vol. 156, no. 1 the american naturalist july 2000 Geographic Variation in Camou¯age Specialization by a Decorator Crab John J. Stachowicz1,* and Mark E. Hay 2 1. Department of Marine Sciences, University of Connecticut, It has been argued that the ecological and evolutionary Groton, Connecticut 06340; and Institute of Marine Sciences, importance of positive interactions among species has University of North Carolina, Morehead City, North Carolina been underappreciated by most biologists (Kareiva and 28557; Bertness 1997). While many factors likely contribute to 2. School of Biology, Georgia Institute of Technology, Atlanta, Georgia 30332-0230 this bias (Boucher et al. 1982; Bertness and Callaway 1994), part of the reason for the modest attention given Submitted February 5, 1999; Accepted February 9, 2000 to the role of mutualism and other positive associations may be that the nature and strength of these interactions can be highly variable. Mutualisms that moderate local biotic or abiotic stresses may become commensal, and abstract: In North Carolina, the decorator crab Libinia dubia cam- commensalisms may become antagonistic when those ou¯ages almost exclusively with the chemically noxious alga Dictyota stresses are relaxed (e.g., Palumbi 1985; Hay 1986; Bron- menstrualis. By placing this alga on its carapace, the crab behaviorally stein 1994; Bertness and Leonard 1997; Holzapfel and sequesters the defensive chemicals of the plant and gains protection Mahall 1999; Stachowicz and Hay 1999a). Although from omnivorous consumers. However, Dictyota is absent north of many biologists recognize that biotic interactions vary in North Carolina, whereas Libinia occurs as far north as New England. outcome with local conditions, few studies examine how Crabs from three northern locations where Dictyota is absent (Rhode these changes may alter species' behaviors and evolution Island, Connecticut, and New Jersey) camou¯aged to match their environment, rather than selectively accumulating any one species. over geographic scales. Such studies of geographic var- When D. menstrualis was offered to crabs from northern sites, they iation may serve as a ªlensº through which many im- did not distinguish between it and other seaweeds for camou¯age, portant ecological and evolutionary questions can be whereas crabs from Alabama and two locations in North Carolina pro®tably examined (Travis 1996). used D. menstrualis almost exclusively. In addition, in winter and As an example, geographic variation in the outcome spring, when Dictyota was seasonally absent in North Carolina, Li- of interspeci®c interactions is thought to be important binia selectively camou¯aged with the sun sponge Hymeniacidon he- to the evolution of specialization (Thompson 1994). liophila, which was chemically unpalatable to local ®shes. Thus, Studies of plant-insect interactions have demonstrated southern crabs were consistent specialists on chemically defended species for camou¯age, while northern crabs were more generalized. that conspeci®c populations differ in the extent to which The geographic shift in crab behavior away from specialization co- they specialize; some populations become highly spe- incides with a reported decrease in both total predation pressure and cialized for the interactions as others remain or become the frequency of omnivorous consumers. These shifts in the nature less specialized (e.g., Janzen 1973; Rickson 1977; Whi- and intensity of predation pressure may favor different camou¯age tham 1983; Abrahamson et al. 1989; Steiner and White- strategies (generalist vs. specialist), contributing to the observed geo- head 1990; Tauber et al. 1995). In particular, populations graphic differences in camou¯age behavior. that fall outside the geographic range of the other species may exhibit few of the morphological or behavioral traits Keywords: decorator crabs, geographic comparison, intraspeci®c var- that enhance the association (Janzen 1973; Rickson iation, local specialization, plant-herbivore interactions, seaweeds. 1977). While it is easy to envision how this scenario may apply for associations involving poorly dispersed organ- isms or physically separated populations, local adaptation * Present address: Section of Evolution and Ecology, University of California, is predicted to be far less common in marine systems Davis, California 95616; e-mail: [email protected]. because they are generally more open and because larval Am. Nat. 2000. Vol. 156, pp. 59±71. q 2000 by The University of Chicago. and adult stages of many species disperse widely (Ja- 0003-0147/2000/15601-0005$03.00. All rights reserved. blonski 1986; Scheltema 1986; Palumbi 1992). Although 60 The American Naturalist geographic differences in allozyme frequencies or mito- Methods chondrial DNA sequences within species have been dem- onstrated in a variety of marine invertebrates with pelagic Organisms and Study Sites larvae (e.g., Avise 1992), geographic variation in behav- ioral traits relevant to the outcome of interspeci®c in- The decorator crab Libinia dubia (Decapoda, Majidae) oc- teractions is poorly known. curs from the Gulf of Mexico to southern New England Specialized positive associations are known among in sounds and saltier estuaries (Williams 1984). As with many marine invertebrates with pelagic dispersal phases several other species of majid crabs (e.g., Kilar and Lou 1986), juveniles of the species ªdecorateº their carapace (Steneck 1982; Glynn 1987; Littler et al. 1995; Stachowicz with materials from the surrounding environment, pre- and Hay 1999a, 1999c). These specialized positive inter- sumably as a defense against predators. Juveniles in North actions may be critical to the maintenance of marine Carolina selectively camou¯age with the chemically nox- communities through their positive effects on producers ious seaweed Dictyota menstrualis; crabs decorated with of biogenic structure such as corals or calci®ed seaweeds this alga experience less predation in the ®eld than crabs (Glynn 1987; Littler et al. 1995; Stachowicz and Hay 1996, decorated with a common seaweed that is not chemically 1999a). Could local specialization play a role in the origin defended from omnivorous consumers (Stachowicz and and maintenance of these commensal and mutualistic Hay 1999c). The crabs prefer D. menstrualis over all other associations, or does the high dispersal potential of these seaweeds, and this preference is cued by the diterpene marine species preclude such responses to local selection? alcohol dictyol E (Stachowicz and Hay 1999c), a metabolite Just as the study of geographic variation in host-plant speci®c to D. menstrualis that deters feeding by omnivo- preferences has increased our understanding of the basis rous ®shes (Hay et al. 1987, 1988; Cronin and Hay 1996), of the evolution of specialization and coevolution among which prey on both seaweeds and crustaceans. This alga insects and plants (e.g., Thompson 1994; Abrahamson occurs from southern Virginia south to the Gulf of Mexico and Weis 1997), a fuller understanding of variation in and throughout the Caribbean (Schneider and Searles specialized marine interactions could enhance our ability 1991). Thus, although the two species co-occur over a large to predict the circumstances under which these associ- portion of their range, northern populations of Libinia do not co-occur with Dictyota. ations arise and come to play an important role in marine We collected Libinia from six locations along the east communities. coast of the United States: Mobile Bay, Alabama (307139N, In this study, we examine spatial and temporal vari- 687029W); Mitchell Village, North Carolina (347449N, ation in the associational defense employed by the dec- 767509W); Drum Shoal, North Carolina (347439N, orator crab Libinia dubia along the east coast of the 767459W); Little Egg Harbor and Great Bay, New Jersey United States. In North Carolina, Libinia is known to (397309N, 747209W); Noank, Connecticut (417199N, enter into a specialized association with the chemically 727599W); and Narragansett, Rhode Island (417259N, noxious seaweed Dictyota menstrualis (Stachowicz and 717279W). The Alabama site, at the inlet to Mobile Bay, Hay 1999c). In this association, the crab escapes pre- was deeper (5.0 m) and more turbid than any of the other dation by placing the alga on its carapace as camou¯age, sites, and it supported few sessile benthic organisms. The behaviorally sequestering the alga's defensive chemicals. North Carolina sites were shoals comprised of a series of However, the crab and alga do not have completely over- sea grass and algal beds, ranging in depth from 0.5 to 1.0 lapping ranges, and it is currently unknown whether m in Bogue Sound, North Carolina, near the University crabs outside the range of the preferred host specialize of North Carolina's Institute of Marine Sciences. The New on other defended algae, generalize in an attempt to Jersey sites, near the Rutgers University Marine Station in blend into the background, or cease decorating alto- Tuckerton, New Jersey, consisted of algal ¯ats that ranged in depth from 1.5 to 2.0 m. The Connecticut site was at gether. Our results demonstrate that although all pop- the former University of Connecticut Marine Lab in ulations of Libinia do camou¯age with seaweeds or sessile Noank and consisted primarily of invertebrate-encrusted invertebrates, the nature and speci®city of camou¯age pilings and rocks that ranged in depth from just subsurface preferences differ greatly between northern and southern