© 1999 Macmillan Magazines Ltd

Predators learn to avoid unpalatable preymore quickly when it is conspicuous

than when it is cryptic1. When they are rare,however, conspicuous individuals suffer agreater risk of discovery by predators thanthe cryptic majority, and their frequency isstill too low for predators to learn to avoidthem2. We might therefore expect selectionto favour unpalatable prey that is cryptic atlow local densities and conspicuous at highlocal densities. I have found that such den-sity-dependent warning coloration occursin Schistocerca emarginata (4lineata)(Orthoptera: Acrididae) grasshoppers, pro-viding insight into the biology of the desertlocust, Schistocerca gregaria, and a new per-spective on the evolution of warning col-oration (aposematism).

Selection on warning coloration by visu-ally hunting predators is predicted to resultin monomorphic aposematic prey3. Despitethis, polymorphisms for aposematismexist4. These polymorphic phenotypes rep-resent variations on a similar aposematictheme, rather than a difference betweencrypsis and conspicuousness. Changes fromcryptic to aposematic morphs have beenthought of as evolutionary change, notpolymorphism, but phenotypic plasticity incoloration can allow this transition to occurwithin a single generation.

Some populations of juvenile S. emar-ginata in Texas feed mainly on Ptelea trifoli-ata (Rutaceae), despite the presence ofother acceptable host plants5, causingnymphs to congregate there. S. emarginatanymphs from populations that feed on Pte-lea exhibit a density-dependent colourpolymorphism: typically, they are green atlow rearing densities, and yellow and blackat high rearing densities (Fig. 1a).

Juveniles from a Ptelea-feeding popula-tion were reared from hatching outdoors inscreen cages as isolated individuals (n428)or under crowded conditions in groups often (n431). Increased rearing density sig-nificantly increased the melanized (black)proportion of the pronotum, wing, femurand abdomen of final-instar insects(P*0.0001, t-tests on arcsine-transformeddata), as well as the frequency of yellow rel-ative to green and yellow/green backgroundcoloration (P*0.0001, Monte Carlo con-tingency table). This polymorphism wasindependent of rearing diet5.

Yellow-and-black S. emarginata wereunpalatable to insectivorous lizards (Anoliscarolinensis) when reared on Ptelea, but notwhen reared on Rubus trivialis (Rosaceae)(P*0.0001, Fisher’s exact test) (Fig. 1b).Unpalatability was mediated by gut contentand was rapidly lost when insects were fed anon-toxic plant. All 14 S. emarginata that

were reared on Ptelea but fed Romaine let-tuce for 24 hours were accepted by lizards(P*0.0001, Fisher’s exact test, comparisonwith Ptelea-only diet treatment in Fig. 1b).Lizards previously fed unpalatable yellow-and-black Ptelea-reared S. emarginatadiscriminated against yellow-and-blackgrasshoppers in favour of green grasshop-pers significantly more often than lizardspreviously fed palatable yellow-and-blackRubus-reared S. emarginata (P*0.05, Fish-er’s exact test; Fig. 1c). The yellow-and-black coloration of S. emarginata nymphs athigh local densities can therefore functionas a warning signal when the insects havebeen feeding on plants containing unpalat-able chemicals.

The desert locust, S. gregaria, alsoexhibits a density-dependent nymphalcolour polymorphism6, the biological role ofwhich remains a mystery. My results indi-cate that the colour patterns of S. gregariamay be related to specific host plants. Innon-outbreak populations, predation can bea strong mortality agent7. Nymphs eat plantscontaining compounds toxic to vertebrates8,and crowding on plants can facilitate colourchange9. Density-dependent aposematismcould reduce predation and populationextinction rates in areas where insects feedon plants that make them unpalatable topredators, and could hasten the initial stagesof population increase under favourableenvironmental conditions.

Phenotypic plasticity in coloration mayhave facilitated evolution of aposematism

in insects. Density-dependent changes fromcryptic to conspicuous phenotypes occur inseveral insect lineages10. Among organismswith the appropriate genotype, density-dependent colour polymorphism couldreduce the initial cost of conspicuousnessonce unpalatability (or some form ofunprofitability) has been achieved. The evolution of this trait would be influencedby many factors, including, but not limitedto, prey population dynamics and localpredator behaviour. Examples of density-dependent aposematism may representsnapshots of this evolutionary process inaction.Gregory A. SwordDepartment of Zoology and Brackenridge FieldLaboratory, University of Texas, Austin, Texas 78712, USAPresent address: Department of Zoology, University of Oxford, South Parks Road, Oxford OX1 3PS, UKe-mail: greg.sword@zoology.oxford.ac.uk

1. Gittleman, J. L. & Harvey, P. H. Nature 286, 149–150 (1980).

2. Mallet, J. & Singer, M. C. Biol. J. Linn. Soc. 32, 337–350 (1987).

3. Fisher, R. A. The Genetical Theory of Natural Selection 2nd edn

(Dover, New York, 1958).

4. Endler, J. A. Phil. Trans. R. Soc. Lond. B 319, 505–523 (1988).

5. Sword, G. A. Host Plant Use and Density-Dependent

Aposematism in Schistocerca (Orthoptera: Acrididae). Thesis,

Univ. Texas (1998).

6. Stower, W. J. Anti-Locust Bull. 32, 1–75 (1959).

7. Greathead, D. J. J. Appl. Ecol. 3, 239–250 (1966).

8. Bouaïchi, A., Simpson, S. J. & Roessingh, P. Physiol. Entomol.

21, 247–256 (1996).

9. Ghaout, S., Louveaux, A., Mainguet, A. M., Deschamps, M. &

Rahal, Y. J. Chem. Ecol. 17, 1499–1515 (1991).

10.Fuzeau-Braesch, S. Comp. Insect Physiol. Biochem. Pharmacol. 9,

549–589 (1985).

NATURE | VOL 397 | 21 JANUARY 1999 | www.nature.com 217

Density-dependent warningcolorationscientific correspondence

FFiigguurree 11 Avoidance ofSchistocerca emarginata byAnolis carolinensis lizards.a, The green and yellow-and-black morphs of S. emarginata. b, Ptelea-conferred unpalatability.Adult A. carolinensis naivewith respect to S. emargina-ta were controlled for moti-vational state and offeredyellow-and-black S. emar-ginata nymphs reared oneither Ptelea (n419) orRubus (n419). Consump-tion without regurgitationwas scored as acceptance.Attacking and releasing orregurgitation within 24hours was scored as rejec-tion. Each lizard was usedonce. c, Lizards discriminated against yellow-and-black in favour of green grasshoppers after experiencewith yellow-and-black Ptelea-reared grasshoppers. Lizards from the palatability assay were offered a simulta-neous-choice test between a green Chortophaga viridifasciata nymph and a similar-sized yellow-and-blackS. emarginata nymph. C. viridifasciata were used because it was not possible to rear green S. emarginata insufficient numbers. Lizards eating both insects were omitted from the analysis.

4

0

8

12

16

0

Ptelea Rubus

Inse

cts

20

Ptelea RubusLizard’s previous experience

acceptedrejected

greenyellow/black

Diet

b c

a