Chemosensory responses to foods by an herbivorous acrodont lizard, <Emphasis Type="Italic">Uromastyx aegyptius</Emphasis>

Actively foraging lizards use the lingual-vomeronasal system to identify prey by chemical cues, but insectivorous ambush foragers do not. The major clade Iguania includes numerous herbivores and omnivores; among them, two iguanid and one agamine species identify plant and animal foods by tongue flicking, and data suggest that the leiolepidine Uromastyx acanthinurus may as well. We conducted experiments on chemosensory response to food by the herbivorous U. aegyptius. When chemical stimuli were presented on cotton balls in experiment 1, the lizards exhibited greater responsiveness (tongue-flick attack scores) to chemical stimuli from crickets and a preferred plant food (dandelion flowers) than from deionized water. When chemical stimuli were on ceramic tiles in experiment 2, the lizards exhibited greater total tongue flicks to cricket stimuli than to any other stimuli, and to dandelion than to deionized water. Lizards bit more frequently in response to cricket and dandelion cues than to stimuli from a nonpreferred plant (carrot) and deionized water. Tongue-flick attack scores were greater in response to cricket and dandelion stimuli than to carrot or water stimuli. These findings are consistent with the hypothesis that herbivores, even those having ambush-foraging ancestors, use chemical cues to identify potential foods. The data support the hypothesis that chemosensory responses correspond to diet. Because most lizards are generalist predators, studies of herbivorous species can provide important information on possible evolutionary adjustment of chemosensory response to dietary shifts. Electronic Publication     

Generalization of predator recognition: Velvet geckos display anti-predator behaviours in response to chemicals from non-dangerous elapid snakes

Many prey species detect chemical cues from predators and modify their behaviours in ways that reduce their risk of predation. Theory predicts that prey should modify their anti-predator responses according to the degree of threat posed by the predator. That is, prey should show the strongest responses to chemicals of highly dangerous prey, but should ignore or respond weakly to chemicals from non-dangerous predators. However, if anti-predator behaviours are not costly, and predators are rarely encountered, prey may exhibit generalised antipredator behaviours to dangerous and non-dangerous predators. In Australia, most elapid snakes eat lizards, and are therefore potentially dangerous to lizard prey. Recently, we found that the nocturnal velvet gecko Oedura lesueurii responds to chemicals from dangerous and non-dangerous elapid snakes, suggesting that it displays gen-eralised anti-predator behaviours to chemicals from elapid snakes. To explore the generality of this result, we videotaped the be-haviour of velvet geckos in the presence of chemical cues from two small elapid snakes that rarely consume geckos: the nocturnal golden-crowned snake Cacophis squamulosus and the diurnal marsh snake Hemiaspis signata. We also videotaped geckos in tri-als involving unsceted cards (controls) and cologne-scented cards (pungency controls). In trials involving Cacophis and Hemi-aspis chemicals, 50% and 63% of geckos spent long time periods (> 3 min) freezing whilst pressed flat against the substrate, re-spectively. Over half the geckos tested exhibited anti-predator behaviours (tail waving, tail vibration, running) in response to Ca-cophis (67%) or Hemiaspis (63%) chemicals. These behaviours were not observed in control or pungency control trials. Our re-sults support the idea that the velvet gecko displays generalised anti-predator responses to chemical cues from elapid snakes. Generalised responses to predator chemicals may be common in prey species that co-occur with multiple, ecologically similar, dangerous predators.     

Chemoreceptive and behavioural responses of the common lizard Lacerta vivipara to snake chemical deposits

Behavioural cues were used to assay the capacity of common lizards to detect chemical deposits of snakes. The lizards were observed in cages that had been previously inhabited either by one of two species of snake that feed on lizards (the viper Vipera berus and the smooth snake Coronella austriaca), or the grass snake (Natrix natrix), which does not feed on lizards. As a control, the lizards were tested both in a clean cage and in one sprayed with a pungent odorant. The lizards responded to the snakes' chemicals by increased tongue-flick rates, with the highest rates being given in response to the deposits of their predators. The chemosensory examination of the snakes' odours induced a shift in general behaviour in response to the predator, but not to the non-predator chemical cues. This behavioural response consisted mainly of a disruption of the locomotor patterns. Our findings strongly suggest that lizards detected and distinguished between the chemicals deposited by three species of snake. Behavioural performances were highly variable among individual lizards in all trials, but the relative scores of individuals tended to be similar in response to different stimuli.     

Can Wall Lizards Combine Chemical and Visual Cues to Discriminate Predatory from Non‐Predatory Snakes Inside Refuges?

The ability to use multiple cues in assessing predation risk is especially important to prey animals exposed to multiple predators. Wall lizards, Podarcis muralis, respond to predatory attacks from birds in the open by hiding inside rock crevices, where they may encounter saurophagous ambush smooth snakes. Lizards should avoid refuges with these snakes, but in refuges lizards can also find non‐saurophagous viperine snakes, which lizards do not need to avoid. We investigated in the laboratory whether wall lizards used different predator cues to detect and discriminate between snake species within refuges. We simulated predatory attacks in the open to lizards, and compared their refuge use, and the variation in the responses after a repeated attack, between predator‐free refuges and refuges containing visual, chemical, or visual and chemical cues of saurophagous or non‐saurophagous snakes. Time to enter a refuge was not influenced by potential risk inside the refuge. In contrast, in a successive second attack, lizards sought cover faster and tended to increase time spent hidden in the refuge. This suggests a case of predator facilitation because persistent predators in the open may force lizards to hide faster and for longer in hazardous refuges. However, after hiding, lizards spent less time in refuges with both chemical and visual cues of snakes, or with chemical cues alone, than in predator‐free refuges or in refuges with snake visual cues alone, but there were no differences in response to the two snake species. Therefore, lizards could be overestimating predation risk inside refuges. We discuss which selection pressures might explain this lack of discrimination of predatory from similar non‐predatory snakes.     

Fixed prey cue preferences among Dusky Pigmy Rattlesnakes (<Emphasis Type="Italic">Sistrurus miliarius barbouri</Emphasis>) raised on different long-term diets

Chemoreception is often crucial to the interaction between predators and their prey. Investigating the mechanisms controlling predator chemical preference gives insight into how selection molds traits directly involved in ecological interactions between species. In snakes, prey cue preferences are influenced by both direct genetic control and experience-based plasticity. We assessed prey preference in a group of Dusky Pigmy Rattlesnakes that had eaten only mice or lizards over a 5 year period to test whether genetics or plasticity primarily determine the preference phenotype. Our results provide evidence for genetic determination of preference for lizard chemical cues in pigmy rattlesnakes. Snakes preferred the scent of lizards, regardless of their initial diet, and the response to mouse scent did not differ from the water-only control. We discuss these findings in light of previous studies that manipulated snake diets over shorter timescales.     

Tasty figs and tasteless flies: plant chemical discrimination but no prey chemical discrimination in the cordylid lizard<Emphasis Type="Italic"> Platysaurus broadleyi</Emphasis>

Lizards use visual and/or chemical cues to locate and identify food. The ability to discriminate prey chemical cues is affected by phylogeny, diet, and foraging mode. Augrabies flat lizards (Platysaurus broadleyi) are omnivorous members of the lizard clade Scleroglossa. Within Scleroglossa, all previously tested omnivores are capable of both prey and plant chemical discrimination. At Augrabies Falls National Park, P. broadleyi feed on both insects (black flies) and plant material (figs), and as scleroglossans, are predicted to discriminate both plant and prey chemicals. However, Platysaurus broadleyi use visual, not chemical cues, to detect and capture black flies, which occur in large concentrations in the study area. We tested free-ranging individuals for the ability to discriminate insect and plant chemicals from controls. There was a significant stimulus effect such that lizards tongue-flicked fig-labelled tiles significantly more than the remaining stimuli, spent more time at the fig-labelled tile, and attempted to eat fig-labelled tiles more often than tiles labelled with control or insect stimuli. Platysaurus broadleyi is exceptional in being the first lizard shown to possess plant chemical discrimination but to lack prey chemical discrimination. We suggest that an absence of prey chemical discrimination may be a consequence of foraging behaviour and environmental effects. Because insect prey are highly clumped, abundant, and aerial, profitable ambushing using visual cues may have relaxed any selective pressure favouring insect prey chemical discrimination. However, a more likely alternative is that responses to figs are gustatory, whereas as prey chemical discrimination and plant chemical discrimination are usually mediated by vomerolfaction.Communicated by P.K. McGregor     

Responses to chemical cues from animal and plant foods by actively foraging insectivorous and omnivorous scincine lizards

If tongue-flicking is important to lizards to sample chemical cues permitting identification of foods, tongue-flicking and subsequent feeding responses should be adjusted to match diet. This hypothesis can be examined for plant foods because most lizards are insectivores, but herbivory/omnivory has evolved independently in many lizard taxa. Here we present experimental data on chemosensory responses to chemical cues from animal prey and palatable plants by three species of the scincine lizards. When tested with chemical stimuli presented on cotton swabs, the insectivorous Eumeces fasciatus responded strongly to prey chemicals but not to chemicals from plants palatable to omnivorous lizards or to pungent or odorless control stimuli. Two omnivorous species, E. schneideri and Scincus mitranus, responded more strongly to chemical cues from both prey and food plants than to the control chemicals. All available data for actively foraging lizards, including these skinks, show that they are capable of prey chemical discrimination, and insectivores do not exhibit elevated tongue-flicking or biting responses to chemical cues from palatable plants. In all of the several species of herbivores/omnivores tested, the lizards show elevated responses to both animal and plant chemicals. We suggest two independent origins of both omnivory and plant chemical discrimination that may account for the evolution of diet and food chemical discriminations in the eight species of skinks studied, five of which are omnivores. All data are consistent with the hypothesis that acquisition of omnivory is accompanied by acquisition of plant chemical discrimination, but data on a broad diversity of taxa are needed for a definitive comparative test of the evolutionary hypothesis. J. Exp. Zool. 287:327-339, 2000.     

Balearic lizards use chemical cues from a complex deceptive mimicry to capture attracted pollinators

Deceptive flowers from several plant species emit odors that mimic oviposition cues and attract female insects seeking for a laying site. Helicodiceros muscivorus is a species that emits an odor mimicking the foul smell of rotting meat and thereby attracts blowflies that usually oviposit on carcasses but are deceived into pollinating the plant. Thus, H. muscivorus is a striking case of pollination by brood‐site deception. The Balearic lizard, Podarcis lilfordi, exhibits remarkable interactions with dead horse arum. Balearic lizards, which sometimes forage on carcasses, are attracted to blooming dead horse arum. We showed experimentally that P. lilfordi can detect chemical cues from carcasses on cotton swabs and exhibits elevated tongue‐flick rates to carcass chemical cues compared to control stimuli. Lizards also detected and located hidden carcasses using only airborne chemical cues. The responses of lizards to chemical cues from the spadix of blooming dead horse arum were qualitatively and quantitatively similar to those to carcass odors. Therefore, the decay‐like odor that attracts blowflies for the plant's benefit also attracts lizards. This attraction may initially have been somewhat favorable for lizards that eat blowflies, but slightly unfavorable for plants because the lizards ate some pollinators. We suggest that lizards attracted by odor may have learned later to use the plant for thermoregulation and then consume its fruits, making the association more positive for lizards and benefitted arum by seed dispersal.     

Wall lizards combine chemical and visual cues of ambush snake predators to avoid overestimating risk inside refuges

The threat sensitivity hypothesis assumes that multiple cues from a predator should contribute in an additive way to determine the degree of risk-sensitive behaviour. The ability to use multiple cues in assessing the current level of predation risk should be especially important to prey exposed to multiple predators. Wall lizards, Podarcis muralis, respond to predatory attacks from birds or mammals by hiding inside rock crevices, where they may encounter another predator, the smooth snake, Coronella austriaca. We investigated in the laboratory whether chemical cues may be important to wall lizards for detection of snakes. The greater tongue-flick rate and shorter latency to first tongue-flick in response to predator scents indicated that lizards were able to detect the snakes' chemical cues. We also investigated the use of different predatory cues by lizards when detecting the presence of snakes within refuges. We simulated successive predator attacks and compared the propensity of lizards to enter the refuge and time spent within it for predator-free refuges, refuges containing either only visual or chemical cues of a snake, or a combination of these. The antipredatory response of lizards was greater when they were exposed to both visual and chemical cues than when only one cue was presented, supporting the threat sensitivity hypothesis. This ability may improve the accuracy of assessments of the current level of predation risk inside the refuge. It could be especially important in allowing lizards to cope with threats posed by two types of predators requiring conflicting prey defences.     

Apparent decoupling of prey recognition ability with prey availability in an insular snake population

Numerous studies on the feeding behavior of snakes have reported the consistency of tongue-flick responses with their natural diets. For representatives of widely distributed, dietary generalist species from particular localities, we can expect that their tongue-flick responses to potential prey unavailable in their original habitats have been reduced whereas those to prey common in the habitats have been enhanced. To test this hypothesis, intraspecific variation in tongue-flick responses to prey chemicals was examined using ingestively naive snakes (Elaphe quadrivirgata) from dietarily different populations: populations from the main Japanese island, where the snakes' diet predominantly consists of sympatric frogs, and from Mikura-jima Island, where no frogs occur and the snakes thus chiefly prey on lizards. We presented chemical stimuli from six items including those from their natural and potential prey (fish, frog, lizard, mouse, water, and cologne) to newborn snakes. Significant effects of stimuli on the tongue-flick responses were detected. On the other hand, effects of population and interaction between stimuli and population were not significant, and individual comparisons revealed no significant interlocality differences in responses to either frog or lizard chemicals. Thus, our hypothesis was not supported. However, in the Mikura-jima sample, significantly fewer snakes responded to frog chemicals than in the main island sample. The significance of the inconsistency between prey recognition ability and prey availability in the Mikura-jima population are discussed. Received: October 17, 2000 / Accepted: December 14, 2000     

Feeding Experience Modifies the Assessment of Ambush Sites by the Timber Rattlesnake, a Sit-and-Wait Predator

To effectively ambush prey, sit‐and‐wait predators must locate sites where profitable prey are likely to return. One means by which predators evaluate potential ambush sites is by recognizing high‐use areas through chemical cues deposited inadvertently by their prey. However, it is unknown whether ambush predators can use chemical cues associated with past prey items in the assessment of potential ambush sites. I examined selection of ambush sites by timber rattlesnakes (Crotalus horridus) exposed to trails made from chemical extracts of the integument of various prey species. I evaluated the role of feeding experience in ambush site selection by comparing the behavior of timber rattlesnakes before and after feeding experience with different sized prey items. Timber rattlesnakes are more likely to select ambush sites adjacent to chemical trails from prey with which they have had feeding experience, but only those fed relatively large prey showed an increase in responsiveness. Increased responsiveness after feeding experience was exhibited in experiments using integumentary extracts of mammals (the natural prey of timber rattlesnakes), but not in those using extracts of fish. These results indicate that ambush predators may learn to recognize chemicals on the integument of profitable food items, and use that experience when subsequently selecting ambush sites. Additionally, these findings provide evidence that size‐dependent predation by snakes may be, in some species, a result of active prey selection.     

Food chemical discriminations by an herbivorous lizard, Corucia zebrata

Adjustment of chemosensory response to diet should be apparent in evolutionary changes corresponding to dietary shifts. Because most lizards are generalist predators of small animals, relationships between chemosensory behavior and diet are difficult to detect. Nevertheless, the evolution of herbivory by a small number of lizards provides an opportunity to detect any corresponding change in response to plant chemicals. I studied tongue-flicking and biting by the large, herbivorous scincid lizard Corucia zebrata in response to chemical cues from crickets, romaine lettuce, and control stimuli presented on cotton swabs. The skinks exhibited significantly stronger response to plant and animal chemicals than to controls for several variables: greater number of individuals that bit swabs, shorter latency to bite, greater rate of tongue-flicks, and greater tongue-flick attack score. The clearest differences were observed for tongue-flick attack score, a composite variable that combines the effects of tongue-flicking and biting. An insectivorous member of the same subfamily, Scincella lateralis, shows strong tongue-flicking and biting response to chemical prey cues, but not to plant chemicals. This suggests that response to plant chemicals by C. zebrata may have evolved in tandem with the incorporation of plants into the diet and that response to cricket chemicals has been retained, perhaps due to similarities between plant and animal food. The findings support the hypothesis that dietary shifts induce corresponding changes in chemosensory response, but provide only a single independent contrast for a study of correlated evolution between plant diet and chemosensory response to plants. J. Exp. Zool. 286:372-378, 2000.     

Antipredator behaviour of invasive geckos in response to chemical cues from snakes

Antipredator behaviours and the ability to appropriately assess predation risk contribute to increased fitness. Predator avoidance can be costly; however, so we expect prey to most strongly avoid predators that pose the greatest risk (i.e., prey should show threat sensitivity). For invasive species, effectively assessing the relative risk posed by predators in the new environment may help them establish in new environments. We examined the antipredator behaviour of introduced Asian house geckos, Hemidactylus frenatus (Schlegel), by determining if they avoided shelters scented with the chemical cues of native predatory snakes (spotted pythons, Antaresia maculosa [Peters]; brown tree snakes, Boiga irregularis [Merrem]; common tree snakes, Dendrelaphis punctulata [Grey]; and carpet pythons, Morelia spilota [Lacépède]). We also tested if Asian house geckos collected from vegetation vs. anthropogenic substrates (buildings) responded differently to the chemical cues of predatory snakes. Asian house geckos did not show a generalised antipredator response, that is, they did not respond to the chemical cues of all snakes in the same way. Asian house geckos avoided the chemical cues of carpet pythons more strongly than those of other snake species, providing some support for the threat‐sensitivity hypothesis. There was no difference in the antipredator behaviour of Asian house geckos collected from buildings vs. natural vegetation, suggesting that individuals that have invaded natural habitats have not changed their antipredator behaviour compared to urban individuals. Overall, we found some evidence indicating Asian house geckos are threat‐sensitive to some Australian predacious snakes.     

The Asian grass lizard (Takydromus sexlineatus) does not respond to the scent of a native mammalian predator

Lacertid lizards use chemical cues emitted by saurophagous snakes to evade predation. Whether these lizards can detect and respond to the chemical cues of predatory mammals has not been studied. As many mammals carry distinct body odours and/or use chemical cues for intraspecific communication, lizards can be expected to use these chemicals as early warning cues. To test this idea, we observed the behaviour of Asian grass lizards (Takydromus sexlineatus) that had been transferred to an unfamiliar test arena containing one of four scent treatments. No particular scent was applied to the arena in the control situation. Diluted aftershave served as a pungency control. In the snake treatment, scent of the Oriental whip snake (Ahaetulla prasina) was applied. We included this treatment to learn how Asian grass lizards react to predator chemical cues. Finally, in the mongoose treatment, the lizards were confronted with scent cues of several small Indian mongooses (Herpestes auropunctatus). Snake scent elicited foot shakes, startles and tail vibrations. These are behaviours that in lacertid lizards are associated with stressful situations such as predatory encounters. Surprisingly, lizards confronted with mongoose scent exhibited none of these stress-indicating behaviours. In fact, their behaviour did not differ from that of lizards subjected to an odourless control treatment. These results raise concern. Mongooses are rapidly invading ecosystems worldwide. If lizards that have co-evolved with mongooses are unable to detect these predators’ presence through chemical cues, it seems highly unlikely that evolutionary naïve lizards will develop this ability rapidly.     

Chemosensory Response of Desert Iguanas (Dipsosaurus dorsalis) to Skin Lipids from A Lizard-Eating Snake (Lampropeltis getula californiae)

Several species of lizards respond to chemicals from sympatric lizard‐eating snakes with increased tongue‐extrusion rates. These substances also elicit antipredator behavior indicating that they have important ecological functions and the resulting behavior can have serious implications for individual fitness of lizards. However, the source and type of snake chemical cues that elicit these behavioral changes in lizards have yet to be determined. We tested the ability of adult desert iguanas (Dipsosaurus dorsalis) to detect and identify a potential predator by exposing them to lipids extracted from shed snakeskins. Lipids were extracted from cast skins of a known lizard‐eating snake, the California kingsnake (Lampropeltis getula californiae), using chloroform and methanol. Test subjects were presented with skin lipids as well as clean, pungent, and chloroform controls on cotton‐tipped applicators in random order. Desert iguanas directed significantly more tongue extrusions toward applicators bearing snakeskin lipids when compared with controls. In addition, overall tongue‐extrusion frequency increased following exposure to lipids during the 5‐min trials. Desert iguanas clearly detected snakeskin lipids, but this stimulus failed to elicit changes in body posture and movement patterns previously observed in experiments using chemical cues from live snakes. Increased tongue flicking by lizards in response to snakeskin lipids may represent a generalized response to this class of chemicals. Additional potential sources of chemicals used in the detection of lizard‐eating snakes are discussed.     

Juvenile western toads, Bufo boreas, avoid chemical cues of snakes fed juvenile, but not larval, conspecifics

Previous investigations have demonstrated the importance of predator diet in chemically mediated antipredator behaviour. However, there are few data on responses to life-stage-specific predator diets, which could be important for animals like amphibians that undergo metamorphosis and must respond to different suites of predators at different life-history stages. In laboratory choice tests, we investigated the chemically mediated avoidance response of juvenile western toads, Bufo boreas, to four different chemical stimuli: (1) live conspecific juveniles; (2) live earthworms; (3) snakes fed juvenile conspecifics; and (4) snakes fed larval conspecifics (tadpoles). Juvenile toads avoided chemical cues from snakes that had eaten juvenile conspecifics, but did not respond to the other three stimuli, including chemical cues from snakes fed larval conspecifics. In addition, the response to cues from snakes fed juveniles differed significantly from that of snakes fed larvae. To our knowledge, this is the first study to demonstrate the importance of diet in predator avoidance of juvenile anurans and the ability of juvenile toads to distinguish between chemical cues from predators that have consumed larval versus juvenile conspecifics. Copyright 2000 The Association for the Study of Animal Behaviour.     

Food chemical cues elicit general and population-specific effects on lingual and biting behaviors in the lacertid lizard Podarcis lilfordi

Actively foraging lizards are capable of identifying prey using only chemical cues sampled by tongue-flicking, and the relatively few omnivorous and herbivorous lizards tested similarly can identify both animal and plant foods from chemical cues. Whether lizards that eat plants respond to cues specific to preferred plant types and whether there is geographic variability in responses to cues from various plants correlated with the importance of those plants in local diets is unknown. In three populations of an omnivorous lacertid, the Balearic lizard Podarcis lilfordi, we studied chemosensory sampling and feeding responses to chemical cues from plant and animal foods presented on cotton swabs. Each lizard population is endemic to one islet off the coast of Menorca, Balearic Islands, Spain. Lizards in all three populations discriminated chemical cues from plant and animal foods from control substances. Our results extend findings of prey chemical discrimination and plant chemical discrimination in omnivores, increasing confidence that correlated evolution has occurred between plant diet and chemosensory response to palatable plants. There were no consistent differences among populations in tongue-flicking and biting responses to stimuli from flowers of syntopic and allopatric plant species. The lizards may respond to cues indicative of palatability in a wide range of plant species rather than exhibiting strong responses only to locally available plant species. Nevertheless, tongue-flicking and biting frequencies varied among plant species, perhaps indicating food preferences. In addition, there were differences among populations in tongue-flick rates, latency to bite, and licking behavior. Licking was observed in only one lizard population as a response to floral chemicals from only one of the plants species tested, raising the possibility of a population-specific linkage between identification of a particular plant species and performance of an appropriate feeding response.     

Effects of Chemical Cues on Foraging in Damselfly Larvae, <Emphasis Type="Italic">Enallagma antennatum</Emphasis>

Animals experiencing a trade-off between predation risk and resource acquisition must accurately predict ambient levels of predation risk to maximize fitness. We measure this trade-off explicitly in larvae of the damselfly Enallagma antennatum, comparing consumption rates in the presence of chemical cues from predators and injured prey. Damselflies distinguished among types of chemical cues based on species of prey injured or eaten. Injured coexisting heterospecific and unknown heterospecific chemical cues did not reduce foraging relative to starved predator cues, while cues arising from predators eating a coexisting heterospecific did decrease foraging. This study shows a cost in terms of reduced foraging in response to chemical cues and further defines the ability of prey to respond discerningly to chemical cues.     

Perceived predation risk as a function of predator dietary cues in terrestrial salamanders

Prey often avoid predator chemical cues, and in aquatic systems, prey may even appraise predation risk via cues associated with the predator's diet. However, this relationship has not been shown for terrestrial predator-prey systems, where the proximity of predators and prey, and the intensity of predator chemical cues in the environment, may be less than in aquatic systems. In the laboratory, we tested behavioural responses (avoidance, habituation and activity) of terrestrial red-backed salamanders, Plethodon cinereus, to chemical cues from garter snakes, Thamnophis sirtalis, fed either red-backed salamanders or earthworms (Lumbricus spp.). We placed salamanders in arenas lined with paper towels pretreated with snake chemicals, and monitored salamander movements during 120 min. Salamanders avoided substrates preconditioned by earthworm-fed (avoidanceX+/-SE=91.1+/-2.5%, N=25) and salamander-fed (95.2+/-2.5%, N=25) snakes, when tested against untreated substrate (control). Salamanders avoided cues from salamander-fed snakes more strongly (75.2+/-5.5%, N=25) than earthworm-fed snakes when subjected to both treatments simultaneously, implying that salamanders were sensitive to predator diet. Salamanders tended to avoid snake substrate more strongly during the last 60 min of a trial, but activity patterns were similar between salamanders exposed exclusively to control substrate versus those subject to snake cues. In another experiment, salamanders failed to avoid cues from dead conspecifics, suggesting that the stronger avoidance of salamander-fed snakes in the previous experiment was not directly due to chemical cues emitted by predator-killed salamanders. Salamanders also did not discriminate between cues from a salamander-fed snake versus a salamander-fed snake that was recently switched (i.e. <14 days) to an earthworm diet. Our results imply that terrestrial salamanders are sensitive to perceived predation risk via by-products of predator diet, and that snake predators rather than dead salamanders may be largely responsible for the release of such chemicals. Copyright 1999 The Association for the Study of Animal Behaviour.