mitra 0.348 + 0.04I 0.42 + 0.09 0.802 NS Balanus spp. 0.67 + 0.32 0.64 + 0.44 O.O55 NS Serpulorbis 2.148 + 0.376 1.17 + 0.22 1.765 NS Polychaete worms 6.255 + 1.55 1.59 NS 2.55 + 1.795 c. ligatum 1.I5 NS 14.52 + 3.2 IU.12 + 1.64 Ceratostoma 0.119 + 0.045 811 NS 0.02 + 0.0 Conus 0.014 + 0.0076 972 N 0.04 + 0.03 Petaloconchus 18.83 1 6.35 J.081 (p80.0I) Table 2: Relative abundance of prey species in the diet and in the habitat. Petaloconchus and polychaete percent cover measurements were used for relative abundance. Relative abundance of individually counted species was calculated separately. Nassarius abundance numbers were not available. Numbers in parentheses are actual numbers of observations, Prey Species Diet Habitat Tegula pulligo 7.585 (5) -0.559 26.8. eula montereyi 10. 4.555 [3 -0.390 7.93 Tegula brunnea -1.0 3.035 [2) Astraea gibberosa U.17 0.894 Mitra ae 5 (I) 1.065 O.I7 28.795 (19) Balanus spp. 2.045 U.868 22.14 25 (15 6.5 Serpulorbis squamigerus U.552 153 (I0) Polychaete worms 6.2 0.416 44.2 Calliostoma ligatum 6.065 (4) -0.759 Ceratostoma toliatum 3.035 [2) 0.365 0.788 onus calornica 0.435 -1.0 Petaloconchus montereyensis 05 N/A Nassarius mendicus 1.515 (I) N/A N/A Bivalve 1.515 (I) N/A N/A 2 137 Undentified N/A N/A Table 3: Relative abundance of prey in the diet and in the habitat at the time of the previous study (Harrold, unpublished data). Nassarius mendicus was not included in the proportional abundance in the habitat for better comparison with recent data, since Nassarius abundance was not measured in the present study, Prey Species Diet Habitat 5.775 (I5) Tegula pulligo 34.775 -0.71 10.7 Tegula montereyi -0.512 3.465 191 ula brunnea 15 6) 5.90 2. -0.4. Astiea gibberosa 2.318 16) U.345 0.743 Mitra dae 1.593 5 (I) -0.610 Balanus spp. 14.65 (38) 2.42 0.716 serpulorbis squamigerus 10.85 [28) 4.425 0.419 Polychaete worms 11.165 (29) 2.55 U.634 Calliostoma ligatum 2.695 77 38.2 -0.869 Ceratostoma toliatum -0.610 3.085 (8) O.085 Conus caliornca 1.155 (3 -0.135 0.705 Petaloconchus montereyensis 27.75 72) 0.I 18.85 4.23 (II Nassarius mendicus N/A N/A other 10.45 (277 N7A N/A Figure 1: The location of the study site at Hopkins Marine Station. The hatched area marks the kelp canopy, and the dark square marks the study site. Figure 2: Relative frequencies of prey species in the diet of Pisaster giganteus. The most frequently eaten prey were Balanus spp., Serpulorbis squamigerus and polychaete worms. Percentages rounded to nearest whole number. Other" includes Nassarius mendicus and a mussel. Figure 3: Changes in diet composition, shown as relative frequengy in 1996 minus relative frequency in Harrold (1981). The largest differences in the diet between Harrold (1981) and the results of my study were in the proportions of Petaloconchus, Serpulorbis, Balanus and polychaete worms. Figure 4: Changes in diet compared to changes in abundance for species with the largest observed dietary change, Changes in abundance calculated as relative abundance in 1996 minus relative abundance in Harrold (1981). Serpulorbis and Balanus increased more in the diet than in the habitat, while the increases of polychaetes and Petaloconchus matched the abundance changes, Figure 5: Electivity index comparison of several species (Harrold, unpublished data). Balanus and Serpulorbis electivity indices increased, as expected from data in Figure 4. 4 Son Francisco L L Santo Cruz PMoss Londing Pacific Grove Monterey Hopkins Marine Station 50m, Pacific Grove, CA cable g kelp eee study A 111 00 site a gs 8 —o 19 . e 1 Figur Unidentified Other Ceratostoma foliatum 3% Calliostoma ligatum 6% Polychaetes 15% Serpulorbis squamigerus 22% T. pulligo 8% Figure 2 T. montereyi Astraea gibberosa 3% Mitra idae 2% Balanus spp. 28% 19 ane Nastarius mendicus Ceratquetoma foliatum Callipstoma ligatum Polychaetes Serpulorbis squamigerus Balanus spp. Mitra idae Astaea gibberosa T. montereyi T. brunnda T. pulligo -10 Frequency Change (%) Figure 3 20 15 5 10 -15 -20 -30 1 2 Prey Species Figure 4 Diet c Habitat 0.8 0.6 0.4 0.2 02 -0.4 -0.6 -0.8 1 1 Prey Species Figure 5 Q1981 21996 22