Substrate Selection in Caprellid Amphipods of Southern California, with Emphasis on Caprella californica Stimpson and Caprella equilibra Say (Amphipoda) !

DONALD E. KEITH 2

ABSTRACT: The substrate affinities of Southern Californian caprellids were studied with principal interest in two species, Caprella califol'11ica and Caprella equilibra. Experiments designed to test the selectivity for three substrates at Long Beach Marina showed C. calijornic« to "prefer" the bryozoan Bugula neritina over the algae Polysiph onia pacifica and V iva lobata. Caprella equilibr« showed no preference between Bugul« neritina and Polysiphonia pacifica, but selected these substrates over VI va lobata. Selectivity of Caprellu calijornica was attributed to its cryptic adapta- . tion.

THE CAPRELLIDEA CON STITUTE a highly spe­ grahami Wigley and Shave, as a commensal on cialized suborder of amphipoda. Previous pub­ the starfish Asterias [orbesi (Desor.) . lished works have dealt primarily in systematics The apparent dependence of most caprellids and very little information is available concern­ on other living organisms for support and pro­ ing caprellid ecology. tection makes this relationship an obvious place Caprellids are most commonly found clinging to begin an ecological investigation. The primary to some living substratum. Among those most importance of this study is to establish whether frequently cited in the literature are: algae, substrate affinities exist among certain associated sponges, hydroids, and bryozoans. According to species of caprellids and to evaluate factors Mayer ( 1882, pp. 170-171) there are some which could influence selectivity of these ani­ caprellids which do not live on other organisms. mals. Pseudolirius kroyeri (Halleri), for example, During the summer of 1961 at the Virginia lives exclusively on sand or muddy sand. This Institute of Marine Science, I obtained some species cannot clasp large objects since the hind evidence to support the thesis that substrate legs do not possess a spine fold for the claw. affinities do exist in some species. The data col­ Another species, Pariambus minatns (Mayer), is lected from the various habitats showed the most known to live, at least for the most part, on common caprellid of the lower York River, dur­ sand. Mayer does agree, however, that the vast ing the months of June and July, to be Caprella majority of caprellids do prefer a living sub­ geometric« Say, which was found in abundance strate. Caprella equilibra was reported by Mayer on nearly all of the substrates examined. This from the tunicate Ciona intestinalis (L), where species was particularly dominant on hydroids, he observed it sitting near the incurrent opening bryozoans, and sponges. Paracaprella tenuis where it could utilize the water current and the Mayer was the species next in abundance during outer muddy surface for feeding. Patton ( 1968, these 2 months, being dominant on bryozoa. p. 119-120) described the caprellid , Caprella The third species was Caprella equllibra, which was not present on any substrate examined prior 1 Based on a portion of a dissertation submitted to early August, at which time it replaced C. by the auth or in partial fulfillment of the require­ geometries as a dominant on sponges. Para­ ments for the doctoral degree in Biological Sciences caprella tennis seemed to show an affinity for at the University of Southern California. Manuscript the brown bryozoan Victorella pavida Kent. received June 30, 1970. 2 Tex as Christian University, Department of Biol­ This species also occurred in large numbers on ogy, Fort W orth, Texas 76129. the sponge, H aliclona permollis Bowerbank, as 387 388 PACIFIC SCIENCE, Vol. 25, July 1971

it began to grow over the bryozoan in late July. tected. Samples from Punta Penasco, Mexico, The sample number was insufficient for statis­ and Santa Catalina Island were also utili zed. tical verification. The results did encourage Caprellid abundance was expr essed in number fur ther investigations into the caprellid-substrate per unit volume for each substrate. relationships. Locomotion should be considered in studies Seasonal Effects at Long Beach Marin a of substrate selection since the ability of an orga­ Long Beach Marina is situated in Alamitos nism to move from one place to another might Bay; it is a small body of water used primarily influence the results of experiments. The as­ for recreational purposes by the city of Long sumption is made that an individual is able to Beach. For a further description and history of move from one substrate to another, selecting Alamitos Bay, see Reish and Winter, 1954, pp . the one which is most suitable. 106--107. Caprellids move over their substrate with ease. To determine seasonal effects on the total They are also capable of swimming in an ante­ caprellid populati on and their abundance on riorly directed manner by quick, jerky move­ various substrates, samples were taken from ments involving sudden ventral flexing and November 1965 through October 1967 from straightening of the body. the sides of the floats which support the wooden When placed in an aquarium, the caprellids boat docks. The floats are constructed of rein­ often appear to have no direction to their swim­ forced molded fiberglass plastic (Reish, 1964, ming ; they sometimes swim to the top of the p. 126) . A total of 144 samples was taken at water where they become trapped by the surface this station from April 1966 to October 1967. tension. In their natural habitat, however, ca­ Twen ty-five samples were taken durin g each prellids swim with a sense of direction. When sampling period and preserved separately in 10 the bryozoan Bugula neritina (L) was detached percent formalin. The caprellids were removed, from the dock float at Long Beach Marina so counted, and stored in 70 percent alcohol. It that it began to sink, the caprellids swam up­ was found that the most effi cient means of re­ ward from the Bttgttla to the other bryozoan moving the caprellids was to shake the preserved colonies attached to the floats. sample in a half-gallon jar filled about one-third Observations of caprellids kept in the lab­ with water, and then quickly removing the sub­ oratory show that they move from substrate to strate and pouring the water through a screen substrate, stopping for varying length s of time. with a pore size of .0070 mm. This process was Th eir ability to swim and to move readily over repeated until caprellids could no longer be ob­ the substrate eliminated locomotion as a problem tained in this manner. The substrate was exam­ in substrate selection experiments. ined under a dissecting microscope and any Feeding studies by Keith (1969) showed that remaining caprellids removed. Caprellid abun­ caprellids in captivity feed principally on detri­ dance was expressed as number per cubic centi­ tus and diatoms which are associated with the meter of substrate. During each samplin g period substrates. Food was removed from the sub­ the water temperature and relative substrate strates prior to each experiment to eliminate it abundance were noted. as a possible factor influencing selection. Substrate Selection

METHODS AND MAT ERIAL S LABORATORY STU DIES : Caprellids were col­ lected along with their substrates at Long Beach Occurrences and Substrates Marina from July 5 to September 8, 1967 to Samples were collected along the Southern determine if substrate affinities could be detected. California coast duri ng Ap ril 1966 from Dana There were three principal substrates on which Point to San Simeon, and all of the information caprellids could be found in abundance at this available from the literature on caprellid occur­ location : the bryozoan Bngnl« neritina and the rences was tabulated in anticipation that certain algae Polysiphonia pacifica H ollenberg and Viva caprellid-substrate relationships might be de- lobata (Kiitzing) .Samples of these substrates Substrate Selection in Caprellid Amphipods-KEITH 389 were collected and placed in large plastic bags the influence of one species on the substrate along with a generous supply of seawater and selection of the other. transported to the laboratory in Styrofoam Approximately equal volumes of each of the containers. The caprellids and their respective three substrates, from which all of the caprellids substrates were kept in separate aquar ia. The and food had been removed, were placed in the environmental temperature was determined each experimental tanks so that they made contact time specimens were collected so that the tem­ with one another. Twelve specimens were ob­ perature of the experimental tanks could be tained from Bngula neritina and four were adjusted to that of the environment. placed on each of the three substrates and ob­ Th e experimental tanks consisted of four 3­ served over a 48-hour period. Seven observa­ gallon plastic Aqu aflair aquaria partially sub­ tions, which consisted of counting the number merged in a 60-X -7-inch insulated Plexiglas of caprellids on each substrate, were made dur­ tank designed to hold four of the small aquaria ing this period at approximately equal intervals end to end. The large vessel was filled with during the daytime. These numbers were then distilled water and acted as a water bath to averaged to give a figure representing one trial. maintain the experimental tanks at the desired Averaging the number of independent counts temperature. help ed to reduce the error due to periodic wan­ Th e water bath was cooled by circulating dering of individuals. Th e substrates were care­ distilled water from a beverage cooler through full y separated prior to counting to prevent the one-half-inch plastic tubing by means of a model caprellids from moving from one to the other 2U March submersible water pump (Bakus, dur ing the count. Af ter counting, the substrates 1965, pp. 230-231) . The tubing made two were again placed in contact. Ten trials were loops in the bottom of the water bath and completed during one experiment and a t-test occupied almost its entire length . The tempera­ was used to determine the significance of the ture was controlled by using a thermostat on difference in numbe rs occurring on the thr ee the cooler and also by adjusting the flow of substrates. The entire experiment was repeated water through the tubing. Th e temperature was with specimens originally obtained from Poly­ kept as near as possible to that of the natural siphonia pacifica and again with specimens from environment and ranged from 18° C on July 5 V Iva lobata to determine if the "parental" sub­ to 22 ° C on September 16, 1967, when the last strate was influencing the results of the selec­ selection experim ent was completed. tion experiments. Seawater for the aquaria was obtained from Experiments were repeated with Caprella Marineland of the Pacific. Air was supplied to equilibra. Because of the scarcity of this species each aquarium by an electric pump. A fiber­ at Long Beach Marina, a separate experiment glass-charcoal filter was employed in conjunction was not conducted for specimens taken from the with the air supply to keep the water free of three substrates. Compensations were made for suspended particles. The tanks were cleaned the influence of the original substrate by using daily by removing fecal pellets and other settled individuals collected from each of the substrates material from the bottom with a syringe. At rath er than from a single one for all 10 trials. the end of each trial of an experiment, one-half To determine the influence of one species on the volume of each tank was removed and re­ the substrate selection of the other, two indi­ placed with fresh seawater. The total volume viduals of each species were placed on each of was changed following the end of the second the three substrates. The same observational trial in the same tank. procedure was used as for the previous experi­ Experiments were conducted to determine: ments. (1) if Caprell« califomica and Caprella equi­ FIE LD STUDIES: Ten samples of the thr ee libra would show an affi nity for one of the prin­ principal substrates, Bugnla neritina, Polysi­ cipal substrates, ( 2) the influence on the selec­ ph onia pacifica, and VI va lobata were taken tivity of the caprellids of the substrate from from Long Beach Marina during October 1967. which they were originally obtained, and (3 ) The abundance of Caprella calii ornic« and C. 390 PACIFIC SCIENCE, Vol. 25, July 1971 equilibra on each of the substrates was compared The differences in substrate associations over a statistically and the results correlated with those fairly extensive area obscure the detection of previously obtained in the laboratory, caprellid-substrate affi nities. Results from this study made it evident that, if substrate prefer­ Size and Substrate Selection ences were to be detected, quantitative studies To determine if any correlation existed be­ would have to be conducted within specific tween size and substrate, 72 samples were col­ communities where environmental parameters lected during August 1967 at Long Beach were relatively uniform at a given time. Marina. A single caprellid species, C. calij or­ nica, was used. The average size of the indi­ Seasonal Effects at Long Beach Marina viduals was determined for each of 10 samples The most dense population of C. calijornica from the three substrates. Measurements were occurred on all substrates during the spring. made from the anterior edge of the cephalon The lowest number recorded per unit volume to the tip of the abdomen. occurred during the winter. The total population It was decided that a valid average for one of C. equilibra varied littl e in 1967, except for sample should consist of approximately 100 in­ a slight increase during the spring. dividuals or more. If a sample had less than 80 Temperatures recorded at Long Beach Ma­ specimens, another sample was combined, and rina during the 1967 sampling periods ranged so forth, until the population approached 100. from 15.5° C in February to 22 ° C in August. Every specimen 2 mm or larger in length was The abundance of the substrates Bsgul« neritina measured. In most cases, specimens smaller and Polysiphonia pacifica varied considerably. than 2 mm could not be positively identified to Du ring the winter their densities were very low, species. Ten averages were obtained for each but as spring approached they began to increase. of the three substrates. The number of indi­ Polysipbonia reached its peak of abundance in viduals per sample ranged from 85 to 214. A April and Bugula reached its peak in July, be­ t-test was applied to the data to determine the coming the dominant growth on the floats. A significance of the difference in mean size of rapid decline in both of these substrates occurred the caprellids living on each of the three sub­ in August. V iva lobata showed only a slight in­ strates. crease during the winter and began to decrease again as temperatures reached about 19 ° C in July. RESULTS AND DISCUSSION Seasonal variations in the total caprellid pop­ Occurrence and Substrates ulation did not appear to be controlled directly Results obtained from collections along the by fluctuations in substrate abundance, because Southern California coast showed that caprellids the number of caprellids per unit volume of the occupy a very wide variety of substrates. The substrate was low during the winter when the principal ones are algae, bryozoans, hydroids, substrate abundance was also low. If substrate sponges, tunicates, eelgrass, and gorgonians. availability had been the limiting factor, the Although found primarily on living substrates, number of caprellids per unit volume would caprellids are also found in marine muds and in have increased as the amount of substrate de­ some instances on the frayed ends of ropes hang­ creased. Reproduction occurs year around, but ing into the water from boat docks. the maximum rate in Caprella calijornica must Caprellids are most commonly found inter­ occur in the early spring when an enormous tidally in tide pools, subtidally, and among foul­ population increase is evident. ing communities on pier pilings and dock floats. They have been reported, however, from depths Substrate Selection of 3,501-4,00 4 meters (McCain, 1968, p. 91). LABORATORY STUDIES : The results of experi­ Local differences in environmental conditions ments to determine if C. calijornic« exhibits a such as wave action and temperature seem to be substrate affinity, and the influence of the sub­ the primary factors affecting species distribution. strate from which the caprellid was taken on Substrate Selection in Caprellid Amphipods-KEITH 391 its selection, are summarized in Tables 1, 2, ference in abundance to arise between this sub­ and 3. strate and Viva lobata (P < .001). Table 1 indicates a highly significant differ­ Table 3 shows that Capretta calijornic« ob­ ence (P < .001) between the number of C. tained from Ul va also exhibited an affinity for californica selecting Bugula and the other two Btlgttla ( P < .001) , but again the influence of substrates. There was no significant selection the parental substrate can be seen.The number shown, however, between Polysiphonia and of individuals selecting V iva was greater among V iva. specimens originally collected from V iva (P < W hen specimens were taken from Polysipho­ .001) . nia pacifica (Table 2), a strong affinity still The results of experiments to show if Ca­ existed for Bugul« neritina (P < .001) . A def­ pretta equilibra exhibited an affinity for any of inite effect of the parental substrate, however, the three substrates are summarized in Tabl e 4. can be seen since a greater percentage of indi­ This table shows that C. equilibra did not select viduals selected Bugul« when they were orig­ Btlgula neritlna over Polysiphonia pacifica as inally taken from this substrate (P < .001). did Capretta californica. Both of these sub­ The increase in numbers of individuals selecting strates, however, were favored over V iva lobata Polysiphonia pacifica caused a significant dif- ( P <.OOl ) .

T A BLE 1

AVERAGE NUMBER OF Caprella calijornica O CCURRIN GON EAC HSUBSTRATE

TRIALS SELECTED SUBSTRATES 2 3 4 5 6 7 8 9 10 Bugula neritina 10 9 11 11 11 11 10 11 10 9 Polysiphonia pacifica 2 2 1 1 0 1 2 1 0 1 Ulva Lobata 0 1 0 0 1 0 0 0 2 2

NOTE: Specimens were obtained from Bngula nerltina ,

T ABLE 2

AVERAGE N UM BER OF Caprella cali/ornica OCCURRIN G ON EACH SUBSTRATE

TRIALS SELECTE D SUBSTRATES 2 3 4 5 6 7 8 9 10 Bugula neritina 7 8 9 8 8 7 8 9 9 8 Polysiphonia pacifica 3 4 3 3 3 3 3 3 3 4 V iva lobata 1 1 0 0 1 1 1 0 0 0

NOTE: Specimens were obtained from Polysipbonla pacifica.

T A BLE 3

AVERAGE N UMBEROF Caprella calijornica OCCURRIN G ON EA CH SUBSTRATE

TRIA LS SELECTE D SUBSTRATES 1 2 3 4 5 6 7 8 9 10 Bugula neritina 6 7 7 7 6 7 7 8 7 8 Polysiphonia pacifica 4 1 4 3 3 1 3 1 3 2 Ulva lobata 2 3 1 3 4 4 2 3 2 1

NOTB: Specimens were obtained from VI va lobata. 392 PACIFIC SCIENCE, Vol. 25, July 1971

TABLE 4

AVERAGE N UMBER OF Caprella equilibra OCCURRING ON EACH S UBSTRATE

TRIALS SEL ECTED SU BSTRATES 2 3 4 5 6 7 8 9 10 Bugula neritina 6 3 4 3 3 6 6 4 3 6 Polysipho nia pacifica 3 5 5 8 7 5 4 5 8 4 Viva lobata 2 4 1 1 1 0 1 3 1 1 NOTE : Specimens were obtained from all three substrates.

The same results were obtained when the two least two means of regulating its color: (1) by species were placed together as when they were cuticular tanning and (2) by physiological color separated, indicating that, at least under condi­ changes brought about by expansion and con­ tions of the experiment, there was no interaction traction of chromatophores. C. equilibra was not between the two species which affected their observed to undergo tanning and was unable to substrate selection. Caprella californica selected adapt completely to dark substrates. Bugula over Polysiphonia (P < .01) , Caprella C. californica was able to adapt somewhat to equilibra showed no selectivity between Bagul« the color of the substrate within about 5 hours, and Polysiphonia, and both species exhibited but adaptation did not appear to be complete the least affinity for Viva. until after ecdysis. The cuticular coloration ap­ The preference of Caprella californica for pears to be permanent. Bu gttla seems to be best explained by the mor­ Ul va appeared to be the least "attractive" of phological similarities between this species and the three substrates. The broad thallus probably B. neritina. The coloration and external anatomy makes it more difficult for caprellids to grasp of this caprellid species appears quite similar to than that of filamentous forms. It also seems the bryozoan. The margins of the segments are that the caprellids studied could not adapt cryp­ often darker giving the illusion of thecae. In tically as well to V Iva as the other two sub­ females of this species, tubercles occur on the strates. Many specimens were found, however, dorsal surface of the thoracic somites which on Viva at Long Beach Marina. also aid in this illusion. Caprella equilibra did not show a preference I suspect that, as conditions become crowded, for Bugula or Polysiphonia. This result was not certain members of the Caprella californica surprising since Caprella equilibra is not as population move from Btlgttla to other sub­ highly adapted morphologically to Bltgula neri­ strates such as Polysiphonia and Viva. Because tina as is Caprella californica. It should be the data from selection experiments show that pointed out, however, that in other areas where the parental substrate affects the substrate selec­ different substrates occur, C. equilibra might tion of some individuals, one might infer that have strong affinities for certain ones. Because caprellids can adapt somewhat to their new of its transparent, slender body, this species substrate, resulting in less affinity for Btlgula. appears to be well adapted to hydroids. Adaptation, however, to the degree that the These two species did not seem to affect one affinity of Caprella californica for Btlgula is another's selection of a substrate under labora­ reduced, may occur over a relatively long period tory conditions, but it was suspected that, if of time since most of the individuals on Poly­ conditions had been crowded, C. equilibra would siphonia pacifica and Ulva lobata still select have been more abundant on Polysiphonia than Bltgula neritina. Bugal«. It is logical to assume that, since Ca­ Studies on adaptive coloration supported this prella equilibra showed no difference in its thesis. Caprella calif ornica and C. equilibra affinity for these two substrates, it would be possess two types of chromatophores: rnelano­ the first to leave Btlgttla under crowded condi­ phores and guanoph ores. C. californica has at tions. Substrate Selection in Caprellid Amphipods-KEITH 393

TABLE 5 principal substrates is summarized in Table 6. NUMBER OF Caprella calijo rnica AND Caprella Each number represents an average of 85 to equilibra P ER C UBIC C ENTIMETER OF S UBSTRATE 214 specimens in terms of ocular micrometer AT LO NG B EACH M ARIN A DURIN G F ALL 1967 units. These units may be converted to milli­ meters by dividing by six. A total of 3,756 SUBSTRATES C. calijornica C. equilibra specimens were measured. 51 33 33 1 The results show that no significant difference 41 36 1 7 exists in the size of specimens found 'on Bugul« Bugul« neritina 57 52 0 1 44 34 1 2 neritina and Polysiph onia pacifica. A significant 28 46 4 18 difference does exist, however, between the mean 17 16 5 15 size of individuals occurring on Ulva lobata and 13 20 23 8 the other substrates, those on Ulva being larger Polysipbonia pacifica 30 19 3 22 (P< .Ol). 4 9 12 10 20 21 9 6 CONCLUSIONS 3 2 0 0 18 17 0 0 1. California caprellids are able to occupy a U/ va /obata 21 24 0 0 7 10 0 0 wide variety of substrates. The majority of the 13 16 0 0 substrates are living organisms such as hydroids, bryozoans, algae, and sponges; but living sub­ FIELD STUDIES: Data obtained during the fall strates are not a requirement. of 1967 at Long Beach Marina to evaluate the 2. Seasonal studies showed density fluctua­ laboratory findings are summarized in Table 5. tions in both the caprellids and their substrates. Each number represents the abundance of caprel­ Substrate abundance showed a positive correla­ lids expressed in number per cubic centimeter tion with the total caprellid density but was not of substrate. The results showed that Caprel/a considered to be the principal factor limiting calijornica was more abundant on Bugttla neri­ the caprellid population size. tina than Polysiphonia pacifica or Ulva (P < 3. The most important substrates for caprel­ .001). No significant difference, however, was lids at Long Beach Marina seemed to be the found in numbers occurring on Polysiph onia bryozoan Bugula neritina and the algae Poly­ pacifica and Ulva. Caprel/a equilibra showed no siphonia pacifica and Ulva lobata. significant preference for Polysiph onia or Bu­ 4. Two caprellid species, Caprel/a calijornica gula, but was more abundant on either of these and Caprella equilibra, occurred at Long Beach substrates than Ulva . The conclusions of the Marina. C. calijornica showed a highly signifi­ field studies agreed with those obtained from the cant preference for the bryozoan Bugula neri­ selection experiments. tina. Caprella equilibra did not demonstrate any selective preference between Bugula and Poly­ Effect oj Size on Substrate Selection siphonia but did seem to prefer these to Ulva. The average size of all Caprel/a calijornica 5. The substrates from which the caprellids occurring in each of 10 samples of the three were collected did have some influence on their

TABLE 6

AVERAGE SIZE OF Caprella calijornica

SAMPLE S

SUBSTRAT ES 2 3 4 5 6 7 8 9 10 Bugula neritina 34 34 34 34 39 36 33 40 38 32 Polysipbonia pacifica 30 34 36 25 30 38 23 42 36 28 U/va /obata 35 31 35 49 43 47 45 44 37 40

NOTE: Size given in terms of ocular micrometer units. Units may be converted to millimeters by dividing by six. 394 PACIFIC SCIENCE, Vol. 25, July 1971 substrate selections in the laboratory, but the LITERATURE CITED preference for Bugula by Caprella califomica BAKUS, G. J. 1965. A refrigerated seawater was apparently so strong that it was evident system for marine organisms. Turtox N ews, even in individuals taken from the other sub­ vol. 43, no. 9, pp. 230-231. strates. KEITH, D. E. 1969.Aspects of feeding in Ca­ 6. UIva lobata was found to have individuals prella califomica Stimpson and Caprella eqni­ of larger average size than the other two sub­ libra Say (Amphipoda). Crustaceana, vol. 16, strates. This phenomenon was attributed to the no. 2, pp. 119- 124. nature of the thallus. MAYER, P. 1882. Die Caprelliden des Golfes 7. Th e affi nity which was demonstrated by von N eapel. Fauna and Flora des Golfes von Caprella califomica for Bugula nerltina was N eapel, vol. 6, pp. 1- 201. attributed to its cryptic adaptations to that sub­ MCCAIN, J. C. 1968. The Caprellidae (Crus­ strate. tacea: Amphipoda) of the Western North Atlantic. Bulletin. United States N ational Museum, vol. 278, pp. 1- 147. PATTON, W. K. 1968. Feeding habits, behavior ACKNOWLEDGMENTS and host specificity of Caprella grahami, an amphipod commensal with the starfish Aste­ Appreciation is expressed to the Allan Han-. rias forbesi. Biological Bulletin, W oods cock Foundation, University of Southern Cali­ Hole, vol. 134, no. 1, pp. 148-1 53. fornia, for the research facilities and financial REISH, D. J. 1964. Studies on the M ytiltls edulis support which made this study possible, and to community in Alamitos Bay, California: 1. Drs. Gerald J. Bakus and John S. Garth, both Development and destruction of the com­ of the Allan Hancock Foundation, for their munity. Veliger, vol. 6, no. 3, pp. 124-131. advice. The author also acknowledges the in­ REISH, D. J., and H. A. WINTER. 1954. The fluence of Dr. Marvin L. W ass of the Virginia ecology of Alamitos Bay, California, with Institute of Marine Science, who stimulated his special reference to pollution. California Fish interest in substrate selection among caprellids. and Game, vol. 40, no. 2, pp. 105-121.