Persistence and Transport of Fauna on Drifting Kelp (Macrocystis Pyrifera (L.) C

Home , Melibe leonina, Pugettia producta

Journal of Experimental Marine Biology and Ecology 253 (2000) 75±96 www.elsevier.nl/locate/jembe

Persistence and transport of fauna on drifting kelp (Macrocystis pyrifera (L.) C. Agardh) rafts in the Southern California Bight

Alistair J. Hobday1 Scripps Institution of Oceanography, 9500 Gilman Drive, La Jolla, CA 92093, USA Received 9 April 1999; received in revised form 19 June 2000; accepted 29 June 2000

Abstract

Drifting rafts of Macrocystis pyrifera may connect isolated kelp forests in the Southern California Bight. To determine which species might utilize this dispersal mechanism, faunal samples from natural drifting rafts and attached M. pyrifera plants were collected during ®ve cruises between March 1995 and December 1997. These rafts, which can be considered as ¯oating islands, were aged and the macroinvertebrate assemblage enumerated. There was no signi®cant relationship between raft age and species richness, or between species richness and distance offshore, which contrasts with predictions based on island biogeography. Species richness, however, was related to raft weight. Patterns of species presence and density were investigated relative to raft age for the species most frequently associated with rafts. Only one species, the isopod Idotea resecata, was found on all sampled rafts. Some species increased in frequency with raft age and others decreased, but only one relationship, a decline in the frequency of the anemone Epiactis prolifera with raft age was signi®cant. When species density was examined over all cruises, only I. resecata had a signi®cant change in density (an increase) with raft age, but additional signi®cant relationships were found when species density patterns were considered by cruise. The results of all the tests were combined to provide a measure of ``raft success''. Nine of the most frequent 19 species had a positive score, indicating a favorable response to rafting, seven were unaffected, and two species had negative responses to rafting. Extinction times were calculated using species abundance and raft age relationships. Two species (E. prolifera and Paracerceis cordata), were predicted to persist on rafts for only about 100 days, which is the maximum estimated raft lifetime. All other species were predicted to persist for longer periods if the rafts ¯oated longer. Kelp fauna that begin rafting appear to be largely unaffected by rafting, and hence dispersal on kelp rafts is possible for many members of the kelp forest community.  2000 Elsevier Science B.V. All rights reserved.

Keywords: Algal rafts; Extinction rates; Faunal dispersal; Island biogeography

E-mail address: [email protected] (A.J. Hobday). 1Present address: CSIRO Marine Research, P.O. Box 1538, Hobart, Tasmania 7001, Australia.

1. Introduction

Attached algae provide habitat for many invertebrates (e.g. Mukai, 1971; Gunnill, 1982; Edgar, 1983; Smith et al., 1996). These habitats and their invertebrate assemblage are often isolated from similar patches by varying distances of non-algal habitat. Alga that detaches from the substrate, if it ¯oats, may drift at the ocean surface for a period of time before beaching or sinking (Hobday, 1998, 2000b). The possibility that algal rafts act as a faunal dispersal mechanism and connect isolated populations has been recognized for a long time (Vallentin, 1895; Fell, 1962; Benech, 1978; Highsmith, 1985; Johannesson, 1988; Helmuth et al., 1994; Worcester, 1994). Because movements between algal habitats can occur by active movements or free dispersal in particular life stages (e.g. planktonic larvae), animals with completely non-planktonic life cycles may rely more heavily on rafts for dispersal. While rafts have been studied for a long time, the majority of studies have simply documented the ®sh and invertebrate fauna (Senta, 1966; Gooding and Magnuson, 1967; Hunter and Mitchell, 1967; Ida et al., 1967; Mitchell and Hunter, 1970; Weis, 1968; Fine, 1970; Dooley, 1972; Kingsford and Choat, 1985; Tully and Ceidigh, 1986; Nakata et al., 1988; Fedoryoka, 1989; Kingsford, 1992; Davenport and Rees, 1993; Ingolfsson, 1995), however, changes in community composition are likely as the raft ages and drifts farther from the point of origin (Helmuth et al., 1994; Hobday, 1998; Ingolfsson, 1998). Macrocystis pyrifera is a large brown alga growing in temperate waters in both the northern and southern hemisphere and forms large forests inhabited by a rich ®sh and invertebrate fauna (e.g. Jones, 1971; Ebeling et al., 1980; Coyer, 1984; North, 1994). Attachment to the substrate is with a root-like holdfast and a canopy composed of stipes and blades extends to the surface (Clendenning, 1971; North, 1971). The holdfast- associated faunal community has attracted most study (Andrews, 1945; Ghelardi, 1971; Jones, 1971; Ojeda and Santelices, 1984), while the canopy community has achieved less attention, due in part to the dif®culty of sampling the large biomass (but see Coyer, 1984). Rafts of M. pyrifera form when plants become detached from the substratum and ¯oat to the surface (Dayton, 1985; Harrold and Lisin, 1989; Tegner et al., 1995; Hobday, 1998, 2000b). Studies of detached M. pyrifera plants have also focused on the more easily sampled holdfast community and have often considered temporal changes following detachment (Ojeda and Santelices, 1984; Edgar, 1987; Vasquez, 1993). Edgar (1987) noted successional and dominance changes in inhabitants of the kelp holdfasts, perhaps due to the absence of kelp forest predators such as ®sh. Perhaps the most complete study to date of the fauna of drifting kelp rafts is that of Bushing (1994) who sampled 109 drifting brown algal rafts (81 of M. pyrifera), within a few kilometers of Santa Catalina Island, California, between 1969 and 1972. A total of 179 invertebrate and 25 vertebrate species were associated with those drifting rafts. The faunal sampling, however, was not quantitative and age or size of drifting rafts was not measured. Like Highsmith (1985), Bushing postulated that for species with limited larval dispersal, drifting rafts might be the most important dispersal mechanism. Rafting will only be a successful dispersal mechanism for those species that can persist on the ¯oating object. Initially the raft community is likely to be similar to that of A.J. Hobday / J. Exp. Mar. Biol. Ecol. 253 (2000) 75 ±96 77 attached algae. Some losses may result during detachment and mobile animals may subsequently leave the drifting raft. As the raft drifts, changes in predation and competition may also in¯uence species abundance. Members of the pelagic ¯otsam community, colonizing as larvae, juveniles or adults may join the raft community (Helmuth et al., 1994). Pelagic species may primarily use rafts as substrate for completing a portion of their life cycle rather than for transport (Hobday, 1998). Because of the large size of M. pyrifera rafts complete sampling has never before been attempted. In this study patterns of macrofaunal species presence and density on natural rafts of M. pyrifera were examined with regard to raft age. The methodology for aging M. pyrifera rafts is described in Hobday (1998, 2000b).

2. Methods

2.1. Sample collection

Drifting M. pyrifera rafts were collected during cruises on ships of opportunity in the Southern California Bight between March 1995 and December 1997. Raft location was recorded with the ship's differential GPS system and the distance to the closest land (island or mainland) calculated. The focus of this study was the macroinvertebrate assemblage associated with kelp rafts, which dictated the sampling style, although juveniles of some ®sh species were also well collected. Two different net designs, using a modi®ed Sigsbee trawl with mesh size of 6 mm, were used to collect rafts. The ®rst design, the Rumsey Net, had a ®xed 3.3 3 3 m rectangular mouth and a tail about 10 m long, and operated as a large neuston net. The second version, the Giant Rumsey Net, was implemented to better capture the associated juvenile ®sh that often school beneath kelp rafts. This net was similar to a large dip net, with the net mouth attached to a rigid 2.5 m diameter circular frame (Hobday, 1998). When a raft of suitable size was observed, the ship slowed and deployed the net using the ship crane and maneuvered alongside the drifting raft which drifted into (design 1) or over (design 2) the net mouth. The mouth was then lifted clear of the surface and net and raft craned aboard. Rafts were transferred through the net cod-end to a large sorting bin (1 m 3 2m3 1 m) with two layers of coarse screens at the bottom (6 cm 3 3cmand1cm3 1cm).A combination of salt and fresh water was used to wash the mobile animals from the kelp, through the coarse screens, out a hole in the bin base and into a 333 micron sieve. Representatives (and the largest) of each of the attached invertebrate species (e.g. Lepas barnacles) were removed by hand as the kelp was removed for weighing. Raft weight was measured in a bin with a spring scale. Stipes with terminal ends (between 1±10 stipes per raft) were set aside to determine the age of each raft according to methods described in detail in Hobday (1998, 2000b). Brie¯y, raft age was calculated using the average blade length on collected rafts, the average blade length on attached plants, and the rate of change in blade length as rafts drift. If raft age was estimated as negative because the raft had longer blades than the average attached plant, raft age was set at 78 A.J. Hobday / J. Exp. Mar. Biol. Ecol. 253 (2000) 75 ±96 zero. A maximum of two to three rafts could be collected at sea per day using these methods. In order to interpret the faunal patterns observed on drifting rafts, an understanding of the community prior to detachment is desired. However, because rafts were collected from ships of opportunity, dedicated ship time to support nearshore sampling of the potential source of attached plants was possible on only one occasion. In February 1996 attached M. pyrifera plants were collected from the four northern Channel Islands which are a likely source area for many of the Southern California Bight rafts (Hobday, 1998, 2000a). Attached samples were collected by scuba divers using a 2 m diameter 10 m long 333 micron mesh net. The canopy of a plant of suitable size was covered at the surface with the net; two divers then swam the net down over the plant, detached the plant at the base of the holdfast, and sealed it inside the net. The sample was brought to the surface inside the net and transferred to the ship where the kelp was weighed and fauna removed as described for the raft samples. Faunal samples were preserved in formalin buffered with sodium borate. Within three weeks samples were washed and transferred to alcohol for sorting at a later date. Large samples were divided with a plankton splitter during sorting. All ®sh and a variety of macrofaunal invertebrate taxa were either enumerated (mobile taxa), or recorded as present (colonial and sessile taxa). Copepods, amphipods (with the exception of Caprella californica) and polychaetes were not identi®ed, as the sampling may not have been complete for these groups. The total length of all ®sh collected was recorded, as was the carapace width of crabs, the length of isopods, and the maximum size of Lepas spp.ina sample. A collection of voucher specimens was made and archived at the Scripps Institution of Oceanography.

2.2. Analyses

ANOVA and MANOVA were used to evaluate differences between cruises in raft weight, age, number of taxa and the distance offshore that rafts were collected. Taxa that were present in eleven or more of the total samples (n 5 58) were analyzed further, hereafter called the ``top species''. Top species patterns as a function of raft age were analyzed with linear regression. Linear regressions were used as the simplest model, even though in some situations exponential or polynomial models may have provided a better ®t. Because the raft age is estimated with error (Hobday, 1998, 2000b), regressions using raft age as the independent variable are Model II problems (Sokal and Rohlf, 1981). If, as is the case here, the error associated with raft age is small compared to the error in the density measures regressed on raft age, then the error in raft age can be ignored and Model I least squares regressions is appropriate (Draper and Smith, 1981). Species presence on rafts divided into age class bins for all cruises and samples combined showed the frequency of rafts in an age range that will have the species aboard. Frequency data was arcsine (œp) transformed before statistical analyses. The density of each species was calculated for each sample to correct for differences in raft weight. Differences in species density on rafts were compared between cruises with MANOVA, with cruise and age as factors. Samples from all cruises were combined, and A.J. Hobday / J. Exp. Mar. Biol. Ecol. 253 (2000) 75 ±96 79 linear regression analyses used to examine changes in the density of each of the top species with raft age. To identify seasonal patterns the relationship between the densities of each of the top species and raft age was also examined with linear regression for each cruise. All density data was log(x 1 1) transformed which corrected for non-normality. Relationships between average size of the top species and the age of the raft were examined with linear regression. Combining the test results for the top species summarized the strength of the response to rafting. Species were ®rst grouped by habitat preferences. If a species was more than 50% more frequent in one category of samples (raft or attached), it was allocated to that habitat. Species that did not differ in frequency of occurrence by more than 50% were grouped as ``both''. Raft (attached plant) species were given a score of 1 (21), while those common on both were scored as 0. Secondly, species that had a statistically signi®cant positive (negative) relationship between frequency of occurrence and raft age were given a score of 1 (21). Thirdly, the species that had a signi®cant positive (negative) slope of density with raft age were given a score of 1 (21). Finally, for each of the species-by-cruise analyses, species that had a signi®cant positive (negative) slope of density with raft age were given a score of 1 (21). These four scores were summed to provide an index between 8 and 2 8. Positive scores indicated species that were successful on rafts (colonizing or increasing), while species with negative scores were unsuccessful on rafts (declining or extinct). Scores of zero indicated that the species was unaffected by rafting (persistent). These arbitrary scores divide the species that were successful, unaffected, or unsuccessful regardless of the relative value given to each test. The extinction times for species on rafts was evaluated using the linear relationships between raft age and the three measures of abundance. The point where the intercept was at zero was de®ned as the extinction time. All tests were carried out using Systat 5.0. No correction for multiple testing of raft samples for each species was made. An alpha of 0.05 was used as the level of statistical signi®cance.

3. Results

3.1. Sample collection

A total of 50 rafts were collected during ®ve cruises within the Southern California Bight between March 1995 and December 1997 (Fig. 1). Twenty six rafts (samples 7±32) were collected with the Rumsey Net and 24 (samples 33±56) were collected with the Giant Rumsey Net. Age was determined for these samples and ranged between 0 and 72 days (Hobday, 1998, 2000b). Rafts were collected at an average distance of 33.2 km from the nearest coast (range, 5.56±82.4 km) and weighed an average of 55.5 kg (range

2±160 kg). There was no difference between cruises for rafts weight (ANOVA F4,45 5 0.516, P 5 0.725), age (ANOVA F4,45 5 1.915, P 5 0.124), or the distance offshore rafts were collected (ANOVA F4,45 5 1.659, P 5 0.176) (Table 1). Two plants growing in water depth of 15±18 m were collected from each of the four northern-most channel 80 A.J. Hobday / J. Exp. Mar. Biol. Ecol. 253 (2000) 75 ±96

Fig. 1. Location of M. pyrifera raft samples collected during ®ve cruises between March 1995 and December 1997. Two attached plants were collected from each of the four northern Channel islands during the 9602 cruise. Island names are shown.

Table 1 Collection summary of M. pyrifera rafts and attached plants by cruisea Cruise Season Sample [ Average Average Average age Average number weight distance to coast (days) of species (kg) (6S.D.) (kin) (6S.D.) (6S.D.) (6S.D.) 9503 Winter 7±16 53.6638.1 37.1622.1 30.4625.1 7.163.54a 9509 Summer 17±32 62.5645.9 38.1616.6 10.1618.3 14.766.22c 9602 Winter 33±39 54.3630.3 44.7630.1 30.2627.4 12.763.55bc 9602 Winter Attached 1±8 57.3624.0 0 0b 14.562.67 9706 Summer 40±49 59.5638.2 41.1610.6 15.1618.5 12.56337bc 9712 Winter 50±56 37.8632.1 27.86 8.0 15.5623.9 9.663.82ab a The cruise code represents the year and month of the cruise, e.g. 9503 is March 1995. Sampling periods that were not signi®cantly different in terms of species per raft (Fishers LSD, P , 0.05) are denoted by the same letter superscript. b Average age of attached plants is de®ned as 0 days. A.J. Hobday / J. Exp. Mar. Biol. Ecol. 253 (2000) 75 ±96 81 islands during the February 1996 cruise (Fig. 1). These plants weighed an average of 57.3 kg (range 23±102 kg) (Table 1).

3.2. General species patterns

A total of 72 macrofaunal species or taxa were enumerated from the attached (n 5 36) and raft samples (n 5 64) (Table 2). At small sizes two members of the shrimp genera Hippolyte (H. clarki and H. californiensis) could not be distinguished and so counts for these two species were pooled. Sixteen species of ®sh were collected as juveniles or adults and the eggs of Cypselurus californicus (¯ying ®sh) were also found attached to kelp rafts collected during summer cruises.

Both cruise (F4,425 4.86, P , 0.003) and weight (F 1,42 5 10.95, P , 0.002) were signi®cant predictors in a MANOVA model of the species richness on rafts, but age and distance offshore were not (P . 0.367 for both). The two interaction terms examined (cruise*weight, and cruise*age) were both non-signi®cant (P . 0.610 for both) and dropped from the ®nal model. A post-hoc Fishers LSD test showed no clear pattern in terms of richness and cruise season (Table 1). Because there was no difference in raft age or weight by cruise, samples were pooled for some additional analyses. There was no signi®cant relationship between the age of the rafts and the number of taxa recorded (Regression F1,48 5 1.15, P 5 0.290) (Fig. 2A). There was a signi®cant positive relationship between the weight of the rafts and the number of taxa (Regression F1,48 5 10.90, P , 0.002), and a positive non-signi®cant relationship between the weight of attached plants and the number of taxa recorded

(Regression F1,6 5 2.14, P 5 0.193) (Fig. 2B). A signi®cant negative relationship was found between the age of rafts and their weight (Regression F1,48 5 5.28, P , 0.027) (Fig. 2C). If attached samples were included in these regressions there was no change to the signi®cance level of the relationships.

3.3. Individual species patterns

3.3.1. Frequency of occurrence and raft age Nineteen species were found on more than ten of the 58 attached and raft samples (Table 2). Raft and attached samples were grouped into seven age classes (Table 3), and the frequency of species on rafts in each class was regressed against the average age of rafts in the age class. Only one species, the isopod Idotea resecata, was found on all 58 samples (Fig. 3A); the second most ubiquitous kelp raft taxa was the shrimp Hippolyte spp., occurring on 95% of all rafts (Fig. 3C). The anemone (Epiactis prolifera) showed the only signi®cant slope, a decline in mean frequency from 40% to 0% (P , 0.005, R2 5 0.84) (Fig. 3E). The ophiuroid Ophiothrix spiculata (initial frequency 58%, ®nal frequency 20%), and the isopod Paracerceis cordata (initial frequency 40%, ®nal frequency 3%), decreased non-signi®cantly with raft age. Several species increased in frequency non-signi®cantly with raft age, including the crabs Planes cyaneus (initial frequency 0%, ®nal frequency 78%), Pugettia producta (initial frequency 37%, ®nal frequency 44%), crab megalope (initial frequency 20%, ®nal frequency 57%), and the pelagic barnacle Lepas spp. (initial frequency 2%, ®nal frequency 80%). For the 82 A.J. Hobday / J. Exp. Mar. Biol. Ecol. 253 (2000) 75 ±96

Table 2 Presence of taxa on M. pyrifera rafts and attached plantsa Taxa Primary Mobility if Offspring Total Attached Rafts Habitat detached (n)(n)(n) Idotea resecata Canopy Swim Brooded 58 8 50 Hippolyte spp.(clarki and californiensis) Canopy Swim Planktonic 55 8 47 Lepas spp. Canopy Swim Planktonic 37 1 36 Sebastes diploproa Canopy Swim Planktonic 36 0 36 Melibe leonina Canopy Swim Planktonic 31 0 31 Caprella californica Canopy Sink Brooded 29 7 22 Ophiothrix spiculata Holdfast Sink Planktonic 29 7 22 Tripoha maculata Canopy Sink Planktonic 29 8 21 Mitrella carinata Canopy Sink Planktonic 25 5 20 Crab Megalope Canopy Swim Planktonic 21 1 20 Pugettia producta Canopy Sink Planktonic 19 5 14 Planes cyaneus Canopy Swim Planktonic 18 0 18 Medialuna californiensis Canopy Swim Planktonic 18 0 18 Snail B Holdfast Sink ? 18 5 13 Paracercies cordata Holdfast Sink Brooded 16 6 10 Epiactis prolifera Holdfast Sink Brooded 15 3 12 Lysniata californica Holdfast Swim Planktonic 14 1 13 Rimicola muscarum Canopy Swim Planktonic 12 5 7 Chromis punctipinnis Canopy Swim Planktonic 11 0 11 Euphausia paci®ca 10 0 10 Portunidae 7 0 7 Sebastes serriceps 70 7 Sygnathus californiensis 70 7 Cypselurus californicus 70 7 Majidae juveniles 7 2 5 Heterostichus rostratus 73 4 Synidotea harfordi 60 6 Sebastes rubrivinctus 60 6 Snail C 6 1 5 Strongylocentrotus purpuratus 63 3 Xanthidae 5 0 5 Alpheus clamator 50 5 Lima sp. 51 4 Heptacarpus spp. 51 4 Holothuroidea unid. 5 3 2 Pisaster sp. 40 4 Mytilus sp. 41 3 Paralabrax clathratus 41 3 Taliepus nuttalli 42 2 Calliostoma annulatum 43 1 Leptopecten monotimeris 30 3 Idotea fewkesi 30 3 Notoacmea inessa 30 3 Synalphaeus lockingtoni 30 3 Ophiactis simplex 31 2 Megabalanus californicus 32 1 Cancer sp. 32 1 Liparis spp.(¯orae and mocosus) 33 0 A.J. Hobday / J. Exp. Mar. Biol. Ecol. 253 (2000) 75 ±96 83

Table 2. Continued Taxa Primary Mobility if Offspring Total Attached Rafts Habitat detached (n)(n)(n) Ulvicola sanctaerosae 33 0 Patriella miniata 20 2 Seriola lalandi 20 2 Pachycheles pubecens 22 0 Gibbonsia sp. 22 0 Pycnopodium sp. 10 1 Dossima facicularis 10 1 Stenoplax sp. 10 1 Pachygrapsus crassipes 10 1 Lytechinus anamesus 10 1 Gymnothorax sp. 10 1 Lythrupnus dalli 10 1 Merluccius productus 10 1 Eupentacta quinquesemita 10 1 Idotea montereyensis 10 1 Collisella pelta 10 1 Crepidula norrisi 10 1 Octopus bimaculoides 10 1 Norrissia norrisi 10 1 Tegula brunnea 10 1 Nassarius sp. 11 0 Paraxanthius taylori 11 0 Pugettia dalli 11 0 Octopus micropyrsus 11 0

Total taxa 72 36 64 a A total of 50 rafts and eight attached plants were sampled. The top nineteen taxa were collected on more than ten samples, and some potentially important characters for persistence on kelp rafts are shown for these species. The entries in the ``mobility if detached'' column indicate whether the species will sink or swim if separated from the kelp plant or raft. remaining eleven species, there was no discernible trend in frequency of occurrence with raft age (Hobday, 1998).

3.3.2. Species density, sizes, and raft age Two of the 17 top species showed a signi®cant response in density to raft age when all samples were combined (two of the nineteen most frequent species, Lepas spp. and Caprella californica, were only recorded as present or absent). The densities of I. resecata (initial density 5 kg21 , ®nal density . 100 kg212 , P , 0.005, R 5 0.152, Fig. 3B) and P. producta (initial density 0.01 kg21 , ®nal density . 0.1 kg21 , P , 0.001, R2 5 0.253) were signi®cantly positively related to the age of rafts, although the variation explained was small. Increased non-signi®cant densities were found for P. cyaneus (initial density 0 kg21 , ®nal density 0.3 kg21 ). Hippolyte spp. had no trend with raft age (average density 3.8 kg21 , range 0±14.2 kg21 , Fig. 3D). No signi®cant negative density trends were detected, although declines in density were observed for E. prolifera 84 A.J. Hobday / J. Exp. Mar. Biol. Ecol. 253 (2000) 75 ±96

Fig. 2. Relationships between M. pyrifera raft age, weight, and number of species. A. Estimated age of rafts (triangles, n 5 50, R2 5 0.03) and the number of macrofaunal species recorded. Attached plants (circles, age 0) are shown at the left end of the graph for comparison. The regression line is for rafts only. B. Weight of rafts (triangles, n 5 50, R225 0.18) and attached plants (circles, n 5 8, R 5 0.26) and the number of macrofaunal species recorded. The slope for rafts (lower line) is signi®cant (P , 0.002), as is the slope for all samples combined (not shown) (n 5 58, P , 0.009), but not for attached plants alone (upper line) (P 5 0.193). The two slopes are not signi®cantly different, nor are their intercepts. C. Estimated age of rafts (triangles, n 5 50, R2 5 0.10) and raft weight. Attached plants (circles, age 0) are shown at the left end of the graph for comparison. The regression line is for rafts only. A.J. Hobday / J. Exp. Mar. Biol. Ecol. 253 (2000) 75 ±96 85

Table 3 Raft and attached M. pyrifera samples grouped according to estimated age for analysis of species frequency by raft agea Age class (days) 0 0±10 11±20 21±30 31±40 41±60 61±80 Average weight (kg) 57.33 65 61 50 41 48 22 N samples 8 26 6 5 4 4 5 Average age (days) 0 2 15 25 34 51 67 a Attached plant age was de®ned as 0 days (0 age class).

(initial density 0.16 kg21 , density 0 kg21 after 38 days, Fig. 3F) and Ophiothrix spiculata (initial density 3.2 kg21 , ®nal density 0.1 kg21 ). Density descriptions for all species are reported in Hobday (1998).

Fig. 3. Example relationships between species frequency and M. pyrifera raft age (A. Idotea resecata,C. Hippolyte spp.,E.Epiactis prolifera), and between overall species density and raft age (B. Idotea resecata,D. Hippolyte spp.,F.Epiactis prolifera). 86 A.J. Hobday / J. Exp. Mar. Biol. Ecol. 253 (2000) 75 ±96

Signi®cant differences in density between cruises were found for four of the top species (Hippolyte spp., Sebastes diploproa, S. serriceps, Melibe leonina) (MANOVA, cruise P , 0.05, raft age n.s.). A ®fth species had marginally non-signi®cant differences in density between cruises (kelp snail, Mitrella carinata). Raft age was a co-factor in these analyses, as in the regression analyses only I. resecata and P. producta had signi®cant density variation with raft age (Hobday, 1998). The densities of the top taxa were each regressed against raft age for each cruise to examine seasonal differences. Eight of these species were found in all cruises, seven from four cruises, and two from three cruises (Table 4). There were no clear patterns in presence in terms of summer and winter cruises. Seven species were found to have signi®cant density±age relationships (Table 4). Six of the nine signi®cant slopes (out of 74 tests) were from winter cruises and three were from summer cruises. Only the nudibranch M. leonina had a signi®cant negative slope and it also had a signi®cant positive slope in a summer cruise. The other signi®cant positive summer relationships were for the ®sh S. diploproa and the shrimp Lysmata californica. The pelagic grapsid crab P. cyaneus, showed a signi®cant increase in density with raft age in one winter cruise and was almost signi®cant in another winter cruise, while I. resecata had two

Table 4 Results of individual regressions on top species density-by-cruise and M. pyrifera raft agea Taxa 9503 9509 9602 9706 9712 S W Idotea resecata 0.973 0.998 0.001 0.383 0.016 02 Hippolyte spp. 0.648 0.298 0.918 0.400 0.942 0 0 Sebastes diploproa 0.069 0.011 0.706 0.745 0.329 1 0 Melibe leonina 0.763 0.737 0.001 0.029 11 Ophiothrix spiculata 0.626 0.566 0.846 0.873 0.535 0 0 Triopha maculata 0.304 0.376 0.035 0.700 0.846 0 1 Mitrella carinata 0.495 0.803 0.456 0.662 0.359 0 0 Crab megalope 0.067 0.360 0.449 0.277 0 0 ]] Pugettia producta 0.070 0.403 0.001 0.133 0 1 ]] Planes cyaneus 0.059 0.548 0.601 0.306 0.027 01 Medialuna californiensis 0.183 0.090 0.075 0.330 0 0 Snail B 0.761 0.298 0.696 0.535 0 0 Paracercies cordata 0.324 0.997 0.422 0.571 0 0 Epiactis prolifera 0.949 0.400 0.397 0.159 0.535 0 0 Lysmata californica 0.934 0.063 0.003 0.535 1 0 ]] Rimicola muscarum 0.730 0.213 0.715 0 0 Chromis punctipinnis 0.424 0.053 0.535 0 0

Signi®cant differences 0132336 a Taxa are listed in order of occurrences in samples. The ®ve cruises are indicated by the year month (yrmo) code at the top of each column. Taxa in bold font are those which were found to have differences between cruises in the pooled analyses of density and raft age. Signi®cant positive slopes are indicated by P values in bold italics, while signi®cant negative slopes are indicated by bold P values. A few marginally nonsigni®cant values are underlined in normal or italic corresponding to negative and positive slopes respectively. The number of signi®cant differences found for summer cruises (S) (9509, 9706) and winter cruises (W) (9503, 9602, 9712) is totaled. Blank cells indicate that the species was not collected on a cruise. A.J. Hobday / J. Exp. Mar. Biol. Ecol. 253 (2000) 75 ±96 87 signi®cant density relationships in winter cruises. The nudibranch Triopha maculata also showed a positive response to rafting in one winter cruise. P. producta had one signi®cant and one almost signi®cant positive response in winter cruises. Eight of nine species analyzed had a positive size relationship with raft age when all the samples were combined (Fig. 4). Only the kelp cling®sh Rimicola muscarum had a signi®cant positive size relationship with raft age (P , 0.001, R2 5 0.63). S. serriceps had an almost signi®cant increase in size with raft age (P 5 0.079, R2 5 0.39). No signi®cant trend was found for the rest of the ®sh, or crustaceans, although when adult and juvenile I. resecata were considered separately adult size was found to increase signi®cantly with raft age (Hobday, 1998).

3.3.3. Successful raft species When the results of all the tests involving species patterns and raft age were combined to provide a single score, two of the nineteen top species were classed as unsuccessful raft fauna (Table 5). E. prolifera, and P. cordata had negative scores (21 out of a possible 2 8). Eight species were unaffected by rafting (score 0), and are classed as persistent, although T. maculata had one positive and one negative score, giving an overall total of 0. The remaining nine species showed a positive response to rafting, with scores between 1 and 3 out of 8. The species with the strongest response to rafting was the most abundant member of the macrofaunal raft community, I. resecata. P. producta, P. cyaneus, and S. diploproa were the next most ``successful'' rafting species, with positive responses to rafting in two tests.

3.3.4. Extinction times Extinction times for species on kelp rafts were calculated using each of the three regression methods used to explore the species raft age relationships. The regressions from the frequency of occurrence analysis provide a method of estimating the time species will remain on a raft. As reported above, E. prolifera had the only signi®cant decline in frequency and had a regression estimated extinction time of 63 days after raft formation. An extinction time of 84 days for P. cordata was found using the non- signi®cant regression relationship. For the density and raft age regressions, only extinction times for the three species classi®ed as unsuccessful on kelp rafts were calculated (Table 5). When the overall density regressions were used, the extinction time for P. cordata and E. prolifera the extinction times were 112 and 140 days respectively, but the regression line for T. maculata was slightly positive and no extinction time exists. Finally, the time to extinction for the three unsuccessful raft species could also be calculated using the regressions from each of the density and raft age relationship for each cruise. Disregarding the single positive slope found for one cruise for T. maculata and P. cordata. the average times to extinction for the two species were 73 (range 50±109) and 125 (range 36±298) days respectively. E. prolifera had an average extinction time of 113 days (range 36±379). Extinction times in summer were shorter than in winter for these species. 88 A.J. Hobday / J. Exp. Mar. Biol. Ecol. 253 (2000) 75 ±96

Fig. 4. Relationships between species size and M. pyrifera raft age. The maximum size for Lepas spp. on each raft is shown (as sampling was non-quantitative), the average sizes are shown for all other species. Only the relationship for Rimicola muscarum (panel H) is signi®cant (P , 0.035). Average size is only possible for the rafts that had the particular species. A.J. Hobday / J. Exp. Mar. Biol. Ecol. 253 (2000) 75 ±96 89

Table 5 Result of raft success summarya Species Attached Rafts Preferred Habitat Signi®cant age trends Overall habitat score score n % n % % Occ Density By cruise Idotea resecata 8 100 50 100 Both 0 1 2 3 Pugettia producta 5 63 14 28 Both 0 1 1 2 Planes cyaneus 0 0 18 36 Raft 1 1 2 Sebastes diploproa 0 0 34 68 Raft 1 1 2 Lepas spp.b 1 13 36 72 Raft 1 1 Melibe leonina 0 0 31 62 Raft 1 1 Medialuna californiensis 0 0 18 36 Raft 1 1 Chromis punctipinnis 0 0 11 22 Raft 1 1 Lysmata californica 1 13 13 26 Both 0 1 1 Megalope 1 13 19 38 Both 0 0 Hippolyte spp. 8 100 47 94 Both 0 0 Caprella californicab 7 88 22 44 Both 0 0 Ophiothrix spiculata 7 88 22 44 Both 0 0 Mitrella carinata 5 63 20 40 Both 0 0 Snail B 5 63 13 26 Both 0 0 Rimicola muscarum 5 63 7 14 Both 0 0 Triopha maculata 8 100 21 42 Attached 2110 Paracercies cordata 6 75 10 20 Attached 21 21 Epiactis prolifera 3 38 12 24 Both 0 21 21 a This index combines all of the statistical test on the response of individual species to raft age. The preferred habitat is that on which the species was 50% more frequent. The signi®cant age trends are from the three analyses, percentage of rafts occupied by the species (% Occ), the relationship between species density and raft age for all samples, and the relationship between species density and raft age for each cruise. Positive (negative) scores are for increased (decreased) frequency of occurrence or density with raft age. The overall score is the sum of the four scores, and species are ranked from highest to lowest score. More details on the scoring are provided in the text. b Lepas spp. and Caprella californica were not enumerated in samples and so regressions of species density and raft age were not possible.

4. Discussion

Large M. pyrifera rafts were collected throughout the Southern California Bight over a three year period, but there was no difference between cruises in the age or size of the collected rafts. Older rafts were smaller than younger rafts which indicates that rafts were loosing material as they drift (Table 3).

4.1. Species richness

Rafts of M. pyrifera are similar in species richness to attached plants (McLean, 1962; Coyer, 1984; Bushing, 1994); the lower richness on attached plants in this study is due to the smaller number of attached plants collected. In other studies the invertebrate fauna associated with drifting algae has been found to be rich and diverse; the Atlantic Sargassum drift community is comprised of at least 100 species (Weis, 1968; Fine, 1970; Dooley, 1972; Stoner and Greening, 1984), while 39 species occur on small algal 90 A.J. Hobday / J. Exp. Mar. Biol. Ecol. 253 (2000) 75 ±96 rafts in Iceland (Ingolfsson, 1995). In the results presented here, 64 macrofaunal taxa were identi®ed from rafts (seven to ®fteen species per raft) (Table 2), but actual species richness was higher. For example, polychaetes and amphipods were common in the samples but they were not included as they may not have been completely sampled. In other studies, amphipods were numerically important species on attached plants (20 canopy species, Coyer, 1984; 37 holdfast species, Snider, 1985). Adding other groups not considered (e.g. non-juvenile ®sh) would push the total of M. pyrifera raft associated taxa to about 150 macrofaunal species. This compares with more than 200 species of ®sh and invertebrate species identi®ed on Macrocystis, Nereocystis and Pelagophycus rafts by Bushing (1994). The complexity of the holdfast and canopy are important factors in the richness of Macrocystis communities (Coyer, 1984) and this complexity is retained and even intensi®ed in drifting rafts (Hobday, 1998). The sympatric annual kelp species Nereocystis leutkeana is much simpler in structure and has a more depauperate community (McLean, 1962). In this study there was no signi®cant trend in species richness and the distance offshore rafts were collected. Classical island biogeography would suggest that similar sized rafts farther from land would have fewer species than closer ``islands''. Kelp rafts, however, in contrast with the islands in the ``equilibrium theory'' of MacArthur and Wilson (1967), all originate close to shore and distance changes quickly over time and is not signi®cantly related to age (Hobday, 1998). This situation is reminiscent of the ``vicariance theory'' where fauna evolves in situ on mobile continents and islands (MacArthur and Wilson, 1967). In another algal raft study, however, the number of species on small algal clumps ( , 1 kg) did decrease with distance offshore (Ingolfsson, 1995). Older islands should have more species than younger islands according to island biogeography, but kelp rafts in this study did not support this prediction. As the age of rafts increases they may lose kelp forest species and be colonized by the planktonic larvae of other species. Storms before and after detachment may have dramatic effects on species numbers, turnover rates, and direction of drift paths. Few species appear to be completely lost from drifting rafts once aboard, however, changes in composition may occur at the point of detachment when large heavy invertebrates like snails may be dislodged (Hobday, 1998). The area prediction from island biogeography was met as larger rafts collected in this study did have more species than smaller rafts. In studies of attached M. pyrifera richness was also greater on larger plants (Coyer, 1984). In general, the patterns of species richness indicate the M. pyrifera rafts are not relic assemblages and are utilized by both nearshore and pelagic species. The goal of this study, however, was not to determine total species richness for kelp rafts, but to explore the changes in species abundance on rafts and determine which species might be successfully dispersed by rafts.

4.2. Species abundance, density, sizes, and raft age

The most abundant kelp raft species was the isopod Idotea resecata which was found on all rafts and signi®cantly increased in density with raft age and was twenty times A.J. Hobday / J. Exp. Mar. Biol. Ecol. 253 (2000) 75 ±96 91 more abundant on old rafts than on attached plants (Fig. 3A, B). Other successful taxa included the crabs Pugettia producta, Planes cyaneus, crab megalope, the pelagic barnacle Lepas spp., a shrimp Lysmata californica, a nudibranch Melibe leonina, and three ®sh S. diploproa, M. californiensis, and C. punctipinnis (Table 5). Boehlert (1977a) also reported that I. resecata and L. californica were the only two invertebrate species found in greater abundance on M. pyrifera rafts than on attached plants. He attributed this increase to release from predation from the kelp forest ®shes Oxyjulus californica and Brachystichus frenatus, which are not associated with rafts. Nearshore species that increase in abundance on the rafts are likely to contribute numerically to existing kelp forests and colonize new ones if rafts return to these habitats. A number of other species showed no decline in frequency or density, which also makes them candidates for dispersal on kelp rafts. Because the focus in this study was to collect the whole raft and the associated invertebrates, the techniques used provided an incomplete sampling of the ®sh fauna. Most of the seventeen ®sh species collected in this study were juveniles that were closely associated with the rafts. The larger, less closely associated individuals were observed evading the net. The most common ®sh in this study was juvenile S. diploproa, as found in two other M. pyrifera raft studies (Mitchell and Hunter, 1970; Boehlert, 1977a). They occurred in association with rafts up to 100 km offshore, despite the fact that the adults are demersal inshore ®shes (Boehlert, 1977b). Rafts, while not essential habitat for juvenile ®sh, are likely to improve survival during that life stage (Kingsford, 1993). Most of the nine species analyzed for size relationships increased in average size with raft age (Fig. 4). This might be due to growth of individuals on the rafts, or recruitment of larger individuals over time. Continuous recruitment from the surrounding waters and increased survival of these juveniles is likely for Lepas spp., P. cyaneus and P. producta, but this would tend to reduce the average size. For brooding species, however, no colonization from surrounding waters can occur. Production and survival of juveniles is higher on rafts than on attached plants for some of these species (Hobday, 1998), and may mask even larger increases in overall average size. The ®ve species of ®sh associated with the rafts that were measured recruit to rafts as juveniles and so any growth during the raft association will also increase the average size, as observed (Fig. 4). Increases in size with raft age also indicate that individuals remain with the rafts once separated from the kelp forest. If the individuals were moving freely between rafts, size should not be related to with raft age unless larger individuals initially seek out and prefer old rafts, while smaller individuals do the opposite, which seems unlikely. Average faunal size and growth rates can be used to determine the length of time that a particular species may have been associated with the raft, and even provide an alternative for aging the raft (Helmuth et al., 1994; Hobday, 1998). For example, using average Lepas paci®ca growth rate of 0.22 mm day21 and a settlement size of 1.5 mm (Hobday, unpublished data), and assuming that no Lepas occur on attached plants, biological estimates of raft age can be made. Such an aging methodology is, however, biased by the abundance of the species in question when the raft forms and is not independent of the fauna, an important criteria when aging rafts to infer faunal relationships (Hobday, 1998). 92 A.J. Hobday / J. Exp. Mar. Biol. Ecol. 253 (2000) 75 ±96

4.3. Successful species and persistence times

Kelp rafts may ¯oat for approximately 110 days (Hobday, 1998, 2000b), and so fauna that persist longer than this will be found on rafts, while less persistent species may be lost. The relationships between raft age and top species frequency of occurrence, overall density, and density by cruise indicated that nine species would increase in abundance and another eight would persist until the raft sank or drifted to another kelp forest, while two species were classed as unsuccessful (Table 5). Extinction times were calculated using the three relationships between fauna and raft age, and differed slightly with each approach. T. maculata, while not classed as ``unsuccessful fauna'', was predicted to go extinct using the density-by-cruise regression relationships. An average of the three methods indicates that E. prolifera, P. cordata, and T. maculata, would be lost after 103, 107, and 72 days respectively. E. prolifera and P. cordata are both species found mainly on the holdfasts of M. pyrifera (Table 2), and E. prolifera has very limited movement. Older rafts may lose their holdfasts before they sink (Hobday, 1998), and so holdfast species may be lost if the holdfast separates from the canopy. For any species to be dispersed between kelp forests the journey must be shorter than the extinction times, however, most persist for about the maximum estimated ¯oating time of M. pyrifera rafts (Hobday, 1998, 2000b). No relationship between rafting success and a variety of life history and behavioral characters could be determined, in fact each rafting species appeared to be successful or unsuccessful with a different combination of traits (Hobday, 1998). Not surprisingly, the ability to swim if separated from the raft was related to success and persistence (successful 78% swim, unsuccessful 100% sink) (Table 2). This ability may have allowed the species to return to the raft if it was dislodged, or to ®nd a new raft when the old raft sinks. Brooding was anticipated to be an important factor, however, brooding species were both the most successful (I. resecata), and the least successful (E. prolifera and P. cordata) rafting species (Table 2 and 5). Vasquez (1993) reported that only brooding amphipods and isopods increased in density on detached holdfasts, but this factor alone can not account for raft success. Additional analyses using various life history traits and measures of raft success did not reveal any other pattern (Hobday, 1998).

4.4. Sampling biases and analyses caveats

Differences in the capture of associated invertebrates using the two net designs is unlikely, but may have in¯uenced the capture success of the more mobile ®sh found below the kelp raft. These ®sh may have been undersampled by the skimming action of the Rumsey Net compared with the vertical capture action of the Giant Rumsey Net, however, differences between cruises did not support this. The differences in the nets and techniques used to collect attached and raft samples might also be a cause for the lower abundance of ®sh collected with attached plants. Additional observations during collection of attached plants for another set of experiments showed that of all the closely associated ®sh of a variety of sizes were captured with these methods (Hobday, 1998). The behavior of kelp associated juvenile ®sh when frightened is too swim to the center A.J. Hobday / J. Exp. Mar. Biol. Ecol. 253 (2000) 75 ±96 93 and in amongst the fronds, which would also facilitate collection in the nets, and suggest that the ®sh patterns are due to real differences in abundance and association.

4.5. Conclusions and implications for faunal dispersal

Kelp forest fauna that remains on kelp rafts is not destined to certain death on a sinking raft and has the potential to be dispersed between isolated kelp forests (Fell, 1962; Helmuth et al., 1994). Hobday (2000a) used satellite drifter data to suggest that up to 45% of M. pyrifera rafts in Southern California may return to the coast after an average of ten days and a displacement of 56 km. Rafting may be an important process in maintenance of viable kelp forest faunal populations, especially as kelp forests periodically disappear (Cribb, 1954; Dayton et al., 1984; Seymour et al., 1989), and not all kelp forest fauna has pelagic dispersal. In some situations a short generation time may be important in the persistence of some species on rafts, however, the lifetime of the rafts is shorter than the generation time of most of the species living on both attached and raft M. pyrifera. Pelagic species with larval dispersal, such as Lepas spp. and P. cyaneus, may complete a cycle of larval settlement, growth, and reproduction before the raft sinks or arrives at the coast (Hobday, 1998). If the rafts arrive back at the coast or kelp forests, these pelagic species do not survive, probably due to predation (Bernstein and Jung, 1979). Adults of these pelagic species require ¯oating substrate and could persist as ``raft-hoppers'', although alternative more persistent habitats like turtles (Davenport, 1994; Hernandez-Vazquez and Valadez-Gonzalez, 1998) may be the primary habitat. The results presented here suggest most of the common attached M. pyrifera fauna is unaffected by rafting and are able to persist for the ¯oating time of the rafts. In general the lifetime of the raft and not the persistence time of the individual species will limit faunal dispersal on kelp rafts in Southern California.

Acknowledgements

The large amount of ®eldwork undertaken in this research would not have been possible without the assistance of many people at Scripps. Scott Rumsey provided invaluable diving support for this work. Ship-based raft collections and surveys of the island kelp forests was supported by Paul Smith, Craig Smith and the captains and crews of the R/V Sproul, R/V New Horizon and R/V David Starr Jordan. The support and assistance of the science parties aboard the cruises in capturing and processing samples is greatly appreciated. Megan Johnson, Amy Ritter, Brad Bagan and Doriena Jone provided assistance with processing samples. Paul Dayton provided space and resources for laboratory work. Review by George Sugihara. Peter Franks, Paul Dayton and two anonymous reviewers improved the clarity of this manuscript. Financial support for this work was provided by an ONR grant to George Sugihara, Scripps Student Ship Funds, and a PADI grant to the author. This research was completed in partial ful®llment of the requirements for a Ph.D. at the Scripps Institution of Oceanography. 94 A.J. Hobday / J. Exp. Mar. Biol. Ecol. 253 (2000) 75 ±96

References

Andrews, H.L., 1945. The kelp beds of the Monterey region. Ecology 26, 24±37. Benech, S.V., 1978. Ocean transport of a community of the grapsid crab Plagusia dentipes (De Haan, 1833). Crustaceana 35 (1), 104. Bernstein, B.B., Jung, N., 1979. Selective pressures and coevolution in a kelp canopy community in Southern California. Ecol. Monog. 49 (3), 335±355. Boehlert, G.W., 1977a. Physiological and morphological adaptations in the surface-to-benthic migration of Sebastes diploproa: (Pisces, Scorpaenidae). Ph.D. University of California, San Diego. Boehlert, G.W., 1977b. Timing of the surface -to-benthic migration in juvenile rock®sh, Sebastes diploproa, off Southern California. Fish. Bull. 75, 887±890. Bushing, W.W., 1994. Biogeographic and ecological implications of kelp rafting as a dispersal vector for marine invertebrates. California Islands Symposium: Update on the Status of Resources 4, 103±110. Clendenning, K.A., 1971. Gross composition of kelp. Nova Hedwigia (suppl) 32, 197±209. Coyer, J.A., 1984. The invertebrate assemblage associated with the giant kelp, Macrocystispyrifera at Santa Catalina Island, California: A general description with emphasis on amphipods, copepods, mysids and shrimps. Fish. Bull. 82 (1), 55±66. Cribb, A.B., 1954. Macrocystis pyrifera (L.) Ag. in Tasmanian waters. Aust. J. Mar. Fresh Res. 5, 1±34. Davenport, J., 1994. A cleaning association between the oceanic crab Planes minutus and the loggerhead sea turtle Caretta caretta. J. Mar. Biol. Assoc. UK 74, 735±737. Davenport, J., Rees, E.I.S., 1993. Observations on neuston and ¯oating weed patches in the Irish Sea. Est. Coast. Shelf Sci. 36, 395±411. Dayton, P.K., Currie,V., Gerrodette, T., Keller, B., 1984. Patch dynamics and stability of some Californian kelp communities. Ecol. Monog. 54 (3), 253±289. Dayton, P.K., 1985. Ecology of Kelp Communities. Ann. Rev. Ecol. Syst. 16, 215±245. Dooley, J.K., 1972. Fishes associated with the pelagic sargassum complex, with a discussion of the sargassum community. Cont. Mar. Sci. 16, 1±32. Draper, N.R., Smith, H., 1981. Applied regression analysis. John Wiley and Sons, New York, NY, second edition, 709pp. Ebeling, A., Larson, R.J., Alevizon, W.S., Bray, R.N., 1980. Annual variability of reef-®sh assemblages in kelp forests off Santa Barbara, California. Fish. Bull. 78 (2), 361±377. Edgar, G.J., 1983. The ecology of south-east Tasmanian phytal animal communities. 3. Patterns of species diversity. J. Exp. Mar. Biol. Ecol. 70 (2), 181±203. Edgar, G.J., 1987. Dispersal of faunal and ¯oral propagules associated with drifting Macrocystis pyrifera plants. Mar. Biol. 95, 599±610. Fedoryoka, B.I., 1989. A comparative characteristic of oceanic ®sh assemblages associated with ¯oating debris. J. Ichthy. 29 (2), 128±137. Fell, H.B., 1962. West-wind-drift dispersal of echinoderms in the Southern Hemisphere. Nature 193, 759±761. Fine, M.L., 1970. Faunal variation on pelagic Sargassum. Mar. Biol. 7, 112±122. Ghelardi, R.J., 1971. ``Species'' structure of the animal community that lives in Macrocystis pyrifera holdfasts. Nova Hedwigia (suppl) 32, 381±420. Gooding, R.M., Magnuson, J.J., 1967. Ecological signi®cance of a drifting object to pelagic ®shes. Pac. Sci. 21, 486±497. Gunnill, F.C., 1982. Effects of plant size and distribution on the numbers of invertebrate species and individuals inhabiting the brown algae Pelvetia fastigata. Mar. Biol. 69, 263±280. Harrold, C., Lisin, S., 1989. Radio-tracking rafts of giant kelp: local production and regional transport. J. Exp. Mar. Biol. Ecol. 130, 237±251. Helmuth, B.S., Veit, R.R., Holberton, R., 1994. Long-distance dispersal of a subantarctic brooding bivalve (Gaimardia trapesina) by kelp rafting. Mar. Biol. 120 (3), 421±426. Hernandez-Vazquez, S., Valadez-Gonzalez, C., 1998. Observations of the epizoa found on the turtle Lepidochelys olivacea at La Gloria, Jalisco, Mexico. Ciencias Marinas 24 (1), 119±125. Highsmith, R.C., 1985. Floating and algal rafting as potential dispersal mechanisms in brooding invertebrates. Mar. Ecol. Prog. Ser. 25, 169±179. A.J. Hobday / J. Exp. Mar. Biol. Ecol. 253 (2000) 75 ±96 95

Hobday, A.J., 1998. Faunal Patterns and Dispersal on Kelp Rafts in Southern California. Ph.D. University of California, San Diego. Hobday, A.J., 2000a. Abundance and dispersal of drifting kelp (Macrocystis pyrifera) rafts in the Southern California Bight. Mar. Ecol. Prog. Ser. 195, 101±116. Hobday, A.J., 2000b. Age of drifting Macrocystis pyrifera (L.) C. Agardh rafts in the Southern California Bight. J. Exp. Mar. Biol. Ecol. 253, 97±114. Hunter, J.R., Mitchell, C.T., 1967. Association of ®shes with ¯otsam in the offshore waters of Central America. Fish. Bull. 66 (1), 13±29. Ida, H., Hiyama, Y., Kusaka, T., 1967. Study on ®shes gathering around ¯oating seaweed-I Abundance and species composition. Bull. Jap. Soc. Sci. Fish. 33 (10), 923±929. Ingolfsson, A., 1995. Floating clumps of seaweed around Iceland: natural microcosms and a means of dispersal for shore fauna. Mar. Biol. 122, 13±21. Ingolfsson, A., 1998. Dynamics of macrofaunal communities of ¯oating seaweed clumps off western Iceland: a study of patches on the surface of the sea. J. Exp. Mar. Biol. Ecol. 231 (1), 119±137. Johannesson, K., 1988. The paradox of Rockall: why is a brooding gastropod (Littorina saxatilis) more widespread than one having a planktonic larval dispersal stage (L. littorea). Mar. Biol. 99, 507±513. Jones, L.G., 1971. Studies on selected small herbivorous invertebrates inhabiting Macrocystis canopies and holdfasts in Southern California kelp beds. Beiheft Nova Hedwigia 32 (chapter 17), 343±367. Kingsford, M.J., 1992. Drift algae and small ®sh in coastal waters of northeastern New Zealand. Mar. Ecol. Prog. Ser. 80, 41±55. Kingsford, M.J., 1993. Biotic and abiotic structure in the pelagic environment: importance to small ®shes. Bull. Mar. Sci. 53 (2), 393±415. Kingsford, M.J., Choat, J.H., 1985. The fauna associated with drift algae captured with a plankton-mesh purse seine net. Limn. Oceanog. 30 (3), 615±618. MacArthur, R.H., Wilson, E.O., 1967. The theory of island biogeography. Princeton University Press, Princeton, New Jersey. McLean, J.H., 1962. Sublittoral ecology of kelp beds of the open coast area near Carmel, California. Biol. Bull. 122 (1), 95±114. Mitchell, C.T., Hunter, J.R., 1970. Fishes associated with drifting kelp, Macrocystis pyrifera, off the coast of southern California and northern Baja California. Cal. Fish Game Bull. 56 (4), 288±297. Mukai, H., 1971. The phytal animals on the thalli of Sargassum serratifolium in the Sargassum region, with reference to their seasonal ¯uctuations. Mar. Biol. 8, 170±182. Nakata, H., Takeuchi, H., Hirano, T., 1988. A ®eld experiment with drifting hoop nets to collect the small ®sh gathering around drift algae. Nippon Suisan Gakkaishi 54 (11), 1899±1906. North, W.J., 1971. The biology of giant kelp beds (Macrocystis) in California. Nova Hedwigia (suppl) 32, 1±98. North, W.J., 1994. Review of macrocystis biology. In: Akatsuka, I. (Ed.), Biology of economic algae. SPB Academic Publishing, The Hague, pp. 447±527. Ojeda, F.P., Santelices, B., 1984. Invertebrate communities in holdfasts of the kelp Macrocystis pyrifera from southern Chile. Mar. Ecol. Prog. Ser. 16, 65±73. Senta, T., 1966. Experimental studies on the signi®cance of drifting seaweeds for juvenile ®shes-I. Experiments with arti®cial drifting seaweeds. Bull. Jap. Soc. Sci. Fish. 32 (8), 639±642. Seymour, R.J., Tegner, M.J., Dayton, P.K., Parnell, P.E., 1989. Storm wave induced mortality of giant kelp, Macrocystis pyrifera, in Southern California. Est. Coast. Shelf Sci. 28 (3), 277±292. Smith, S.D.A., Simpson, R.D., Cairns, S.C., 1996. The macrofaunal community of Ecklonia radiata holdfasts: Description of the faunal assemblage and variation associated with differences in holdfast volume. Aust. J. Ecol. 21 (1), 81±95. Snider, L.J., 1985. Demersal zooplankton of the giant kelp Macrocystis pyrifera: patterns of emergence and the population structure of three gammarid amphipod species. Ph.D. University of California, San Diego. Sokal, R.R., Rohlf, F.J., 1981. Biometry. Freeman, New York. Stoner, A.W., Greening, H.S., 1984. Geographic variation in the macrofaunal associates of pelagic sargassum and some biogeographical implications. Mar. Ecol. Prog. Ser. 20, 185±192. Tegner, M.J., Dayton, P.K., Edwards, P.B., Riser, K.L., 1995. Sea urchin cavitation of giant kelp (Macrocystis pyrifera C. Agardh) holdfasts and its effects on kelp mortality across a large California forest. J. Exp. Mar. Biol. Ecol. 191 (1), 83±99. 96 A.J. Hobday / J. Exp. Mar. Biol. Ecol. 253 (2000) 75 ±96

Tully, O., Ceidigh, P.O., 1986. The ecology of Idotea species (Isopoda) and Gammarus locusta (Amphipoda) on surface driftweed in Galway Bay (west of Ireland). J. Mar. Biol. Assoc. UK 66, 931±942. Vallentin, R., 1895. Some remarks on the dispersal of marine animals by means of seaweeds. The annals and magazine of natural history 16, 418±423. Vasquez, J.A., 1993. Effects on the animal community of dislodgement of holdfasts of Macrocystis pyrifera. Pac. Sci. 47 (2), 180±184. Weis, J.S., 1968. Fauna associated with pelagic Sargassum in the Gulf Stream. Am. Mid. Nat. 80 (2), 554±558. Worcester, S., 1994. Adult rafting versus larval swimming: dispersal and recruitment of a botryllid ascidian on eelgrass. Mar. Biol. 121 (2), 309±317.