Generalist predation on forest tent caterpillar varies with forest stand composition: an experimental study across multiple life stages

Ecological Entomology (2012), 37, 13–23 DOI: 10.1111/j.1365-2311.2011.01330.x

Generalist predation on forest tent caterpillar varieswith forest stand composition: an experimental studyacross multiple life stagesA M Y E . N I X O N and J E N S R O L A N D Department of Biological Sciences, University of Alberta,

Edmonton, Canada

Abstract. 1. Generalist enemies can regulate low-density forest insect populations,and are widely considered to cause greater mortality in more diverse habitats. Foresttent caterpillars (Malacosoma disstria Hubner; FTC) are a major defoliator of aspen(Populus tremuloides Micheaux) in the boreal forest, a region with a mosaic of foreststand types. This heterogeneity may influence FTC outbreaks if generalist predationor parasitism differs among stands of different tree composition.

2. Using exclusion experiments we estimate predation and parasitism of FTC acrossmultiple life-history stages in low-density populations occupying both aspen (lowdiversity) and mixedwood stands (high diversity).

3. Arthropod and avian generalist predators were responsible for most naturalenemy-caused mortality of immature FTC, but their relative impacts varied amongFTC life-history stages. Contrary to expectation, predation on late instar larvae andpupae was higher in the less diverse aspen stands and early instar mortality did notdiffer.

4. By considering multiple life-history stages, our results provide a morecomprehensive view of natural enemy-caused morality of immature FTC. Becausegeneralist predation on FTC was higher in aspen than in mixedwood stands, wesuggest that FTC populations may be slower to reach outbreak levels in aspen stands.

Key words. Exclusion experiment, forest composition, forest insect, forest tent cater-pillar, generalist predation, habitat diversity, natural enemies, population dynamics.

Introduction

Interactions with natural enemies are widely considered anessential component of the dynamics of forest insect popu-lations. Natural enemies including predators, parasitoids, andpathogens have been proposed both as regulators of low-density endemic populations in some forest insect populations(Gould et al., 1990; Elkinton et al., 1996; Klemola et al., 2002)and as drivers behind the quasi-periodic outbreaks of others(e.g. Berryman, 1996; Turchin et al., 2003; Klemola et al.,2010). Generalist enemies, especially generalist predators, fre-quently have their greatest impact in low-density populationsand predation may be responsible for maintaining forest defo-liator populations at endemic densities (Gould et al., 1990;Elkinton et al., 1996; Klemola et al., 2002). In contrast, lagged

Correspondence: Jens Roland, Department of Biological Sciences,University of Alberta, Edmonton, Alberta T6G 2E9, Canada. E-mail:[email protected]

density-dependent interactions with specialist enemies such asparasitoids or pathogens, in which enemy populations respondnumerically to prey populations but with a time-lag, may drivedefoliator population cycles but cause very little mortality inlow-density populations (Berryman, 1996; Turchin et al., 2003;Klemola et al., 2010). Because both specialist and general-ist enemies can affect forest insect dynamics, spatial variationin natural enemy communities can have consequences for thedynamics of forest insect populations and result in geographi-cal gradients in dynamics or variation in dynamics associatedwith habitat heterogeneity (e.g. Roland & Taylor, 1997; Kle-mola et al., 2002).

The forest tent caterpillar (FTC), Malacosoma disstriaHubner (Lepidoptera: Lasiocampidae), is a widespread cyclicdefoliator of aspen, Populus tremuloides Micheaux (Saliaceae)in the boreal forest. Forest tent caterpillar populations cyclewith an approximately 10-year periodicity across much oftheir range and localised outbreaks typically last 2–3 years,although both cycle length and outbreak duration vary

© 2012 The AuthorsEcological Entomology © 2012 The Royal Entomological Society 13

14 Amy E. Nixon and Jens Roland

geographically (Cooke et al., 2009). A variety of mechanismsmay contribute to the cyclic dynamics of FTC populations,including: maternal effects (Myers, 1990), weather, (Rolandet al., 1998), pathogens (Stairs, 1966), and natural enemies,especially parasitoids (Parry, 1995; Roland & Taylor, 1997;Roland, 2005).

Forest tent caterpillars are attacked by a suite of dipteranand hymenopteran parasitoids (Witter & Kulman, 1972; Parry,1995). Declines from peak density have been associated withhigh parasitism by the sarcophagid fly Arachnidomyia aldrichiParker and the tachinid fly Leschenaultia exul Townsend(Witter & Kulman, 1979; Parry, 1995). In addition, analysesof short population time series indicate the presence of laggeddensity-dependent processes (Roland, 2005), further suggestingparasitism may be partly responsible for promoting FTCpopulation cycles. A variety of generalist arthropod and avianpredators also attack FTC throughout their life cycle (Witter& Kulman, 1972; Parry et al., 1997; Glasgow, 2006) and,although predation on pupae does not appear to regulateFTC populations, generalist predators may still cause highFTC mortality in low-density populations (Glasgow, 2006).Interactions between FTC and their natural enemies have beenwidely studied across a range of FTC densities (e.g. Parry,1995; Parry et al., 1997; Roland & Taylor, 1997; Rothman& Roland, 1998; Glasgow, 2006; Roth et al., 2006), howeverindividual studies have typically focused on a single guild ofnatural enemies (i.e. predators, pathogens, or parasitoids), andon only one or a few life stages (but see Rothman & Roland,1998). It is therefore unclear how specialist and generalistnatural enemies interact across the immature stages of FTClife history to contribute to generational mortality, especiallyin low-density populations.

Fragmentation of deciduous forests by agriculture influencesFTC population dynamics, with outbreaks typically lastinglonger in regions where fragmentation is high (Roland, 1993),and disruption of host–parasitoid interactions may be partlyresponsible for this variation in dynamics in response tolandscape heterogeneity (Roland & Taylor, 1997). In themixedwood boreal forest of northern Alberta, forest standcomposition and diversity is highly variable, and aspen canbe found in pure stands or in mixedwood stands withother deciduous and coniferous trees. Although the effects ofdeciduous forest fragmentation on FTC population dynamicshave been examined, the mixedwood boreal forest alsocomprises a large part of the range of FTC and habitateffects on FTC populations in this ecosystem have never beenthoroughly considered (but see Parry et al., 1997).

The ‘natural enemies hypothesis’ (Root, 1973; Russell,1989) proposes that the impacts of natural enemies on her-bivore insect populations are greater in more diverse habitats,in particular habitats with greater vegetational species diver-sity, because more diverse habitats support a higher abundanceor diversity of natural enemies. More diverse habitats typicallyprovide a greater variety of prey species and microhabitats forgeneralist natural enemies, a greater diversity of resources foradult parasitoids, or a greater abundance of refuges for preyspecies that allows for persistence of specialist enemy popu-lations (reviewed by Jactel et al., 2005). Although the natural

enemies hypothesis was originally motivated by observationsfrom agricultural systems, observations of reduced forest pestdamage in more diverse forest stands has advanced its con-sideration in forest systems (e.g. Jactel et al., 2005; Korichevaet al., 2006; Jactel & Brockerhoff, 2007). For other forest defo-liators, such as the spruce budworm, Choristoneura fumiferanaClemens (Lepidoptera: Tortricidae), and autumnal moth, Epir-rita autumnata Borkhausen (Lepidoptera: Geometridae), pre-dation and parasitism, especially by generalists, are higher inmore diverse forest stands (Cappuccino et al., 1998; Quayleet al., 2003; Riihimaki et al., 2005), although the responseby specialist parasitoids may be opposite (Herz & Heitland,2005). Forest stand diversity may therefore alter the inter-actions between FTC and its natural enemies, with potentialimplications for the dynamics of FTC populations in the mixed-wood boreal forest.

The objectives of this study were to determine: (i) therelative contribution of generalist predators and both specialistand generalist parasitoids to mortality of low-density FTCpopulations, and (ii) if natural enemy-caused mortality ofFTC is greater in more diverse forest stands as predictedby the natural enemies hypothesis. We use several enemyexclusion treatments to identify the contribution of predatorsand parasitoids to mortality of FTC larvae and pupae andcompare natural enemy-caused mortality between aspen (lowdiversity) and mixedwood (high diversity) forest stands. Weexpect generalist enemies to be responsible for most ofthe natural enemy-caused mortality in the low-density FTCpopulations. Furthermore, we expect greater natural enemy-caused mortality of FTC in the more diverse mixedwood standscompared to the less diverse aspen stands.

Materials and methods

Study location and site selection

Field experiments were conducted through the spring andsummer of 2009 and 2010 in the mixedwood boreal forestof north-central Alberta, Canada (55◦31′N, 113◦27′W). Foresttent caterpillar populations in the region do not exhibit regularcycles (Cooke, 2001). Following a sustained outbreak, FTCpopulations in the region collapsed in 1989 and have sinceremained low, with no detectable defoliation (J. Roland, pers.obs.). Upland forest in this region is dominated by standsof aspen and balsam poplar, Populus balsamifera Linnaeus(Saliaceae), and deciduous-conifer mixedwood stands that alsoinclude white spruce, Picea glauca (Moench) Voss (Pinaceae),and balsam fir, Abies balsamea (Linnaeus) Miller (Pinaceae);the proportion of coniferous species present typically reflectstime since disturbance.

Twenty forest stands (10 aspen stands and 10 mixedwoodstands) were selected in the spring of 2009 based onforest composition (assessed visually), accessibility, and theavailability of aspen saplings suitable for exclusion treatments.Selected sites extended over a transect approximately 100 kmlong and were a minimum of 1 km apart. Stands ranged insize between 0.9 and 27.5 ha (median: 2.86 ha). Aspen standshad canopies consisting of >80% aspen and <10% coniferous

© 2012 The AuthorsEcological Entomology © 2012 The Royal Entomological Society, Ecological Entomology, 37, 13–23

Forest tent caterpillar predation and forest composition 15

species. Mixedwood stands were aspen dominated (>50%aspen in the canopy), but contained a minimum of 20% whitespruce in the canopy.

Yearly male moth FTC abundance was estimated by theaverage trap-catch from two pheromone traps at each site(Schmidt et al., 2003). Two traps baited with pheromone lures(Cantech Enterprise Inc., Delta, British Columbia, Canada)were hung between 1.5 and 2 m above ground level, 50 m apartbefore the flight period at the beginning of July and collectedat the end of August in both years.

Experimental exclosures

To determine the relative impacts of generalist predators andparasitoids on FTC mortality, we established four exclusiontreatments on aspen saplings at each site. Treatments wereas follows: Enemy exclusion: All natural enemies (predatorsand parasitoids) were excluded by enclosing the sapling in alight-coloured, fine mesh bag and applying an approximately20 cm-wide band of Tanglefoot (The TangleFoot Company,Grand Rapids, Michigan) around the tree base. Velcro orzippers on one side of the bag allowed access for sampling.Predator exclusion: Both epigaeic arthropod predators andavian predators were excluded, while still allowing accessby parasitoids. Epigaeic arthropods were excluded withTanglefoot. Birds were excluded using a bamboo frame drapedwith 1-inch mesh gill netting erected around the saplingcanopy. A 15-cm tall aluminium funnel was installed aroundthe base of the exclosure to prevent larvae from escapingdown the tree trunk. These exclosures were not effectiveat preventing predation by spiders (Araneae) or pentatomids(Hemiptera: Pentatomidae), which were removed by handwhen they were observed. Arthropod exclusion: Epigaeicarthropods were excluded while allowing access by birds andparasitoids by applying Tanglefoot around the base of thesapling. Spiders and pentatomids were removed by hand whenobserved. Open: All natural enemies were allowed access toFTC larvae and pupae. All treatments were established at theonset of the experiment with the exception of the arthropodexclusion treatment, which was added in 2010 to distinguishbetween arthropod and avian predation.

Saplings selected for exclusion treatments were of similarsize (approximately 3.5 m tall) and were a minimum of 10 minto each forest stand from the edge. The enemy exclusionand predator exclusion treatments were placed near each other,but a minimum of 50 m from the arthropod exclusion andopen treatments (which were placed together). This separationprevented the conspicuous exclosures from affecting predationand parasitism at the less conspicuous treatments.

In endemic FTC populations, early instar larvae are foundhigh in the canopy where females laid egg masses theprevious summer and, while late instar larvae and pupaeare progressively more dispersed vertically through the foreststrata, the majority remain in the canopy (Batzer et al., 1995).The use of exclusion treatments in the understorey in thisstudy may introduce bias into observations of predation andparasitism; for example, parasitism of FTC by both tachinid

and sarcophagid flies is higher in the understorey than in thecanopy (Witter & Kulman, 1979; Parry, 1995). Conducting asimilar experiment on FTC larvae and pupae in the canopy,however, would be impractical if not impossible, and anyimposed bias is consistent among experimental treatments.

Experimental assessment of natural enemy-caused mortalityof FTC

Natural enemy-caused mortality in aspen and mixedwoodforest stands was assessed using exclusion treatments for threelife-history stages of FTC separately: early instar larvae (instars1–3), late instar larvae (instars 4–5), and pupae.

In late April of both years, each experimental tree wasstocked with two egg masses obtained the previous winterfrom high-density FTC populations in northern Alberta (2009)or near Prince George, British Columbia (2010). Early instarlarvae hatching from each of the two egg masses weremonitored twice weekly for abundance and instar until thesecond observation of third-instar larvae or until moult tothe fourth instar, whichever occurred first. At the end of theobservation period, remaining larvae were counted and re-collected from the saplings. In 2009, approximately 50 larvaewere left and allowed to pupate in each enemy exclusiontreatment, and were subsequently used in the assessment ofpupal mortality factors (see below). The initial number offirst-instar FTC larvae on each tree was determined fromhatched eggs in egg masses. Early instar larvae rarely dispersefrom their natal colonies (Fitzgerald, 1995) so all losses ofthese larvae from the experimental trees were interpreted asmortality, with the source inferred according to the exclusiontreatment.

Because fourth- and fifth-instar FTC larvae disperse fromtheir natal colonies (Fitzgerald, 1995), it was necessary totether these larvae to the experimental trees to ensure recoveryand fate determination. Tethers, consisting of cotton threadadhered transversely to the abdomens of the larvae with cyano-acrylate adhesive (Krazy Glue�), were attached to larvae for12–24 h prior to deployment. Because defoliation is used as anoviposition cue by some tachinid parasitoids of FTC (Mondor& Roland, 1998), larvae were tethered to experimental treeswith approximately 20 cm of slack thread, and within easyaccess to leaves previously eaten by FTC larvae. Larvae werefrequently tangled to some extent upon retrieval. In late Juneof 2010 only, fifth-instar larvae were tethered in sets of two orthree per tree for 24 h on two separate days, for a total of five,fifth-instar larvae per exclusion treatment per site. After 24 h,we recorded the fate of the larvae (dead, alive, or preyed on).Partial larval remains were classified as having been preyedon. Live larvae were collected and subsequently classifiedas being healthy (if an adult moth eclosed), parasitised (ifa parasitoid emerged from the larva or pupa), or dead fromunknown sources. Unknown mortality may result from viralor fungal pathogens (Stairs, 1966) or, in this study, as a sideeffect of the tethering protocol, but specific causes were notidentified.

To assess natural enemy-caused mortality of FTC pupae,large fifth-instar cage-reared FTC larvae not previously



exposed to parasitoids were placed in mesh bags on branchesof each experimental tree and allowed to spin cocoons andpupate in the leaves. Within 3 days of pupation, mesh bagswere removed to expose pupae to predation and parasitism.In July 2009, between three and nine (median = 4.5) pupaewere deployed using mesh bags on each tree in the predatorexclusion and open treatments. Pupae in the enemy exclusiontreatment developed from the approximately 50 FTC larvaethat remained in the exclosures since hatch. All pupae werecollected over 2 days in early August. In the predator exclusionand open treatments, exposure times of pupae ranged from12 to 26 days (median: 17 days) depending on phenology. InJuly 2010, five pupae were deployed in all three exclusiontreatments using mesh bags. All pupae were re-collected afterhaving been exposed for 12 days.

Pupal fates were identified based on a suite of characteris-tics: Healthy : a healthy moth eclosed or the puparium had cleansutures and the silk cocoon had an exit hole with scales remain-ing; Parasitised : presence of parasitoid larva, pupa, or adult;characteristic parasitoid exit hole from puparium or characteris-tic staining of cocoon silk (Hodson, 1939); Preyed on: absenceof puparium but silk cocoon remaining or silk cocoon not intactand puparium not intact; Unknown mortality : all pupae thatcould not be otherwise classified. Fates were assigned blindlywith respect to treatment.

Parasitism rates of tethered larvae exposed for 24 h werevery low (see Results) so additional collections of FTClarvae were made in 2010 to better characterise the parasitoidcommunity. Between five and seven egg masses were placedon four or five additional saplings at each site to facilitatelarval collections. At least 30 late fourth- or early fifth-instarlarvae were collected from all but one site (median = 73)between 17 June and 27 June. Re-collected tethered fifth-instar larvae and larvae collected to estimate parasitism rateswere reared in plastic cups (Solo�) in groups of three andfed fresh aspen foliage every other day until the emergence ofparasitoids, pupation, or death. Re-collected pupae were storedat room temperature in plastic cups until emergence or death.All parasitoids recovered from larvae or pupae were identifiedusing published keys (Sippell, 1961; Dasch, 1971; Goulet &Huber, 1993; Williams et al., 1996).

Data analysis

We compared male FTC moth abundance between foreststand types and years using two-sample and paired t-tests,respectively. We used linear mixed effects models to comparemortality of all FTC life-history stages between forest standtypes and among exclusion treatments, unless otherwiseindicated. In all linear mixed models, site nested withinforest type was included as random factor. For analysis oftethered larvae, tethering date was additionally nested withinsite. All proportions were arcsin-square-root transformed.The significance of main effects and their interactions wereevaluated with F -ratio tests, and likelihood ratio tests ofnested models were used for model simplification. Significantmain effects or interactions were further investigated using

a posteriori orthogonal contrasts evaluated at α/n where n isthe number of contrasts.

We examined the effects of exclusion treatment and foresttype on mortality of early instar FTC larvae using linearmixed models. Total early instar larval mortality in 2009and 2010 were analysed separately because of the additionof the arthropod exclusion treatment in 2010. Damage tosaplings and premature larval dispersal reduced replicationof the predator exclusion treatment in both the aspen andmixedwood forest (n = 9 each) in 2009, and replication of thepredator exclusion treatment in aspen stands (n = 6), the opentreatment in mixedwood (n = 7) and aspen (n = 9) stands, andthe arthropod exclusion treatments in aspen stands (n = 9) in2010.

Of the total observed mortality of late instar larvae, unknownmortality comprised 45% in both the open and arthropodexclusion treatments, 77% in the predator exclusion treatment,and 100% in the enemy exclusion treatment. Because of thishigh unknown mortality, we only consider larval fates thatcould be conclusively identified (predation and parasitism).The effects of exclusion treatment and forest stand type on theproportion of larvae preyed on were analysed using a linearmixed model, and only those treatments in which predationoccurred were considered (e.g. the enemies exclusion treatmentwas omitted).

Predation and mortality from unknown sources mayhave caused larval death before the emergence of para-sitoids so that actual parasitism rates may be obscuredby these other mortality factors. To adjust the appar-ent parasitism rate to account for the contemporaneouseffects of predation and unknown mortality, we calcu-lated marginal parasitism rates (mParasitism; Elkinton et al.,1992):

mParasitism = pParasitism

1 − mUnknown − pPredation

where:mUnknown = pUnknown

1 − pPredation(1)

where px is the observed proportion of larvae dying fromcause x. This calculation assumes that parasitoids are entirelyout-competed by pathogens or other factors causing unknownmortality, and corrects for predation preventing the detectionof both unknown mortality and parasitism. Parasitism of teth-ered larvae was rare, so we compared apparent and marginalparasitism rates between forest types and among exclusiontreatments using Kruskal–Wallis tests. If larval predation isindependent of parasitism, marginal parasitism rates should bethe same for larvae protected from predators and those exposedto predation.

In 2009, a low incidence (8.3%) of pupal parasitism in theenemy exclusion treatment occurred primarily because manylarvae spun cocoons on the exclusion cage rather than inthe foliage; these pupae were excluded from the analysis.All other pupal data from both years were combined. Theeffects of exclusion treatment and forest stand type and yearon total pupal mortality, the proportion of pupae preyed onand the apparent and marginal parasitism rates (calculated as



for tethered larvae) were analysed using separate linear mixedmodels.

Unless otherwise indicated, all analyses were conducted withα = 0.05 and were conducted in R v 2.12.1 (R DevelopmentCore Team, 2010, Vienna, Austria) using functions availablein the stats, nlme, lattice, and gmodels packages.

Results

Male moth FTC abundance did not differ between aspenand mixedwood stands in either year (2009: t18 = 0.46, P =0.645; 2010: t18 = 1.83, P = 0.083). Abundance increasedsignificantly between years (2009: 5.1 ± 0.9 moths per trap;2010: 16.5 ± 2.1 moths per trap; t19 = –6.46, P < 0.0001),but was low in both years compared to outbreak levels (approx.100 moths per trap; Roland, 2005).

Early instar larval mortality

Enemy exclusion treatments reduced early instar mortalityin both years (2009: F2,36 = 25.07, P < 0.0001; 2010: F3,48 =15.71, P < 0.0001; Fig. 1a,b). In 2009, excluding all gener-alist predators reduced mortality from 66.5% ± 5.6% (opentrees) to 28.8% ± 2.1% (predator and enemy exclusions; t52 =6.79, P < 0.0001). In 2010, excluding only arthropod preda-tors reduced mortality from 62.1% ± 4.9% (open trees) to37.9% ± 2.3% (all other treatments; t63 = 6.64, P < 0.0001).Further excluding avian predators had no additional effecton early instar mortality (t63 = 0.39, P = 0.698), indicatingthat arthropod predators are the primary source of natu-ral enemy-caused mortality for early instar larvae. No par-asitism of early instar larvae was observed in any treat-ment in either year. There was no difference in early instarlarval mortality between forest types in either year (2009:F1,18 = 3.05, P = 0.098; 2010: F1,18 = 0.77, P = 0.393;Fig. 1a,b).

Late instar larval predation and parasitism

Predation on late instar larvae was greater in aspen standsthan in mixedwood stands (F1,18 = 11.90, P = 0.003; Fig. 2).A small amount of predation (8.7% ± 4.0%) occurred in thepredator exclusion treatment, indicating that some predator wasnot successfully excluded (Fig. 2). Predation rate was affectedby exclusion treatment (F2,76 = 7.82, P = 0.0008; Fig. 2).Excluding arthropod predators had no effect on the preda-tion rate (t114 = 0.48, P = 0.629), but additionally excludingavian predators reduced predation (t114 = 3.51, P = 0.0006),indicating that birds were the primary source of late instarlarval predation. Exclusion of bird predators reduced larvalpredation in aspen stands only, leading to a significant inter-action between forest type and exclusion treatment (treat-ment × forest: F2,76 = 3.22, P = 0.046; Fig. 2), indicatingthat predation was higher in aspen stands only.

Pro

po

rtio

n P

reye

d O

n

0.0

0.2

0.4

0.6

0.8

1.0

AspenMixedwood

a

b

a

b b

b

OpenArthropod Exclusion

Predator Exclusion

Treatment

Fig. 2. Mean (± SE) predation of tethered late instar forest tentcaterpillar larvae in aspen and mixedwood forest stands in 2010.Treatments as in Fig. 1. Letters indicate homogeneous groups.

Pro

po

rtio

n D

ead

0.0

0.2

0.4

0.6

0.8

1.0

AspenMixedwood

a

b b

OpenPredator Exclusion

Enemy Exclusion

Treatment

2009(a)

Pro

po

rtio

n D

ead

0.0

0.2

0.4

0.6

0.8

1.0

AspenMixedwooda

b b b

OpenArthopod Exclusion

Predator Exclusion

Enemy Exclusion

Treatment

2010(b)

Fig. 1. Mean (± SE) mortality of early instar forest tent caterpillar larvae in aspen and mixedwood forest stands in (a) 2009 and (b) 2010. Exclusiontreatments were designed to exclude all natural enemies, both parasitoids and predators (enemy exclusion), exclude avian and arthropod predators(predator exclusion), or to exclude only arthropod predators (arthropod exclusion; 2010 only). Larvae in the open treatment were exposed toall sources of natural enemy-caused mortality. Letters indicate homogeneous groups within years where there are significant differences betweentreatments.



Parasitism of tethered late instar larvae over 24 h wasrare (median: 0%; maximum: 40%), but occurred in allexcept in the enemy exclusion treatment and in both aspenand mixedwood stands. There was no effect of exclusiontreatment or forest type on the proportion of larvae parasitised(treatment: K3 = 1.48, P = 0.478; forest type: K1 = 0.85,P = 0.357; Fig. 3a,b). Marginal parasitism rates were slightlyhigher (median: 0%; maximum 100%) but still did notdiffer between exclusion treatments or forest stand types(treatment: K2 = 1.70, P = 0.429; forest type: K = 1.58, P =0.209; Fig. 3c,d). Although parasitism rates were very lowfor tethered larvae, similar apparent parasitism rates in thepredator exclusion treatment and the open treatment suggeststhat parasitoids do not cause additional mortality in the absenceof predators (i.e. was not compensatory), and similar marginalparasitism rates between these treatments suggests that birds donot differentiate between parasitised and non-parasitised FTClarvae.

Late instar larval parasitism rates determined from thebroader collections made in 2010 were highly variable,ranging from 0% to 66% among all sites. Across allsites, the most widespread parasitoids were the tachinid flyCarcelia malacosomae Sellers and the ichneumonid waspAgrypon anale Say. Leschenaultia exul (Diptera: Tachinidae)and Hyposoter fugitivus Say (Hymenoptera: Ichneumonidae)caused high parasitism rates at some sites

Pupal mortality

Total pupal mortality did not differ between years (F1,37 =2.71, P = 0.108), nor between forest stand types (F1,37 =

Ap

par

ent

Par

asit

ism


Predator Exclusion

00.

51

Treatment

(a)

Ap

par

ent

Par

asit

ism

Aspen Mixedwood

00.

51

Forest Type

(b)

Mar

gin

al P

aras

itis

m


Predator Exclusion

00.

51

Treatment

(c)

Mar

gin

al P

aras

itis

m

Aspen Mixedwood

00.

51

Forest Type

(d)

Fig. 3. Apparent (a, b) and marginal (c, d) parasitism rates of tetheredlarvae in each exclusion treatment (a, c; treatments as in Fig. 1) and inaspen and mixedwood stands (b, d). Heavy horizontal lines indicate themedian in each group, with box lengths and whisker lengths indicatingthe interquartile range (IQR) and 1.5 IQR, respectively. Open symbolsindicate outliers. A small amount of noise was added to the outliers toseparate overlapping points.

0.075, P = 0.786; Fig. 4a). Pupal mortality differed amongexclusion treatments (F2,71 = 55.21, P < 0.0001). Preda-tor exclusion reduced pupal mortality from 75.3% ± 4.0%to 51.6% ± 4.6% (t103 = 4.128, P < 0.0001), and excludingparasitoids reduced mortality further to 14.3% ± 2.8% (t103 =6.31, P < 0.0001), indicating that both predation and para-sitism contribute to FTC pupal mortality.

Predation on pupae only occurred in the open treatment, andwas consistently greater in aspen than mixedwood forests inboth years (forest type: F1,36 = 4.60, P = 0.039; year: F1,36 =1.00, P = 0.323; Fig. 4b). In both years, pupal parasitism wasdominated by the sarcophagid fly Arachnidomyia aldrichi anda suite of ichneumonid wasps, primarily Itoplectis quandricin-gulata Provancher, I. conquisitor Say, and Theronia atalan-tae fulvescens Cresson. Neither the apparent nor the marginalpupal parasitism rates differed between years or between foreststand types (apparent parasitism – year : F1,37 = 0.04, P =0.834; forest type: F1,37 = 0.13, P = 0.720; marginal para-sitism – year: F1,37 = 0.31, P = 0.582; forest type: F1,37 =0.002, P = 0.966). However, the apparent parasitism rate ofpupae was lower when predators were excluded compared towhen both predators and parasitoids were present (F1,38 =12.01, P = 0.0013; Fig. 4c). The reduced parasitism rate inthe predator exclosure relative to the open treatment suggeststhat parasitoids were negatively affected by the presence of thepredator exclosure (i.e. a ‘cage effect’). Marginal parasitismrates were also lower in the predator exclosure than in theopen treatment (F1,38 = 11.79, P = 0.0015; Fig. 4d), possiblyindicating that predation was non-random with respect to para-sitism and that predators preferentially attacked non-parasitisedpupae in the open treatment.

Discussion

Most natural enemy-caused mortality in the low-density FTCpopulations in our study was from generalists, both generalistpredators and generalist parasitoids. Among generalists how-ever, the guilds that contributed most to FTC mortality differedacross the life history of FTC larvae and pupae. Natural enemy-caused mortality of early instar FTC larvae was similar in aspenand mixedwood stands but, contrary to our expectation, pre-dation on late instar larvae and pupae was higher in aspenstands.

Predation

Early instar FTC larvae are vulnerable to predation bybirds and arthropods, including spiders, ants, beetles, andpentatomids. In 2009, the exclusion treatments indicated thatpredators of early instar FTC larvae could reduce larvalabundance by approximately 40%. Although predation bypentatomids can contribute to mortality of early instar FTClarvae, especially if hatch is delayed, and birds have beenimplicated in the disappearance of entire colonies (Parryet al., 1998), we found no evidence that pentatomids or birdscontribute to early instar larvae FTC mortality. The addition ofthe arthropod exclusion treatment in 2010 demonstrated that



Total Pupal Mortality

0.0

0.2

0.4

0.6

0.8

1.0

AspenMixedwood

a

b

c

(a)


Enemy Exclusion

Treatment

Pro

po

rtio

n D

ead

Pupal Predation

0.0

0.2

0.4

0.6

0.8

1.0

AspenMixedwood

a

b

(b)

Open

Treatment

Pro

po

rtio

n P

reye

d O

n

Apparent Pupal Parasitism

0.0

0.2

0.4

0.6

0.8

1.0

AspenMixedwood

a

b

(c)


Treatment

Ap

par

ent

Par

asit

ism

Rat

e

Marginal Pupal Parasitism

0.0

0.2

0.4

0.6

0.8

1.0

AspenMixedwood

a

b

(d)


Treatment

Mar

gin

al P

aras

itis

m R

ate

Fig. 4. Mean (± SE) (a) total mortality, (b) predation, (c) apparent parasitism rate, and (d) marginal parasitism rate of forest tent caterpillar pupaein aspen and mixedwood forest stands. Treatments as in Fig. 1. Letters indicate homogeneous groups where there are significant differences amongtreatments or forest types.

loss of early instar larvae was almost exclusively caused byarthropods, including spiders and carabid beetles, accessinglarval colonies from the ground.

The tethering protocol we used on late instar larvae causedhigh unknown mortality; many larvae died after re-collectionfrom the effects of the tether adhesive. However, tethers werenecessary to ensure recovery and fate determination becauseof the dispersal behaviour of these later instars (Fitzgerald,1995), which may in fact be an adaptation for predator andparasitoid avoidance (Parry et al., 1997). Because of the diffi-culty of quantitatively assessing predation on late instar larvae,it is otherwise only anecdotally reported in the literature fromdirect observation (Parry et al., 1997) and avian stomach con-tents (Witter & Kulman, 1979). Predation and parasitism rateswe observed over 24 h are likely elevated above natural levelsand cannot be directly scaled across the entire duration of thefourth- and fifth-instar stages. The tether effects were consis-tent among treatments and forest types however, so inferencesabout effects of forest stand type on late instar larval mortalityremain tenable.

In contrast to high arthropod predation on early instar larvae,the exclusion treatments on late instar larvae indicated thatpredation was largely due to avian predators, particularly in

aspen stands. Although early instar larvae are palatable tobirds (e.g. Pelech & Hannon, 1995), later instars are physicallydefended and are therefore generally considered less palatableto avian predators (Parry et al., 1997). The high incidenceof bird predation (up to 55% per day) on late instar larvaerelative to other mortality was therefore unexpected, giventhat unpalatable prey items are unlikely to be preferred andtargeted by predators, especially at low densities. The timingof late instar larval predation corresponds to the appearance ofbird nestlings and increased demand for food resources amonginsectivorous birds (Parry et al., 1997), thus making late instarFTC larvae more appealing prey. Predation in the predatorexclusion treatment was likely due to pentatomids, whichwe observed attacking tethered larvae on several occasions,and which left behind characteristically exsanguinated larvalcarcasses.

Generalist predation on FTC pupae ranged from 10% to30% in the endemic FTC populations of this study. Duringoutbreaks, predation typically accounts for a similarly smallamount (<10% to 20%) of pupal mortality (Hodson, 1943;Stark & Harper, 1982). In contrast, Parry et al. (1997)reported >90% predation on pupae by avian predators inendemic FTC populations in Alberta, and Glasgow (2006)



observed pupal predation rates of 25–45% over 2 years atintermediate densities, and identified predation from botharthropod and avian predators. Therefore, predation on FTCpupae may be similar to predation on early instar larvae inthat it is the cumulative result of several guilds of generalistpredators.

Parasitism

There was no parasitism on early instar larvae. Onlyone species of parasitoid, Aleiodes malacosomatus Mason(Hymenoptera: Braconidae), attacks FTC prior to the fourthinstar in Alberta (Parry, 1995). This braconid wasp typicallyattacks late second- to early third-instar FTC larvae, subse-quently emerging from late third instar and early fourth instars.Our study design prevented quantitative assessment of para-sitism by A. malacosomatus, because monitoring of early instarlarvae ended prior to emergence of this parasitoid. Previousstudies of FTC parasitism indicate that A. malacosomatus is acommon parasitoid in low-density FTC populations and mayexert density-dependent mortality (Parry, 1995; Parry et al.,1997; Roland, 2000).

Parasitism of late instar larvae, as determined from thebroader collections, was highly variable, and each species ofparasitoid was common at only a few sites. High parasitismrates by the ichneumonids Agrypon anale and Hyposoterfugitivus at several sites were unexpected, considering previousstudies have found only very low or no larval parasitism byichneumonid wasps (Witter & Kulman, 1979; Parry, 1995;Roth et al., 2006). In contrast, the tachinids Leschenaultiaexul and Carcelia malacosomae are commonly recoveredfrom low-density FTC populations (Witter & Kulman, 1979;Parry, 1995). Hymenopteran parasitoids of FTC larvae areat least oligophagous, if not broad generalists (Goulet &Huber, 1993) and, although they are more specialised (Parry,1995), the tachinid parasitoids may persist in low-density hostpopulations by attacking FTC while larvae are still foragingas a colony (Witter & Kulman, 1979; Parry, 1995). Thus,generalist parasitoids, or those that are adapted to and commonin low-density host populations, were common parasitoids oflate instar larvae in this study.

Apparent pupal parasitism was high (45%) in the opentreatment, and considerably more consistent across sitesthan was parasitism of FTC larvae. Ichneumonid parasitoidsattack FTC pupae in outbreaking populations, but have notbeen previously reported in high numbers from endemicpopulations (Witter & Kulman, 1979; Parry, 1995). In contrast,Arachnidomyia aldrichi is a common facultative parasitoidof FTC pupae in Alberta and elsewhere across a range ofhost densities (Hodson, 1939; Witter & Kulman, 1979; Parry,1995). Similar to late-larval parasitoids, pupal parasitoids werecomprised entirely of generalists whose populations can bebuffered against low densities of FTC by the presence ofalternative prey (Schmidt & Roland, 2006). Differences inapparent parasitism between the open and predator exclusiontreatments (Fig. 4c) suggest that parasitoids may have beennegatively affected by the presence of the predator exclusiontreatment.

Interactions between predators and parasitoids

Neither apparent nor marginal parasitism rates of late instarlarvae differed in the presence or absence of predators, sug-gesting that generalist predators do not or cannot differentiatebetween parasitised and unparasitised larvae. FTC larvae par-asitised by tachinid flies typically show no external physicalevidence of parasitism so birds, as visual predators, may not beable to distinguish between those parasitised and unparasitised.

In contrast to FTC larvae, predators of FTC pupae doappear to distinguish between parasitised and unparasitisedindividuals, resulting in a higher marginal parasitism rate ofpupae exposed to predators. In a similar study examining birdand beetle predation on FTC pupae Glasgow (2006) also foundevidence that predators, especially birds, avoided parasitisedFTC pupae. Selection by birds against pupae parasitised byArachnidomyia aldrichi has also been observed directly (Parryet al., 1997). Selective predation on non-parasitised prey isnot limited to interactions between generalist predators andFTC pupae; beetle larvae preferentially prey on unparasitisedwinter moth, Operophtera brumata Linnaeus (Lepidoptera;Geometridae) pupae in the soil rather than pupae parasitised bythe tachinid fly Cyzenis albicans Fallen (Roland, 1990). Suchselection may reflect lower palatability of pupae parasitisedby A. aldrichi to generalist predators, possibly because larvaeof A. aldrichi break down tissues of FTC pupae upon hostpenetration (Hodson, 1939).

Exclusion experiments provide an excellent opportunity toexamine compensatory mortality among natural enemy guilds(Campbell & Torgersen, 1983). Parasitoids did not attack moreFTC pupae when predators were excluded, suggesting thatparasitoids do not compensate for the absence of predators,and that generalist predation is a unique source of mortalityfor FTC pupae. In contrast, avoidance of parasitised pupae bygeneralist predators suggests that birds may attack relativelymore pupae in the absence of parasitoids, thus compensatingfor the absence of parasitoids, although we were unable to testthis directly.

Forest composition and natural enemy-caused mortalityof FTC

According to the natural enemies hypothesis, natural enemy-caused mortality of insects should be greater in morevegetatively diverse habitats as a result of the capacity of thesehabitats to support a greater diversity or abundance of thosenatural enemies (Root, 1973; Russell, 1989). Contrary to ourexpectations, we found no evidence that natural enemy-causedmortality of FTC was greater in mixedwood stands comparedto aspen stands. In fact, bird predation on late instar larvaeand generalist predation on pupae was higher in aspen standsthan in mixedwood stands. Natural enemy-caused mortalityof FTC does vary with forest composition, but possibly moreas a consequence of specific differences in the composition ofthe generalist predator community rather than species richness,diversity or abundance per se (Nixon, 2011).

There is accumulating evidence that landscape-scale hetero-geneity may have a greater influence than local vegetation



diversity on natural enemy communities and their impacts onprey populations (Roland, 2000; Barbaro et al., 2005; Cronin& Reeve, 2005). For example, bird species diversity respondspositively to both the presence of deciduous stands surround-ing pine plantations (Barbaro et al., 2005) and heterogeneity inforest cover measured over 100 ha (Drolet et al., 1999). Par-asitoids also respond to landscape-scale vegetation diversity.In many agricultural studies, parasitoid diversity and para-sitism rates of pests in diverse agricultural landscapes aregreater (Cronin & Reeve, 2005; but see Menalled et al., 1999).Although evidence from forest systems is scarce, spruce bud-worm, Choristoneura fumiferana, parasitism rates are higherin more diverse forest landscapes (Cappuccino et al., 1998),and landscape-scale diversity is associated with reduced infes-tations of forest insect pests (Koricheva et al., 2006). It maytherefore also be useful to consider parasitism and predationon FTC in a landscape-diversity, rather than a stand-diversity,context (Roland, 2000).

Implications for endemic FTC populations

In forest insect populations, natural enemies may functionboth as regulators of low-density endemic populations anddrivers behind the quasi-periodic dynamics of others (e.g.Gould et al., 1990; Berryman, 1996; Klemola et al., 2002,2010; Turchin et al., 2003). Observations and theory suggestthat mortality from generalist enemies can maintain low-density prey populations, in part because persistence ofgeneralists does not depend on the abundance of a singleprey species (Southwood & Comins, 1976; Gould et al.,1990; Tanhuanpaa et al., 1999). However, the contribution ofgeneralist enemies to insect mortality is frequently studied inonly one life-history stage of the prey (e.g. Glasgow, 2006).Understanding the full implications of natural enemy-causedmortality on insect populations however requires considerationof impacts across all life-history stages, and the nature of thedensity dependence of each mortality factor.

Our study was limited to FTC populations at low density,so no inferences can be made about the density dependenceof natural enemy mortality. Whereas previous studies of FTChave only considered mortality of individual life-history stages(e.g. Parry, 1995; Glasgow, 2006) however, our results providethe first complete study of natural enemy-caused mortality ofimmature FTC in low-density FTC populations. To addressthe nature of density dependence of these interactions, similarexperiments may be performed in regions or years withintermediate and high FTC abundances. Enemy exclusiontreatments targeting specific guilds of enemies (parasitoids,arthropod predators, and bird predators) allowed for inferencesregarding sources of mortality even when the result ofenemy attack was simply the disappearance of the prey.As anticipated, we found that most natural enemy-causedmortality in low-density FTC populations was from generalists,both generalist predators and generalist parasitoids. Amonggeneralists, the guilds that contributed most to mortalitychanged across the life history of FTC larvae and pupae;arthropod predators caused high mortality in early instar

larvae, mortality of late instar larvae was predominantly fromavian predators and predators and generalist parasitoids eachcontributed to mortality of pupae. This outcome highlights theadvantage of examining natural enemy-prey interactions acrossseveral life-history stages to obtain a more comprehensive viewof the contribution of separate guilds of natural enemies tototal mortality. Natural enemy-caused mortality of FTC variedwith forest composition, but contrary to expectation, mortalitywas higher in the less-diverse aspen habitat. Because generalistpredation on FTC was higher in aspen forest stands, FTCpopulations may be slower to reach outbreak levels in aspencompared to mixedwood stands, but this expectation remainsto be tested.

Acknowledgements

Field studies were supported by grants to AEN fromthe Alberta Conservation Association, Canadian CircumpolarInstitute, and grants to JR from Alberta Pacific Forest Indus-tries and the National Scientific and Engineering ResearchCouncil. Comments from two anonymous reviewers improvedthe manuscript.

References

Barbaro, L., Pontcharraud, L., Vetillard, F., Guyon, D. & Jactel, H.(2005) Comparative responses of bird, carabid, and spider assem-blages to stand and landscape diversity in maritime pine plantationforests. Ecoscience, 12, 110–121.

Batzer, H.O., Martin, M.P., Matison, W.J. & Miller, W.E. (1995)The forest tent caterpillar in aspen stands: distribution and densityestimation of four life stages in four vegetation strata. ForestScience, 41, 99–121.

Berryman, A. (1996) What causes population cycles of forest Lepi-doptera? Trends in Ecology & Evolution, 11, 28–32.

Campbell, R.W. & Torgersen, T.R. (1983) Compensatory mortalityin defoliator population dynamics. Environmental Entomology, 12,630–632.

Cappuccino, N., Lavertu, D., Bergeron, Y. & Regniere, J. (1998)Spruce budworm impact, abundance and parasitism rate in a patchylandscape. Oecologia, 114, 236–242.

Cooke, B.J. (2001) Interactions between climate, trembling aspen, andoutbreaks of forest tent caterpillar in Alberta. PhD thesis, Universityof Alberta, Edmonton, Canada.

Cooke, B.J., Lorenzetti, F. & Roland, J. (2009) On the durationand distribution of forest tent caterpillar outbreaks in east-centralCanada. Journal of the Entomological Society of Ontario, 140, 3–18.

Cronin, J.T. & Reeve, J.D. (2005) Host–parasitoid spatial ecology: aplea for a landscape-level synthesis. Proceedings of the Royal SocietyB, 272, 2225–2235.

Dasch, C.E. (1971) Ichneumon-Flies of America North of Mexico:Subfamily Mesochorinae. American Entomological Institute, AnnArbor, Michigan.

Drolet, B., Desrochers, A. & Fortin, M. (1999) Effects of landscapestructure on nesting songbird distribution in a harvested borealforest. The Condor, 101, 699–704.

Elkinton, J., Buonaccorsi, J., Bellows, T. & Van Driesche, R. (1992)Marginal attack rate, k -values and density dependence in the analysisof contemporaneous mortality factors. Researches on PopulationEcology, 34, 29–44.



Elkinton, J.S., Healy, W.M., Buonaccorsi, J.P., Boettner, G.H., Haz-zard, A.M. & Smith, H.R. (1996) Interactions among gypsy moths,white-footed mice, and acorns. Ecology, 77, 2332–2342.

Fitzgerald, T.D. (1995) The Tent Caterpillars. Cornell UniversityPress, Ithica, New York.

Glasgow, M.A. (2006) The effects of forest fragmentation on generalistpredation on forest tent caterpillars (Malacosoma disstria). MScthesis, University of Alberta, Edmonton, Canada.

Gould, J.R., Elkinton, J.S. & Wallner, W.E. (1990) Density-dependentsuppression of experimentally created gypsy moth, Lymantriadispar (Lepidoptera: Lymantriidae), populations by natural enemies.Journal of Animal Ecology, 59, 213–233.

Goulet, H. & Huber, J.T. (1993) Hymenoptera of the World: AnIdentification Guide to Families. Centre for Land and BiologicalResources Research, Agriculture Canada, Ottawa, Canada.

Herz, A. & Heitland, W. (2005) Species diversity and niche separationof cocoon parasitoids in different forest types with endemicpopulations of their host, the Common Pine Sawfly Diprion pini(Hymenoptera: Diprionidae). European Journal of Entomology, 102,217–224.

Hodson, A.C. (1939) Sarcophaga aldrichi Parker as a parasite ofMalacosoma disstria Hbn. Journal of Economic Entomology, 32,396–401.

Hodson, A.C. (1943) Birds feeding on forest insects. Flicker, 15,50–51.

Jactel, H. & Brockerhoff, E.G. (2007) Tree diversity reduces herbivoryby forest insects. Ecology Letters, 10, 835–848.

Jactel, H., Brockerhoff, E. & Duelli, P. (2005) A test of the biodi-versity – stability theory: meta-analysis of tree species diversityeffects on insect pest infestations, and re-examination of responsiblefactors. Forest Diversity and Function: Temperate and Boreal Sys-tems (ed. by M. Scherer-Lorenzen, C. Korner and E.-D. Schulze),pp. 235–262. Springer, New York, New York.

Klemola, T., Tanhuanpaa, M., Korpimaki, E. & Ruohomaki, K. (2002)Specialist and generalist natural enemies as an explanation forgeographical gradients in population cycles of northern herbivores.Oikos, 99, 83–94.

Klemola, N., Andersson, T., Ruohomaki, K. & Klemola, T. (2010)Experimental test of parasitism hypothesis for population cycles ofa forest lepidopteran. Ecology, 91, 2506–2513.

Koricheva, J., Vehvilainen, H., Riihimaki, J., Ruohomaki, K., Kai-taniemi, P. & Ranta, H. (2006) Diversification of tree stands asa means to manage pests and diseases in boreal forests: myth orreality? Canadian Journal of Forest Research, 36, 324–336.

Menalled, F.D., Marino, P.C., Gage, S.H. & Landis, D.A. (1999) Doesagricultural landscape structure affect parasitism and parasitoiddiversity? Ecological Applications, 9, 634–641.

Mondor, E.B. & Roland, J. (1998) Host searching and oviposition byLeschenaultia exul, a tachinid parasitoid of the forest tent caterpillar,Malacosoma disstria. Journal of Insect Behavior, 11, 583–592.

Myers, J.H. (1990) Population cycles of western tent caterpillars:experimental introductions and synchrony of fluctuations. Ecology,71, 986–995.

Nixon, A.E. (2011) Interactions between the forest tent caterpillar( Malacosoma disstria Hubner) and its natural enemies: the effects offorest composition and implications for outbreak spread. MSc thesis,University of Alberta, Edmonton, Canada.

Parry, D. (1995) Larval and pupal parasitism of the forest tent cater-pillar, Malacosoma disstria Hubner (Lepidoptera: Lasiocampidae),in Alberta, Canada. Canadian Entomologist, 127, 877–893.

Parry, D., Spence, J.R. & Volney, W.J.A. (1997) Responses of naturalenemies to experimentally increased populations of the foresttent caterpillar, Malacosoma disstria. Ecological Entomology, 22,97–108.

Parry, D., Spence, J.R. & Volney, W.J.A. (1998) Budbreak phenol-ogy and natural enemies mediate survival of first-instar forest tentcaterpillar (Lepidoptera: Lasiocampidae). Environmental Entomol-ogy, 27, 1368–1374.

Pelech, S. & Hannon, S.J. (1995) Impact of tent caterpillar defoliationon the reproductive success of black-capped chickadees. TheCondor, 97, 1071–1074.

Quayle, D., Regniere, J., Cappuccino, N. & Dupont, A. (2003) Forestcomposition, host-population density, and parasitism of spruce bud-worm Choristoneura fumiferana eggs by Trichogramma minutum.Entomologia Experimentalis et Applicata, 107, 215–227.

R Development Core Team (2010) R: A Language and Environmentfor Statistical Computing. R Foundation for Statistical Computing,Vienna, Austria.

Riihimaki, J., Kaitaniemi, P., Koricheva, J. & Vehvilainen, H. (2005)Testing the enemies hypothesis in forest stands: the important roleof tree species competition. Oecologia, 42, 90–97.

Roland, J. (1990) Interaction of parasitism and predation in the declineof winter moth in Canada. Population Dynamics of Forest Insects(ed. by A. D. Watt, S. R. Leather, M. D. Hunter and N. A. Kidd),pp. 289–302. Intercept, Andover, U.K.

Roland, J. (1993) Large-scale forest fragmentation increases theduration of tent caterpillar outbreak. Oecologia, 93, 25–30.

Roland, J. (2000) Landscape ecology of parasitism. Parasitoid Popu-lation Biology (ed. by M. E. Hochberg and A. R. Ives), pp. 83–99.Princeton University Press, Princeton, New Jersey.

Roland, J. (2005) Are the ‘seeds’ of spatial variation in cyclicdynamics apparent in spatially-replicated short time-series? Anexample from the forest tent caterpillar. Annales Zoologici Fennici,42, 397–407.

Roland, J. & Taylor, P.D. (1997) Insect parasitoid species respond toforest structure at different spatial scales. Nature, 386, 710–713.

Roland, J., Mackey, B.G. & Cooke, B. (1998) Effects of climate andforest structure on duration of forest tent caterpillar outbreaks acrosscentral Ontario, Canada. Canadian Entomologist, 130, 703–714.

Root, R.B. (1973) Organization of a plant-arthropod associationin simple and diverse habitats: the fauna of collards (Brassicaoleracea). Ecological Monographs, 43, 95–124.

Roth, D., Roland, J. & Roslin, T. (2006) Parasitoids on theloose – experimental lack of support of the parasitoid movementhypothesis. Oikos, 115, 277–285.

Rothman, L.D. & Roland, J. (1998) Forest fragmentation and colonyperformance of forest tent caterpillar. Ecography, 21, 383–391.

Russell, E.P. (1989) Enemies hypothesis: a review of the effectof vegetational diversity on predatory insects and parasitoids.Environmental Entomology, 18, 590–599.

Schmidt, B.C. & Roland, J. (2006) Moth diversity in a fragmentedhabitat: importance of functional groups and landscape scale in theboreal forest. Annals of the Entomological Society of America, 99,1110–1120.

Schmidt, B.C., Roland, J. & Wakarchuk, D. (2003) Evaluation ofsynthetic pheromones for monitoring forest tent caterpillar (Lep-idoptera: Lasiocampidae) populations. Environmental Entomology,32, 214–219.

Sippell, W.L. (1961) Key to the puparia of common dipterous parasitesof the forest tent caterpillar, Malacosoma disstria Hbn., in Ontario.Canadian Entomologist, 93, 433–436.

Southwood, T.R.E. & Comins, H.N. (1976) A synoptic populationmodel. Journal of Animal Ecology, 45, 949–965.

Stairs, G.R. (1966) Transmission of virus in tent caterpillar popula-tions. Canadian Entomologist, 98, 1100–1104.

Stark, E.J. & Harper, J.D. (1982) Pupal mortality in forest tentcaterpillar (Lepidoptera: Lasiocampidae): causes and impact on



populations in southwestern Alabama. Environmental Entomology,11, 1071–1077.

Tanhuanpaa, M., Ruohomaki, K., Kaitaniemi, P. & Klemola, T. (1999)Different impact of pupal predation on populations of Epirritaautumnata (Lepidoptera; Geometridae) within and outside theoutbreak range. Journal of Animal Ecology, 68, 562–570.

Turchin, P., Wood, S.N., Ellner, S.P., Kendall, B.E., Murdoch, W.W.,Fischlin, A. et al. (2003) Dynamical effects of plant quality andparasitism on population cycles of larch budmoth. Ecology, 84,1207–1214.

Williams, D.J.M., Parry, D. & Langor, D.W. (1996) Sampling andIdentification of Forest Tent Caterpillar Parasitoids in the Prairie

Provinces, Information Report NOR-X-345. Canadian ForestService, Northern Forestry Centre, Edmonton, Canada.

Witter, J.A. & Kulman, H.M. (1972) A Review of the Parasites andPredators of Tent Caterpillars (Malacosoma spp.) in North America,Agricultural Experiment Station Technical Bulletin 289. Universityof Minnesota, Minneapolis, Minnesota.

Witter, J.A. & Kulman, H.M. (1979) The parasite complex ofthe forest tent caterpillar in northern Minnesota. EnvironmentalEntomology, 8, 723–731.

Accepted 13 October 2011


Documents

Generalist predation on forest tent caterpillar varies with forest stand composition: an experimental study across multiple life stages