Skip to main content
Open AccessThe Neuroecology of Chemical Defense

Neuroecology, Chemical Defense, and the Keystone Species Concept

Received 9 May 2007; accepted 7 July 2007.


Neuroecology unifies principles from diverse disciplines, scaling from biophysical properties of nerve and muscle cells to community-wide impacts of trophic interactions. Here, these principles are used as a common fabric, woven from threads of chemosensory physiology, behavior, and population and community ecology. The “keystone species” concept, for example, is seminal in ecological theory. It defines a species whose impacts on communities are far greater than would be predicted from its relative abundance and biomass. Similarly, neurotoxins could function in keystone roles. They are rare within natural habitats but exert strong effects on species interactions at multiple trophic levels. Effects of two guanidine alkaloids, tetrodotoxin (TTX) and saxitoxin (STX), coalesce neurobiological and ecological perspectives. These molecules compose some of the most potent natural poisons ever described, and they are introduced into communities by one, or only a few, host species. Functioning as voltage-gated sodium channel blockers for nerve and muscle cells, TTX and STX serve in chemical defense. When borrowed by resistant consumer species, however, they are used either in chemical defense against higher order predators or for chemical communication as chemosensory excitants. Cascading effects of the compounds profoundly impact community-wide attributes, including species compositions and rates of material exchange. Thus, a diverse array of physiological traits, expressed differentially across many species, renders TTX and STX fully functional as keystone molecules, with vast ecological consequences at multiple trophic levels.


A wide range of critical ecological interactions are mediated by chemistry. Biological responses to environmental chemical stimuli abound. Sensory perception of chemical signals, for example, strongly influences predation (Nevitt et al., 1995; Zimmer-Faust et al., 1995; Baldwin et al., 2006), courtship and mating (Dussourd et al., 1991; Painter et al., 1998; Bray and Amrein, 2003; Johnston, 2003; Riffell et al., 2004), aggregation and school formation (Blackburn et al., 1998; Keeling et al., 2003), and habitat colonization (Zimmer-Faust and Tamburri, 1994; Swanson et al., 2004; Lecchini et al., 2005; Dreanno et al., 2006). Additionally, chemical defenses produced by prey organisms (animals, plant, and microbes) render their tissues unpalatable or toxic to consumers (Whittaker and Feeny, 1971; Janzen, 1977; Lindquist and Hay, 1996; Nagle and Paul, 1998; Cruz-Rivera and Villareal, 2006). Despite the crucial ecological importance of such molecules, underlying mechanisms are lacking for most processes that structure communities. Natural products chemistry and chemical ecology have emphasized studies of secondary metabolites acting as toxins and antifeedants (Hay and Fenical, 1988; Pawlik, 1993; Daly, 1995; McClintock and Baker, 1997; Eisner et al., 2000). Yet there are numerous outstanding examples of secondary metabolites serving multiple roles and regulating the behavioral or physiological responses of individuals at lower trophic levels (Weller et al., 1999; Arnold and Targett, 2002; Bernays et al., 2002a, b; Camacho and Thacker, 2006; Steinke et al., 2006). Transferred to consumers at higher trophic levels, these effects have profound consequences for the distributions and abundances of organisms.

Unifying principles in ecology can provide conceptual frameworks for chemical defense and signaling. Optimal defense theory, Jensen's inequality, and the growth-differentiation balance hypothesis are benchmark intellectual achievements in chemical ecology, with roots embedded in a larger ecological context (McKey, 1974; Rhodes and Cates, 1976; Herms and Mattson, 1992; Ruel and Ayers, 1999). Similarly, the “keystone” species concept is intrinsically valuable for directing research and integrating studies on chemical defenses, chemosensory systems, behavior, and population and community dynamics.

A keystone species is one whose impacts on a community are greater than would be predicted from its relative abundance or total biomass (Paine, 1966; Estes and Duggins, 1995; Power et al., 1996; Menge et al., 2004). Sea otters (Enhydra lutris), for example, are carnivores in giant kelp forests of the northeast Pacific Ocean. Although their population densities are low, their predation on herbivores drastically reduces grazing pressure on kelp, leading to a highly diverse community of plants and associated invertebrate and fish species (Estes and Duggins, 1995; Steinberg et al., 1995). Removal of otters from kelp forest habitats, in contrast, facilitates herbivore survivorship and reproduction, accelerating grazing pressure and resulting in barren grounds as an alternative, stable community state (Estes and Palmisano, 1974; Duggins et al., 1989). Red-banded newts (Notophthalmus viridescens) similarly have dramatic effects in streams of the southeastern United States. Foraging by naturally small populations of these salamanders impacts composite community attributes, including species diversity and total prey biomass (Kurzava and Morin, 1998; Davic and Welsh, 2004; Smith, 2006). Moreover, newt predation significantly affects both vertebrate and invertebrate communities in the water column and on the stream bed. Grizzly bears (Ursus arctos) and beavers (Castor canadensis) are also keystone species, although the ways in which they modify community dynamics differ from those of otters and newts: bears alter rates of material (primarily nitrogen) exchange between aquatic and terrestrial environments, and beavers restructure physical habitats (Jones et al., 1994, 1997; Hilderbrand et al., 1999; Helfield and Naiman, 2006).

Neurotoxins are rare within natural communities, but they exert profound effects on species interactions at multiple trophic levels and thus could function in keystone roles (Williams et al., 2004; Ferrer and Zimmer, 2007a, b). The guanidine alkaloids, in particular, compose some of the most potent natural poisons ever described, with devastating effects as feeding deterrents at extremely dilute concentrations. Toxin-mediated processes that inhibit the generation of action potentials in nerve and muscle tissues are well understood at molecular and cellular levels (Noda et al., 1989; Terlau et al., 1991; Satin et al., 1992; Cestéle and Catterall, 2000). These toxins also affect trophic interactions, biomagnification, and evolved resistances in higher order consumer species (Brodie and Brodie, 1990; Llewellyn, 1997; Kvitek and Bretz, 2004; Williams et al., 2004; Bricelj et al., 2005; Llewellyn et al., 2006). Few investigations have connected the dots, however, for vertically integrating neural effects and ecological consequences at individual, population, and community levels. Such is the purpose of this synthesis on neuroecology, chemical defense, and the keystone species concept.

Here, the guanidine alkaloids tetrodotoxin (TTX) and saxitoxin (STX) are used to illustrate recent research findings that coalesce neurobiological and ecological paradigms. We know more about the integrative biology of these toxins than about any other poison. Each of our examples begins with a brief explanation of the cellular principles underlying toxicity. Next, we consider the ecological consequences of each toxin and focus specifically on their realized, as well as hypothesized, roles in mediating trophic interactions and community dynamics. Such interactions are best understood, and therefore emphasized, for organisms living in riparian stream and coastal ocean environments.

Historically, TTX and STX were known as inhibitors of nerve and muscle function (Kao, 1966; Evans, 1972; Hille, 1975); contemporarily, we also know them as mediators of both chemical defense and chemical communication (Kvitek et al., 1991; Kvitek and Beitler, 1991; Daly, 1995; Matsumura, 1995; Hwang et al., 2004; Camacho and Thacker, 2006; Zimmer et al., 2006). The chemoreceptors of some species apparently evolved for toxin interactions that stimulate, rather than suppress, conduction of nerve impulses (Yamamori et al., 1988; Zimmer et al., 2006). This varied array of physiological effects, expressed differentially across many species, has community-wide impacts, ultimately rendering TTX and STX as keystone molecules of substantial ecological importance.

The Neuroecology of Chemical Defense in Riparian Communities: Tetrodotoxin and Arginine

Cellular basis for tetrodotoxin toxicity and chemical defense

Tetrodotoxin (TTX) is a heterocyclic guanidine alkaloid harbored naturally by microbes and metazoans of phylogenetically diverse origins (Fig. 1a; Kim et al., 1975; Sheumack et al., 1978; Miyazawa et al., 1986; Thuesen et al., 1988). This neurotoxin acts as a chemical defense by inhibiting the transmission of signals between electrically excitable cells in consumers, and thus inducing paralysis and respiratory failure (Li, 1963; Kao and Fuhrman, 1963; Brodie, 1968). The molecule binds specifically and with high affinity to voltage-gated sodium channels on membranes of nerve and muscle cells (Hille, 1975; Kao, 1986; Catterall, 1992). In normal, unaffected cells, action potentials are generated and propagated by the influx of extracellular sodium ions through voltage-gated sodium channels, leading to conduction of an electrical impulse (Hille, 1984; Catterall, 1988). Bound TTX blocks the transport of sodium ions through the channel, effectively eliminating depolarization and impulse transmission (Hille, 1975; Kao, 1986).

Figure 1.
Figure 1.

Pathways hypothesized for biosynthesis of (A) tetrodotoxin and (B) saxitoxin, from an arginine precursor (Kotaki and Shimizu, 1993; Plumley, 1997).

Mechanisms of TTX binding and the characterization of its receptor site have particular relevance to the structure and function of voltage-gated sodium channels (Fig. 2A). The major subunit of a sodium channel, the α-subunit, is composed of four homologous domains (DI–DIV), each having six transmembrane segments (S1–S6). The TTX binding site—toxin receptor site 1—is located on the extracellular side of the sodium channel protein at a pore loop (SS2) that connects segments S5 and S6 on each of the four domains (Terlau et al., 1991). Studies using site-directed mutagenesis of sodium channels from brain, skeletal muscle, and cardiac muscle have identified amino acid residues on the SS2 loop critical in TTX binding (Noda et al., 1989; Terlau et al., 1991; Backx et al., 1992; Satin et al., 1992; Penzotti et al., 1998). Positions 385 of domain I in brain sodium channels and 401 of domain I in skeletal muscle sodium channels consist of phenylalanine and tyrosine, respectively. These aromatic amino acids confer high TTX binding affinity, and thus sensitivity to the toxin, in brain and skeletal muscle tissues. Cardiac muscle, however, possesses a substituted cysteine at the analogous residue 374 and consequently reduces TTX sensitivity 100- to 1000-fold (Chen et al., 1992; Satin et al., 1992; Lipkind and Fozzard, 1994). Additionally, toxin binding is eliminated completely by neutralization of glutamic or aspartic acid residues thought to interact electrostatically with the guanidinium moiety of TTX (Noda et al., 1989; Terlau et al., 1991).

Figure 2.
Figure 2.

Binding sites on (A) the SS2 loop (domains I–IV) of a voltage-gated sodium channel, (B) an olfactory receptor, and (C) a gustatory receptor. The selectivity hypothesized for each receptor type is indicated: filled circle, tetrodotoxin ( TTX) and saxitoxin (STX) bind at the same site; open circle, TTX and arginine bind at the same site; hatched circle, TTX and STX bind at related, but different sites (see Penzotti et al., 1998; Cestelé and Catterall, 2000; Yamamori et al., 1988; Pin et al., 2003; Zimmer et al., 2006). (D) Structures of TTX, STX, and arginine. Guanidinium moieties, highlighted with broken circles, are largely responsible for receptor-binding interactions.

More specifically, the pore binding site is composed of two paired sets of important residues, each pair residing on the SS2 loops of domains I through IV. The four respective SS2 loops collectively form a binding pore, consisting of important inner and outer residue rings. The outer ring is composed of glutamic acid, glutamic acid, methionine, and aspartic acid from domains I–IV, respectively. In contrast, aspartic acid, glutamic acid, lysine, and alanine make up the inner ring of the binding pore and function as the DEKA selectivity filter of the channel. TTX binding at this inner ring, or mutation of any of these residues, eliminates ionic conductance (Terlau et al, 1991; Heinemann et al., 1992), suggesting that TTX physically occludes the channel and thus blocks ion influx. It does so by interacting with the receptor site through guanidinium and C9, C10, and C11 hydroxyl groups (Lipkind and Fozzard, 1994; Penzotti et al., 1998; Choudhary et al., 2003).

Ecology of tetrodotoxin chemical defense and resistance

When disturbed by predators, many amphibians secrete a variety of toxins, including TTX, from glands along their dorsum (for reviews, see Daly, 1995; Toledo and Jared, 1995). These compounds, and the animals that produce them, are critical components of freshwater and riparian communities. Newts of the genus Taricha are especially meaningful candidates for studies on neuroecology. They reside in freshwater ponds and streams along the west coast of North America, and they use TTX as a potent chemical defense during many stages in their life cycle (Buchwald et al., 1964; Hanifin et al., 2002, 2003).

When ingested, sufficiently high concentrations of TTX inhibit neuromuscular function, resulting in paralysis and death. These effects have been observed in a wide array of vertebrate carnivores (Brodie, 1968; McAllister et al., 1997; Mobley and Stidham, 2000). Although Taricha newts are chemically well defended by TTX, sympatric garter snakes of the genus Thamnophis have evolved a resistance to the compound (Brodie and Brodie, 1990, 1999; Brodie et al., 2005). The most tolerant snakes can eat many toxic newts in a short period of time (Brodie, 1968). Consequently, predation by garter snakes may significantly impact Taricha populations, particularly where habitat loss and other anthropogenic effects have already reduced newt densities (Jennings and Hayes, 1994; Riley et al., 2005).

The molecular basis for the immunity of garter snakes to toxic newts reveals phenotypic adaptation (Geffeney et al., 2002, 2005). TTX resistance in snake skeletal muscle arises from mutations altering amino acid residues on receptor-binding sites for voltage-gated, sodium channel proteins (Fig. 3A). A single amino acid substitution, from isoleucine to valine, at residue 1561 confers low levels of resistance. Conversely, snakes with high resistance have undergone mutations altering amino acid residues at as many as four separate sites (Geffeney et al., 2005). In fact, snake resistance to TTX has evolved independently in different populations through small changes in sodium channel sequences.

Figure 3.
Figure 3.

(A) Amino acid sequences on domain IV of the sodium channels of skeletal muscle cells in garter snakes (Thamnophis sirtalis) with varying resistance to tetrodotoxin (TTX) (modified from Geffeney et al., 2005). (B) Amino acid sequences of domains I–IV of the sodium channels in skeletal muscle cells of puffer fish (Tetraodon nigroviridis) and the nerve cells of soft clams (Mya arenaria) with varying resistance to TTX or saxitoxin (STX), respectively (modified from Soong and Venkatesh, 2006; Bricelj et al., 2005). As yet, sodium channel genes for puffer fish sensitive to TTX or STX have not been sequenced. Residue substitutions conferring resistance are highlighted in grey, and analogous snake, fish, and clam domain IV residues are grouped in a box.

Ecological interactions at higher trophic levels also are affected by TTX. After garter snakes ingest poisonous newts, TTX accumulates in the tissues of the snakes for several weeks at concentrations within the lethal ranges of higher order predators (Williams et al., 2004). After a diet of newts, Thamnophis snakes—natural prey of many raptors—pose a threat to their predators. Furthermore, owls and waterfowl have been found dead with TTX-laden newts lodged in their esophageal tracts (Pimentel, 1952; McAllister et al., 1997; Mobley and Stidham, 2000). Mortality of these apex predators due to TTX consumption could generate cascading effects throughout freshwater and riparian communities.

Tetrodotoxin as a chemosensory cue of predation risk

Larval newts, unlike adults, juveniles, and embryos, are not chemically defended and are vulnerable to a number of vertebrate and invertebrate predators (Kats et al., 1992; Gamradt and Kats, 1996). Although TTX is absent from the skin of larvae, TTX released from adult newts is detected by larvae and stimulates predatory-avoidance and refuge-hiding behaviors (Table 1; Zimmer et al., 2006). Because aquatic adults exhibit intense cannibalism on larvae when the abundance of alternative prey is low, TTX is a reliable indicator of predation threat and alerts young newts to seek refuge. Once larvae visually detect a refuge, they move rapidly and on a linear trajectory to a hiding place. From the point of TTX contact, they swim directionally, upstream or downstream depending on the location of the refuge; thus the behavior is not simply an aversive reaction to a noxious chemical (Zimmer et al., 2006). Hence, TTX plays a dual role in freshwater habitats, serving both as a chemical defense and a predator-avoidance cue.

Table 1.

Effects of neurotoxins and arginine on behavior of adult and larval newts (Taricha torosa) (data are previously unpublished or taken from Zimmer et al., 2006; Ferrer and Zimmer, 2007a,b)

Life history stageCompound and concentration releasedNumber of animals
Hiding in refugeHaving muscle spasmsSwimming upstream to sourceNot responding
Larvae (laboratory)10−7 mol l−1 tetrodotoxin9001
10−7 mol l−1 saxitoxin1009
10−7 mol l−1 μ-conotoxin GIIIB0703
10−7 mol l−1 arginine00010
10−7 mol l−1 arginine + 10−7 mol l−1 tetrodotoxin00010
None (tap water control)00010
Adults (field)10−5 mol l−1 arginine*00114
None (filtered stream water)00213

*Trials substituting fluorescent dye for arginine indicate a mean concentration of 2.6 × 10−9 mol l−1 in contact with adults (see Ferrer and Zimmer, 2007a).

Electrophysiological assays in complement with behavioral experiments on newt larvae demonstrate the role of olfaction in mediating predator avoidance. Olfactory receptor cells of the larvae are excited by applications of TTX at 1 μmol l-1 (or lower), and refuge-hiding behavior is evoked by TTX at concentrations as dilute as 1 nmol l-1 (Fig. 4A; Zimmer et al., 2006). Whereas larvae perceive TTX emitted from cannibalistic adult newts as olfactory information, they exhibit muscle tremors and morbidity when exposed to unnaturally high concentrations of the toxin (at and above 10 μmol l-1). Such opposing physiological effects are dose-dependent, inhibiting or stimulating larval neuromotor activity at high or low concentrations, respectively. This disparity may arise because of differences in the TTX binding affinities for olfactory receptors (high) and voltage-gated sodium channels (low). In fact, protein sequences of sodium channels varying by even one amino acid residue express notably different TTX binding affinities among salamanders (Kaneko et al., 1997). Low binding affinity for sodium channel receptor sites, resulting in TTX resistance, allows animals to detect the alkaloid and respond behaviorally to dilute concentrations by using high-affinity, TTX-sensitive olfactory receptors. Therefore, in resistant organisms, TTX can convey chemosensory information that mediates trophic interactions.

Figure 4.
Figure 4.

Electrophysiological responses of olfactory or taste receptor cells to guanidine neurotoxins. (A) Reactions of newt (Taricha torosa) larval olfactory epithelium to artificial fresh water (AFW), forskolin (an adenyl cyclase activator), or tetrodotoxin (TTX) (data from Zimmer et al., 2006). Because forskolin does not bind to receptor proteins, it takes longer for this compound to diffuse across olfactory cell membranes and activate a cAMP transduction pathway. Alternatively, application of AFW control does not induce a change in electrical activity. (B) Reactions of rainbow trout (Salmo gairdneri) gustatory nerve to the free amino acid l-proline (a behavioral feeding stimulant) or increasing concentrations of saxitoxin (STX) (data from Yamamori et al., 1988). Arrows denote the time of initial stimulus application.

Global significance of tetrodotoxin as a combined signal/defense molecule

Like newts, other organisms have evolved mechanisms for recognizing TTX as a source of information (Table 2). Male puffer fish, for example, are attracted to TTX after its emission from gravid females (Matsumura, 1995). Moreover, resistant carnivorous snails exhibit feeding attraction to TTX-laden prey, presumably as a means of sequestering the poison from a dietary source (Hwang et al., 2004). There is, however, no physiological evidence of the specific sensory mechanisms underlying these behaviors.

Table 2.

Effects of tetrodotoxin (TTX) and saxitoxin (STX) on animal behavior

TTXMale puffer fish (Fugu niphobles)Released by gravid female puffer fish and attracts sexually mature conspecific malesMatsumura, 1995
Marine snails (Natica spp.)Present in the tissues of various prey species and stimulates feeding in predatory snailsHwang et al., 2004
California newt larva (Taricha torosa)Released by adult cannibals and causes larval prey to flee and hide in refugesZimmer et al., 2006
STXStaghorn sculpin (Leptocotus armatus)Present in the tissues of toxic butter clams and conditions feeding aversion in predatory fishKvitek, 1991
Sea otter (Enhydra lutris)Present in the tissues of toxic butter clams and alters feeding behavior and diet of predatory ottersKvitek et al., 1991
Oystercatcher (Haematopus bachmani)Present in the tissues of toxic mussels and alters feeding behavior, diet, and distributions of predatory birdsKvitek and Bretz, 2005
Amphipod (Hyalella azteca)Released by dinoflagellates and stimulates feeding activity in amphipod grazersCamacho and Thacker, 2006

The community-wide consequences observed for TTX in riparian habitats await discovery in other systems. Pervasive effects of the molecule seem inevitable, because TTX has a nearly cosmopolitan biogeographical distribution within marine, freshwater, and terrestrial organisms (Kim et al., 1975; Sheumack et al., 1978; Miyazawa et al., 1986; Thuesen et al., 1988; Matsumura, 1995; Kogure et al., 1996; Ritson-Williams et al., 2006). Moreover, it functions globally as a chemical defense for prey, a predator venom to subdue prey, a predator-avoidance cue, and a sex pheromone. Such remarkable flexibility arises from the contrasting impacts of different chemical concentrations, and points to TTX as a bioactive molecule of considerable ecological significance.

Arginine as a chemosensory cue of reduced predation risk and as a feeding attractant

The basic amino acid arginine, structurally similar to TTX, has very different but equally critical effects on trophic relationships (Table 1 and Fig. 2D). As feeding generalists, adult Taricha torosa dine on a taxonomically diverse prey assemblage, including primarily insects, worms, snails, and other small invertebrates (Stebbins, 1972; Hanson et al., 1994; R. P. Ferrer and R. K. Zimmer, unpubl. data). When adult newts feed on invertebrate prey, arginine is released at elevated concentrations into surrounding stream water (Ferrer and Zimmer, 2007a). Arginine concentrations in amphibian tissues and blood are at least 10–20 times lower than those in stream invertebrates (Gallardo et al., 1994; Emelyanova et al., 2004). Consequently, arginine is much more likely to signal the presence of injured invertebrates than of larval conspecifics.

California newt adults feed preferentially on worms over conspecific young, and there is no evidence for adult adaptations specifically for cannibalism (Elliott et al., 1993; Kerby and Kats, 1998). The cannibal-avoidance response in larvae is therefore suppressed when arginine from injured invertebrate prey mixes with TTX from adults (Table 1; Ferrer and Zimmer, 2007b). Experimental results suggest that the guanidinium moiety, present on both arginine and TTX, is likely to compete for common chemoreceptor-binding sites (Fig. 2B). Hence, the presence of arginine signals a reduced predation threat, and mixture suppression decreases fitness costs by inhibiting antipredator behavior. In contrast, arginine released from damaged prey stimulates foraging behavior in adult newts (Table 1; Ferrer and Zimmer, 2007a). Attractant chemical plumes function as a road map for the adults to find prey, thus facilitating further predation on invertebrate populations.

TTX and its congener, arginine, tightly link trophic levels ranging from deposit feeders and detritivores, such as worms and aquatic insects, to apex predators including hawks and owls (Fig. 5). The predator-prey interactions that shape both vertebrate and invertebrate communities are mediated by these compounds due to their opposing roles. Whereas TTX defends adult newts and resistant garter snakes by inhibiting neuromuscular function in predators, the toxin stimulates olfactory receptor cells in larval newts, eliciting avoidance behavior and reducing predator-driven mortality. Conversely, arginine suppresses larval antipredator responses to TTX, while activating food search and feeding behavior in adults. Although olfactory, gustatory, and vomeronasal organs function throughout a newt's lifetime, an ontogenetic shift in larval and adult chemosensory ability changes behavioral expression, hence reflecting the unique selection pressures that act at each life-history stage (Ferrer and Zimmer, 2007a,b).

Figure 5.
Figure 5.

Energy flow within a simplified trophic web, coupling riparian stream and mountain communities along the Pacific coast of North America. Hypothesized interactions between species are denoted by arrows, including those mediated directly by tetrodotoxin (black) or arginine (grey). Numbered organisms are (1) red tailed hawk (Buteo jamaicensis), (2) great horned owl (Bubo virginianus), (3) garter snake (Thamnophis spp.), (4) Western grebe (Aechmophorus occidentalis), (5) ground squirrel (Spermophilus beecheyi), (6) field mouse (Peromyscus spp.), (7) Pacific treefrog (Hyla regilla), (8) newt adult (Taricha spp.), (9) wild barley (Hordeum spontaneum), (10) newt larva (Taricha spp.), (11) earthworm (Eisenia rosea), (12) stonefly larva (Calineuria californica), and (13) mayfly larva (Hexagenia spp.).

The Neuroecology of Chemical Defense in Coastal Marine Communities: Saxitoxin

Cellular basis for saxitoxin toxicity and chemical defense

Communities within coastal marine habitats worldwide are shaped by phytoplankton chemical defenses. These molecules significantly modify material exchange rates among organisms in the water column, and between benthic and pelagic environments. The paralytic alkaloid saxitoxin (STX) is a potent neurotoxin synthesized by marine dinoflagellates (Fig. 1B; Schantz et al., 1966; Proctor et al., 1975; Bates et al., 1978; Shimizu, 1987) and certain freshwater cyanobacteria (Jackim and Gentile, 1968; Negri and Jones, 1995). During algal blooms, STX in surrounding waters can reach harmful concentrations, resulting in major die-offs of fish (White, 1980, 1981) and benthic invertebrates (Nagai et al., 2000; Yamatogi et al., 2005).

Like tetrodotoxin (TTX), STX is a heterocyclic guanidine alkaloid that binds with high affinity to sodium channel proteins and prevents the influx of sodium ions into excitable cells (Fig. 2A and D; Hille, 1975; Kao, 1986; Catterall, 1992; Lipkind and Fozzard, 1994). The two poisons are about the same size, with one or two positively charged guanidinium groups (Hille, 1975; Kao and Walker, 1982; Kao, 1986). They competitively inhibit each other in binding assays (Hansen-Bay and Strichartz, 1980; Sherman et al., 1983). Moreover, specific amino acid substitutions at toxin receptor site 1 of sodium channels produce similar effects on STX and TTX binding (Noda et al., 1989; Terlau et al., 1991; Kontis and Goldin, 1993). Thus, the two compounds are largely complementary in structure and function. Site-directed mutagenesis studies reveal that STX 7, 8, 9 and TTX 1, 2, 3 guanidinium groups each bind with an aspartic acid, glutamic acid, lysine, and alanine (DEKA) selectivity filter (at the inner region of the pore) of voltage-gated sodium channel proteins (Penzotti et al., 1998). Whereas TTX has a stronger interaction at amino acid residue 401 (tyrosine), STX interacts more effectively with the more extracellular residues. Unique moieties of each toxin are relegated to interactions with receptors at secondary binding sites on the SS2 loop. Ultimately, sodium ion influx into excitable cells is blocked principally by STX/TTX guanidinium interaction with the selectivity filter, despite subtle differences in toxin structures and sodium channel receptor associations.

Ecology of saxitoxin chemical defense and resistance

Dinoflagellate population growth can be tightly regulated by planktonic grazers and benthic, suspension-feeding invertebrates (Blasco, 1977; Turner and Anderson, 1983; Uye, 1986). Common consumers include ciliates (Stoecker et al., 1981), copepods and other zooplankton crustaceans (Fenchel, 1988), larval and adult fish (Last, 1980; Stoecker and Govoni, 1984; Gosselin et al., 1989; Robineau et al., 1991), and larval and adult macroinvertebrates (Robineau et al., 1991). In many cases, grazing pressure imposed by these organisms is significantly reduced when STX and related compounds are produced at high concentrations in dinoflagellates or cyanobacteria (Fiedler, 1982; Ives, 1985; Huntley et al., 1986; Haney et al., 1995; Marsden and Shumway, 1995; Smayda, 1997).

The physiological effects elicited by STX on its consumers vary among and within taxonomic groups. Vertebrates, such as fish, birds, and mammals, that take in STX exhibit loss of neuromuscular control, suppressed breathing, regurgitation, and abnormal behavior; in many cases, mortality results (White, 1981; Kvitek, 1991; Kvitek et al., 1991; Shumway et al., 2003). These effects are consistent with the selective inhibition of sodium ion influx associated with STX and TTX sodium channel binding (Li, 1963; Kao and Fuhrman, 1963). Grazing ciliates, however, undergo ciliary reversal, swelling, and cell lysis when exposed to STX (Hansen, 1989; Hansen et al., 1992). Furthermore, whereas some invertebrate grazers show decreases in consumption and swimming rates after ingesting STX (Ives, 1987; Sykes and Huntley, 1987; Haney et al., 1995), other closely related groups are unaffected (Teegarden and Cembella, 1996) or exhibit feeding stimulation by the toxin (Table 2). Changes in feeding rates can be attributed to chemosensory mechanisms (Huntley et al., 1986; Haney et al., 1995; Turriff et al., 1995; Teegarden and Cembella, 1996).

Many marine organisms are capable of accumulating STX in their tissues by consuming toxin-producing dinoflagellates (Twarog et al., 1972; Bricelj and Shumway, 1998; Turner et al., 2000; Llewellyn et al., 2006). The poison and its associated metabolites are present in a highly biodiverse assemblage of invertebrates, such as bivalves, gastropods, crustaceans, and echinoderms, in temperate (Jonas-Davies and Liston, 1985) and tropical waters (Llewellyn et al., 2006). Moreover, vertebrate predators, including several species of fish, also accumulate STX in their tissues, digestive tracts, and eggs (Nakamura et al., 1984; Llewellyn et al., 2006). The presence of these paralytic compounds in omnivores and carnivores indicates that they can be transferred among resistant consumers across multiple trophic levels and even to apex predators.

Although isolated and dinoflagellate-associated bacteria are capable of producing STX (Kodama et al., 1988, 1990; Gallacher et al., 1997; Baker et al., 2003), accumulation of the toxin in higher organisms occurs via dietary intake or exposure to surrounding toxin-laden waters during algal blooms (Schantz, 1986; Bricelj et al., 1990). This mechanism contrasts with that of most animals that sequester TTX through direct associations with bacterial symbionts (Noguchi et al., 1986; Ritchie et al., 2000). Select species with evolved resistance concentrate high doses of STX while maintaining normal nerve and muscle function (Kvitek and Beitler, 1991; Bricelj et al., 2005). Such resistance facilitates feeding on potentially poisonous prey, while borrowing the consumed toxin for their own chemical defense.

Immunity to STX and TTX in soft clams (Mya arenaria) and puffer fish (Tetraodon nigroviridis), respectively, offers a spectacular case of convergent evolution (Fig. 3B). For both species and toxins, a high degree of resistance is conferred by a single point mutation at the same amino acid residue (#758), located on the outer vestibule of sodium channel proteins in nerve and muscle cells (Bricelj et al., 2005; Soong and Venkatesh, 2006). An aspartic acid is substituted for a glutamic acid, causing a 1500- to 3000-fold decrease in STX/TTX binding affinity as compared to the ancestral form. Shared selection pressures and constraints acting on protein structure-function relationships promote a final, common pathway for the independent evolution of toxin resistance in distantly related species.

Interactions in the plankton

Dinoflagellates make up a significant percentage of phototrophic biomass in plankton communities and serve as important links between primary production and energy transfer throughout the trophic web (Lessard and Swift, 1985; Anderson and Sørenson, 1986; Fenchel, 1988; Mallin and Pearl, 1994). Periods of rapid cell growth facilitate increases in primary production and promote heightened energy exchange with consumers. Zooplankton grazing and nutrient limitations are key agents regulating dinoflagellate blooms (Blasco, 1977; Turner and Anderson, 1983; Uye, 1986; Berninger and Wickham, 2005) and phytoplankton composition. However, when STX and other paralytic poisons are released from cells, zooplankton feeding rates decay and monospecific blooms of chemically defended phytoplankters arise in response (Wyatt and Horwood, 1973; Fiedler, 1982; Huntley, 1982; Uye and Takamatsu, 1990).

Experimental evidence indicates that grazing pressure and the deterrent effects of STX are species-specific (Teegarden and Cembella, 1996). Whereas many zooplankton species undergo feeding suppression in harmful blooms, others are capable of consuming toxin-producing dinoflagellates. These grazers accumulate STX in their tissues and serve as vectors for toxin transfer throughout the planktonic food web (White, 1981; Boyer et al., 1985; Teegarden and Cembella, 1996; Turner et al., 2000). Consequently, STX plays a pivotal role in mediating multi-trophic interactions.

Consumption of STX-laden zooplankton or their incapacitated predators can have dramatic effects on top pelagic predators. Vertebrates such as fish (Adams et al., 1968; White, 1980, 1981), seabirds (Nisbet, 1983; Shumway et al., 2003), and marine mammals (Geraci et al., 1989; Reyero et al., 1999; Doucette et al., 2006) are much more sensitive to STX and its derivatives than are invertebrate grazers. Consequently, after dinoflagellate blooms, large-scale vertebrate mortality arises from ingestion of STX-laden planktonic organisms. Massive die-offs of top pelagic predators such as right whales (Doucette et al., 2006), monk seals (Reyero et al., 1999), and several species of fish (White, 1980, 1981) can lead to dramatic cascading effects throughout entire planktonic communities (Carpenter et al., 1985; Myers and Worm, 2003; Bruno and O'Connor, 2005).

Interactions linking plankton with benthos

A combination of STX toxicity and resistance results in profound effects at multiple trophic levels, coupling planktonic and benthic communities (Fig. 6). This phenomenon is illustrated in the Alaskan coastal ocean (Schantz et al., 1957, 1966; Quayle, 1969; Boyer et al., 1986; Kvitek and Beitler, 1991; Kvitek and Bretz, 2004). There, suspension-feeding butter clams (Saxidomus giganteus) are abundant in soft sediments, exhibit STX resistance, and sequester toxin in select tissues (Quayle, 1969; Beitler and Liston, 1990; Smolowitz and Doucette, 1995). STX and its derivatives are produced by dinoflagellates of the genus Alexandrium (previously classified as Gonyaulax or Protogonyaulax) during periodic algal blooms along the coast of southeastern Alaska (Horner et al., 1997; Van Dolah, 2000). Appropriate environmental conditions result in massive dinoflagellate blooms, and thus in elevated toxin concentrations (White, 1978; Boyer et al., 1987; Plumley, 1997; Smayda, 1997). In areas where harmful algal blooms occur seasonally (during late spring and early fall), S. giganteus retains STX at extremely high concentrations over the entire calendar year (Bricelj and Shumway, 1998). As a result, clam populations in these habitats are chemically well defended against higher order consumers.

Figure 6.
Figure 6.

Energy flow within a simplified trophic web, coupling benthic and pelagic communities in coastal habitats of the northeast Pacific Ocean. Hypothesized interactions between species are denoted by arrows, including those mediated directly by saxitoxin (black). Numbered organisms are (1) humpback whale (Megaptera novaeangilae), (2) killer whale (Orcinus orca), (3) staghorn sculpin (Lepticotus armatus), (4) oystercatcher (Haematopus bachmani), (5) sea otter (Enhydra lutris), (6) mackerel adult (Scomber scombrus), (7) mackerel larva (Scomber scombrus), (8) copepod (Calanus spp.), (9) krill (Euphausia spp.), (10) butter clam (Saxidomus giganteus), (11) whelk (Nucella spp.), (12) sea urchin (Strongylocentrotus spp.), (13) kelp crab (Pugettia producta), (14) mussel (Mytilus spp.), (15) abalone (Haliotis spp.), (16) toxic dinoflagellate (Alexandrium spp.), and (17) giant kelp (Macrocystis spp).

Sea otters are major carnivores of suspension-feeding and herbivorous prey animals in both soft-sediment and kelp-forest communities of the northeastern Pacific (Kvitek et al., 1992; Estes and Duggins, 1995; Steinberg et al., 1995). Effects are especially well described for kelp forests, where otter predation considerably reduces herbivore abundance, largely releasing plants from grazing pressure. In Alaskan waters, otter predation causes significant local decline in butter clam population density and thus promotes a community dominated by alternative suspension-feeding species (Kvitek and Oliver, 1992). Historically, otters were primarily distributed offshore in areas where blooms of STX-producing dinoflagellates were infrequent (Kvitek and Oliver, 1992). Nontoxic butter clams are present at these sites and make up a majority of otter diets (Kvitek and Oliver, 1992). Recently, however, otter populations have expanded into regions of harmful blooms, where butter clams accumulate high STX concentrations. Sea otters at these overlapping sites exhibit dramatic prey switching. They avoid toxic clams and feed on other, less common marine invertebrates (Kvitek and Bretz, 2004), leading ultimately to a shift in the species composition and structure of soft-sediment communities.

Saxitoxin as a chemosensory cue

Organisms can respond behaviorally to STX (Table 2). When STX is detected, invertebrates, including several species of zooplankton grazers, reduce feeding rates or reject dinoflagellates prior to consumption (Huntley et al., 1986; Haney et al., 1995; Teegarden and Cembella, 1996). Although rejection preempts ingestion, grazers typically discard the toxic cells after physical contact, suggesting that gustatory reception is responsible for STX perception. Similarly, vertebrate predators first sample potentially dangerous prey tissues orally before switching to less toxic individuals or organs. This behavior has been observed in animals such as bivalve-siphon-nipping fish (Kvitek, 1991), sea otters (Kvitek et al., 1991; Kvitek and Bretz, 2004), and shorebirds (Goss-Custard, 1996; Bustnes, 1998; Kvitek and Bretz, 2005). Sea otters, for example, show highly specific behavioral responses to butter clams of varying STX concentrations. Prior to feeding, otters break open the bivalves and test specific tissues for STX. They readily and completely consume clams of low toxicity but discard STX-laden tissues of moderate and high toxicity (Kvitek et al., 1991; Kvitek and Bretz, 2004).

Sampling behavior in vertebrates and zooplankton indicates that taste reception mediates a conditioned aversion to STX. This hypothesis is supported by electrophysiological experiments in which gustatory receptors of anadromous fish exhibit highly sensitive and specific responses to STX (Figs. 2C and 4B; Yamamori et al., 1988). Taste receptors are differentially tuned for STX or TTX, signifying a binding mechanism different from that of voltage-gated sodium channels (Hille, 1975; Kao, 1986). Olfactory-mediated responses in Taricha newt larvae also reveal discrimination by chemosensory cells between STX and TTX (Table 1; Zimmer et al., 2006). It thus appears that the nearly identical binding interactions of STX and TTX at sodium channels differ from the highly specific interactions at gustatory and olfactory receptors. Not all receptor-binding sites are created equal.

Recapitulation and Synthesis

The guanidine alkaloids tetrodotoxin (TTX) and saxitoxin (STX) play keystone roles in natural communities. Both have multiple, opposing physiological effects with strong, but contrasting, ecological consequences. The presence of STX in phytoplankton determines the habitat and prey choices of higher order consumers, significantly impacting species compositions of coastal ocean communities (Kvitek, 1991; Kvitek and Bretz, 2004, 2005). Large, episodic die-offs of predatory fish and mammals also modify primary plant-herbivore relationships, and thus regulate trophic cascades in both benthic and pelagic environments (Carpenter et al., 1985; Myers and Worm, 2003; Bruno and O'Connor, 2005). Similarly, TTX has profound effects on trophic interactions that connect riparian stream and coastal mountain communities. Used as a chemical defense by adult newts, this compound also triggers escape reactions in conspecific larval prey and protects resistant snake species from avian raptors (Williams et al., 2004; Zimmer et al., 2006). The behavioral responses of adult and larval newts are modified further by the free amino acid arginine, in association with alternative invertebrate prey species (Ferrer and Zimmer, 2007b). Studies have demonstrated that snakes, newts, and raptors have large impacts on riparian species assemblages (Marti et al., 1993; Kurzava and Morin, 1998; Jones et al., 2001; Davic and Welsh, 2004; Smith, 2006). Our own study, for example, revealed that predation by newts depresses invertebrate prey populations (R. P. Ferrer and R. K. Zimmer, unpubl. data). Other investigations have shown that selective foraging by adult newts reverses competitive hierarchies among prey species and significantly changes community composition (Kurzava and Morin, 1998). The toxins TTX and STX thus have classic keystone characteristics. Whereas trace concentrations are typically introduced by a single species, ultimately these molecules have large impacts on many species at many trophic levels.

Concluding Remarks and Future Directions

The guanidine alkaloids are only two of several molecules potentially playing keystone roles within natural communities. In open-ocean habitats at polar latitudes, for example, dimethylsulfoniopropionate (DMSP) and its metabolites (dimethyl sulfide and acrylate) convey information among several trophic levels, including large pelagic predators, in planktonic food webs (Nevitt et al., 1995; Wolfe et al., 1997; Steinke et al., 2006; Pohnert et al., 2007). The DMSP signaling/defense pathways are vital for maintaining material exchange between primary producers, grazers, and carnivores, and have critical roles in the microbial loop (Zimmer and Butman, 2000). Alternatively, the pyrrolizidine alkaloids act as chemical defenses, mate attractants, and gustatory stimuli among many plant and resistant consumer species in terrestrial systems (Dussourd et al., 1989; Eisner and Eisner, 1991; Schulz et al., 1993; Trigo et al., 1996; Weller et al., 1999; Eisner et al., 2000; Bernays et al., 2002a, b).

Greater understanding of neuroecological phenomena will be gained through increased interdisciplinary efforts to bridge the gaps between processes that affect individuals and higher order ecological interactions. Thus far, nearly all investigations have focused at only one level of biological organization and have provided limited synthesis. Consequently, a great deal is known about parts without a clear understanding of the whole within a unified neuroecological perspective. The guanidine alkaloids tetrodotoxin (TTX) and saxitoxin (STX) are notable exceptions.

Combined studies of cell physiology and autecology (individual organisms) are especially tractable for direct experimental analysis. Hence there is an impressive body of literature on effects of chemical defense and signaling compounds (e.g., Howe, 1976; Vickers et al., 2001, Bernays et al., 2003; Kicklighter et al., 2005). Considerably less is known, however, about the community-wide impacts of such molecules.

Deciphering the roles of chemical signal/defense molecules in mediating population and community-wide processes will prove challenging. Experimental treatments would benefit from genetic manipulations to eliminate the biosynthetic capacity of primary producers, herbivores, and higher order consumers for compounds hypothesized to be keystone molecules (Baldwin et al., 2006; Izaguirre et al., 2006). Such studies could also independently involve manipulations of genes that control toxin resistance, code for the binding properties of chemosensory receptors, or both. This experimental program is a tall order indeed, and may raise concerns about using genetically engineered organisms in the field.

Alternatively, if the interactions are sufficiently strong, removing a single species may reveal a keystone effect. The elimination, for example, of Plocamium red algae high in terpenes (as chemical defenses) resulted in habitat colonization by competitively subordinate soft coral species and caused significant change in community structure (de Nys et al., 1991). Even this removal, however, did not isolate chemical defense from other potential factors mediating community-wide impacts. Similarly, experimental microcosm and mesocosm experiments in the laboratory and field have proven amenable for assigning the outcomes of chemical interactions among selected species at two or three trophic levels (Dicke et al., 2003; McIntosh et al., 2003; Linhart et al., 2005).

Coupled with these experiments, investigations could exploit biogeographical gradients in distributions of signal/defense molecules. Such gradients will provide natural laboratories for discovering the impacts of signal/defense compounds on community organization. The population density of chemically defended phytoplankton, for example, varies from high to low along a transect from inshore to offshore in Alaskan ocean waters. This natural variation has been used effectively to establish the community-wide consequences of STX (Kvitek and Bretz, 2004). Similarly, genetically distinct populations of the rough-skinned newt are found among equivalent habitats in northern California, Oregon, Washington, and southern British Columbia. The populations differ considerably in TTX concentration, from none to highly toxic levels (Hanifin et al., 1999; Brodie et al., 2002). Through careful selection of populations and habitats, it may be possible to identify community-wide properties that vary uniquely as a function of TTX concentration.

By using these natural laboratories, it will be possible to establish broad community-wide patterns that motivate controlled experiments, in field or laboratory, on targeted individual or species interactions. Thus, integrating a wide repertoire of quantitative, natural historical, and experimental approaches would establish a composite neuroecological picture from its physiological, behavioral, and ecological parts. The resulting knowledge will define the effects of chemical signal/defense molecules on cellular processes and determine their consequences within natural communities. It also will determine the extent to which the keystone concept can be applied as a basic neuroecological principle. Finally, it will establish where impacts attributed to keystone species arise not from biological interactions, but as a consequence of chemistry.


We thank C.A. Zimmer and members of the UCLA graduate seminar in Chemical Ecology (A. E. Nichols, G. A. Ferrier, A. J. Corcoran, and S. B. Olssen) for their assistance in developing ideas linking chemical defenses, chemical signals, and the keystone species hypothesis. C. D. Derby was an early and gracious collaborator in hatching the neuroecology concept. His invitation for a seminar, as well as strong encouragement, led to organizing these thoughts into a collective body of work. Comments by J. B. McClintock and an anonymous reviewer greatly improved an earlier draft of the manuscript. This research was supported by awards from the National Science Foundation (OCE 02-42321) and California Sea Grant (Project # R/F-147).

Literature Cited