Experimental design and the outcome of preference-performance assays, with examples from mitoura butterflies (lycaenidae)

2008·4.4 MB·English

by  Matthew L Forister

Volume 62, Number 2 99
JournaloftheLepidopterists’Society
62(2),2008,99-105
EXPERIMENTAL DESIGN AND THE OUTCOME OF PREFERENCE-PERFORMANCE ASSAYS,
WITH EXAMPLES FROM MITOURA BUTTERFLIES (LYCAENIDAE)
Matthew L Forister
Dept,ofNaturalResources&EnvironmentalScience/MS 186, 1000ValleyRoad,UniversityofNevada, Reno, U.S.A;
email: [email protected]
ABSTRACT. Investigationsintoadulthostpreferenceandtheperformanceoflarvaeondifferenthostplantshaveplayedacentralrolein
ecologicalandevolutionaryplant-insectresearch. HereIpresenttwosetsofexperimentsthataddressaspectsoftheexperimentaldesignofpref-
erence-performanceassays, usingawell-studiedsystemol lycaenidbutterflies. First, I compareresultsfrom sequential,no-choiceopposition
assaystopreviousresults reportedfromsimultaneouschoicetestswithMitouranelsoni. Second, I describeanexperimentinwhichthelarvae
oftwocloselyrelatedspecies(M. nelsoniandMitouramuiri)wererearedinparallelonplantsinthelaboratoryandinthefieldtoassessthepo-
tentialinfluenceofenvironmentalconditionsonperformance. Resultsfromtheno-choicepreferenceassaysareconsistentwithpreviousresults,
suggestingthat,atleastinthissystem, thetwotypesofexperimentaldesignleadtosimilarconclusions. Theexperimentrearinglarvaeinthe
fieldandinthelaboratoryrevealedasignificanteffectofenvironmentonpupalweights,butdidnotdetectaspeciesbyenvironmentinterac-
tion. Thus forpupalweights, a laboratory-based studyis sufficientto compare performance between M. nelsoni andM. muiri. However, a
speciesbyenvironmentinteractionwasobservedfordevelopmenttime,whichhasimplicationsforhost-associatedspeeiationinthisgroupthat
wouldnothavebeendetectedinasolelylaboratory-basedstudy.
Additional keywords: Callophrys,specialization,choicetest,no-choicetest.
Preference-performance assays are used to address a related issues not discussed here, see Courtney et al.
range of questions in the ecology and evolutionary (1989), Singer& Lee (2000), Barton Browne &Withers
biology of herbivorous insects (Dethier 1954; (2002), Singer et al. (2002) Van Driesche & Murray
Thompson 1988; Jaenike 1990; Waekers 2007; Craig & (2004), and Mereader& Scriber(2007)). Inchoice tests,
Itami 2008). Preference refers to the choices made by hostplants are presented toan adult female orgroupof
ovipositing females or feeding individuals for different females in an array and the response is typically the
host plant species (Singer 2000), and performance number of eggs laid on the different plants in a set
refers to the development of juvenile stages on specific amount of time. In a no-choice assay, the behavioral
hosts. The questions addressed by preference- response (oviposition) is scoredwith plants in isolation,
performance experiments may be as simple as; will a oftensequentially,with plantsbeingpresentedone after
species ofinsectaccept aparticularspecies ofplant as a the othertoadults. Simultaneous choicetestshavebeen
host, and is the same plant a suitable host for larval criticized as being unrealistic, as different host plant
development? Questions may also involve genetic species may not be in immediate physical proximity in
variation and correlations among genetic elements: in the wild (Singer et al. 1992). On the other hand, an
particular, is preference for a particular plant species argument can be made that simultaneous choice tests
geneticallycorrelatedwith the abilityoflarvae to utilize are conservative: the juxtaposition of plants in an
the same host species (Via 1986; Thompson 1988; experimental arena could make it more difficult for an
Mayhew 1997). Experiments involving preference and ovipositing female to make a choice (since information
performance are also central to the practice of gatheredfromvolatileplantcues maybeoverlappingor
biocontrol, in which behavioral and physiological host mixed).
range must be determined before the release of a In any event, the two types of test, choice and no-
controlagent (Marohasy 1998). Theseand relatedtopics choice, potentially provide different and
havebeen reviewedbymanyauthors, includingJaenike complementary information (Withers & Mansfield
(1990), Thompson & Pellmyr (1991), Craig & Itami 2005). Consider a simple, hypothetical scenario: two
(2008), and Berenbaum & Feeny (2008). The goal of host plants (A and B) are used by a particular insect
this paper is to address two methodological and herbivore. In choice tests, plant A is overwhelmingly
experimental issues involved in preference- preferred to the exclusion of plant B, but in no-choice
performance assays: choiceversusno-choicepreference tests both plants receive a comparable number ofeggs
tests, and the influence of laboratory versus field from ovipositing females. It might be the case that the
conditions on performance experiments. volatile and tactile cues that characterize plant A are
Two ofthe more common ways in which preference sufficientlymorestimulatingtoovipositingfemales such
assays can be constructedinclude choice and no-choice that B is ignored in the presence of A. While in the
tests (for a review of other experimental designs and absence ofA, B is recognized as a suitable host andwill
100 Journalofthe Lepidopterists’ Society
be utilized. The choice test tells us not onlywhat could multiple hosts, as expressed in pupal weight and
happen in the wild when plants are interdigitated or in survival, has been previously described (Forister 2004,
veryclose proximity, but it tells us something about the 2005a). Here I focus on one host, a host ofM. muiri,
inherent ranking of host cues by the herbivore (e.g. and askifdifferencesbetween the twobutterflyspecies
Thompson 1993). The no-choice test on the otherhand in performance on that host are consistent between
mightgiveaclearerpictureofwhatcouldhappeninthe laboratoryandfieldenvironments.
wildas afemale moves from one isolatedpatchofplants
to another. Choice tests are more common in the Materials and Methods
literature, perhaps because they are logistically more Butterflies and plants. M. nelsoni andM. muiri are
efficient. What is not often tested (and which I address part ofacomplexofhost-specific lyeaenid butterflies in
herewith one butterflyspecies) is howoften the results North America associated with plants in the family
from choice and no-choice tests provide different lines Cupressaceae which have been the focus of recent
ofinformation (as in the hypothetical example above), investigations into the ecology of speciation (Nice &
or howoften results are congruent or redundant. Shapiro 2001; Forister 2004, 2005a, 2005b). M. nelsoni
Preference experiments are often rather contrived in is found in association with incense cedar at low to
that females are typically presented plants under middle elevationsin mesicforests from southern British
artificialconditions (cages orpreference arenas), andin Columbia to Baja California. M. muiri is an edaphic-
arrays or sequences that they might never encounter in endemic associated with cypress hosts (primarily
the field (though more realistic preference tests have MacNab cypress, Cupressus macnabiana A. Murray,
been conducted, e.g. Singer & Thomas 1988). In and Sargentcypress, Cupressussargentii Jepson) onlow
contrast to this, performance experiments need not be elevation, ultramafic soils such as serpentine in
quite so highlyabstracted from natural conditions: it is California (Gervais & Shapiro 1999).
possible to rear larvae in the field by confining them to The experiments described here used M. nelsoni
small cages or bags. Despite this, the majority of adults in preference experiments, and cateipillars of
performance experiments have addressed the both species in performance experiments. The M.
performance of larvae in laboratory conditions, often nelsoni adults consisted ofwild-caught and laboratory-
with larvae reared singly in petri dishes (Zalueki et al. reared individuals. Wild-caught individuals were taken
2002). Whatever measure of performance is taken from thefollowinglocationsin2004onthewestslopeof
(pupal weight, development time, etc.), it seems the Sierra Nevada Mountains near interstate 80: Drum
intuitively obvious that results may be biased by Powerhouse RoadandFangCrossing(see Forister2004
laboratory conditions. For example, the architecture oi for more details on these locations). Faboratory-reared
a given species ofplant might provide a microclimate adults were part ofa colony that was being maintained
thatallows larvae to feedthroughouttheheatofthe day, for other experiments at the University of California,
resultingin fasterdevelopmentthan on ahostthat does Davis. These individuals were the offspring of females
not have the same architecture (Alonso 1997). This collected from a number of populations in the Sierra
effect would only be apparent iflarvae were reared in Nevadaand North Coast Rangesinthepreviousseason.
the field. Other environment-dependent effects could Farvae used in performance experiments were
include interactionswith predators andparasitoids. generated from individuals reared and mated in the
I used two species of lyeaenid butterflies, Mitoura laboratory. For both M. nelsoni and M. muiri larvae
,
nelsoni Boisduval and Mitoura muiri H. Edwards, to were pooled from multiple lines without regard to
address these issues in the design of preference and genetic backgroundwithin species. In otherwords, M.
performance experiments. The oviposition behavior of nelson larvae were the product ofmatings between M.
M. nelsoni females in choice tests has been previously nelsoni adults from a number of locations throughout
described: they have consistent preferences for their California (and the same for M. muiri). These matings
host incense cedar Calocedms decurrens Torrey), are describedin detail in Forister (2005a).
(
laying the most eggs on that host in both four-way and Three host plant species were involved in these
two-waychoicetests involvingotherhosts ofMitoura in experiments: incense cedar (the host of M. nelsoni),
Northern California (Forister 2004, 2005a). Here I ask Sargent cypress and MacNab cypress (hosts of M.
if the preferences of M. nelsoni females for incense muiri). For preference experiments, incense cedar and
cedarareexpressedinno-choicetests as awillingness to Sargent cypress were collected from Goat Mountain in
layeggs on incense cedar and areticence to layeggs on the North Coast Range of California, where the two
an alternate host when encountered in isolation. The hosts grow sympatrically. For the performance
larval performance of M. nelsoni and M. muiri on experiments, MacNabcypresswas usedbothinthefield
Volume 62, Number 2 101
and through collection from one location, Knoxville cuttingsfrom thesetrees broughtbacktothelaboratory.
Public Lands, alsoin the North Coast Range. Caterpillars in the laboratorywere reared in groups of
Preference assays. Inordertoassessthe oviposition fiveinlargedrinkingcups nestedwithin smallercups so
behaviorofM. nelsoni inno-choiceassays, femaleswere that the cut ends ofbranches could be pushedthrough
confined individually with sprigs of host plants in holes in the larger cup and into water held in the
oviposition arenas (cylinders ofwire mesh, 3600 cm3 smaller cup. Upon pupation, pupae were weighed on a
).
Theywere exposed to one host for 24 hours, and then Mettler Toledo microbalance to the nearest hundredth
switchedtotheotherhost for24hours (thetwohosts, as ofamilligram. Caterpillarsthatbecamepartofthe field
mentioned above, were incense cedar and Sargent component were reared initially in the laboratory
cypress). The switch from one host to the other was through the first instar. They were then transferred to
done in the early morning of the second day, before the field, where theywere rearedto pupation in groups
butterflies were active. Experiments were only offive inspun meshbags enclosingtreebranches. Each
conducted for 48-hour periods because previous ofthe ten trees in the field had two bags (one M. muiri
experience with Mitoura butterflies had shown that bag and one M. nelsoni bag). Caterpillars in bags were
females become considerably less vigorous and egg- checked weekly and moved to new branches on the
laying begins to drop offafter 48 hours when they are same trees when foliage had been depleted. Upon
keptinagreenhouse in full sun (Forister, pers. obs.). At pupation,pupaewere removedfrom bags, broughtback
the start of the experiment, each female was to the laboratory and weighed. In addition to pupal
haphazardly assigned to one of two groups, with one weight, survival and days topupationwere recordedfor
group being confined first with incense cedar, and the both the laboratoryand field-reared individuals.
second group being confined first to Sargent cypress. Analyses of variance (ANOVA) using restricted
Sugar water was applied to the cages as an artificial maximum likelihood (REML) mixed models were used
nectar source that was readily consumed by butterflies to analyze results from performance experiments
throughout the experiment. The number of eggs on (Littell et al. 1996). Fixed factors in models included
plants was counted at the end of each interval as a species, location (fieldorlaboratory), andaninteraction
measure of host preference (Mitoura butterflies very between species and location. Random factors were
rarely ovisposit on any surface in preference arenas tree, and interactions between tree and species, and
otherthanthehostplants; andif eggswerefoundonthe between tree and location. Rearing group is not
side ofthe cage theywere not counted). included in models because values within groups (for
Results from preference assays were analyzed in two pupal weight, development time and survival) were
ways. First, the number ofeggs laid by each female on simply averaged prior to analysis (individuals within
the two hosts was treated as a pair in a nonparametric groups are not statisticallyindependent).
Wileoxon matched-pairs test. This analysis addressed No transformations were found to be necessary to
the question: which host received more eggs without meet the assumptions of ANOVA for pupal weight or
reference to the orderofthe hosts? Second, aWileoxon development time. Residual error from analysis of
rank-sum test was used to ask: does the first host survival (the fractionofindividuals survivingtopupation
encountered affect the number ofeggs laid on incense withineach rearinggroup)washighlynon-normal(even
cedar? In this case, each female is represented by one following arcsine transformation) due to the large
data point (the number of eggs laid on cedar), and number of groups in which survival was 100%.
females are identified as belonging to either the Therefore, two separate nonparametricWileoxon rank-
treatment that received incense cedar first or Sargent sum testswere performed tocompare survival between
cypress first. the two species in the laboratory and in the field. IMP-
Performance assays. The goal of performance IN software, version 7.0 (SAS Institute, Cary, NC,
assayswas to askifdifferences in performance between U.S.A.), and Kaleidagraph, version 3.6 (Synergy
the two butterfly species observed in the laboratory Software, Reading, PA, U.S.A.), were used for
(Forister2004, 2005a) are also observed in the field. To nonparametric analyses (both for survival data and
address this question, ten trees ofMacNab cypress, the preference results, described above), and PROC
host ofM. muiri, were selected at a field site that has MIXED in SAS, version 9.1 (SAS Institute, Cary, NC,
been studied previously (Knoxville, see Forister 2004). U.S.A.), was used for REMLanalyses ofvariance.
Trees were selected haphazardly within a small area
Results
(approximately 100 square meters), and caterpillars of
both M. muiri and M. nelsoni were reared to pupation Preference assays. A total of45 M. nelsoni females
simultaneously on these trees in the field and on were testedin no-choice assays usingincense cedar, the
102 Journalofthe Lepidopterists’ Society
host ofM. nelsoni, and Sargent cypress, the host ofM. (between the two species) is not affected by rearing
muiri. As can be seen in Fig. la, females laidamajority environment (Fig 2a, and note the insignificant species
oftheireggs on incense cedarin these no-choice assays by location interaction in Table 1). In contrast, rearing
(T = 4.63, P < 0.0001). The behavioroffemaleswas not environment did have a differential effect on the
influenced bythe orderin which plantswere presented development time of the two species: in the field, M.
to them: a comparable number of eggs was laid on muiri individuals reach pupation 4.62 days earlier than
incense cedar regardless of whether that host was M. nelsoni individuals (Fig. 2b, Table 2). In general,
presented first or second in sequence (Fig lb; Tj = - larvae of both species developed more slowly in the
1.11, P = 0.26). field, and this might be because they did not feed at
Performance assays. A total of 185 larvae were night: when checking the bags in the early morning, I
reared to pupation in 39 rearing groups (20 in the found larvae to be inactive, while larvae in the
laboratoryand 19 in the field; larvae from oneM. muiri laboratoryare capable offeeding throughout the night.
group in the field escaped). As has been observed in There were no significant differences between the
previous work (Forister 2004), M. nelsoni individuals
develop topupalweights that are greaterthan M. muiri
(on average 10% greater), even though the host in (a)
question is the natal host of M. muiri. The results
reported here demonstrate that this difference
(a)
cedar cypress
(b)
(c)
cedar cypress
Fig. 1. Resultsfrompreferenceassaysillustratedasboxplots.
Thesamedataisshownintwodifferentwaysin(a)and(b):data
shownin (a) isthe fraction ofeggs laidon thetwohosts,while
(b) showstheinfluenceofexperimental sequence in sequential
no-choiceassaysonovipositionbehavior. Inotherwords, in (b), Fig.2. Meansandstandarderrorsfromassaysofperformance
the datashown is the fraction ofeggs laidon incense cedarfor in the laboratory and in the field. Statistical results for pupal
femaleswhichwereexposedtothatplantfirst(theleftbox),and weight(a) anddaystopupation (b) areshowninTables 1 and2
forfemaleswhichwereexposedtocypressfirst(therightbox). respectively. Seetextformoredetailsrelatedtosurvival(c).
Volume 62, Number 2 103
survival ofM. nelsoni and M. muiri caterpillars in the
85
laboratory(T = 0, P = 1.0) orin the field (T = 0.51, and
; ;
P = 0.61) (Fig. 2c). T) 80
Differences among individual trees had a significant .£
effect on pupal weight (Table 1), but not on o> 75
development time (Table 2). Although tree had an E
effect on pupal weight, this was not influenced by 70
rearingenvironment, norwasthere asignificantspecies O)
'55
by tree interaction. In other words, larvae of both 3 65
species did better on certain trees, and this was true 7a5
whether larvae were reared in the field or on cuttings a3 60
from the same trees in the laboratory. In ordertobetter
visualize the influence of individual trees on pupal 55
70 75 80 85 90 95 100 105 110
weight. Fig. 3 shows thecorrelation betweenweights of
larvae reared in the field and in the laboratory. One Pupal weight (mg), lab
outlier has been excluded from the relationship shown
in Fig. 3: one M. nelsoni rearing group had high mean Fig. 3. Comparisonofpupalweights in the fieldversuspupal
weights in the laboratory. Eachpointcorresponds toagroupof
pupalweight in the field (80.03 mg), but unusually low larvaerearedonfoliagefromasingleplantinthefieldandin the
weight in the laboratory (66.9 mg). With the outlier laboratory. Eachhostplantisshowntwice,onceasthehostofM.
excluded, thecorrelationissignificant: Pearson product- Mmu.irmiuilrairvasequaanrdesonacreeMas.tnheelshoonsit).ofAMs.inngellesoonuitlliaerrvwaaes(ceixrcclleusdeadr,e
momentcorrelationof0.73, P = 0.0006;with theoutlier seetext f,ordetails.
includedthe correlation is 0.37, P = 0.12.
informationtono-choice tests. There are twoimportant
Discussion
caveatstothisconclusion. First, these results shouldnot
M. nelsoni females express a clear preference for be used to infer that the two types of choice test are
their natal host, incense cedar, in both choice tests equivalentinothersystems. Rather, the results reported
(Forister 2004, 2005a), and no-choice tests, as reported here highlightthe utilityol exploringbothtypes ofassay,
here (Fig. 1). Choice tests are more efficient from the andthepossibilitythatin somesystemschoicetests may
point of view ol experimenter effort: there is less be sufficient. Second,while itistruethatchoice andno-
manipulation inchoicetests, asplantsdonot needtobe choice assayswith M. nelsoni leadto similarconclusions
changedpartwaythroughthetest (ascomparedtoano- aboutthe relative rankingofthetwohosts byovipositing
choice design with sequential replacement of hosts). females, there may be situations in which no-choice
The results reported here suggest that, at least in the tests would still be uniquely useful. For example, no-
Mitoura system, choice tests provide equivalent choice tests couldbeusedtosurveyforvariationamong
Table 1. Results from analysisof pupalweights. Degrees of Table2. Resultsfrom analysisofdevelopmenttime (days to
freedom andF ratios are reportedforfixedeffects, covariance pupation). Degrees offreedom and F ratios are reported for
estimates andstandarderrors forrandomieffects. SignificantP fixedeffects, covariance estimates andstandarderrors forran-
valuesareshowninboldtext. domeffects. SignificantPvaluesareshowninboldtext.
Source NDF DDF F P Source NDF DDF F P
Species 1 26.3 16.8 0.00040 Species 1 13.3 13.2 0.00300
Location 1 26.3 97.3 < 0.0001 Location 1 26.3 97.3 < 0.0001
Speciesx Speciesx
location 1 26.3 0.21 0.65 location 1 26.3 0.21 0.24
Covariance SE P Covariance SE P
Tree 19.0 13.1 0.0095 Tree 0.354 0.93 0,33
Treexspecies 0 - - Treexspecies 0.945 1,36 0.24
Treexlocation 0 Treexlocation 0.836 1.34 0.26
.
104 Journalofthe Lepidopterists’ Society
females in preference for a less preferred host, while The present study suggests a previously undetected
such variation could potentially be harder to detect in component of local adaptation in M. muiri-. faster
choice tests where females always spend a majority of development thanM. nelsoni on MacNabcypress inthe
theirtime ovipositingon the preferred host. No-choice field. Whythis differencewould onlybe manifest in the
tests could also be used to study factors (such as egg field is not known, though one possibility is that M.
load) which may influence “motivation” and lead to the muiri larvae may be able to feed over a slightly wader
acceptance ofan otherwise less-preferred host (Singer range of temperatures than M. nelsoni larvae. Faster
etol. 1992). growth may reduce exposure to natural enemies
With the performance results reported here, it is (Williams 1999), orextreme climaticevents (Fordyce &
apparent thatacomparisonbetween thetwospecies for Shapiro 2003). In particular, faster development at low
at least one element of lanal performance (pupal elevations in the diy, inner North Coast Range of
weight) is not greatly influenced by rearing California might allow larvae to pupate before
environment. M. nelsoni pupaearebiggerthanM. muiri temperatures become unfavorably high (three days
pupae, and individuals reared in the laboratory are before the end of the experiment, a maximum daily
bigger than individuals reared in the field (Fig. 2a), but temperature of40 degrees Celsius was recorded at the
being reared in the laboratory or the field does not field site). Although the adaptive significance offaster
change the relative sizes of M. nelsoni and M. muiri developmentin the fieldisunknown,this is adifference
pupae. The foliage quality of individual trees was also between M. nelsoni and M. muiri that would not have
consistent across rearing environments (Fig. 3). The been observed in asolelylaboratory-basedstudy.
vast majority of performance experiments are done in
Acknowledgements
the laboratory (Zalucki et al. 2002), thus the results
reported here are heartening: not only may laboratory I thank S. L. Thrasher for assistance in rearing larvae and
tendingtoadultbutterflies, andA. M. Shapiro, E. A. Legerand
performance (as measured by pupal weight) be an G. W. Foristerforhelpincollectingplantsandbutterflies.
accurate reflection ofperformance in the field (at least
in the absence of natural enemies), but intraspecific Literature Cited
variation in plant quality may in some cases also be Alonso, C. 1997. Choosing aplace to grow. Importance ofwithin-
reasonablystudiedunderlaboratoryconditions. Osieret plantabioticmicroenvironmentforYponorneutamahalebella En-
tomol. Exp.Appl.83: 171-180.
al. (2000) reported a similar consistency between Barton Browne, L., & T. M. Withers. 2002. Time-dependent
performance in the laboratory and in the field on changesinthehost-acceptancethresholdofinsects:implications
particularplant genotypes using gypsy moth larvae and forhostspecificitytestingofcandidatebiologicalcontrolagents.
BiocontrolSci.Tech. 12: 677-693.
quakingaspen clones. Berenbaum, M. R., & P. Feeny. 2008. Chemical mediationofhost-
The performance results reported here are also plant specialization: The papilionid paradigm. In K. Tilmon
interesting in the light ofa scenario ofhost-associated (ed.).Theevolutionarybiologyofherbivorousinsects:s|pecializa-
tion,speciation,andradiation. Berkeley: Univ.ofCalif. Press.
speciation that has been described in Mitoura. Berlocher,S. IT,& L. Feder.2002. Sympatricspeciationinphy-
Differences in host preference are believed to be a key tophagousinsects:| Movingbeyondcontroversy?Ann. Rev. Ento-
mechanism in the diversification of this group (Nice & mol.47: 773-815.
Shapiro 2001; Forister 2004, 2005a), as has been Courftonreiyn,diSv.idPu,alG.hoKs.tCsehleecnt,io&n.AO.ikGoasrd55n:er5.5-6159.89.Ageneralmodel
suggested for a number of other phytophagous insect CraigT. P, & J. K. Itami. 2008. Evolutionofpreferenceandperfor-
systems in which adults mate and oviposit on their host mancerelationships.In K. |.Tilmon(ed.).Theevolutionarybiol-
plants (Berlocher & Feder 2002; Dies & Mallet 2002). ogyofherbivorous insects: specialization, speciation, and radia-
tion. Berkeley: Univ.ofCalif. Press.
Divergent host preferences are expected to evolve in Dethier, V. G. 1954. Evolution of feeding preferences in phy-
association with host-specific larval adaptations, tophagousinsects. Evolution8:33-54.
particularlywhendivergenceis insympatryorparapatry Drest.heMi.r,i&mpJ.orMtaalnlceet.in20s0y2m.paHtorsitcrsapceecsiaitnipolna.ntP-hfile.edTirnagnisn.seRc.tsSaoncd.
(Fry 2003) (which appears to be the case for Mitoura). Lond. B Biol. Sci.357:471-492.
M. nelsoni fits this model nicely: females have strong Fordyce.J.A., &A. M. Shapiro.2003. Anotherperspectiveonthe
slow-growth/high-mortality hypothesis: chillingeffects on swal-
preferences and larvae attain considerably larger pupal lowtaillarvae. Ecology84:263-268.
weights on incense cedar (largerthan M. nelsoni larvae Forister,VI.L.2004.Oppositionpreferenceandlarvalperformance
rearedonotherhosts ofMitoura in northern California, withinadiverginglineageoflycaenidbutterflies. Ecol. Entomol.
29:264-272.
and larger than other Mitoura larvae reared on incense . 2005a. Influence ofhost plant phenology on Mitoura nelsoni
cedar). In contrast, M. muiri females have strong host (Lepidoptera Lycaenidae). Ann. Entomol. Soc. Am. 98:
:
preferences but M. muiri larvae do not attain greater 295-301.
2005b. Independent inheritance ofpreference and perfor-
pupal weights or have higher survival on their natal m.anceinhybridsbetweenhostracesofMitourabutterflies(Lep-
cypresses relativetoM. nelsoni larvaeonthesamehosts. idoptera Lycaenidae). Evolution59: 1149-1155.
:
Volume 62, Number 2 105
Fry,J. D.2003. Multilocusmodelsofsympatricspeciation: Bushver- ,C. Stefanescu, &I. Pen. 2002. Whenrandom samplingdoes
susRiceversusFelsenstein. Evolution57: 1735-1746. notwork:standarddesignfalselyindicatesmaladaptivehostpref-
Gervais, B. R., &A. M. Shapiro. 1999. Distributionofedaphic-en- erencesinabutterfly. EcologyLett.5: 1—6.
demicbutterfliesintheSierraNevadaofCalifornia.GlobalEcol. Thompson, N. 1988. Evolutionaryecologyoftherelationshipbe-
J.
Biogeography8: 151-162. tween opposition preference and performance ofoffspring in
Jaenike, 1990. Host specialization in phytophagous insects. Ann. phytophagousinsects. Entomol. Exp.Appl.47:3-14.
J.
Rev. Ecol. Syst.21:243-273. . 1993. Preferencehierarchiesandtheoriginofgeographicspe-
Littell,R.C.,W.W. Milliken,&R. D.Wolfinger. 1996.SASsys- cialization in host use in swallowtail butterflies. Evolution 47:
tem formixedmodelscomputerprogram,version ByLittell, R. 1585-1594.
C„W.W. Milliken,andR. D.Wolfinger,Cary, NY. ,&O.Pellmyr. 1991.EvolutionofoppositionbehaPorandhost
Marohasy, [. 1998.Thedesignandinterpretationofhost-specificity preferenceinLepidoptera.Ann. Rev. Entomol.36:65-89.
tests forweedbiologicalcontrolwithparticularreferencetoin- Van Driesche, R. G., & T. Murray. 2004. OverPew oftesting
J.
sectbehaviour. BiocontrolNewsandInformation 19: 13-20. schemesanddesignsusedtoestimatehostranges. In R. G.Van
Mayhew, R 1997. Adaptivepatternsofhost-plantselectionbyphy- Driesche,andR. Reardon (eds Assessinghostrangesforpara-
J. .),
tophagousinsects.Oikos79:417P28. sitoidsandpredatorsusedforclassicalbiologicalcontrol:aguide
Mercader, R. & M. Scriber.2007. Diversificationofhostusein tobestpractice. ForsterHealthTechnologyEnterpriseTeam.
J., J.
twopolyphagousbutterflies: differencesinoppositionspecificity Via, S. 1986. Geneticcovariancebetweenoppositionpreferenceand
orhostrankhierarchy?Entomol. Exp.Appl. 125:89-101. larvalperformanceinaninsectherbivore.Evolution40:778-785.
Nice,C. C., &A. M. Shapiro. 2001. Populationgeneticevidenceof Wackers, F. L„J. Romeis,and P.van Rijn. 2007. Nectarandpollen
restrictedgeneflowbetweenhostracesinthebutterflygenusMi- feedingbyinsectherbivoresandimplicationsformultitrophicin-
toura (Lepidoptera : Lycaenidae). Ann. Entomol. Soc. Am. 94: teractions.Ann. Rev. Entomol.52:301-323.
257-267. Williams,I. S. 1999.Slow-growth,high-mortality-ageneralhypoth-
Osier,T. L.,S.Y. Hwang,&R. L. Lindroth.2000.Effectsofphyto- esis,orisit?Ecol. Entomol.24: 490-495.
chemicalvariationinquakingaspen Populustremuloidesclones Withers, T. M., & S. Mansfield. 2005. Choice orno-choicetests?
on gypsy moth Lyinantria clisparperformance in the field and Effectsofexperimentaldesignontheexpressionofhostrange.In
laboratory. Ecol. Entomol. 25: 197-207. M. Hoddle (ed.), Proceedings, 2nd international symposium of
Singer, M. C., D. Vasco, C. Parmesan, C. D. Thomas, & D. Ng. biologicalcontrolofarthropods. USDA ForestSerPce,WestVir-
1992. Distinguishingbetweenpreferenceandmotivationinfood ginia.
choice: An example from insect opposition. Anim. Behav. 44: Zalucki, M. P, A. R. Clarke, & S. B. Malcolm. 2002. Ecologyand
463-471. behaPoroffirstinstarlarvalLepidoptera.Ann.Rev.Entomol.47:
. 2000. Reducing ambiguity in describing plant-insect interac- 361-393.
tions: "preference", "acceptability"and"electiPty". EcologyLett.
3: 159-162. Receivedforpublication24November2007;revisedandaccepted
& ]. R. Lee. 2000. Discrimination within and between host 15April'2008
species bya butterfly: implications fordesign ofpreference ex- .
periments. EcologyLett.3: 101-105.