HMO.“
42.. .. n...
0%.»?
‘Hlilr
$17,

1...:

w w «a? ,1. {3.

1.931.!

$45M 5).!»

I...

 

in “EL yu‘Lrau‘ . . . , z c .
,,u..u-s . . _ uni” $33

A. cl...‘

2 r LIBRARY
205 6 Michigan State
University

This is to certify that the
dissertation entitled

SPECIES RANGES, HOST SELECTION, AND
HYBRIDIZATION: HOW INCREASED HYBRIDIZATION IS
LEADING TO HOST USE DIVERGENCE IN A
POLYPHAGOUS SIBLING SPECIES PAIR

presented by

RODRIGO J. MERCADER

has been accepted towards fulfillment
of the requirements for the

Ph. D degree in Entomology, Ecology,
Evolutionary Biology and
Behavior

 

iii/Mafia

 

7 Major Professor's Signature
5 M7 2%

Date

MSU is an aflinnative-acfion, equal-opportunity employer

 

 

 

PLACE IN RETURN BOX to remove this checkout from your record.
TO AVOID FINES return on or before date due.
MAY BE RECALLED with earlier due date if requested.

 

DATE DUE DATE DUE DATE DUE

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

5/08 K lProlecc8Pres/CIRC/Date0ue.indd

 

SPECIES RANGES, HOST SELECTION, AND HYBRIDIZATION: HOW
INCREASED HYBRIDIZATION IS LEADING TO HOST USE DIVERGENCE IN A
POLYPHAGOUS SIBLING SPECIES PAIR.

By

Rodrigo J. Mercader

A DISSERTATION

Submitted to
Michigan State University
in partial fulfillment of the requirements
for the degree of

DOCTOR OF PHILOSOPHY

Department of Entomology
and

Program in Ecology, Evolutionary Biology, and Behavior

2008

ABSTRACT

SPECIES RANGES, HOST SELECTION, AND HYBRIDIZATION: HOW
INCREASED HYBRIDIZATION IS LEADING TO HOST USE DIVERGENCE IN A

POLYPHAGOUS SIBLING SPECIES PAIR.
By
Rodrigo J. Mercader

In this dissertation I use the P. glaucus and P. canadensis system to understand
how climate change will affect interactions between organisms. In particular, how
increased genetic introgression caused by climate change may alter plant-insect
associations.The sibling species Papilio glaucus and Papilio canadensis meet in a narrow
hybrid zone believed to be maintained by temperature thresholds acting independently on
both species. However, due to recent climate change an increased movement of Papilio
glaucus into historically P. canadensis territory has been observed, which has lead to
greater genetic introgression (Scriber 2002), including the formation of a delayed
emerging phenotype or “late flight” first observed in 1999 (Scriber and Ording 2005;
Scriber et al. 2008). To understand how climate change will affect future interactions
between these butterflies I first tested the assumption that the P. canadensis range is
limited by rare high temperatures in late summer and/or autumn on pupal survival. Short
periods of high temperatures did not induce the high mortality rates required to be the key
factor limiting the range of P. canadensis. However, P. canadensis did exhibit a
considerably lower tolerance to high temperature extremes and conditions simulating
shorter/warmer winters than P. glaucus. Subsequently, I tested how increased genetic

introgression will likely affect a central interaction in plant feeding insects, host use. To

accomplish this I first investigated what the difference in host use between P. glaucus
and P. canadensis is. The results from this investigation indicated that the primary
difference between P. glaucus and P. canadensis is limited to a Z-linked shift in host
rank hierarchy due to an acceptance of Populus tremuloides Michx. (Salicaceae) and
reduced specificity for Liriodendron tulipifera L. (Magnoliaceae) in P. canadensis.
Finally, I assessed how the recombination of the parental genomes leading to the
formation of the “late flight” phenotype may have facilitated another major ecological
shifi, host use divergence. The results of this investigation indicate that the ovipositional
preference of this hybrid swarm is identical to that of the introgressing parental species,
P. glaucus. Due to the absence of the preferred hosts of P. glaucus where the “late flight”
occurs, this ovipositional pattern implies an apparent functional shift onto a secondary
host of both parental species, F. americana. In contrast, the larval host use abilities
represent a mixture of P. glaucus and P. canadensis, indicating divergence in larval host
use abilities has not taken place. However, high genetic variability (CV0) is present for
grth on F. americana and tradeoffs for larval performance on the preferred hosts of the
two parental species are evident; indicating a strong potential for specialization in larval
host use abilities to occur. This current scenario represents an instance where a shift in a
major ecological trait, host use, is likely occurring as a byproduct of a shift in an

unrelated trait (delayed emergence) leading to partial reproductive isolation.

To Becky Mercader

iv

TABLE OF CONTENTS

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

List of Tables _ ________ vii

List of Figures ..................................... ix
Chapter 1

General Introduction ............................................................................................................... 01

Introduction to the system .................................................................... 04
Chapter 2

Asymmetrical thermal constraints on the parapatric species boundaries of two widespread

generalist butterflies._ _ ______ 06

Abstract _ _ ....................................... 07

Introduction ______ 08

Materials and Methods __ 11

Insects ............................................... ll

Experiment 1: Short Term Heat Stress Simulations 12

Measurements Taken 12

Experiment 2: Prolonged Summer Simulations ,,,,,,,,,,,,,,, 13

Measurements Taken 14

Experiment 3: Warm Winter Simulation 14

Statistical Analyses _ _ _ ___________ 17

Results 18

Experiment 1: Short Term Heat Stress Simulations l8

Experiment 2: Prolonged Summer Simulations ,,,,,,,,,,,,,,,,,, 22

Experiment 3: Warm Winter Simulation 24

Discussion ______ 28

Conclusion ............... - ....................................................................... 32
Chapter 3

Diversification of host use in two polyphagous butterflies: differences in oviposition

specificity or host rank hierarchy? 34

Abstract ........................................ 35

Introduction _____________ _ 36

Materials and Methods ..................... - 41

Insects. _ _ 41

General methods for oviposition assays 43

Papilio canadensis oviposition preference 43

Differences in oviposition preference hierarchy between Papilio glaucus

and Papilio canadensism 45

Seven choice assays (2006) _ _ 45

Two choice assays (2006) 48

Seven choice assays (1993-1998) 48

Ovposition behavior of the “late flight” population 49

Results 50

 

 

Papilio canadensis oviposition preference _ 50
Differences in oviposition preference hierarchy between Papilio glaucus

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

and Papilio canadensis 52
Seven choice assays (2006) 52
Two choice assays (2006)___ _ 54
Seven choice assays (1993-1998)__ _ - - 60
Ovposition behavior of the “late flight” population 60
Discussion 66
Chapter 4
Hybridization leads to a host shift in a polyphagous butterfly sibling species pair ,,,,,,,,,,,, 74
Abstract __ _ _ _ 74
Introduction .................. 75
Materials and Methods _ _ _ __ __ ____________ 79
Study System ..... 79
Insect and Plant Sources _ _ 81
Oviposition Assays ___________ 82
Larval Growth Assays .............................. 83
Statistical Analyses ........................................... 84
P0pulation Contrasts ............................. . 84
Oviposition ...... 84
Survival 84
Growth... - - -- ........... 84
Analyses Within the “late flight” ______ __ 85
Results_ 87
Oviposition Assays ......................... 87
Larval Growth Assays -_ 87
Survival____ 87
Larval Growth 91
Estimates of genetic variance for larval performance 93
Discussion .................... 97
Oviposition Preference 97
Larval survival and growth (between populations) 99
Estimates of genetic variance for the “lateflight” population ,,,,,,,,,,,,,,,,,,,, 101
Conclusion __ 104
Chapter 5
General Conclusron -- 105
Appendix .......................................................... 107
Appendix 1 (Voucher SpecimeHS) ............................................ 108
Appendix 2 (Abstracts of Related Work) ...................................................... 110
References ...................................................................................................... l 15

 

vi

LIST OF TABLES

Table 2.1. Summary of temperature conditions applied in the three experiments presented
in chapter 2 ___________ 16

Table 2.2 The effect of spring emergence temperatures, high temperature spikes in
autumn, initial weight, and initial body condition on the successful emergence (survival)
of P. canadensis pupae. Analysis of Deviance performed by applying a GLM (Family =
Binomial, logit link) using R (R Development Core Team 2006) _ 19

Table 2.3. Summary of results for measurements taken on pupae and butterflies of P.
glaucus and P. canadensis from Experiment 1: Short Term Heat Stress Simulations.
Values in the measurement column represent the mean :1: SE for males and females. P-
values and 0323 in the autumn and spring treatment effect columns are derived from
ANCOVAs using Type “III” sums of square including sex as an independent variable and
initial body weight and initial body condition as covariates in R (R Development Core
Team 2006) ............ 21

Table 3.1. Factor loadings for the first three principal component axes of the proportion
of eggs laid in seven choice bioassays for the Papilio canadensis early flight” (n = 36)
and northern Michigan (n = 20) populations, and the Papilio glaucus Pennsylvania (n =
20) and Georgia populations (11 = 15) 47

Table 3.2. Bootstrapped variance estimates of the proportion of eggs laid by the Papilio
canadensis “early flight” population and the Spearman correlations between the
proportion of eggs laid for the three hosts that indicated any significant correlations
performed of the same population _____ 51
Table 3.3. Differences between the Papilio canadensis (Pc) early flight” (n = 26) and
northern Michigan (n = 20) populations, and the Papilio glaucus (Pg) Pennsylvania (n =
20) and Georgia populations (n = 15) for the individual scores from the first axis of the
PCA and the eighth axis of the Varimax rotation performed on the proportion of eggs laid
on seven choice arenas.)__ g ______________ 53
Table 3.4. Axes formed by Varimax rotation of the factor loadings for the principal
component axes of the proportion of eggs laid on seven-choice bioassays for the Papilio
canadensis early flight (EF) (n = 36) and northern Michigan (MI) (11 = 20) populations,
and the Papilio glaucus Pennsylvania (PA) (n = 20) and Georgia (GA) populations (n =
15) 55

vii

Table 3.5. Factor loadings for the first three principal component axes of the proportion
of eggs laid on three choice oviposition assays for the Papilio canadensis early flight” (n
== 94) and northern Michigan (11 = 27) populations, the Papilio glaucus Pennsylvania (11 =
93) and Georgia populations (11 = 20), and the “late flight” hybrid swarm population (11 =
29) ............................................... _ _ _ _ - _ 63

Table 3.6. Differences between the Papilio canadensis (Pc) early flight” (n = 94) and
northern Michigan (n = 27) populations, the Papilio glaucus (Pg) Pennsylvania (11 = 93)
and Georgia populations (n = 20), and the “late flight” hybrid swarm population (HS) (11
= 29) for the individual scores from the first two axis of the PCA performed on the
proportion of eggs laid on three choice arenas. Populations with different letters within a
column indicate significant differences from Bonferroni corrected Mann-Whitney tests
(P<0.05) __ 64

Table 4.1. P-values for pairwise Non-Parametric MANOVA’s contrasting the
ovipositional behavior in seven choice arenas. Upper triangle represent uncorrected P-
values and the lower triangle Bonferroni corrected P-values____ ______ 89
Table 4.2. Summary of variance estimates derived from the “late flight” population for
weights (mg) after six days of growth on five plants using GLMMs. Parameters
represented are the square root of genetic variance (\/VG) and residual variance (\IVR) 3:
standard deviation (from MCMC estimates), and the corresponding coefficient of genetic
variance (CVG), coefficient of residual variance (CVR), and broad sense heritability
(H2). _ _ __ _ _ _ 94

viii

LIST OF FIGURES

”(EF) population, 20 P. canadensis females from the northern Michigan (NMI) Figure
2.1. Percent survival of P. glaucus and P. canadensis pupae placed in environments
simulating heat waves in early Autumn (3 7°C days and 22°C nights for 0, l, or 2 weeks)
and emerged in environments simulating long springs, “cool” spring emergence
conditions (16°C days and 10°C nights) or short springs, “warm” spring emergence
conditions (24°C days and 20°C nights). Dark lines represent P. glaucus and gray lines P.
canadensis. For each species solid lines represent pupae emerged in chambers set at cool
emergence conditions, and dashed lines pupae emerged in chambers at warm emergence
conditions ................. - _ 20

=““= “ “ ““ u==OOOOOOIIIIIQI====“‘ ‘ ““ v==w===-‘

Figure 2.2. Percent survival of P. canadensis pupae maintained at 18°C, 22°C, or 26°C
(a) for one month or (b) two months during autumn. Pupae were subsequently emerged
“cool” spring emergence conditions (14°C) or “warm” spring emergence conditions
(22°C). Survival significantly lower than expected values based on binomial tests are
represented by * for 90%, ** for 60%, *** for 50%, and **** for 40% 23

Figure 2.3. Percent survival of P. glaucus and P. canadensis pupae maintained in
chambers set at continuous 4°C (4), alternating 12 hours at 10°C and 12 hours at 6°C
(10:6), or continuous 10°C (10) from midwinter to early spring. 26

Figure 2.4. Percent weight loss of P. glaucus and P. canadensis pupae maintained in
chambers set at continuous 4°C (4), alternating 12 hours at 10°C and 12 hours at 6°C
(10:6), or continuous 10°C (10) from midwinter to early spring. Within each species
means followed by the same letter are not significantly different at the p<0.05 level
(Tukeys HSD) ............................. - - __ __ ......... - -_ __ 27
Figure 3.1. Mean percentage + SE of eggs laid by 93 female Papilio glaucus from
Lancaster Co., PA, and 94 female Papilio canadensis from the “early flight” found in the
Battenkill River Valley in the New York/ Vermont border. Leaves present in the bioassay
were L. tulipifera, P. serotina, and P. tremuloides. The “other” category represents eggs
laid on the paper or plastic of the oviposition arena _ __ _ 40

Figure 3.2. Mean percentage + SE of eggs laid by the 36 Papilio canadensis (Pc) females
from the “early flightpopulation, 20 Papilio glaucus (Pg) females from the Pennsylvania
population, and 15 P. glaucus females from the Georgia (GA) population. Leaves
present in the bioassay were L. tulipifera, P. tremuloides, F. americana, P. trifoliata, B.
papyriflzra, P. serotina, and L. benzoin. The “other” category represents eggs laid on the
Paper or plastic of the oviposition arena ........................................................... - _ 56

ix

Figure 3.3a. Median proportion of eggs laid, with interquartile ranges, by Papilz'o glaucus
(Pg) and Papilio canadensis (Pc) butterflies on L. tulipzfizra (Lt) in two—choice assays.
Leaves present in the bioassays were L. tulipifera and P. serotina (Ps) or L. tulipifera and
P. tremuloides (Pt). For each leaf the pairwise differences between the proportions of
eggs laid on it in the different arrays (host and butterfly set-ups) were analyzed using
Mann-Whitney tests. Arrays with the same letter did not have a significant difference at a
Bonferroni corrected a<0.05 __ 57

Figure 3.3b. Median proportion of eggs laid, with interquartile ranges, by P. glaucus and
P. canadensis butterflies on P. serotina in two-choice assays. Leaves present in the
bioassays were P. serotina and L. tulipifera or P. serotina and P. tremuloides. For each
leaf the pairwise differences between the proportions of eggs laid on it in the different
arrays (host and butterfly set-ups) were analyzed using Mann-Whitney tests. Arrays with
the same letter did not have a significant difference at a Bonferroni corrected a<0.05_,,,58

Figure 3.30. Median proportion of eggs laid, with interquartile ranges, by P. glaucus and
P. canadensis butterflies on P. tremuloides in two-choice assays. Leaves present in the
bioassays were P. tremuloides and L. tulipifera or P. tremuloides and P. serotina. For
each leaf the pairwise differences between the proportions of eggs laid on it in the
different arrays (host and butterfly set-ups) were analyzed using Mann-Whitney tests.
Arrays with the same letter did not have a significant difference at a Bonferroni corrected
a<0.05 ,_ 59

Figure 3.4. Mean percentage of eggs laid by Papilio glaucus (n = 38) from Ohio and
Kentucky, Papilio canadensis (n = 20) from Alaska and northern Michigan, and F1
hybrids (n = 30) from P. canadensis mothers and P. glaucus fathers. Leaves present in
the bioassay were L. tulipiflzra, P. tremuloides, F. americana, P. trifoliata, L. benzoin, P.
serotina, and R. cathartica===== _ 61

Figure 3.5. Mean percentage + SE of eggs laid by 94 Papilio canadensis (Pc) females
from the “early flight” population (EF), 27 P. canadensis females from the northern
Michigan population (NMI), 29 females from the “late flight” (LF) hybrid swarm (HS)
population, 25 Papilio glaucus (Pg) females from the Georgia population (GA), and 93 P.
glaucus females fi'om the Pennsylvania (PA) population. Leaves present in the bioassay
were L. tulipifera, P. serotina, and P. tremuloides. The “other” category represents eggs
laid on the paper or plastic of the oviposition arena ,,,,,, 62

Figure 4.1. Mean percentage + SE of eggs laid by the 36 Papilio canadensis females
from the “early flight” population (2006), 26 females from the “late flight” hybrid swarm
(2007), and 20 Papilio glaucus (Pg) females from Pennsylvania (2006). Leaves present in
the bioassays were L. tulipifera, P. tremuloides, F americana, P. trifoliata, B. papyrifera,
P. serotina, and L. benzoin. The “other” category represents eggs laid on the paper or
plastic of the oviposition arena _ __ - __ _ 88

 

Figure 4.2. Mean percent survival + SE of the “late flight” population (LF), of the P.
canadensis Northern Michigan population (NMI), of the P. glaucus Georgia population,
and of the Michigan hybrid zone population (HZ). The five leaf species on which assays
were performed are L.tulipifera, P. tremuloides, F. americana, P. serotina, and P.
trifoliata. Within each host species population means followed by the same letter are not
significantly different at the p<0.05 level (Tukeys HSD)" _ , _ 90

Figure 4.3. Mean Z-score for larval growth + SE of the “late flight” population (LF), of
the P. canadensis Northern Michigan population (NMI), of the P. glaucus Georgia
population, and of the Michigan hybrid zone population (HZ). The five leaf species on
which assays were performed are L. tulipz‘fera, P. tremuloides, F. americana, P. serotina,
and P. trifoliata. Within each host species population means followed by the same letter
are not significantly different at the p<0.05 level (Tukeys HSD)“ 92

Figure 4.4. Figure 4. Spearman correlations of mean family Z-scores for growth,

performed between pairs of five host species. Host labels are L. tulipifera (Lt), P.
tremuiloides (P. tre), F. americana (F. a), P. serotina (P. s), and P. trifoliata (P. tri) ,,,,,,,,,, 9 6

xi

CHAPTER 1

GENERAL INTRODUCTION

The recent environmental changes taking place have highlighted our need to
better understand how interactions between organisms evolve. Rapid environmental
changes have not only lead to population extinctions, but the novel selection pressures
placed upon organisms have lead to several examples of rapid evolution (e.g. examples
reviewed in Thompson 1998; Palumbi 2001 ). These instances of rapid evolution have
included well documented shifts in the interactions between plant feeding insects and
their hosts due to plant invasions (Strauss et al 2006; Carroll 2007; Singer et a1. 2008) and
land use change (Singer et al.2008). However, little is known about how climate change
may affect plant-insect associations, other than through local extinctions. In particular,
how increased gene flow may affect host use traits and alter the associations between
plant and insects. This issue is of particular importance in the Lepidoptera as
hybridization is very common in this group, occurring in > 15% of all species (Sperling
1990; Presgraves 2002). Global climate change has already affected the geographic
ranges of numerous species, with warming most often resulting in shifts in species
distributions of insects toward higher latitudes or altitudes (Thomas et al.2001; Hill et
al.2002; Parmesan and Yohe 2003; Root et al.2003; Walther 2005; Parmesan 2006), thus
greatly increasing the likelihood and/or rate of hybridization.

Due to their high diversity, plant feeding insects have been the focus of multiple
studies on speciation and adaptive radiation. This high diversity of plant feeding insects

has often been attributed to their diversification on the wide chemical diversity of plants

(e.g. Ehrlich & Raven, 1964; Mitter et al., 1988; Futuyma, 1989; Weingartner et al.,
2006). Although divergence through specialization has often been considered a primary
mechanism for diversification in plant feeding insects, recent reviews have indicated a
high level of transition from specialist to generalist phases in insect lineages (Janz et
al.2001; Nosil 2002). In addition, as described by Janz et al. (2006) diversification
through specialization alone would quickly reach a dead end as lineages become
increasingly specialized. Instead, J anz et al.(2006) propose a hypothesis whereby insect
lineages oscillate between specialized and generalized forms, in which initial divergence
occurs in the generalized phase and subsequent specialization leads to greater
diversification.

Relatively little is known about how host shifts occur during polyphagous stages
of insect lineages, as the vast majority of this work has been conducted on monophagous
or oligophagous species (e.g., Carroll & Boyd, 1992; Radtkey & Singer, 1995; Menken,
1996; F eder et al. , 1998; Abraharnson et al. , 2001; Hora et al. , 2005; Ohshima &
Yoshizawa, 2006; Stastny et al., 2006; but see Thompson, 1998; Janz et al., 2001 for
shifts between specialized and generalized forms). This bias is likely a reflection of the
greater number of specialist insects and the relatively easier task of detecting host shifts
in specialized insects.

The polyphagous and widely distributed species of the Papilio glaucus group are
probably the most widely studied non-pest generalist insect group, particularly the sibling
species Papilio glaucus and Papilio canadensis. The large ecological knowledge base we
have for these insects makes them an ideal model system to study how climate warming

will affect the interactions between plant feeding insects and their host plants. These two

sister species share many ecological similarities, can readily produce fertile hybrid
offspring (e. g. Scriber 1998), and until recently were considered the same species (Hagen
etal.1991). However, despite their similarities they exhibit significant differences in host
plant use. In particular, tulip tree, Liriodendron tulipifera (Magnoliaceae), the preferred
host of P. glaucus, is toxic to P. canadensis larvae, while quaking aspen, Populus
tremuloides (Salicaceae), the preferred host of P. canadensis, is toxic to P. glaucus.
Historically these two species have had a narrow, but extensive, hybrid zone occurring at
the ecotone between temperate deciduous and boreal forests. Recent climate change has
lead to an increased emigration of P. glaucus into historically P. canadensis territory,
which in turn has lead to increased hybridization between the two species. This has
created an allochronically separated hybrid swarm population (Scriber and Ording 2005,
Scriber et al.2008).

In this dissertation I use the P. glaucus and P. canadensis system as a model
system to understand how climate change is likely to affect the interaction between
organisms. In particular, I explore how increased genetic introgression caused by climate
change may alter plant-insect associations. I do this by first testing the factors limiting the
distribution of these insects (Chapter 2), then identifying the factors underlying the
differences in host selection between these two species (Chapter 3), and finally by
assessing the host preference and performance of the hybrid swarm population produced

by the genetic introgression of P. glaucus into P. canadensis (Chapter 4).

INTRODUCTION TO THE SYSTEM

The Eastern Tiger Swallowtail, P. glaucus (L.) and the Canadian Tiger
Swallowtail, P. canadensis are generalists in both the number of tree species they
consume (30-40 species in 14 plant families; Scriber 1988) and in the temperature
regimes they experience (combined ranges extend from Southern Florida USA. to
Central Alaska USA). The Northern range limit of P. glaucus coincides with the
Southern range limit of P. canadensis in a narrow but extensive hybrid zone at the
ecotone between boreal and temperate deciduous forests (Scriber er al. 2003). This zone
coincides with the thermal limit for the completion of two generations (1556-1611 C
Degree days base 10 °C), and not surprisingly marks the range of the facultative
diapauser P. glaucus with a minimum of two generations per year (Scriber 1996). This
hybrid zone also delimits the Southern range of the obligate diapauser P. canadensis,
which is limited in part by lower pupal survival in warmer conditions (Scriber et al. 2002;
Mercader and Scriber 2008a) as well as likely interactions with P. glaucus and shifts in
host plant availability.

Despite being highly polyphagous, P. glaucus and P. canadensis have strong
ovipositional preferences for particular hosts, including local adaptation in ovipositional
specificity in both species (Bossart and Scriber 1995; Scriber 1996b, 2002b), and
significant differences between the two species (Scriber et al. 1991, 1995, 2003; Mercader
and Scriber 2007). In addition, they exhibit significant differences in their growth rates
on different hosts, including significant differences between each other (Reviewed in

Scriber etal.1995; Scriber 1996; Scriber et al. 2003). In particular, the preferred host of P.

glaucus, tulip tree, Liriodendron tulipifera L. (Magnoliaceae), is toxic to the larvae of P.
canadensis, whereas one of the most common hosts of P. canadensis, quaking aspen,
Populus tremuloides Michx. (Salicaceae), is toxic to the larvae of P. glaucus.
Additionally, larval host use abilities differ significantly for some secondary hosts such
as spicebush, Lindera benzoin (L.) Blume (Lauraceae), which is a marginal host for P.
glaucus and toxic for P. canadensis and paper birch, Betula papyrifera Marshaff
(Betulaceae), which is an excellent host for P. canadensis, but only marginal for P.
glaucus. Despite these differences, these two species do have multiple secondary hosts in
common. Among these are white ash, Fraxinus americana L. (Oleaceae), black cherry, P.
serotina, and Hop tree, Ptelea trifoliata L. (Rutaceae) [notez P. canadensis will rarely
encounter P. trifoliata], which are all excellent hosts for the larvae of P. canadensis and
P. glaucus. In addition, adult female ovipositional preferences for these secondary hosts
are essentially identical (Mercader and Scriber 2007).

A recent increase in immigration of P. glaucus into historically P. canadensis
territory due to climate warming has lead to increased hybridization between these two
species. This has resulted in the formation of a univoltine allochronically separated
hybrid swarm population (Scriber and Ording 2005, Scriber et al.2007). This hybrid
swarm has been noted, at the population level, to have larvae survive on both L. tulipifera
and P. tremuloides (Scriber et al. 2007), and three choice oviposition bioassays indicate

its ovipositional behavior may follow that of P. glaucus (Mercader and Scriber 2007).

CHAPTER 2

Mercader RJ, Scriber JM (2008) Asymmetrical thermal constraints on the parapatric
species boundaries of two widespread generalist butterflies. Ecological
Entomology (In Press).

ASYMETRICAL THERMAL CONSTRAINTS ON THE PARAPATRIC SPECIES

BOUNDARIES OF TWO WIDESPREAD GENERALIST BUTTERF LIES.

Abstract

The sibling species Papilio glaucus and Papilio canadensis meet in a narrow
hybrid zone believed to be maintained by temperature thresholds acting independently on
both species. Here we test if this assertion is true for the cold adapted species, P.
canadensis, which is presumed to be limited by rare high temperatures in late summer
and/or autumn on pupal survival. Three experiments were conducted examining the
effects of 1) short periods of high temperature stress in autumn, 2) prolonged warm
temperatures in autumn, and 3) temperatures simulating warmer winters/ longer springs
upon the survival of P. canadensis and P. glaucus. Results indicated that short periods of
high temperatures did not induce the high mortality rates required to be the key factor
limiting the range of P. canadensis. However, P. canadensis did exhibit a considerably
lower tolerance to high temperature extremes and conditions simulating shorter/warmer
winters than P. glaucus. In combination, differences in temperature tolerance throughout
the pupal stage are likely to be a significant factor in maintaining the southern range limit
of P. canadensis. Further warming as may occur during climate change, particularly in
winter and spring, will likely affect the dynamics of southerly populations of P.

canadensis.

Introduction

The factors determining species ranges have been a central focus of ecological
research, as they represent the thresholds of tolerance for organisms to biotic and abiotic
factors. Current changes in global climate patterns present a new challenge to organisms
as increases in global temperature are likely to increase the likelihood of both population
extinctions (e.g. McLaughlin et al., 2002; Hoyle & James, 2005; Shoo etal., 2005) and
species range expansions (e. g. Parmesan et al. , 1999; Thomas et al. , 2001; Walther et al.,
2002; Crozier, 2004; Battisti et al. , 2005). These shifts are likely to drastically alter the
species composition of regions, particularly those at ecotones where many range limits
overlap. Therefore, as the abiotic and biotic environment change, our ability to predict
shifts in insect community composition will depend on our understanding of the factors
that limit their ranges.

The environmental factors that define range limits in organisms, and the
mechanisms through which they act are necessary in order to understand how species
distributions may change. For temperate insects two major environmental constraints are
the length of the growing season and large fluctuations in temperatures throughout the
year. Constraints in the length of the growing season can be seen in the number of
organisms exhibiting geographical variation in the number of generations per year
(Tauber & Tauber, 1981; Ayres & Scriber, 1994). Constraints imposed by fluctuating
temperatures can be seen in the prolonged periods of inactivity most insects undergo in
stages that can tolerate environmental extremes (e. g. diapausing pupae). These
adaptations allow many species to maintain extensive geographic ranges, often with

locally adapted populations differing in their thresholds for diapause induction, duration

of diapause, and rate of diapause and non-diapause development (Tauber &Tauber,
1981). The factors that define the ranges of these widespread species are likely to signify
important physiological and/or ecological tradeoffs.

The Eastern tiger swallowtail, Papilio glaucus (L.) and the Canadian tiger
Swallowtail, Papilio canadensis (R & J) (Lepidoptera: Papilionidae, these
Papilio =Pterourus) are extreme generalists in both the number of tree species they
consume (30-40 species in 14 plant families; Scriber, 1988) and in the temperature
regimes they experience. Their extensive ranges, which combined extend from Southern
Florida to Central Alaska USA, cover a vast range of biotic and abiotic conditions. The
Northern range limit of P. glaucus coincides with the Southern range limit of P.
canadensis in a narrow but extensive hybrid zone at the ecotone between boreal and
temperate deciduous forests (Scriber, 1994). While the northern range limit of P. glaucus
is defined by the thermal environment that allows the completion of two generations
(Scriber, 1994), the factors determining the southern range limits of the univoltine P.
canadensis are not clearly understood.

The first flight of P. glaucus and the only flight P. canadensis occur as spring
begins, and by late May the flights of both species co-occur near the hybrid zone (and
potentially from the central United States to Alaska; Scriber 1996). Although the flights
of both species co-occur and both overwinter as pupae, P. glaucus has two or more
generations per summer and is therefore less likely to encounter high temperatures in an
overwintering pupal stage than the univoltine P. canadensis. Scriber et al. (2002) found
extremely high mortality rates (near 100%) of P. canadensis and P. glaucus pupae

exposed to high temperatures during emergence in spring (30, 33, & 36 ° C). The low

tolerance observed for high temperature in P. canadensis pupae would pose a strong
limiting factor on the southern range of P. canadensis even if high temperature extremes
were relatively rare. This lead Scriber et a1. (2002) to conclude that rare periods of
extreme high temperatures during late summer or early Autumn were the key factor
determining the southern range limit of P. canadensis (Scriber et al., 2002). Extreme
temperature events have been increasing and are expected to continue increasing (e. g.
Easterling et al. , 2000; Clark et al. , 2006) and P. canadensis populations above the
hybrid zone may begin to encounter such conditions with greater frequency. If rare
periods of extreme high temperatures are the key factor limiting the range of P.
canadensis, southern populations of P. canadensis could be at risk of local extinction.

However, while summer and autumn temperatures may occasionally be at or near
30° C, spring emergence temperatures at the P. canadensis southern range limit are
unlikely to reach daily temperature averages above 30° C. For example, in Kalamazoo
Michigan (located South of the hybrid zone) between 1948 and 1998 there were no days
where the maximum and minimum temperature averaged 28° C or more between early
April and early June (KBS, LTER site weather station). In contrast, during the same time
period there were 10 separate years in which daily temperatures averaged 28° C or more
for at least three consecutive days between late July and the end of September (KBS,
LTER site weather station).

The tolerance for extreme temperatures is likely to be significantly higher prior to
the onset of winter when diapausing pupae are in a quiescent phase than in the spring just
prior to emergence (when the imago is forming inside the pupa). Therefore, in this study

we re-evaluate the hypothesis that mortality induced by short term heat stress prior to

10

winter is the primary factor determining the southern range limit of P. canadensis. As
regional changes in temperature are not only likely to differ in high temperature spikes,
we also tested the impact of other forms of heat stress that are likely to be encountered by
P. canadensis as climate patterns change.

In particular, global climate change is likely to create shorter winters in temperate
regions, where winter temperatures are increasing at a higher rate than summer
temperatures (Walther et al., 2002; Schwartz et al., 2006). Therefore, pupae will not only
be exposed to higher incidences of temperature spikes prior to winter, but also to
prolonged periods at or near metabolic thresholds (longer springs and auturnns). Here we
evaluate the effects of prolonged warm temperatures prior to the onset of winter (long
autumn) on P. canadensis pupae and the effects of prolonged periods at temperatures in
between winter conditions and those generally experienced prior to emergence (long

spring) on pupae of P. canadensis and P. glaucus.

Materials and Methods

Insects

Pupae were obtained by mass rearing larvae on Prunus serotina branches from
eggs laid by wild-caught females in 2005 and 2006. P. canadensis females were collected
from the first flight in the Battenkill River Valley area at the New York/Vermont border

U.S.A (43.2° N latitude), and P. glaucus females were collected in Lancaster Co. in

south-eastem Pennsylvania, USA (40° N latitude).

ll

Experiment 1 .' Short Term Heat Stress Simulations

On 27 September, 2005 pupae of P. canadensis (n= 240) and P. glaucus (n=240)
were randomly divided into three groups containing 80 pupae of each species. The first
group was kept for 14 days in a chamber maintained at 12 hrs of light at 37°C and 12 hrs
of dark at 22°C, the second group was kept in the same chamber for 7 days and then
placed in a chamber set at 22°C (12: 12 hrs Light:Dark) for 7 days, and the third group
was placed for 14 days in the chamber maintained at 22°C (12:12 hrs Light:Dark). After
this period all pupae were kept for 3 days at 22°C (12:12 LzD), followed by 3 days at
14°C (18:6 LzD), and subsequently maintained in the dark at 3-4°C throughout the
remainder of the winter.

On 24 April 2006 pupae were removed from winter storage and placed at 14°C (18:6
LzD) for three days and then equally separated into four chambers. Two chambers were
set to “warm” emergence conditions set at18 hrs of light at 24°C and 6 hrs of dark at
20°C (avg. = 22°C), and two chambers set to “cool” emergence conditions at 18 hrs of
light at 16°C and 6 hrs of dark at 10°C (avg. = 14°C). Emergence conditions were
selected to represent approximate average temperatures encountered during early spring

(“cool”) and late spring (“warm”) in areas near the southern edge of the hybrid zone.

Measurements taken. On September 26 & 27, 2005 and on April 23 & 24, 2006 all
pupae were weighed and their length was recorded using a slide calliper. Pupal weight
loss was recorded as the weight prior to treatments minus the pupal weight after
overwintering. Pupal condition (used as a covariate) was estimated separately for P.

canadensis and P. glaucus as the residuals of a regression of weight on length (P.

12

canadensis r2 = 0.87, P. glaucus r2 = 0.83). In the spring of 2006 the emergence success
and condition of the wings of the emerging adults was recorded. The condition was
recorded as not crumpled (=normal), ‘A crumpled, l/2 crumpled, % crumpled, and
completely crumpled. Pupae that were dead and butterflies that were unable to fully
emerge or unable to fly (wings 3/4 or completely crumpled) were considered dead.

As butterflies emerged they were allowed to dry at which time they were placed
in envelopes and kept in a refrigerator at approximately 6°C for 4-8 hours and then their
weights, wing length, and abdomen size were recorded. Abdomen size was recorded by
measuring the abdomen length (1), width (w), and thickness (t) with a slide calliper and
calculating the size as an ellipsoid V=7rlwt/6. Percent weight lost was calculated as (initial

pupal weight - emergence weight)/initial pupal weight.

Experiment 2: Prolonged Summer Simulations

From 19 August, 2005 to 19 September, 2005 pupae of P. canadensis were placed
in chambers maintained at a photoperiod of 18hrs light and 6 hrs dark and temperatures
of either 18°C (n= 84), 22°C (n= 146), or 26°C (n= 96). On 19 September, 2005 half the
pupae at each temperature were removed and placed at 14°C for 3 days and subsequently
kept in the dark at 3-4°C throughout the winter. The remaining half was kept for an
additional 30 days at their respective temperatures to simulate a prolonged summer and
then slowly cooled to simulate a progressive drop in temperature. Pupae were slowly
cooled by sequentially placing pupae for two weeks in chambers set at 4°C lower than the
previous chamber until they were placed at 14°C, and subsequently kept in the dark at 3-

4°C throughout the winter. Therefore pupae starting at 26°C were placed for two weeks

13

at 22°C, 18°C, 14°C, and finally at 3-4°C for the winter; while pupae starting at 18°C
were only placed at 14°C for two weeks and then placed at 3-4°C for the winter.

On 2 May 2006 pupae were placed at 14°C (18:6 L:D) for three days and then
equally separated into four chambers for spring emergence of adults. As in the previous
experiment pupae were placed in two chambers set to “cool” emergence conditions at
14°C (18:6 L:D), and two chambers set to “warm” emergence conditions at 22°C (18:6
L:D). As in the previous experiment emergence conditions were selected to represent
approximate average temperatures encountered during early spring (“cool”) and late

spring (“warm”) in areas near the southern edge of the hybrid zone.

Measurements taken. On 19 & 20 September 2005, and again on 1 & 2 May 2006 all
pupae were weighed and their length measured and pupal condition estimated as
described in as described in Experiment 1. As butterflies emerged their condition, weight,

and wing length were recorded as in Experiment 1.

Experiment 3: Warm Winter Simulations

In these studies we considered “cold” winter temperatures to be our standard
winter temperature (3-4°C), and “cool” winter temperatures to be temperatures between
our standard winter temperature and 10°C. We chose 10°C as an upper limit due to the
very slow development of P. glaucus pupae at temperatures near 10°C [e. g. average
emergence time for P. glaucus pupae at 11°C is over 130 days for males and 280 days for

females (Scriber unpublished results)].

14

We first tested the impact of prolonged “cool”, rather than “cold” winter
temperatures on the survival of P. canadensis and P. glaucus pupae in 2006. On 2 May
2006 pupae of P. canadensis and P. glaucus kept at the same autumn and winter
conditions as controls in the short term heat stress simulations were placed at 6-8°C in the
spring of 2006 for 14 (n=40 pupae), or 42 (n=110 pupae) days. They were then placed at
14°C (18:6 L:D) for three days, and subsequently at 22°C (18:6 L:D) for adult
emergence. The emergence success and condition of the wings of the emerging adults
was recorded as above. The pupae used in this experiment had undergone a complete
overwintering period (Mid-October to April/May) under standard conditions and may
have been more sensitive than pupae only experiencing a warmer winter.

In 2007 we re-evaluated the response of P. canadensis and P. glaucus pupae to
“cool” winter conditions. However, in this experiment we used pupae that had not yet
experienced a complete overwintering period. On 1 February, 2007 overwintering pupae
(maintained at 3-4°C ) of P. canadensis and P. glaucus were weighed and separated into
three group each composed of 60 P. canadensis (n= 180) and 30 P. glaucus (n=90). The
first group was maintained at 3-4°C as a control, the second group was placed in a
chamber set at 10°C 12: 12 L:D, and the second group was placed in a chamber
maintained at 12 hrs of light at 10°C and 12 hrs of dark at 6°C. On April 13 2007 all
pupae were weighed and placed at 10°C 12:12 L:D for 1 day, then at 14°C 12:12 L:D 2
days, and finally at 22°C 18:6 L:D until they emerged. On August 25 2007 all remaining

dead pupae were removed and sexed.

15

 

.mcoEE—oo
35325 :53 3 com own—a =<

4:828

cos a .32 “an: ooeé message

wczmfiozm .cobcoov Dov .553 as wfiam
base 8 6.5.5-2.: Bob coon—a own?—

.Uov a meta—~33; .5“
.8 one 5.3.8 BEA—Bo =8:
.00 M: B “Sofia—HES 8a.:— =<

meets—=85
.5255 8.33

 

.mcoE—ocoo
85325 38 .8 :53 on com 895

0% cm voESEeE 89a =<

0%
an watoafizgo com “cm can...
.00 on no .mm .2 «a 858%».
5:2: 3 SSE—a 85 .8
3.5896 5.8:. 3 .5883. mo
use 05 :25 uoESEaE own?—

2552585
2322—808
BEE-i cease—o...—

 

£53250
oocowcoEo 38 8 8.83 8 com 895

0% ca 35859: 895 =<

0%
8 mew—353.85 c8 com :05

EB 836:8 238388
SE a 303 N a ._ .o

 

agate—55m 3.9.5
23—.— Each. team

 

wfiEm

 

 

.353

5553..

Iii

.353th macoatomxo 025 05 E com—mam 82558 cadfiomfioo mo beach—m ._.m 2an

 

16

Statistical Analyses

Survival of P. canadensis pupae in the short term heat stress simulations,
prolonged summer simulations, and warm winter simulations were analysed as Analyses
of Deviance by applying a GLM (Family = Binomial, logit link) using R (R Development
Core Team 2006). Due to exceptionally high survival of P. glaucus Analyses of Deviance
were not possible as assumptions of dispersion were violated. Therefore survival of P.
glaucus in all experiments was analysed as Binomial tests.

Wing length, abdomen size and percent weight loss for butterflies that emerged
with wings less than ‘A crumpled in the short term heat stress simulations were analysed
separately for P. canadensis and P. glaucus using AN COVAs with autumn and spring
temperature conditions (and their interaction) and sex as fixed factors and initial pupal
condition and initial weight as covariates. As all AN COVAs included the same factors
and covariates for these analysis estimates of partial omega square ((02) are comparable
between these analyses. (02 is a measure of effect size ranging from 0 to 1 (as long as the
F -ratio is greater than 1), for which the larger the value the larger the effect of the
treatment on the measured variable.

Pupal weight loss and emergence times for each species in the warm winter
simulation experiment were analysed as separate ANCOVAs, using initial weight as a
covariate and temperature treatment and sex as fixed factors. All ANCOVAs and (02
estimates were based on Type “III” sums of square using the CAR package available for
R (Fox, 2006). In all analyses performed no significant differences were found between

replicate chambers, therefore it was dropped as a factor in the analyses.

17

Results

Experiment 1: Short Term Heat Stress Simulations

The results from this study indicate that while unusually high autumn
temperatures did have a significant impact upon P. canadensis survival (Table 2.2 and
Fig. 2.1); it was much lower than the near complete mortality reported by Scriber et
al. (2002) for similar temperatures imposed on pupae during emergence in spring. Results
from Scriber et al. (2002) indicated that even 4 days at 36 °C during spring were
essentially lethal for P. canadensis pupae and also induced a high mortality on P. glaucus
pupae.

Table 2.3 provides a summary of the ANCOVAs performed separately for P.
canadensis and P. glaucus for the percent weight loss from pupae to emergence, wing
lengths, and abdomen size. These measurements provide a method to assess the stress
induced by autumn and spring temperature treatments on the two species. An AN COVA
of pupal weight loss of P. glaucus pupae indicated a significant effect of high autumn
temperature (F2359 P= 0.002). However, the effect size was fairly small (02 =0.036 and
survival of P. glaucus was excellent under all treatment conditions (Fig. 2.1). In contrast
P. canadensis survival was significantly lowered by both heat stress in autumn and cool
spring emergence conditions (Fig. 2.1 and Table 2.2). As with P. glaucus an ANCOVA
of pupal weight loss of P. canadensis for pupae in the short term heat stress simulation
was significant (F1253 P< 0.001), but the effect size was considerably larger (02 =0.3

(larger effect size = greater treatment effect).

18

 

 

m Se 53% am :3 N asap caesfaee macaw

83 SM. _ S :N S; a 8380 boom 3:5

83 2 3% Na ~85 _ Ease 3:5

8e 3 _ as mm 5.8 N ease :53.

$3 5% ms 33 _ ease macaw
338 as .522
853%

 

3.23

as Seam

35305 he bfiuaoscom 3ch 250,—.

 

.Goom as"; 28 225235 6 m was 3:: as .3805

u band”: 336 a watchman 3 Sanctum oo§_>oQ mo main: demon antennas .& .«o €3ng cocowcoao 58083 05 so
comic—coo boon REE use .EmEB 325 £833 5 moo—Em oESoQEB Ami $058095“ cocowbfio macaw mo “cube of. .N.N Bang.

19

£85980 853080 8.33 3 £38.30 5 vowaofiu 39a 8:: 353 was £82950 3592.5

Boo as How £3555 E Rumbas 89a E8058 was: 38 86on some Sm datuhgg N won: 3% EB §u§~w N Samoa?— mos:
yuan AmeE Doom 98 29% Uovmv £83980 ooaowuoao watam :53}. .mwfiam roam Ho Aflswma 002 ES 99% Doc: 223280
359080 mam—am Loco: awn—tam w:o_ was—£25m 3:08:835 E 33080 28 A8303 N .8 4 ho 8m 3&8 Oomm can 98¢ Oohmv
§3< >13 5 $225 “so: mafia—28mm 3:088:25 E coon—a 89a mmmamhgcu N can ”3.28% .& 90 3323 3083 AN oSwE

 

 

 

 

 

 

533 c. .95» :2: s 9.82.
N F o
p L b O
- 9
68+ 2835 .Q
§m>>+ .. ON
.80}. a macmumcwo .Q . on d
ggifil. m
dEo... oocouhoEm octam . ov m
w
. on S
m
- om M.
A
e
- E. I
- ow
/ I, ,, . om
- o9

 

 

 

 

20

 

omod M Na :5.on Ed .I. N8 <5.on ES Nd H flow ”m ES md a N: ”.mv &ENBEE .&

 

 

:5 n Ne :5.on <2 n me $3”; 88 2 a 3m ”m as no a ”.3 ”m 3:5» m
533 was,
<2 n ma 3:1 <2 u Ne <8.on mas o a 3mm ”m mes m a 32 ”m Eggs .m
So M Ne :55"; <2 u N3 8».on nae w a 3: “m ”as c a 3% ”m ”3:5...“ .m
onm 580?? 0353*
So n ma Scum mod u N3 sodvm .2 no a a? “m .2 we a 3m Mm 353:8 .m
<2 u N8 $.on <2 u N8 among .2 no a 0.? ”m .x. no a 3m ”2 gas» a

Avocowbfim 8 89:: $04 Emma? 252$

 

«oatm— Eo—anfi. watam tuba Eon—32H =EE=< E05333: 8325

 

 

.Goom Each 200 EoEQOESG 5 m E 88938 3 225280 33 REE Ba 2363 309 3:3 98 2925» 3020325 aw
mm 23 mafia—0E 032cm «0 2:3 :5: 09¢. wfims m<>OUZ< 88m 3352 as 22228 “coho 23523.5 macaw can 263% 05 E #8
Es moggi any—~80,“ Ea v.22: 8m mm H :88 2: 880.58 5228 30883808 05 E 823/ .maowmigm mmobm “mom Ebb roam

“_ Eon—toaxm Soc 2.223828 N 98 23.38% .m we mom—tots: 28 own—E :o :83 35822308 8m 3:52 we ng .md 2an

21

An interesting result observed was that when we compare the effect size ((02) for
spring emergence conditions on wing length and abdomen size we see a greater treatment
effect of emergence conditions for P. glaucus than P. canadensis (Table 2.3). This
difference in the reaction to emergence conditions may be a reflection of the larger
reserves present in P. glaucus. It is likely that P. glaucus is capable of buffering the
effects of longer (cooler) spring emergence conditions by reallocating resources, as has
been seen in other insects in response to various forms of stress (e. g. Angelo and Slansky
1984, Boggs and Freeman 2005, Jannoteta1.2007).

The overall results from this experiment indicate a difference in tolerance to heat
stress between pupae of P. canadensis and P. glaucus prior to winter (prior to imago
development), but far less severe than observed during emergence in spring (Scriber et
al.2002) and insufficient for rare heat waves to set a southern range limit for P.
canadensis.

Experiment 2: Prolonged Summer Simulations

In this experiment we found that extending the exposure to warm
temperatures did significantly increase the mortality rate of P. canadensis. An Analysis
of Deviance indicated survival was significantly reduced by higher autumn temperatures
(deviance = 51.57, d.f. = 2, P0?) < 0.001), prolonged higher autumn temperatures
(deviance = 8.99, d.f. = 1, P0?) = 0.003), lower spring temperatures (deviance = 13.55,
d.f. = 1, P08) < 0.001), and a significant interaction between spring and autumn
temperatures (deviance = 9.21, d.f. = 2, P0?) = 0.01). More detailed observations of the
results from this experiment are summarised in Figure 2.2. In Figure 2.2a we can see the

survival of P. canadensis pupae kept for 31 days (August 19-September 19) at

22

(a)

 

 

 

 

 

 

 

 

 

 

 

 

Placed for1 Month
100 ~
a
_ 80 «
g 70 -
g 60 1
‘3 50 -
E 40 - Spring Emergence Temp. **
g 30 ‘
a 20 q . Cool
10 - -o—Warm
0 fl 1 1
18°C 22°C 26°C
Autumn Temperature
(b)
Placed for 2 Months
100 -
90 a _
_ 80 ~ o—r 4w
% 7o 1
a 60 -
2 5° ‘
40 - .
g 30 - Sprung Emergence Temp. .. -* * *
- . Cool
20 ##ch
10 r -o—Warm
0 f l
18°C 22°C 26°C

Autumn Temperature

 

 

 

Figure 2.2. Percent survival of P. canadensis pupae maintained at 18°C, 22°C, or 26°C
(a) for one month or (b) two months during autumn. Pupae were subsequently emerged in
“cool” spring emergence conditions (14°C) or “warm” spring emergence conditions
(22°C). Survival significantly lower than expected values based on binomial tests are
represented by * for 90%, ** for 60%, *** for 50%, and **** for 40%.

23

temperatures likely to be observed during this period (18°C and 22°C) had excellent
survival. Only pupae exposed to both 26°C and cool emergence conditions (14°C)
experienced a significant decrease in survival. This indicates that exposure to 26°C in
autumn significantly stressed the pupae of P. canadensis. This effect became more
evident when pupae were maintained for 61 days (August 19- October 19) at 18°C, 22°C,
or 26°C (Fig. 2.2b), as is evidenced by significantly lower survival under both emergence
conditions.

The results from this experiment indicate that prolonged exposure to warm

temperatures can significantly stress the pupae of P. canadensis.

Experiment 3: Warm Winter Simulations

Successfirl emergence of P. glaucus placed for 14 or 42 days at 6-8 °C prior to
being set for emergence in 2006 was high (95% and 93.5% respectively) and not
surprisingly binomial tests indicated they were not significantly lower than 95%
(P=0.64). For P. canadensis successful emergence of pupae left for 14 days at 6-8 °C was
lower, but not unusual (80%). A binomial test did not indicate a significant difference
than the survival of the controls (90%) (P= 0.13). However, the number of successful
emergences of P. canadensis placed at 6-8 °C for 42 days prior to emergence was very
low, 42%, and significantly lower than 60% (P=0.005).

The emergence of pupae placed at 10 °C or temperatures alternating between 10
°C and 6 °C from 1 February, 2007 to 13 April, 2007 indicated a very similar pattern to
those placed at 6-8 °C for 42 days in 2006. In particular we found no significant

difference in emergence success for P. glaucus (all treatments had emergence success >

24

95%) and a significantly lower survival for P. canadensis in our treatments relative to our
control (deviance = 29.5, d.f. = 2, PM) < 0.001) (Fig. 2.3). In addition, we found a
significantly greater percent weight loss in P. canadensis pupae placed at 10 °C and
temperatures alternating between 10 °C and 6 °C relative to the control (172,174 = 61.8, P<
0.001; Fig. 2.4), and no significant difference in percent weight loss between P. glaucus
pupae in different temperature regimes (F230 = 2.0, P=0.15; Fig. 2.4).

Finally, when pupae were placed at 22 °C for emergence in 2007 an AN COVA
indicated significantly different emergence time for P. canadensis pupae placed at 10 °C
(9.4 :t 0.4 days), temperatures alternating between 10 °C and 6 °C (11.8 :i: 0.3 days), and
the 4 °C control (16.0 :1: 0.4 days) (F1113 = 81.52, P< 0.001). Tukeys HSD indicated
significant differences between the three temperature conditions at a=0.05. For P.
glaucus, we also found a significant difference in emergence times between pupae placed
at the three temperature regimes (F1113 = 14.0, P< 0.001). However, Tukeys HSD
indicated that differences were only significant between pupae placed at 10 °C (15.2 i:
0.85 days) and the other two treatments; temperatures alternating between 10 °C and 6 °C
(19.5 i 0.57 days) and the control (19.7 i 0.63 days). In addition, the effect size (032) for
the treatment effect on P. canadensis is 0.56 while that for P. glaucus is only 0.24,
indicating a much stronger effect of temperature treatment on the emergence time of P.

canadensis than P. glaucus.

25

 

I P. canadensis

l P. glaucus

  

90
80 -
70 ‘
60
50 ~
40 -
30 ,
20 4
10 -
0 -__

% Survival

 

 

 

4:4 10:6 10:10
Chamber Temperatures °C 12hr:12hr

'1

 

 

 

Figure 2.3. Percent survival of P. glaucus and P. canadensis pupae maintained in
chambers set at continuous 4°C (4:4), alternating 12 hours at 10°C and 12 hours at 6°C
(10:6), or continuous 10°C (10:10) from midwinter to early spring.

26

 

 

     

I P. canadensis A
14 ~
B lP.gIaucus

12 ~
3 10 -
3
E 8 ‘ C
'5 a
3 6 —
X

4 _

2 r

0 —~— —r———

4:4 10:6 10:10
Chamber Temperatures °C 12hr:12hr

 

 

 

Figure 2.4. Percent weight loss of P. glaucus and P. canadensis pupae maintained in
chambers set at continuous 4°C (4), alternating 12 hours at 10°C and 12 hours at 6°C
(10:6), or continuous 10°C (10) from midwinter to early spring. Within each species
means followed by the same letter are not significantly different at the p<0.05 level
(Tukeys HSD).

27

Discussion

Our results indicate that unusually hot late summer/autumn temperatures are not the
primary cause of the southern range limit of P. canadensis. While pupae of P. canadensis
exposed to short periods of unusually high autumn temperatures did have a significantly
lower survival (Fig. 2.1), it was much lower than the near total mortality for high spring
temperatures reported by Scriber et al.(2002). Extreme responses, such as the near
complete mortality of P. canadensis exposed to 4 days at 36 °C during spring (Scriber et
al., 2002), are likely to be sufficient for rare events to pose a significant selective
pressure. However, our results did not indicate such an extreme response for P.
canadensis (Fig. 2.1). The temperatures we exposed pupae to (37 °C days and 22 °C
nights) are extreme, and although possible, these temperatures are rare and unlikely to be
present for prolonged periods of time (e.g. 1-2 weeks). The discrepancy between this
study and the one presented by Scriber et al. (2002) lies in the physiological state of the
pupae prior to overwintering and after overwintering. Prior to overwintering pupae are in
a quiescent phase more likely to tolerate temperature extremes, while during spring,
diapause is broken and formation of the imago inside the pupae begins, and sensitivity
during this stage is likely a representation of temperature thresholds during imago
development. Temperature averages exceeding 30 °C may occur in the summer, but are
highly unlikely to occur in the spring when the imago is developing. Therefore it is
unlikely that spring thermal stress is the primary cause of the Southern range limit of P.
canadensis.

We did find a greater tolerance to high temperatures in P. glaucus than P.

canadensis that is likely to be a reflection of differences in metabolic thresholds and/or

28

energy reserves in the two species. In contrast to P. canadensis, P. glaucus did not
experience a significant decrease in survival rate or weight due to our short term heat
stress simulations (Fig. 2.1 and Table 2.3). In addition, cool emergence temperatures did
not significantly impact P. glaucus survival and weight, while it did have a significant
effect upon P. canadensis (Fig. 2.1 and Table 2.3). Metabolism during diapause is
significantly reduced, but once diapause is broken, metabolism rates become higher and
prolonged periods of development are likely to significantly reduce energy reserves.
Pupae placed for emergence in 2006 at cool emergence temperatures (14 °C average)
took approximately twice as long to emerge as those placed at warmer emergence
temperatures (22 °C average). This extended period of development significantly reduced
the survival of P. canadensis, while that of the larger P. glaucus was not affected (Fig.
2.1). This result may be, at least in part, due to the fact that in general, smaller
individuals are believed to have fewer metabolic resources and have higher metabolic
rates (Peters, 1983; Gillooly et al., 2001).

Although, P. glaucus did not have a reduction in survival, spring emergence
temperature did cause a significant reduction in P. glaucus abdomen size and a stronger
negative effect on the wing length of emerging adults (Table 2.3). These differences are
indicative of larger reserves in P. glaucus dampening the negative effects of prolonged
developmental time on survival. In addition, although there were significant reductions in
abdomen size and wing length due to emergence conditions, there was no significant
effect of emergence conditions on P. glaucus overall weight (Table 2.3). This result is a
strong indication that P. glaucus is capable of buffering the effects of longer (cooler)

spring emergence conditions by reallocating resources, as has been seen in other insects

29

in response to various forms of stress (e. g. Angelo and Slansky 1984, Boggs and Freeman
2005, Jannot et al.2007).

In addition to potentially higher metabolic reserves being present in P. glaucus, it
is also apparent that P. canadensis has a lower metabolic threshold. In our warm winter
simulations P. canadensis weight loss was significantly higher relative to the controls,
while that of P. glaucus was not significantly different (Fig. 2.4). In addition, emergence
times were significantly shorter for P. canadensis exposed to warm winters than for P.
glaucus exposed to warm winters (see results). Together these results indicate that P.
canadensis had lower developmental thresholds in our warm winter simulations than P.
glaucus, which is likely to be the cause of the high mortality rates observed (Fig. 2.3).
The ability to have a faster imago development is likely adaptive in the majority of the P.
canadensis range as the available growing season is constrained, but in more southern
latitudes it may pose a significant cost.

The prolonged summer simulations indicated that P. canadensis is also sensitive
to prolonged periods at relatively mild warm temperatures prior to winter conditions (Fig.
2.2). The survival rate of individuals exposed to one or two months of warm conditions
was far lower than those exposed to short periods of high heat (Figs. 2.1 and 2.2).
Although diapausing pupae have lowered metabolic rates, some metabolism is still
present and high temperatures can be metabolically costly (Hahn & Delinger, 2007). The
low survival rates of P. canadensis exposed to prolonged periods of warm temperature or
slightly above overwintering conditions (Figs. 2.2 and 2.3) indicate that mild but
prolonged shifts in temperature may be a more significant stress than occasional short

heat waves prior to overwintering.

30

Although short periods of unusually hot summers do not appear sufficient to act
as the primary limit to the southern distribution of P. canadensis, higher temperatures do
appear to pose a significant cost to P. canadensis that could limit its range. However, the
effects of temperature appear to be gradual (e. g. increased warm temperature conditions)
and we would not expect a sharp southern boundary to the range as is observed (Scriber
et al., 2003). It is likely that the sharp boundary observed for P. canadensis is due to a
combination of environmental factors that shift at the ecotone between boreal and
temperate deciduous forests, one of these being the temperature regime.

Other potential factors associated with the sharp boundary include the shift in host
plants available and the presence of P. glaucus. At this ecotone the most common host of
P. canadensis, Populus tremuloides (Salicaceae), begins to decline and the preferred host
of P. glaucus, Liriodendron tulipifera (Magnoliaceae), becomes more prevalent. While L.
tulipifera is toxic to P. canadensis larvae female P. canadensis will readily oviposit on it.
The presence of P. glaucus is also likely to affect that of P. canadensis through factors
such as apparent competition (e. g. through common predators) and/or significantly
reducing successful matings. Both P. canadensis and P. glaucus males strongly prefer
mating with P. glaucus females (Deering & Scriber, 2002). Although P. glaucus and P.
canadensis crosses produce fertile offspring (e. g. Scriber et al., 1995; Scriber et al.,
2003), the female offspring from these crosses emerge 2-3 weeks after the flight period of
P. canadensis and P. glaucus males (Scriber & Ording, 2005; Scriber et al., 2007),
significantly reducing the likelihood of mating for such hybrid females.

An indication that the southern range limit for P. canadensis may not be primarily

maintained by abiotic factors alone is that prior to 1998, P. canadensis-like traits were

31

observed farther south into P. glaucus populations than P. glaucus traits were into P.
canadensis populations (Scriber, 1990). As the P. glaucus northern range limit is defined
by the thermal environment that allows the completion of two generations (Scriber,
1994), an environmental barrier to movement was present for P. glaucus. In contrast,
such a sharp environmental barrier may not have been present for P. canadensis; instead
a combination of factors (including temperature, host availability, and the presence of P.
glaucus) may have all acted to create the southern range limit of P. canadensis. However,
since 1998 the rapid increases in temperature have changed the previous pattern by
increasing the geographic area where P. glaucus can complete 2 generations (Scriber &
Ording, 2005; Scriber et al., 2007). This environmental change has lead to the
introgression of P. glaucus traits into P. canadensis populations including the formation
of a late emerging hybrid swarm population (Scriber, 2002; Scriber & Ording, 2005;

Scriber et al., 2007).

Conclusion

Occasional very hot summer temperatures do not appear to be the primary cause of the
southern range limit of P. canadensis. However, P. canadensis does exhibit a
considerably lower tolerance to high temperature extremes and conditions simulating
shorter/warmer winters than P. glaucus. These results suggest that differential
temperature tolerance may be a factor in maintaining the parapatric species borders in
this hybrid zone. In particular the results suggest that the expenditure of metabolic
reserves is likely to be an important component of the southern range limit of P.

canadensis. As the climate warms, cold temperatures are expected to be shorter, in

32

particular transitions between cold temperatures to hot temperatures are expected to be
longer and arrive earlier (i.e. shorter winters and longer springs) (Schwartz et al., 2006).
Therefore it is likely that climate change will greatly increase the likelihood of local
extinctions of cold adapted insects that diapause as pupae, even if they can tolerate

temperature spikes and don’t emerge out of synchrony with their hosts.

33

CHAPTER 3

Mercader RJ, Scriber J M (2007) Diversification of host use in two polyphagous
butterflies: Differences in oviposition specificity or host rank hierarchy?
Entomologia Experimentalis et Applicata 125: 89-101.

34

DIVERSIFICATION OF HOST USE IN TWO POLYPHAGOUS BUTTERF LIES:
DIFFERENCES IN OVIPOSITION SPECIFICITY OR HOST RANK

HIERARCHY?

Abstract

Novel host use may represent an initial step towards diversification or radiation
onto novel hosts within an evolutionary lineage, particularly if a shift in host plant
preference ranking takes place. Polyphagous stages of evolutionary lineages may
represent transitional states in which novel host associations are more likely to develop,
but may be more difficult to detect experimentally. The polyphagous sister species
Papilio glaucus L. and Papilio canadensis (Lepidoptera: Papilionidae; these Papilio =
Pterourus) are known to exhibit differences in host plant use, despite significant overlap
in host use abilities, providing an opportunity to examine how host shifts in polyphagous
species may occur and what the implications for future divergence may be. In particular,
we were interested in l) determining whether differences in oviposition behavior of these
species were due to changes in specificity or shifts in host plant hierarchy, 2) whether the
varying preference for primary hosts also affected the preference for secondary hosts, and
3) what the oviposition preferences of a new hybrid swarm population are. We examined
more than 40,000 oviposition bouts from more than 400 P. glaucus, P. canadensis, and
hybrid females placed in seven, three, or two choice assays. In each of the choice assays,
leaves from plants in different plant families of varying suitability for P. glaucus and P.
canadensis larvae were used. We found the primary difference between P. glaucus and P.

canadensis to be limited to a Z-linked shift in host rank hierarchy due to an acceptance of

35

Populus tremuloides Michx. (Salicaceae) and reduced specificity for Liriodendron
tulipifera L. (Magnoliaceae) in P. canadensis. In addition, we found the absence of the Z-
linked oviposition acceptance of P. tremuloides in a recently formed allochronically
separated hybrid swarm population found in P. canadensis territory at the northern border

of the P. glaucus and P. canadensis hybrid zone.

Introduction

The divergence of insect species through host plant specialization has been proposed as a
mechanism to explain the great diversity of plant feeding insects, but specialization alone
would be a dead end process. Janz et al.(2006) proposed a scenario of oscillating host
expansion and specialization to help resolve this apparent dilemma. Within this scenario,
polyphagy is a transitional state that allows the incorporation of new hosts. It thus
prevents specialization from reaching a dead end and may even facilitate radical host
shifts (Janz et al. , 2001; Weingartner et al., 2006). In addition, colonization of new hosts
leading to radiation, rather than co-cladogenesis, has been gaining support as the reason
for the association between plant feeding insects and their hosts (Janz et al., 2001; Nosil,
2002; Percy et al., 2004; Braby & Trueman, 2006; Lopez- Vaamonde et al., 2006; and
Murphy & Feeny, 2006). Therefore, host shifts in polyphagous insects are of particular
interest, as they may represent the initial steps towards diversification or the acquisition
of novel hosts within an evolutionary lineage.

Perhaps due to the greater abundance of specialized insects and an easier
detection of host shifts within them, the majority of studies on host shifts have

concentrated on monophagous or oligophagous species (e. g., Carroll & Boyd, 1992;

36

Radtkey & Singer, 1995; Menken, 1996; F eder et al., 1998; Abrahamson et al., 2001;
Hora et al., 2005; Ohshima & Yoshizawa, 2006; Stastny et al., 2006; but see Thompson,
1998; Janz et al., 2001 for shifts between specialized and generalized forms).The closely
related sibling species Papilio glaucus L. and Papilio canadensis R & J (Lepidoptera:
Papilionidae; these Papilio = Pterourus) offer the unusual opportunity to examine a shift
in host use between two polyphagous forms. These sibling species exhibit the highest
degree of polyphagy amongst the 560+ species of Papilionidae (Scriber, 1984), as
together their larvae can consume 30-40 species in 14 plant families (Scriber, 1988).
Despite their high level of polyphagy and a significant host overlap, both species exhibit
differences in suitability of some of their primary hosts for neonate larvae (Scriber,
1996a). In particular, the most common host plant family of P. canadensis, Salicaceae, is
toxic to P. glaucus larvae, while the preferred host of P. glaucus, tulip tree, Liriodendron
tulipifera L. (Magnoliaceae), is toxic to P. canadensis larvae.

For plant feeding insects with larvae that develop on a single host, such as tree
feeding Papilio, changes in oviposition behavior of adults are necessarily required for
host range shifts. Behavioral adaptations to use new hosts have been proposed to precede
physiological adapations (F utuyma et al., 1984; Janz et al., 1994) and to be evolutionarily
labile (Wasserrnan & Futuyma, 1981; Gassman et al., 2006). In addition, the common
observance of oviposition “mistakes” (Straatman, 1962; Wiklund, 1975; Chew, 1977;
Berenbaum, 1981; Scriber et al., 1991; Scriber, 1993; Larsson & Ekbom, 1995; Renwick,
2002; Graves & Shapiro, 2003), lack of concordance between preference and
performance in many organisms (Thompson & Pellmyr, 1991; Mayhew, 1997), and

variation in oviposition profiles due to thermal constraints (e. g., Scriber & Lederhouse,

37

1992; Scriber, 1996b, 2002a) support the notion that the apparently close associations
between plant feeding insects and their host plants are actually rather flexible.

Shifts in oviposition behavior may be due to changes in specificity in which the
proportion of eggs laid on lower ranking hosts differs, or due to shifts in the host
preference ranking (Courtney et al., 1989). Changes in specificity are likely to lead to
shifts between specialist and generalist forms which may lead to diversification within
the associated plant lineage. However, changes in specificity are unlikely to significantly
alter host plant associations, because the primary host associations remain. In contrast,
shifts in host preference rank hierarchy can lead to shifts away from associated plant
lineages towards the plant lineage of secondary hosts, altering primary host plant
associations within an insect lineage. In some instances, shifts in host preference rank
hierarchy may also include the acceptance of a novel host or plant family unused within
the evolutionary lineage of the organism. Such shifts are likely to open an entirely novel
niche which could lead to adaptive radiation. However, these shifts may be uncommon as
is reflected in the evolutionary conservatism of host plant affiliations in plant feeding
insects (Ehrlich & Raven, 1964; Janz et al., 2001; Weiblen et al., 2006).

Within the P. glaucus species group there has been significant diversification in
associations with plant families used (Scriber, 1996a) and significant differences in
oviposition behavior have been noted between the closely related P. glaucus and P.
canadensis (Scriber et al. , 1991; Scriber, 1994). Studies comparing the oviposition
behavior of P. glaucus and P. canadensis have primarily used three choice oviposition
arenas with leaves of L. tulipifera (Magnoliaceae), Populus tremuloides Michx.

(Salicaceae), and a common host for both species, Prunus serotina L. (Rosaceae). These

38

studies have indicated that in general P. glaucus individuals will lay most of their eggs on
L. tulipifera and very few eggs on P. tremuloides, while P. canadensis individuals will
lay roughly similar quantities of eggs on all three hosts (Figure 3.1).

Despite the intensive work on these species it is still unknown whether the
observed differences in oviposition behavior represent differences in specificity or a shift
in host plant hierarchy. In addition, it is unknown whether changes in oviposition
behavior observed are also found amongst hosts in the other lower ranked plant families
used by these polyphagous species. These aspects of their shifts in oviposition behavior
are relevant for the potential future divergence within these species or as these two
species hybridize.

In recent years gene movement has been seen to increase across a hybrid zone
delineating the range limits of P. glaucus and P. canadenis (Scriber, 2002b). This increase
in hybridization has produced an allochronically separated hybrid swarm population
(Scriber & Ording, 2005) remarkably similar to a population found in northern Georgia
believed to represent a separate species Papilio appalachiensis (Pavulaan & Wright,
2002). The larval host use abilities of this delayed hybrid swarm population represent a
mixture of the two genomes (Scriber et al. , unpublished data) and their oviposition
behavior is likely an important indicator of the ecology and the evolutionary potential of
this hybrid swarm population if it becomes/remains reproductively isolated.

In order to understand what the difference in oviposition between P. canadensis

and P. glaucus is and what its implications for firture divergence are, we assayed P.

39

 

 

 

 

   

 

  

 

 

70 I L tulip. ”
3 60 D P. tram
a 50 I P. semt
é l aOther
0 40 “
a:
5 30 J
i

20 ~

10 T 7

o | _ // ___.,

 

P. glaucus P. canadensis

Figure 3.1. Mean percentage + SE of eggs laid by 93 female Papilio glaucus from
Lancaster Co., PA, and 94 female Papilio canadensis from the “early flight” found in the
Battenkill River Valley in the New York/ Vermont border. Leaves present in the bioassay
were L. tulipifera, P. serotina, and P. tremuloides. The “other” category represents eggs

laid on the paper or plastic of the oviposition arena.

40

canadensis and P. glaucus populations near the northern and southern border of the
hybrid zone on arenas containing seven, three, or two hosts. We were also able to
compare individuals from the newly formed hybrid swarm population in three choice
assays to populations of both P. canadensis and P. glaucus. Using these data we
specifically asked the following questions 1) Does P. canadensis exhibit a preference for
P. tremuloides or simply a decreased specificity for L. tulipifera?; 2) Does P. canadensis
exhibit a different oviposition preference than P. glaucus for hosts other than P.
tremuloides?; and 3) What is the oviposition pattern of the newly formed late flight

hybrid swarm?

Materials and methods
Insects

Butterflies used in seven choice and two-choice studies in 2006 were reared in
2005 or 2006 from eggs laid by wild-caught females, with the exception of the northern
Michigan individuals which were wild-caught individuals. Females of the “early flight”
P. canadensis were reared on P. serotina by either mass rearing in sleeves placed on tree
branches or reared independently in a growth chamber at 22 °C in 2005, fiom eggs of
females collected from the first flight (late May-early June) in the Battenkill River Valley
area at the New York/Vermont border USA (43° N latitude and 73° W longitude). The
P. glaucus females from the Pennsylvania population were field-reared in 2005 on P.
serotina from butterflies collected in Lancaster Co. in south-eastern PA, USA (40° N
latitude and 76° W longitude). All females were kept at 3-4 °C throughout the fall and

winter of 2005, until they were emerged the spring of 2006. Females from the Georgia

41

population were field-reared on P. serotina in 2006 from females collected in Oglethorpe
Co., GA, USA (34° N latitude and 83° W longitude). After eclosion, butterflies were fed
a 15-20% honey water solution and stored at 3-4 °C for a maximum of 4 days until they
were hand paired and setup in oviposition arenas. Females from the northern Michigan
population were field collected in 2006 from Cheboygan and Emmet Counties in MI,
USA. (45° N latitude and 84° W longitude). These butterflies were transported in coolers
to the MSU laboratory where they were fed a honey water solution and placed in
oviposition arenas.

Butterflies used in the three choice assays were a combination of females
collected in 2003, 2004, and 2005, and individuals reared on P. serotina during 2002,
2003, and 2004. The populations used were the same as those above, with the exception
of the “late flight” population. Females from the “late flight” population were derived
from wild-caught females from the “false-second” flight in the Battenkill River Valley
area at the NY/V T border, USA as described in Scriber & Ording (2005).

Butterflies used in seven choice bioassays in 1993, 1996, and 1998 were a
combination of wild-caught and lab reared females. Females were collected in 1993 from
Fairbanks, Alaska, USA (65° N latitude), and in 1996 from Menifee Co., Kentucky,
(37.9° N latitude) and Lawrence Co., Ohio, USA (38.5° N latitude) and were shipped
overnight to our MSU facilities. In 1998, wild-caught P. canadensis females from
Charlevoix Co., Emmet Co., and Dickinson Co., MI, USA, (45.5-47° N latitude) were
transported in coolers to our MSU facilities. All females were fed honey water solution
upon arrival to MSU and setup in oviposition arenas. In 1996, four hybrid families were

produced by crossing laboratory reared P. canadensis females from northern Michigan

42

with wild-caught males from Menifee Co. KY, USA. The resulting females were then

mated and placed in oviposition arenas.

General methods for oviposition assays

The general methods used were similar to those described by Scriber (1993); arenas
consisted of clear round plastic containers placed on a rotating platform (approximately
10 revolutions per h) in front of incandescent lights on a L4:D4 h photocycle to maximize
egg laying, with leaves of each plant species placed in floral aquapicks. Butterflies were
fed a honey-water solution daily and allowed to oviposit until they were too weak to fly
(4-12 days). The number of eggs laid on leaves of each plant or non-leaf portions were
counted daily and the leaves replaced with new leaves as necessary. No leaf remained for
more than 2 days in the assays.

Leaves were collected from trees growing in Ingham County, MI, USA, from
locations known to be pesticide free, placed in aquapicks in plastic bags, and placed in a
refrigerator for up to 3 days to ensure leaf quality remained high. Leaves used in the
assays were collected from at least four trees, with the exception of L. benzoin and B.
papyrifera for which two and one plants were used, respectively, since plants known to

be pesticide free were more limited for these two hosts.

Papilio canadensis oviposition preference
Ideally, distinguishing oviposition specificity and host plant rank hierarchy would be
accomplished using a combination of sequential no choice trials and simultaneous choice

trials. However, the behavior of these butterflies provide a few important experimental

43

constraints to the use of no-choice trials and interpretations of simultaneous choice trials:
1) the number of eggs laid is highly variable among days and individuals, which requires
oviposition data to be analyzed as proportions, and 2) the degree of host specificity
exhibited by individual females (even within full sibling families) is highly variable.
These limitations prohibit the use of single host arenas and decrease the likelihood of
distinguishing between shifts in specificity and host preference rank hierarchy from
arenas with multiple hosts, which make direct tests of preference hierarchies difficult.
However, we can test for preference hierarchies indirectly by making the assumption that
if there is a preference for a particular host then individuals with greater specificity for
that host should lay a greater number of eggs on that host and individuals with lower
specificity for that host should lay eggs more evenly across hosts. Therefore, given
enough choices we should be able to distinguish whether the cause of an increased
average proportion of eggs laid on a host are due to the presence of a preference for that
host or the reduction in specificity for other hosts.

To test for this relationship, we placed butterflies in oviposition arenas containing
leaves from plants in seven different families with varying larval suitability for P.
canadensis (see above); L. tulipifera, L. benzoin, P. tremuloides, P. trifoliata, B.
papyrifera, F. americana, and P. serotina. These seven choice bioassays were conducted
between 15 May and 17 June 2006 using P. canadensis individuals from the Vermont
early flight population (11 = 57) for which only females that laid more than 20 eggs were
used (11 = 36). Using these data we performed two tests of the variance in response

between individuals for each of the hosts.

44

First, if individuals with greater specificity lay a greater number of eggs on a
preferred host and individuals with lower specificity lay eggs more evenly across hosts
we should expect that greater variance should be present between individuals for the
proportion of eggs laid on the preferred host(s) than for less preferred hosts. We
examined the variance between individuals by performing bootstrapped variance
estimates for the proportion of eggs laid on each host (1000 non-parametric bootstrap
replications) to identify if a higher variance was present for specific hosts. Second, we
examined the relationship between hosts by testing for significant correlations between
the hosts in the seven choice arenas. A strong preference for a particular host should be
evidenced by significant negative correlations between the preferred host and the other
common hosts. In contrast, there should be weak or positive correlations amongst the
other potential hosts.

Bootstrapped variance estimates, Spearman correlations, and all other analyses
presented in this study were performed using the R statistical package V 2.4 (R

Development Core Team, 2006).

Differences in oviposition preference hierarchy between Papilio glaucus and Papilio
canadensis

Seven choice assays (2006). In order to compare differences in host rank hierarchy
between species and populations, we performed seven choice bioassays identical to the
ones mentioned above using females from the P. canadensis northern Michigan
population (11 = 54 set up, 20 used), and the P. glaucus Pennsylvania (11 = 44 set up, 20

used) and Georgia (n = 27 set up, 15 used) populations. We then combined the proportion

45

of eggs laid by each butterfly on each host leaf in these seven choice bioassays to those of
the P. canadensis “early flight” population used above. As responses are likely to be
correlated, we performed a Principal Component Analysis (PCA) on this data set to form
PCA scores with which we could compare populations along the major axes of variance.
Based on the broken stick approach presented in Peres Neto et al. (2003), we created a
reference to identify which variables (hosts) were relatively more strongly correlated with
particular axes. We did this by squaring loadings [the sum of squared loadings across a
row (a single variable) is l] and identifying which loadings exceeded the expected values
under the broken stick distribution.

The first three eigenvectors accounted for more than 80% of the variance (Table
3.1), and were therefore used for further analysis. We tested for differences between the
four populations along these three main axes by performing Kruskal-Wallis tests on the
individual scores for each butterfly along these axes, and contrasted the populations using
Bonferroni corrected Mann-Whitney U-tests. We were also interested in testing whether
differences for minor hosts were present between the populations, and these were
difficult to interpret from the eigenvectors produced. We performed a Varimax rotation
on the PCA factor loadings to form axes that were easier to interpret. Using the broken
stick model approach mentioned earlier we identified four axes which were not primarily
correlated with the proportion of eggs laid on either of the most acceptable hosts, P.
tremuloides or L. tulipifera (axes 5, 6, 7, and 8). We analyzed the individual scores from

these four axes as above.

46

Two choice assays (2006). We were interested in the basis of the difference in responses
between P. canadensis and P. glaucus previously seen in three choice oviposition arenas
(Scriber et al. , 1991; Figure 3.1). In particular we were interested in whether P.
canadensis has a lower preference than P. glaucus for L. tulipifera , and if P. canadensis
has a greater specificity for P. tremuloides than L. tulipifera . To test for these differences
we conducted two-choice tests consisting of combinations of L. tulipifera (Lt), P.
serotina (Ps), and P. tremuloides (Pt) between 21 June 2006 and 22 July 2006 on
females from the early flight (EF) and Pennsylvania (PA) populations. Only individuals
that laid more than 10 eggs were used for the analysis. In total, 21 EF females and 25 PA
individuals were placed in arenas with T1" and BC (15 and 18 used, respectively). The
BF and PA individuals that successfully laid eggs were placed into arenas with QA and
BC; nine EF and four PA individuals laid more than 10 eggs. An additional eight EF and
15 PA individuals were placed in arenas with QA and BC, of which six EF and six PA
laid more than 10 eggs. The 14 EF and 10 PA individuals were analyzed together. Finally
eight EF and 12 PA individuals were placed in arenas with QA and TT, of which four EF
and 11 PA laid more than 10 eggs. To test for differences we contrasted the proportion of
eggs laid on each host separately for our two choice bioassays using Bonferroni corrected

Mann-Whitney tests.

47

Table 3.1. Factor loadings for the first three principal component axes of the proportion
of eggs laid in seven choice bioassays for the Papilio canadensis early flight” (n = 36)
and northern Michigan (n = 20) populations, and the Papilio glaucus Pennsylvania (11 =
20) and Georgia populations (n = 15).

Principal component axis

 

 

 

Leaf 12 2 3

L. tulipifera 0.680'I 0.6221 0.03
L. benzoin -0.054 -0.152 -0.289
P. tremuloides 0699‘ - 0.5031 0.296
P. trifoliata 0.022 -0.307 -0.333
B. papyn'fera -0.016 -0.058 -0.135
F. americana 0.184 0473‘ -0.768‘
P. serotina -0.105 -0.125 -0.330
Other -0.012 -0.011 -0.006

% Variance o o o
Explained 51.7 /o 18.9% 11.0/o

 

lValues higher than expected under the broken-stick model.

2Scores between populations for that axis were significantly different between
populations (Kruskal-Wallis P<0.05).

48

Seven choice assays (1 993-1998). Here we include data from seven choice bioassays
performed in 1993 and 1996 using a slightly different host array to compare oviposition
responses of F1 hybrids. The methodology used in these bioassays was similar to that
used in 2006, with the exception of a small difference in the host array used. In these
bioassays all hosts remained the same except that instead of B. papyrifera, the invasive
species buckthom, Rhamnus cathartica L. (Rhamnaceae) was used in 1993 and 1996. In
1993, 13 females collected in Alaska placed in the oviposition arrays laid more than 20
eggs, and in 1996, three females from Kentucky, 35 females from Ohio, and 30 hybrid
females laid more than 20 eggs. In 1998, two females from Charlevoix Co., one female
from Dickinson Co., and four females from Emmett Co., laid more than 20 eggs. Due to
the small sample sizes and high variance among individuals from the P. canadensis

populations we limit this data to a qualitative comparison.

Ovposition behavior of the “late flight” population.

The oviposition behavior of the “late flight hybrid population” was contrasted to
populations of P. canadensis and P. glaucus using three choice oviposition bioassays
consisting of L. tulipifera, P. serotina, and P. tremuloides conducted during the months
of May, June, and July of 2003, 2004, and 2005. A total of 124 P. canadensis females
from the early flight population, 35 individuals from the late flight hybrid population, 34
individuals from the P. glaucus Georgia population, 40 individuals from the P.
canadensis northern Michigan population, and 118 individuals from the P. glaucus
Pennsylvania population were set up. Data were expressed as proportions of eggs laid on

each host and only those that laid more than 20 eggs were used (94, 29, 25, 27, and 93

49

females for the “early flight”, “late flight”, Georgia, northern Michigan, and
Pennsylvania populations, respectively). As with the seven choice assays, we were
interested in comparing populations along the major axes of variance, so we performed
Kruskal-Wallis tests on the scores from the first three axes of a PCA performed on the
combined proportion data of the five populations (which accounted for >99% of the
variance), and contrasted the populations using Bonferroni corrected Mann-Whitney U-
tests. As with seven choice assays we used the broken stick model approach to identify

which axes were more strongly associated with particular variables.

Results

Papilio canadensis oviposition preference

Variance estimates were higher for P. tremuloides than for any other host (Table
3.2) supporting the notion that there is a preference for P. tremuloides. Spearman
correlations between hosts in seven choice arenas for P. canadensis indicated no
correlations between host pairs that did not include P. tremuloides, L. tulipifera, or F.
americana. Significant negative correlations and weak correlations were found between
the number of eggs laid on P. tremuloides and the other hosts, as was predicted if a
preference for P. tremuloides was present, again supporting that there is a preference for
P. tremuloides. An unexpected response was the significant positive correlation between
the number of eggs laid on L. tulipifera and F. americana. This correlation is likely a
reflection of a reduced specificity for P. tremuloides being exhibited as an increased

proportion of eggs laid on the other preferred hosts in the oviposition hierarchy.

50

Table 3.2. Bootstrapped variance estimates of the proportion of eggs laid by the Papilio
canadensis “early flight” population and the Spearman correlations between the

proportion of eggs laid for the three hosts that indicated any significant correlations
performed of the same population.

 

Spearman correlations

 

 

Plant Variance t SE P. tremuloides L. tulipifera F. americana
Populus tremuloides 0.057 :1: 0.017 1 x x
“73%;?” 0.026 :t 0.007 -0751 1 x
Fraxinus americana 0.027 1: 0.014 -0.721 0.541 1
Ptelea trifoliata 0.006 :1: 0.001 -0.18 0.29 0.09
Betula papyn'fera 0.008 r: 0.002 -0.01 0.16 -0.07
Lindera benzoin 0.009 :1: 0.003 -0.06 -0.04 -0.07
Prunus serotina 0.014 1 0.008 0.24 -0.33 -0.16
Other 0.001 1 0.0003 0.31 -0.35 -0.11

 

lSignificant correlations (P<0.05).

51

Differences in oviposition preference hierarchy between Papilio glaucus and Papilio
canadensis

Seven choice assays (2006).The first three axes of the PCA performed on the seven
choice oviposition data of the “early flight”, northern Michigan, Pennsylvania, and
Georgia populations accounted for 85% of the variance in the data. The first principal
component was primarily correlated to the number of eggs laid on L. tulipifera and P.
tremuloides (Table 3.1), with signs in opposite directions. Individual scores along this
axis were significantly different among populations (x2= 29.79, d.f. = 3, P < 0.001), and
Mann-Whitney U-tests indicated significant differences between the P. canadensis
populations and the P. glaucus populations (Table 3.3). The second axis was primarin
correlated in one direction with L. tulipifera and P. tremuloides and in the opposite
direction with P. trifoliata and F. americana . This second axis appears to relate to
differences in preference for secondary hosts relative to the preferred hosts of the two
species. No significant differences were observed between populations along this axis (12
= 3.44, d.f. = 3, P = 0.33). Finally, the third axis was primarily correlated with F.
americana in one direction and a weaker correlation with P. trifoliata in the other
direction and as with the previous axis no difference was observed among populations (12

= 0.23, d.f. = 3, P = 0.97).

52

Table 3.3. Differences between the Papilio canadensis (Pc) early flight” (11 = 26) and
northern Michigan (11 = 20) populations, and the Papilio glaucus (Pg) Pennsylvania (n =
20) and Georgia populations (11 = 15) for the individual scores from the first axis of the
PCA and the eighth axis of the Varimax rotation performed on the proportion of eggs laid
on seven choice arenas.

 

 

Population Principal component 1 Varimax axis 8
Early flight (Pc) A A
Northern Ml (Pc) AB AB
Pennsylvania (Pg) C B
Georgia (Pg) BC B

 

Populations with the same letters within a column indicate no significant differences from
Bonferroni corrected Mann-Whitney tests (P<0.05)

53

Varimax rotation yielded five axes (axes 4, 5, 6, 7, and 8, Table 3.4) that were not
strongly correlated to either L. tulipifera or P. tremuloides. Axis 4 was primarily
correlated to P. trifoliata, Axis 7 was primarily correlated to P. serotina, and axis 6 was
primarily correlated to B. papyrifera and F. americana. There were no significant
differences between the populations for these three axes (x2 = 3.72, d.f. = 3, P = 0.29; x2 =
5.88, d.f. = 3. P = 0.12; and x2 = 4.23, d.f. = 3, P = 0.24, respectively). Axis 5 was
primarily correlated to P. trifoliata and B. papyrzfera, and although a Kruskal-Wallis test
identified a significant difference between populations (x2 = 10.83, d.f. = 3, P = 0.013),
Mann-Whitney tests failed to detect significant differences between populations. In
contrast, significant differences between populations (x2 = 13.49, d.f. = 3, P = 0.004;
Table 3.4) were observed along axis 8 which was primarily correlated with B. papyrifera
and F. americana in opposite directions.The proportion of eggs laid on each host by each

population is represented in Figure 3.2.

Two choice assays (2006). Results from two-choice oviposition assays indicated
significant differences between the P. canadensis “early flight” and P. glaucus
Pennsylvania populations. The proportion of eggs laid on L. tulipifera was significantly
different between the “early flight” and Pennsylvania populations and within each
population between the L. tulipifera -P. tremuloides and L. tulipifera -P. serotina
oviposition arrays (Figure 3.3A). Similarly, we found significant differences between the

two populations in the proportion of eggs laid on P. serotina , but we only found

54

Amodvm 050343050 0:20.238 c0953 0:20.030 baa—0053mm 803 £03 00:0 .80. mcouflsmom 50253 0080mm
4.2008 mouméoxob 05 020:: 03? Emma: 0:800 0&8on 05 03% $20?
.302: 0.030-365 05 0025 029 $093 om80>0 05 2690 0020? «so

 

 

 

 

.000? 000$ $0.9 $0.3 $0.9 000.0 800.0. .000? 85900 80000> s
00.0- 00.0 00.0 00.0- 00.0 00.9 00.0- 00.0. .050
2.0 .000 :0. 9.0 9.0- 0 F0 :0 9.0 00.0008 0055
P000 000 30 S0- 000. 00.0 00.0 000 008.0050 asset
N00.0. 8.0- N~00 P000 20. B0 B0 00.0 000$qu seem
E0 0.? «0. N30 .20 00.0 3.0- 2.? 0000005 020E
00.0- 5.0 2.0 00.0- 000 .R0 000 000 000039000 003000
2.0- 5.0 00.0 2.? 5.0 “ts? P~00 N00.0 58000 000::
00.0 00. ~00- :0 ~00 0 F .0- P000. P000 $000.05 09000000:
.0 s 0 0 v 80 s N m 000..
008 metm>

 

 

A2 .I. 5 macaw—anon 29 «$800 050. SN n 5 2nd «Emirwgom 0.333% 0.530%
05 98 0003003050” 8N u 5 a): gwfloaz 0.55.5: Ea Gm u 5 Q5 Ema base 000.5338 0.550% 09 com 9300005 0065.530
co EB mwwo mo :omtomoa 05 mo 088 80:09:00 08653 2: .8.“ $5002 0800.“ on... we cease §§> 3 38.5.0 mox< in 030,—.

.0880 80200820 05 mo 00.003 8 890m

05 so 20— mwwo 088082 bemoan”. 1050.. BE. .8983 4 93 62.2900. N 6.050050% .m .8835...» .& 6:32.086 .k. 60303808

N 6.08830 4 0.83 1800005 05 8 8000.8 0963 doze—anon 20v 8980 05 80¢ 02080.0 0.3.383. N 2 080 doze—30a
0802383 05 89¢ 00380.0 35 0.308% 0.530% cm 8300—288 927C 59:32 805.5: 05 80¢ 02080.“ 8303350

.m cm dots—smog £62035 have 05 80¢ 03080.“ 85 80.208028 0.530% cm 05 3 BE mwmo .00 mm + 0383th 832 .N.m 0.88m

occasion.
an: E as: :22 on: “m

 

W

a

e

. u
__ 0000.0: o/o
. _ a
6

6

s

O? m.-

m

_- om

row

 

56

SD

—\
I.
0..

0.9 C i

0.8 l
0.7 e
0.6 1
0.5 1
0.4 l d

.l
3.3 i

 

Proportion of eggs laid on L. tulipifera

 

P. canadensis P. glaucus

Figure 3.3. (A) Median proportion of eggs laid, with interquartile ranges, by Papilio
glaucus (Pg) and Papilio canadensis (Pc) butterflies on L. tulipifera (Lt) in two-choice
assays. Leaves present in the bioassays were L. tulipifera and P. serotina (Ps) or L.
tulipifera and P. tremuloides (Pt). For each leaf the pairwise differences between the
proportions of eggs laid on it in the different arrays (host and butterfly set-ups) were
analyzed using Mann-Whitney tests. Arrays with the same letter did not have a
significant difference at a Bonferroni corrected a<0.05.

57

 

GB

8 11. c

g 0.9 4

"2 0.8l

‘L 1

c 0.7

0 0.61

E I b

g 0.51

§ 0.4 I b

w 0.3 i

Q- | a

2 0'2l

.2 0.1 l i

t

a O-l— — , 1 ____ -1?

g Ps-Pt Ps-Lt Ps-Pt ; Ps-Lt
l
l Pc Pc 1 Pg | pg

P. canadensis P. glaucus

Figure 3.3. (B) Median proportion of eggs laid, with interquartile ranges, by P. glaucus
and P. canadensis butterflies on P. serotina in two-choice assays. Leaves present in the
bioassays were P. serotina and L. tulipifera or P. serotina and P. tremuloides. For each
leaf the pairwise differences between the proportions of eggs laid on it in the different
arrays (host and butterfly set-ups) were analyzed using Mann-Whitney tests. Arrays with
the same letter did not have a significant difference at a Bonferroni corrected 0.<0.05

58

Proportion of eggs laid on P. tremuloides

0.8

\

P. canadensis

 

b
C
___‘___m _ __,_.
Pt-Lt ! Pt-Ps
Pg l Pg
P. glaucus

Figure 3.3. (C) Median proportion of eggs laid, with interquartile ranges, by P. glaucus
and P. canadensis butterflies on P. tremuloides in two-choice assays. Leaves present in
the bioassays were P. tremuloides and L. tulipifera or P. tremuloides and P. serotina. For
each leaf the pairwise differences between the proportions of eggs laid on it in the
different arrays (host and butterfly set-ups) were analyzed using Mann-Whitney tests.
Arrays with the same letter did not have a significant difference at a Bonferroni corrected
a<0.05.

59

differences between arrays for the Pennsylvania population (Figure 3.3B). Finally, for the
proportion of eggs laid on P. tremuloides, we again saw significantly different responses
between populations, but only significant differences between arrays for the Pennsylvania

population (Figure 3.3C).

Seven choice assays (1993-1998). Oviposition behavior of the females from 1993, 1996,
and 1998 indicated similar results to those found in three choice assays (Scriber et al.,
1991; Scriber, 1994). Here we again observed that hybrids had a similar oviposition
pattern to that of the fathers genotype. In this case F1 hybrids of P. canadensis females

by P. glaucus males had a similar oviposition pattern to that of P. glaucus (Figure 3.4).

Ovpositional behavior of the “late flight” population.

The first three axes of the PCA performed on the combined three choice oviposition data
of the “early flight”, northern Michigan, Pennsylvania, Georgia, and ”late flight”
populations accounted for >99% of the variance in the data. The first principal
component was primarily correlated to the proportion of eggs laid on L. tulipifera and a
weaker correlation with the proportion of eggs laid on P. tremuloides (Table 3.5), with
signs in opposite directions. This axis is likely a representation of the lower acceptance of
L. tulipifera characteristic of the difference between P. glaucus and P. canadensis.
Individual scores along this axis were significantly different between populations ()8 =
32.02, d.f. = 4, P<0.001), and Mann-Whitney U-tests indicated significant differences
between the “early flight” and northern Michigan P. canadensis populations and the two

P. glaucus and “late flight” populations (Table 3.6).

60

100*

 

 

 

 

 

 

 

 

 

llllllllllllllllllllilillllJl

90 —

80 A 7/////A ”I R. oath
E 70 J IP. semi
8’ 60 I EL. benz
8, 50 R up. tn'f
i 40 4 IF. amer
g 30 1 DP. trem

2° ‘1 _!_.’-_-_.‘.‘£’1’P_.

10 l

o .1__ L +_ g

 

P. canadensis F1 Hybrid P. glaucus

Figure 3.4. Mean percentage of eggs laid by Papilio glaucus (n = 38) from Ohio and
Kentucky, Papilio canadensis (n = 20) from Alaska and northern Michigan, and F1
hybrids (n = 30) from P. canadensis mothers and P. glaucus fathers. Leaves present in
the bioassay were L. tulipifera, P. tremuloides, F. americana, P. trifoliata, L. benzoin, P.
serotina, and R. cathartica.

61

 

Mean % eggs laid

N
O

_s
o

 

EF (PC) NMI (PC) LF (HS) GA (Pg) PA (Pg)

Figure 3.5. Mean percentage + SE of eggs laid by 94 Papilio canadensis (Pc) females
from the “early flight” population (EF), 27 P. canadensis females from the northern
Michigan population (N MI), 29 females from the “late flight” (LF) hybrid swarm (HS)
population, 25 Papilio glaucus (Pg) females from the Georgia population (GA), and 93 P.
glaucus females from the Pennsylvania (PA) population. Leaves present in the bioassay
were L. tulipifera, P. serotina, and P. tremuloides. The “other” category represents eggs
laid on the paper or plastic of the oviposition arena.

Table 3.5. Factor loadings for the first three principal component axes of the proportion
of eggs laid on three choice oviposition assays for the Papilio canadensis early flight” (11
= 94) and northern Michigan (11 = 27) populations, the Papilio glaucus Pennsylvania (11 =
93) and Georgia populations (11 = 20), and the “late flight” hybrid swarm population (11 =
29).

Principal component axis

 

 

 

13 23 33
Leaf
L. tulipifera 0.8441 -0149 0.124
P. serotina -0.255 0.6532 0.5092
P. tremuloides 0451 0713‘ 0.197
Other -0.138 0.209 0829‘
% Variance Explained 61.1% 21.0% 17.9%

 

lValues above the average largest value under the broken-stick model.
2Values above the average second largest value under the broken-stick model.
3 Axes for which individual scores were significantly different between populations.

63

Table 3.6. Differences between the Papilio canadensis (Pc) early flight” (11 = 94) and
northern Michigan (11 = 27) populations, the Papilio glaucus (Pg) Pennsylvania (n = 93)
and Georgia populations (11 = 20), and the “late flight” hybrid swarm population (HS) (11
= 29) for the individual scores from the first two axis of the PCA performed on the
proportion of eggs laid on three choice arenas. Populations with different letters within a

column indicate significant differences from Bonferroni corrected Mann-Whitney tests
(P<0.05)

 

 

Axis
Population 1 2
Early flight VT (Pc) B B
Northern Ml (Pc) B C
Pennsylvania (Pg) A A
Georgia (Pg) A A
Late flight VT 4H8) A A

 

64

The second axis was primarily correlated in one direction with P. tremuloides and
to a lesser extent P. serotina in the opposite direction (Table 3.5). Significant differences
were observed between populations along this axis (78.; = 55.45, P < 0.001), not
surprisingly as the acceptance of P. tremuloides is also a characteristic difference
between P. canadensis and P. glaucus. However, significant differences along this axis
were not only observed between the “early flight” and northern Michigan and the other
populations, but also between the “early flight” and northern Michigan population (Table
3.6). The differences observed between the “early flight” and northern Michigan
populations and the other three populations are likely to be a representation of the very
low number of eggs laid on P. tremuloides by individuals other than P. canadensis. On
the other hand the difference between the “early flight” and the northern Michigan
population are likely due to differences in specificity between the two populations.

The third axis is difficult to interpret as it is primarily correlated with the
proportion of eggs laid on non-leaf portions of the arena and to a lesser extent 0n P.
serotina in the opposite direction (Table 3.5). This axis could be construed as a
representation of the types of “mistakes” made by insects with a certain level of residual
excitation, as this axis represents the proportion of eggs laid on a host low on the
hierarchy (Figure 3.2) and eggs laid on non-leaf portions of the arena. As seen with the
previous axes, significant differences were observed among populations (x2 = 30.67, d.f.

= 4, P<0.001; Table 3.6).

65

Discussion

The selection pressure for specialization as a function of encountering the wrong host has
been noted since the first model used to explain the predominance of specialist insects
was developed (Levins & MacArthur, 1969). The probability that P. canadensis females
will encounter most of the toxic hosts used by its sister species (e. g., L. tulipifera) are
very low. It is therefore feasible that due to the low likelihood of encountering acceptable
toxic hosts, a lower overall specificity may be sufficient to incorporate a host for which
there previously was very limited acceptability (e. g., P. tremuloides). Under such a
scenario, the oviposition differences between P. glaucus and P. canadensis seen in three
choice oviposition assays (Scriber et al., 1991; Figure 3.1) could simply be a change in
specificity. Courtney et al.(1989) noted that changes in specificity are far more likely to
occur due to the higher level of genetic variance likely to exist for specificity while host
rank hierarchies are more likely to remain stable. In P. glaucus this appears to be the
case, as genetic variance between populations is seen for specificity of oviposition
preference (Bossart & Scriber, 1995), while its host plant preference hierarchy is
maintained throughout its range (Mercader & Scriber, 2005). However, in this study we
detected a shift in the preference rank hierarchy between P. canadensis and P. glaucus
(Tables 3.2, 3.4 and Figure 3.2). This illustrates an example of a shift in the oviposition
preference rank hierarchy during the polyphagous stage of a lineage, without reducing the
acceptance rates of secondary hosts.

Whether this shift in preference rank hierarchy between P. glaucus and P.

canadensis was due to an acceptance of plants in the Salicaceae by P. canadensis or a

66

loss of acceptance for plants in the Salicaceae by P. glaucus is difficult to determine.
Within the Papilio glaucus group, another generalist species closely related to P. glaucus
and P. canadensis, Papilio rutulus, also naturally uses plants in the Salicaceae (Scriber,
1996), and it is not possible to resolve whether the acceptance of the Salicaceae by
ovipositing females arose independently in P. rutulus and P. canadensis or once in the
ancestor of P. rutulus, P. canadensis, and P. glaucus and was then subsequently lost in P.
glaucus. However, ovipositing females of both P. multicaudatus [closely related and
basal to P. glaucus, P. canadensis, and P. rutulus (Zhakharov et al. , 2004)] and P. rutulus
in oviposition arenas containing suitable host plants will accept L. tulipifera (RJ
Mercader and J M Scriber unpubl.), although it is toxic to their larvae (Scriber, 1996). In
contrast, P. multicaudatus, like P. glaucus, has a very low acceptance rate for P.
tremuloides, suggesting that within the P. glaucus group acceptance of Magnoliaceae
hosts by ovipositing females is likely primitive, while acceptance of Salicaceae hosts is
likely derived.

In addition, the oviposition behavior observed for P. glaucus and P. canadensis
was consistent with what would be expected for the acceptance of a new host as
described by Courtney et al.(l 989). In particular, we observed the acceptance of a new
host and the ancestral host by P. canadensis, and very limited acceptance of the novel
host by P. glaucus (Figure 3.2). Surprisingly, the acceptance of P. tremuloides did not
significantly alter the response of P. canadensis for most of the secondary hosts also used
by P. glaucus (Figure 3.2 and Tables 3.2, 3.3, and 3.4). While we found significant
differences in the seven choice assays between populations for axes relating to L.

tulipifera and/or P. tremuloides (Tables 3.2, 3.3, and 3.4), the only difference along any

67

axes we found, that was not primarily related to P. tremuloides or L. tulipifera, was the
difference between the “early flight” P. canadensis population and the two P. glaucus
populations along the eighth Varimax axis (Tables 3.3 and 3.4). This axis was primarily
related to F. americana in one direction and to a lesser extent B. papyrifera in the
opposite direction, which may be an indication of a higher acceptance of F. americana in
P. glaucus and higher B. papyrifera acceptance in P. canadensis. Unfortunately, we are
unable to distinguish if this small exception is due to the strong correlation between L.
tulipifera and F. americana or if it is an independent response. However, even with this
small exception our results strongly indicate that the primary difference between the two
species is a reduced acceptance of L. tulipifera and a preference for P. tremuloides in P.
canadensis relative to P. glaucus. Our two-choice oviposition assays also supported this
notion. In these assays P. canadensis females laid a significantly lower number of eggs
on L. tulipifera and a significantly higher proportion of eggs on P. tremuloides than P.
glaucus (Figure 3.3). It is interesting to note that no significant differences were found
amongst P. canadensis females in the number of eggs laid on P. serotina in assays
containing L. tulipifera or P. tremuloides (Figure 3.38).

The Papilio glaucus group is unique amongst the Papilionidae in its high degree
of polyphagy and by its “escape” from the host specializations on tropical plant families,
such as Lauraceae, Magnoliaceae, and Rutaceae, which along with the Aristolochiaceae,
Annonaceaee, and Apiaceae constitute the diet of 90-95% of all Papilionidae (Scriber,
1996). The shift in host plant use in the Papilio glaucus group (e. g., to include the
Salicaceae, Betulaceae, Rosaceae, and Oleaceae) may have been facilitated by the alleles

coding for these preference differences having likely been Z-linked (as seen for the

68

Salicaceae in P. canadensis (Scriber, 1994; Figure 3.4)). Z-linked differences in
oviposition preference rank hierarchies may be an important component of butterfly host
associations as has been observed in other related and unrelated groups (Thompson,
1988, 1993, 1994; Janz, 1998; Nygren et al. , 2006). As proposed for female preference of
male ornaments (Reeves & Pfennig, 2003), Z-linked traits under positive selection
expressed only by females in species with a ZZ/ZO sex determination (female
heterogamety) are less likely to be lost to genetic drift, and therefore it is more likely for
these rare alleles to become fixed in populations.

Scriber (1994), using F1 P. glaucus and P. canadensis reciprocal hybrids, found
the difference in three choice oviposition trials (similar to the ones reported here) were Z-
linked. Similarly, in this study we also found corresponding responses in seven choice
assays between P. glaucus and hybrids with P. canadensis mothers and P. glaucus fathers
(Figure 3.4). This result further supports the notion that the primary difference observed
between P. glaucus and P. canadensis is their relative preferences for L. tulipifera and P.
tremuloides, which appears to be Z-linked. It is unknown whether the Z-linked
oviposition difference is due to a single Z-linked preference for P. tremuloides in P.
canadensis and a separate preference for L. tulipifera in both species of uncertain
inheritance, or two Z-linked traits; one for the acceptance of P. tremuloides and L.
tulipifera in P. canadensis and one for only the acceptance of L. tulipifera in P. glaucus.
Previous studies using three choice oviposition assays of P. glaucus, similar to the ones
reported here, suggest that the acceptance of P. tremuloides may be independent from
specificity for L. tulipifera (Mercader & Scriber, 2005). In particular, they indicate a very

weak negative relationship between the proportion of eggs laid on L. tulipifera (a

69

measure of specificity) and P. tremuloides (adjusted r2 = 0.09) compared to the negative
relationship between the proportion of eggs laid on L tulipifera and P. serotina (adjusted
3:06m.

In addition to differences between P. glaucus and P. canadensis, differences in
oviposition profiles have been noted between populations of P. canadensis in natural
field conditions (Scriber, 2002a) and choice assays not containing L. tulipifera (Scriber &
Lederhouse, 1992; Scriber, 1996b). The differences in oviposition profiles constituted
reductions in the proportion of eggs laid on P. tremuloides and increases in the proportion
of eggs laid on F. americana in populations located in areas where local cold
temperatures, “Cold Pockets”, significantly reduced the quality of the available P.
tremuloides. These observed shifts in oviposition profiles appear to represent multiple
independent changes in host rank hierarchy in these “Cold Pocket” populations.
However, the significant correlation between the proportions of eggs laid on L. tulipifera
and F. americana and the significant negative correlation between the proportion eggs
laid on P. tremuloides and these two hosts in P. canadensis (Table 3.2), hints at an
alternative mechanism by which this shift may have occurred. A reduced specificity for
P. tremuloides would likely be reflected as a preference for L. tulipifera or F. americana
in P. canadensis, which in oviposition assays without L. tulipifera would be represented
as an apparent preference for F. americana. In this case it is likely that changes in
specificity for P. tremuloides have led to an apparent shift in host rank hierarchy for the
locally highest quality host. Whether changes in specificity and host ranking are simply
differential expressions of the same genes, as has been proposed for Polygonia c-album

(Janz, 2003), or due to different genes is unknown.

70

Irrespective of the exact mechanism behind the differences in specificity and host
preference ranking, the combination of the acceptance of a novel host family (Salicaceae)
and the potential for changes in specificity provide a likely avenue for diversification of
P. canadensis. One potential reason for the lack of divergence observed is the low
selection pressure for plant specialization predicted for populations of univoltine species
with relatively long growing seasons (Scriber & Lederhouse, 1992) and the absence of L.
tulipifera in the vast majority of P. canadensis range. In addition to not encountering
ancestral hosts, it is likely that even if acceptance of toxic hosts via contact
chemoreception exists, changes in other senses may be likely to change the functional
response in the field. For example, responses to extracts from dried leaves of L. tulipifera
elicit three-fold higher electroantennagram responses in P. glaucus than P. canadensis
(Mercader et al. In Press). It is possible that changes in the perception of visual or
olfactory cues could lead to functionally specialized populations in the field which we are
unable to distinguish in our oviposition arenas.

The recently described “late flight” hybrid swarm population (Scriber & Ording,
2005; Scriber et al., 2007) provides another avenue that may lead to specialization in this
group. This population is thought to have formed due to recent increased introgression of
P. glaucus alleles into P. canadensis populations leading to Z-chromosome recombinant
hybrids which resulted in an allochronically separated population. This population
represents a mixture of the P. canadensis and P. glaucus genomes including a
recombined Z-chromosome with Z-linked allozymes diagnostic of both P. glaucus and P.
canadensis (Scriber & Ording, 2005). A priori, we would have expected either a P.

canadensis oviposition pattern or a mixture of P. canadensis and P. glaucus oviposition

71

patterns amongst females. This expectation was due to another Z-linked P. canadensis
trait, obligate diapause, being essential for survival in this region where two generations
are not possible. In this “late flight” population we found the oviposition profile in three
choice assays to be that of P. glaucus. In particular we found significant differences with
P. canadensis and not P. glaucus populations along PCA axis 1 and 2 (Table 3.6). Axis I
was primarily related to proportion of eggs laid on L. tulipifera, while axis 2 was
primarily related to the proportion of eggs laid on P. tremuloides in one direction and to a
lesser extent the number of eggs laid on P. serotina in the opposite direction (Table 3.5).
Together these axes represent the main difference between P. glaucus and P. canadensis
observed in the seven choice assays (a lowered preference for L. tulipifera and a
preference for P. tremuloides) and indicate that the “late flight” hybrid swarm population
has a P. glaucus oviposition profile.

This oviposition profile has significant implications for the ecology and potential
diversification of this population as plants in the Magnoliaceae are absent in or north of
the historical hybrid zone where the “late flight” population occurs (Scriber & Ording,
2005; Scriber et al., 2007), making L. tulipifera an unlikely host. However, F. americana
is abundant where the “late flight” occurs, and although we did not assay this population
for F. americana, both P. glaucus and P. canadensis use F. americana as a host.
Therefore it is highly likely that the “late flight” hybrid swarm population may be
primarily utilizing F. americana, which if it becomes/remains reproductively isolated
could lead to specialization on F. americana. Future studies on the host use of this
Papilio hybrid swarm population and populations from the putative species P.

appalachiensis (Pavulaan & Wright, 2002) may help elucidate whether such a change is

72

likely and provide an avenue to study the process of specialization and possible

constraints to specialization in these organisms.

73

CHAPTER 4

HYBRIDIZATION LEADS TO A HOST SHIFT IN A POLYPHAGOUS

BUTTERFLY SIBLING SPECIES PAIR.

Abstract

Recent increased movement of Papilio glaucus into historically P. canadensis territory,
due to climate warming, has led to the formation of an allochronically separated hybrid
population with a delayed emerging phenotype or “late flight” first observed in 1999 in
the Battenkill River Valley, at the New York-Vermont border, USA. Here we assess how
the recombination of the parental genomes leading to the formation of this phenotype
may have facilitated another major ecological shift, host use divergence. We first contrast
the ovipositional profiles of the “late flight” population to that of the parental species P.
glaucus and P. canadensis. Second we contrast larval survival and grth on five hosts
of the “late flight”, a P. canadensis population, a P. glaucus population, and another
population from the Northern edge of the hybrid zone. Our results indicate that the
ovipositional preference of this hybrid swarm is identical to that of the introgressing
parental species, P. glaucus. Due to the absence of the preferred hosts of P. glaucus
(Liriodendron tulipifera L. and Ptelea trifoliata L.) where the “late flight” occurs, this
ovipositional pattern implies an apparent functional shift onto a secondary host of both
parental species, F raxinus americana L. In contrast, the larval host use abilities represent
a mixture of P. glaucus and P. canadensis, indicating divergence in larval host use

abilities has not taken place. However, high genetic variability (CV0) is present for

74

growth on F. americana in the “late flight” hybrid swarm and tradeoffs for larval
performance on the preferred hosts of the parental species are evident (L. tulipifera and
Populus tremuloides Michx.); indicating a strong potential for specialization in larval
host use abilities to occur. This current scenario represents an instance where a shift in a
major ecological trait, host use, is likely occurring as a byproduct of a shift in an

unrelated trait (delayed emergence) leading to partial reproductive isolation.

Introduction

Recent anthropogenic changes to the environment are creating novel selection
pressures upon organisms that have already lead to several examples of rapid evolution
(e.g. examples reviewed in Thompson 1998; Palumbi 2001). These changes not only
affect individual populations or species, but can affect entire communities by altering the
interactions between organisms. This has been seen in the interactions between plant
feeding insects and their hosts due to plant invasions (Strauss et al. 2006; Carroll 2007;
Singer et al.2008) and land use change (Singer et al.2008). The interactions between
plants and plant feeding insects are vital for community composition and function (e. g.
Weisser and Siemann 2004), and are arguably the most common interaction between
multicellular organisms in terrestrial systems (plants and insects compose approximately
40% of terrestrial organisms; Price 2002). Due to the ubiquity of plant insect interactions
and their important role in community composition, understanding the potential (and
documented) impacts of environmental change on plant-insect associations is of

significant importance.

75

The high diversity of plant feeding insects is believed to be due to their radiation
onto the wide chemical diversity of plants (e. g. Ehrlich and Raven 1964; Mitter et
al. 1988; Futuyma 1989; Weingartner et al.2006). In particular, this divergence is
believed to be the result of the colonization of novel hosts, making novel niches available
for adaptive radiation to take place (e.g. Jermy 1984; Janz et al.; 2001; Jenny and Szentsi
2003; Percy et al.2004; Braby and Trueman 2006; Lopez— Vaamonde et al. 2006; Wheat
et al.2007). Although specialization would appear to be an evolutionary dead end,
transitions between generalist and specialist forms within lineages are believed to be
common (Janz et al.2001; Nosil 2002; Scriber et al. 2008b) and a hypothesis of
oscillating host expansion and specialization has been proposed to help explain the great
diversity of plant feeding insects (Janz et al.2006; Janz and Nylin 2008). However, how
the genetic variation leading to these host shifts arises and is maintained is not clearly
understood for most systems.

One probable source of genetic variation leading to isolation and host shifts is
hybridization, as has been recently observed in Rhagoletis fruit flies (Schwarz et al.
2005). The notion that hybridization can form novel phenotypes leading to speciation is
not a new idea, and can be traced as far back as Linnaeus (Arnold 1997; Coyne and Orr
2004). However, while botanists have acknowledged the importance of hybridization as
an evolutionary force generating novel genotypes, zoologists have historically viewed
hybridization primarily as maladaptive (reviewed in Arnold 1997; Coyne and Orr 2004).
In recent years the role of hybridization as an evolutionary force generating novel
genotypes has been gaining support, and begun to be accepted in animal taxa as well (e. g.

reviews by Seehausen 2004; Mallet 2007). The importance of hybridization as an

76

evolutionary force centers on the view that the formation of novel, transgressive,
phenotypes by hybridizing organisms allows for the colonization of open niches,
facilitating speciation and adaptive radiations (Seehausen 2004; Mallet 2007).

Implicit in hybridization is the recombination of genomes, which will
undoubtedly lead to the recombination of most traits in organisms, not only those that
may potentially lead to reproductive isolation. If hybridization leads to a reproductively
isolated or partially isolated population, the evolutionary trajectory of such a population
would be constrained by the genetic variability inherited from the parental populations.
This may or may not be an equal representation of both parental populations. The
importance of phylogenetic constraints on insect radiations has been widely recognized in
the study of plant-insect interactions (e. g. Ehrlich and Raven 1964; Price 2008). In the
case of reproductively isolated (or partially isolated) populations resulting from
hybridization events, the recombination of traits that takes place may lead to novel
constraints and/or the release from constraints (relative to the parental populations)
unrelated to the trait(s), leading to isolation.

A recent instance of a hybridization event leading to a transgressive phenotype
has occurred in the Battenkill River Valley, at the New York-Vermont border, USA. In
this region, recent climate warming has greatly increased the gene movement across a
hybrid zone delineating the range limits of the Eastern Tiger Swallowtail, Papilio
glaucus, and the Canadian Tiger Swallowtail, Papilio canadensis (Scriber 2002). Prior to
recent warming trends, this region was near the Southern edge of the univoltine P.
canadensis range, where only one generation of P. canadensis or P. glaucus could

develop starting at the beginning of the growing season. However, after 1998 this region

77

warmed significantly and became capable of supporting an early emerging population
(typical P. canadensis or P. glaucus phenotype) as well as a potentially late emerging
population (Scriber et al.2008a). The increased movement in P. glaucus traits into the
region has produced a transgressive phenotype with delayed emergence, or “late flight”,
forming an allochronically separated hybrid swarm population that can exploit this novel
thermal zone (Scriber and Ording 2005; Scriber et al.2008a).

The host use abilities of this “late flight” are relatively unknown as the two
parental species P. glaucus and P. canadensis are both highly polyphagous species, but
have differing ovipositional preferences (Scriber etal.1991; Mercader and Scriber 2007)
and larval host use abilities (Scriber 1996). Preliminary evidence from three choice
oviposition assays indicate that the “late flight” population has a similar ovipositional
profile to P. glaucus (Mercader and Scriber 2007), and the ability to survive on hosts of
both P. glaucus and P. canadensis (Scriber et al. 2008a). In this study we assess how the
formation of this delayed emerging phenotype may have affected host use traits in this
“late flight” population. We use bioassays for both oviposition and larval host use
abilities on plants of variable quality for both parental species. We first contrast the
ovipositional profiles of the “late flight” population to P. glaucus and P. canadensis
using oviposition assays containing host plants of P. glaucus and/or P. canadensis from
seven different plant families. Secondly we contrast survival and growth of the “late
flight”, P. glaucus, P. canadensis, and a population located at the Northern edge of the P.
glaucus and P. canadensis hybrid zone on leaves from five plant families. Finally, we

present a quantitative genetic analysis for larval host use abilities of the “late flight”

78

population using estimates of the broad sense heritability and coefficients of genetic

variation from larval grth assays on five hosts.

Materials and Methods

Insect and Plant Sources

Female butterflies from the “late flight” population used in seven choice oviposition
assays were a combination of wild-caught females and individuals reared from eggs laid by wild-
caught females. All collections of “late flight” individuals took place in mid to late July 2006
(mothers of mass reared individuals) and 2007 (wild-caught females) in the Battenkill River
Valley area at the NYN T border, USA (43° N latitude and 73° W longitude) as described in
Scriber & Ording (2005). All individuals were mass reared on P. serotina in sleeves placed on
tree branches in the Battenkill River Valley in 2006. Reared females were mated by hand pairing
them to a single wild caught male or lab reared male (not from the same sib group), wild females
were assumed to have been mated in the wild and were subsequently not hand paired. Eggs laid
by females ovipositing in our seven choice bioassays were collected daily and the emerging
larvae were used in our growth assays as described below.

Larvae fiom the remaining populations were collected from adult females ovipositing in
three-choice or two-choice arenas. In addition, the ovipositional patterns in seven choice assays
are well established for the P. canadensis and P. glaucus populations used (Mercader and Scriber
2005, 2007, 2008b). Larvae of P. glaucus from a Georgia population were obtained from the eggs
of wild captured females collected in Oglethorpe Co., GA, USA in 2007 (34° N latitude and 83°
W longitude). Larvae of P. canadensis from a Northern Michigan population were obtained from

eggs of field collected females in 2007 from Cheboygan and Emmet Counties MI, USA, (45° N

79

latitude and 84° W longitude). Finally, larvae from a population in the Michigan hybrid zone
were obtained from field collected females in 2007 from Isabella County MI, USA, (43° N
latitude and 84° W longitude) located at the Northern edge of the historical hybrid zone (Scriber
et al.2003). All wild caught butterflies were transported in coolers to our MSU laboratory where
they were fed a 15-20% honey water solution prior to placement in the oviposition arenas.

Leaves were collected from trees growing in Ingham County, MI, USA, from locations
known to be pesticide free. These were placed in aquapicks inside plastic bags, and kept in a
refrigerator for no more than 3 days to ensure leaf quality and turgor remained high. Leaves used
in the assays were collected from at least four different trees, with the exception of L. benzoin and
B. papyrifera for which two and one plants were used, respectively, since plants known to be

pesticide free were more limited for these two hosts.

Oviposition Assays

We placed butterflies in oviposition arenas containing leaves from plants in seven
different families of varying larval suitability for P. canadensis and P. glaucus (see
above); L. tulipifera (Magnoliaceae), P. tremuloides (Salicaceae), F. americana
(Oleaceae), P. trifoliata (Rutaceae), P. serotina (Rosaceae), B. papyrifera (Betulaceae),
and L. benzoin (Lauraceae). These seven choice bioassays were conducted between 15
May and 24 July 2007 using individuals from the Vermont “late flight” population (11 =
50). Only females that laid more than 20 eggs were utilized for later analysis (11 = 26).
The general methods used were similar to those described by Scriber (1993); arenas
consisted of clear round plastic containers placed on a rotating platform (approximately
10 revolutions per h) in front of incandescent lights on a L4:D4 hrs photocycle to

maximize egg laying. Leaves of each plant species were placed in floral aquapicks to

80

maintain turgidity. Butterflies were fed a honey-water solution daily and allowed to
oviposit until they were too weak to fly (4-12 days). The number of eggs laid on leaves of
each plant or non-leaf portions were counted daily and the leaves replaced as necessary.
No leaf remained for more than 2 days in the assays.

Ovipositional responses from these individuals were contrasted to those of P.
glaucus individuals from Lancaster Co. in South-Eastern PA, USA (40° N latitude and
76° W longitude) and P. canadensis individuals from the “early flight” in the Battenkill
River Valley area at the New York/Vermont border in 2006 for a previous study using
identical methodology (Mercader and Scriber 2007). Unfortunately we were unable to
procure sufficient specimens from these populations in 2007 to run the assays

simultaneously.

Larval Growth Assays

Eggs collected from ovipositing females were maintained at 25 °C and checked at least
twice daily for hatching larvae. To increase within family replication we did not include
all hosts used in the oviposition arenas, and limited our larval assays to the five most
commonly used hosts of P. glaucus and P. canadensis; L. tulipifera (Magnoliaceae), P.
tremuloides (Salicaceae), F. americana (Oleaceae), P. trifoliata (Rutaceae), and P.
serotina (Roseaceae). We allocated all recently hatched neonate larvae onto the five host
plants using a fine camel hair brush. Leaf petioles were supported in rubber-capped
water-filled vials to maintain turgor and leaves were checked daily to ensure high quality.
All larvae were maintained at a constant 25 °C with a 18:6 hrs (Light:Dark) photoperiod.

After three days larvae were checked for survival and instar achieved. After six days all

81

larvae were checked again for survival and instar achieved. They were then frozen at -20
°C. At the conclusion of the experiment all larvae were weighed to the nearest 0.1 mg
while still frozen to prevent water loss.

In total, 1388 larvae from 36 families of the “late flight” population, 401 larvae
from 13 families of the Northern Michigan population, 251 larvae from 8 families of the
Georgia population, and 417 larvae from 11 families of the Michigan hybrid zone
population were set up. For our grth studies we only used families for which at least
five larvae survived after six days. This reduced the number of larvae used in the grth
assays to 945 larvae from 25 families of the “late flight” population, 310 larvae from 11
families of the Northern Michigan population, 172 larvae from 7 families of the Georgia

population, and 326 larvae from 10 families of the Michigan hybrid zone population.

Statistical Analyses

Population contrasts

Oviposition: We contrasted the overall ovipositional profiles of P. glaucus, P. canadensis
“early flight”, and the “late flight” hybrid swarm by performing a one-way non-
parametric MANOVA (NP MAN OVA) on the proportion of eggs laid on each leaf by
individual butterflies in the seven choice bioassays. MANOVA has been shown to be an
effective tool to differentiate oviposition preference ranking using proportion data from
choice trials similar to our own (Thompson, 1988; Bossart and Scriber, 1995). However,
we were unable to satisfy assumptions of MANOVA through transformations and
therefore used a Non Parametric MANOVA (or Permutation MANOVA) as described by

Anderson (2001), and implemented in the PAST program (Hammer et al. , 2001), using

82

the Bray-Curtis distance matrix and 10,000 iterations. We then used pair-wise NP
MAN OVAs as post-hoe tests to determine significant differences between the three
populations. Significance for pairwise NP MANOVAs was accepted at Bonferroni

corrected P<0.05.

Survival: Survival between populations was analyzed separately for each host as repeated
measures analyses using Generalized Linear Mixed Models (GLMM) (Family =
Binomial, logit link) in the lme4 package (Bates and Sarkar 2007) of R (R Development
Core Team 2007). In these analyses the state of the larvae after six days (alive or dead)
was the dependent variable, population the fixed effect, and family was considered the
random effect. By declaring family as the random effect, family was considered our
“subject”, thereby constraining our analysis to comparisons of family performances,
rather than individual performances, between populations. We performed pairwise
contrasts of populations in these analyses using Tukey’s HSD for pairwise contrasts in

the multcomp package (Hothom et al.2007).

MU); Here we were interested in contrasting the relative performance of individuals
from families from different populations on each host, rather than overall size differences
between families or populations. In order to contrast the relative performance of
individuals on different hosts we took the Z-score of each individual weight by
subtracting the family average weight (on all hosts) from each individual weight (on a
single host) and dividing this value by the standard deviation for each family (on all

hosts). We then contrasted the Z-scores between populations for each host as for survival

83

using GLMM (Family = Gaussian) with population as the fixed effect and family as the

random effect and Tukey’s HSD for pairwise contrasts.

Analyses Within the Late Flight
We were interested in two primary comparisons within the “late flight” hybrid
swarm population.

1) Is there high genetic variation for growth on the different hosts?

As this “late flight” population is a known hybrid swarm from two species of
organisms with known differences in host use we are inclined to expect significant
variation in host use abilities. However, this may not necessarily be the case for all (or
any) of the hosts assayed. While significant differences have been observed in growth
rate on P. glaucus and P. canadensis, P. canadensis has not shown significant differences
in growth rate between populations in Michigan and Alaska (Ayres 1991), indicating a
potentially limited variation in detoxification abilities in P. canadensis. Secondly, it is
unknown if the recombination events and strong divergent selection on the X(Z)-
chromosome (leading to the formation of the “late flight”) population selectively
removed variation leading to reduced variation in host use ability for particular hosts.
Finally, we were unaware if there would be lower or greater variation for growth on L.
tulipifera and P. tremuloides relative to the common hosts F. americana and P. serotina,
or for P. trifoliata.

Quantitative genetic analyses of hybrids can be difficult to interpret, particularly
when considering the F. generation. Gordon (1999) derived the quantitative genetic

expectations for F 1 hybrid swarm populations and identified that additive genetic

84

variance and narrow sense heritability cannot be defined for these populations. However,
he found the genotypic variance and broad sense heritability to be defined for these
populations and useful determinants of genetic advance from selection. In addition,
Gordon (2000) derived the theoretical expectations for an allogamous F2 population,
derived from F1 crosses and determined that allogamous F 2 populations behaved in an
identical fashion to the randomly fertilized populations on which classical quantitative
genetic analysis is based. In this study we do not estimate additive genetic variation, as
we are using full-sib variance estimates, and are deriving our estimates for a likely F2+
population. Therefore, our main obstacle is the same as that for any other study based of a
full sib design; we are unable to separate additive and non-additive genetic components.

In order to determine the level of genetic variation within the “late flight” we
performed Mixed Model analyses for each host again using the lme4 package and
Markov Chain Monte Carlo estimates of variability for the parameters measured using
the mcsamp function in the arm package (Gelman et al.2008) for R. In these analyses we
included date of measurement as a fixed effect (to account for potential differences in leaf
quality as the season progressed) and family as a random effect. Using these data we
measured broad sense heritability Hz, the genetic coefficient of variation CV6, and
residual coefficient of variation CVR following Houle (1992). As we were only able to
sample full sibling families our estimates differ from those of Houle (1992), in our use of
total genetic variance (V0) rather than purely additive genetic variance (VA). The use of
V0 rather than VA will likely lead to an inflated value relative to the additive genetic
variance coefficient CVA and a deflated value for the CVR.

2) Are there significant interactions between host use abilities?

85

In order to determine if there were significant interactions between host use abilities
we performed two tests. First we ran linear Mixed Models using the lme4 package (Bates
and Sarkar 2007) for each host pair combination using larval weight as the dependent
variable, family as the random factor, host as the fixed factor, and date as a covariate. We
ran these models with and without an interaction between family and host and performed
a log likelihood ratio test for each host pair combination between the model with an
interaction between family and host and the model without the interaction. Our second
test consisted of taking the average family Z-score (calculated as above) for each host and
performing Spearman correlations between family averages on each host pair

combination.

Results

Oviposition Assays

The ovipositional responses of P. glaucus were virtually identical to those of the
“late flight” hybrid swarm population, whereas those of the P. canadensis “early flight”
population were different (Figure 4.1). Not surprisingly the NP MANOVA performed
found a significant overall difference in the ovipositional behavior of these three
populations (F259 = 11.8, P<0.001), and pairwise comparisons indicated significant
differences in the overall ovipositional responses between the “early flight” population

and the P. glaucus and “late flight” populations (Table 4.1).

Larval Growth Assays

86

_S_u__rv_iLal_: The results for 6-day survival for the four populations sampled are summarized
in Figure 4.2. Log likelihood ratio tests between models with family as a random effect
and with or without population as a fixed effect indicated a significant effect of
population for all hosts tested (L. tulipifera x 2 = 28.6, P < 0.001; P. tremuloides x 2 =
35.8, P < 0.001; F. americana x 2 = 18.0, P < 0.001; P. serotina x 2 = 20.2, P < 0.001; P.
trifoliata x 2 = 10.1, P =0.02). Within L. tulipifera we found significant differences in
survival between all populations except between the “late flight” and the Michigan
Hybrid zone population (Figure 4.2). Of particular note is the significant difference
observed between the “late flight” population and the Georgia P. glaucus population,
indicating reduced survival on L. tulipifera in this hybrid swarm relative to the pure P.
glaucus type (Figure 4.2). A similar result was observed for P. tremuloides with the
exception that no significant difference was observed between the Michigan hybrid zone
population and the Northern Michigan population (Figure 4.2). As with L. tulipifera the
“late flight” population had a lower survival on P. tremuloides than the reference P.
canadensis population (Figure 4.2). However, for F. americana, P. serotina, and P.
trifoliata significant differences were only found between the Northern Michigan
population and the “late flight” population (Figure 4.2). This may reflect a lower overall
survival in the “late flight” population. However, lower sample sizes for the Georgia and
Hybrid zone populations preclude us from being able to discern whether this difference is
due to high survival in the Northern Michigan population or low survival in the “late

flight” population.

87

.0000 08:88.5 05 m0 0:003 00 00000 08 :0 2.2 ammo 08000800 @0900

20050.. 00:. 8.80000 4 000 62.2300. N .Egksmum .m. 80.03.08 N 6208080 N 0030380.: N 605830 .N 0003 0.0000005 05

8 80008 m0>00q .Goomv 080>§maa0m 80¢ m0_080¢ AME 0.008% Smack om 080 A808 8030 2.5.3 2055 000? 05 80¢ 00—080¢
cm .688 08003000 :EmE 3.80.. 05 80¢ mag—08¢ 0000000200 0,0550% cm 08 .3 22 www0 ¢0 Mm + $080800 8002 .34 08me

200000000 .1

000003 .& n:

 

 

 

.050 0
$0800 4 8
02.0800 .1 n.

pgel $569 % ueew

0000503 .m N
0000055 .& a
0:00.580 .4 I

 

80.63800 .0 n
800005 .0 I

 

 

88

Table 4.1. P-values for pairwise Non-Parametric MANOVA’s contrasting the
ovipositional behavior in seven choice arenas. Upper triangle represent uncorrected P-
values and the lower triangle Bonferroni corrected P-values.

 

 

Population
Population Late Flight P. canadensis P. glaucus
Late Flight <0.001 0.518
P. canadensis <0.001 <0.001
P. glaucus 1 <0.001

 

89

.953 3003.5 ~0>0_ modvm 05 00 00000m$0 0:000meme 000 000 0030— 2:00 06 .3 0032—8 00008

0030—0000 360% 0000 0000 03:? 000.2052 M 000 62.20200 N 0200020220 E 603032202 N 020503004 000 00808009 0003
000000 0033 00 006000 .«02 02.0 000. .ANEV 00003000 0000 2.53 gwfiouz 05 m0 000 00003000 0&0000 §000~m N 05 («0 .3326
00003000 003032 0008002 022000200 N 05 .00 "E15 0000—0000 2035 000—: 05 m0 mm H H0323 002:0— 0000000 0002 .Né 05me

0%.: N 0020.... N .2020 M 0220.0 .& 0030 Q

 

 

W
3
m
w
m
w
s
n
W
N:
<0-
0
.220 a n0 0 0 0

u...-

90

Larval growth (Between Populations): A summary of the relative growth of the
caterpillars on each host as measured by family Z-Scores is illustrated in Figure 4.3.
These Z-Scores represent the growth of larvae in each family on one host relative to the
overall performance of that family on all hosts; therefore they are quantitative measures
of host suitability ranking within families. They are by necessity also only representative
of those caterpillars that survived. Log likelihood ratio tests between models with family
as a random effect and with or without population as a fixed effect indicated a significant
effect of population for L. tulipifera (x 2 = 14.7, P = 0.002), P. tremuloides (x 2 = 21.4, P <
0.001), F. americana (x 2 = 11.7, P = 0.009), and P. trifoliata (x 2 = 10.4, P = 0.015), but
not for P. serotina (x 2 = 2.7, P < 0.44). For L. tulipifera significant differences were
observed between the Northern Michigan population and the other three populations, but
no significant differences between any of the other populations (Figure 4.3). For P.
tremuloides significant differences were observed between most populations with the
exception of the Michigan hybrid zone population compared to the “late flight” or the
Northern Michigan populations (Figure 4.3). Finally, significant differences within P.
trifoliata were only observed between the Georgia P. glaucus population and the other

three populations (Figure 4.3).

91

.Emm 00003: ~0>0_ modV0 05 00 0000000000 3000200030 000 000 0000— 00000 05 3 0032—00 00008 00003000

0000000 0000 2000 00023 000.200.: N 000 02.20200 N 0200020220 N 002020002200 N 020003004 00 000000000 0003 000000 0033
00 0000000 000—0 02.0 00H .ANEV 0000—0000 0000 2.53 0030002 000 00 000 00303000 0&0006 00000~M N 05 00 .9206 0000—0000
00w2002 000500 Z 00020200200 N 06 00 A06 003000000 afiwa 000—: 05 00 mm + 538w RES 000 00000-N 0002 .m.0 00030

   
   

 

 

 

 

 
 
 

0X00 N 00200 N .2020 N 020.0 N N000 4 - N-
m M m . m
m m m m - me-
m M W m .0
m m m m H . F-
m M M m m
0 m u u m - md- e
. u u u U
a m n m u a.
m HHI w u u r O 0Q.»
. n n n O
u m m n n u . J
n u a n u - no a
NI! " n 0 u _ 0 0
<00 m m
n u n u H u _.
E‘ZU " n u m

 

u..- - we

92

Estimates oflgenetic variance for larval performance: We found fairly low estimates of
broad sense heritability for larval growth on all hosts measured (Table 4.2). However, we
found relatively high coefficients of genetic variability (CVG) for grth on P.
tremuloides (17.6) and P. trifoliata (23.4), lower for F. americana (11.1) and P. serotina
(11.01), and lowest for L. tulipifera (3.81). Although the CV0 for growth on L. tulipifera
was lower than that for the other four hosts it is still similar to those measured for adult
body size in Drosophila (Wayne et al. 1997). Morphological traits have ofien been found
to have high heritability estimates and low coefficients of genetic variance (Houle 1992).
In contrast, life history traits tend to have low heritability estimates and high coefficients
of genetic variability (Houle 1992). As in other studies (Houle 1992) the low heritability
and high CVG in this study is likely due to the presence of high residual variance
observed for grth on all hosts measured (CVR; see Table 4.2). These results indicate
that in this hybrid swarm population there is a high level of genetic variability for this
important life history trait (larval host use ability), but not equal levels of genetic
variability for the ability to use different hosts.

In addition to high levels of genetic variability, we detected significant family by
host interactions indicating significant genotype by environment correlations. In
particular, log likelihood ratio tests between models with family as a random effect, host
as a fixed effect, and with or without host by family interactions indicated a significant
interaction effect for L. tulipifera and P. tremuloides (x 2 = 6.18, P = 0.046), L. tulipifera
and P. trifoliata (x 2 = 10.6, P =0.005), P. tremuloides and P. trifoliata (x 2 = 18.8, P

<0.001), and F. americana and P. trifoliata (x 2 = 11.3, P =0.004). In contrast, we found

93

 

Nd N06 06 N06 mod N:

 

00.3 00.0. 8.00 5:. $.00 ”50
$.00 3.: :.: 8.: 5.0 86
0.0002 0.30.0 0.32: 0.00.0 Sid >030 “ NE,
0 :00 0.300 0. :3 034.0 03.0.0 >200 a o>>
3 a 08 0o 0 EN 0.0 0 08 0o 0. 0.2 I H 0.00 200
a 0002
$0: 2 :0: 2 £5 E 0.5 E :5 MN 00 020.50

 

5000st» N 00.00000 N 05.0.2050 N 00303500 N 30.00.05 4

 

.000 0:585: 35... 085 Ea .058

000000.» 0002000 00 0006000000 A0>Uv 000000> 000000w 00 0000000000 w0€00000000 05 000 A00005000 0202 0005 0000300
00000000 “m A02; 000000? 0002000 000 0026 00000000, 0:000w 00 0000 0000.00 05 00 0000000000 0000000000 02215 w0000 0000—0
02.0 00 530% 00 0000 000 00000 $03 003003 .50 0000—0000 afiwa 000—: 05 8000 003.000 00000000 0000002, 00 00000005 :3» 0300.

94

no significant interaction effects between L. tulipifera and P. serotina (x 2 = 0.6, P =0.74),
L. tulipifera and F. americana (x 2 = 0.29, P =0.86), P. tremuloides and F. americana (x 2
= 3.5, P =O.17), P. tremuloides and P. serotina (x 2 = 0.92, P =0.63), F. americana and P.
serotina (x 2 = 0.9, P =0.64), and P. serotina and P. trifoliata (x 2 = 4.59, P =0.1).

Spearman correlations between family means for the Z-scores indicated
significant negative correlations between L. tulipifera and P. tremuloides (Spearman
correlation= -0.56, P=0.006) and between P. tremuloides and P. trifoliata (Spearman
correlation= -0.54, P=0.04l). No significant correlations were found for any other host
pair combination (Figure 4.4). The lack of significant correlations between L. tulipifera
and P. trifoliata and between F. americana and P. trifoliata, is an indication that the
significant interactions found above were random or weakly directional (i.e. some were
positive and some were negative). In contrast, the significant negative correlations
between L. tulipifera and P. tremuloides and between P. tremuloides and P. trifoliata
indicate that the genotype X environment interactions observed were strongly directional,
an indication of potential tradeoffs in host use ability between P. tremuloides and L.

tulipifera and P. trifoliata.

95

050 35300 .0 0% .30 2.023. .0 .8 .5 38.550 .0 .035 832.350: .0 .0 .0 E00005 .0
0.00 0000— 0003 0000000 0000 030 00 0.000 0002500 000000000 .530& .000 00.0000..N 0:000 0008 00 00000—00000 0000000m 0.0 03E

 

0.000
0.0.3.0 3.0.0.0 0.0-5. .2334 .334 3.1.0 2.0.8.0 00.0.0.0 00.0.0.0 00.013
... ._. t .0
I 0 S
c. d
a
e
J
w
B
- .o u
0 3
O
J
J
6|
8
n...
u .0 0
c. u

 

0'1

96

Discussion

Ovipositional Preference

In plant feeding insects where larvae complete development on a single host, such
as most tree feeding insects, the ovipositional behavior of the adult female determines
larval host use. For this reason shifts in ovipositional behavior are necessary for host use
divergence or specialization to occur in most tree feeding insects. Our results indicate an
identical ovipositional profile in the “late flight” hybrid swarm to that of the introgressing
parental species, P. glaucus (Figure 4.1 and Table 4.1). In particular, the “late flight”
does not accept the most common host (P. tremuloides) of the parental species (P.
canadensis) found in this region In addition, neither the preferred host, L. tulipifera, nor
the second most common host of P. glaucus, P. trifoliata, are present where the “late
flight” occurs. In fact, of those hosts on which it laid over 5% of its eggs, only F.
americana is present in the Battenkill River Valley (United States Geological Survey,
http://esp.cr.usgs.gov/data/atlas/little). Therefore a functional shift in host preference is
likely taking place in this population towards this secondary host of both parental species.
An important consideration is that as global warming proceeds tree distributions will
likely be affected. Iverson et al. (2008) modeled the potential shift in 134 tree species
under six climate scenarios, and indicated the potential for forest types (Oak-Hickory)
associated with significant L. tulipifera densities to move into the Battenkill River Valley
by 2100. However, under all the climate scenarios Iverson et al. (2008) modeled L.
tulipifera is not expected to invade this region, while F. americana is expected to

increase in abundance (http://www.nrs.fs.fed.us/atlas/tree/tree_atlas.html).

97

Previous work using three choice assays had also indicated a similar ovipositional
behavior of the “late flight” hybrid swarm relative to that of P. glaucus (Mercader and
Scriber 2007). In the same study, using a combination of seven choice, three choice, and
two choice oviposition assays, we determined that the primary difference in the
ovipositional behavior of P. glaucus and P. canadensis, is a Z-linked acceptance of P.
tremuloides as an ovipositional substrate and a reduced level of specificity in P.
canadensis. The results presented in the current study along with our previous results
indicate that the shift in ovipositional profile in this hybrid swarm, relative to the local
parental species, is due to the absence of a Z-linked trait for the acceptance of P.
tremuloides. This result is striking, because the late-flight also possesses a Z-linked trait
present in P. canadensis and not P. glaucus, obligate diapause, which is essential to
prevent a second generation incapable of completing development in the Battenkill River
Valley (Scriber and Ording 2005; Scriber et al.2008a). It is highly unlikely that this
phenotype can be attributed to natural selection, due to the short period of time this
population has existed (first observed in 1999 in a regularly sampled region). Therefore
the presence of the P. glaucus ovipositional behavior and the P. canadensis obligate
diapause trait on the same chromosome indicate that this shift in ovipositional behavior is
likely a byproduct of the recombination event leading to the delayed emergence
phenotype that forms the “late flight”.

Functional differences in ovipositional behavior due to differing environments,
rather than in the underlying behavior, are not unique to this hybrid swarm population.
Like P. canadensis, the Western species of the P. glaucus group, P. multicaudatus, P.

rutulus, and P. eurymedon will also accept L. tulipifera as an ovipositional substrate even

98

though their larvae are incapable of developing on it (Mercader and Scriber 2008b in
Press). However, this potential host shift represents the first case for which we know the
origin of the genetic variation. Several studies have demonstrated how hybridization can
lead to speciation by forming transgressive phenotypes capable of utilizing novel
environments (e. g. Riesberg 1997; Schwarz et al.2005, 2007; Mavarez et al.2006;
Gompert et al.2007). Here we demonstrate how the recombination leading to a
transgressive phenotype, delayed emergence, has also lead to a major ecological shifi

(host use) unrelated to the novel environment (thermal niche).

Larval survival and growth (between populations)

Both hybrid populations tested had high survival on three host plants tested that
were suitable to both parental species, F. americana, P. serotina, and P. trifoliata.
However, we found significantly lower survival for both of these hybrid populations
relative to the introgressing parental species, P. glaucus, on its preferred host (L.
tulipifera) relative and significantly lower survival for the “late flight” hybrid swarm
relative to the other parental species, P. canadensis, on its most common host (P.
tremuloides) (Figure 4.2). L. tulipifera and P. tremuloides are also the two hosts for
which P. glaucus and P. canadensis show strong differences in larval survival (Scriber
1996; Figure 4.2). These results demonstrate that the “late flight” population, and to a
lesser extent the Michigan hybrid zone population, maintain a mixture of larval host use
abilities from both parental species, unlike what is seen for ovipositional behavior.

In addition to differences in survival, we found significant differences in the

relative host suitability of the surviving larvae from these populations (Figure 4.3). For

99

these contrasts we accounted for overall family size differences across hosts by using Z-
scores (see Materials and Methods). This approach allowed us to contrast the relative
ability to use the five hosts assayed between two species with inherently different grth
rates and initial hatching weights and their hybrids. Therefore these results are not
contrasts of differences in absolute size between populations, but rather of the relative
suitability of each host for the families assayed from each population.

L. tulipiflera is known to have a wide variety of defensive compounds particularly
sesquiterpene lactones (Doskotch 1983; Lindroth et al. 1986; Scriber etal.1987; Barbosa
et al. 1989), which P. glaucus is capable of handling, but P. canandensis is not (Lindroth
et al.1986). The ability of P. glaucus to feed on L. tuilipifera appears to be at least partly
due to its ability to excrete unaltered sesquiterpene lactones (Frankfater et al.2005). In
this study we found the relative growth of the “late flight” on L. tulipifera was positive,
indicating it was a comparatively good host. In addition, we found no significant
difference in the relative suitability of L. tulipifera for P. glaucus and the “late flight”
population (Figure 4.3), indicating that the surviving “late flight” larvae may have
inherited from P. glaucus the ability to detoxify or excrete sesquiterpene lactones.

In contrast, for both the “late flight” population and the Michigan hybrid zone
population, performance on P. tremuloides was negative and significantly lower than for
the Northern Michigan P. canadensis population (Figure 4.3). The ability to consume P.
tremuloides by P. canadensis appears to be primarily based on the ability to detoxify
phenolic glycosides by high esterase activity (Lindroth et al.1986; Lindroth et al.1988;
Scriber etal.1989; Scriber et al. 1999). Studies using backcrosses between P. glaucus and

P. canadensis have indicated that the ability to use P. tremuloides acts in a dose

100

dependent manner, where hybrids have intermediate abilities to detoxify these
compounds (Scriber et al. 1989), as is also seen in these natural hybrid populations.

The likely host for the “late flight”, F americana, appeared to be an excellent
host for both hybrid populations and P. canadensis, but was not as good a host for P.
glaucus (Figure 4.3). This apparently low performance for P. glaucus larvae is likely due
to the excellent growth observed on P. trifoliata, reducing its Z—score relative to the other
three populations. P. glaucus is known to have a higher ability to detoxify
furanocoumarins present in Rutaceae hosts, such as P. trifoliata, than is P. canadensis
due to cytochrome P-450 enzymes (Li et al.2001, 2002, 2003; Mao et al.2007). Although
no significant difference was observed in the relative suitability of P. trifoliata between
the “late flight” and P. canadensis from Northern Michigan, it is still possible that the P.
glaucus like cytochrome P-450 enzymes are present in this population, but not in all
families (see below).

These results again indicate that the larval host use abilities of the “late flight”
population are a mixture of the host use abilities of both parental species. The
combination of detoxification abilities suggests that this population currently remains
unspecialized in its larval host use abilities. Future specialization will depend on the

genetic variability present for these traits (see below)

Estimates of genetic variance for the “late flight ” population
The ability to specialize depends on both preference for a host and performance
on that host. Ovipositional results indicated that preference in the “late flight” is towards

a functional shift onto a secondary host, F. americana. Our genetic analysis of larval

101

weight indicates that while there is a low heritability for growth on F. americana there is
a relatively high level of genetic variation for this trait (Table 4.2). The low levels of
heritability are due to the high levels of residual variance observed, which are an
indication that large within-family variation is likely.

In this study we are unable to separate additive and non-additive genetic
components due to our use of a full-sib family design. However, the large within-family
variation and the intermediate mortality within families on P. tremuloides and L.
tulipifera relative to P. canadensis and P. glaucus (Figure 4.2) are an indication of non-
additive effects segregating out in these hybrid individuals. Within classic quantitative
genetics theory non-additive effects are considered to be of little importance. However,
recent studies indicate the opposite may be true as non-additive effects have been found
to be exceedingly common and of significant importance (Roff and Emerson 2006),
including important roles in host race formation and speciation (Carroll et al.2001; Wade
2002; Rego et al. 2007).

High genetic variation was also found for P. tremuloides, P. serotina, and P.
trifoliata, but relatively low genetic variation for L. tulipifera (Table 4.2). The lower
genetic variability observed for L. tulipifera may be due to a relatively uniform
detoxification system in P. glaucus (although this is unlikely, given results in Bossart and
Scriber 1995), mortality of individuals exhibiting alternate enzymatic combinations, or a
limited introgression of detoxification abilities of P. glaucus. At this point we are
uncertain as to why we should see relatively lower levels of genetic variation for grth

on L. tulipifera compared to the ability to growth on the other four hosts (Table 4.2).

102

However, although the level of genetic variability for grth on L. tulipifera is
lower, it exists and is of a similar degree to that found for morphological characters in
Drosophila spp. (Wayne et al.1997). In fact, we see a significant genotype by host
interaction in the ability to use L. tulipifera and P. tremuloides (Results). Spearman
correlations between family Z-score means for these two hosts indicate a strong negative
correlation (Figure 4.4). The primary difference in host use that has been observed
between P. glaucus and P. canadensis is a distinct difference in their abilities to detoxify
Magnoliaceae (L. tulipifera) and Salicaceae (P. tremuloides) hosts (Scriber 1996; Scriber
et al.2003). The presence of this strong negative genetic correlation found here indicates
that a tradeoff in host use ability may have maintained the historically sharp boundary for
host use ability along the P. glaucus and P. canadensis hybrid zone. In addition, we also
found significant genotype by host interactions between P. trifoliata and L. tulipifera, F.
americana, and P. tremuloides (Results). However, there were no significant correlations
observed between P. trifoliata and L. tulipifera or F. americana (Figure 4.4), indicating a
weak (or lack of) directionality for these interactions. In contrast, there was significantly
negative correlation between P. trifoliata and P. tremuloides. The significant number of
genotype by host interactions including P. trifoliata is indicative of the presence of P.
glaucus and P. canadensis like cytochrome P450 enzymes, however this needs to be
tested.

An important consideration for the likely evolutionary trajectory of the
detoxification abilities of the “late flight” hybrid swarm population is the ovipositional
profile described above. In particular, this hybrid swarm is unlikely to oviposit on P.

tremuloides and will not encounter P. trifoliata or L. tulipifera in its current range.

103

Therefore any loss or retention of host use abilities for these hosts will largely depend

upon genetic drift.

Conclusion

Climate change has caused species ranges to shift significantly and is expected to
continue to do so (e. g. Parmesan 1999, 2006), bringing formerly separated population
ranges into closer contact. This situation has created a greater need to understand how
gene flow between locally adapted populations of the same species or closely related
species will affect the evolution of these organisms. This study illustrates how increased
hybridization due to climate change can lead to significant rapid ecological divergence
for traits that are not involved in reproductive isolation. We currently do not know
whether the “late flight” hybrid swarm population will become completely reproductively
isolated or not. However, the results presented here represent an instance where a shift in
a major ecological trait, host use, is likely occurring as a byproduct of a shift in an
unrelated trait leading to partial reproductive isolation. These changes in host use
behavior include a Z(X)-linked trait (oviposition) present in a recombined Z(X)-

chromosome and autosomally inherited non-additive larval detoxification abilities.

104

CHAPTER 5

GENERAL CONCLUSION

In this dissertation I have used the P. glaucus and P. canadensis system as a
model system to understand how climate change is likely to affect the interaction
between organisms. The results presented in Chapter 2 have indicated that while P.
canadensis is significantly impacted by higher temperature regimes, occasional very hot
summer temperatures do not appear to be the primary cause of the southern range limit of
P. canadensis. This situation implies other factors are likely contributing to the range
limit of P. canadensis, likely to include interactions with P. glaucus. In addition, the
results presented in chapter 2 indicate that the expenditure of metabolic reserves, caused
by increased temperatures, is likely to significantly reduce the fitness of P. canadensis.
Therefore climate warming will not only increase movement of P. glaucus into P.
canadensis territory, decreasing P. canadensis “mating success”, but is also likely to
decrease the survival of P. canadensis due to thermal stress.

The increased movement of P. glaucus into P. canadensis territory has already
been observed in trait clines (Scriber 2002) and the formation of the “late flight” (Scriber
and Ording 2005; Scriber et al.2008a). This situation has permitted the evaluation of how
increased genetic introgression caused by climate change may alter plant-insect
associations. As I noted in Chapter 3, the difference in ovipositional behavior of P.
canadensis and P. glaucus is caused by a Z-linked trait for the acceptance of P.
tremuloides. However, this small difference can have profound implications for the

evolutionary trajectory of the plant-insect associations in this polyphagous group. In

105

particular, the results presented in Chapter 4 indicate that recombination leading to the
“late flight” may also be leading to a host shifi onto a secondary host of both parental
species. This situation highlights our need to understand how gene flow between locally
adapted populations of the same species or closely related species will affect the
evolution of these organisms, and in general a need to understand how interactions

between organisms will be affected by climate warming.

106

APPENDIX

107

Appendix 1

Record of Deposition of Voucher Specimens”
The specimens listed on the following sheet(s) have been deposited in the named museum(s) as
samples of those species or other taxa, which were used in this research. Voucher recognition
labels bearing the Voucher No. have been attached or included in fluid-preserved specimens.
Voucher No.: 2008-06
Title of thesis or dissertation (or other research projects):
SPECIES RANGES, HOST SELECTION, AND HYBRIDIZATION: HOW INCREASED

HYBRIDIZATION IS LEADING TO HOST USE DIVERGENCE IN A POLYPHAGOUS SIBLING
SPECIES PAIR.

Museum(s) where deposited and abbreviations for table on following sheets:

Entomology Museum, Michigan State University (MSU)

Other Museums:

Investigators Name(s) (typed)
Rodrigo Jose Mercader

 

 

Date 05/01/2008

*Reference: Yoshimoto, C. M. 1978. Voucher Specimens for Entomology in North America.
Bull. Entomol. Soc. Amer. 24: 141-42.

Deposit as follows:
Original: Include as Appendix 1 in ribbon copy of thesis or dissertation.

Copies: Include as Appendix 1 in copies of thesis or dissertation.
Museum(s) files.
Research project files.

This form is available from and the Voucher No. is assigned by the Curator, Michigan State
University Entomology Museum.

108

 

 

 

80th .23

 

 

 

.8 29:80am 35. 98m 65 328mm

 

wouooom .oz bozozo>

5305.2 .32. area—com
€098 3952 95597.95
Snowmen: 2 20me .2363 $3

 

Appendix 1.1
Page _1_ of 1 Pages

Voucher Specimen Data

N

.528 ._.>0>z
>m=m> 82m. .5533 05 5 32. End
.29... em... 05 see 882.8 $350.. See
$5093 352$ Emco xoflm co poamom

.528 E>z
>m=m> 52m Ecofim 9: s 652.6%
Ema 2.69. 9: E0: 882.8 mmfiEfi E9.—
Améeg mutant Eoco xomfi :0 850m

.m_cm>_>mccmn_
.oo 868:3 E 098:8 83th Eat
AmEEEmm 2.50% Emco xom_m co ooh—mom

Emam 25»...
95nt 0.00..me Emcobmcmo 0.5me

_. w x $658ch oSme

a. 2633 20..qu

 

 

 

 

 

 

 

 

Museum
deposited
Adults 6‘
Adults £2
Pupae
Nymphs
Larvae
Eggs

where

Other

8.683 new new:
.0 0202.8 33.8on .2 Emu .33

 

:98» .050 .o 920on

 

L-

 

 

 

109

Appendix 2

Abstracts/ Summaries of closely related work on the ovipositional behaviour of the P.
glaucus group not included in this dissertation.

Mercader RJ, Scriber JM (2005) Phenotypic plasticity of host selection in adult tiger
swallowtails; Papilio glaucus (L.). In T.N. Ananthakrishnan and D. Whitman [ed.]

Insects and phenotypic plasticity. pp. 25-57. Science Publishers, Enfield. ,,,,,,,,,,,,,,,,,,,,,,, 1 ll

Mercader RJ, Stelinski LL, Scriber JM (In Press). Differential antenna] sensitivities of the
generalist butterflies Papilio glaucus and Papilio canadensis to host plant and non-host
Plant extractS- Journal of LepidopteristS’ Society. .............................................................. 113

Mercader RJ, Scriber JM (In Press) Divergence of ovipositional behavior in the Papilio
glaucus gIOUP- Insect Science- ......................................................... _ 114

110

Appendix 2.1
Summary for

Mercader RJ, Scriber JM (2005) Phenotypic plasticity of host selection in adult tiger
swallowtails; Papilio glaucus (L.). In T.N. Ananthakrishnan and D. Whitman [ed.] Insects
and phenotypic plasticity. pp. 25-57. Science Publishers, Enfield.

In summary, oviposition preferences of insects are determined by a large array of external
and internal stimuli. Natural selection has acted on the genetic basis of oviposition ranking
behavior, but phenotypic plasticity also plays a very important functional role. Behavioral
preference induction in adults and larvae, as well as the effects of fecundity, age, and other
factors may modify the ‘acceptability threshold’ for different hosts and determine how far down a
ranking-order the insect will go (i.e. its ‘specificity’).

The eastern tiger swallowtail butterfly (P. glaucus) is the most polyphagous of all 560
species of Papilionidae (Scriber 1984a; 1995). Here we show that the ranking order of oviposition
preference for this generalist appears basically constant in seven family host choice arrays for
populations from the northernmost to southernmost parts of its geographic range (Fig 6). Local
‘phenotypic differentiation’ (as opposed to local ‘adaptation’) may be accomplished largely by
phenotypically flexible (plastic) ‘specificity’ in host selection by individual females, thereby
alleviating the directional selection for rank-order changes that might be expected in various local
populations. The lack of local adaptations for changing rank-order of preference is not due to a
lack of genetic variation in oviposition behavior; it is clear that individual females vary in their
placement, with 100% of their eggs to less than 10% of their eggs on (the putatively ancestral)
tulip tree favorite (Fig. 7). ‘Plasticity’ appears to dampen (lessen the need for) directional
selection (that might take a ranking-order change to fixation). Ovipositional flexibility for local
hosts or ecological conditions is thus (theoretically) possible without need for extensive gene

flow because of the specificity adjustments in this Papilio species. In contrast, the avoidance of

111

aspen (both ranking order and specificity) may be genetically and behaviorally invariant, with
almost every individual female placing few, if any eggs on this species (Fig 7).

The benefits of being polyphagous should exceed those of being specialized. The
behavioral efficiency of locating the best host is generally less for generalists (perhaps due to
neural limitations; Bemays 2001). Plasticity in behavior would be useful (e.g. where choices of
hosts vary in availability in non-predictable fashion (as seen in generalists). Such non-genetic
(inducible) responses may have permitted an ancient generalist (P. glaucus) to persist in
ecologically and phytochemically diverse habitats trather than evolving numerous local host-
specializations which would seem clearly possible, given the high levels of genetic variability for
oviposition preferences that persists across the range of this generalist species (Bossart 2003,
Bossart and Scriber 1995, 1999,; Fig. 7). Clearly, the complete ecology of the species needs to be
addressed in order to understand the factors preserving this persistent polyphagy.

In addition to host plant and enemy-related factors, genetically-based behavioral
oviposition divergence in time and location have been documented due to abiotic factors such as
thermal selection as shown in highland/lowland Drosophila (Dahlgaard et 01.2001) and in
climatic ‘cold pockets’ for Papilio canadensis (Scriber 2002b). As seen here with P. glaucus, it
may be that plasticity in specificity allows different genetically-determined rank orders among
related Papilio species (Scriber et al. 2003), but usually not within Papilio species. For example,
heritable variation has been shown to exist for P. glaucus toward less preferred hosts rather than
ranking order (Bossart and Scriber 1999). Both ranking and specificity of butterfly oviposition
may be largely X-linked and genetically-controlled (Janz 2003; Singer 2003). However, the role
of phenotypic plasticity in determining oviposition variation observed between individuals,
populations, and congeneric species will continue to be an important prerequisite for our
understanding of host plant (and adult preference/larval performance correlations; Mayhew 2001;

Bossart 2003) relationships in phtyophagous insects.

112

Appendix 2.2
Abstract for

Mercader RJ, Stelinski LL, Scriber JM (In Press). Differential antennal sensitivities of the
generalist butterflies Papilio glaucus and Papilio canadensis to host plant and non-host
plant extracts. Journal of Lepidopterists’ Society

It is likely that olfaction is used by some generalist insect species as a pre-
alighting cue to ameliorate the costs of foraging for suitable hosts. In which case,
significantly higher antennal sensitivity would be expected to the volatiles of preferred
over less or un-preferred host plants. To test this hypothesis, antennal sensitivity was
measured by recording electroantennogram (EAG) responses from intact antennae of the
generalists Papilio glaucus L. and P. canadensis R & J (Lepidoptera: Papilionidae) to
methanolic leaf extracts of primary, secondary, and non-host plants. BAGS recorded from
antennae of P. glaucus were approximately four fold higher than those of P. canadensis
in response to extracts of its most suitable host plant, Liriodendron tulipifera
(Magnoliaceae). Likewise, BAG responses of P. canadensis to its preferred host, Populus
tremuloides (Salicaceae), were significantly higher than those of P. glaucus. In addition,
P. glaucus exhibited significantly higher (approximately three fold) EAG responses to its
preferred host, L. tulipifera, than to its less-preferred hosts, Ptelea trifoliata, Sassafi'as
albidum, and Lindera benzoin. The results from this study indicate a significant
divergence in the olfactory system of two closely related generalist butterfly species,

including a strong specialization in the olfactory system of P. glaucus.

113

Appendix 2.3

Abstract for
Mercader RJ, Scriber JM (In Press) Divergence of ovipositional behavior in the Papilio
glaucus group. Insect Science

This study contrasts the ovipositional profiles of 4 members of the P. glaucus

group, P. glaucus, P. multicaudatus, P. canadensis, and P. rutulus. We use seven choice
oviposition bioassays containing leaves from hosts in seven plant families utilized by
members of the P. glaucus group. Specifically, we contrast the overall ovipositional
profiles of these species and their acceptance of a host in a novel plant family (Populus
tremuloides: Salicaceae) and a host in a putatively ancestral host plant family
(Liriodendron tulipifera: Magnoliaceae). Significant differences were observed between
the ovipositional profiles of P. glaucus and P. multicaudatus relative to each other and to
P. canadensis and P. rutulus. In contrast, no significant differences were observed
between the ovipositional profiles of P. canadensis and P. rutulus, which were also the
only species that accepted P. tremuloides. Unlike the acceptance of P. tremuloides, the
acceptance of L. tulipifera was present throughout the group despite the inability of the
larvae of most species in the group to utilize this host. These results support the
prediction of the “hierarchical threshold model” (Courtney et al., 1989) that ancestral
host plants are likely to be retained in the ovipositional hierarchy while novel hosts

should only be accepted by derived populations.

114

REFERENCES CITED

Abrahamson WG, Eubanks MD, Blair CP, Whipple AV (2001) Gall flies, inquilines, and
goldenrods: A model for host-race formation and sympatric speciation. American
Zoologist 41: 928-938.

Anderson MJ (2001) A new method for non-parametric multivariate analysis of variance.
Austral Ecology 26: 32-46.

Angelo MJ, Slansky F (1984) Body building by insects- Trade-offs in resource allocation
with particular reference to migratory species. Florida Entomologist 67: 22-41.

Arnold ML (1997) Natural Hybridization and Evolution. Oxford Series in Ecology and
Evolution, Oxford University Press, Oxford, UK.

Ayres MP, Scriber J M (1994) Local adaptation to regional climates in Papilio canadensis
(Lepidoptera:Papilionidae). Ecological Monographs 64:465-482.

Barbosa P, Gross P, Provan GJ, Pacheco DY, Sterrnitz FR (1990) Allelochemicals in
foliage of unfavored tree hosts of the gypsy month Lymantria dispar L .1.
Alkaloids and other components of Liriodendron tulipifera L. (Magnoliaceae).
Acer rubum (Aceraceae), and Cornusflorida (Cornaceae). Journal of Chemical

Ecology 16: 1719-1730.

Bates D, Sarkar D (2007) lme4: Linear mixed-effects models using S4 classes. Available
online at: http://www.r-project.org.

Battisti A, Stastny M, Netherer S, Robinet C, Schopf A, Roques A, Larsson S (2005)
Expansion of geographic range in the pine processionary moth caused by
increased winter temperatures. Ecological Applications 15: 2084-2096.

Berenbaum MR (1981) An oviposition “mistake” by Papilio glaucus (Papilionidae).
Journal of the Lepidopterists’ Society 35: 75.

Boggs CL, Freeman KD (2005) Larval food limitation in butterflies: effects on adult
resource allocation and fitness. Oecologia 144: 353—361

Bossart J L, Scriber J M (1995) Maintenance of ecologically significant genetic variation
in the tiger swallowtail butterfly through differential selection and gene flow.
Evolution 49: 1163-1171.

Bowers MD, Stamp NE, Collinge SK (1992) Early stage of host range expansion by a
specialist herbivore, Euphydras phaeton (N ymphalidae). Ecology 73: 526-536.

115

Braby MF, Trueman J WH (2006) Evolution of larval host plant associations and adaptive
radiation in pierid butterflies. Journal of Evolutionary Biology 19: 1677-1690.

Carroll SP, Boyd C (1992) Host race radiation in the soapberry bug: Natural history with
the history. Evolution 46: 1052-1069.

Carroll SP, Dingle H, Famula TR, Fox CW (2001) Genetic architecture of adaptive
differentiation in evolving host races of the soapberry bug, Jadera haematoloma.
Genetica 112-113: 257—272.

Carroll SP (2007) Natives adapting to invasive species: Ecology, genes, and the
sustainability of conservation. Ecological Research 22: 892-901.

Chew FS (1977) Coevolution of pierid butterflies and their cruciferous foodplants. II. The
distribution of eggs on potential foodplants. Evolution 31: 568-579.

Clark RT, Brown SJ, Murphy JM (2006) Modeling Northern Hemisphere summer heat
extreme changes and their uncertainties using a physics ensemble of climate

sensitivity experiments. Journal of Climate 19: 4418-4435.

Courtney SP, Chen GK, Gardner A (1989) A general-model for individual host selection.
Oikos 55: 55-65.

Coyne JA, Orr HA (2004) Speciation. Sinauer Associates, Sunderland, MA.

Crozier L (2004) Warmer winters drive butterfly range expansion by increasing
survivorship. Ecology 85: 231-241 .

Deering MD, Scriber JM (2002) Field bioassays show heterospecific mating preference
asymmetry between hybridizing North American Papilio butterfly species
(Lepidoptera: Papilionidae). Journal of Ethology 20: 25-33.

Doskotch RW, Wilton JH, Harraz FM, Fairchild EH, Huang CT, Elferaly FS (1983) 6
additional sesquiterpene lactones from Liriodendron tulipifera. Journal of Natural
Products 46: 923-929.

Easterling DR, Meehl GA, Parmesan C, Changnon SA, Karl TR, Mearns L0 (2000)
Climate Extremes: Observations, Modeling, and Impacts. Science 289: 2068-
2074.

Ehrlich PR, Raven PH (1964) Butterflies and plants: A study in coevolution. Evolution
18: 586-608.

F eder JL, Berlocher SH, Opp SB (1998) Sympatric host race formation and speciation in
Rhagoletis (Diptera: Tephritidae): A tale of two species for Charles D. In: S

116

Mopper, Strauss SY (eds) Genetic Structure and Local Adaptation in Natural
Insect Populations, Chapman and Hall, New York, NY, USA, pp 408-441.

Fox J (2006). car: Companion to Applied Regression. R package version 1.1-2.
http://www.r-project.org, http://socserv.socsci.mcmaster.ca/ifox/.

Frankfater C, Schtihly W, Fronczek FR, Slattery M (2005) Processing of a sesquiterpene
lactone by Papilio glaucus caterpillars. Journal of Chemical Ecology 31 :2541-
2550.

F utuyma DJ (1989) Macroevolutionary consequences of speciation: inferences fiom
phytophagous insects. In: Otte D, Endler JA (eds) Speciation and Its
Consequences. Sinauer Associates, Sunderland, Massachusetts, USA, pp 557—
578.

Futuyma DJ, RP Cort RP, van Noordwijk I (1984) Adaptation to host plants in the Fall
Cankerworm (Alsophila pometaria) and its bearing on the evolution of host
affiliation in phytophagous insects. American Naturalist 123: 287-296.

Gassmann AJ, Levy A, Tran T & F utuyma DJ (2006) Adaptations of an insect to a novel
host plant: a phylogenetic approach. Functional Ecology 20: 478-485.

Gelman A, Su YS, Yajima M, Hill J, Pittau MG, Kerrnan J, ZhenT (2007) arm: Data
Analysis Using Regression and Multilevel/Hierarchical Models. Available online
at: http://www.r-proiect.og.

Gompert Z, F ordyce JA, F orister ML, Shapiro AM, Nice CC (2006) Homoploid hybrid
speciation in an extreme habitat. Science 314: 1923-1925.

Gordon J. (1999) Quantitative genetics of intraspecies hybrids. Heredity 83: 757-764.

Gordon IL (2000) Quantitative genetics of allogamous F2: an origin of randomly
fertilized popualtions. Heredity 85: 43—52.

Graves SD, Shapiro AM (2003) Exotics as host plants of the California butterfly fauna.
Biological Conservation 1 10: 413-4333.

Gillooly JF, Brown J H, West GB, Savage VM, Charnov EL (2001) Effects of size and
temperature on metabolic rate. Science 293: 2248-2251.

Hammer 0, Harper DAT, Ryan PD (2001) PAST: Paleontological Statistics Software
Package for Education and Data Analysis. Palaeontologia Electronica 4(1): 9pp.

Hahn DA, Denlinger DL (2007) Meeting the energetic demands of insect diapause:
Nutrient storage and utilization. Journal of Insect Physiology 53: 760-773.

117

Hora KH, Roessingh P, Menken SB] (2005) Inheritance and plasticity of adult host
acceptance in Yponomeuta species: implications for host shifts in specialist
herbivores. Entomologia Experimentalis et Applicata 115: 271-281.

Hothom T, Bretz F, Westfall P, Heiberger RM (2007). multcomp: Simultaneous
inference for general linear hypotheses. Available online at: http://www.r-

projectorg.

Houle D (1992) Comparing evolvability and variability of quantitative traits. Genetics
130: 195-204.

Hoyle M, James M (2005) Global warming, human population pressure, and viability of
the world's smallest butterfly. Conservation Biology 19: 1113-1124.

Iverson LR, Prasad AM, Matthews SN, Peters M (2008) Estimating potential habitat for
134 eastern US tree species under six climate scenarios. Forest Ecology and
Management 254: 390-406.

J annot JE, Bruneau E, Wissinger SA (2007) Effects of larval energetic resources on life
history and adult allocation patterns in a caddisfly (Trichoptera: Phryganeidae)
Ecological Entomology 32: 376—383.

Janz N, Nylin S, Wedell N (1994) Host plant utilization in the comma butterfly: sources
of variation and evolutionary implications. Oecologia 99: 132-140.

J anz N (1998) Sex-linked inheritance of host-plant specialization in a polyphagous
butterfly. Proceedings of the Royal Society of London B Biological Sciences.
265: 1675-1678.

J anz N, Nyblom K, Nylin S (2001) Evolutionary dynamics of host plant specialization: a
case study of the tribe Nymphalini. Evolution 55: 783-796.

J anz N (2003) Sex-linkage of host plant use in butterflies. Butterflies: Ecology and
Evolution Taking Flight. In: Boggs CL, Watt WB, EhrlichPR (eds), University of
Chicago Press, Chicago, IL, USA, pp 229-239.

Janz N, Nylin S & Wahlberg N (2006) Diversity begets diversity: host expansions and
the diversification of plant-feeding insects. BMC Evolutionary Biology 6: 4.

J anz N, Nylin S (2008) The oscillation hypothesis of host-plant range and speciation. In:
Tilmon KJ (ed) Specialization, Speciation, and Radiation: The Evolutionary
Biology of Herbivorous Insects. University of California Press, Berkley , CA, pp
203-215.

J ermy T (1984) Evolution of insect host plant relationships. American Naturalist
124:609-630.

118

Jenny T, Szentsi R (2003) Evolutionary aspects of host plant specialisation - a study on
bruchids (Coleoptera : Bruchidae). Oikos 101:196—204.

Larsson S, Ekbom B (1995) Oviposition mistakes in herbivorous insects: confusion or a
step towards a new host plant? Oikos 72: 155-160.

Levins R, MacArthur R (1969) An hypothesis to explain the incidence of monophagy.
Ecology 50: 910-911.

Li W, Berenbaum MR, Schuler MA (2001) Molecular analysis of multiple CYP6B genes
from polyphagous Papilio species. Insect Molecular Biology 31: 999-110.

Li W, Petersen RA, Schuler MA, Berenbaum MR (2002) CYP6B cytochrome P450
monooxygenases from Papilio canadensis and Papilio glaucus, potential

contributions of sequence divergence to host plant associations. Insect Molecular
Biology 11: 543—551.

Li W, Berenbaum MR, Schuler MA (2003) Diversification of furanocoumarin-
metabolizing cytochrome P450s in two papilionids: specificity and substrate

encounter rate. Proceedings of the National Academy of Science USA 100:
14593—14598.

Lindroth RL, Scriber JM, Hsia MTS (1986) Differential responses of tiger swallowtail
subspecies to secondary metabolites from tulip tree and quaking aspen. Oecologia
70: 13-19.

Lindroth RL, Scriber J M, Hsia MTS (1988) Chemical ecology of the tiger swallowtail:
Mediation of host use by phenolic glycosides. Ecology 69: 814-822.

Lopez-Vaamonde C, Wikstrom N, Labandeira C, Godfray HCJ, Goodman SJ, Cook JM
(2006) Fossil-calibrated molecular phylogenies reveal that leaf-mining moths

radiated millions of years after their host plants. Journal of Evolutionary Biology
19: 1314-1326.

Mallet J (2007) Hybrid speciation. Nature 446: 279-283.

Mao W, Schuler MA, Berenbaum MR (2007) Cytochrome P4503 in Papilio
multicaudatus and the transition from oligophagy to polyphagy in the
Papilionidae. Insect Molecular Biology 16: 481-490.

Mavarez J, Salazar CA, Bermingham E, Salcedo C, Jiggins CD, Linares M (2006)
Speciation by hybridization in Heliconius butterflies. Nature 441: 868-871.

119

Mayhew JP (1997) Adaptive patterns of host-plant selection by phytophagous insects.
Oikos 79: 417-428.

McLaughlin JF, Hellmann JJ, Boggs CL, Ehrlich PR (2002) Climate change hastens
population extinctions. Proceedings of the National Academy of Science USA 99:
6070-6074.

Menken SBJ (1996) Pattern and process in the evolution of insect-plant associations:
Yponomeuta as an example. Entomologia Experimentalis et Applicata 80: 297-
305.

Mercader RJ & Scriber J M (2005) Phenotypic plasticity of host selection in adult tiger
swallowtails; Papilio glaucus (L.). In: Ananthakrishnan TN, Whitman D (eds)
Insects phenotypic plasticity: Diversity of responses, Science Publishers, Enfield,
NH, USA, pp 25-57.

Mercader RJ, Scriber JM (2007) Diversification of host use in two polyphagous
butterflies: Differences in oviposition specificity or host rank hierarchy?
Entomologia Experimentalis et Applicata 125: 89-101.

Mercader RJ, Scriber J M (2008a) Asymmetrical thermal constraints on the parapatric
species boundaries of two widespread generalist butterflies. Ecological
Entomology (In Press).

Mercader RJ, Scriber J M (2008b) Divergence of ovipositional behavior in the Papilio
glaucus group. Insect Science (In Press).

Mitter C, Farrel B, Wiegmann B (1988) The phylogenetic study of adaptive zones: has
phytophagy promoted insect diversification? American Naturalist 132:107—128.

Murphy SM & Feeny P (2006) Chemical facilitation of a naturally occurring host shift by
Papilio machaon butterflies (Papilionidae). Ecological Monographs 76: 399-414.

Nosil P (2002) Transition rates between specialization and generalization in
phytophagous insects. Evolution 56: 1701-1706.

Nygren GH, Nylin S, Stefanescu C (2006) Genetics of host plant use and life history in
the comma butterfly across Europe: varying modes of inheritance as a potential
reproductive barrier. Journal of Evolutionary Biology 19: 1882-1893.

Ohshima I, Yoshizawa K (2006) Multiple host shifts between distantly related plants,
Juglandaceae and Ericaceae, in the leaf-mining moth Acrocercops leucophaea
complex (Lepidoptera : Gracillariidae). Molecular Phylogenetics and Evolution

38: 231-240.

120

Palumbi SR (2001) The evolution explosion: How humans cause rapid evolutionary
change. W.W. Norton, New York, NY, USA.

Pavulaan H, Wright DM (2002) Pterourus appalachiensis (Papilionidae: Papilioninae), a
new swallowtail butterfly from the Appalachian region of the United States.
Taxonomic Report 3: 1—20.

Parmesan C, Ryrholm N, Stefanescu C, Hill JK, Thomas CD, Descimon H, Huntley B,
Kaila L, Kullberg J, Tammaru T, Tennent J, Thomas J A, Warren M (1999)
Poleward shift of butterfly species’ ranges associated with regional warming.
Nature 399: 579-583.

Percy DM, Page RDM, Cronk QCB (2004) Plant-insect interactions: double-dating
associated insect and plant lineages reveals asynchronous radiations. Systematic
Biology 53: 120-127.

Peres-Neto PR, Jackson DA, Somers KM (2003) Giving meaningful interpretation to
ordination axes: Loading significance in Principal Component Analysis. Ecology
84: 2347—2363.

Peters RH (1983) The ecological implications of body size. Cambridge University Press,
Cambridge, UK.

Price PW (2002) Resource-driven terrestrial interaction webs. Ecological Research 17:
241—247.

Price PW (2008) Adaptive radiation: Phylogenetic constraints and ecological
consenquences. In: Tilmon KJ (ed) Specialization, Speciation, and Radiation: The
Evolutionary Biology of Herbivorous Insects. University of California Press,
Berkley , CA., pp 174-187.

R Development Core Team (2006) R: A Language and Environment for Statistical
Computing. Royal Foundation for Statistical Computing, Vienna, Austria. ISBN
3-90005 1-07-0, http://www.R-project.org.

 

Radtkey RR, Singer MC (1995) Repeated reversals of host-preference evolution in a
specialist insect herbivore. Evolution 49: 351-359.

Renwick JAA (2002) The chemical world of crucivores: lures, treats, and traps.
Entomologia Experimentalis et Applicata 104: 35-42.

Rieseberg LH (1997) Hybrid origins of plant species. Annual Review of Ecology and
Systematics 28: 359-389.

121

Rockey SJ, Hainze JH, Scriber J M (1987) Evidence of a sex-linked diapause response in
Papilio glaucus subspecies and their hybrids. Physiological Entomology 12: 181—
184.

Rego C, Santos M, Matos M (2007) Quantitative genetics of speciation: additive and
non-additive genetic differentiation between Drosophila madeirensis and
Drosophila subobscura. Genetica 131: 167-174.

Schwartz MD, Ahas R, Aasa A (2006) Onset of spring starting earlier across the Northern
Hemisphere. Global Change Biology 12: 343-351.

Schwarz D, Shoemaker KD, Botteri NL, McPheron BA (2007) A novel preference for an
invasive plant as a mechanism for animal hybrid speciation. Evolution 61: 245-
256.

Scriber J M (1984) Larval foodplant utilization by the world Papilionidae (Lepidoptera):
Latitudinal gradients reappraised. Tokurana (Acta Rhopalocerologica) 2: 1-50.

Scriber, JM (1988) Tale of the tiger: Beringial biogeography, binomial classification, and
breakfast choices in the Papilio glaucus complex of butterflies. In: Spencer KC
(ed) Chemical Mediation of Coevolution, Academic Press, New York, NY, USA,
pp 240-301.

Scriber JM, Hsia MTS, Sunarjo P, Lindroth RL (1987) Allelochemicals as determinants
of insect damage across the North American continent. In: Waller GR (ed)
Allelochemicals: Role in agriculture and forestry. American Chemical Society,
Washington DC, USA, pp 439-448.

Scriber JM, Lindroth RL, Nitao J (1989) Differential toxicity of a phenolic glycoside
from quaking aspen to Papilio glaucus butterfly subspecies, hybrids, and
backcrosses. Oecologia 81: 186-191.

Scriber JM, Giebink BL & Snider D (1991) Reciprocal latitudinal clines in oviposition
behavior of Papilio glaucus and P. canadensis across the Great Lakes hybrid
zone: possible sex-linkage of oviposition preferences. Oecologia 87: 360-368.

Scriber J M (1993) Absence of behavioral induction in oviposition preference of Papilio
glaucus (Lepidoptera: Papilionidae). Great Lakes Entomologist 26: 81-95.

Scriber J M (1994) Climatic legacies and sex chromosomes: latitudinal patterns of
voltinism, diapause size and host-plant selection in 2 species of swallowtail
butterflies at their hybrid zone. In: Danks HV (ed) Insect life-cycle
polymorphism: theory, evolution and ecological consequences for seasonality and
diapause control, Kluwer Academic Publishers, Dordrecht, Netherlands, pp 133-
171.

122

 

Scriber JM, Tsubaki Y, Lederhouse RC (1995) Swallowtail butterflies: Their ecology &
evolutionary biology. Scientific Publishers, Gainesville, Fl USA.

Scriber JM, Lederhouse RC, Dowell RV (1995) Hybridization studies with North
American Swallowtails. In: Scriber JM, Tsubaki Y, Lederhouse RC (eds)

Swallowtail butterflies: Their ecology and evolutionary biology. Scientific
Publishers, Gainesville, USA, pp 269-281.

Scriber JM (1996a) Tiger tales: Natural history of native North American swallowtails.
American Entomologist 42:19-32.

Scriber JM (1996b) A new cold pocket hypothesis to explain local host preference shifts
in Papilio canadensis. 9th International Symposium on Insects and Host Plants.
Entomologia Experimentalis et Applicata 80: 315-319.

Scriber JM, Weir K, Parry D, Deering J (1999) Using hybrid and backcross larvae of
Papilio canadensis and Papilio glaucus to detect induced phytochemical
resistance in hybrid poplar trees experimentally defoliated by gypsy moths.
Entomologia Experimentalis et Applicata 91: 233-236.

Scriber J M (2002) The evolution of insect-plant relationships; Chemical constraints,
coadaptation and concordance of insect/plant traits. Entomologia experimentalis
et applicata 104: 217-235.

Scriber J M (2002a) Latitudinal and local geographic mosaics in host plant preferences as
shaped by thermal units and voltinism in Papilio spp. (Lepidoptera). European
Journal of Entomology 99: 225-239.

Scriber JM (2002b) Evolution of insect-plant relationships: chemical constraints,
coadaptation, and concordance of insect/plant traits. Entomologia Experimentalis
et Applicata 104: 217-235.

Scriber J M, Keefover K, Nelson S (2002) Hot summer temperatures may stop movement
of Papilio canadensis butterflies and genetic introgression south of the hybrid
zone in the North American Great Lakes Region. Ecography 25: 184-192.

Scriber JM, Stump A, Deering M (2003) Hybrid zone ecology and tiger swallowtail trait
clines in North America. In: Boggs CL, Watt W, Ehrlich PR (eds) Ecology and
evolution taking flight: Butterflies as model study systems, University of Chicago
Press, Chicago, USA, pp 367-391.

Scriber J M, Ording GJ (2005) Ecological speciation without host plant specialization;

possible origins of a recently described cryptic Papilio species. Entomologia
Experimentalis et Applicata 115: 247-263.

123

 

Scriber J M, Ording GJ, Mercader RJ (2008a) Hybrid introgression and parapatric
speciation in a hybrid zone. In: Tilmon K (ed) Specialization, Speciation, and
Radiation: The Evolutionary Biology of Herbivorous Insects, University of
California Press, Berkley, CA, USA, pp 69-87.

Scriber JM, Larsen ML, Allen GR, Walker PW, Zalucki MP (2008b) Interactions
between Papilionidae and ancient Australian angiosperms: evolutionary
specialization or ecological monophagy? Entomologia experimentalis et applicata
(In Press)

Seehausen O (2004) Hybridization and adaptive radiation. Trends in Ecology and
Evolution 19:198—207.

Shoo LP, Williams SE, Hero JM (2005) Potential decoupling of trends in distribution
area and population size of species with climate change. Global Change Biology
11: 1469-1476.

Singer MC, Wee B, Hawkins S, Butcher M (2008) Rapid narural and anthropogenic diet
evolution: Three examples from checkerspot butterflies. In: Tilmon KJ (ed)
Specialization, Speciation, and Radiation: The Evolutionary Biology of
Herbivorous Insects. University of California Press, Berkley, CA,USA, pp 311-
324.

Stastny M, Battisti A, Petrucco-Toffolo E, Schlyter F & Larsson S (2006) Host-plant use
in the range expansion of the pine processionary moth, Thaumetopoea
pityocampa. Ecological Entomology 31: 481-490.

Straatmaan R (1962) Notes on certain Lepidoptera ovipositing on plants which are toxic
to their larvae. Journal of the Lepidopterist’s Society 16: 99-103.

Strauss SY, Lau JA, Carroll SP (2006) Evolutionary responses of natives to introduced
species: what do introductions tell us about natural communities? Ecology Letters
92354-371.

Tauber CA, Tauber MJ (1981) Insect seasonal cycles: Genetics and evolution. Annual
Review of Ecology and Systematics 12: 281-308.

Thomas CD, Bodsworth EJ, Wilson RJ, Simmons AD, Davies ZG, Musche M, Conradt L
(2001) Ecological and evolutionary processes at expanding range margins. Nature
411: 577-581.

Thompson JN (1988) Variation in preference and specificity in monophagous and
oligophagous swallowtail butterflies. Evolution 42: 118-128.

Thompson JN, Pellmyr O (1991) Evolutionary genetics of oviposition behavior and host
preferences in Lepidoptera. Annual Review of Entomology 36: 65-69.

124

 

Thompson JN (1993) Preference hierarchies and the origin of geographic specialization
in host use in swallowtail butterflies. Evolution 47: 1585-1594.

Thompson JN (1994) The Coevolutionary Process. University of Chicago Press, Chicago,
IL, USA.

Thompson JN (1998) The evolution of diet breadth: Monophagy and polyphagy in
swallowtail butterflies. Journal of Evolutionary Biology 11: 563-578.

Thompson JN (1998b) Rapid evolution as an ecological process. Trends in Ecology and
Evolution 13: 329-332. '

Wade MJ (2002) A gene’s eye view of epistasis, selection and speciation. Journal of
Evolutionary Biology 15: 337—346.

Walther, GR, Post E, Convey P, Menzel A, Parmesan C, Beebee TJC, Fromentin JM,
Hoegh-Guldberg O, Bairlein, F. (2002) Ecological responses to recent climate
change. Nature 416: 389-395.

Wasserrnan SS, F utuyma DJ (1981) Evolution of host plant utilization in laboratory
populations of the southern cowpea weevil; Callosobruchus maculates F abricius
(Coleoptera, Bruchidae) Evolution 35: 605-617.

Wayne ML, Hackett JB, Mackay TF C (1997) Quantitative genetics of ovariole number in
Drosophila melanogaster. l. Segregating variation and fitness. Evolution 51:
1156-1163.

Weiblen GD, Webb CO, Novotny V, Basset Y, Miller SE (2006) Phylogenetic dispersion
of host use in a tropical insect herbivore community. Ecology 87: S62-S75.

Weingartner E, Wahlberg N & Nylin S (2006) Dynamics of host plant use and species
diversity: a phylogenetic investigation in Polygonia butterflies (N ymphalidae).
Journal of Evolutionary Biology 19: 483-491.

Weisser WW, Siemann E (2004) Insects and ecosystem function. Springer-Verlag,
Berlin, Germany.

Wheat CW, Vogel H, Wittstock U, Braby MF, Underwood D, Mitchell-Olds T (2007)
The genetic basis of a plant-insect coevolutionary key innovation. Proceedings of
the National Academy of Science USA 104: 20427-20431.

Wiklund C (1975) The evolutionary relationship between adult oviposition preferences
and larval host plant range in Papilio machaon L. Oecologia 18: 185-197.

125

 

Zakharov EV, Caterino MS & Sperling FA (2004) Molecular phylogeny, historical
biogeography, and divergence time estimates for swallowtail butterflies of the
genus Papilio (Lepidoptera: Papilionidae). Systematic Biology 53: 193—215.

126

llulllillillljljllljllii