Oecologia (1996) 105:377-387 i 0 &ringer-Verlag 1996

John J. Stachowicz - Mark E. Hay

Facultative mutualism between an herbivorous crab and a coralline alga: advantages of eating noxious seaweeds

Received: 1 May 1995 / Accepted: 1 1 September 1995

Abstract Because encrusting coralline algae rely -on herbivory or low light levels to prevent being overgrown by competitively superior fleshy algae, corallines are rel- atively rare in shallow areas with low rates of herbivory. In contrast to this general trend, the branching coralline alga Neogoniolithon stricturn occurs primarily in shallow seagrass beds and along the margins of shallow reef flats where herbivory on macrophytes is low. This alga appar- ently persists in these habitats by providing refuge to the herbivorous crab Mithrax sculptus at mean densities of 1 crab per 75 g of algal wet mass. When crabs were re- moved from some host corallines,, hosts without crabs supported 9 times the epiphytic growth of hosts with crabs after only 30 days. Crabs without access to a coral- line alga were rapidly consumed by reef fishes, while most of those tethered near a host alga survived. These results suggest that the crabs clean their algal host of fouling seaweeds and associate with the host to minimize predation. However, to effectively clean the host, the crab must consume the wide array of macroalgae that commonly co-occur with coralline algae in these habi- tats, including chemically defended species in the genera Halimeda, Dictyota, and Laurencia. Crabs did readily consume these seaweeds, which were avoided by, and are chemically defended from, herbivorous fishes. Even though crabs readily consumed both Halimeda and Dic- tyota in whole-plant feeding assays, chemical extracts from these species significantly reduced crab feeding, suggesting that factors other than secondary chemistry (e.g., food value, protein, energy content), may deter- mine whole-plant palatability. Having the ability to use a wide variety of foods, and choosing the most profitable rather than the least defended foods, would diminish for- aging time. increase site fidelity, and allow the crab to

J.J. Stachowicz . M.E. Hay University of North Carolina at Chapel Hill. Institute of Marine Sciences. 3431 Arendell St.. Morehead City NC 28557 USA, Phone: 919-726-6841, Fax: 919-726-2426. Internet: jaystach@email.unc.edu (J. Stachowicz) seaweed@email.unc.edu (M. Hay)

function mutualistically with the host alga. Despite the obvious benefit of associating with N. stricturn, M. sculp- tus did not prefer it over other habitats offering a struc- turally similar refuge, suggesting that these crabs are not N . stricturn specialists, but rather occupy multiple habi- tats that provide protection from predators. Structurally complex organisms like N. stricturn may commonly sup- press competitors by harboring protective symbionts like M. sculptus. It is possible that diffuse coevolution has occurred between these two groups; however, this seems unlikely because both herbivore and host appear to re- spond most strongly to selective pressures from preda- tors and competitors outside this association.

Key words Algal chemical defenses . Competition. Fouling . Mutualism . Plant-herbivore interaction

Introduction

Trade-offs between competitive ability and resistance to herbivores are well known for plants in both marine (Lu- bchenco and Gaines 1981; Hay 1985; Lewis 1986; Estes and Steinberg 1988) and terrestrial (Cates and Orians 1975; Connell and Slatyer 1977; Crawley 1989) systems. Herbivore-resistant plants in low-herbivory environments are often at a competitive disadvantage and may be sus- ceptible to overgrowth by superior competitors that are vulnerable to herbivores (Hay 198 1 a, 1984a, 1985; Gaines and Lubchenco 1982; Steneck 1982; Carpenter 1986; Lewis 1986; Sousa and Connell 1992). When this overgrowth is epiphytic, it can harm the host plant by: (1) reducing growth through shading and competition for nutrients (Orth and van Montfrans 1984; Brawley 1992; Williams and Seed 1992), (2) increasing drag on the plant, leading to a greater possibility of dislodgment or breakage by physical forces (Sousa 1979; D' Antonio 1985), or (3) increasing the probability that the host will be consumed by large herbivores if it is overgrown by more palatable epiphytes (Wahl and Hay 1995).

378 OECOLOGIA 105 (1996) 0 Springer-Verlag

Large herbivores like fishes and sea urchins are known to have a considerable impact on marine algal communities (e.g., Hay 1981a, 1984a; Lewis 1986; Car- penter 1986), but small, less apparent grazers like amphi- pods, dipteran larvae, and gastropods can also alter com- munity composition by removing fast-growing, grazer- susceptible plants and epiphytes, allowing dominance by competitively inferior, grazer-resistant species (Sousa 1979; Brawley and Adey 1981; Dethier 1981; Robles and Cubit 1981; Lubchenco 1983; McBrien et al. 1983; Ste- neck et al. 1991). In marine systems, predation pressure is commonly high on these small grazers and can select for specialization on chemically defended plants in order to reduce encounters with, and susceptibility to, preda- tors (Hay et al. 1989, 1990a,b; Duffy and Hay 1991a,b, 1994; reviewed in Hay 1992; Hay and Steinberg 1992). However, using a plant as both food and shelter imposes constraints on the amount of food available for consump- tion. If an herbivore eats too much of its host, its shelter may be compromised. Herbivores that associate with par- ticular plants for shelter may be forced to reduce feeding on host tissue in order to maintain the integrity of the shelter. This might be accomplished by: (1) foraging away from the shelter when predators are less active, (2) lowering metabolic rates, (3) evolving alternative energy sources such as sequestered chloroplasts (Hay et al. 1989), or (4) consuming primarily epiphytes. When an herbivore restricts feeding to epiphytes, a mutualistic in- teraction between the host-plant and the herbivore could arise if alternative means of removing overgrowth, such as shedding of surficial cells (Moss 1982; Masaki et al. 1984; Johnson and Mann 1986) or production of alle- lopathic chemicals (Sieburth and Conover 1965; Schmitt et al. 1995), are not completely effective.

Coralline algae are slow-growing (Littler and Arnold 1982), heavily calcified macroalgae that may be rapidly overgrown by epiphytes and other fleshy algae in the ab- sence of herbivores (e.g., Littler and Doty 1975; Wan- ders 1977; Steneck 1982, 1986; Hay and Taylor 1985). These calcified algae are major components of coral reefs where herbivory is intense (Hay 1981b, 1991; Lew- is 1986; Carpenter 1986; Steneck 1986), but their rela- tive cover is generally low in habitats such as reef flats, seagrass beds, and damselfish temtories that serve as spatial escapes from intense herbivory (Hay 1985, 1991).

In contrast to this general pattern, the coralline alga Neogoniolithon stricturn (Foslie) Setchell and Mason (as Goniolithon strictum in Taylor 1960) thrives on portions of shallow reef flats and in tropical grass bed habitats where herbivory is relatively low (Ogden et al. 1973; Hay 1984a,b; 1985). How N . stricturn is able to avoid overgrowth by foulers and rapidly growing macroalgae in such a low herbivory environment is unknown. Al- though sloughing of surficial cells by some corallines has been interpreted as an anti-fouling trait, Keats et al. (1994) have recently cast doubt on the general effective- ness of sloughing in removing overgrowth. In contrast to this possible physiological mechanism, the herbivorous crab Mithrar sculptus Lamarck. which commonly lives

*

among the branches of N. stn'cfum, might clean its host of epiphytes much as ants remove competing plants from near their Acacia hosts (Janzen 1966). Previous work has demonstrated that these crabs keep small corals free of seaweed overgrowth (Coen 1988a). Whether these crabs can serve as effective cleaners of N. stricrum would de- pend, in part, on whether they can consume the wide range of seaweeds, including several chemically defend- ed species in the genera Dictyotu, Hulimeda, and Lau- rencia (reviewed by Hay 1991; Paul 1992), which com- monly overgrow corallines in habitats where herbivory is reduced (Hay and Taylor 1985; Lewis 1986; Momson 1988).

Many studies describing apparently mutualistic asso- ciations demonstrate a tangible benefit to the host, but fail to provide direct evidence for a reciprocal benefit to the associate (Cushman and Beattie 1991). In this study, we show that not only does M. sculptus protect its host by preventing overgrowth by other seaweeds, but that the crab also benefits from the association by reduced preda- tion. Despite the reciprocal benefits involved, this rela- tionship is not obligate, and we argue that even diffuse coevolution (sensu Fox 198 1) between coralline algae and small grazers that remove foulers is unlikely in this case, because the crab and coralline alga respond primar- ily to selective pressures from predators and competitors outside the association (Vermeij 1983, 1994).

Methods

Study site and organisms

This research was conducted on several patch reef and grass bed habitats.on the seaward side of Key Largo, Florida, United States during July and August 1993 and July 1994. Coralline algae and crabs for all experiments were collected in < I m of water in a grass bed on the south and east sides of Rodriguez Key, Florida. Other algae for feeding assays were collected from a nearby mixed stand of the seagrass Thalussiu restudinurn Koenig and numerous species of algae in 1 .O m of water.

The crab Mithrax sculprus is a member of the family Majidae, and is commonly found from the Bahamas and Miami throughout the Caribbean to Abrolhos Islands, Brazil (Rathbun 1925). The species occurs intertidally, under rocks at low tide, and to a depth of 55 m on grass, sand, shell, or mud bottoms (Rathbun 1925; Powers 1977). Males are generally larger than females (Hartnoll 1965), but no differences in feeding preferences among sexes or crabs of different sizes have been observed (Coen 1988b). Individ- uals used in this study ranged from 0.40 g to 2.00 g in wet weight and 8.2 mm to 15.6 mm in carapace width.

Neogoniolithon stricturn (hereafter referred to as Neogonioli- rhon) is a branching crustose (i.e., not articulated) coralline alga that commonly grows in shallow water in coral reef and grass bed habitats throughout the Caribbean (Taylor 1960; Littler et al. 1981). Although the branching morphology of this species is vari- able (e.g., Bosence 1985), the spaces between branches are usually large enough to provide a habitat for a diverse community of in- vertebrates including sessile (ascidians, sponges) and mobile @to- matopods. decapod crabs) groups.

Crab abundance

To determine the density of crabs on Neogoniolirhon, we haphaz- ardly collected 40 separate clusters of Neogoniolirhon from a shal-

OECOLOGIA 105 (1996) 0 Springer-Verlag 379

(Gmelin) Montagne. Despite a recent revision of the taxonomy of the genus Dicfyota (Hornig et al. 1992), we were not able to un- ambiguously assign the Diczyota to a specific species. Dasya and Padina were in limited supply and were used only in the choice assay with crabs. Because stress can affect algal palatability to herbivores (Renaud et al. 1990; Cronin and Hay in press), we min- imized algal stress by placing the algae in coolers with fresh sea water for transport to the lab, sorting and holding the algae in flow-through seawater tanks upon arrival (within 0.5 h of collec- tion), and beginning all assays within 6 h of collection.

To determine the relative dietary preferences,of Mithrax, we placed one crab in each of 20 separate 1.4-1 bowls, each of which held a 60-80 mg piece of all seven seaweed species collected (i.e., a choice assay). As a control for changes in seaweed mass unrefat- ed to herbivory, 20 bowls without crabs contained pieces of the same species of seaweed. Within each replicate, treatment and control pieces of algae were taken from the same algal thallus to provide an accurate estimate of changes in algal mass due to fac- tors unrelated to the activities of the crab. After 48 h of grazing, each piece of alga was reweighed, and changes in mass of treat- ment algae were compared to changes in their paired control por- tions using a paired two-tailed t-test.

To determine the willingness of crabs to consume the various seaweeds when no alternative choice was available (i.e., a no- choice assay), we conducted Mithrax feeding assays in which each crab was offered only one species of alga. Pieces of five species of algae (Dictyota sp., Halimeda sirnufans, and the three Luurencia species) between 1.00 and 2.00 g wet weight with appropriate controls were offered to Mithrax as described previously, with the exception that each crab had access to only one species of alga. There were eight replicates for each seaweed species, and assays for all five seaweeds ran simultaneously for 48 h. We tested for significant differences in mass change in grazed versus control al- gae for each species using the Mann-Whitney U-test.

To calculate net mass loss due to crab feeding for each algal species in the choice and no-choice assays, we corrected for changes in mass unrelated to herbivory using the formula, [Ti x (CdC,)] - T,, where Ti and T, are the initial and final masses of the seaweed portion in the container with a crab, and Ci and C, are the initial and final masses of the seaweed portion in the paired control container. Preliminary data indicated that there was no re- lationship between the size of the crab and the amount of algae consumed (see also Coen 1988b), so results are reported as milli- grams of alga consumed per 48 h. Data from no-choice assays were analyzed by one-way ANOVA, but choice assays cannot be analyzed this way due to the non-independence of treatments (see Peterson and Renaud 1989). We therefore analyzed choice data us- ing the non-parametric Friedman’s two-way test on ranked data, which allows for dependence among treatments, provided that each replicate is independent (Conover 1980; Coen 1988b). Among-species comparisons were made using Friedman’s multi- ple comparisons test (Conover 1980) with Sidak’s correction (So- kal and Rohlf 1981).

Preferences of reef fishes for the seaweeds used in the no- choice crab feeding assay were determined by offering all five species to a natural assemblage of reef fishes in the field. .We placed a 5-cm-long piece of each seiweed species between the strands of a 0.5-m-long piece of bi ided polypropylene rope (n = 23), and fastened these ropes at a depth of 2-3 m on Pickles Reef, near Key Largo. Grazing by reef fishes was allowed to pro- ceed for 1.5 h, and then each species on each rope was recorded as either still present or totally consumed (see Hay 1984a; Paul and Hay 1986 for more details of this method). Among-species differ- ences in the frequency of total consumption were determined by contingency table analysis using the G-statistic with Sidak’s cor- rection for multiple comparisons.

low (<I m) grass bed near Rodriguez Key, and placed each indi- vidually in plastic bags underwater. These were returned to the laboratory where each was inspected carefully for crabs. We re- corded the wet mass of each clump and the number of Mithrar oc- cupying it.

Fouling experiment

We tested whether Mithrax prevents seaweed overgrowth of Neo- goniofithon by removing crabs from some algal clusters and com- paring the mass of epiphytes on algal hosts without crabs to the mass of epiphytes on control clusters with crabs. We constructed a floating rack that held forty 1.4-1 containers, each with four 1.2- cm-diameter holes equally spaced along the wall of each container to allow lateral exchange of water with the environment. This rack was anchored in a grass bedalgal flat (1 .O m deep at low tide) on the seaward side of Key Largo. The rack was deployed such that the top of each container floated at the sea surface, and the bogom (where the crab and coralline alga were) was 15 cm below the sur- face. Wave action provided water movement through the holes in the containers. We placed in each container a 70-100 g cluster of Neogoniolithon that was determined by visual examination to be free of epibionts (this approximated the mean-sized cluster on which we found one crab). We placed one crab in each of 20 hap- hazardly selected containers (the other 20 containers served as controls), and covered all containers with a 0.4 cm mesh to pre- vent loss of the crabs. There was no significant difference in the initial weight of the Neogoniolithon clusters with and without crabs (two-tailed t-test, equal variances; P = 0.8137) and treatment and control replicates were interspersed randomly in the rack.

After 30 days, the contents of each container were removed, sealed in individual plastic bags with seawater, and transported to the laboratory for quantification of epiphyte mass. We were able to remove some of the epiphytic growth from Neogoniolithon by im- mersing each cluster in sea water and shaking vigorously for 15 s. Some remaining epiphytes were removed using a jet of sea water from a squirt bottle (volume used = 200 ml per cluster). However, thorough removal of epiphytes could only be achieved by breaking the cluster into pieces and manually removing them. Basal por- tions of some epiphytes remained on the more fouled host-plants even after cleaning. This, combined with the slow growth rate of coralline algae in general (Adey and McKibbin 1970; Adey and Vassar 1975; Littler and Arnold 1982; Steneck 1986), prevented us from obtaining reliable estimates of coralline growth in mass over this 30-day period. For this reason, we compared only the mass of epiphytes, not growth rates of Neogoniolithon, between treat- ments.

The water containing the epiphytes removed from Neogoniofi- rhon was filtered in a 0.5-mm nitex sieve to remove larger epi- phytes, and again in a 0.2-mm nitex sieve to separate smaller epi- phyte fragments. Seagrass and other obvious drift items, such as bits of plastic, were removed from the filtered material, and the mass of epiphytes dried in an oven at 65°C for 30 h, then weighed to the nearest 1 mg. To standardize the mass of epiphytes for hosts of different size, the data were expressed as dry mass of epiphytes as a percentage of host wet mass. To achieve homoscedasticity of variances, the data were arcsine transformed before performing statistical analysis by an unpaired two-tailed t-test.

Whole-plant palatability to crabs and fishes

To assess the potential of M. sculptus to be an effective cleaner of Neogoniolithon we used both choice and no-choice assays to de- termine the crab’s feeding preferences for macroalgae that com- monly co-occur with Neogoniolithon. We collected seven of the most abundant species growing subtidally at the site of the fouling experiment: the calcareous green alga Hulimeda simuluns Howe: the lightly calcified brown alga Pudina gymnosporu (Keutzing) Sonder, and another brown alga in the genus Diccora; and the red algae Laurenciu pupiffosa (C. Agardh) Greville, L. intricatu Lam- ouroux, L. poitei (Lamouroux) Howe, and Dasyu baillouviuna

Effects of algal extracts on crab feedine

To determine how seaweed chemical defenses might affect herbi- vory, Iipophilic crude extracts of each alga were obtained by gnnding a known fresh volume of each species in a blender in 2 1

380 OECOLOGIA 105 (1996) Q Springer-Verlag

dichloromethane (DCM): methanol (MeOH). Solids were filtered and solvents removed by rotary evaporation. Water soluble materi- als in this extract were removed via dichloromethane/water parti- tion and discarded. We did not test the water-soluble portion of the extract for deterrent effects on crab feeding because (1) the crabs feed slowly, and water-soluble materials would likely leach out of the artificial food before the completion of the assay, thereby com- promising the results of these tests; and (2) previous assays of li- pophilic and water soluble extracts from tropical marine algae have indicated that deterrent activity is common in the lipophilic extract, but rare in the water soluble extract (Bolser and Hay, in press).

To qualitatively determine the presence of lipid soluble sec- ondary metabolites in each seaweed species, we examined the contents of each crude lipophilic extract by thin layer chromatog- raphy (TLC). TLC plates spotted with the lipophilic extracts of each species were developed using three different solvent sequenc- es: (1) 100% hexane only, which allowed us to optimally resolve the least polar metabolites in the extract; (2) 100% hexane fol- lowed by 1:l hexane/ether, which allowed us to clearly view com- pounds of intermediate polarity, and (3) 100% hexane followed by 1:l hexane/ether and then 100% ether, to resolve the more polar compounds in the lipophilic extract. Developed plates were sprayed with 50% sulfuric acid, then heated with a hot-air gun. This caused both primary and secondary metabolites to be con- verted to colored decomposition products for easy visualization of separated compounds. Additionally, the color after acid charring can be characteristic of certain types of compounds.

To test if feeding preferences observed in whole-plant assays were related to plant secondary chemistry, we fed crabs artificial foods incorporating the lipophilic crude extract of each algal spe- cies using a method modified from Hay et al. (1994). The artifi- cial food was made of freeze-dried, powdered Ulva (a green alga palatable to many herbivores) mixed in an agar base, and formed onto a strip of window screen. The screen provided a matrix to hold the food, as well as a grid that allowed us to quantify the amount of food removed by the herbivore by counting empty squares (see Fig. 1 in Hay et al. 1994). Algal secondary metabo- lites were added to the food by dissolving the lipophilic crude ex- tract from 2.0 g of seaweed in anhydrous ether, and adding this solution to 2.0 g of dry Ulva powder in a small flask. Enough ether was added to cover the powder with liquid. The solvent was then removed by rotary evaporation, resulting in a uniform coat- ing of the extract on the algal particles. Control foods were treated identically, but without the addition of extract to the ether. For more detailed description of the preparation of the artificial food, see Hay et al. (1994).

The sections of window screen holding the artificial foods with and without seaweed extract were cut into pieces measuring 10 x 10 squares, and offered together to individual crabs in 1.0 1 dishes containing fresh seawater. Experiments were monitored regularly, and feeding was allowed to continue until either (1) over 50% of either the treatment or control food was eaten; or (2) the termination of the experiment (for replicates with low rates of feeding - usually after 12-24 h). If the herbivores did not feed or ate all of both foods between monitoring intervals, then that repli- cate was excluded from analysis because it provided no informa- tion on relative palatability of the foods. The mean difference be- tween the number of treatment and control squares eaten for each species was analyzed by a two-tailed paired sample t-test.

Crab susceptibility to predation

To determine if Neogoniolithon could provide a refuge from pre- dation for Mirhrax, we tethered crabs both within reach of Neo- goniolithon clusters and in the open on a patch reef 3-4 m deep at Dry RocksAVhite Banks near Key Largo. We tied a 10-12 cm length of 0.18-mm-diameter monofilament line to a 10-cm-long galvanized nail and affixed the other end of the line to the center of the carapace of the crab using Duro Quick Gel no-run super glue (Loctite Corp.). We blotted the crab dry with a paper towel,

then placed a single droplet of glue on the back of the carapace. We allowed the drop to sit for about 30 s to gel, then placed the end of the monofilament line into the glue, and held it in place for 30 s to allow the glue to set. The crabs were then paired by size and each pair put in a small container full of seawater and covered with a lid for transport. One crab of each pair was tethered such that it had access to a cluster of Neogoniolirhon that we had placed on the reef, and the other was tethered within 50 cm of the first, but without access to a Neogoniolithon cluster. We monitored both crabs for the first few minutes of the experiment to be sure that the tether was holding, and then checked again after 0.5 and 5 h, re- cording the presence or absence of the crabs in each treatment. The frequency of consumption of crabs tethered in the open was compared to that of those tethered near Neogoniolithon by Fisher’s exact test.

Substrate choice

To determine if Mithrax sculptus preferred Neogoniolithon over other habitats, we placed crabs in 1.4-1 round bowls (one crab per bowl) with two substrates having approximately equal surface area (visually estimated). One of the two substrates in every trial was a piece of Neogoniolithon judged visually to be free of large epi- bionts. Alternative substrates were (1) a piece of Halimeda sp. (n = 14; a mix of H . monile and H. incrassuta); (2) dead fragments of the coral Porites sp. encrusted with a filamentous brown algae ( n = 16); or (3) N. strictum encrusted with ascidians (n = 30). Each substrate was placed haphazardly within the container and the two choices were placed far enough apart (not touching) to allow us to unambiguously determine to which substrate the crab moved. We waited between 0.5 and 2 h to allow the crabs to choose be- tween the habitats, and then recorded which habitat housed the crab in each replicate. For each pair of substrate types, the ob- served frequency of habitat selection was compared to that expect- ed if there were no significant preference between substrates using a G-test.

Results

Crab density and fouling experiment

M. sculptus was the dominant mobile organism associat- ed with Neoguniolirhon at our study site, although sever- al other crabs, stomatopod shrimp, and small fishes were also present at much lower frequencies. Of the 40 Neo- goniolithon clusters examined, 30 contained at least one Mithrax sculptus, and the mean mass of alga per crab was 75.1 g.

After 30 days in the field, clusters of Neogoniolithon without crabs were covered with filamentous algae, and were pale in color, while clusters with crabs appeared free of epiphytes and were pink in color. Crabs were missing from seven of the containers that initially held crabs, so these were excluded from the analysis. Neo- goniolithon clusters without crabs sepported a mass of epiphytes 9 times greater than clusters with crabs (Fig. 1, P < 0.0001, two-tailed t-test).

Whole-plant palatability

All species of algae exposed to crab grazing showed greater mass loss than controls for both choice ( P < 0.05, two-tailed t-test, n = 20) and no-choice assays ( P < 0.05,

OECOLOGIA 105 (1996) 0 Springer-Verlag 38 1

P < .o001 h 3 1

P<.ooo1 1m-I A N=23

Mithrax Mithrax absent present

Fig. 1 Dry mass of fouling organisms (mean 2 ISE) as a percent- age of host wet mass on Neogoniolithon clusters with and without resident crabs after 30 days in the field. Analysis by two-tailed t- test; n is given at the base of each bar

Mann-Whitney U-test, n = 8 for each species), indicating that crabs fed on all species of algae (Figs. 2B,C). When offered a choice between all seven species of algae, crabs showed relatively few significant preferences among al- gal foods (Fig. 2B). Dasya baillouviana was eaten less than the four species that were most preferred and Lau- rencia papillosa was eaten significantly less than two of the more heavily consumed species (Fig. 2B, P < 0.05, Friedman’s multiple comparisons test with Sidak’s cor- rection). Thus, in the choice assay, there was no signifi- cant preference expressed between the following poten- tial foods: Padina gymnospora, L. poitei, Halimeda sirnulam, L. intricara, or Dictyota sp. In the no-choice assay, feeding differed significantly among the various foods (Fig. 2C, P = 0.0039, one-way ANOVA), but this effect was produced solely by L. papillosa being con- sumed more than Dictyota, Halimeda, L. poitei, or L. in- tricata [Fig. 2C, P < 0.05, Fisher’s protected least squar- es difference (PLSD)]. These last four species did not differ from each other in either choice or no-choice as- says. L. papillosa was an intermediate to low preference species in the choice assay, yet consumed at high rates in the no-choice assay, possibly indicating compensatory feeding on L. papillosa when alternate foods are unavail- able.

Parrotfishes (Scaridae) and surgeonfishes (Acanthur- idae) were commonly observed feeding on the algae that we transplanted onto Pickles Reef, and these fishes fed selectively (Fig. 2A, P < 0.0001, G-test). L. papillosa and L. intricata were most preferred, L. poitei was of in- termediate preference, and Dictyota sp. and Halimeda simulans were low preference foods (Fig. 2A; P < 0.05; pairwise G-tests, holding the significance level constant at 0.05 using Sidak’s correction). Feeding preferences clearly differed between crabs and fishes, as there is no significant correlation between the feeding preference ranks of crabs and fishes, when both were given a choice of algal foods (Kendall rank correlation tau = -0.400; P = 0.3272; n = 5) .

d0-l P

Fig. 2A-C Comparison of feeding preferences of reef fishes and the crab, Mithrax sculptus. A Frequency of total consumption of algae exposed to a natural assemblage of reef fishes at Pickles Reef, Florida. Analysis by G-test with Sidak’s correction for mul- tiple comparisons (k0.05). B Wet mass of tissue consumed by crabs (corrected for autogenous changes unrelated to herbivory, see Methods) for 7 species of seaweed offered simultaneously for a 48 h period (mean 2 ISE). Analysis by Friedman’s rank test (P<O.OOO I), and multiple comparisons made using Friedman’s multiple comparison test with Sidak’s correction (P<0.05). Initial masses of all pieces of algae were between 60 and 80 mg wet weight. C Wet mass of tissue consumed by crabs (corrected for autogenous changes unrelated to herbivory, see Methods) for 5 species of seaweed offered separately for a 48-h period (mean f ISE). Analysis by one-way ANOVA, followed by Fisher’s PLSD at P<0.05. Initial masses of all pieces of algae were be- tween 1 .OO and 2.00 g wet weight

Crude extract feeding assays

TLC indicated that L. papillosa lacked secondary metab- olites, but that all other algae contained several lipophilic secondary metabolites. Dictyota sp. and L. intricata were most chemically rich, each containing at least five obvi- ous (producing large, brightly colored spots on the TLC plate) secondary metabolites. Both Halimeda simulans and L. poitei contained one or two obvious secondary metabolites and several that were less apparent (produc-

OECOLOGIA 105 (1996) Q Springer-Verlag

tmatment controi tethered intheopen

tethered in Neogonidithon

P < .oO01 P =.0062

Fig.3 Number of squares of artificial foods consumed by crabs (mean k ISE). Treatment is artificial food with seaweed extract; control is artificial food without extract (see Methods). Analysis by paired sample r-test (twetailed); n is given at the base of each pair of bars

ing less distinct or smaller spots on the TLC plate). The genus Dicryoru has recently undergone taxonomic revi- sion (Hornig et al. 1992), so prior reports of secondary chemistry in Dicfyora by species may be confused. How- ever, the genus Dicfyota is, in general, very rich in sec- ondary chemistry, including several diterpene alcohols (Faulkner 1993, and references cited therein) which are known to inhibit feeding by various herbivores (Hay and Fenical 1992). L. inrricuru is known to contain many non-polar secondary metabolites, but the bioactivity of these compounds has not been reported (Faulkner 1986, 1991, 1993). Many species of Hulimedu, including H. simuluns, produce the secondary metabolites halimed- atrial and halimedatetraacetate in significant quantities; both of these compounds deter feeding by reef fishes (Paul and Fenical 1983; Hay et al. 1988; Paul and Van Alstyne 1992). L. poirei contains several secondary me- tabolites including poitediol, dactyol and poitediene, but the bioactivity of these compounds has not been reported (Faulkner 1984, 1986). No secondary metabolites have been reported from L. pupillosu (Faulkner 1993, and ref- erences cited therein). Four of the five species tested contain obvious secondary metabolites, but only the ex- tracts of Dictyotu sp. and H . simuluns significantly de- terred feeding by Mithrax (Fig. 3; P = 0.0406 for Dicty- otu and P = 0.0497 for Hulimedu; two-tailed r-tests).

0.5 5.0

Hours since tethering

Fig.4 Number of crabs consumed by reef fishes when tethered with or without access to Neogoniolithon clusters (total number of crabs of each treatment = 11). Analysis by Fisher's exact test

Fig. 5 Percent of crabs choosing Neogoniolithon versus an alter- native habitat. Analysis by G-test; n is given at the base of each pair of bars

In laboratory assays, crabs showed no preference be- tween Neogoniolirhon that was free of fouling organisms and Neogoniolirhon partially overgrown by ascidians (Fig. 5, P = 0.1201, G-test). Crabs also showed no pref- erence between clean Neogoniolithon clusters and Po- rires fragments (Fig. 5, P = 0.4795, G-test). However, a strong preference for Neogoniolifhon over Halimedu was observed (Fig. 5, P < O.OOO1, G-test), despite the fact that Hulimeda was a high-preference food (Fig. 2B), suggesting that food may not be the most important fac- tor in habitat selection.

Tethering experiment and habitat choice Discussion

Crabs tethered near Neogoniolithon clusters were preyed upon significantly less than crabs tethered in the open af- ter both 0.5 and 5 h (Fig. 4; P < 0.0001 and P = 0.0062, respectively; Fisher's exact test). We observed blue- headed wrasse (Thalassoma bifasciutum) and gray angel- fish (Pomucanrhus arcarus) attacking crabs that were tethered in the open.

Herbivores are most often thought of as the natural ene- mies of plants, but there are situations where grazers can benefit the plants they associate with or consume. Some of the more common examples involve an increase in propagule release or settlement due to feeding on repro- ductive tissue that survives gut passage (Santelices

. 1992). As an example, the amphipod Hyale media feeds preferentially on the mature cystocarpic tissue of the red alga Irideaea laminarioides resulting in an increase in spore release over ungrazed plants (Buschmann and San- telices 1987). Settlement of algal spores to the bottom is enhanced by spore incorporation into herbivore fecal ma- terial after ingestion (Santelices and Ugarte 1987), which may increase growth and development of the spore by providing a nutrient rich microenvironment (Santelices 1992). This same process can also enhance dispersal of algal propagules (Buschmann and Vergara 1993), much as fruit and seed predation by mobile animals can in- crease seed dispersal if seeds survive gut passage (review by Janzen 1983). Removal of plant biomass may in- crease the productivity of surviving plants and plant por- tions in terrestrial (McNaughton 1983, 1985; Paige and Whitham 1987), freshwater (Flint and Goldman 1975; Porter 1976; Lamberti and Resh 1983) and marine (Car- penter 1981, 1986; Klumpp et al. 1987) environments, especially at intermediate grazing intensities. In most of these cases, the increased production is thought to result from decreased self-shading or increased nutrient avail- ability from herbivore excretions. However, the long- term effects of herbivory are poorly understood, and short-term increases in productivity may be achieved by depletion of reserves or reduced reproductive output (Belsky 1987).

Grazing that has some positive effects may also in- volve a cost to the plant due to a loss of tissue, but where herbivores feed primarily on epiphytes the host plant may suffer little, if any, direct damage from associated grazers (Brawley and Adey 1981; Brawley 1992; Duffy 1990). Additionally, even if herbivores do directly graze a host plant, the overall effect on the host can be positive if the herbivore does even greater damage to nearby competitors of the host (Steneck et al. 1991). Small amounts of epiphytic growth can have beneficial effects on host plant growth by minimizing photoinhibition in shallow waters (Norton and Benson 1983). However, most of the effects of epiphytes on host plant growth and survivorship are negative (Sousa 1979; Orth and van Montfrans 1984; DAntonio 1985; Brawley 1992; Wil- liams and Seed 1992; Wahl and Hay 1995). Following the removal of grazing crabs from Neogoniolithon, epi- phytes completely covered the surface of the host plant, shading it from light. Such overgrowth of coralline algae frequently results in mortality (Littler and Doty 1975; Steneck 1982, 1986) or at minimum a competitive su- pression of growth (Steneck et al. 1991).

Coralline algae are slow-growing (Adey and McKib- bin 1970; Adey and Vassar 1975; Littler and Arnold 1982; Steneck 1986) and highIy resistant to herbivory (Steneck 1986, 1992), but may be rapidly overgrown in the absence of herbivores (e.g., Paine and Vadas 1969; Littler and Doty 1975; Wanders 1977; Brock 1979; Ste- neck 1982; Hay and Taylor 1985). Thus, in environments like seagrass beds where herbivory on seaweeds is rela- tively low (Ogden et al. 1973; Hay 1984a,b), corallines are competitively disadvantaged relative to faster-grow-

OECOLOGIA 105 (1996) 0 Springer-Verlag 383

ing fleshy species. This appears to be true for Neo- goniolifhon, as clusters isolated from grazing for only 30 days developed a heavy coating of epiphytes (Fig. 1). By harboring herbivorous crabs, Neogon!olithon increases localized, grazing pressure, potentially tipping the com- petitive balance in its favor. However, because Neo- goniolithon commonly grows in sedimentary environ- ments where it may be subject to frequent burial, it may also be able to tolerate some degree of shading without fatal consequences.

These crabs can be effective cleaners of Neogonioli- thon because they readily consume a wide ,range of sea- weeds (Fig. 2B), including species that are defended from other grazers by secondary metabolites [e. g., Ha- limeah, Dictyota, and some species of Laurencia (Hay et al. 1987; Hay 1991; Paul 1992)l. Fishes consumed al- most no Halimeda or Dictyotu in whole-plant assays (Fig. 2A), and, correspondingly, the extracts or pure sec- ondary metabolites from these algae have been repeated- ly shown to deter feeding by reef fishes (reviewed by Hay 1991; Paul 1992). Interestingly, extracts from Dicty- ora and Halimeda deterred Mithrax feeding (Fig. 3) even though Mithrax readily consumed both species in whole- plant assays (Fig. 2B,C).

Several hypotheses might explain this apparent incon- gruity between Mithrax feeding in the whole-plant as- says (Fig. 2B,C) and the effects of algal extracts on feed- ing (Fig. 3). First, food nutritional quality can interact with chemical defenses to determine the overall accept- ability of a food to reef consumers like fishes and ur- chins (Duffy and Paul 1992; Hay et al. 1994). Nutrition- al quality might be even more important for less mobile grazers such as crabs. Small crustacean herbivores are highly susceptible to predation, so their foraging time and prey choices can be much more constrained than are those of fishes (Duffy and Hay 1991b, 1994). Thus, crabs might need to bias their dietary choices in favor of those foods that are high in value (protein, essential nu- trients, energy), and be less constrained by secondary metabolite content. As a possible example, Laurenciu papillosa was among the lower-preference species in choice assays (Fig. 2B), but its extract did not deter crab feeding (Fig. 3) and in no-choice assays Mithrax con- sumed twice as much L. papillosa as any other species offered (Fig. 2C). Coupled with the findings that L. pa- pillosa has high water content and low energy content relative to other seaweeds (Coen 1988b), these patterns suggest that Mithrax may avoid L. papillosa when more valuable species are available (Fig. 2B), but can com- pensate for its low value by increased feeding where al- ternative foods are unavailable (Fig. 2C). Herbivorous fishes, in contrast, prefer L. pupillosu over most other species offered in choice assays (Fig. 2A), possibly be- cause it was the only seaweed species without secondary metabolites, and secondary metabolites from Hulimedu, Dictyota and other Laurencia spp. are known to deter feeding by reef fishes (Hay 1991; Paul 1992). Although these crabs appear to exhibit compensatory feeding in the laboratory (Fig. 2C), this could be constrained in the

384 0 E C 0 L 0 GI A 105 ( 1996) 0 Springer-Verlag

field if increased foraging time involved increased risk of predation. Under these conditions, crabs that tolerate seaweed secondary metabolites may be advantaged if doing so allows them to consume more valuable foods and reduce the predation risk associated with increased foraging.

The differences in palatability between a whole plant and its chemical extract could also be resolved if crabs can deactivate plant chemical defense mechanisms be- fore they take effect. This might occur in the same way that proteins in the saliva of mule deer bind tannins, eliminating their digestibility reducing effects (Robbins et al. 1987). As a possible example, Mithrax might cir- cumvent the defenses of Hulimeda by deactivating the mechanism that converts the less-deterrent secondary metabolite halimedatetraacetate to the more-deterrent ha- limedatrial upon crushing of plant tissue (Paul and Van Alstyne 1992). When Hulirnedu is extracted by the meth- od used in this study, the conversion to the more deter- rent compound takes place during the extraction, and thus would preclude the crab from interfering with this reaction. Regardless of the reason, M. sculptus consumes a wide range of algae and appears to use factors other than plant chemical defenses in choosing food, suggest- ing that it could prevent chemically defended seaweeds from overgrowing Neogoniolithon.

Despite reciprocal benefits between the crab and cor- alline alga, Mithrax is not a specialist on Neogoniolithon, and it is unlikely that the relationship is coevolved. Habi- tat architecture, including spatial complexity, shape, and color, is known to play an important role in habitat selec- tion by small crustaceans such as amphipods (Hacker and Steneck 1990) and shrimps (Hacker and Madin 1991). Habitat choice for Mithrax also appears to be based on the physical structure of the habitat and the de- gree of protection it affords from predators, as the crab chooses with equal frequency several substrates that pro- vide a similar structural refuge (Fig. 5) . M. sculptus is also commonly associated with the branching coral Po- rites, where it removes encroaching seaweeds (Coen 1988a). Similarly, Neogoniolithon is known to harbor an- other herbivorous crab species (M. coryphe) that appears to exhibit similar food choice patterns to M. sculptus (J. Stachowicz and M. Hay, personal observation), and may also serve a cleaning function.

Association with a structurally complex sessile organ- ism provides a refuge from predation for Mithrax, as crabs which were given access to Neogoniolithon clus- ters were far more likely to survive than crabs on ex- posed substrate (Fig. 4). Gut analyses of tropical fishes (Randall 1967) showed that 22 species of fishes in 13 families contained crab parts identified as belonging to Mithrax species, so predation by a wide variety of fishes is likely to be a major source of mortality for these crus- taceans in tropical habitats. Many other crustaceans ex- perience high predation pressure from predatory and om- nivorous reef fishes (Randall 1967), and both crabs (Hay et al. 1989, 1990b) and amphipods (Hay et al. 1990a; Duffy and Hay 1991b, 1994) reduce encounter rates with

predators and increase survivorship by associating with seaweeds that are unpalatable to fishes.

Mutually beneficial plant-animal relationships have previously been documented in both marine (Steneck 1982; Branch et al. 1992) and terrestrial (Janzen 1966; O’Dowd and Wilson 1991; Cushman and Beattie 1991) environments. An herbivore of low mobility that cannot easily relocate may be likely to develop a close associa- tion with its host (Steneck 1982, 1992), especially when that herbivore uses the host as an important refuge from predation (Duffy and Hay 1991a,b, 1994; reviewed in Hay and Steinberg 1992). In these cases, the herbivore may be expected to minimize damage to the host plant in order to preserve its host’s value as a refuge. Such rela- tionships may be most common when the host provides a spatial refuge from potential predators, but is in an envi- ronment where it competes poorly with other sessile or- ganisms, thus providing an ample food source for resi- dent herbivores through epiphytes and encroaching com- petitors.

Predation pressure rather than feeding specialization appears to drive crustaceans to seek these associations with herbivore-resistant seaweeds (Hay 1992; Hay and Steinberg 1992; Duffy and Hay 1994). Because of their susceptibility to predation, crabs and other small herbi- vores may be unable to spend as much time foraging as larger, more mobile, or better defended, herbivores such as fishes or urchins. Crabs may, therefore, need to choose the most profitable foods rather than the least de- fended ones. These feeding preferences may be driven by predator avoidance strategies, but they coincidentally benefit the host alga by removing both palatable and chemically defended competitors. It is doubtful that Neo- goniolithon has evolved in response to selective pressure by M. sculptus, as coralline algae evolved present morphologies well before the advent of modem herbi- vores (Steneck 1992), and other small herbivores (e.g., other crabs, chitons, amphipods, etc.) may also be capa- ble of removing epiphytes from Neogoniolithun. While it is possible that diffuse coevolution (Fox 1981) between small herbivorous crustaceans and branched coralline al- gae has occurred to enhance the relationship initiated by chance, this appears unlikely because the participants in this association do not appear to be responding to selec- tive pressure from each other, but to pressure from pre- dators or competitors outside the association (Vermeij 1983).

Acknowledgements This investigation was supported by NSF grant OCE 92-02847 to M. E. Hay, and NOAA grants NA88AA- D-UR004 and NA36RU0060 (to N. Lindqllist and M. E. Hay) through the UNCWMURC lab at Key Largo, Florida. J. Stacho- wicz was supported by an NSF Graduate Research Fellowship. Austin Williams identified our crabs. Dave Colby helped with sta- tistical advice. Robin Bolser, Mike Deal, Mike Klompas and Greg McFall provided field assistance. Comments from Charles Peter- son, Niels Lindquist, Robert Steneck, Phyllis Coley, and an anony- mous reviewer improved the manuscript.

0 E C 0 L 0 G I A 1 05 ( 1996) 0 Springer-Verlag 3 85

Duffy JE (1990) Amphipods on seaweeds: partners or pests? Oecologia 83:267-276

Duffy JE, Hay ME (1991a) Amphipods are not all created equal: a reply to Bell. Ecology 72:354-358

Duffy JE, Hay ME (199lb) Food and shelter as determinants of food choice by an herbivorous marine amphipod. Ecology 72: 1286-1298

Duffy JE, Hay ME (1994) Herbivore resistance to seaweed chemi- cal defense: the roles of mobility and predation risk. Ecology

Duffy JE, Paul VJ (1992) Prey nutritional quality and the effec- tiveness of chemical defenses against tropical reef fishes. Oecologia 90:333-339

Estes JA, Steinberg PD (1988) Predation, herbivory, and kelp evo- lution. Paleobiology 14: 19-36

Faulkner DJ (1984) Marine natural products: metabolites of ma- rine algae and herbivorous marine mollusks. Nat Prod Rep

Faulkner DJ (1986) Marine natural products. Nat Prod Rep 3: 1-33 Faulkner DJ (1991) Marine natural products. Nat Prod Rep

Faulkner DJ (1993) Marine natural products. Nat Prod Rep

Flint RW, Goldman CR (1975) The effects of a benthic grazer on the primary productivity of the littoral zone of Lake Tahoe. Li-- mnol Oceanogr 20:935-944

Fox LR (1981) Defense and dynamics in plant-herbivore systems. Am Zoo1 2 1 :853-864

Gaines SJ, Lubchenco J (1982) A unified approach to plant-herbi- vore interactions. 11. Biogeography. Annu Rev Ecol Syst

Hacker SD, Madin LP (1991) Why habitat architecture and color are important to shrimps living in pelagic Sargassum: use of camouflage and plant-part mimicry. Mar Ecol Prog Ser 70: 143-155

Hacker SD, Steneck RS (1990) Habitat architecture and the abun- dance and body-size-dependent habitat selection of a phytal amphipod. Ecol71:2269-2285

Hartnoll RG (1965) The biology of spider crabs: a comparison of British and Jamaican species. Crustaceana 9:l-16

Hay ME (1 98 1 a) The functional morphology of turf-forming sea- weeds: persistence in stressful marine habitats. Ecology

Hay ME (1981b) Herbivory, algal distribution, and the mainte- nance of between-habitat diversity on a tropical fringing reef. Am Nat 1 18:520-540

Hay ME (1984a) Predictable spatial escapes from herbivory: how do these affect the evolution of herbivore resistance in tropical marine communities. Oecologia 64:396-407

Hay ME (1 984b) Patterns of fish and urchin grazing on Caribbean coral reefs: are previous results typical? Ecology 65:44&454

Hay ME (1985) Spatial patterns of herbivore impact and their im- portance in maintaining algal species richness. Proc 5th Int Coral Reef Congr 4:29-34

Hay ME (1991) Fish-seaweed interactions on coral reefs: effects of herbivorous fishes and adaptations of their prey. In: Sale PF (ed) The ecology of fishes on coral reefs. Academic Press, San Diego, pp 96- I 19

Hay ME (1992) The role of seaweed chemical defenses in the evo- lution of feeding specialization and in the mediation of com- plex interactions. In: Paul VJ (ed) lcological roles of marine natural products. Comstock Press, Ithaca, pp 93-1 18

Hay ME, Fenical W (1992) Chemical mediation of seaweed-herbi- vore interactions. In: John DM, Hawkins SJ, Price JH (eds) Plant-animal interactions in the marine benthos (Systematics association special volume 46). Clarendon Press, Oxford, pp

Hay ME, Taylor PR (1985) Competition between herbivorous fish- es and urchins on Caribbean reefs. Oecologia 65591-598

Hay ME. Steinberg PD (1992) The chemical ecology of plant-her- bivore interactions in marine versus terrestrial communities. In: Rosenthal GA, Berenbaum MR (eds) Herbivores: their in- teractions with secondary plant metabolites. 2nd edn. vol 11.

75: 1291-1306

1125 1-280

8~97- 147

10497-539

13:111-138

62:739-750

3 19-337

References

Adey WH, McKibbin DL (1970) Studies on the maerl species Phymatolithon calcareum (Pallas) nov. comb. and Lith- othamnium coralloides Crouan in the Ria de Vigo. Bot Mar 13:100-106

Adey WH, Vassar JM (1975) Colonization, succession and growth rates of tropical crustose coralline algae (Rhodophyta, Crypt- onemiales). Phycologia 14:55-69

Belsky A1 (1987) The effects of grazing: confounding of ecosys- tem, community, and organism scales. Am Nat 129:777-783

Bolser R, Hay ME (in press) Are tropical plants better defended? Palatability and defenses of temperate versus tropcial sea- weeds. Ecology

Bosence DWJ (1985) The morphology and ecology of a mound- building coralline alga (Neogoniolithon strictum) from the Florida Keys. Palaeontology 28: 189-206

Branch GM, Harris JM, Parkins C, Bustamante RH, Eekhout S (1992) Algal ‘gardening’ by grazers: a comparison of the eco- logical effects of territorial fish and limpets. In: John DM, Hawkins SJ, Price JH (eds) Plant-animal interactions in the marine benthos (Systematics association special volume 46). Clarendon Press, Oxford, pp 405-423

Brawley SH (1992) Mesoherbivores. In: John DM, Hawkins SJ, Price JH (eds) Plant-animal interactions in the marine benthos (Systematics association special volume 46). Clarendon Press, Oxford, pp 235-263

Brawley SH, Adey WH (1981) The effect of micrograzers on algal community structure in a coral reef microcosm. Mar Biol

Brock RE (1979) An experimental study on the effects of grazing by parrotfishes and the role of refuges in benthic community structure. Mar Biol51:381-388

Buschmann A, Santelices B (1987) Micrograzers and spore re- lease in lridaea laminarioides Bory (Rhodophyta: Gigartin- ales). J Exp Mar Biol Ecol 108:171-179

Buschmann A, Vergara PA (1993) A field study of the role of rocky intertidal amphipods as algal spore dispersal agents in southern Chile. J Phycol 29:154-159

Carpenter RC (1981) Grazing by Diadema anrillarum (Philippi) and its effects on the benthic algal community. J Mar Res

Carpenter RC (1986) Partitioning herbivory and its effects on cor- al reef algal communities. Ecol Monogr 56:345-363

Cates RG, Orians GH (1975) Successional status and the palat- ability of plants to generalized herbivores. Ecology 56:410-418

Coen LD (1988a) Herbivory by crabs and the control of algal epi- bionts on Caribbean host corals. Oecologia 75: 198-203

Coen LD (1988b) Herbivory by Caribbean majid crabs: feeding ecology and plant susceptibility. J Exp Mar Biol Ecol

Connell JH, Slatyer RO (1977) Mechanisms of succession in natu- ral communities and their role in community stability and or- ganization. AmNat 111:1119-1142

Conover WJ (1980) Practical nonparametric statistics, 2nd edn. Wiley, New York

Crawley MJ (1989) The relative importance of vertebrate and in- vertebrate herbivores in plant population dynamics. In: Ber- nays EA (ed) Insect-plant interactions, vol I . CRC Press, Boca Raton, pp 45-7 1

Cronin G. Hay ME (in press) Susceptibility to herbivores depends on recent history of both the plant and animal. Ecology

Cushman JH, Beattie AJ (1991) Mutualisms: assessing the bene- fits to hosts and visitors. Trends Ecol Evol 6:193-195

D‘ Antonio C (1985) Epiphytes on the rocky intertidal red alga Rhodomela lark (Turner) C. Agardh: negative effects on the host and food for herbivores? J Exp Mar Biol Ecol

Dethier M ( 1 98 1 ) Heteromorphic algal life histories: the seasonal pattern and response to herbivory of the brown crust, Rolfsia californica. Oecologia 49:333-339

61 : 167-1 77

39~749-765

122:251-276

86~97-218

386 OECOLOGIA 105 (1996) 0 Springer-Verlag

Evolutionary and ecological processes. Academic Press, San - - Diego, pp 371-413

Hav ME. Duffv JE. Pfister CA. Fenical W (1987) Chemical de- , I . , fense agaiist different marine herbivores: are amphipods in- sect equivalents? Ecology 68:1567-1580

Hay ME, Paul VJ, Lewis SM, Gustafson K, Tucker J, Trindell RN (1988) Can tropical seaweeds reduce herbivory by growing at night? Die1 patterns of growth, nitrogen content, herbivory, and chemical versus morphological defenses. Oecologia 75:233-245

Hay ME, Pawlik JR, Duffy JE, Fenical W (1989) Seaweed-herbi- vore-predator interactions: host-plant specirtlization reduces predation on small herbivores. Oecologia 8 1 :418-427

Hay ME, Duffy JE, Fenical W (1990a) Host-plant specialization decreases predation on a marine amphipod: an herbivore in plant’s clothing. Ecology 7 1 :733-743

Hay ME, Duffy JE, Paul VJ, Renaud PE, Fenical W (1990b) Spe- cialist herbivores reduce their susceptibility to predation by feeding on the chemically defended seaweed Avrainvilleu longicuulis. Limnol Oceanogr 35: 1734-1743

Hay ME, Kappel QE, Fenical W (1994) Synergisms in plant de- fenses against herbivores: interactions of chemistry, calcifica- tion, and plant quality. Ecology 75:1714-1726

Hornig I, Schnetter R, hd’homme van Reine WF (1992) The genus Dicryotu (Phaeophyceae) in the North Atlantic. I. A new generic concept and new species. Nova Hedwigia 54:45- 62

Janzen DH (1966) Coevolution of mutualism between ants and acacias in Central America. Evolution 20:249-275

Janzen DH (1983) Dispersal of seeds by vertebrate guts. In: Fu- tumya DJ, Slatkin M (eds) Coevolution. Sinauer, Sunderland,

Johnson CR, Mann KH (1986) The crustose coralline alga, Phy- mutotithon Foslie, inhibits the overgrowth of seaweeds wifhout relying on herbivores. J Exp Mar Biol Ecol96: 127-146

Keats DW, Wilton P, Maneveldt G (1994) Ecological significance of deep-layer sloughing in the eulittoral zone coralline alga, Spongites yendoi (Foslie) Chamberlain (Corallinaceae, Rhodo- phyta) in South Africa. J Exp Mar Biol Ecol 175: 145-154

Klumpp DW, McKinnon D, Daniel P (1987) Damselfish temto- ries: zones of high productivity on coral reefs. Mar Ecol Prog Ser 40:41-51

Lamberti GA, Resh VH (1983) Stream periphyton and insect her- bivores: an experimental study of grazing by a caddisfly popu- lation. Ecology 64: 1 124-1 135

Lewis SM (1986) The role of herbivorous fishes in the oganiza- tion of a Caribbean reef community. Ecol Monogr 56: 183-200

Littler DS, Littler MM, Bucher KE, Noms IN (1981) Marine plants of the Caribbean. Smithsonian Institution, Washington Dc

Littler MM, Arnold KE (1982) Primary productivity of marine functional-form groups from Southwestern North America. J

Littler MM, Doty MS (1975) Ecological components structuring the seaward edges of tropical Pacific reefs: the distribution, communities and productivity of Porolithon. J Ecol

Lubchenco J (1983) Lirtorina and Fucus: effects of herbivores, substratum heterogeneity, and plant escapes during succession. Ecology 64:11161123

Lubchenco J, Gaines SD (1981) A unified approach to marine plant-herbivore interactions. I. Populations and communities. Annu Rev Ecol Syst 12:405-437

Masaki T, Fujita D, Hagen NT (1984) The surface ultrastructure and epithallium shedding of crustose coralline algae in the ‘Isoyaki’ area of southwestern Hokkaido, Japan. Hydrobiolo- gia 116/117:218-223

McBrien H, Harmsen R, Crowder A (1983) A case of insect graz- ing affecting plant succession. Ecology 64: 1035-1039

McNaughton SJ (1983) Compensatory plant growth as a response to herbivory. Oikos 40:329-336

McNaughton SJ (1985) Ecology of a grazing ecosystem: the Serengeti. Ecol Monogr 55:259-294

pp 232-262

PhyCOl 18:307-3 1 1

63: 117-1 29

Momson D (1988) Comparing fish and urchin grazing in shallow and deeper coral reef algal communities. Ecology

Moss BL (1982) The control of epiphytes by Hulidrys siliquosa (L.) Lyngb. (Phaeophyta, Cystoseiraceae). Phycologia

Norton TA, Benson MR (1983) Ecological interactions between the brown seaweed Sargassum muticum and its associated fau- na. Mar Biol75: 169-1 77

O’Dowd DJ, Wilson MF (1991) Associations between mites and leaf domatia. Trends Ecol Evol6:179-182

Ogden JC, Brown RA, Salesky N (1973) Grazing by the echinoid Diadem antillarum Philippi: formation of halos around West Indian patch reefs. Science 182:7 15-7 17

Orth RJ, Montfrans J van ( I 984) Epiphyte-seagrass relationships with an emphasis on the role of micrograzing: a review. Aquat Bot 18:43-69

Paige KN, Whitham TG (1987) Overcompensation in response to mammalian herbivory: the advantage of being eaten. Am Nat 129:407416

Paine RT, Vadas RL (1969) The effect of grazing by sea urchins, Strongylocentrotus spp., on benthic algal populations. Limnol Oceanogr 14:710-719

Paul VJ (1992) Seaweed chemical defenses on coral reefs. In: Paul VJ (ed) Ecological roles of marine natural products. Comstock

69: 1367-1 382

21 185-1 91

Press, Ithaca, & 24-50 Paul VJ, Fenical W (1983) Isolation of halimedatrial: chemical de-

fense adaptation in the calcareous reef-building alga, Ha- limeda. Science 221:747-749

Paul VJ, Hay ME (1986) Seaweed susceptibility to herbivory: chemical and morphological correlates. Mar Ecol Prog Ser

Paul VJ, Van Alstyne KL (1992) Activation of chemical defenses in the tropical green algae Hulimeda spp. J Exp Mar Biol Ecol

Peterson CH, Renaud PE (1989) Analysis of feeding preference

Porter KL (1976) Enhancement of algal growth and productivity

Powers LW (1977) A catalogue and bibliography to the crabs of

Randall JE (1967) Food habits of reef fishes of the West Indies.

Rathbun MJ (1925) The spider crabs of America. U S Natl Mus

Renaud PE, Hay ME, Schmitt TM (1990) Interactions of plant stress and herbivory: intraspecific variation in the susceptibili- ty of a palatable versus an unpalatable seaweed to sea urchin grazing. Oecologia 82:217-226

Robbins CT, Mole S , Hagerman AE, Hanley TA (1987) Role of tannins in defending plants against ruminants: reduction in dry matter digestion? Ecology 68:1606-1615

Robles CD, Cubit J (1981) Influence of biotic factors in an upper intertidal community: dipteran larvae grazing on algae. Ecolo- gy 62:15361547

Santelices B (1992) Digestion survival in seaweeds: an overview. In: John DM, Hawkins SJ, Price JH (eds) Plant-animal interac- tions in the marine benthos (Systematics association special volume 46). Clarendon Press, Oxford, pp 363-384

Santelices B, Ugarte R (1987) Algal life-history strategies and re- sistance to digestion. Mar Ecol Prog Ser 55:267-275

Schmitt TM, Hay ME, Lindquist N (1995) Constraints on chemi- cally mediated coevolution: multiple functions for seaweed secondary metabolites. Ecology 76: 107-1 23

Sieburth JM, Conover JT (1965) Sargassum tannin, an antibiotic which retards fouling. Nature 20852-53

Sokal RR, Rohlf FJ (1981) Biometry, 2nd edn. Freeman, New York

Sousa WP ( I 979) Experimental investigations of disturbance and ecological succession in a rocky intertidal community. Ec01 Monogr 49:227-254

Sousa WP, Connell JH (1992) Grazing and succession in marine algae. In: John DM, Hawkins SJ, Price JH (eds) Plant-animal

33 :255-264

160: 191-203

experiments. Oecologia 80:82-86

by grazing zooplankton. Science 192: 1332-1334

the Gulf of Mexico. Contrib Mar Sci Suppl20 1-190

Stud Trop Oceanogr 5:665-847

Bull 129:l-613

interactions in the marine benthos (Systematics association special volume 46). Clarendon Press, Oxford, pp 425-441

Steneck RS (1982) A limpet-coralline alga association: adapta- tions and defenses between a selective herbivore and its prey.

Steneck RS (1986) The ecology of crustose corallines: convergent patterns and adaptive strategies. Annu Rev Ecol Syst

Steneck RS (1992) Plant-herbivore coevolution: a reappraisal from the marine realm and its fossil record. In: John DM, Hawkins SJ, Price JH (eds) Plant-animal interactions in the marine ben- thos (Systematics association special volume 46). Clarendon Press, Oxford, pp 477-491

Steneck RS, Hacker SD, Dethier MN (1991) Mechanisms of com- petitive dominance between crustose coralline algae: an herbi- vore mediated competitive reversal. Ecology 72:938-950

Taylor WR (1960) Marine algae of the eastern tropical and sub- tropical coasts of the Americas. University of Michigan Press, Ann Arbor

Ecology 63~507-522

171273-303

OECOLOGIA 105 (1996) 0 Springer-Verlag 387

Vermeij GJ (1983) Intimate association and coevolution in the sea. In: Futumya DJ, Slatkin M (eds) Coevolution. Sinauer, Sun- derland, pp 3 1 1-327

Vermeij GI (1994) The evolutionary interaction among species: selection, escalation, and coevolution., Annu Rev Ecol Syst

Wahl M, Hay ME (1995) Associational resistance and shared 25:2 19-236

doom: effects of epibiosis on herbivory. Oecologia 102~329-340

Wanders BW (1977) The role of benthic algae in the shallow reef of Curacao (Netherlands Antilles). 3. The significance of graz- ing. Aquat Bot 3:357-390

Williams GA, Seed R (1992) Interactions between macrofaunal epiphytes and their host algae. In: John DM, Hawkins SJ, Price JH (eds) Plant-animal interactions in thC marine benthos (Systematics association special volume 46). Clarendon Press, Oxford, pp I 89-2 1 I