Evolutionary Biology

, Volume 36, Issue 3, pp 267–281

The Ancient Chemistry of Avoiding Risks of Predation and Disease


  • M. Yao
    • Department of BiologyMcMaster University
  • J. Rosenfeld
    • Department of Pathology and Molecular MedicineMcMaster University
  • S. Attridge
    • Department of BiologyMcMaster University
  • S. Sidhu
    • Department of BiologyMcMaster University
  • V. Aksenov
    • Department of BiologyMcMaster University
    • Department of BiologyMcMaster University
Research Article

DOI: 10.1007/s11692-009-9069-4

Cite this article as:
Yao, M., Rosenfeld, J., Attridge, S. et al. Evol Biol (2009) 36: 267. doi:10.1007/s11692-009-9069-4


Illness, death, and costs of immunity and injury strongly select for avoidance of predators or contagion. Ants, cockroaches, and collembola recognize their dead using unsaturated fatty acids (e.g., oleic or linoleic acid) as “necromone” cues. Ants, bees, and termites remove dead from their nests (necrophoric behavior) whereas semi-social species seal off corpses or simply avoid their dead or injured (necrophobic behavior). Alarm and avoidance responses to exudates from injured conspecifics are widespread. This involves diverse pheromones, complex chemistry and learning. We hypothesized that necromones are a phylogenetically ancient class of related signals and predicted that terrestrial Isopoda (that strongly aggregate and lack known dispersants) would avoid body fluids and corpses using fatty acid “necromones.” Isopods were repelled by crushed conspecifics (blood), intact corpses, and alcohol extracts of bodies. As predicted, the repellent fraction contained oleic and linoleic acids and authentic standards repelled several isopod species. We further predicted a priori that social caterpillars (lacking known dispersants) would be repelled by their own body fluids and unsaturated fatty acids. Both tent caterpillars and fall webworms avoided branches treated with conspecific body fluid. Oleic and linoleic acids were also strongly avoided by both species. Necromone signaling appears widespread and likely traces to aquatic ancestors pre-dating the divergence of the Crustacea and Hexapoda at least 420 million years ago.


IsopodaCaterpillarsDeath recognitionShelter selectionNecromonesBehaviorPredationDiseaseFatty acids


Predation and contagion are among the most powerful agents of natural selection and health is also a predominant criterion for sexual selection (Hamilton and Zuk 1982). In addition to contagion, microbes sequester food resources with toxins that animals had best avoid (Janzen 1977). Diverse phylogenies of unicells, plants, and animals employ chemically-mediated detection, communication, and responses to predation, herbivory, parasitism or disease (Tollrian and Harvell 1999; Wisenden 2003; Kavaliers et al.2006; Gelperin 2008). Wilson et al. (1958) demonstrated that “necrophoric” removal of corpses by ants was elicited by oleic acid. Linoleic acid was also effective (Akino and Yamaoka 1996). Such behavior occurs in many ants and bees and could serve to remove sources of contagion, or avoid disease or toxins associated with decomposition (Rollo et al.1994, 1995; Julian and Cahan 1999; Masterman et al.2001; Ayasse and Paxton 2002).

In ants, sensitivity to oleic acid is caste-specific, soldiers being refractory (Lopez-Riquelme et al.2006). Detection of oleic and linoleic acids and oleic acid ester were enhanced in “mite-resistant bees” (Martin et al.2002). Bees avoid flowers where conspecifics may have been killed (Dukas 2001; Abbott 2006) or those treated with extracts of conspecific body parts (Stout et al.1998). Necromone recognition, undertaking and avoidance of dead and injured conspecifics may serve to reduce risks of predators and contagion that are reliably associated with such signals (e.g., Wilson-Rich et al. 2009). Bees, termites, beetles, lobsters, and mice variously discriminate infected conspecifics (e.g., viral, bacterial, protozoan, nematode) via olfactory cues (Ayasse and Paxton 2002; Worden and Parker 2005; Behringer et al.2006; Kavaliers et al.2006).

Blood, intact corpses and alcohol extracts of conspecifics are highly repellent to cockroaches. Exploration of possible mechanisms highlighted death recognition as the most likely function (Rollo et al.1994, 1995). Cockroaches are ancestral to termites that are also repelled by their dead and remove or bury corpses (Crosland and Traniello 1997; Su 2005). The hypothesis that cockroaches recognize their dead led us to Wilson et al.’s (1958) ant study. Even in the early 1990s it seemed improbable that insects as phylogenetically distant as ants and cockroaches (at least 200 million years) might share common recognition mechanisms for death. This highlights a major shift in evolutionary theory from expectations of progressive divergence of features and associated genes to recognition of remarkable conservation across genetic, biochemical and physiological levels (Rollo 2006). We postulated that the active fraction avoided by cockroaches might contain fatty acids as in ants. This proved to be exactly the case with linoleic and oleic acids (the signals employed by ants) identified as the specific signals.

John Borden (Simon Fraser University) suggested the term “necromones” to designate this new class of behavior-modifing chemistry. This has remained relatively unexplored, possibly because conservation of ecological chemistry is rarely considered in a world of species-specific pheromones and defensive products. Regardless, a necromone function for unsaturated fatty acids (linoleic) was recently extended to Collembola, a primitive order predating ants by ~300 million years (Nilsson and Bengtsson 2004a, b).

We hypothesized that fatty acid “necromones” are fundamentally associated with injury and death and consequently may serve as reliable cues of risk across wide phyla of organisms. This may be particularly true of species that aggregate (Briones-Fourzan et al. 2008). As a strong test of this idea we predicted that terrestrial Isopoda that strongly aggregate in shelters and express a shelter-marking aggregation pheromone (Kuenen and Nooteboom 1963; Takeda 1984, this study) would be repelled by injured or dead conspecifics. We employed crushed bodies to simulate predator-induced injury, and intact corpses that could indicate the presence of disease. We further hypothesized that, as in cockroaches, alcohol extracts of bodies will prove repellent even though no endogenous isopod repellents are known. We further predicted a priori that the repellent fraction of these extracts would contain fatty acids, and of these, tests with authentic oleic and linoleic acids would prove to specifically mediate repellency. As with cockroaches, we tested isopods belonging to several genera to ensure generality within the Isopoda.

All of these unlikely predictions are strongly confirmed here. We chose isopods because they are crustaceans and the phylogenetic divergence of insects and crustaceans dates to >420 million years ago (Gaunt and Miles 2002). This supports a truly ancient origin of risk avoidance utilizing fatty acid necromones. Given the strength of these results and likely generality we postulated that necromones have simply been overlooked or ignored because repellency of body extracts from species that aggregate and employ aggregation pheromones seems paradoxical. Our original extraction of cockroach bodies was expected to obtain aggregation.

To test this we examined responses of two well studied social caterpillar species to body fluids and oleic and linoleic acids. Tent caterpillars (Malacosoma americanum) and fall webworms (Hyphantria cunea) belong to different families and differ in life histories and host plants. Although extensively studied, no record of endogenous repellents was found for either species. Tent caterpillars regurgitate enteric fluids in ant defense that contain hydrogen cyanide and other materials repellent to ants. However, they are immune to such materials and actually seek out leaves containing the highest amounts (Peterson et al.1987; Fitzgerald 1995). Remarkably, our preliminary assessment found profound avoidance of body fluids and unsaturated fatty acids in both lepidopteran species.

Experimental Procedures


To ensure generality of results in Isopoda we utilized species belonging to four different genera. Oniscus asellus (L.),Armadillidium vulgare (Latreille), Porcellio scaber (Latreille), and Porcellionides pruinosus (Brandt) were collected locally or purchased from Ward’s Natural Science, St. Catherines, ON. Holding tanks were maintained at 22 ± 1°C and a photoperiod of 12 h light: 12 h dark, the same conditions employed for experiments. Protocols adhered to Canada Council on Animal Care guidelines.

Standard Bioassays

The bioassay arena was delimited by a circular plastic barrier 15 cm high and 30 cm across. The floor consisted of absorbent paper moistened with distilled water. Eight shelters constructed from 5.5 cm diameter Petri® dishes were inverted and arranged in a circle surrounding a central dish of food (wood, potato, and carrot slices, and dog chow). A 2.0 cm hole in the side of each Petri® dish served as a doorway. Initially doorways were oriented toward the food dish to ensure equidistance from food. Woodlice tend to follow the edge of the arena, however, and orienting shelter entrances toward the arena wall facilitated discovery.

Filter papers were placed on the bottom of each shelter (top half of the inverted Petri® dish) and moistened with distilled water. This prevented test material from diffusing out of shelters. Aluminum (“tart”) cups placed over the shelters minimized light, and were easily lifted to count residents. Woodlice were released into the center of the arena 0.5 h before dark onset and numbers in shelters were counted the next day.

For each assay, test materials (body extracts, corpses, crushed bodies, authentic chemicals) were placed on filter papers and any solvent was allowed to completely evaporate. Control (evaporated solvent only) or treated papers were placed in every other shelter to obtain a balanced distributed. Care was taken to ensure uniform lighting as woodlice prefer darker shelters. The number of animals tested in an assay varied with availability but usually 15–25 animals of a single species were observed. Each experiment involved 3–15 assays.

Aggregation Pheromone

Woodlice deposit aggregation pheromone in feces although this has been little studied (Kuenen and Nooteboom 1963; Takeda 1984). We examined intra- and inter-specific responses to fecal material to quantify attractiveness of aggregation pheromone. This was of interest because presence of aggregation pheromone means that repellency would need to be sufficiently strong to override this signal (i.e., alcohol extracts, crushed bodies, intact whole bodies). Feces were scraped from the paper liners of holding tanks containing large numbers of woodlice. About 1/2 tsp of feces was smeared on filter papers in experimental shelters. Controls received no treatment.

Crushed Conspecifics

Crushed conspecifics were considered to test responses to possible predation where blood and dismembered body parts might be encountered. Experiments with intact and crushed dead bodies used Petri® dish shelters 4 cm in diameter. P. pruinosus were killed by freezing and two bodies were crushed and spread on papers in experimental shelters. Six assays were carried out with 20 isopods each. O. asellus were killed by freezing and two bodies were crushed and spread across papers placed in experimental shelters. Four assays were carried out for O. asellus. P. scaber and A. vulgare were similarly tested against crushed O. asellus.

Intact Dead Bodies

Intact dead bodies may be indicators of contagion but lack cues associated with injury such as blood. Removal of dead and diseased brood by bees is considered defence against infection, and brood simply killed by freezing elicit such responses (Wilson-Rich et al.2009). O. asellus were killed by freezing and four thawed intact bodies were placed in each experimental shelter. Responses of O. asellus, P. scaber, and A. vulgare to dead O. asellus were examined. Responses of P. pruinosus to dead conspecifics were also examined.

Body Extracts

We previously found that repellent constituents of cockroaches were soluble in alcohol (Rollo et al. 1994, 1995) and that refrigerated alcohol extracts maintained repellency for at least a month. Generality of necromones would be supported if alcohol extracts were repellent in isopods, since they normally aggregate. Extracts of P. pruinosus were made with 100 adult animals in 20 ml of 90% ethanol. Extracts of other species were made with 100 animals in 40 ml of 95% ethanol. Experiments with crushed conspecifics, dead bodies or extracts employed freezing as euthanasia. Extractions were refrigerated for 3 days with daily agitation, following which bodies were removed. Concentrations applied in experiments were calculated as woodlouse body equivalents (1 BE = material from one woodlouse).

Intraspecific Responses to Ethanol Body Extracts

We tested the repellency of conspecific extracts across species to assess generality but also whether the degree of response might vary. Extracting repellent constiuents was also a critical aspect for chemical identification. For experiments with ethanol extracts, control papers received equivalent amounts of ethanol as shelters treated with extract. All papers were air-dried until all ethanol evaporated. Papers were then moistened with distilled water for the assay. For P. pruinosus experimental shelters received 2 ml (~10 BE) of extract.

Eighteen assays utilizing 15–20 woodlice each were carried out. In addition three assays were run where all shelters were treated with body extract (i.e., no controls). For this experiment the number of animals inside treated shelters was compared to those that remained outside during the photophase. The repellency of species-specific body extracts were examined for O. asellus,P. scaber, and A. vulgare across a range of 1–5 BE of extract (Fig. 1). Because we did not test all species at 1 and 5 BE, ANOVA (Statistica®) was limited to concentrations of 2–4 BE to obtain a balanced design. There were 15 assays/dose and 45 assays per species.
Fig. 1

Dose-response relationships of the isopods O. asellus, P. scaber, and A. vulgare to conspecific body extract. There were species-specific differences in the degree of repellency to conspecific extract (see text)

Interspecific Responses to Body Extracts

Woodlice often form multi-species aggregations although our laboratory observations also noted strong species-specific association. Utilization of specific fatty acids signals would be expected to yield cross-species reactivity and detecting risks in related species associates could prove adaptive. Alternatively, we considered that other cues (e.g., species recognition) might modulate repellent responses. We carried out an ANOVA restricted to responses of O. asellus and P. scaber to doses of 3 and 4 BE of O. asellus extract. We also directly compared the responses of P. scaber to its own extract versus that of O. asellus at a dose of 3 BE. Further, we examined the response of A. vulgare to 5 BE of its own extract, compared to a weaker response to O. asellus extract at 5.5 BE.

Partitioning of Woodlouse Extract

Once a repellent fraction was obtained we proceeded to identify the active fraction and the identitiy of its chemical constituents. The active fraction of woodlouse extract was identified via testing fractions with the standard bioassay. Twenty milliliter of ethanol extract from O. asellus was transferred to a round bottom flask and the ethanol was evaporated using a rotator evaporator. The dried residue was twice reconstituted with 10 ml of ether and poured into a separatory funnel to retain the original volume of extract. Ten milliliter of 1% ammonium hydroxide (NH4OH) in water was then added. After partitioning, the alkaline aqueous bottom layer was drained into a clean flask. This extraction was repeated three times and the remaining ammonical aqueous fractions were combined in the flask and covered with tinfoil and Parafilm for future use. The ether layer was drained from the separatory funnel into an Erlenmeyer flask with sodium sulfate (Na2SO4) to dry any remaining water. The ether extract was then decanted into a volumetric flask using a pasture pipette and filter funnel, and diluted with ether to achieve a total volume of 50 ml. To test activity, 4 ml (4 BE) of ether extract was applied to treated shelters with ether alone to controls.

The NH4OH extract was acidified by adding 2 ml of HCl and transferred to a clean separatory funnel. Ten milliliter of ether was added to the system and shaken. After the mixture partitioned, the bottom layer was drained into the original flask and the top layer was drained into a clean flask. The contents of the original flask were poured back into the separatory funnel and the partitioning procedure was repeated twice more. All of the resulting top layers were combined in the clean flask and mixed with Na2SO4 to remove water from the system. The supernatant was decanted into a volumetric flask using a pasture pipette and filter funnel, and then diluted with ether to a total volume of 50 ml. The activity of the extract was tested using 4 ml (4 BE)/treated shelter as above.

Chemical Analysis

One milliliter of the ether extract was evaporated under a slow stream of nitrogen and the dry residue derivatized with pentaflourobenzyl bromide by standard procedures (Arneson and Roberts 2007; Rosenfeld 2002, 2003). The derivatized extract was analyzed by gas chromatograph mass spectrometry (GCMS) [Varian® gas chromatograph (GC) and Varian® quadrupole mass spectrometer]. The GC employed a Factor 4 column (VF-5 ms) 30 M long, with 0.25 mm inner diameter and a phase thickness of 0.25 μm. The injector port was kept at 320°C with a split ratio of 25 (i.e., the flow was split with one part entering the column and 25 parts vented to air). The temperature program was as follows: Initial temperature 150°C, hold for 0.5 min; First ramp to 250°C at 30°C/min.; Second Ramp to 285°C at 10°C/min.; Third ramp to 320°C at 40°C/min. Hold for 1 min. Mass spectra were obtained in negative chemical ionization (CI−) mode with accelerating voltage at −150 eV with the ion source at 200°C.

Authentic Unsaturated Fatty Acids

The ultimate test of phylogenetic generality is whether the specific fatty acid signals mediating repellency are the same in isopods and insects. For cockroaches, linoleic acid was most repellent and consequently we initially characterized responses of P. pruinosus to linoleic acid (≥99% pure, Sigma-Aldrich Inc.) at concentrations of 50, 30, 20, 10, and 4% (diluted with 90% ethyl alcohol). Concentrations of 50–10% were assessed using 15 animals tested individually. For the 4% concentration, 12 assays were completed with 10–20 woodlice/assay. P. pruinosus were similarly tested in nine assays where all shelters were treated (i.e., no control shelters). Woodlice are strongly light aversive so the number of animals that failed to enter shelters was compared to those that did.

Since linoleic acid (two double bonds) was more repellent to cockroaches than oleic acid (one double bond), we hypothesized that linolenic acid (three double bonds) might be even more repellent. Synthetic oleic, linoleic, linolenic, stearic, and palmitic acids (≥99% pure, Sigma-Aldrich Inc.) were assayed using 15 replicates for each (24 woodlice/assay). These are all acids that occurred in the active fraction of cockroach extracts and also here for woodlice (see below). Only O. asellus and P. scaber were assessed. Acids were diluted to 1% concentration which represents ~10% of the body mass for one adult O. asellus (i.e., equivalent to a few dead woodlice). Treated filter papers received 1 ml of 1% fatty acid in 95% ethyl alcohol whereas controls received 1 ml of 95% alcohol.

Social Caterpillars

We tested the strength of the necromone hypothesis by predicting that two social caterpillars would be repelled by conspecific body extracts and oleic and linoleic acids. Tent caterpillars (Malacosoma americanum) were collected from the field in spring-early summer, and fall webworms (Hyphantria cunea) were obtained in late summer and fall. Branches from host trees (cherry and walnut, respectively) were collected and washed in distilled water. The bioassay employed two branches matched for size and number of leaves, placed in a plastic container lined with moistened paper (Fig. 4a). Escape was prevented by a thin layer of Vaseline® around the upper rim of the container. Each branch protruded through holes in the container floor into flasks of water. For each assay, 100 caterpillars were placed on the arena floor equidistant from either branch. Caterpillars on branches were counted 24 h later for H. cunea and 3 h later for M. americanum. Each experiment consisted of 8–14 replicates.

Alcohol extracts would negatively impact leaves so we employed raw body fluids from caterpillars. The bodies of 100 late instar caterpillars were thoroughly mashed upon thawing and the slurry was directly applied onto all stems and leaves of the test branch (immediately before testing). Control branches received no treatment. For oleic or linoleic acids, we mixed three parts of either acid (Sigma-Aldrich Inc) in 97 parts of distilled water. The actual concentrations of fatty acids were somewhat less than 3% because for these preliminary experiments the oleic acid used was 90% pure and linoleic was 60% pure. This would achieve levels of these acids within physiological ranges. Although fatty acids are not water soluble they were suspended by vigorously shaking in distilled water. The test branch was immersed and stirred in this solution. The control branch was similarly treated in distilled water.


Repellency or attraction was calculated as [(N/2−T)/(N/2)] * 100, where N = the number of woodlice tested and T = the number in treated shelters. This yields a range of values where 0% repellency is equivalent to a 50–50% distribution of animals between test and control shelters. Preliminary experiments suggested that individual animal choices were independent with respect to treated versus control shelters even though there was obvious aggregation among control shelters. Regardless, we applied statistics at the level of group assays rather than individual choices to ensure independence of observations. Regression, ANOVA, and Chi Square analyses were performed using Statistica®. Chi Square was performed using raw counts whereas for other analyses proportional data (e.g., repellency for ANOVA) were transformed (ArcSin(Sqrt(X))) to improve normality as suggested by Zar (1974). For interpretation all descriptive statistics are presented as untransformed (means ± standard error (SE)).


Aggregation Pheromone

For A. vulgare tested with its own feces (seven replicates) only 24/112 chose untreated shelters (Mean Attraction = 57.14%, χ2 = 30.624, d.f. = 6, P < 0.00003). When A. vulgare was tested against combined feces from O. asellus and P. scaber, however, results of four replicates suggested at best a weak trend for attraction (46/77 choices) that proved non-significant (Mean attraction = 5.133%, χ2 = 6.0412, d.f. = 3, P > 0.10). For P. pruinosus the presence of feces in shelters was significantly attractive, 75 of 108 choosing shelters treated with conspecific feces (Mean attraction = 38.89%, χ2 = 13.369, d.f. = 5, P < 0.021).

Responses to Crushed Conspecifics

When P. pruinosus were killed by freezing and two bodies were then crushed and spread across papers in experimental shelters (simulated predation) only three of 113 woodlice selected treated shelters (Mean repellency = 94.6%, χ2 = 52.7, d.f. = 5, P < 0.00001). Responses of O. asellus, P. scaber, and A. vulgare to crushed O. asellus (two crushed bodies per experimental shelter) were all highly significant. Of 142 choices by O. asellus, only 16 chose treated shelters (Repellency = 77.465%, χ2 = 44.27, d.f. = 3, P < 0.00001). Although P. scaber showed weak responses to intact bodies of dead O. asellus (see below) only two out of 45 chose shelters treated with crushed O. asellus (Mean repellency = 91.11%, χ2 = 18.911, d.f. = 3, P < 0.0003). A. vulgare was highly repelled by crushed O. asellus, only one out of 29 choosing treated refuges (Mean repellency = 93.1%, χ2 = 12.667, d.f., = 3, P < 0.005).

Responses to Intact Dead Bodies

Initial observations with P. pruinosus showed highly variable and non-significant responses to intact bodies of dead conspecifics (potential contagion) despite strong responses to crushed bodies. Subsequently, four assays were run with the other species using treated shelters containing intact bodies of four O. asellus. Immediate responses to freshly thawed bodies were highly variable. Some fresh corpses were actually fed upon. After 24 h, however, treated shelters were occupied by only 27 of 151 O. asellus (Mean repellency = 64.2%, χ2 = 31.89, d.f. = 3, P < 0.00001). Repellency of dead O. asellus to A. vulgare was similar, only eight of 50 choosing treated shelters (Mean repellency = 68.0%, χ2 = 12.32, d.f. = 3, P < 0.0064). P. scaber, however, showed relatively weak responses to bodies of O. asellus, 18 of 49 choosing shelters containing dead O. asellus (Mean repellency = 26.53%, χ2 = 3.907, d.f. = 3, P > 0.27, ns).

Intraspecific Responses to Body Extracts

No endogenous repellent is known for woodlice. Body extracts of P. pruinosus (~10 body equivalents per shelter) were highly repellent to conspecifics, only five of 151 animals choosing treated shelters (Mean Repellency = 93.38%, χ2 = 142.12, d.f. = 17, P < 0.00001). If no control shelters were provided (i.e., all shelters treated), 9/57 animals entered treated shelters and the remainder stayed outside during the photophase (Repellency = 68.42, χ2 = 14.82, d.f. = 2, P < 0.006). If control shelters were then provided, 100% of woodlice chose control shelters.

We examined the responses of O. asellus, P. scaber, and A. vulgare to various concentrations of their own body extracts. Each species showed a unique dose-response curve across concentrations of 1–5 BE (Fig. 1). We performed ANOVA limited to a range of 2–4 BE to obtain a balanced design. ANOVA identified significant differences among species (P < 0.00001), doses (P < 0.00001) and a species-dose interaction (P < 0.00001). Newman-Keuls statistically resolved all three species from one another (P < 0.05 in all cases).

The mean repellency of A. vulgare across these doses (50.93% ± 6.17 SE) was lowest and differed from the other species by at least P < 0.00002. The mean repellency of P. scaber (84.42% ± 3.06) was greatest and differed significantly from O. asellus (72.79% ± 4.22) to a lesser degree than from A. vulgare (P < 0.022). Differences among specific doses (Newman-Keuls) were complex (Fig. 1). A. vulgare displayed exceptionally low repulsion at the lowest dose of 2 BE, differing from the other species by at least P < 0.00009. P. scaber exhibited much higher repulsion than the other species at 3 BE (P < at least 0.009). All species showed repulsion greater than 87% at 4 BE.

Interspecific Responses to Ethanol Extracts

We were interested in whether necromone signals may be modulated by other species-specific factors. Responses of P. scaber and A. vulgare to extracts of O. asellus were markedly different from intra-specific results (Fig. 2). ANOVA was performed for P. scaber and O. asellus to confirm differences in response at 3 and 4 BE. ANOVA detected highly significant differences between both species (P < 0.00001) and doses (P < 0.00001). Newman-Keuls found that P. scaber responses at 4 BE of O. asellus, did not differ from O. asellus responses at 3 BE. At 3 BE P. scaber showed very low repellency (6.10% ± 4.33) compared to O. asellus (71.68% ± 2.03), P < 0.0002). At 4 BE O. asellus achieved nearly full repellency (98.34% ± 1.66) whereas repellency of P. scaber remained moderate (62.24% ± 6.67) (P < 0.002).
Fig. 2

Dose-response relationships of isopods to body extracts of O. asellus. The response relationship of P. scaber and the single assessment for A. vulgare indicated much weaker repellency to extracts of O. asellus compared to conspecific extract

To further clarify, we compared responses of P. scaber to its own extract and responses to O. asellus extract (3 BE in either case). The response of P. scaber to its own body extract was remarkably strong (92.85% ± 12.59) compared to that of O. asellus (6.10% ± 4.33) (P < 0.00001, t-test on ArcSin data, d.f. = 18). We also compared the strong response of A. vulgare to its own extract at 5 BE (93.35% ± 2.63) to its response to O. vulgare extract at 5.5 BE (66.70% ± 3.51) (P < 0.00005, t-test).

Chemical Analyses

Upon acid-base fractioning of the O. asellus extract, the basic and neutral portions showed no repellent activity whereas the acidic portion expressed full activity when tested in bioassays. The repellent fraction showed three strong peaks corresponding mainly to palmitic, stearic, and C18 fatty acids (i.e., oleic, linoleic, linolenic) (Fig. 3). Structures were determined by GCMS based on comparison of retention times of compounds in the sample with authentic samples as well as by mass spectra. The peak attributed to palmitic acid showed single mass at 255 corresponding to the M-1 (molecular weight less 1 H). This is formed by the loss pentaflourobenzyl (PFB) group from the PFB ester of palmitic acid that is typical of the CI- fragmentation. GCMS retention times did not discriminate among C18 unsaturated fatty acids but the mass chromatograms indentified the M-1 ions of: stearic acid at 283; mono-unsaturated (oleic acid, MUSA) at 281; di-unsaturated (linoleic acid DUSA) at 279; and tri-unsaturated (linolenic acid, TUSA) at 277. The approximate ratio was 283/281/279/277 = 8/8/8/1 indicating that oleic and linoleic acid were present in roughly equivalent amounts and linolenic acid was about 8-fold lower. Likely structures of the unsaturated acids were oleic (MW = 282), linoleic (mw = 280), and linolenic acid (MW = 278). Without complete chromatographic separation of the PFB derivatives of the acids it was not possible to more rigorously assign structure since mass spectra did not differentiate cis and trans acids.
Fig. 3

Gas chromatograph separation of the active fraction of body extract from O. asellus. Showing peaks for palmitic, stearic, and C18 fatty acids). Mass spectrum analysis and verification with authentic acids identified palmitic and stearic acids and detected oleic, linoleic, and linolenic acids as C18 constituents

Unsaturated Fatty Acids

Dilutions of linoleic acid of 50, 30, 20, and 10% (2 ml/shelter) were all completely repellent (P < 0.01 in all cases) to P. pruinosus (15 animals tested individually at each concentration). Using groups of 10–20 animals in 12 assays, a 4% solution of linoleic acid was highly repellent, zero of 130 woodlice choosing treated shelters (Repellency = 100%, χ2 = 70.0, d.f. = 11, P < 0.00001). If all shelters were treated 62/170 woodlice entered shelters and the rest remained outside during the photophase. Thus, shelters were still significantly avoided even in the presence of light (Repellency = 27.06%, χ2 = 42.76, d.f. = 8, P < 0.00001). When control shelters were re-established, 100% of the animals (20) choose control shelters.

O. asellus showed no significant response to 1% stearic acid (Repellency = −0.037%, χ2 = 4.960, d.f. = 11, P > 0.93) or 1% palmitic acid (Repellency = 0.053%, χ2 = 3.5217, d.f. = 8, P > 0.89). In contrast, 1% linoleic acid was strongly repellent to O. asellus (Repellency = 76.98%, χ2 = 77.11, d.f. = 11, P < 0.00001). In further experiments comparing repellency of oleic, linoleic, and linoleic acids to O. asellus and P. scaber, ANOVA (arcsin transformed) detected significant differences among the acids (P < 0.00022) but not species (P > 0.93). Newman-Keuls indicated that linoleic and oleic acids were differentiated from linolenic acid, but not from one another.

Examination of the specific responses of each species to the unsaturated acids using raw counts and Chi square analyses showed that despite a strong trend, repellency of linolenic acid did not reach statistical significance for either O. asellus (Repellency = 29.98% ± 5.66, χ2 = 5.899, d.f. = 4, P > 0.21) or P. scaber (Repellency = 37.74% ± 4.44, χ2 = 5.497, d.f. = 4, P > 0.24). Linoleic acid was most repellent and most significant to both O. asellus (Repellency = 61.68% ± 3.34, χ2 = 22.199, d.f. = 4, P < 0.0002) and P. scaber (Repellency = 53.36 ± 5.46, χ2 = 12.938, d.f. = 4, P < 0.0114). Oleic acid repellency was intermediate and statistically significant for both O. asellus (Repellency = 48.34 ± 5.54, χ2 = 14.112, d.f. = 4, P < 0.0069) and P. scaber (Repellency = 48.88 ± 2.74, χ2 = 10.366, d.f. = 4, P < 0.035).

Social Caterpillars

Both social caterpillar species were significantly repelled by conspecific body slurries, and by both oleic and linoleic acids (Table 1). The caterpillar counts were conservative as many left the untreated branches as they became defoliated. Photographs of fall webworms (Fig. 4a–d) illustrate dramatic protection of leaves by all test materials. Caterpillars avoided walking on treated branches. Once the untreated branch was defoliated, most caterpillars descended and wandered about the arena floor. Only a few climbed treated stems.
Table 1

Impact of conspecific body fluids and oleic and linoleic acids on foraging of eastern tent caterpillars (Malacosoma americanum) and fall webworm (Hyphantria cunea)



Number per branch (Mean ± S.E.)


Probability (t-test)

M. americanum

(a) Control

28.7 ± 2.1



(b) Body extract

2.4 ± 0.6a



(a) Control

13.0 ± 1.7



(b) Oleic acid

1.6 ± 0.3



(a) Control

15.0 ± 1.3



(b) Linoleic acid

2.9 ± 0.4



H. cunea

(a) Control

12.5 ± 3.6



(b) Body extract

1.1 ± 0.5



(a) Control

10.9 ± 1.7



(b) Oleic acid

1.1 ± 0.4



(a) Control

10.1 ± 2.7



(b) Linoleic acid

1.3 ± 0.3



aNo more than three tent caterpillars were observed on the treated branch until the control branch had been entirely defoliated. For all replicates 100 caterpillars were originally placed on the arena floor between the branches. Counts refer to observations 3 or 24 h following release for M. americanum and H. cunea, respectively. Counts are conservative as caterpillars left the untreated branch as it became defoliated

Fig. 4

a Initial setup for bioassays for testing caterpillar responses to various treatments. b Typical bioassay for body exudates 24 h after releasing 100 fall webworms. T = treated in all frames. c Typical bioassay for oleic acid 24 h after releasing 100 fall webworms. d Typical bioassay for linoleic acid after releasing 100 fall webworms. Results for tent caterpillar body extract were similar except cherry branches were used instead of walnut


Responses to Crushed Conspecifics

All isopod species were strongly repelled by crushed conspecifics. Woodlice entered shelters containing crushed conspecifics but vacated them following inspection. This argued against alarm pheromones as these are generally volatile, influence relatively large areas and dissipate relatively quickly. Some literature attributing avoidance of dead or injured conspecifics to alarm pheromones may in fact involve necromones. Rather than a mutual signal between individuals, necromones likely involve biochemicals that are reliably associated with death, injury and perhaps other pathologies. Evolution of specific recognition and behavioral responses to ubiquitous and reliable signals of risk would be highly adaptive. Unlike alarm pheromones, unsaturated fatty acids are relatively non-volatile and persistent (forest beetles were repelled for 22 days by oleic acid) (Nijholt 1980). The strength of responses suggests that woodlice likely recognize and avoid sites where isopods have been killed.

Responses to Intact Dead Bodies

Responses to intact conspecific corpses (possibly indicative of contagion) were generally weaker than to crushed bodies or extracts. Interestingly, P. pruinosus, did not respond to conspecific corpse at all and P. scaber was repelled much less by intact bodies of O. asellus than crushed O. asellus. Alternatively, A. vulgare responded strongly to both crushed and intact corpses of O. asellus. P. pruinosus and P. scaber may experience higher risks from predation (blood) than disease (intact corpses).

Responses to O. asellus corpses increased with time, with strong avoidance evident within 12–24 h. Akino and Yamaoka (1996) suggested that high production of oleic acid in ant corpses occurred over 48 h consistent with increased likelihood of removal (Wilson et al. 1958). Howard and Tschinkel (1976) suggested that death recognition involving fatty acids occurred within an hour in fire ants but Visscher (1983) observed death recognition within ~10 min in honey bees. Leakage of hemolymph might influence responses and relative risks of unhygienic behavior (and sensitivity to necromones) likely shows species variation. Attraction to low-level contagion may serve to obtain immunity. Ants benefit from contact with infected nestmates and in termites contact with individuals with acquired immunity may convey “social immunization” (Traniello et al.2002; Ugelvig and Cremer 2007). Such mechanisms may explain paradoxical attraction to low-level pathology despite general repellency to frank disease (Surinov 2007).

Intraspecific Responses to Body Extracts

Repellency of body extracts strongly supported our necromone hypothesis because we would otherwise expect to extract aggregation pheromone. Extracts even overrode the avoidance of light otherwise paramount to isopods. P. pruinosus rarely remained outside shelters during the photophase, but when all shelters were treated with extract only 9/57 entered (P < 0.006). Dose-responses to alcohol extracts of frozen bodies obtained significant differences among O. asellus, P. scaber, and A. vulgare, with P. scaber showing greatest sensitivity and A. vulgare the least (Fig. 1). Variation may reflect differences in body size, differences in sensitivity to signaling constituents or differences in fatty acid composition among species. O. asellus and A. vulgare were larger than P. scaber so simple differences in amounts of signaling material/BE is unlikely.

Interspecific Responses to Ethanol Extracts

Responses of P. scaber and A. vulgare to body extract of O. asellus were significant but greatly diminished compared to conspecific extracts (compare Figs. 1 and 2). Thus, fatty acids are generally repellent but species-specific responses suggest additional cues.

Chemical Characterization

The active fraction of O. asellus extract (Fig. 3) showed strong peaks for palmitic, stearic, and C18 fatty acids (Fig. 3) but only oleic and linoleic acid were significantly repellent. Further increases in C18 fatty acids might be derived following death by enzymatic or bacterial actions (not yet examined). Thus, isopods share oleic or linoleic acid necromones with ants, cockroaches, and springtails as predicited.

Responses to Authentic Fatty Acids

Linoleic acid was 100% repellent to P. pruinosus using 1 ml/shelter of fatty acid solution at concentrations ranging from 50% to 4%. No fatty acid at 1% concentration obtained 100% repulsion. Using 1% solutions, authentic linoleic acid appeared most repellent to woodlice although not statistically resolved from oleic acid. Repellency of linoleic, and oleic acids, but no repellency by palmitic or stearic acids parallels results with cockroaches (Rollo et al. 1994, 1995). Saturated fatty acids were also not repellent to Collembola (Nilsson and Bengtsson 2004a).

Our hypothesis that repellency reflects the degree of unsaturation was rejected. Hwang et al. (1984) made a similar conclusion for mosquito oviposition repellents. For woodlice, linolenic acid (three double bonds) was not significantly repellent whereas oleic (one double bond) and linoleic acid (two double bonds) were highly repellent (P < 0.05). Given that repellency of linolenic acid was ~30%, a larger sample size or higher concentration might obtain statistical resolution. Results with cockroaches obtained repellency with oleic acid, but linoleic acid was more effective (Rollo et al.1994). Collembola strongly responded to linoleic acid but not oleic acid (Nilsson and Bengtsson 2004a).

Signaling Complexities

Aggregation pheromone is present in feces of woodlice and pillbugs (Kuenen and Nooteboom 1963; Takeda 1984) although our data indicate relatively high species specificity. To be effective, necrophobia must override aggregation pheromone (an important signal providing safety against desiccation). Species extracts were universally repellent at sufficient doses, but displayed species-specific dose-response relationships (Figs. 1 and 2). In contrast, despite variation in repellency among authentic acids, no species-specific differences were detected. Additional signals could be aggregation pheromones (see above) or other kin-recognition cues. Rather than antagonism between attractive and repellent signals, aggregation pheromone may even synergize with necromones to convey that “the dead are us.”

The quality, quantity, and context of fatty acid mixtures may be important. Ants show context-dependent utilization of unsaturated fatty acid signals for recognizing corpses or assessing food (Gordon 1983). In a pervasive coevolutionary interaction, more than 3,000 plants (spanning 80 families) express “elaisome” structures on their seeds that chemically elicit ant transport. The lipid balance (which includes oleic and linoleic acids) may mimic dead insects (Hughes et al.1994).

Nestmate recognition in bees involves multi-channel signaling that includes oleic, linoleic, palmitoleic linolenic acids, methyl esters, and diverse hydrocarbons (Breed 1998; Buchwald and Breed 2005; Slessor et al.2005; Cremer and Sixt 2009). Wilson et al. (1958) noted that an ester as well as oleic acid induced ant necrophoric behavior. Fatty acid combinations may contribute to a recognition template reflecting colony membership. Major changes in constituents of this “identity badge” alerts guards (Buchwald and Breed 2005). Death would severely distort this template, possibly converting nestmate to corpse recognition. Thus, guarding and undertaking may simply require recognizing differences in the balance of common cues. Interestingly, bees may recognize infected individuals and guards may bar their entry (Cremer and Sixt 2009). Combinatorial codes using a conserved, reliable alphabet may represent a quintessential “unity amidst diversity” in the evolution of recognition systems.

Phylogenic Breadth of Necromones

Fatty acid necromones span wide phylogenies of insects (e.g., Hymenoptera to Collembola). Bees and ants arose >100 MYA (Wilson and Hölldobler 2005; Poinar and Danforth 2006). Cockroaches were present by the late Carboniferous, 323 MYA and derived termites ~150 MYA (Krishna and Grimaldi 2003). Insects resembling Collembola date to ~400 MYA (Gaunt and Miles 2002). Similar necromones in ants, cockroaches, and springtails suggests an origin predating the Hexapoda. Shared necromones by insects and crustaceans suggests a truly ancient origin or remarkable convergence on reliable cues of death and contagion. Insects and fairy shrimp (Anostraca) shared a common ancestor ~420–430 MYA (Gaunt and Miles 2002) suggesting necromones arose in aquatic species.

Predator and Disease Avoidance in Crustacea

Many aquatic crustaceans show strong responses to chemicals signaling predation and disease. Amphipods exhibit alarm responses to predators and injured conspecifics (Williams and Moore 1985; Wudkevich et al.1997; Wisenden et al.2001). Amphipods and isopods reduce their activity in habitats harbouring fish (Andersson et al.1986; Short and Holomuzki 1992; Holomuzki and Hatchett 1994). Intertidal barnacles deploy predator-resistant morphology against predaceous snails (Lively et al. 2000) and shrimp alter their growth and antennal morphology in response to crushed conspecifics (Nga et al.2006).

Crayfish and several crabs exhibit alarm responses or reduced feeding in response to conspecific exudates and some respond to heterospecific cues (McKillup and McKillup 1992; Rittschof et al.1992; Hazlett 1994, 2003; Aqistapace et al. 2005; Hazlett and McLay 2005). Blue crabs (Callinectes sapidus) suffer high mortality when released from hatcheries due to small carapace spines and inadequate burying behavior but this normalizes over time (Davis et al.2004). They avoid injured conspecifics but not injured stone crabs (Ferner et al.2005). Similarly, dead spider crabs in lobster traps exclude conspecifics but not lobsters (Stachowicz 2001).

Panulirus interruptus lobsters aggregate in refuges but are repelled by excised body parts and their own dead (Zimmer-Faust et al.1985). Panulirus argus responded with alarm (necrophobia, retreat, increased sheltering, suppressed feeding) to conspecific blood. This species showed mixed alarm and appetitive behaviors to P. interruptus, whereas blood from blue crabs induced appetitive behavior (Shabani et al. 2008). P. argus even avoided conspecifics with viral infections (Behringer et al.2006). Panulirus guttatus showed no avoidance of water from injured conspecifics but P. argus did (Briones-Fourzan et al.2006, 2008). Such examples highlight that death recognition is generally coupled to additional cues conveying relatedness as found for woodlice.

Fatty acids and cell membrane processes are good candidates for localized signaling of contagion (disease) or demise (predation) even in submerged environments. Although lipids spread as surface films, their insolubility has excluded them from consideration as aquatic signals. Regardless, kin-recognition and schooling in catfish is rapidly mediated by the phospholipid, phosphatidylcholine (Matsumura et al. 2004). Recent findings also reveal that body fluids, including urine and blood, contain significant amounts of membrane-bound nanovessicles (exosomes) that are packaged with proteins and secreted from cells (Knepper and Pisitkun 2007). Exosomes serve in inter-cellular communication, their contents vary with specific stressors and those produced by immunocytes (containing major histocompatability class I and II peptides) elicit immunological activation and even vaccination (Andre et al. 2004; Chaput et al. 2004; Keller et al. 2007; Qazi et al. 2009).

Exosomes, membrane fragments and their associated fatty acids suspended in urine or associated with cell lysis have been overlooked as possible inter-individual signals because they only separate from suspension by ultracentrifugation. Exosome-mediated immunological alterations in mice were achieved via intra-nasal administration (Prado et al. 2008). Considerable social information in both crustaceans and fish is secreted in urine (Moore et al. 1994; Zulandt Schneider and Moore 2000; Almeida et al. 2005; Horner et al. 2006) and irradiated fish induce alterations at the cellular level in unexposed bystanders (Mothersill et al. 2007). Further, mother-fetal immunological communication is mediated by exosomes in amniotic fluid (mostly fetal urine) (Keller et al. 2007). Lipid transport and packaged delivery of peptides represents a virtually unexplored inter-individual signaling paradigm.

Daphnia expresses morphological, behavioral, and life history alterations in response to signals from predators (e.g., Parejko and Dodson 1990; von Elert and Stibor 2006), crushed conspecifics (Pijanowska 1997; Pijanowska and Kowalczewski 1997; Laforsch et al.2006) and even heterospecific extracts (Laforsch et al.2006). Laforsch et al. (2006) concluded that unspecific alarm cues interact with predator-derived signals.

Stabell et al. (2003) found that thawed extracts of Daphnia did not induce defensive responses for ~12 h. They suggested that “latent alarm pheromones” were activated from Daphnia ingested by predators, perhaps via bacteria or digestion. This is similar to Wilson et al.’s (1958) observation that ants only remove aging corpses and the weak responses of our woodlice to freshly thawed conspecific bodies. Microbes certainly contribute to oxylipid alterations and signals of advanced purification. Commensal microbial associates may even produce volatile cues of genetic identity and aspects of recognition reflecting the major histocompatability complex (Lanyon et al.2007). Bacterial associates of predators may generate cues used by aquatic crustaceans to alter their vertical distribution (Forward and Rittschof 1993; McKelvey and Forward 1995; Beklioglu et al.2006). Janzen (1977) noted that microorganisms increased the free fatty acids in beans from 1.4% up to 62%, thus generating reliable signals of contamination.

Although bacterial contributions are likely, crushed bodies, alcohol extracts of frozen bodies and authentic oleic and linoleic acids were immediately repellent to woodlice. Wilson et al. (1958) also obtained immediate responses to oleic acid.

Phylogenetically conserved endogenous oxylipid chemistry involves NAD(P)H oxidases, nitric oxide synthases, cyclooxygenases, lipooxygenases, dioxygenases, phospholipases, lyases, and p450 enzymes. Many function in cell-free extracts (some associated with membrane fragments) following cell death or rupture. Derivatives of arachidonic, phosphatidic, oleic, linoleic, linolenic, and docosahexanoic acids as well as phosphatidylcholine are associated with dysregulated, senescent or dying cells. Rapid release of free unsaturated fatty acids by phospholipases may be the “deciding step” in initiating defense responses (see Pohnert 2002). Large amounts of oleic acid are enzymatically released from triglycerides following death of ants (Akino and Yamaoka 1996). Thus, endogenous mechanisms capable of rapidly generating “necromone” signals are well known and span immense phylogenies of unicells, plants, and animals.

Phosphatidylserine, dubbed the “death knell,” is the clinical biomarker of apoptosis (Fadok et al.2001). Oxylipid externalization in cell membranes of apoptotic cells (phosphatidylserine, arachidonic, and linoleic acids) allows sensing by the lipid scavenger receptor CD36 on immunocytes (Greenberg et al.2006). Significantly, multiple homologs of CD36 detect lipids on insect antennae (Nichols and Vogt 2008).

Learning and Necromones

Responses to predators may involve signals from injured conspecifics and/or associative learning of predator cues. General signals of injury, death or contagion risk require no coevolutionary history or maintenance of complex or expensive recognition systems for diverse enemies. Damselfly larvae learn predator odors associated with cues from injured conspecifics (Wisenden et al.1997) and Dukas (1998) found that Drosophila larvae learned to avoid odors paired with disturbance and crushed conspecifics. Thus, general death and injury signals can be linked to specific predators or pathogen risks given the associative learning capacity of a maggot. A focus on which chemicals of predators signal alarm may overlook an important role of necromones in the learned association.

Social Caterpillars

Our unlikely predictions that social caterpillars would avoid body fluids and unsaturated fatty acids were dramatically supported (Fig. 4, Table 1). Caterpillars avoided climbing treated stems suggested repellency rather than a palatability effect. Moreover, the body slurry undoubtedly contained trail-marking pheromone used in social recruitment to food and aggregation in tents (Peterson 1987; Fitzgerald and Peterson 1988; Fitzgerald 1995). Tent caterpillars travelled in parades around the arena floor. Applications of synthetic trail pheromone disrupts social cohesion of young caterpillars, but was not repellent (Fitzgerald 2009). Thus, repellency of our test materials appears to override such signals. Characterizing the chemistry of repellent fractions from caterpillar extracts and potential impacts on tent relocation are pending.


Shared necrophobic behaviors and chemistry by diverse insects and crustaceans suggests ancient evolution and widespread occurrence of surveillance and responses to predation and contagion. A review in progress in fact suggests relevant mechanisms may trace to unicells. Terrestrial isopods are an ideal research model based on availability, interspecific variation, ease of maintenance, and effective assay. Outstanding questions pertain to age, gender, maturity, season, life history plasticity, associative learning, aquatic relatives, tradeoffs, and responses to infection or immunological activation. Evolutionary conservation of necromones might well be expected since effective recognition of injury and death would likely be advantageous to even the most primitive of organisms. Fatty-acid necromones might be profitably considered by those working in the broadest contexts of predation, disease, pheonotypic plasticity, life history tactics, stress responses, and pest management.


This research was supported by the Natural Sciences and Engineering Research Council of Canada. Mary Ann dela Cruz, Melanie Prosser, Tara Ladd, and Cathy Woods contributed to the research effort. I thank our editor, Dr. Benedikt Hallgrimsson and two anonymous reviewers who greatly improved this paper.

Copyright information

© Springer Science+Business Media, LLC 2009