Abstract Previous studies of the woolly sculpin, Clinocottus analis, have revealed homing behavior and site fidelity in this wide-ranging intertidal fish. The purpose of such behavior has been attributed to the increased chance of survival for fishes that can recognize and relocate suitable habitat during the daily fluctuations of submersion within intertidal zone. It should be equally important to the survivorship of fishes to be able to respond to adverse conditions in tidepools to which they home. The purpose of this study was to determine whether C. analis would continue to home under experimental conditions that were expected to be unfavorable, i.e. the periodic draining of the majority of water in tidepools. A field experiment was conducted at Hopkins Marine Station in Pacific Grove, California, during May 2001. Eleven pools between 0.430 m and 1.002 m above MLLW in the moderately-protected region of the intertidal zone were selected and assigned to three treatment groups: an unmanipulated control group, a group that was drained by 78-98% on a daily basis for one week, and a group that was drained by the same amount on three of the seven days of the experimental period. Fish in all pools were marked with subcutaneous paint spots. Pools were surveyed twice during the experiment. Draining was expected to increase maximum pool temperatures, density of fish and other tidepool organisms and decrease rugosity of pools; and all of these changes were observed. Contrary to expectations however, fish in all treatments maintained - 34% site fidelity/homing and did not respond in any statistically significant way to experimental treatment. These results imply either that fish were not adversely affected by the increased maximum temperatures, densities, and decreased rugosities introduced, or that fish were affected, but homing behavior was not sufficiently plastic within the time-scale of the experiment to show significant treatment effects between pools. INTRODUCTION As habitat for intertidal organisms, tidepools are extremely variable. Basic conditions such as temperature, salinity, submersion, and pool size change with each low tide. Longer term shifts in tidal patterns occur seasonally and interannually and less cyclic environmental disturbances, such as El Ninno occur at longer intervals. All of these factors make a tidepool a relatively unpredictable habitat. Yet, for smaller fish, tidepools serve as refuges from the intense predation by larger fish in the more stable subtidal zone and they serve as refuges within the intertidal during low tide, allowing the fish to avoid desiccation and other perils of emersion. In comparison with sessile intertidal organisms, such as barnacles, mussels and anemones, sculpins have the ability to adjust their location behaviorally within the intertidal. Habitat selection by intertidal cottids has been well studied. Degree of cover-whether it be vegetation or rock structures—is an important factor. Mollick (1968) found that C. analis prefers dark, densely-matted plant substrates and crevices. Richkus (1978,1980) observed that amount of cover was correlated with distribution of C. analis, and found that fish in artificial tidepools in the lab tended to select deeper and more protected pools. Wells (1986) observed higher population densities in pools with moderate to heavy cover. And Davis (2000) also concluded that a large amount of rocky cover was one of the preferred characteristics of C. analis habitat. Habitat selection implies the existence of a behavioral response to environmental conditions. In sculpins this behavior has an added layer of complexity. Not only do sculpins actively distribute themselves among suitable habitats, but they also remember these habitats and return to them. This behavior is termed homing. Intertidal sculpins leave pools at high tide and return, or "home," to these pools at successive low tides (Williams, 1957; Richkus, 1978; Yoshiyama, 1992). The closely related tidepool sculpin, Oligocottus maculosis, homes to its pool of capture when displaced 100 meters away or after a period of captivity of six months (Green, 1971). The mechanism that sculpins use to home has not been identified, although olfaction, and to a lesser extent, vision play significant roles (Khoo, 1974). Previous research has attributed homing behavior to the necessity of avoiding the large area of the intertidal that become uninhabitable during low tide—either due to emersion or adverse thermal or chemical conditions in tidepools. Although many tidepools with favorable conditions are available as the tide goes out, not all will remain so, and certain changes, such as maximum temperatures. ultimate low-tide pool size, and duration of isolation, are difficult to predict with any other means than experience, i.e. presence in the pool at a previous low tide. Thus, the reliance on pools with favorable conditions based on previous experience with those pools is hypothesized to be the motivation for homing behavior. The purpose of this study was to assess the plasticity of homing behavior by observing responses of C. analis to unfavorable conditions imposed experimentally on pools to which they home. I hypothesized that if fish can recognize and relocate favorable pools, they should be able to recognize and avoid unfavorable conditions should they arise in a fish's home pool. Experimental tidepools were made unfavorable by regular draining at low tide. Substantial changes in temperature, fish density and rugosity were seen but without a concurrent change in the degree of homing. MATERIALS AND METHODS Sculpin homing was observed during May, 2001, in a region of the rocky intertidal zone at Hopkins Marine Stations in Pacific Grove, California. The study site consisted of a series of tidepools, within a 30 m radius in a moderately-protected zone. Pools which were too large, deep, or crevice-filled to ensure high capture rates were not used in the experiment. Pools were observed for one week. Due to logistic and time constraints, only four pools at most could be surveyed in a low tide. The week-long experiments were staggered over a period of two weeks for different sets of pools, from May 14th-May 27th. Low tides during this time were in the middle of the day to late afternoon until May 22nd, and in the early morning thereafter until May 28th. Experimental treatment consisted of draining pools one to two hours before the low tide, and leaving those pools drained for the duration of the low tide. Sufficient volume was left for fish to be fully submerged and mobile, but draining usually reduced their habitat to a very shallow and exposed body of water at the bottom of the tidepool. The treatment was conducted at three levels: daily draining, moderate draining, and no draining. Four pools were not drained at all and served as controls (“controls:” F, H, J, and O). Five additional pools were drained 3 days of the 7-day observation period ("moderate:" N, E, P, Q, R), and two pools were drained every day of the observation period (“drained:" G, M). Pools were surveyed at three times: before treatment ("preliminary"), on the first day of treatment ("initial"), at day three ("mid¬ treatment"), and at day seven (“final"). In a few pools, preliminary surveys were not made, due to time constraints. At each survey, pools were thoroughly searched with dip-nets and probes for fishes. I examined crevices, algal cover and any other shelter that might conceal fishes. Fishes typically darted out of hiding when disturbed and either retreated to another crevice or remained motionless in the open. In the latter case, the fish was surrounded with two medium-sized dip nets at an angle and driven into them with a hand motioning from a third direction. This process was repeated until 1 was confident that nearly all fish were caught. Capture rates were not in all cases 100%. An additional 5-10% of fish probably evaded capture, since certain fish would take refüge in visible crevices, from which they would not move even when prodded. As visibility is essential for capture, fishes that exhibited this behavior when out of sight, were undetectable. Although this is a serious possible bias in the experiment, it is somewhat lessened by the apparent lack of a "capture bias" in pools with 100% estimated capture rate. Capturing fish in these pools was no more difficult than capturing marked fish; it is therefore hoped that samples of the entire population in other pools would be representative, if not complete. All captured fish were measured, identified to species, and released. During the preliminary and initial surveys all fish were also marked During preliminary surveys (except in the case of pools N, O, P, Q and R) clove oil was used to partially tranquilize fish (M. Webster pers. comm.). In order to apply a relatively undiluted dosage, pools were partway drained, a few drops of clove oil added, and all fish caught. Fish were transferred to a bucket of fresh seawater, taken from the same pool prior to treatment, and appeared to recover very quickly. Fish were then identified to species, measured to the nearest .01 cm, tagged and returned to the tidepools. Tidepools were thoroughly flushed with fresh sea-water to dilute traces of clove oil. Tranquilization was later abandoned because of feasibility of capturing untranquilized fish and the desire not to alter temperatures in drained pools by flushing after application of clove oil. Also, it was feared that partial tranquilization would aggravate the motionless behavior described above and actually compromise capture rates. Fish were marked using silicon-based elastamer paints, injected from an insulin syringe Marks were subcutaneous, on the ventral surface, to either side of the anal fin. Marking on the dorsal surface to facilitate recognition of fish from above without disturbing them by capture was ineffectual because of the high concentration of skin pigments there. Ventral marking has minimal effect on swimming abilities and is inconspicuous from above. All fish, with the exception of fish in pools F and G (released within 3 m of their pools of capture) were released into their pools of capture after preliminary surveys. Temperatures were recorded using programmable temperature recorders (i-buttons '94. Dallas Semiconductor). Volumes were taken when pools were fully isolated by pumping water into calibrated 18-liter buckets. Population densities were calculated based on the number of fishes present and the volume of a pool at a particular survey; these values were compared to homing rates of the following survey. Depths were measured at the lowest point of the pool. Rugosity was measured qualitatively, following Bennett and Griffiths (1984), on a scale from one to five—five being the most rugose (pools which were covered entirely in uneven rocky shelter) and one being the least rugose (pools with entirely smooth sand or rock bottoms and sides). Statistics were carried out using analysis of variance on Systat 8.0. RESULTS The average of homing in all pools was 34% (Table 1). Average size of all fish was 4.38 cm and of homing fish, 3.78cm. Of the 101 fish tagged in experimental pools, 39, or 38% were never seen again. Three fish were seen in pools other than those in which they were captured. All three were found in other experimental pools less than 12 m away and were smaller than 3.7 cm. There were no significant differences in homing between treatment groups or between times (two-factor ANOVA, all P 2.53; Table 2). Fish in drained and moderate pools homed to the same extent as fish in unmanipulated controls. Since the interaction of treatment factors and time was not significant, the interaction and error Mean Squares were pooled for tests of main effects. The lack of significant differences in homing among treatments led me to examine the degree of actual disturbance introduced by the experimental treatments. Drained pools reached higher maximum temperatures than control and moderate pools, which were more similar to one another (Fig. 1). Maximum temperatures in moderate pools on May 18th, May 20th, and May 24th were unusually low. A downward trend in maximum temperatures is apparent during the experiment and is the result of the times of low tides shifting towards from mid-day to evening, and eventually early morning (May 24). Volumes were reduced by 78-98% in pools that were drained. This necessarily resulted in a large change in density of fish (Fig. 2). Densities increased from less than 0.5 fish per liter at normal volume to up to 5 fish per liter at drained volume. No pattern was evident between original volume of the pool and density of fish. No aggressive interactions were observed between fish in drained pools. There was no apparent relationship between homing and fish density (Fig. 3). Other intertidal fish observed in pools were Clinocottus recalvus and juvenile pricklebacks. Increased density of other intertidal species caused heightened activity in the small volume of water, especially among hermit crabs, which were observed grabbing with their claws for fish, causing the fish to dart away. This was not observed in pools at normal volumes. In one instance, a small Pachygrapsus successfully captured an unmarked C. analis Qualitative measures of rugosity taken for each pool at normal and drained volumes (except for the control) showed that treatment caused general decreases in rugosity (Fig. 4). Depending on the topography of the individual pools, draining reduced the number of crevices and rocks available for cover. In certain pools (E, N, and R) draining restricted fish to sand- covered low spots with very little rocky cover. DISCUSSION The lack of any difference in rates of homing between treatment groups, while unexpected, might be the result of a number of things. Disturbance of fish by capture and marking may have played a role in the homing rates seen. Since handling techniques were the same in all pools the effect of this factor would be equal for all treatments. If the effect of handling were very large, then it may have obscured more subtle trends in homing between the different treatments. Unfortunately, the creation of a wholly unmanipulated set of controls was unsuccessful in the early stages of this experiment. Richkus (1978) reported homing rates of 30% after two weeks, compared to 34% in my experiment at the end of one week. Because both homing and total sculpin population decreased over time, Richkus (1978) asserted that handling was not an influence on homing. The higher rates of homing in my experiment suggest a similar lack of disturbance from handling. Ideally, replicate studies with procedural controls must be done to confirm this. Graham (1970) found that lethal temperatures for C. analis are between 25° and 27° C. The upper temperature of tidepools in which C. analis naturally is found is 22°C (Wells, 1986) Temperatures in my experimental pools did not exceed 22°C at any time and were several degrees cooler on every day but May 16. Thus lack of a thermal stress may explain why fish continued to home to drained pools. Although early observations both in the laboratory (Richkus, 1980) and in the field (Richkus, 1978) suggested that C. analis was highly non-aggressive, more recent studies have suggested that C. analis is in fact one of the more aggressive species of the cottid family. It has recently been suggested that (no comma) in the large area of California coast where it is the most abundant species, C. analis might be competitively excluding other cottids from tidepool habitat (Gibson, 1999). Such aggression occurs at the intraspecific level as well. Size may be a factor correlated with aggression. Yoshiyama (198 1) observed marked intraspecific aggression by larger fish in laboratory settings. Small fish were nonassertive, but upon reaching 5-6 cm fish would invariably develop aggressive behavior. If this is true in the field, it may offer insight into why homing behavior did not change in my experiment. Aggressive interactions may have been minimal, given that the majority of fish in each pool were under 5 cm. Draining pools may have incited little change in stress due to aggression, and therefore little effect on the homing behavior of experimental pool populations. Interspecific density was a large source of potential disturbance. Heightened activity of other tidepool organisms had the effect of harassing resident sculpins. Why it did not cause distinct changes in homing rates is unclear, especially considering the predatory nature of most of this activity. However, only one actual injury was observed during the entire experiment (inflicted by a Pachygrapsus) so it is assumed that most attacks were unsuccessful, and did not cause sufficient disturbance to alter homing behavior. Rugosity is an important factor in the habitat selection of C. analis. That substantial changes in rugosity did not result in any subsequent change in homing behavior is puzzling- especially given that previous experimental manipulations decreasing rugosity in the field have resulted in definite declines in pool populations of C. analis (Davis, 2000). This result implies one of three things: 1) that qualitative measurements of rugosity were not accurate enough,2) that rugosity changes, however large, did not cause a significant impact on fishes, or 3) that the time scale of the experiment was not sufficient to observe a downward trend in homing rates in drained pools. Of these possibilities, the latter is the most probable. In conclusion, it is quite plausible that the extremely limited scope of this experiment—in both time and space—may not have allowed me to observed actual changes in homing resulting from the experimental conditions introduced. The clarity of the experiment would have benefited from the addition of pools to the three treatment groups. With degrees of freedom being as low as one for the drained group, the power of statistical tests was considerably compromised. And it is highly possible that the limited time-scale of the experiment—one week—may have been too short to reveal any changes in homing behavior. ACKNOWLEDGEMENTS 1 am grateful to the many people at Hopkins who make undergraduate research possible, worthwhile and fun. I would especially like to thank Jim Watanabe for guidance in all aspects of this project; for helping me to understand the value of and challenging me to use good experimental design, statistics, and critical thinking. Thank you to Joanna Nelson for encouragement and help in the field, and Luke Hunt for help with statistics and field surveying Thank you to Eric Sanford and Michael Webster for providing the materials and methods for effective marking; and to Freya Sommer for helping me refine my species identification in the first weeks. Finally I would like to express my gratitude to all 175H professors for providing an intense and amusing introduction to real research. LITERATURE CITED Bennett, B. A. and Griffiths, C.L. 1984. Factors affecting the distribution, abundance, and diversity of rock-pool fishes on the Cape Peninsula, South Africa. S. Afr. J. Zool. 19: 97- 104 Davis, J. L. D. 2000. Spatial and seasonal patterns of habitat partitioning in a guild of southern California tidepool fishes. Mar. Ecol. Prog. Ser. 196: 253-268. Horn, M and M. Chotkowski, eds. 1999. Intertidal fishes: a life in two worlds. Academic Press. San Diego. Graham, J. B. 1970. Temperature sensitivity of two species of intertidal fishes. Copeia 1970: 49-56. Green, J. M. 1971. High tide movements and homing behavior of the tidepool sculpin Oligocottus maculosus. J. Fish. Res. Bd. Canada 28: 383-389. Khoo, H. W. 1974. Sensory basis of homing in the intertidal fish Oligocottus maculosus Girard. Can. J. Zool. 52: 1023-1029. Mollick, R. S. 1968. The distribution of Clinocottus analis Girard in tide pools as related to substrate preference. San Diego State College, MS thesis, 86p. Mollick, R.S. 1969. The distribution and behavior of an intertidal fish. Virg. J. Sci. 20:113. Richkus, W. A. 1978. A quantitative study of inter-tidepool movement of the wooly sculpin Clinocottus analis. J. Mar. Biol. 49: 277-284. Richkus, W. A. 1980. Laboratory studies of intraspecific behavioral interactions and factors influencing tidepool selection of the wooly sculpin, Clinocottus analis. Calif. Fish and Game. 67: 187-195. Wells, A. W. 1986. Aspects of ecology and life history of the woolly sculpin, Clinocottus analis, from southern California. Calif. Fish and Game. 72: 213-226. Williams, G. C., 1957. Homing behavior of California rocky shore fishes. Univ. Calif. Publ. Zool. 19: 294-284. Yoshiyama, R. M.; Gaylord, K. B.; Phillipart, M. T.; Moore, S. R.; Jordan, J.R.; Coon, C. C.; Schalk, L. L., Valpey, C. J., Tosques, I. 1992. Homing behavior and site fidelity in intertidal sculpins (Pisces: Cottidae). J. Exp. Mar. Biol. Ecol. 160: 115-129. Yoshiyama, R. M. 1981. Distribution and abundance patterns of rocky intertidal fishes in central California. Env. Biol. Fish. 6: 315-332. Table 1. Homing rates. Treatment groups and the corresponding homing rates in replicate pools at day 3 and day 7. Homing rates are calculated by dividing total number of tagged fish at a particular survey over the total number of original tagged fish. The average of homing over time and treatments is 34%. HOMING Total Day TREATMENT Pool Tagged Fish Day ? 0.3 control 0.22 0.50 0.33 0.29 0.07 0.69 0.62 0.2 0.38 moderate 0.50 0.33 0.2: 0.00 0.29 0.43 R 0.50 0.00 drained G 0.45 0.36 M 10 0.40 0.40 0.36 11 101 0.32 0.34 Table 2. Two-way ANOVA examining treatment, time and interaction. No significant differences were found between homing rates in different treatments or at different times (p20.50), Interaction of treatment and time was also non-significant; these mean squares were pooled to test the main effects. SOURCE MS F-ratio 0.654 0.022 Treatment 0.007 0.215 63 Time 0.959 Treatment x Time 0.032 40 16 0.033 Within FIGURE LEGEND Fig. 1. Maximum low tide temperatures in control, moderate, and drained pools. Times of low tides shift from the middle of the day on May 16 to evening on May 22. May 24 is an early morning low tide. Fig. 2. Density changes were substantial: 5-47-fold increases in number of fish per liter. Fig. 3. Density vs. Homing. Slopes of both lines are near-zero; at survey 2(day 3) y = 0.0009x +0.3932. At survey 3 (day 7) y = 0.0149x + 0.3014 (P22.05 in both cases). Fig. 4. Rugosity measurements are qualitative, based on a scale from one to five-five being the most rugose (pools which were covered entirely in uneven rocky shelter) and one being the least rugose (pools with entirely smooth sand or rock bottoms and sides). For moderate and drained pools, filled bars are pre-treatment and shaded bars are post- treatment. Fig. 1 Temperature (C) 245 23 ne 20 . V CONTRO 17 NODERATE 16- 17- 18- 19- 20- 21- 22- 24- DATE (May, 2001) Fig. 2 Fish Density in Drained and Normal Volume Pools Edrained normal volume FHJOENPORGM Pool Fig. 3 Homing vs. Fish Density 0.80 060 § 0.40 0 0.20 —— 0.00 6.00 8.00 2.00 4.00 0.00 Fish Density at Previous Survey Survey 2 * Survey 3 Fig. 4 2 contro moderate drained