rc

Received 17 February 2009Initial acceptance 20 March 2009

ietye fu

hypothesis that the body-enclosing faecal case created and lived in by the immature stages of Neo-chlamisus leaf beetles reduces their risk of predation. We especially focus on the case of N. platani, which

generalist predators (crickets, soldier bugs and lynx spiders) that represent different feeding strategies

ght of5, pag

immediately after it is laid (Erber 1988; Brown & Funk 2005).Larvae later cut away the distally attened roof of this egg case,thus creating an opening from which they extend their heads and

tional structural aspects into its faecal case. Specically, faecal casesof N. platani are, to varying degrees, fuzzy in appearance because ofthe relative abundance of plant hairs (i.e. trichomes) that areattached externally and also incorporated into the faecal matrix ofcase walls (Popenoe & Marlatt 1889; Brown & Funk 2005; Chabooet al. 2008; Fig. 1c). Other species of Neochlamisus incorporatetrichomes in their cases, but rarely to the degree of N. platani(Chaboo et al. 2008). These short, stellate, nonglandular trichomesderive from the leaves of N. platanis host plant, the American

* Correspondence: D. J. Funk, Department of Biological Sciences, VanderbiltUniversity, 7264 MRB3 Biological Sciences Building, 465 21st Avenue South,Nashville, TN 37235-1634, U.S.A.

Contents lists availab

Animal Be

journal homepage: www.els

Animal Behaviour 79 (2010) 127136E-mail address: daniel.j.funk@Vanderbilt.edu (D.J. Funk).a builder to expand control over some aspect of its environment(Hansell 2000, 2005). Animals of a variety of taxa show amazingfeats of architecture in a multitude of media. One type of animalarchitecture that has only recently received attention is thebuilding of structures from an animals own waste materials(Weiss 2006). An elaborate example of such faecal architecture isprovided by the casebearing leaf beetles (Coleoptera: Chrys-omelidae: Camptosomata). Most camptosomate cases are initiallyconstructed by an ovipositing female that uses individual plates offaeces to build a sealed, bell-shaped structure around each egg

Fig. 1). Camptosomate larvae never leave their cases, but system-atically add their own faeces to enlarge them and thus accom-modate growth (see details in Brown & Funk 2005). Prior topupation, the case opening is sealed and the case is attached to thesubstrate. During this immobile stage of the life cycle, pupae areparticularly vulnerable to predation, and eviscerated pupal casesare observed in the eld (Flinte & Macedo 2004; D.J.F., personalobservation). Ultimately, the mature adult cuts itself free of itscase with its mandibles.

One camptosomate, Neochlamisus platani, incorporates addi-Keywords:building behaviourcasebearerChrysomelidaefaecesimmature insectinsectplant interactionleaf beetleNeochlamisuspredatorprey interactiontrichome

Animal architecture can be thoubehaviour to materials (Hansell 2000003-3472/$38.00 2009 The Association for the Studoi:10.1016/j.anbehav.2009.10.010were used in our individual-level repeated observation behavioural trials. Results strongly demonstratedthat the faecal case itself greatly reduced predation risk for all combinations of beetle species, life historystage and predator. Additional evidence indicated that both trichomes and attics further and indepen-dently reduced predation risk. Variation in results among treatments was also informative. For example,the capacity of faecal case components to reduce predation sometimes varied markedly among predatorsand between larval versus pupal life stages. Patterns of predator behaviour provided no evidence thatcaseless larvae have alternative means of defence. This study further presents a rare example of theco-option of a physical plant defence (trichomes) by an herbivore. 2009 The Association for the Study of Animal Behaviour. Published by Elsevier Ltd. All rights reserved.

as the application ofe 33) and often enables

legs for feeding and movement, while the rest of the body remainswithin the case (LeSage 1984; Erber 1988; Brown & Funk 2005;Available online 5 November 2009MS. number: A09-00097Rin the case apex. Here, we separately evaluated the effects of case, trichomes and attic on each of severalbehavioural stages of predator attack using N. platani and N. bimaculatus larvae and pupae. ThreeFinal acceptance 17 September 2009 is externally covered with host-plant trichomes, and includes a distinct trichome-lled chamber (attic)Antipredatory properties of an animal athwart arthropod attack

Christopher G. Brown, Daniel J. Funk*

Department of Biological Sciences, Vanderbilt University

a r t i c l e i n f o

Article history:

Animals create a wide varintriguing structure whosdy of Animal Behaviour. Publishehitecture: how complex faecal cases

of structures to deal with abiotic and biotic challenges. We evaluated annction has never been thoroughly tested. Specically, we evaluated the

le at ScienceDirect

haviour

evier .com/locate/anbehavd by Elsevier Ltd. All rights reserved.

wit(e)al c

al BeTrichome attic

External trichomes

(a) (b)

(e)

(f)

Figure 1. (ad) Typical appearance of N. platani larvae: (a) without a case, (b) in a casefrom which the trichome attic has been removed (i.e. note dark area in apex of case).component parts of a trichome attic. (g) Cross section of a trichome attic from a pup(2005).

C.G. Brown, D.J. Funk / Anim128sycamore, Platanus occidentalis. The trichomes of various plantsfunction as a physical or chemical barrier that deters smallarthropods from feeding on the plant surface (Bernays 1991;Valverde et al. 2001; Andres & Connor 2003), sometimes evenkilling these would-be herbivores (Gilbert 1971). However, someinsects are able to bypass these defences (Moran 1986; Medeiros &Moreira 2002).Neochlamisus platani, for example, is not deterred bythe plant hairs of sycamore, instead clearing paths through themwith their mandibles in order to feed (C. G. Brown & D. J. Funk,unpublished data). Neochlamisus platani performs another buildingbehaviour that has not otherwise been documented in casebearersand that is known to be absent in other Neochlamisus (Brown &Funk 2005; Chaboo et al. 2008). Specically, a cross section of theapex of pupal cases reveals an extra internal compartment, or attic,that is lled with trichomes (Brown & Funk 2005; Fig. 1eg). Thiscompartment sits above the main chamber that houses the pupa,from which it is separated by a thin layer of faecal material. Themethod by which this attic is constructed is not known.

Camptosomate faecal cases represent highly integrated andphysically robust structures. As a building material, these faecesshare advantages with other secreted substances, requiring rela-tively low production or collecting costs, as all organisms mustproduce waste, and providing uniform and malleable buildingmaterials (Olmstead & Denno 1992; Hansell 2005). Additionally,Neochlamisus faeces include undigested trichomes (C.G.B., personalobservation). Such composite materials, as in reinforced concrete,better resist breaking under tension and compression (Hansell2005). Faeces may also show the same thixotropic properties asmud, a material used by many animal builders. Indeed, faeces areadvantageously used for a diversity of purposes (e.g. Seymour 1974;Weaver et al. 1989; Anbutsu & Togashi 2002; Grasso et al. 2005;Borg-Karlson et al. 2006; Weiss 2006).External trichomes

Outer layer offaecal material

Trichome attic

Inner attic wall

Inner chamberwith larva

(c) (d)

(g)

h few external trichomes, (c) in a case with dense external trichomes and (d) in a caseDiagrammatic representation of a larva in a case with a trichome attic. (f) Diagram ofase. Parts of this gure have been modied from Figures 4 and 7 in Brown & Funk

haviour 79 (2010) 127136None the less, building, maintaining and carrying a casethroughout larval development, as inNeochlamisus, is itself likely tobe time consuming and energetically costly (e.g. Stevens et al. 1999;Venner et al. 2003; McKie 2004; Hansell 2005). In addition, femalesare vulnerable during the lengthy (half-hour; Brown & Funk2005) period required for the construction of each of the dozens ofindividual egg cases that they construct (Funk 1998). Most gener-ally, faeces are typically discarded by insects since they can stim-ulate growth of harmful fungi and bacteria and attract predatorsand parasites (Muller & Hilker 1999; Weiss 2003, 2006). The exis-tence of the casebearing habit despite such costs hints at the like-lihood of compensating benets, yet we know of only three priorstudies, each modest in scale, that have investigated possiblebenecial functions for camptosomate faecal cases (Wallace 1970;Root & Messina 1983; Flinte & Macedo 2004). Thus, the potentialtness advantages underlying the evolutionary origins and main-tenance of this complex behaviour remain almost entirely unex-plored. Animal constructions often aid in protection (from bothbiotic and abiotic factors), prey capture and communication (Han-sell 2005). For example, animal architectures can protect buildersor their young from attack by reducing detection, as in birds thatadd lichen to their nest to disrupt its outline (Hansell 1996), or byreducing the probability of capture after detection, as in thefunnelled burrow entrances of Paralastor wasps (Smith 1978).Notably, several lineages of non-casebearing leaf beetles are knownto retain faecal material as protective, if less complex, coverings fortheir eggs and larvae (Olmstead 1994; Muller & Hilker 2003, 2004;Chaboo et al. 2008). Faecal shields, for example, present physically(Eisner et al. 1967; Olmstead & Denno 1993; Eisner & Eisner 2000;Nogueira-de-Sa & Trigo 2002) or chemically (Morton & Vencl 1998;Vencl & Morton 1998; Muller & Hilker 1999, 2003) repellentbarriers to predators.

The faecal cases of casebearers such as Neochlamisus, as well asthe external trichomes and trichome attics of N. platani, mightsimilarly protect their builders from arthropod predators, reducingaccess to the otherwise ill-protected immatures, whose soft bodieshave fewer hairs, spines and hard sclerotized plates than otherjuvenile leaf beetles (Moldenke 1971; LeSage 1982; Root & Messina1983). Since trichomes protect plants from herbivorous arthropods,incorporating them into faecal cases may similarly protect N. pla-tani from predaceous arthropods. The trichome attic might furtherincrease physical protection by presenting predators witha concentration of these trichomes and by increasing the faecalbarriers that must be penetrated to gain access to the immatureanimal through the case apex. Field-collected pupal cases of variousNeochlamisus taxa show damage to this particular region of thecase, suggesting that it is indeed a site where predators seek entry

repeated observation experiments to quantify the effects of these

Nashville, TN. Ovipositing adults and developing immatures weremaintained in these containers to provide test animals for ourexperiments. Tests involved two life stages that are similar in size,but differ in mobility and case completion: last-instar larvae, whosehead and legs can emerge from an opening in the case, thusenabling feeding and walking, and immobile pupating animals incases cemented to the substrate such that their openings arecompletely sealed (Fig. 1). Neochlamisus platani and N. bimaculatuscases differ primarily in that the former shows variable butconsiderable amounts of external trichomes and the trichome attic,while the latter does not. These closely related species commonlylive together in the same habitats and thus may experience similarselective pressures and predators.

Since little is known about the natural predators of casebearers(Karren 1964; Cox 1996), three generalist predators of insects,

into prey, then consume it. These were collected from Rubus plants

at

C.G. Brown, D.J. Funk / Animal Behaviour 79 (2010) 127136 129architectural traits on the threat posed by three disparatearthropod predators to active larvae and immobile pupae ofNeochlamisus.

METHODS

Study Animals

Neochlamisus platani and N. bimaculatus are univoltine leafbeetles, about 4 mm, that develop and live entirely on one hostplant, American sycamore and blackberry (Rubus spp.), respec-tively, across the American southeast (Karren 1972). Eggs, larvaeand adults of these species were collected from host plants in andaround Davidson County, TN, U.S.A., during the springs of 2006 and2007 for our studies. These animals were placed on host foliage inplastic boxes lined with moistened paper towels in an incubatorat 25 C and a 14:10 h light:dark cycle at Vanderbilt University,

N. bimaculatus N. platani

Larvae Pupae Larvae Pupae

-c +c -c +c-c

-t-a -t+a

+t-a

+t+a -c -t -t+a-a

+-a+t

+

Predator:

Beetle species:

Beetle stage:

Treatment:

Crickets(D.J.F., personal observation).Camptosomate faecal cases have long intrigued biologists, and

a protective function is often assumed (e.g. Riley 1874; Popenoe &Marlatt 1889; Cockerell 1891; Donisthorpe 1902). Here, we aimedto determine whether Neochlamisus casebearers benet from theirunusual building behaviours by testing the long-standing hypoth-esis that their cases reduce the success of predator attack.Furthermore, we separately evaluated the contributions of faecalcases per se, trichome incorporation and the trichome attic toantipredator defence in N. platani. As we had insufcient infor-mation to rigorously evaluate whether these three aspects repre-sent adaptations, let alone evolutionarily independent adaptations,we focused instead on evaluating their current functional conse-quences for predation resistance. Therefore, we used manipulativeFigure 2. Basic structure of the experimental design for the repeated observation trials.a attic. See text for further details.at one of the sites where N. bimaculatus were also collected for ourstudy. Spiders were maintained individually on mealworms. Theavailability of N. platani larvae and pupae and of N. bimaculatuspupae was low when spiders were collected, preventing our eval-uation of these predatorprey combinations.

Experimental Treatments

Except as noted above, individuals of each predator were pairedwith an individual N. bimaculatus larva or pupa representing one oftwo experimental treatments: (1) larva/pupa removed from itsfaecal case (i.e. caseless), or (2) larva/pupa in its unmanipulatedcase (which, again, naturally possesses no external trichomes orattic). By contrast, N. platanis more complex case architectureencouraged the evaluation of additional treatments: (1) caseless,(2) both trichomes and attic removed, (3) trichomes removed butattic intact, (4) trichomes intact but attic removed and (5) unma-nipulated casewith trichomes and attic intact (Fig. 2). Several of our

N. bimaculatus N. platani

Larvae Pupae Larvae Pupae

-c +c -c +c-c

-t-a -t+a

+t-a

+t+a -c -t -t+a-a

+a-a+t

+t

Soldier bugs Lynx spiders

N. bimaculatus

Larvae

-c +crepresenting the variety of predatory taxa and feeding morphol-ogies present in the beetles habitats, were chosen for our experi-ments: (1) the common house cricket, Acheta domestica, is a largebiting/chewing insect with robust mandibles. These insects alsofeed on vegetation, and after purchase, were maintained on breadand potatoes. Although this cricket species does not normally occurin the habitats of our study beetles, other cricket species commonlydo and represent likely predators. (2) The spined soldier bug,Podisus maculiventris, is a native hemipteran that uses its beaklikepiercing/sucking mouthparts to penetrate the cuticle of prey andconsume their uids. These were purchased as nymphs and rearedto maturity on mealworms. (3) Green lynx spiders, Peucetia sp., aresit-and-wait predators with chelicerae that inject digestive uidsc caseless; c cased; t no trichomes; t with trichomes; a no attic;

al Betreatments are not encountered in nature. Nevertheless, the phys-ical manipulations and subsequent treatment comparisonsdescribed here represent a long tradition of using these approachesto isolate the functional contributions of particular traits to insectbehaviour (Bernays & Chapman 1994; Chapman 1998).

All test animals were removed from their rearing containers anddivided into treatments on the day their particular trials wereconducted. Treatments were assigned in a manner designed toavoid potential bias. For caseless treatments, cases were cut awayfrom living larvae and pupae with forceps and a razor blade. Theseanimals are not physically attached to their cases and were notharmed during case removal. For trichome-free treatments, theexternal layer of trichomes was gently removed with a scalpel(Fig.1b). For attic-free treatments, the outer wall of the attic and thetrichomes inside were removed with a scalpel and forceps,respectively, while the inner faecal layer separating the attic fromthe encased animal was left intact (Fig. 1d). Unmanipulated cases(Fig. 1c) were rotated between the ngers for several seconds tosimulate and control for the physical manipulation of other treat-ments. All individuals were handled while wearing latex gloves toavoid exposure to human-derived substances. Following their usehere, test animals were frozen for potential molecular work, or, ifdead, discarded with other nontoxic biological materials.

Experimental Design and Protocols

All trials took place in awalk-in environmental chamber at 24 Cand more than 80% relative humidity. Each trial arena consisted oftwo 10 cm petri dish bottoms taped together to form an enclosedchamber with adequate room for larvae and predators to movenaturally. This had a oor of paper towel cut to t the dish, towhich300 ml of water was added to prevent desiccation. All predatorswere starved for at least 48 h prior to use in a trial. Each predatorwas acclimated to its arena for 2 h prior to testing. Each cricket wasused in only a single trial. Each soldier bugwas used in either one ortwo trials. To avoid systematic effects of previous experience, bugindividuals used in multiple trials were distributed across preyspecies and treatments in an unbiased fashion and rested anaverage of 12 days between trials. Most spiders were used in onetrial but a few were used in two, with 24 days between trials. Nolarva or pupa was used in more than one trial. A maximum of 30arenas was observed at a time.

Repeated Observation Trials

To begin each trial period, one larva/pupa of a given experi-mental treatment was placed in the centre of the arena with itsresident predator (Fig. 2). Observations were then recorded every5 min for 2 h by point sampling. During each observation, the levelof threat posed to the larva/pupa by the predator was scored ona scale of 15 as follows: 1 predator ignoring larva/pupa;2 predator approaching larva/pupa; 3 predator investigatinglarva/pupa (i.e. touching it with legs, antennae, or mouthparts);4 predator attacking larva/pupa (i.e. biting or piercing it); and5 predator killed larva/pupa. When an animal was killed prior tothe conclusion of the trial, a value of ve was entered for theremaining observations, thus accounting for the rapidity withwhich death occurred in evaluating threat level (see below).Sample sizes are provided in tables (see Results).

Attic/No-attic Repeated Observation Choice Trials

In the hope of more powerfully teasing out effects of thetrichome attic in reducing predation threat, we conducted choice

C.G. Brown, D.J. Funk / Anim130tests using crickets and pairs of pupal N. platani (N 42 pairs).White craft glue was used to attach two cased pupae in theirnatural orientation, near the centre of a lter paper placed on thebottom of the arena. For one pupa in each pair, the trichome atticwas intact and for the other it was removed. External trichomeswere removed from both pupal cases to better isolate attic effects.We scored threat levels to each pupa from a cricket in the arenaevery 5 min for 2 h as above.

Statistical Analyses

Several kinds of information were extracted and analysed fromthe data collected in our repeated observation experiments. First,the 24 threat levels (representing the 24 observations) recordedacross each 2 h trial were averaged to obtain the mean threat levelexperienced by a given test animal during the trial. Second, wequantied the proportion of test animals that were investigated,attacked and killed by predators during the trials. Third, wedetermined the time between behaviours (described above) thatwere associated with successively higher threat levels. Specically,the time between trial initiation and initial investigation of the testanimal by the predator was interpreted as reecting its long-distance attraction to, or recognition of, the larva/pupa. Likewise,the time between initial investigation and predator attack wasinterpreted in terms of the degree of predator interest/motivationfollowing prey contact. Last, the time to death after initial attackwas evaluated as the speed or handling time required by a pred-ator to dispatch its prey. Total time to death and number of attackssurvived by the larva/pupa provided additional data on predatorresistance.

All analyses used nonparametric methods because of deviationsof the data from normality. We used KruskalWallis test to evaluatedifferences among treatments in continuous variables (Dunn 1964;Zar 1996). Proportions were analysed using G tests to evaluategoodness of t. For the attic/no-attic choice test, we performeda Wilcoxons signed-ranks test for two groups, with the dataarranged as paired observations (Sokal & Rohlf 1995). Since wea priori hypothesized that each faecal case trait (i.e. case, trichomes,attic) would independently lower predation risk, all testscomparing two treatments (one with and one without the trait)were one tailed. Means are presented SE. Trends refer to talliedresults across our analyses, with each such result reduced to a yesor no depending on whether comparing absolute treatmentmeans indicated greater or lesser resistance to predation, respec-tively. These trends revealed general patterns with respect toparticular biological factors (e.g. in X of Y analyses, cases withtrichomes afforded more resistance to predation than thosewithout, illustrating a trend). We did not statistically analyse thesetrends because of issues of nonindependence. Rather, we providea summary of data patterns across the many biological axes (i.e.beetle species, immature life history stage, predator type, case trait)and analyses of our study. Findings described below that do notrefer to a trend represent statistically signicant results.

RESULTS

General Patterns

Our analyses indicated important contributions of cases,trichomes and attics to predator resistance (Table 1, Fig. 3,Appendix). Here we focus on major results and general trends thatillustrate this support (Table 2). Critically, all three predators readilyate both species and life stages of caseless Neochlamisus, indicatingthe acceptability of caseless Neochlamisus as prey and their lack ofeffective case-independent defences, and the suitability of our

haviour 79 (2010) 127136choice of predators for these experiments. Complementarily, the

a fun

%

111

1

al BeTable 1Percentages of immatures that experienced varying levels of threat from predators asanalyses combined data from larvae and pupae)

Predator/beetle stage N % Investigated

G PN. bimaculatusCaseCricketsLarvae 10 10 100 70 4.70 0.03Pupae 10 10 100 90 1.44 0.23Pooled 20 20 100 80 5.99 0.07

Soldier bugsLarvae 4 4 100 75 1.53 0.22Pupae 4 4 75 50 0.54 0.46Pooled 8 8 88 62 1.38 0.12

Lynx spiders

C.G. Brown, D.J. Funk / Animconsiderable effectiveness of the case as an antipredatory structureis illustrated by the near uniformity with which absolute predationthreat was lowered in cased compared to caseless treatments.Despite sometimesmodest sample sizes, this trendwas observed in76 of 77 analyses across all study species, life stages and predators.

Further analyses of N. platani detected no signicant differencesin threat level between the four cased treatments. (These analysesused two-tailed tests in the absence of a priori hypotheses on therelative antipredatory contributions of different case traits.)Therefore, we pooled related treatments (e.g. those for which atticswere present) and analysed the following comparisons: (1) casedversus caseless animals, (2) external trichomes present versusremoved and (3) attic present versus removed. Doing so revealeda trend whereby trichome-bearing animals appeared highly resis-tant to crickets (15/16 comparisons) but were susceptible to soldierbugs (2/13). Another pair of contrary trends was presented by thepresence of attics in larvae, which improved resistance in only 6 of

Larvae 12 11 83 82 0.01 0.92

N. plataniCaseCricketsLarvae 17 64 100 88 4.00 0.05Pupae 15 60 93 83 1.12 0.29Pooled 32 124 97 86 4.00 0.02

Soldier bugsLarvae 13 52 100 87 3.33 0.07 1Pupae 4 16 75 25 3.40 0.07Pooled 17 68 94 72 4.58 0.02

TrichomesCricketsLarvae 32 32 90 84 0.58 0.45Pupae 30 30 83 83 0.00 1.00Pooled 62 62 87 83 0.26 0.30

Soldier bugsLarvae 26 26 88 85 0.17 0.68Pupae 8 8 25 25 0.00 1.00Pooled 34 34 74 71 0.07 0.39

AtticsCricketsLarvae 32 32 78 97 5.71 0.98Pupae 30 30 93 73 4.58 0.03Pooled 62 62 86 86 0.00 1.00

Soldier bugsLarvae 26 26 85 89 0.17 0.32Pupae 8 8 38 13 1.38 0.24Pooled 34 34 78 71 0.07 0.40

Differences in percentages in the absence () versus presence () of each trait were calcnonsignicant differences (0.10 P > 0.05) are italicized. For these data, effectiveness ocolumn than in the column. Level of threat from predator: investigation < attack < dction of beetle species, case-associated trait, predator and life history stage (pooled

Attacked % Killed

G P G P

00 40 10.97 0.0009 100 10 21.02

ldie

al BeCrickets So

N. bimaculatus

Larvae Pupae

* * *5

4

3

2

1

Cas

e

at

+ + +Larvae

N. platani

C.G. Brown, D.J. Funk / Anim132means of evaluating predation threat and prey resistance (Table 1,Fig. 3, Appendix).

Cased N. bimaculatus larvae were less frequently investigated bycrickets. In most analyses, cases reduced attack frequency and preymortality, lowered overall threat and increased survival time.Indeed, no soldier bugs or spiders killed cased individuals. Cricketsand soldier bugs took longer to investigate cased larvae. Cricketsattacked caseless larvae and pupae almost immediately afterinvestigation, but took much longer to attack cased larvae/pupaeand longer to kill cased pupae. Cases also increased soldier bughandling time and the total survival time of larvae tested withthem. With respect to both predators and both larvae and pupae,N. platani cases lowered the frequency of investigation and thelikelihood of being attacked and killed, while reducing overallthreat and increasing time to attack and total survival time. Cricketsand soldier bugs took longer to investigate cased larvae and pupae,respectively. Cases also increased predator handling time forcrickets and soldier bugs attacking pupae and larvae, respectively.

*

**

5

4

3

2

1

5

4

3

2

1

NS NS

NS

Cri

cket

sSo

ldie

r bu

gs

Mea

n t

hre

+ + +

Larvae Pupae Larvae

Case Tr

Figure 3. Because of variation in the suite of treatments evaluated for the two beetlea symmetrical gure. Neochlamisus bimaculatus (rst row) lacks trichome and attic case trarows) possesses all case traits but was not tested against spiders. This gure depicts the meover the course of each repeated observation trial. These values are plotted as a function odifferences in the mean threat level ( SE) experienced by individual larvae lacking () versutests (H). For all comparisons df 1. Signicant differences are indicated with asterisks. Aindicated by a lower value for the bar than for the bar.r bugs Spiders

*NS

+ + +Pupae Larvae Pupae

Pupal datanot taken

haviour 79 (2010) 127136Cased larvae also escapedmore attacks per trial when facing soldierbugs. External trichomes signicantly beneted pupal but notlarval N. platani when exposed to crickets. Trichomes lowered theproportion of pupae killed and the mean threat, and increased thetime between investigation and attack as well as survival time.Attics lowered investigation frequency and mean threat for pupaefacing crickets. Larvae with attics also required longer handlingtimes and escaped more often from crickets.

DISCUSSION

This study demonstrates the success of a complex and atypicalexample of animal architecture, the faecal case of camptosomateleaf beetles, at resisting predation. It does so for each of two testedNeochlamisus species and both larval and pupal stages against threetrophically disparate arthropod predators. While evidence for theantipredatory properties of the case itself proved compelling, wefurther demonstrate the contributions of external trichomes and

* *NS

NS NS NS

+ + +

Pupae Larvae Pupae

ichomes Attic

species, the plots in this gure are organized differently for each species to createits, but was tested against all three predators. Neochlamisus platani (second and thirdan predation threat level (see text) experienced by individual immatures as measuredf beetle species, predator species, case-associated trait and life history stage. Potentials possessing () a given case trait were evaluated using KruskalWallis nonparametric treduction in predation threat associated with the possession of a given case trait is

contrast, as faecal cases can easily be repaired (Brown & Funk2005), a faecal means of storing chemicals would allow an indi-

al Bethe trichome attic in the more complex faecal case of N. platani.We thus provide the initial documentation of a specic function forthese immature insect constructions. Our study system furtheroffers a rare example whereby physical (rather than chemical)antiherbivore plant structures have been co-opted for the herbi-Table 2Number of analyses (summed across Table 1, Fig. 3, Appendix) in which a positive(Yes) versus negative (No) effect on resistance to predation was observed for eachcase-associated trait as a function of beetle species, predator, case-associated traitand life history stage

Beetle stage/treatment Crickets Soldierbugs

Lynxspiders

Sum

No Yes No Yes No Yes No Yes

LarvaeCasesN. bimaculatus 0 9 0 9 0 7 0 25N. platani 0 9 0 9 0 18

Trichomes 1 8 7 2 8 10Attics 6 3 5 3 11 6

PupaeCasesN. bimaculatus 1 8 0 7 1 15N. platani 0 9 0 9 0 18

Trichomes 0 7 4 0 4 7Attics 1 8 2 4 3 12

Sum of above:Cases 1 35 0 34 0 7 1 76Trichomes 1 15 11 2 12 17Attics 7 11 7 7 14 18

Results from each analysis were scored as yes or no irrespective of statisticalsignicance. Comparisons involving equivalent means (i.e. ties) are not included.These tallies are provided to illustrate general tendencies that might not beapparent from individual statistical analyses based on modest sample sizes; theyfurther facilitate inspection of patterns associated with particular biological factors.

C.G. Brown, D.J. Funk / Animvores own defence.

Faecal Cases

Primary defences lower rates of detection and attack by pred-ators (Ruxton et al. 2004). Neochlamisus faecal cases appeared toprovide a primary defence against crickets and soldier bugs, whichoften investigated cased larvae/pupae less frequently and lessquickly than they did caseless individuals. Here, this primarydefence probably represents a visual and olfactory masqueradesince cases resemble common, yet inedible, components of theenvironment (Ruxton et al. 2004). Although some predators usefaeces to locate prey (Muller & Hilker 2004; Weiss 2006), it seemsunlikely that generalist predators would expect prey to be withinthe faeces itself. Indeed, even after investigating prey, predators inour study often delayed much longer between investigating andattacking those with cases, thus increasing opportunities for preyescape. In nature, faecal cases may play yet larger and additionalroles in primary defence than indicated by our studies, since oursmall petri dish test arena articially increased the frequency ofphysical encounters between predators and cased prey and did soin an unnatural environment. Perhaps, for example, faecal casestructure reects in part their having evolved to masquerade asbuds or other host-plant-associated structures (Briggs 1905; D.J.F.,personal observation).

Again bearing in mind arena space constraints, we note that thegreat majority of test larvae/pupae were eventually investigated bypredators, offering the opportunity to evaluate faecal cases aspotential secondary defences, that is, those that reduce attacksuccess after detection (Ruxton et al. 2004). Consistent with thisviduals case (rather than the individual itself) to be sampled bya predator without permanent harm. Finally, caseless larvae werereadily eaten by all predators, consistent with (although notdemonstrating) a lack of case-independent chemical defences. Thechemistry of camptosomate faecal cases and larvae is presentlyunknown and in need of investigation.

Likewise, behavioural components could contribute to anti-predator case functionality. For example, larvae were commonlyobserved to pull their case down ush to the substrate in responseto predators, restricting access to the animal inside. This behaviourseemed particularly suited to defence against soldier bugs, whosebeaks could not penetrate the case and which only killed larvae viaentry through the case opening. Additionally, some threatenedlarvae rapidly wiggled the apex of their cases back and forth,a behaviour that could function to dislodge or startle a predator,and which was anecdotally observed to be followed by cricketretreats.

External Trichomes

The external trichomes on N. platani cases further increaseddefence against the biting/chewing crickets, predominately inpupae. Many plants use trichomes to protect their leaf surfacesfrom herbivory by irritating and/or limiting movement of smallherbivores (Levin 1973). Some herbivores that can survive onpubescent plants indirectly benet from trichomes that also inhibittheir arthropod predators (Roda et al. 2000; Mulatu et al. 2004;Guershon & Gerling 2006). The case-associated trichomes studiedhere provided similar, but more direct, protection. Crickets wereless likely to attack and kill larvae/pupae with trichome-coveredcases, whether because the trichomes limited predator recognitionof a food source, were irritating, or presented a physical barrier.Trichomes were much less effective against soldier bugs, whosebeaks lack the palps (mouthparts) that often bear sensitive chemo-and mechanoreceptors in other insects (Chapman 1998). Thus,perhaps the bugs suffer less trichome-induced irritation. Trichomesdid not increase crickets or soldier bugs time to investigation. Wesuspect, however, that trichomes might indeed reduce casedetection in the natural environment of a trichome-covered syca-more leaf, where mimicry of plant structures or camouage maypossibility, the cases of both beetle species greatly decreased thelikelihood of attack and of mortality in a large fraction of experi-ments, while also increasing predator handling time, as illustratedby the time required for crickets to chew through these cases toreach the immature inside. Cases thus clearly present successfulsecondary defences to predators, while also reducing the energeticbenet to predator attack and increasing opportunities for larvalescape.

Although not investigated here, the secondary defence providedby Neochlamisus faecal cases could have a chemical component.Wet faecal coverings in some other leaf beetles protect larvae fromants via chemicals derived from the host plant rather than thefaeces itself (Morton & Vencl 1998; Vencl & Morton 1998; Venclet al. 1999). Likewise, chemical defence may represent the functionof the unusual faecal chains of Adelpha lepidopteran larvae, whichrest at the distal ends of their frass rods when not feeding (Aiello &Solis 2003). Storing such substances in externally maintainedfaeces reduces the need for anatomical and physiological mecha-nisms of sequestration (Hilker 1992). Additionally, many conven-tionally chemically sequestering species must be wounded beforetheir defensive compounds are secreted (Olmstead 1994). By

haviour 79 (2010) 127136 133come into play.

Unelius, C. R. 2006. Antifeedants in the feces of the pine weevil Hylobiusabietis: identication and biological activity. Journal of Chemical Ecology, 32,

anemone Calliactis tricolor (Le-Sueur) in protecting hermit crabs from

al BeAlthough co-opting (sequestering) chemicals from host plantsas a source of defence may be common in insects (Eisner 2005),our study has identied a rare example of an insect herbivore thatco-opts physical attributes of host defence for its own protection.We could nd no examples that t these criteria. Those that cameclosest include an herbivorous amphipod that builds a protectiveshelter from its toxic algal host (Hay & Duffy 1990) and bagwormsthat use whole leaf clippings (rather than the specic physicaldefences of the host) in making their bags (Rivers et al. 2002).More generally, although various animal taxa protect themselvesby attaching parts of other organisms to their bodies (Millott1955; Brooks 1988; Stachowicz & Hay 1999; Thanh et al. 2003;Boyero & Barnard 2004), considerably fewer actually collect partsof their food source for this purpose (e.g. Eisner et al. 1978; Sta-chowicz & Hay 2000; Brandt & Mahsberg 2002), among themnudibranchs that acquire and defend themselves with undis-charged stinging nematocysts from their cnidarian prey (Frick2003). Coincidentally, a juvenile lacewing also gathers andattaches sycamore trichomes to itself, for protection from bugswith beaks too short to penetrate the trichome covering (Eisneret al. 2002). In this example, however, the insect is a predatorrather than an herbivore, so its trichomes do not derive froma co-opted food source. Thus, beetle and lacewing may playopposing roles in the coevolutionary dynamics of sycamores andtheir insect associates.

Trichome Attic

One of our most novel ndings was that the species-specic,trichome-lled chamber in the case apex of N. platani alsocontributed importantly to predation resistance. However, larvaedid not appear to receive the same attic-associated protection thatpupae did. Although this has not been formally investigated, itseems likely that attics are not created until the case is prepared forpupation during the last larval instar (the stage evaluated here).Otherwise, this complex structure would require repeatedrebuilding as the case is regularly enlarged to accommodatecontinual larval growth (Brown & Funk 2005). If so, attics mayspecically function to protect the immobile pupal stage. That said,the mature larvae in our studies did sometimes benet from theirattics in a manner that pupae could not. Specically, crickets weresometimes observed chewing on a removed attic while the larvaescaped its attacker. In this way, the attic was reminiscent of a liz-ards tail that can be autonomously shed to divert attention whilethe lizard escapes.

Conclusion and Future Directions

In summary, through controlled observational experiments andassociated treatments, our study shows that the faecal cases ofNeochlamisus larvae and pupae are sufcient to signicantly lowerthe threat from functionally and taxonomically varied arthropodpredators. In addition, the extra architectural componentsspecic to N. platani cases (abundant external trichomes andtrichome attics) further this protective function. These results areconsistent with the possibility that predation resistance contrib-uted to the evolution of this behavioural architectural trait, andpresently promotes its selective maintenance. Indeed, clearexamples of fully developed camptosomate cases have been foundin 45 million-year-old amber (Poinar 1999; Grimaldi & Engel2005), indicating that they have long served this or other func-tions for these beetles. Work on the phylogenetics of camptoso-mates is needed to better understand the evolution of their cases.Other areas of faecal case biology in need of study include

C.G. Brown, D.J. Funk / Anim134potential trade-offs between their antipredatory function and theoctopus predators. Journal of Experimental Marine Biology and Ecology, 116,1521.

Chaboo, C. S., Brown, C. G. & Funk, D. J. 2008. Faecal case architecture in thegibbosus species group of Neochlamisus Karren 1972 (Coleoptera: Chrys-omelidae: Cryptocephalinae: Chlamisini). Zoological Journal of the LinneanSociety, 152, 315351.

Chapman, R. F. 1998. The Insects: Structure and Function. Cambridge: CambridgeUniversity Press.

Cockerell, T. D. A. 1891. Case-making coleopterous larvae. Entomologists MonthlyMagazine, 2, 190191.

Cox, M. L. 1996. Insect predators of Chrysomelidae. In: P. H. A. Jolivet & M. L. Cox),pp. 2392. Amsterdam: SPB.

Donisthorpe, H. J. K. 1902. The life history of Clythra quadripunctata, L. Transactionsof the Entomological Society of London, 50, 1124.943957.Boyero, L. & Barnard, P. C. 2004. A Potamophylax larva (Trichoptera: Limnephili-

dae) using other caddisy cases to construct its own case. Journal of NaturalHistory, 38, 12971301.

Brandt, M. & Mahsberg, D. 2002. Bugs with a backpack: the function of nymphalcamouage in the West African assassin bugs Paredocla and Acanthaspis spp.Animal Behaviour, 63, 277284.

Briggs, E. M. 1905. The life history of case bearers: 1. Chlamys plicata. Cold SpringHarbor Monographs, 4, 212.

Brown, C. G. & Funk, D. J. 2005. Aspects of the natural history of Neochlamisus(Coleoptera: Chrysomelidae): fecal case-associated life history and behavior,with a method for studying insect constructions. Annals of the EntomologicalSociety of America, 98, 711725.

Brown, W. J. 1943. The Canadian species of Exema and Arthrochlamys (Coleoptera:Chrysomelidae). Canadian Entomologist, 75, 119131.

Brooks, W. R. 1988. The inuence of the location and abundance of the sea-similar (although counterproductive) protection afforded thelarvae of any parasitoid wasps able to circumvent the cases andoviposit on the immature beetles inside. Indeed, high rates ofparasitism are sometimes observed in Neochlamisus (Brown 1943;Neal 1989; D.J.F., personal observation). Taxonomic, geographicaland habitat-associated variation in case form and function (e.g.trichome incorporation) provide other potentially informativeavenues for research. Clearly, the present study should only be thestart of rigorous investigations on this fanciful faecal trait and theanimals that live within.

Acknowledgments

We thank April Brown, Candace Gay, Mark Mandel, MarkChapman and Jennell Talley for their help in rearing beetles andsetting up experiments. We also thank Scott Egan, Dan Duran andManuel Leal for their input on this study. Finally, we are grateful totheMetropolitan Parks and Recreation Department of Nashville, TNfor their permission to collect on parkland. This work was fundedby a Student Research Grant from the Animal Behaviour Society andan Exploration Grant from the Explorers Club to C.G.B., as well asNational Science Foundation grant IOB 0616135 to D.J.F.

References

Aiello, A. & Solis, M. A. 2003. Defense mechanisms in Pyralidae and Choreutidae:fecal stalactites and escape holes, with remarks about cocoons, camouage andaposematism. Journal of the Lepidopterists Society, 57, 168175.

Anbutsu, H. & Togashi, K. 2002. Oviposition deterrence associated with larval frassof the Japanese pine sawyer, Monochamus alternatus (Coleoptera: Ceram-bycidae). Journal of Insect Physiology, 48, 459465.

Andres, M. R. & Connor, E. F. 2003. The community-wide and guild-specic effectsof pubescence on the folivorous insects of mananitas Arctostaphylos spp.Ecological Entomology, 28, 383396.

Bernays, E. A. 1991. Evolution of insect morphology in relation to plants.Philosophical Transactions of the Royal Society of London, Series B, 333,257264.

Bernays, E. A. & Chapman, R. F. 1994. Host-Plant Selection by Phytophagous Insects.New York: Springer.

Borg-Karlson, A.-K., Nordlander, G., Mudalige, A., Nordenhem, H. &

haviour 79 (2010) 127136Dunn, O. J. 1964. Multiple contrasts using rank sums. Technometrics, 6, 241252.Eisner, T. 2005. For the Love of Insects Cambridge. Massachusetts: Belknap Press.

Eisner, T. & Eisner, M. 2000. Defensive use of a fecal thatch by a beetle larva(Hemisphaerota cyanea). Proceedings of the National Academy of Sciences, U.S.A,97, 26322636.

Muller, C. & Hilker, M. 2004. Ecological relevance of faecal matter in Chrys-omelidae. In: New Developments in Biology of the Chrysomelidae (Ed. by P. Jolivet,J. Santiago-Blay & M. Schmitt), pp. 693705. The Hague: Kugler.

C.G. Brown, D.J. Funk / Animal Behaviour 79 (2010) 127136 135Eisner, T., Vantasse, E. & Carrel, J. E. 1967. Defensive use of a fecal shield by a beetlelarva. Science, 158, 14711473.

Eisner, T., Hicks, K., Eisner, M. & Robson, D. S. 1978. Wolf-in-sheeps-clothingstrategy of a predaceous insect larva. Science, 199, 790794.

Eisner, T., Carrel, J. E., Van Tassel, E., Hoebeke, E. R. & Eisner, M. 2002.Construction of a defensive trash packet from sycamore leaf trichomes bya chrysopid larva (Neuroptera: Chrysopidae). Proceedings of the EntomologicalSociety of Washington, 104, 437446.

Erber, D. 1988. Biology of Camptosomata Clytrinae - Cryptocephalinae - Chlamisi-nae -Lamprosomatinae. In: Biology of Chrysomelidae (Ed. by P. Joviet,E. Petitpierre & T. H. Hsiao), pp. 513552. Boston: Kluwer Academic.

Flinte, V. & Macedo, M. V. 2004. Population ecology of Fulcidax monstrosa (Chla-misinae). In: New Developments in the Biology of the Chrysomelidae (Ed. byP. Jolivet, J. Santiago-Blay & M. Schmitt), pp. 623631. The Hague: Kugler.

Frick, K. 2003. Response in nematocyst uptake by the nudibranch Flabellina ver-rucosa to the presence of various predators in the southern gulf of Maine.Biological Bulletin, 205, 367376.

Funk, D. J. 1998. Isolating a role for natural selection in speciation: host adaptationand sexual isolation in Neochlamisus bebbianae leaf beetles. Evolution, 52,17441759.

Gilbert, L. E. 1971. Butteryplant coevolution: has Passiora adenopoda won theselection race with heliconiine butteries? Science, 172, 585.

Grasso, D. A., Sledge, M. F., Le Moli, F., Mori, A. & Turillazzi, S. 2005. Nest-areamarking with faeces: a chemical signature that allows colony-level recognitionin seed harvesting ants (Hymenoptera, Formicidae). Insectes Sociaux, 52,3644.

Grimaldi, D. & Engel, M. S. 2005. Evolution of the Insects. Cambridge: CambridgeUniversity Press.

Guershon, M. & Gerling, D. 2006. Effects of plant and prey characteristics on thepredatory behaviour of Delphastus catalinae. Entomologia Experimentalis etApplicata, 121, 1521.

Hansell, M. H. 1996. The function of lichen akes and white spider cocoons on theouter surface of birds nests. Journal of Natural History, 30, 303311.

Hansell, M. H. 2000. Bird Nests and Construction Behaviour. Cambridge: CambridgeUniversity Press.

Hansell, M. H. 2005. Animal Architecture. Oxford: Oxford University Press.Hay, M. E. & Duffy, J. E. 1990. Host-plant specialization decreases predation

on a marine amphipod: an herbivore in plants clothing. Ecology, 71,733743.

Hilker, M. 1992. Protective devices of early developmental stages in Pyrrhaltaviburni (Coleoptera, Chrysomelidae). Oecologia, 92, 7175.

Karren, J. B. 1964. Protective coloration and form in the North American genusExema (Chrysomelidae, Coleoptera). Proceedings of the North Central Branch,Entomological Society of America, 19, 7779.

Karren, J. B. 1972. A revision of the subfamily Chlamisinae of America northof Mexico (Coleoptera: Chrysomelidae). University Science Bulletin, 49,875988.

LeSage, L. 1982. The immature stages of Exema canadensis Pierce (Coleoptera:Chrysomelidae). Coleopterists Bulletin, 36, 318327.

LeSage, L. 1984. Immature stages of Canadian Neochlamisus Karren (Coleoptera:Chrysomelidae). Canadian Entomologist, 116, 383409.

Levin, D. A. 1973. Role of trichomes in plant defense. Quarterly Review of Biology, 48,315.

McKie, B. G. 2004. Disturbance and investment: developmental responses oftropical lotic midges to repeated tube destruction in the juvenile stages.Ecological Entomology, 29, 457466.

Medeiros, L. & Moreira, G. R. P. 2002. Moving on hairy surfaces: modications ofGratiana spadicea larval legs to attach on its host plant Solanum sisymbriifolium.Entomologia Experimentalis et Applicata, 102, 295305.

Millott, N. 1955. The covering reaction in a tropical sea urchin. Nature, 175, 561.Moldenke, A. R. 1971. Host-plant relations of phytophagous beetles in Mexico

(Coleoptera- Bruchidae, Chrysomelidae, Curculionidae). Pan-Pacic Entomolo-gist, 47, 105116.

Moran, N. A. 1986. Morphological adaptation to host plants in Uroleucon (Homo-ptera: Aphididae). Evolution, 40, 10441050.

Morton, T. C. & Vencl, F. V. 1998. Larval beetles form a defense from recycled-hostplant chemicals discharged as fecal wastes. Journal of Chemical Ecology, 24,765785.

Mulatu, B., Applebaum, S. W. & Coll, M. 2004. A recently acquired host plantprovides an oligophagous insect herbivore with enemy-free space. Oikos, 107,231238.

Muller, C. & Hilker, M. 1999. Unexpected reactions of a generalist predator towarddefensive devices of cassidine larvae (Coleoptera, Chrysomelidae). Oecologia,118, 166172.

Muller, C. & Hilker, M. 2003. The advantages and disadvantages of larval abdom-inal shields on the Chrysomelidae: a mini-review. In: Special Topics in Leaf BeetleBiology (Ed. by D. G. Furth), pp. 243259. Soa-Moscow: Pensoft.Neal, J. W., Jr. 1989. Bionomics of immature stages and ethology of Neochlamisusplatani (Coleoptera, Chrysomelidae) on American sycamore. Annals of theEntomological Society of America, 82, 6472.

Nogueira-de-Sa, F. & Trigo, J. R. 2002. Do fecal shields provide physical protec-tion to larvae of the tortoise beetles Plagiometriona avescens and Stolaschalybea against natural enemies. Entomologia Experimentalis et Applicata, 104,203206.

Olmstead, K. L. 1994. Waste products as chrysomelid defences. In: Novel Aspectsof the Biology of Chrysomelidae (Ed. by P. H. Jolivet, M. L. Cox & E. Petitpierre),pp. 311318. Boston: Kluwer Academic.

Olmstead, K. L. & Denno, R. F. 1992. Cost of shield defence for tortoise beetles(Coleoptera: Chrysomelidae). Ecological Entomology, 17, 237243.

Olmstead, K. L. & Denno, R. F. 1993. Effectiveness of tortoise beetle larval shieldsagainst different predator species. Ecology, 74, 13941405.

Poinar, G. O., Jr. 1999. Chrysomelidae in fossilized resin: behavioural inferences.In: Advances in Chrysomelidae Biology 1 (Ed. by M. L. Cox), pp. 116. Leiden:Backhurys.

Popenoe, A. M. & Marlatt, C. L. 1889. The sycamore leaf beetle, Chlamys plicata Fab.In: First Annual Report of the Kansas Experiment Station, State Agricultural College,pp. 4041. Topeka, Kansas: G. W. Crane.

Riley, C. V. 1874. Innoxious insects: the Dominican case-bearer- Coscinopteradominacana (Fabr.). In: Sixth Annual Report on the Noxious, Benecial, and OtherInsects of the State of Missouri, 128131. Jefferson City, Missouri: Regan & Carter,State Printers.

Rivers, D. B., Antonelli, A. L. & Yoder, J. A. 2002. Bags of the bagworm Thyr-idopteryx ephemeraeformis (Lepidoptera: Psychidae) protect diapausing eggsfrom water loss and chilling injury. Annals of the Entomological Society ofAmerica, 95, 481486.

Root, R. B. & Messina, F. J. 1983. Defensive adaptations and natural enemies ofa case-bearing beetle, Exema canadensis (Coleoptera: Chrysomelidae). Psyche,90, 6780.

Roda, A., Nyrop, J., Dicke, M. & English-Loeb, G. 2000. Trichomes and spider-mitewebbing protect predatory mite eggs from intraguild predation. Oecologia, 125,428435.

Ruxton, G. D., Sherratt, T. N. & Speed, M. P. 2004. Avoiding Attack: the EvolutionaryEcology of Crypsis, Warning Signals and Mimicry. Oxford: Oxford UniversityPress.

Seymour, R. S. 1974. Convective and evaporative cooling in sawy larvae. Journal ofInsect Physiology, 20, 24472457.

Smith, A. P. 1978. An investigation of the mechanisms underlying nest constructionin the mud wasp Paralastor sp. (Hymenoptera: Eumenidae). Animal Behaviour,26, 232240.

Sokal, R. R. & Rohlf, F. J. 1995. Biometry. 3rd edn. New York: W.H. Freeman.Stachowicz, J. J. & Hay, M. E. 1999. Reducing predation through chemically

mediated camouage: indirect effects of plant defenses on herbivores. Ecology,80, 495509.

Stachowicz, J. J. & Hay, M. E. 2000. Geographic variation in camouage speciali-zation by a decorator crab. American Naturalist, 156, 5971.

Stevens, D. J., Hansell, M. H., Freel, J. A. & Monaghan, P. 1999. Developmentaltrade-offs in caddis ies: increased investment in larval defence alters adultresource allocation. Proceedings of the Royal Society of London, Series B, 266,10491054.

Thanh, P. D., Wada, K., Sato, M. & Shirayama, Y. 2003. Decorating behaviour by themajid crab Tiarinia cornigera as protection against predators. Journal of theMarine Biological Association of the U.K., 83, 12351237.

Valverde, P. L., Fornoni, J. & Nunez-Farfan, J. 2001. Defensive role of leaf trichomesin resistance to herbivorous insects in Datura stramonium. Journal of Evolu-tionary Biology, 14, 424432.

Vencl, F. V. & Morton, T. C. 1998. The shield defense of the sumac ea beetle,Blepharida rhois (Chrysomelidae: Alticinae). Chemoecology, 8, 2532.

Vencl, F. V., Morton, T. C., Mumma, R. O. & Schultz, J. C. 1999. Shield defense ofa larval tortoise beetle. Journal of Chemical Ecology, 25, 549566.

Venner, S., Bel-Venner, M., Pasquet, A. & Leborgne, R. 2003. Body-mass-depen-dent cost of web-building behaviour in an orb weaving spider, Zygiella x-notata.Naturwissenschaften, 90, 269272.

Wallace, J. B. 1970. The defensive function of a case on a Chrysomelid larva. Journalof the Georgia Entomological Society, 5, 1924.

Weaver, D. K., McFarlane, J. E. & Alli, I. 1989. Aggregation in yellow mealworms,Tenebrio molitor L (Coleoptera Tenebrionidae) larvae. 1. Individual and groupattraction to frass and isolation of an aggregant. Journal of Chemical Ecology, 15,16051615.

Weiss, M. R. 2003. Good housekeeping: why do shelter-dwelling caterpillars ingtheir frass? Ecology Letters, 6, 361370.

Weiss, M. R. 2006. Defecation behavior and ecology of insects. Annual Review ofEntomology, 51, 635661.

Zar, J. H. 1996. Biostatistical Analysis. Upper Saddle River, New Jersey: PrenticeHall.

APPENDIX

Timing between successive levels of predation threat as a function of beetle species, case-associated trait, predator and life history stage (pooled analyses combined data from larvae and pupae)

Predator/beetle stage N Time to investigation Time between investigation andattack

Time between attack and death Survival time Number of attacks escaped

H P H P H P H P H PN. bimaculatusCaseCricketsLarvae 10 10 122.9 4816.4 2.81 0.05 00.0 6518.4 11.76 0.0003 61.2 5530.5 0.82 0.18 173.2 1164.5 15.83