3+3} 1.31:}

am§¢-;§.L.... 1.

auteur... WE. L.
I‘m“

V ,1. fi
I._IL n4. . I.‘
1 :1: i

'-. ; 4 -I
u: .3. ‘4

 

J;.‘:‘t “4%
115‘

I
Aunt; ~39?"
if“; "in ‘5: Wigw'fi'f-U 3

‘ J .Lrwuv ’u.‘
”4V J—..\.--~r
M u. u
413'
n —.
"-‘ IL
“I, +
v\‘
.5: “v;
1,- ‘1:

.«ux 1. .

.5...

u

.
.urn'fiif; '5‘!»
I.

Ir l ~ « [5‘14
wit 0 L A

UfquJ' v

:u»...l..

u
'u 1;)"

«"415;
“nor-'1

. . W ' " - .ur
, - :4 W . 4r, . . . m K ‘

. P 3. 50"“ -1"‘?‘('. ’ .L. wfimfi
{it tars? @3944 . 50%;? lag, ”,5 , if '~ ”SEE-.1” "3'? 3"“ v“ .w j 3:315:
Ll" ' ._ c; - v ‘ 1‘ _—... a“.

. ;rr; fifggg‘ <35.“ «4 J r“ “4:? fl .. ”in... :11 -:€?.x£‘ 3
5:23 ’, .

infra.
”Why.
«,1th -

x was",

“’IV‘
u",:

x r r
.,...- .‘.5’.m;-,_"3. n?
m \x-

fi'l‘.
-": n.3,?

v “37-“: “’5
‘0’

" 53‘,‘ u
{ :st.'.}'.v.'.r"" I 1;
H ..L

.r)‘
"- F'nur': n
.u m..,,.,.
.4. w

'1.“
. ut'v-

 

€."."'9§"1'.'. s .q
u' .1

mam-'1
3‘“. It
4-" 1441mm

1“ I
'I'MI'" '5'; .4."

‘L.,.1,.--':,f£$'nub-11,151,, I ”‘5'!

[,l/(i,..,\'" :2::,'.’,-’

4f

 

(’“5'
f, Elfin." i

dESlS

                                                                                                                               

300896 3294

This is to certify that the

dissertation entitled

ADAPTATION AND CONSTRAINT IN PAPILIO CANADENSIS:
GEOGRAPHIC VARIATION IN NUTRITIONAL PHYSIOLOGY
AND TEMPERATURE RESPONSES

 

presented by

MATTHEW P. AYRES

has been accepted towards fulfillment
of the requirements for

PH . D . degree in ENTOMOLOGY

 

 

Major professor

Date 15 NOVEMBER 1991

 

MSU is an Affirmative Action/Eq ual Opportunity Institution 0-12771

 

ADAPTATION AND CONSTRAINT IN PAPILIQ CANADENSIS:
GEOGRAPHIC VARIATION IN NUTRITIONAL PHYSIOLOGY
AND TEMPERATURE RESPONSES

By

Matthew P. Ayres

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 and Evolutionary Biology

1991

 

477» 7632

 

ABSTRACT

ADAPTATION ANI) CONSTRAINT IN PAPILIQ gANADENSIs:
GEOGRAPHIC VARIATION IN NUTRITIONAL PHYSIOLOGY
AND TEMPERATURE RESPONSES
By

Matthew P. Ayres

Papilio canadensis, the Canadian tiger swallowtail butterfly, is a
polyphagous herbivore common throughout the boreal forests of North America.
Populations inhabiting Michigan feed on different tree species, under a different
climatic regime, than those inhabiting interior Alaska. With respect to its
host-use abilities and temperature responses, P. canadensis could be a mosaic of
relatively specialized genotypes, each of which can only succeed in a subset of
environments exploited by the species as a whole, or, it could be single cohesive
genotype that is unusually robust. I evaluated these hypotheses by comparing the
physiological ecology of P. canadensis populations from Alaska and Michigan on
the host species normally encountered by each, and at the temperatures normally
encountered by each.

The ability to consume, detoxify, assimilate, and convert chemically diverse
plant tissue is highly conserved within P. canadensis. There was no differentiation
of host-use abilities between Michigan and Alaskan populations, although there is
little overlap between their host communities. Michigan caterpillars grew as well
as Alaskan caterpillars on Alaskan hosts, and no better than Alaskan caterpillars

on Michigan hosts. There was also no evidence of host specialists within

 

populations. Nonetheless, there were pronounced differences in host-use abilities
between P. canadensis and its presumed progenitor (P. glaucus).

In contrast to host plant use, marked population divergence has occurred
in response to climatic differences between Alaska and Michigan. At low
temperatures (12°C), Alaskan larvae grew faster (fifth instar doubling time of 5.8
versus 9.1 d), and molted faster (fifth molt required 11.8 versus 17.8 d). Increased
egg size and reduced adult size represent additional adaptations Of Alaskan P.
canadensis for short subarctic summers. In the Alaskan climate, Alaskan
swallowtails have an estimated fitness of 3.00 relative to Michigan swallowtails.
Changes in growth-temperature responses made the greatest apparent
contribution to enhanced fitness, followed by increased egg size. Occasional
extreme summers appear to be more important than average summers in shaping
adaptive responses. Regional adaptation to climate allows P. canadensis to
maintain a broader distribution than would Otherwise be possible, but distribution

limits are probably still constrained by summer temperatures.

 

 

ACKNOWLEDGMENTS

This project benefited in many ways from the labors and intellectual
generosity of Bruce Ayres, Darsie Ayres, Janice Bossart, Guy Bush, Ed Debevic,
Bob Hagen, Don Hall, Dan Herms, Kelly Johnson, Bob Lederhouse, Steve
MacLean, Bill Mattson, Jim Miller, James Nitao, and Mark Scriber. Financial
support was provided by NSF BSR 88-01184, USDA 87-CRCR-1-2581, the
Michigan AES (Projects 1640, 1644, and 8072), graduate fellowships from
Michigan State University College Of Natural Sciences and a Barnett Rosenberg
Fellowship. Weather data were provided by the Michigan Department Of
Agriculture, Michigan State University, and the Geophysical Institute, University
of Alaska Fairbanks. My temperature-driven development model was adapted
from an earlier version written by Stephen F. MacLean. The Institute of Arctic
Biology, University of Alaska Fairbanks generously provided laboratory Space and
a rich intellectual environment. I especially acknowledge the indefatiguable

support of Rosemary, Sarah, and Benjamin Ayres.

iv

TABLE OF CONTENTS

LIST OF TABLES ........................................... iix

LIST OF FIGURES ........................................... x

INTRODUCTION ............................................ 1
CHAPTER 1 - Conservation of polyphagous abilities in

oligophagous populations of Papilio canadensis ............ 3

Introduction ............................................. 3

Methods .............................................. 5

Overview ........................................... 5

Sampling of insects and hosts ............................ 6

First instar growth performance ........................... 7

Middle instar growth performance ......................... 8

Final instar growth performance .......................... 9

Pupal mass, larval duration, and survival ................... 10

Statistical analyses .................................... 10

Results ............................................. 12

First instar growth performance: intraspecific variation ......... 12

First instar growth performance: interspecific variation ......... 15

Growth performance during the middle instars ............... 18

Growth performance during the final instar ................. 21

Pupal mass, larval duration, and survival ................... 25

 

Discussion .............................................
Evaluation of hypotheses ...............................
Explanation for invariance of nutritional physiology ...........
Implications for herbivore evolution .......................

CHAPTER 2 - Local adaptation to regional climates
in Papilio canadensis ..............................

Introduction ............................................

Methods .............................................

Population sampling ..................................
Larval growth during the fourth and fifth instars .............. .
The duration Of molts .................................
Basking success ......................................
Statistical analyses ....................................
Results .............................................
Larval growth rates ...................................
Dry mass and nitrogen budgets ..........................
Molting physiology ...................................
Pupal mass .........................................
Survival ...........................................
Development time ...................................

Basking success ......................................

vi

 

 

Discussion ............................................. 73
Adaptive modifications of temperature physiology ............. 73
Conserved attributes .................................. 76
Egg size and adult size ................................ 77
Comparing the ecological worth Of temperature adaptations ..... 78
Tradeoffs .......................................... 82
Model validation ..................................... 85
Northern distribution limits of P. canadensis ................. 86

CONCLUSIONS ............................................. 88
APPENDIX 1. Record of deposition of voucher specimens .............. 93
APPENDIX 2. P. canadensis first instar growth rates ................... 95
APPENDIX 3. Parameters used in P. canadensis development model ....... 96
BIBLIOGRAPHY ............................................ 97

vii

~ ;“t’ ‘-

 

Table 1:

Table 2:

Table 3:

Table 4:

Table 5:

Table 6:

Table 7:

Table 8:

LIST OF TABLES

ANOVA comparing the first instar growth rate of P. canadenszls
larvae from Alaska and Michigan on host species occurring in
Alaska and Michigan ................................... 14

ANOVA comparing the first instar growth rate of P. canadensis
AK and P. glaucus M1 on host species occurring in Alaska and
southern Michigan ..................................... 17

ANOVA results comparing the growth performance P.

canadensis populations from Alaska and Michigan feeding on

nine Alaskan host species. RGRmid is the relative growth rate

during the middle instars (day 3 - 17), and M17 is the larval mass

on dayI 17. .......................................... 20

ANOVA results comparing 5th instar growth-energetics of P.
canadensis from Alaska and Michigan feeding on nine Alaskan
host species. ........................................ 23

ANOVA results comparing the pupal mass and larval period of
P. canadensis from Alaska and Michigan reared on nine Alaskan
host species. ....................................... 26

Cumulative survival of P. canadensis larvae from Alaska and

Michigan reared on nine Alaskan host species. Each host began

with 19 or 20 neonate larvae from each population. Pupal

survival only includes the first half of winter diapause. .......... 29

Summary Of ANOVAS comparing the relative growth rates of

fourth (RGR4) and fifth (RGRS) instar P. canadensis from

Alaska and Michigan on three hosts at four temperatures.

Corresponds to data in Fig. 10. ........................... 51

Nitrogen use efficiency (NUE) of fifth instar P. canadensis from

Alaska and Michigan feeding at four temperatures on three

hosts (Populus tremuloides, Populus balsamifera, and Betula

resim'fera). N = 4-5 larvae (3-5 families) in each treatment
combination (SE in parentheses) ........................... 58

viii

 

Table 9:

Table 10:

Table 11:

Table 12:

Table 13:

Table 14:

Summary of ANOVAS comparing the duration of the fourth

molt (Molt4) in P. canadensis from Alaska and Michigan on

three hosts at four temperatures. Corresponds to data in

Fig. 14. ............................................ 61

ANOVA comparing the duration of the fifth molt in P.

canadensis larvae from Alaska and Michigan at four

temperatures (square root transformed data). Corresponds to

data in Fig. 15. ...................................... 63

ANOVA comparing the pupal mass of male and female P.
canadensis from Alaska and Michigan reared at four
temperatures. Corresponds to data in Fig. 16. ................ 66

Survival Of P. canadensis from Alaska and Michigan reared at

four temperatures in 1989. Larval survival is the percentage of

early fourth instar larvae beginning the experiment that

successfully pupated. Pupal survival is the percentage of pupae

that survived 9 mo Of diapause and emerged as adults the

following spring. Cumulative survival is the product of larval

and pupal survival. ................................... 67

A comparison Of the value in enhanced developmental success

of adaptations exhibited by Alaskan P. canadensis. Percent

larval success indicates the average proportion of larvae

predicted to complete development during 48 seasons in Alaska

(based on data in Figure 20 assuming cohort sizes of 12%, 32%,

27%, 18%, and 11% for cohorts 1-5a). Alaskan attributes were
introduced into the Michigan phenotype one at a time (eg size,
pupation size, molting rates, and growth rates) and then

simultaneously (Alaskan phenotype). ...................... 83

Summary of geographic differentiation between P. canadensis

from Alaska and Michigan. Traits are organized by the

environmental factor hypothesized to select for divergence in the
Alaskan population (low summer temperatures, or unique
phytochemistry of their hosts). Change is expressed in terms Of
percentage, standard deviations, and statistical significance. ...... 89

 

 

Figure 1:

Figure 2:

Figure 3:

Figure 4:

Figure 5:

Figure 6:

LIST OF FIGURES

Growth rate of first instar P. canadensis from Alaska and

Michigan compared on nine Michigan host species (upper) and

nine Alaskan host species (lower). Points near the line Of

equality represent hosts on which the two insect populations
performed equally well. Error bars show :1 SE. Host species

are listed in appendix. ................................. 13

Growth rate of first instar P. canadensis and first instar P.

glaucus compared on six host species from southern Michigan

and nine from Alaska. Points near the line of equality

represent hosts on which the two insect species performed

equally well. Error bars show :1 SE. Host species are listed

in appendix. ........................................ 16

Growth performance of P. canadensis from Alaska and

Michigan compared on nine Alaskan host species. Upper figure

shows relative growth rate during the middle instars (day 3 to

day 17). Lower figure shows larval mass on day 17. Error bars
indicate 95% confidence intervals. Host species, from left to

right, are Populus tremuloides, Populus balsamifera, Salix

alaxensis, Salix bebbiana, Salix glauca, Salix novae-angliae, Betula
resinzfera, Alnus crispa, and Alnus tenuifolia ................. 19

Growth performance of final instar P. canadensis from Alaska

and Michigan compared on nine Alaskan host species. Growth

rate (upper right) equals the product of consumption rate,

apparent digestibility, and ECD. Error bars indicate 95%

confidence intervals. Host species as in Fig. 3. ............... 22

Nitrogen use efficiency of final instar P. canadensis from Alaska
and Michigan compared on nine Alaskan host species. Error
bars indicate 1 SE. Host species as in Fig. 3. ................ 24

Pupal mass and larval duration of P. canadensis from Alaska
and Michigan reared on nine Alaskan host species. Error bars
indicate 95% confidence intervals. Host species as in Fig. 3 ...... 27

Figure 7:

Figure 8:

Figure 9:

Figure 10:

Figure 11:

Figure 12:

Pupal mass as a function of larval duration in P. canadensis
from Alaska and Michigan reared on Alnus tenuzfolia. Each
point represents one pupa. ............................. 28

Frequency distribution of thermal sums accumulated above a

10°C base in Fairbanks, Alaska (48 years) and Chatham,

Michigan (50 years). Michigan weather records included

1931-1980; Alaska records included 1936-1980 (excluding 1969)

and 1987-1990. Annual accumulations were begun on 1 April

and terminated near the time of leaf fall (15 September in

Alaska and 1 October in Michigan). Thermal sums were

calculated from daily maxima and minima using a sine function
recommended by Watanabe (1978). ....................... 39

Distribution of daily mean temperatures encountered by P.

canadensis larvae in Fairbanks, Alaska (48 years) and Chatham,
Michigan (50 years) (weather records as in Fig. 1). The period

of P. canadensis larval development was defined as 350-700

degree days in Michigan and 250-550 degree days in Alaska

(truncated at 15 September when necessary). The percentage

of days in each temperature class was calculated separately for

each season, and the figure shows the average percent across 48

or 50 years of records (otherwise the distribution would be

weighted towards long cold seasons). ...................... 40

Relative growth rates of P. canadensis larvae from Alaska (solid

lines) and Michigan (dashed lines) as a function of temperature.
Experiments included fourth and fifth instars reared on Populus
balsamifera and P. tremuloides in 1988 and P. tremuloides in

1989. Some standard errors are obscured by the data point.

See Table 7 for corresponding ANOVAS. ................... 50

Consumption rate, apparent digestibility, and growth efficiency

(ECD, growth/assimilation) Of P. canadensis larvae from Alaska

(solid lines) and Michigan (dashed lines) as a function of

temperature. Corresponding growth rates appear in middle left

of Fig. 10 (fourth instars feeding on Populus tremuloides in

1988). Some standard errors are obscured by the data point. . . . 53

Consumption rate, assimilation rate, metabolic rate (ADMR),
apparent digestibility, and growth efficiency (ECD,
growth/assimilation) Of P. canadensis larvae from Alaska (solid

lines) and Michigan (dashed lines) as a function of temperature.
Corresponding growth rates appear in lower right of Fig. 10

(fifth instars feeding on Populus tremuloides in 1989). Some

standard errors are obscured by the data point. ............. 55

xi

Figure 13:

Figure 14:

Figure 15:

Figure 16:

Figure 17:

Figure 18:

Growth rate, consumption rate, apparent digestibility, and

growth efficiency (ECD, growth/assimilation) of P. canadensis

fifth instars feeding on three hosts as a function of temperature.

Hosts were Populus balsamifera (dotted lines), P. tremuloides

(solid lines), and Betula resinifera (dashed lines). Some

standard errors are obscured by the data point. ............. 56

The duration of the fourth molt (Molt4, fourth instar to fifth

instar) in P. canadensis from Alaska (solid line) and Michigan

(dashed line) as a function of temperature. Experiments

included larvae reared on Populus balsamifera and P.

tremuloides in 1988 and P. tremuloides in 1989. Some standard

errors are obscured by the data point. See Table 9 for

corresponding ANOVAs. ............................... 60

The duration of the fifth molt (Molts, fifth instar to pupa) in P.
canadensis from Alaska (solid line) and Michigan (dashed line)

as a function of temperature. Larvae were reared on P.

tremuloides in 1989. Some standard errors are obscured by the

data point. See Table 10 for corresponding ANOVA ........... 62

Pupal mass of male and female P. canadensis from Alaska

(solid line) and Michigan (dashed line) as a function of

temperature. Larvae were reared on P. tremuloides in 1989.

See Table 11 for corresponding ANOVA .................... 65

Development time from early fourth instars to pupae of P.

canadensis from Alaska (upper) and Michigan (lower) as a

function of temperature. Durations of the fourth instar, the

fourth molt, the fifth instar, and the fifth molt, are indicated. . . . . 69

Basking success as a function of basking potential in fifth instar
P. canadensis from Alaska (closed circles) and Michigan (open
circles). Basking success equals the difference between larval
temperature and ambient air temperature; basking potential
equals the difference between the temperature of a high
absorption "black body" and ambient air temperature. The
heavy solid line shows the points of equality between success
and potential. The pooled regression function is indicated
(regressions fit to each population separately did not differ:
Fl,112 = 1.12, P > 0.20). Note that most points to the right of
6°C on the x axis represent observations of larvae on detached
branches placed in full sunlight and protected from the wind. . . . . 71

xii

 

Figure 19: The temperatures of Alaskan and Michigan larvae when
allowed to move and bask naturally under field conditions near
Fairbanks, AK. Each point is the mean of 5-10 larvae from
each population. Standard deviations ranged from 05°C at low
temperatures to 4°C at high temperatures. .................. 72

Figure 20: The predicted developmental success of various P. canadensis
phenotypes under two scenarios of host quality during 48
seasons in interior Alaska. Possible developmental success in
each year ranged from 0 successful cohorts (extinction) to 5
successful cohorts (no mortality due to incomplete
development). Under each host quality scenario, I began with
the Michigan phenotype (top), then changed egg mass, pupation
threshold, molting rates, and growth rates (one at a time) to the
condition of the Alaskan phenotype. Bottom figures show
results when the four Alaskan adaptations were combined. ...... 81

xiii

 

INTRODUCTION

Why do organisms occur where they do but not elsewhere? How do some
species maintain very cosmopolitan distributions while others are highly endemic?
Why do some populations respond to environmental change through adaptation
while others go extinct? Is it possible to generalize about the types of ecological
attributes that are likely or unlikely to respond to changes in selection pressure. I
addressed these broad questions through a study of geographic differentiation in a
pair of broadly distributed insect species that encounter very different
environments across their range.

Two closely related species of the tiger swallowtail butterfly, Papilio
glaucus and Papilio canadensis (Lepidoptera: Papilionidae), are abundant and
conspicuous members of the herbivore fauna from southern Florida to interior
Alaska. Their larvae are able to complete development on hosts representing at
least 14 plant families (30-40 species; Scriber 1983, 1984) under summer
temperature regimes that run the gamut from >3500 degree days to <700 degree
days (10°C base). NO single host species occurs throughout this north-south
transect. Even within the species, populations encounter dramatically different
environments from one end of their distribution to the other. How do these
swallowtail populations succeed in completing their life cycle each year in such

disparate environments? Under one scenario, each species has converged on a

2
single cohesive genotype (Mayr 1963) that is sufficiently robust to allow rather
cosmopolitan success (super-genotype hypothesis). In the alternative view, species
are composites of variable genotypes and differentiated populations (Dethier,
1954), any one of which is able tO perform well in only a subset of conditions
encountered by the species as a whole (genetic mosaic hypothesis). The genetic
mosaic hypothesis suggests that local populations are continually being modified
through natural selection to provide a match between the environment and the
phenotype; i.e., the environment determines organism attributes. In contrast,
support for the cohesive genotype scenario would indicate that species attributes
are largely fixed, and suggest instead that an organism’s attributes determine the
environment where it can occur.

Using the experimental format of a reciprocal transplant, I evaluated these
hypotheses by comparing the physiology ecology of P. canadensis populations from
Alaska and Michigan on the host species normally encountered by each (Chapter
1), and at the temperatures normally encountered by each (Chapter 2).

Interspecific comparisons with P. glaucus allowed a comparison of variation within

and between species.

 

CHAPTER ONE

Conservation of polyphagous abilities in
OligOphagous populations of Papilio canadensis

Introduction

When Ehrlich and Raven (1964) formulated what was to become a
dominant paradigm for the evolution of insect-host plant relations, they framed
the model very explicitly in terms of species characteristics. Attributes such as
detoxification abilities were viewed as constants within a species, with significant
evolutionary advances being occasional, rapid, and most often associated with
Speciation events: "The evolution of secondary plant substances and the stepwise
evolutionary responses to these by phytophagous organisms have clearly been the
dominant factors in the evolution of butterflies and other phytophagous groups"
(my italics). The same typological thinking underlies the influential papers
written 22 years later by Feeny (1976) and Rhoades and Cates (1976). Although
the idea that herbivore species may be genetically heterogeneous is old (e.g.,
Brues, 1924; Thorpe, 1930; and Dethier, 1954), the importance of this variability
in understanding insect-host interactions has been commonly overlooked until
recently. In 1978, Edmunds and Alstad emphasized that interactions between
scale insects and ponderosa pine varied widely depending on characteristics of the

particular insect genotype and the individual tree attacked. Since then, various

4

studies have reported variation between populations and/or genotypes in the
ability to use particular hosts (Hsiao, 1978; Moran, 1981; Tavormina, 1982;
Rausher, 1982, 1984; Blau and Feeny, 1983; Service, 1984; Via, 1984a; Haukioja
and Hahnimaki, 1985; Weber, 1985; Bowers, 1986; Hare and Kennedy, 1986;
Futuyma and Philippi, 1987; Horton et al., 1988; Nitao et al. 1991). Fox and
Morrow (1981) and Scriber (1983, 1986) suggested that apparently polyphagous
species may actually be mosaics of relatively oligophagous genotypes. If this
hypothesis were substantiated, it would alter the scale on which we study and
describe plant-herbivore interactions (Ng, 1988; Thompson, 1988). Recent
reviews have addressed intraspecific variability in herbivores (Denno and
McClure, 1983; Diehl and Bush, 1984; Futuyma and Peterson, 1985; Jaenike,
1990; Via, 1990), but its generality and its magnitude relative to variation
between herbivore species remains poorly known.

Papilio canadensis R & J (formerly P. glaucus canadensis, Hagen et al.,
1991) is a polyphagous tree-feeding insect (Lepidoptera: Papilionidae) common
throughout the boreal and north-temperate forests of the Nearctic. Its hosts
include species of Populus and Salix (Salicaceae), Betula and Alnus (Betulaceae),
Prunus and Amelanclzier (Rosaceae), Fraxinus (Oleaceae), Tilia (Tiliaceae), and
probably more (Scriber and Ayres 1990). Its distribution is more extensive than
most of the tree species on which it feeds, hence populations inhabiting the Great
Lakes region of Wisconsin and Michigan necessarily feed on a different suite of
host species than those inhabiting interior Alaska 4000 km distant.

With respect to its nutritional physiology, P. canadensis could be a mosaic

of differentiated populations and variable genotypes that only collectively possess

_j

5
the ability to use all of its reported host plants; alternatively, a single unusually
robust genotype allows its polyphagous habits and broad distribution. I refer to
these as the genetic mosaic and super-genotype hypotheses, respectively. They
were tested by comparing the growth performance of P. canadensis populations
from Alaska and Michigan on nine Alaskan host species and nine Michigan host
species. If genetic variability is large and pervasive it should be evident within
each population as interactions between insect families and host species.
Additionally, the genetic mosaic hypothesis predicts regional differentiation of the
populations such that Alaskan populations should be generally better suited to
Alaskan hosts and Michigan populations better suited to Michigan hosts (Scriber
1983, 1986). To provide context for evaluating such intraSpecific variability, I
included interspecific comparisons with P. glaucus L, the apparent closest relative

of P. canadensis (Scriber et al., 1990; Hagen and Scriber, 1991).

Methods

Overview

In Papilio, host effects are often apparent as differences in neonate
survival (Scriber et al., 1990; Hagen, 1990). My most extensive experiments
compared first instar relative growth rates, which similarly reflect neonate growth
performance, but provide a continuous rather than a discrete variable for analysis.
Results of first instar trials allow a partitioning of variance due to ( 1) interspecific
variation between sister species, (2) intraspecific variation at the level of
populations and full-sib families, (3) effects of host species, and (4) interactions

between insect genotypes and host species. One set of first instar grth trials

 

 

6
(P. canadensis from Alaska and Michigan tested on nine Alaskan host species)
was extended to allow measurements of growth rate in the middle and final
instars, pupal mass, larval duration, and survival by stadia. Fifth instar growth
performance was further compared through measurements of consumption rate,

apparent digestibility, growth efficiency, and nitrogen use efficiency.

Sampling of insects and hosts

During early June 1988, adult P. canadensis females were collected from
sites in interior Alaska (vicinity of Fairbanks) and northern Michigan/Wisconsin
(Gogebic and Iron counties in MI, and nearby Vilas and Iron Counties in WI).
At the same time, adult P. glaucus females were collected from southern
Michigan (Washtenaw, Ingham, and Cass Counties). All adults were transported
to our laboratory in Michigan, fed a 20% honey solution daily, and induced to
oviposit in clear plastic boxes (10 x 20 x 27 cm) with sprigs of appropriate foliage
(100% humidity, 4 h dark at 24° alternating with 4 h bright light at 32°).
Resulting neonate larvae were used for comparisons of P. canadensis and P.
glaucus growth performance on six host species common in southern Michigan. A
slightly later cohort of progeny from the same collection of adults were
transported as eggs to Fairbanks, Alaska where they were used for comparisons
of P. canadensis from Alaska and Michigan on nine Alaskan host species. The
following year (June 1989), adult P. canadensis females were collected from the
same sites in interior Alaska and northern Michigan and their progeny were
tested on nine host species common in northern Michigan and the Great Lakes

region. During July 1989, adult P. glaucus were collected from Gallia County

 

7
Ohio, and their progeny were tested on nine Alaskan host species. Each of the
three Papilio populations was represented in my experiments by 15-20 full sib
families (7-10 families in each year). Voucher specimens have been deposited
with the Michigan State University Department of Entomology museum.

Alaskan hosts were represented by nine tree species in four genera and
two families: Betula resim'fera, Alnus crispa, and Alnus tenuifolia in the Betulaceae;
and Populus tremuloides, Populus balsamifera, Salix alaxensis, Salix bebbiana, Salix
glauca, and Salix novae-angliae in the Salicaceae (see appendix). Foliage was
collected from or near the University Of Alaska Fairbanks arboretum. Northern
Michigan host species were represented by nine species in six genera and five
families: Fraxinus americana (Oleaceae); Prunus serotina and Prunus virginiana
(Rosaceae); Tilia wnen'cana (Tiliaceae); Betula alleghaniensis, Betula papynfera,
and Alnus rugosa (Betulaceae); and Populus grandidentata (Salicaceae). Foliage
was collected from Dunn County, Wisconsin. Southern Michigan host species
were represented by six species in five genera and five families: Liriodendron
tulipifera (Magnoliaceae); Ptelea trifoliata (Rutaceae); Tilia amen'cana (Tiliaceae);
Prunus serotina and Prunus virginiana (Rosaceae); and Fraxinus americana
(Oleaceae). Foliage was collected from Ingham County, Michigan. Each host
species was represented in each first instar growth trial by foliage from five

genets.

First instar growth performance
Recently eclosed larvae were weighed to $0.1 mg (W0), then allocated

individually to clear plastic cups (12 ounce) containing a leaf or shoot of one host

8
species. Plaster-of-paris bases in the cups were saturated with water, which
provided high humidity and maintained leaf turgor throughout the trial. Larvae
from each family were spread equitably across host species. On the third day
(after about 60 h at 24°C, photoperiod Of L:D 18:6), when larvae on the best
hosts were nearly ready to initiate their first molt, larvae were reweighed (W3).
First instar relative growth rate was calculated as RGRI = (ln(W3-O.9) -
ln(W0))/T, where T = the elapsed time in days (Gordon, 1968). W3 was reduced
by 10% (= estimated mass of food in their gut, unpublished data) to make it
comparable to W0 (taken before the larvae had begun to feed). This adjustment
affected no patterns or statistical results but corrected for what would otherwise
be a bias in the growth rate. All population and species comparisons were based
on concurrent measurements using foliage collected from the same trees at the
same time, except for comparisons of P. canadensis and P. glaucus on Alaskan
host species, where foliage was collected from the same sites but at different

times (see appendix).

Middle instar growth performance

On day 17 Of the experiment comparing P. canadensis from Alaska and
Michigan on Alaskan host species, all larvae were reweighed (W17) and their
instar was recorded. At this time, some larvae on the best hosts had just entered
the fifth (final) instar. A relative growth rate encompassing these middle instars
was calculated as RGRmid = (ln(W,7) - ln(W3))/ T, where W, = mass at the end of

the first instar trial, and T = time elapsed from W3 to W17.

Final instar growth performance

Shortly after larvae completed molting into the fifth and final instar (had
resumed feeding and producing frass), they were weighed, given a previously
weighed shoot of foliage, and allowed to feed and grow uninterrupted for the
balance of the instar (ca. 3 (1), when they were reweighed. A matched shoot of
weighed foliage was placed in an adjacent cup and exposed to the same
conditions as the foliage fed to the larva; this served as a control to estimate the
initial dry mass of the experimental foliage. After the trial, the experimental
foliage, the control foliage, and the frass produced during the 3 d trial were
oven-dried and weighed. A subset of the control foliage and frass samples was
analyzed for total nitrogen with standard micro-Kjeldahl techniques (sulfuric acid
digestion followed by colorimetric analysis with a Technicon Auto-analyzer II;
Herms, 1991). Non—experimental fifth instar larvae from Alaska and Michigan
were used to estimate the dry mass of experimental larvae from their fresh mass
(DM = 0.125 - FM, n= 20; dryrwet ratio was unaffected by host species). These
data allowed estimation of fifth instar relative growth rate (RGRV), relative
consumption rate (RCR), relative assimilation rate (RAR), apparent digestibility
(AD), efficiency of conversion of digested matter (ECD), and nitrogen use
efficiency (NUE). Calculations of RCR, RAR, and RGR follow formulae in
Gordon (1968). AD = RAR/RCR, and ECD = RGR/RAR. NUE equalled the
fraction of ingested nitrogen that did not appear in the frass (frass contains both
undigested and excretory nitrogen): NUE = (Ningestcd - Ncm,cd)/Ningcs,cd. All

indices of fifth instar growth performance are in units of dry mass.

10
Pupal mass, larval duration, and survival
Larvae were checked daily, and foliage replaced as needed (every 2-3 d)
until larvae pupated or died. Larval duration was defined as the interval from
initiation of feeding in neonate larvae (W0), until one day before the conspicuous
color change that signals prepupation. The diapausing pupae were moved to 12°
for one month, then weighed and sexed. Larval mortality was recorded by day

and instar.

Statistical analyses

All data were analyzed by ANOVA (GLM procedure, Type III sums of
squares: SAS Institute, 1985) after evaluating assumptions of normality and
homogeneity of variances. I tested for population differentiation in first instar
growth rate with a mixed model design, where insect population (AK and MI),
host community (AK and MI), and host species nested within host community
were all treated as fixed effects. Insect families were nested within insect
population and host community, and were treated as random effects. The
expected mean squares and F—tests follow Scheffe’s formulation of the mixed
model ANOVA (Ayres and Thomas, 1990). I tested for interspecific patterns in
first instar growth with an identical model except insect species was substituted
for insect population and the host communities were from Alaska and southern
Michigan. Both analyses excluded some families not represented on all hosts,
which facilitated tests for family x host interactions, and did not alter the result of
tests involving insect populations or insect species.

Larval mortality subsequent to the first instar trial reduced the number of

 

11

. families that were represented on all hosts and reduced total degrees of freedom.
Consequently, I used a simpler ANOVA model to analyze growth performance in
the middle instars, growth performance in the final instar, pupal mass, and larval
duration: insect population and host species were treated as fixed effects, and the
analysis was run using family means (Hurlbert, 1984). All figures, including first
instar growth rates, show means and confidence intervals (or SE’s) calculated from
family means.

Pupal mass and larval duration tend to be sexually dimorphic in P.
canadensis, which could confound analyses because some treatments yielded
unequal numbers of male and female pupae. Sex could not be readily included
as a factor in the ANOVA model because some treatment combinations lacked
male or female pupae. I evaluated sexual dimorphism by comparing mean male
and female values for the 18 population-host combinations. The larval duration
of females averaged 1.5 d (5%) longer than males (t17 = 3.18, P < 0.001), and
the pupal mass of females averaged 39 mg (6%) greater than males (tl7 = 4.12, P
< 0.001). I developed the following linear regression equations:

Daysfcmalc = -100 + 1.09-Daysmale (P < 0.0001, df = 1,17, r2 = 0.90),

Massfemale = -141 + 1.26-Mass

(P < 0.0001, df 1,17, r2 = 0.82),

male

which were used to adjust male measurements to that expected for females
(Haukioja and Neuvonen, 1985; Ayres et al., 1987). Residuals from the
regressions did not differ between populations or hosts (P > 0.15 for all four
tests). Analyses of larval duration and pupal mass used these adjusted values.

The effects of these adjustments were small (grand means changed by 2-3%) and

 

12
the ANOVA conclusions did not change when I reran the analyses using

unadjusted values.

Results

First instar growth performance: intraspecific variation

Counter to the predictions of the genetic mosaic hypothesis, there was
little evidence for enhanced performance of first instar P. canadensis larvae on
regionally available hosts. Michigan larvae grew no faster than Alaska larvae
when both populations were challenged with foliage from nine Michigan host
species (Figure 1, upper; least square means : SE equalled 0.410 t 0.039 versus
0.418 1' 0.029 mg- mg'1 - d‘1 for Michigan and Alaskan larvae respectively, P =
0.81). Similarly, Alaskan larvae grew only slightly faster than Michigan larvae
when both populations were challenged with foliage from nine Alaskan species
(Figure 1, lower; least square means 1 SE equalled 0.525 1 0.029 versus 0.466 t
0.019 mg-mg’1 - d'1 for Alaska and Michigan respectively, P = 0.06). The
interaction between insect population and host community was nonsignificant
(F120 = 0.64, P = 0.43, Table 1). There was a weak overall tendency for Alaskan
larvae to grow faster than Michigan larvae (F 120 = 3.34, P = 0.083, Table 1).

The main effect of host species on first instar growth performance was
large and significant (1’16,160 = 14.27, P = 0.0001; Figure 1), and there was
significant variation between insect families within populations (F20.357 = 2.72, P
< 0.0001, Table 1). But, counter to the predictions of the genetic mosaic

hypothesis, there was no evidence for within-population interactions between

insect families and host species (me = 0.74, P = 0.99, Table 1). Some

 

l3

0-8j N. MICHIGAN HOSTS

0.6:

0.4: Q4

0-2‘. @- iSE
00

 

 

0-8‘, ALASKA HOSTS
0.6:
0.4:

0.2 ‘

GROWTH RATE OF ALASKA LARVAE (mg-mg‘l-d‘l)

 

 

0.0 . . . , . . r f- - r , . . . ,
0.0 0.2 0.4 0.6 0.8
GROWTH RATE OF MICHIGAN LARVAE

(mg-mg"-d")

Figure 1: Growth rate of first instar P. canadensis from Alaska and Michigan
compared on nine Michigan host species (upper) and nine Alaskan host species
(lower). Points near the line of equality represent hosts on which the two insect
populations performed equally well. Error bars show :1 SE. Host species are
listed in appendix.

 

’

 

9 —_

 

14

Table 1: ANOVA comparing the first instar growth rate of P. canadensis larvae
from Alaska and Michigan on host species occurring in Alaska and Michigan.

 

 

SOURCE DF
A. Insect pOpulation 1
B. Host community 1

C. Insect pop. x Host community 1
D. Host species (community) 16
E. Insect pop. x Host species 16
F. Insect family (p0p., comm.) 20
G. Insect family x Host species 160

H. Error 357

MS - 104
2316
6654

445

162

694

255

3.34

9.59

0.64

14.27

0.86

2.72

0.74

0.083
0.0057
0.43
0.0001
0.50
0.0001

0.99

 

3‘ The F test denominator for sources A, B, and C was MSFamily'
The F test denominator for sources D, and E was MSFam

The F test denominator for sources F, and G was

ily x Host species‘
r

15
families tended to have higher first instar growth rates than other families, but
the pattern among families was unaffected by host species. This test for family x
host interactions was very general in that it simultaneously considered nine host
species from each of two regions. A strong interaction involving any pair of host
species could have been obscured if the other host species contributed little to
the family x host interaction. As a check of this possibility, I tested for family x
host interactions using 72 different subsets of data that compared each possible
pair of hosts from Alaska and from Michigan. None of the 72 F-tests was
significant at P < 0.05 (degrees of freedom were 8-14 for the numerator and

37-47 for the denominator).

First instar growth performance: interspecific variation

In contrast to the absence of variation within P. canadensis, there were
striking differences between P. canadensis and its sister species, P. glaucus. P.
glaucus larvae grew significantly faster than P. canadensis when both species were
challenged with six hosts from southern Michigan (Figure 2, upper; least square
means t SE equalled 0.486 t 0.018 versus 0.355 t 0.026 mg-mg‘l - d‘1 for P.
glaucus and P. canadensis respectively, P = 0.008). The pattern was reversed in
the reciprocal test, where P. canadensis larvae grew significantly faster than P.
glaucus larvae when both species were challenged with nine Alaskan host species
(Figure 2, lower; least square means t SE equalled 0.525 t 0.014 versus 0.326 t
0.012 for P. canadensis and P. glaucus respectively, P < 0.0001). The interaction

between insect species and host community was highly significant (F1.19 = 71.49, P

< 0.0001, Table 2), and in the hypothesized direction (insect species grew fastest

 

 

 

 

03‘. S. MICHIGAN HOSTS
A 05: iSE
...,_, . ~1-
'°’ 0.4:
e .
t.» ‘ »—§—«
§ .
LU 0'2: +
1- .
< 1
LEGO
I
f-
E 0.8-
m 9‘
o _- “9‘
9 0.6‘ g,
(D .
g 0.4- £4?
< .
2 1
E 0.2-

. ALASKA HOSTS
0.0 . . . 'T

 

0.0 0.2 0.4 0.6 , 0.8

GLA ucus GROWTH RATE
(mg-mg“-d")

Figure 2: Growth rate of first instar P. canadensis and first instar P. glaucus
compared on six host species from southern Michigan and nine from Alaska.
Points near the line of equality represent hosts on which the two insect Species
performed equally well. Error bars Show :1 SE. Host species are listed in
appendix.

—

 

17

Table 2: ANOVA comparing the first instar growth rate of P. canadensis AK and
P. glaucus MI on host species occurring in Alaska and southern Michigan.

 

 

 

SOURCE DF MS - 10‘ F’ P

A. Insect species 1 726 4.02 0.060
B. Host community 1 25 0.14 0.71
C. Insect spp. x Host community 1 12926 71.49 0.0001
D. Host species (community) 13 2176 14.38 0.0001
E. Insect species x Host species 13 1061 7.01 0.0001
F. Insect family (A, B) 19 181 0.96 0.51
G. Insect family x Host species 131 151 0.80 0.92
H. Error 209 189

 

°' The F test denominator for sources A, B, and C was MSFamily'
The F test denominator for sources D, and E was MSFamily x Hos, species'
The F test denomlnator for sources F, and G was MSEmr.

 

18
on the hosts that they normally encounter). On southern tree species, the
differences between P. glaucus and P. canadensis were most dramatic on
Liriodendron tulipifera (Magnoliaceae), but P. glaucus also tended to grow faster
on Ptelea tnfoliata (Rutaceae) and Prunus virginiana (Rosaceae) (Figure 2,
Appendix). On boreal tree species, the grth rates Of P. canadensis larvae were

generally elevated relative to P. glaucus larvae (Figure 2).

Growth performance during the middle instars

P. canadensis populations from Alaska and Michigan did not differ in their
relative growth rate spanning days 3 to 17 (Figure 3, upper; population effect in
Table 3). Most larvae were just completing the first instar on day 3. By day 17,
larval development ranged from midway through the third instar (on poor hosts)
to early in the fifth and final instar (on good hosts). Although relative growth
rates were nearly identical between populations, Alaskan larvae had attained
significantly greater mass by day 17 (Figure 3, lower; population effect in Table
3). This was attributable to larger egg size and greater neonate mass in Alaskan
larvae. At hatch, Alaskan larvae were 1.4 times heavier than Michigan larvae
(mean t SE = 1.33 i 0.03 versus 0.98 i 0.03 mg; FL16 = 31.41, P < 0.0001;
based on 360 total larvae from 18 families). By day 17, typical larvae from both
populations had increased their mass by 170 fold, but Alaskan larvae still
averaged 1.5 times heavier than Michigan larvae (mean t SE = 241 i 9 versus

158 t 9 mg). Host species had a large effect on growth performance during the

middle instars (Figure 3, Table 3).

 

19

I P. canadensis AK

E! P. canadensis MI 95% Cl

0.30

0.25

0.20

GROWTH RATE, DAYS 3-17
(m9 m9 1 d 1)

 

 

 

0.1 5 ‘
500
400

300

200
IIIIII

trem ba/s a/ax bebb g/auc nov resin crisp ten

MASS AT DAY 17
(m9)

0

 

 

Popu/us Salix Betu/a A/nus
ALASKA HOST SPECIES

Figure 3: Growth performance of P. canadensis from Alaska and Michigan
compared on nine Alaskan host species. Upper figure shows relative growth rate
during the middle instars (day 3 to day 17). Lower figure shows larval mass on
day 17. Error bars indicate 95% confidence intervals. Host species, from left to
right, are Populus tremuloides, Populus balsamifera, Salix alaxensis, Salix bebbiana,
Salix glauca, Salix novae-angliae, Betula resinifera, Alnus crispa, and Alnus
tenuifolia.

 

 

20

Table 3: ANOVA results comparing the grth performance P. canadensis
populations from Alaska and Michigan feeding on nine Alaskan host
species. RGRmi is the relative grth rate during the middle instars (day
3 - 17), and M17 15 the larval mass on day 17.

 

 

 

RGRmid M17
Source DF Ms- 106 F MS ~ 10" F
Population 1 332 0.71 24014 49.15’"
Host 8 8247 1769'" 12802 2620'"
Pop x Host 8 1076 2.30' 588 1.20
Error 107 468 488

 

‘ P < 0.05; P < 0.0001

 

21

Growth performance during the final instar

The growth rates and efficiencies of final instar larvae were strongly
affected by host species, but were nearly identical between P. canadensis
populations (Figure 4, Table 4). Relative consumption rate ranged from about
1.4 mg- mg"1 - d'1 on Salix alaxensis, Salix bebbiana, Salix glauca, and Alnus cn'spa
to about 2.0 mg- mg'1 - d'1 on Populus balsamifera, Populus tremuloides, Salix
novae-angliae, and Alnus tenuzfolia. The apparent digestibility ranged from 0.23
on Salir bebbiana to 0.40 on Populus tremuloides. On the other hand, the
efficiency with which assimilated matter was converted to new larval tissue (ECD
= growth/assimilation) was relatively insensitive to host species, averaging about
0.40 on eight of nine hosts, but dropping to 0.30 on Salix glauca. Relative growth
rate (the product of consumption rate, apparent digestibility, and ECD) ranged
from about 0.30 mg-mg'1 . d'1 on good hosts, to 0.13 mg- mg'1 - d'1 on nutritionally
inferior hosts. Thus, the time required for fifth instar larvae to double their mass
varied from 2.3 d to 5.3 (1 depending on host, but was independent of the insect
population.

Apparent digestibility and ECD are measures of efficiency based on dry
mass. Nitrogen use efficiency is a physiological analog of apparent digestibility x
ECD, and may be of equal or greater ecological relevance (Mattson, 1980).
Nitrogen use efficiency differed markedly between hosts, but not between P.

canadensis populations (Figure 5): F835 = 7.10, P < 0.0001, host effect; F135 =

3.09, P = 0.09, population effect.

 

.m .wE E mm 362; “we: REESE 85.0956 $3 836.: mean

ectm .Qum can .bzfitmowfi E883 .88 coughing Lo 8:0on 65 fiasco
chC .253 BE 5390 223% “mo: :3me 2:: co 389:8 :meBE
new 82me :5: “@5358 .k 835 REL Lo oucmfieotoa 5390 "v PEEK

wwamnw hmOI <¥m<4< wmamnw thI «36613
2:5: 533 Sam. 333m 3:? 326m Emu. Stick
:2 auto Sm? so: 6:63 33 SE 2.3 Em: cc. tho 598 .6: 6:50 Gown SE 33 Em:

V
d
d
V
U
3
t m.

w m

3

W m .m a
w m
2 mm w 3
2 S
u
m
n
l
A

(Pp l Btu But)
BLVU HLMOHE)
(,_p 11511.1 611:)
ELVH NOIllenSNOO

 
 

‘2 fixmtmbfitflu .1 I
¥< “\mtmbwtmu .Q I

.0 o\omm

 

23

88.0 v m.

I.

me .6235 :32 u m2 6

 

 

 

No9 m SN mom Nwwm Mmtm ow Lotm
mmd m.:. .mwd $3 34 v3 omd mm: wed HNH w “mom x no;
a; oovm 593: mowbm :mde 5.2m :oWQ 89:. 299.2 Hwov w mo:
wmd 3m cod 0 cm." m: NwN $on 9.6 NS — coca—smog
k «m2 k «m2 k «9.4 k «m2 k umZ LO 850w
23* Eznzmoma 8mm Bum
Gum cocazEfim< EoEaQ< coszsmcoU 539.0

 

.352? so; 52ng

2:: :o @558 newEE—z Em «in? E0: maneuvers .& Co motowcocoigoE ESE 5m MEEQEOO m:=3._.<>OZ< 6 63m...

 

24

I P. canadensis AK
E] P. canadensis MI

0.2

(rem ba/s a/ax bebb glauc nov resin crisp ten

+SE

 

NITROGEN USE EFFICIENCY
(mg/mg)
O
0')

 

 

Populus Sa/ix Betula Alnus
ALASKA HOST SPECIES

Figure 5: Nitrogen use efficiency of final instar P. canadensis from Alaska and
Michigan compared on nine Alaskan host species. Error bars indicate 1 SE. Host
species as in Fig. 3.

25
Pnpal mass, larval duration, and survival

By extending the duration of growth, Michigan larvae produced larger
pupae than Alaskan larvae (Figures 6-7, Table 5), even though Alaskan larvae
began their life larger, and tended to grow at the same relative growth rate.
Female pupal mass (: SE) averaged 688 t 9 versus 750 t 9 mg, and female
larval duration averaged 28.0 t 0.3 versus 33.0 i 0.3 d, for Alaska and Michigan
respectively. On all but the worst hosts, larvae growing for a longer time
produced larger pupae; Michigan larvae tended to feed and grow for a longer
time than Alaskan larvae, but the relationship between duration of growth and
pupal mass did not differ between populations (Figure 7).

Pupal mass and larval duration were also strongly affected by host species
(Figure 6, Table 5). On hosts where they grew slowly, larvae extended the
duration of growth but still produced smaller pupae than on high quality hosts.
There were interactions between host species and insect population in that
population differences in pupal mass were least on low quality hosts, and
population differences in larval duration were greatest on low quality hosts.

Larval survival was strongly affected by host species but was very similar
between populations (Table 6). Survival to pupation ranged from 30% on Salix
glauca to about 80% on Populus balsamifera, Populus tremuloides, Betula resim'fera,
and Alnus tenuifolia. Patterns of survival also differed; on Alnus crispa, virtually
all of the mortality occurred during the first instar, while on Salix novae-angliae,
there was some mortality at almost every stage. This suggests acute toxicity as
Opposed to chronic malnutrition. Survival was generally correlated with other

measures of growth performance, but with some exceptions. Betula resinifera had

26

Table 5: ANOVA results comparing the pupal mass and larval period of P.
canadensis from Alaska and Michigan reared on nine Alaskan host species.

 

 

 

Female Female
Pupal Mass Larval Period
Source DF MS - 10'1 F MS ~ 102 F
Population 1 10570 2254'" 69531 122.59'"
Host 8 10286 2194'" 27979 4933'“
Pop x Host 8 12670 2.70" 1255 2.21‘
Error 98 469 567

 

'P < 0.05; " P < 0.01; P < 0.0001

 

27

I P. canadensis AK
III P. canadensis MI

95% CI

PUPAL MASS (mg)

 

 

 

 

 

50
E
5 40
r.
E
D 30 T
Q
i
> 20
a:
<
.1

10

trem ba/s a/ax bebb glauc nov resin crisp ten

 

 

Populus Sa/ix Betu/a Alnus

ALASKA HOST SPECIES

Figure 6: Pupal mass and larval duration of P. canadensis from Alaska and
Michigan reared on nine Alaskan host species. Error bars indIcate 95%
confidence intervals. Host species as in Fig. 3.

 

28

 

0 P. canadensis AK
0 P. canadensis MI

 

 

 

 

 

1100-
‘ O
A 1ooo~
U) ‘ O
E
m _: O O
m 900. o
< . o
E
.1 L 0 O
< 800‘ O o . o
S . -
O
D. . O 000 O
700- 0 O o
. O 0
j 0 o Alnus tenuifolia
600 - t , , . j

20 22 24 26 2'8 ' 30 32
LARVAL DURATION (d)

Figure 7: Pupal mass as a function of larval duration in P. canadensis from
Alaska and Michigan reared on Alnus tenuifolia. Each point represents one pupa.

 

29

Table 6: Cumulative survival of P. canadensis larvae from Alaska and Michigan reared
on nine Alaskan host species. Each host began with 19 or 20 neonate larvae from each
population. Pupal survival only includes the first half of winter diapause.

 

Percent of intial insects surviving
to the stadium indicated

 

Insect
Tree species Population 1 2 3 4 5 Pupae
Populus tremuloides AK 80 80 80 80 80 80
MI 80 75 75 75 75 75
Populus balsamifcra AK 90 85 85 85 75 70

MI 95 90 90 90 90 9O

Salix alaxensis AK 79 79 79 74 68 68
MI 80 75 75 70 60 60
Salix bebbiana AK 75 75 75 70 50 30

M I 70 65 65 60 55 45
Salix glauca AK 55 45 45 45 40 40
MI 55 50 50 35 20 20
Salix novae-angliae AK 90 70 65 65 60 55
MI 90 68 63 63 63 58
Betula resinifera AK 95 84 84 84 79 79
MI 85 85 85 85 85 80
Alnus cnspa AK 58 58 58 58 58 53
M I 70 70 65 65 65 65
Alnus tenuifolia AK 95 95 85 85 85 85

MI 90 85 85 80 75 70

 

 

30
very high survival even though larval growth rates and pupal mass were below
average. Conversely, Salix novae-angliae allowed generally high growth rates and

produced large pupae, even though survival was less than 60%.

Discussion

Evaluation of hypotheses

There was no evidence for population differences in the ability of P.
canadensis larvae to ingest, detoxify, assimilate, or convert the leaf tissue of
various tree species that they encounter throughout their broad geographic
distribution. Populations from Alaska were no better able to use Alaskan hosts,
and no worse able to use Michigan hosts, than populations from Michigan
(absence of population x host community interaction, Table 1, Figure 1).
Similarly, there was no evidence of host specialists among families within
populations (absence of family x host species interaction, Table 1). Instead, all
genotypes appeared equally capable of using all hosts, thereby refuting the
genetic mosaic hypothesis both at the inter- and intra-population level. Among
500+ species in the Papilionidae, P. canadensis is extraordinary in its ability to
exploit so many taxonomically and phytochemically diverse hosts (Scriber, 1984).
The explanation for these polyphagous abilities lies not within the population
structure of the species, but within the genome of any individual within the
species (super-genotype hypothesis).

Apparently, future evolutionary change in the nutritional physiology of P.
canadensis will be constrained by the appearance of novel variation. Thus I

would not expect P. canadensis to respond rapidly to climate-induced changes in

31

the plant community, nor would I expect that other P. canadensis populations, in
other regions of the boreal forest with other communities of hosts, differ in their
host-use abilities. The detection of variation is partly a function of sampling
intensity, but my conclusions seem robust. Population comparisons were based
on nearly 600 larvae representing over 30 families. Given the observed variance,
I was likely to detect population differences in growth rate of > 15% during the
first instar, >3% during the middle instars, and >10% during the final instar.
Furthermore, the same experimental design readily detected interspecific
differences between P. glaucus and P. canadensis (Table 2, Figure 2). Populations
of P. canadensis at the northern and southern edges of the boreal forest have
been associated with different tree communities for most of the Holocene (Webb
et al., 1983). If there were meaningful intraspecific variation in the nutritional
physiology of P. canadensis, even if it were undetectable within populations, I
would have expected measurable divergence between these distant populations.

Although host-use abilities appear to be invariant, there is substantial
variation in other attributes of P. canadensis. Alaskan females lay eggs that are
1.5 times larger than those produced by Michigan females of the same size (which
benefits larval development, Figure 3, Table 2), but only produce 60% as many
(Lederhouse and Ayres, unpub.). Alaskan caterpillars initiate pupation at a
smaller size than their Michigan counterparts; hence both larval duration and
pupal mass are less in the Alaskan population (Figures 6-7, Table 5). Alaskan
caterpillars are capable of more rapid growth at low temperatures (e.g., doubling

time of 6 d versus 9 d for 5th instar larvae at 12°C), and require less time to molt

at low temperatures (6 (1 versus 8 d for the penultimate molt at 12°, and 12 d

 

32
versus 18 d for the final molt) (Chapter 2). Each of these differences is an
apparent response to the selective pressures imposed by short cool Alaskan
summers. P. canadensis populations have diverged in their temperature responses
and life history strategies yet have remained surprisingly invariant in their

nutritional physiology.

Explanation for invariance of nutritional physiology

Endler (1977) considered clinal differentiation as a balance between the
homogenizing effects of gene flow and the diversifying effects of selection
gradients. The biology of P. canadensis seems to match Endler’s gradient
selection model in that suitable habitat is more or less continuous from Michigan
to Alaska (ca. 4000 km), and the dispersal capabilities of individual butterflies are
small with respect to the species distribution. In Endler’s terms, the lack of
differentiation in P. canadensis could be due to high gene flow or shallow
selection gradients.

We lack detailed information on gene flow in P. canadensis, but based on
adult longevity in the laboratory (Lederhouse et al., 1990), behavioral studies of
the closely related P. giaucus (Lederhouse, 1982; Berger, 1986), and my own field
observations, it appears that most mating and oviposition takes place within 20
km of the site of larval development, and I doubt that much occurs beyond 60
km. The vagility of P. canadensis is certainly greater than such insect herbivores
as Euphydryas butterflies (Ehrlich et al., 1975), Alsophila moths (Futuyma et al.,
1984), and Nuculaspis scales (Alstad and Corbin, 1990), but probably less than

others such as Danaus butterflies (Eanes and Koehn, 1978; Malcolm and Brower,

 

 

 

33
1989) and Schistocerca locusts (Chapman, 1990). In any case, it is obvious that
gene flow is not so extensive to preclude differentiation, because these
populations have diverged markedly in other traits (Chapter 2).

Weak selection gradients also seem an unlikely explanation for the lack of
divergence. Host use patterns differ between the populations (Scriber and Ayres,
1990, unpublished), and it is clear that the plant species differ in the secondary
chemistry of their foliage (Palo, 1984). Leaves of anus serotina contain the
cyanogenic glycoside prunasin (Horsley and Meinwold, 1981; Reilly et al., 1986).
Alnus crispa contains the stilbene pinosylvin; A. cn'spa and A. tenuzfolia contain
the triterpene a-amyrin, and A. rugosa contains what may be the shikimate
glycoside oregonin (Bryant, 1981; Clausen et al, 1986; Ayres and Reichardt,
unpublished). Fraxinus contains an assortment Of phenolic acids and flavonoids
(Kowalczyk and Olechnowicz, 1988). Tilia contains the triterpene B-amyrin, plus
a number of flavonoids (Hickok and Anway, 1972; Guelz et al., 1988). Populus
tremuloides and P. grandidentata contain the phenolic glycosides tremulacin and
salicortin (Lindroth et al., 1987); P. balsamifera buds and leaves contain a
complex mixture of compounds including cineol and benzyl alcohol (Reichardt et
al., 1990, pers. comm.). Betula papynfera and B. alleghaniensis differ in their
qualitative composition of flavonoids (Pawlowska, 1983). Salix novae-angliae
contains high concentrations of the flavonoid kaempferol-3-arabinsyl-7-rhamnoside
(T. P. Clausen, pers. comm.) Salix alaxensis, S. bebbiana, S. glauca, S.
novae-angliae, Betula alleghaniensis, B. papynfera, and B. resinifera, all produce

high concentrations of condensed tannins that are structurally distinguishable

(percent procyanidin, stereochemistry of C4 relative to C3, and the number of

 

34
flavonoid subunits per molecule) (Ayres et al., in prep.) Even under favorable
laboratory conditions, larval mortality of P. canadensis was substantial on most
hosts (Table 6), and larval growth rates (neonate to prepupa) were only half that
of specialist relatives in the Papilio troilus group (Nitao et al., 1991; Lederhouse
et al, 1992). Natural selection should favor genotypes that survive better and
grow faster, and it seems that the optimum physiology should differ across this

diverse spectrum of hosts.

Implications for herbivore evolution

I suggest that the nutritional physiology of P. canadensis, Epim'ta
autumnata (Ayres et al., 1987), and perhaps other insect herbivores, has been
canalized such that mutations and recombination seldom lead to phenotypic
variation. Thus natural selection may have little on which to operate. (Endler’s
single locus, two allele, model assumes the pre-existence of variation.) Under this
canalization scenario, substantive evolutionary change is predicted to involve the
occasional breakdown and successful rearrangement of gene interaction systems
(Wright, 1977, pp. 443-473), which would be unlikely except in small populations,
and may be commonly associated with Speciation events. This model reconciles
the apparent stasis of nutritional physiology in P. canadensis with the marked
discontinuity between P. canadensis and its putative ancestor P. glaucus (compare
Figures 1 and 2, Tables 1 and 2). This is not a new idea. It is consistent with
models of stepwise evolution (Ehrlich and Raven, 1964) and punctuated equilibria
(Eldredge and Gould, 1972), and is linked with concepts of "metastable

equilibria" (Haldane, 1931), "coadapted genes" (Dobzhansky, 1955), and "the unity

35
of the genotype" (Mayr, 1963, pp. 263-298).

This view of herbivore evolution emphasizes the importance of epistasis. I
propose that larval growth performance is more complex than the additive effects
of multiple gene products. The role of gene interactions in detoxification
physiology could be tested by experimentally inactivating enzyme systems singly
and in combination and measuring the response of the insect (e.g., Hedin et al.,
1988; Lindroth, 1989a); statistical interactions will indicate epistasis. If nutritional
physiology has been canalized, the composite process of growth should be
buffered from genetic variation in the underlying components. Thus, genetic
variation in relative growth rate is predicted to be less than variation in mixed
function oxidase activity, esterase activity, proteinase activity, permeability of the
peritrophic membrane, pH Of the gut lumen, rate Of gut passage, or the number
and type of carrier proteins in the gut wall. Homeostatic properties relevant to
insect growth may arise at many levels, including gene duplication (8 esterase
isozymes have been electrophoretically resolved in P. canadensis, K. S. Johnson,
unpublished data), Mendelian dominance (Sved and Mayo, 1970), regulation of
gene expression (Lindroth, 1989b), kinetics of biochemical pathways (Kacser and
Burns, 1981), and neuroendocrine control (Bernays and Simpson, 1982).

Broad patterns of nutritional adaptation in herbivorous insects may not be
a simple extension of the microevolutionary processes occurring within most
populations; i.e., intraspecific patterns may be poor predictors of interspecific
patterns. Intraspecific variation provided no hint of the extensive divergence
between P. canadensis and P. glaucus, nor of the strong negative correlations

(apparent tradeoffs) between their growth performance on different hosts.

36
Studies in the Papilio troilus species group have similarly indicated that
intraspecific variation is of a completely different scale and pattern than variation
between closely related species (Nitao et al., 1991; Lederhouse et al., 1992).
Negative genetic correlations in insect growth performance on different hosts are
predicted by prevalent explanations for the evolution of specialization (Levins and
MacArthur, 1969; Futuyma and Moreno, 1988), but recent reviews of intraspecific
studies have noted that such negative correlations are surprisingly rare (Jaenike,
1990; Via, 1990). This is less surprising if intraspecific variation is so low as to be
of modest ecological importance (e.g. Via, 1984b; Bowers, 1986; James et al.,
1988; Nitao et al., 1991). Fair tests of correlation require meaningful variation,
even if it requires the complications of interspecies comparisons. Available data
may even overestimate the importance of intraspecific tradeoffs in host-use
abilities, because most published studies involve systems where there were a prion'
expectations of low gene flow and high intraspecific variation.

Traits that are canalized may still evolve in predictable patterns according
to general principles, but describing the patterns and testing the principles will
require the tools of comparative biology in addition to quantitative genetics
(Ehrlich and Raven, 1964; Stearns, 1980; Cheverud et al., 1985; Huey, 1987; Page]
and Harvey, 1988; Wanntorp et al., 1990). Further studies that compare
variation within and between closely related species will test my suggestion that
nutritional physiology is typically canalized and evolves during uncommon

rearrangements of the genome.

CHAPTER TWO

Local adaptation to regional climates

inBaDiliecanadensjs

Introduction

Why do insect herbivores occur where they do, but not elsewhere?
Herbivore distributions are commonly less extensive than that of their host plants
(MacLean 1983, McClure 1989), implying that they are not a simple function of
host plant distributions. Interspecific competition may occur among insect
herbivores (Fritz et al. 1986, Karban 1986), but competitive exclusion is probably
rare because food resources are not usually limiting (Lawton and Strong 1981,
Fritz and Price 1990). Here I explore the role of abiotic constraints, specifically
summer temperatures, in determining the geographic range of the northern tiger
swallowtail, Papilio canadensis R & J (Lepidoptera: Papilionidae).

P. canadensis is a polyphagous tree-feeding insect that occurs throughout
much of the boreal forests of North America, from New England and upstate
New York, west and north through the Great Lakes states and Canada to interior
Alaska (Scriber 1988). Preliminary information suggested an important role of
temperature in the ecology of P. canadensis. Larval growth rates are strongly
temperature sensitive in swallowtails (Scriber and Lederhouse 1983, Ritland and

Scriber 1985), as in most insects (Scriber and Slansky 1981, Taylor 1981). Also,

37

38
the northern limit of Papilio glaucus L., a southern parapatric sister species
(Hagen et a1. 1991), closely corresponds to the 1400 degree day isotherm (10°C
base), which approximates the thermal sum required for completion of two
generations (Scriber 1988). P. canadensis is obligatorily univoltine while P.
glaucus is typically multivoltine (Hagen and Scriber 1989); if the biology of these
two taxa were otherwise identical, P. canadensis would require about 700 degree
days to complete its development.

As a first step in assessing the importance Of summer temperatures for P.
canadensis, I analyzed the climates encountered by populations near the southern
and northern limits of their distribution. The thermal sums accumulated at
Chatham, Michigan (86° 50’ W, 46° 25’ N) ranged from 760 to 1254 degree days
over 50 years, with an average of 985 degree days (Figure 8). In contrast, the
warmest summer in Fairbanks, Alaska (147° 30’ W, 64° 50’ N) accumulated fewer
degree days than the coldest year in Chatham: ranging from 453 to 747 degree
days over 48 years, with an average of 583 degree days (Figure 8). If we take 700
degree days as the minimum for completion of one generation, the upper
peninsula of Michigan always accumulates sufficient degree days for completion
of one generation, but never two, while only 5 years out of 48 would have
allowed even a single generation in interior Alaska. In fact, P. canadensis
butterflies were nectaring outside my Fairbanks laboratory as I completed these
analyses.

Summers in interior Alaska are relatively cool as well as short. The
average daily mean temperature during the period of P. canadensis larval

development was 144°C compared to 188° in northern Michigan (Figure 9). In

39

 

 

 

 

10‘
s—
8 . FAIRBANKS, AK
ii II
t o I ' ' r fl
>_ .
O .
z .
DJ
8 4
u.I 10:
E
LL
CHATHAM, MI
5—t
o...,...,,,l_lfi
200 600 1000 1400

THERMAL SUM (10° BASE)

Figure 8: Frequency distribution of thermal sums accumulated above a 10°C base
in Fairbanks, Alaska (48 years) and Chatham, Michigan (50 years). Michigan
weather records included 1931-1980; Alaska records included 1936-1980
(excluding 1969) and 1987-1990. Annual accumulations were begun on 1 April
and terminated near the time of leaf fall ( 15 September in Alaska and 1 October
in Michigan). Thermal sums were calculated from daily maxima and minima
using a sine function recommended by Watanabe (1978).

40

14.4°

303 FAIRBANKS, AK

gLAAIAALA

 

 

 

, 18.8°
30{ CHATHAM, MI i

  

FREQUENCY (PERCENT)

 

 

0‘ ' I v I I T
0 6 12 18 24 30

DAILY MEAN TEMPERATURE

Figure 9: Distribution of daily mean temperatures encountered by P. canadensis
larvae in Fairbanks, Alaska (48 years) and Chatham, Michigan (50 years)
(weather records as in Fig. 1). The period of P. canadensis larval development
was defined as 350-700 degree days in Michigan and 250-550 degree days in
Alaska (truncated at 15 September when necessary). The percentage of days in
each temperature class was calculated separately for each season, and the figure
shows the average percent across 48 or 50 years of records (otherwise the
distribution would be weighted towards long cold seasons).

41
Alaska, the daily mean temperature was 19.5° or less on most days (80%), while
in Michigan it was 21° or warmer on nearly half the days (43%).

This climatic comparison led me to turn the original question around and
ask how P. canadensis maintains as broad a distribution as it does. I suggest two
alternative hypotheses: (1) P. canadensis is a composite of regionally adapted
populations that collectively maintain a broader distribution than would otherwise
be possible, or (2) P. canadensis is a single cohesive genotype that is surprisingly
robust in its tolerance of climate. The scenario of regional adaption (H1)
suggests that populations are continually being modified through natural selection
to conform to the local environment; i.e., the environment determines the
genotype and therefore climatic limits to distribution and abundance may be
nonexistent or transient. This contrasts with the notion of a cohesive genotype
(H2), which emphasizes genetic constraints, and argues instead that the genotype
determines the environment where the organism occurs.

These hypotheses were evaluated by comparing the developmental
physiology of P. canadensis from northern Michigan and interior Alaska across a

range of temperatures in the laboratory and in the field.

Methods

Overview

P. canadensis populations were compared with respect to fourth instar
growth rates (RGR4), fifth instar growth rates (RGRS), the duration of the fourth
molt (Molt4, fourth instar to fifth instar), the duration Of the fifth molt (Molts,

fifth instar to pupa), consumption rate (RCR), assimilatiOn rate (RAR), average

42
daily metabolic rate (ADMR), apparent digestibility (AD), the efficiency of
conversion of digested matter (ECD), nitrogen use efficiency (NUE), pupal mass,
and survival to imagos. Studies were conducted during two seasons and included
three host species: the temperature response of Alaskan and Michigan larvae
were compared on Populus tremuloidcs Michx. (quaking aspen) in 1988 and 1989,
and on Populus baLramzfera L. (balsam poplar) in 1988 (both hosts are used
naturally by both populations); 1988 experiments also included the temperature
response of Alaskan larvae feeding on Betula resinifera Britton (Alaska paper
birch). These experiments were conducted under controlled laboratory
temperatures using freshly detached leaves. In addition, I compared the ability of
the two populations to elevate larval body temperatures through basking behavior
under field conditions, and compared the growth rates of larvae feeding naturally
in the field with growth rates of larvae at an equivalent experimental

temperature, feeding on detached leaves in the laboratory.

Population sampling

Female butterflies were collected during early June 1988 and 1989 from
sites in interior Alaska (vicinity of Fairbanks) and northern Michigan/Wisconsin
(Goegebic and Iron counties in MI; Vilas, Taylor, and Chippewa counties in WI),
then transported to the laboratory, fed a 20% honey solution daily, and allowed
to oviposit on sprigs of quaking aspen foliage (Populus tremuloides). Resulting
larvae from the two populations were reared simultaneously and fed the same
foliage (same trees, same leaf samples) until early in the fourth instar when

growth trials began. Larvae destined for experiments with different host species

43
were reared on those hosts from the time of egg hatch. Thus, maternal parents,
eggs, and larvae of both populations were handled with the same protocol, in the
same laboratory, at the same time. This should have precluded any spurious
differences between populations due to differing environments.

Experiments during 1988 included 9 Alaskan families (total of 108 larvae)
and 6 Michigan families (96 larvae); 1989 experiments included 10 Alaskan
families (76 larvae) and 10 Michigan families (55 larvae). Larvae from each
family were allocated evenly across temperatures and hosts, but it was not
possible to maintain equal sample sizes of each family in each treatment, and

mortality resulted in some families being unrepresented in some treatments.

Larval growth during the fourth and fifth instars

Early fourth instar larvae (recently molted but feeding, ca. 100 mg fresh
mass) were weighed (Wk), offered previously weighed leaves, and allocated to
one of four experimental temperatures (12, 18, 24, or 30°C; photoperiod of L:D
18:6). Larvae were individually confined to clear plastic vials (12 ounce cups)
with water-saturated plaster-of-paris bases that provided a high humidity
atmosphere and maintained leaf turgor near natural levels. Paired control leaves
(one sample per larva) were placed in adjacent vials, and exposed to the same
conditions as the foliage fed to the larvae; the ratio of dry mass to wet mass of
control leaves allowed me to estimate the initial dry mass of the foliage fed to
the larvae. When larvae began to approach the fifth molt (ca. 380 mg; after 1.5 d

at 30°, 4-5 (I at 12°), they were reweighed (W0), given fresh foliage, and returned

to the same temperature. The unconsumed leaf tissue and the accumulated frass

 

44
were dried and weighed. The dry mass of experimental larvae was estimated
from their fresh mass as DM = 0.125 - FM. (The dryzwet ratio of P. canadensis
larvae did not differ between populations and was unaffected by host plant; N =
20)

Fourth instar relative growth rates were calculated as:

RGR. = tlntW.,) -1n(W.-)I / 71...:

where ln = natural logarithm, and T4”, = the elapsed time in days. Relative
consumption rate (RCR) and relative assimilation (RAR) were calculated
following formulae in Gordon (1968), which are similar to those of Waldbauer
(1968) (produced values within :5%), but better describe a system of exponential
growth. Apparent digestibility (AD) = RAR/RCR. The efficiency of conversion
of digested matter (ECD) = RGR/RAR. Average daily metabolic rate (ADMR)
was calculated as RAR - RGR. All are expressed in units of dry mass.

When larvae had completed the fourth molt and begun feeding in the fifth
instar (ca. 400 mg), they were reweighed (W5), and again offered fresh, weighed
foliage. Fifth instar trials lasted the majority of the fifth instar (1.5 to 5 (1
depending on temperature), and produced similarly derived estimates of all
parameters measured in the fourth instar. In addition, a subset of the control
leaves and frass samples was analyzed for total nitrogen content with standard
micro-Kjeldahl techniques (sulfuric acid digestion followed by colorimetric
analysis with a Technicon Auto-analyzer II; Herms, 1991). Nitrogen use
efficiencies (NUE) were calculated as the fraction of ingested nitrogen that did

not appear in the frass: NUE = (Ningested - NcgcdeNingcmd.

45
In 1989, larvae were returned to their experimental temperatures after the
fifth instar growth measurements, fed fresh foliage, and monitored daily (12 and
18°) or twice daily (24 and 30°) until they completed the fifth molt and entered
the diapausing pupal stage. Four weeks after pupation, pupae were sexed,
weighed, and stored at 2°C in darkness through the winter. The following spring
they were placed outdoors (Fairbanks) in screen emergence traps until the adults

emerged or the pupae died.

The duration of molts
Because the process of molting between instars is not entirely discrete

from that of growth within instars, its contribution to total development time is
difficult to measure by simple counting the number of days or hours that an
animal appears to be molting (Ayres and MacLean 1987). I estimated the
duration of the fourth molt by calculating, based on fourth and fifth instar growth
rates, how long it would have taken a larva to grow from its early fourth instar
mass to its late fifth instar mass in the absence of molt (T, + T5), and subtracting
this from how long it actually took (Th-5,). The difference was defined to be the
duration of the fourth molt (Molt4). Calculations follow Ayres and MacLean
(1987).

Molt4 = T4,._5f- (T4 + T5).

T4 = [ln(M4max) - ln(M4,-)]/RGR4.

T5 = [19(M5f) ' ln(M5min)]/RGR5'

Msmt‘n = M

4max

- EXUV - RESP.

46
The mass of fifth instar larvae when they have first filled their gut after ecdysis
(Wm) equals the maximum mass of fourth instar larvae (M m) minus exuvial
losses (EXUV) and respiratory expenses during molt (RESP). For P. canadensis,
EXUV was estimated as 2.5 mg and RESP as Mm - 0.05. EXUV and RESP are
difficult to estimate precisely, but halving or doubling them has little effect on
Molt4 (Ayres and MacLean 1987).

For simplicity I refer to the "duration of molt", but because of temporal
overlap between the processes of growth and molt, Molt4 is somewhat less than
the time between the hormonally mediated initiation and completion of molt.
Molt4 is also somewhat longer than the time when a molting larva is obviously
immobile and nonfeeding, because growth slows in preparation for apolysis, and
only gradually resumes after ecdysis.

The duration of the fifth and final molt (Molts) was calculated similar to
Molt4 except was only based on growth rate in the preceding (fifth) instar because
growth ceases after pupation.

upa

The dry mass of pupae (DMp ) were estimated from pupal fresh masses as -18.4

“P“
+ 0.243 ~FMpupa for males and -56.9 + 0.316-FMpupa for females (Ayres,
unpublished data). The dry mass of early fifth instar larvae (DMSi) was calculated
as before (FM - 0.125) with a correction for indigestible gut contents (-15% fresh
mass). As an alternative to this estimate of Molts, I also measured the time

elapsed from when late fifth instar larvae cleared their guts and the completion of

ecdysis.

47
Basking success

During August 1989, I compared the ability of P. canadensis larvae from
Alaska and Michigan to elevate their body temperatures with solar radiation.
Fifth instar larvae from Alaska and Michigan (71 = 18 and 11, respectively) were
placed outdoors on quaking aspen trees where they were allowed to move, feed,
and bask naturally. Temperatures were recorded to the nearest 0.1°C using a
Sensortek BAT-12 telethermometer with a blunt tip, fine gauge
copper/constantan sensor (Sensortek model MT-4, needle diameter 0.33 mm,
time constant 0.025 sec). Caterpillar temperatures were measured non-invasively
by sliding the sensor tip into a natural fold dorso-lateral to the first pair of legs.
With each caterpillar temperature, I also measured air and leaf surface
temperatures, as well as the temperature of a standardized "black body" exposed
to the same insolation as the caterpillar. My black body was a black,
caterpillar-sized piece of open cell foam rubber (30 x 10 x 10 mm), which I
suspended next to each caterpillar until its temperature stabilized under the
ambient radiation (1-4 minutes). Black body temperatures provided a measure of
basking potential. All temperatures were recorded with the sensor tip shaded.

I measured the relative growth rates of larvae in the field (for comparison
with laboratory growth rates) by weighing the Alaskan P. canadensis prior to
being placed outdoors, and 3 (I later. The ambient daily mean temperature
encountered by field larvae was measured with a shaded maximum-minimum

thermometer. A matched set of P. canadensis larvae was weighed one day later

and grown for 3 d in the laboratory at a constant temperature set each day to the

 

48
mean temperature encountered the previous day by the field caterpillars.

Laboratory larvae were fed leaves collected from the same trees.

Statistical analyses

Statistical comparisons of P. canadensis populations from Alaska and
Michigan were based on the number of families representing each population and
the variance among families. Except as noted, figures and tables show population
means and standard errors calculated from family means. RGR,” RGRS, and
Molt4, were analyzed separately at each temperature (due to extreme
heteroscedasticity among temperatures) with an ANOVA model that included
population (AK versus MI) and host (Populus tremuloides in 1988, Populus
balsamifera in 1988, and Populus tremuloides in 1989) as fixed effects, and families
nested within population as a random effect. Molt5 (only measured in 1989) was
analyzed with an identical ANOVA model except that temperature was
substituted for host. Nitrogen use efficiency (incomplete factorial) was analyzed
with two two-way ANOVAS, one comparing two populations across four
temperatures on a single host (main effects = population and temperature), and
one comparing a single population (AK) across three hosts and four temperatures
(main effects = host and temperature). Pupal mass was analyzed with a
three-way ANOVA that included temperature, population, and sex. Family was
not explicitly included in analyses of nitrogen use efficiency and pupal mass
because of limited sample sizes and because variance among families contributed
only trivially to the other models. Molts was square root transformed to correct

for heteroscedasticity (figure shows non-transformed data); all other variables

49
satisfied assumptions of normality and equal variance. ANOVAs were calculated
using the SAS General Linear Models procedure (SAS 1985). The effects of
population and temperature on survival were tested with a two-way contingency

analysis (CATMOD procedure, linear model, SAS 1985).

Results

Larval growth rates

As expected, temperature had an enormous effect on larval growth rate,
with higher temperatures generally resulting in higher growth rates (Figure 10).
In addition there were substantial differences between populations, and
interactions among the effects of temperature, population, and host (Table 7). At
low temperatures (12°C), Alaskan P. canadensis had consistently higher fourth and
fifth instar growth rates than Michigan P. canadensis (Figure 10): least square
means t SE = 0.152 1 0.008 versus 0.114 t 0.007 mg-mg‘lod'1 for fourth instars
and 0.119 x 0.008 versus 0.076 _+. 0.010 mg-mg'l~d'1 for fifth instars (F,23 =
14.10 and F1,19 = 11.69, Table 7). Thus Alaskan fifth instars at 12° were doubling
their mass in 5.8 (1 versus 9.1 d for their Michigan counterparts (doubling time =
ln(2)/RGR). At 18°, Alaskan growth rates were higher in six of six comparisons
(Figure 10), significantly so for fifth instars (Table 7). At high temperatures (24
and 30°), population comparisons differed depending upon the host (Figure 10,
Population x Host interactions in Table 7). In 1988, growth rates of Alaskan
fourth instars feeding on Populus tremuloides declined from 24 to 30°C, while the
growth rate of Michigan fourth instars increased across the full range of

experimental temperatures (Figure 10, upper left); consequently, Michigan larvae

50

   

 

 

   

 

 

 

 

 

FOURTH INSTARS FIFTH INSTARS
0.6 '
0.4
0.2
P. bals. 1988 P. bals. 1988
0-0 I E I f ' I I 7. I
:9 0.6
'0:
5
a)
g 0.4
L“.
<
I: 0.2
I
E P. trem. 1988 a P. trem. 1988
g 0.0 r . I I ‘ ' I I I T
(5
0.6
0.4
0.2
P. (rem. 1989 o P. trem. 1989
0.0 I T T T T I I I
1 2 1 8 24 30 1 2 1 8 24 30

TEMPERATURE (C°)

Figure 10: Relative grth rates of P. canadensis larvae from Alaska (solid lines)
and Michigan (dashed lines) as a function of temperature. Experiments included
fourth and fifth instars reared on Populus balsamifera and P. tremuloides in 1988
and P. tremuloides in 1989. Some standard errors are obscured by the data point.
See Table 7 for corresponding ANOVAS.

51

886 v m . ”8.0 v m. Mmod v m.

.913 3:33 833.; we moouwov Gotm .n

LotomE 5.5 332 263 £52 350 A.aox_:__E£m2 we? coca—dmom e8 8358966 88 K of. .u

 

 

 

 

 

 

 

we; .wQ a: :3 Re Re Ea who 8-: REVERE

.34 ...wo.2 of :28 :.o 2d 8m 8d N 28m A dog

zmvw :02 be. 68 a: .wE SA SA N .8:

$3. as 85 o: 58.8 3; :3: 32.: H 8:288

JOE Joe Joe Joe Joe Joe Joe Joe 6:6 886m
.3. .3 4: .fl

 

«83%; k

 

.2 .3...
E «6% 2 683260 82382.82 50.. E 38: 3:: no sawing—z «Ea 38?. EB. announce“. .k 335
A m3: EE 28 355 £56. .o 8:: .2an 62.22 2: wages m<>oz< Co E8885 ”A. Sea.

52
were growing significantly faster than Alaskan larvae at 30° (mean t SE = 0.533
t 0.043 versus 0.279 t 0.054 mg-mg‘l-d'l, P = 0.0009). Of the remaining five
comparisons at 30°, the populations grew at similar rates in two comparisons, and
Alaskan larvae tended to grow faster in three comparisons (significantly so in

both 1989 comparisons, P < 0.001; Figure 10).

Dry mass and nitrogen budgets

Population comparisons of dry mass consumption rates and conversion
efficiencies were summarized through analysis of two most disparate trials: fourth
instars on Populus tremuloides in 1988, and fifth instars on Populus tremuloides in
1989 (upper left and lower right in Figure 10). At 12° in 1988 (Figure 11), the
high growth rates of Alaskan larvae relative to Michigan larvae was attributable
to 44% higher consumption rates (RCR : = 0.85 i 0.07 versus 0.59 _+. 0.03
mg-mg‘1 - d'l, P < 0.05); neither apparent digestibility (AD) nor the efficiency of
conversion of digested matter (ECD) differed between populations at 12° (note
that RGR = RCR - AD - ECD). At 30°, the reduced growth of Alaskan larvae
relative to Michigan larvae was due to elevated respiratory expenses leading to a
42% reduction in the efficiency with which digested matter was converted into
new larval tissue (ECD t SE = 0.28 i 0.05 versus 0.48 :t 0.06 mg/mg for AK
and MI respectively, P < 0.05); in this trial, neither consumption rate nor
apparent digestibility differed between populations at 30° (Figure 11).

“In trials where the growth rate of Alaskan larvae remained higher than
Michigan larvae across all temperatures (e.g., fifth instars in 1989), it was due to

generally elevated consumption rates (average of 40% higher in Alaskan larvae,

EFFICIENCY (mg/mg)

RATE (mg- mg"- d")

53

.6
0 GROWTH M. CONSUMPTION
05 3.0

0A

20
03

02

 

1.0 .
iSE °'

 

0d

 

 

03 I I I I on
03
05
0A

03

 

 

 

0.2 I T I I 0.2 I T I I
12 18 24 30 12 18 24 30

TEMPERATURE TEMPERATURE

Figure 11: Consumption rate, apparent digestibility, and growth efficiency (ECD,
growth/assimilation) of P. canadensis larvae from Alaska (solid lines) and
Michigan (dashed lines) as a function of temperature. Corresponding growth
rates appear in middle left of Fig. 10 (fourth instars feeding on Populus
tremuloides in 1988). Some standard errors are obscured by the data point.

54
P < 0.001 at each temperature, Figure 12). Digestive efficiency (AD) did not
differ between populations so assimilation rates also averaged 40% higher in
Alaskan larvae (Figure 12). Although respiratory expenses (= average daily
metabolic rate, = assimilation rate - growth rate) averaged 35% higher in
Alaskan larvae, the higher assimilation rate was more than enough to
compensate, and larvae maintained higher growth rates at all temperatures
(Figure 12). Because population differences in metabolic rate balanced
population differences in assimilation rate, there were no population differences
in growth efficiency (= ECD = growth/assimilation) (Figure 12).

A powerful interaction between temperature and host quality was revealed
in comparing the temperature responses of Alaskan P. canadensis on three
different host species (Figure 13). On Populus balsamifera, fifth instar growth
rate increased linearly with temperature from 12 to 30° (010 = 2.16). On Populus
tremuloides, where consumption rates were indistinguishable from P. balsamifera,
growth rate increased from 12 to 24°, then remained constant from 24 to 30°.
The higher growth rates on P. balsamifera compared to P. tremuloides at 18 and
30°, were due to lower metabolic rates (ADMR :t SE at 30° = 0.57 3: 0.08 versus
0.91 i 0.03 mg-mg‘1 - d'l). Apparent digestibility, and therefore assimilation rate,
was equal or lower on P. balsamifera compared to P. tremuloides, so ECD, (=
(RAR - ADMR)/RAR), was higher (Figure 13). On Betula resinifera, growth rate
was wholly insensitive to temperature (Q10 = 1.16 from 12 to 30°), even though
consumption rate increased steadily with temperature (Q10 = 1.68) (Figure 13).
Consumption did not, however, increase as steeply with temperature as on the

other hosts (compare to RCR 010 = 2.21 on Populus balsamzfera). The apparent

 

55

   

 

 

 

   

 

 

 

 

 

 

1 CONSUMPTION RATE ' APPARENT DIGESTIBILITY
—. I 0.5
1? 2~
.'. 4 0)
g j g, 04
m 4
E 11 E
I 0.3
I
0 T T T T 0.2 T T T T
I ASSIMILATION RATE 0-5
1.2~
.. 4 0.5
I? 1
Ta 0.8" g,
E ‘ a 04
a 4
E 04‘ E
' 1 0.3
0.0 r r r T 0.2 r T f r
12 18 24 30
TEMPERATURE (C°)
T1?
'0:
E
U)
E .
0.0 1 1 1 1
12 18 24 30

TEMPERATURE (C°)

Figure 12: Consumption rate, assimilation rate, metabolic rate (ADMR), apparent
digestibility, and growth efficiency (ECD, growth/assimilation) of P. canadensis
larvae from Alaska (solid lines) and Michigan (dashed lines) as a function of
temperature. Corresponding growth rates appear in lower right of Fig. 10 (fifth
instars feeding on Populus tremuloides in 1989). Some standard errors are
obscured by the data point.

56

 

 

 

 

 

 

 

 

 

3 GROWTH CONSUMPTION
05
r 1 3°
.0 .
7' ,0-4‘. :85
m :
E 0.3:
°‘ :
E 0.2:
W I
2 0.11
m :
0.0
A 0.6
‘e”
\m 0.5
E
" 0.4
5
z 0.3
E
2 0.2
u.
u” 1
w 0.1 I Y I Y 0.1 I r T I
12 18 24 30 12 18 24 3o
TEMPERATURE TEMPERATURE

Figure 13: Growth rate, consumption rate, apparent digestibility, and growth
efficiency (ECD, growth/assimilation) of P. canadensis fifth instars feeding on
three hosts as a function of temperature. Hosts were Populus balsamifera (dotted
lines), P. tremuloides (solid lines), and Betula resinifera (dashed lines). Some
standard errors are obscured by the data point.

 

57
digestibility of Betula was low (least square mean t SE = 0.26 t 0.02) and
relatively insensitive to temperature (Figure 13). Consequently, assimilation rate
on Betula was low (RAR : SE at 30° = 0.34 i 0.06 mg'mg'l -d'1), and the
arithmetic difference between assimilation rate and metabolic rate (= growth -
rate), did not change from 12 to 30°C even with a 2.5 fold increase in
consumption rate.

There were no apparent differences in the nitrogen use efficiency of P.
canadensis populations from Alaska and Michigan (Table 8; NUE least square
means 1 SE = 0.67 t 0.02 versus 0.69 _+_ 0.02 mg/mg for Alaska and Michigan
respectively; 1‘71.28 = 1.34, P = 0.26). As with dry mass conversion efficiencies
(AD and ECD), nitrogen use efficiency was relatively insensitive to temperature
(Table 8; F3.44 = 1.94, P = 0.14). However, there were large effects of host
species (Table 8; F144 = 66.71, P < 0.0001). Nitrogen use efficiency may have
been somewhat less on Populus balsamifera than on P. tremuloides (least square
means 1 SE = 0.61 t 0.02 versus 0.67 t 0.02, P = 0.063), but it was much less
on Betula (0.36 t 0.02) than on either Populus species (P < 0.0001). Leaf
nitrogen content (% dry mass) was lower in Betula foliage (mean .t SE = 2.10 1'
0.13%) than in P. tremuloides (2.67 t 0.11%) or P. baLramifera (2.44 t 0.10%),
and larval dry mass consumption rates were also lower (Figure 13).
Consequently, nitrogen consumption rates (NCR) at 30° were 2.3 times higher on
P. balsamifera than on Betula (75.4 versus 33.0 mg-g'1 -d'1), and because of the
reduced nitrogen use efficiencies on Betula, nitrogen accumulation rate (=
NCR - NUE) at 30° was 4.1 times higher than on Betula (47.5 versus 11.6

mg - g" ‘ d").

 

58

 

A38
and

5.8
8d

9.98 A88
Ed Ed

:2 ¥<

 

0cm

 

 

25.8 $0.8

38 and

$0.8 3.8

SE Be

25.8 :59 A58 A38

:6 So So one

E O? .2 OZ
.2 a:

bcomoEm 3.3 :OwOEZ

 

 

A38

m3 at .m

A88

one 3 m
398 35.8
who Re .8. m
E V2 28:

.2

 

.Amomoficuuma E mmv 529.358 EoEEO: some E @0583 93 oatm—
mé u 2 .3652}? 35.5 was .EmbEOflg SSEOK 6833352. 32:33 Enos 3:: :O MEEEOQEB SO“
E @558 53:22 was 8.82 :8: 33.0323 .m 335 SE .O AmDZV 5:2qu om: :OwOEZ “m 23%

59

Molting physiology
The fourth molt (Molt4) required 14-50% more time for Michigan larvae

than for Alaskan larvae (Figure 14). Population differences were significant at

+

12° (least square means : SE = 8.13 .. 0.39 versus 6.58 t 0.38 d for Michigan
and Alaska respectively), at 18° (3.12 t 0.28 versus 2.08 .t 0.25 d), and at 30°
(1.18 t 0.09 versus 0.86 t 0.11 (1) (Table 9). Molt duration was highly sensitive
to temperature (required 7.2 times longer at 12° than at 30°, Figure 14), but was
relatively insensitive to host (no significant host effects or population x host
interactions in Table 9).

The fifth molt (Molts, Figure 15) required 50% longer for Michigan larvae
than for Alaskan larvae at 12° (P < 0.0025), 42% longer at 18° (P = 0.02), 32%
longer at 24° (P < 0.0025), and actually required slightly less time at 30° (P =
0.18) (population effect and temperature x population interaction in Table 10).
The duration of the fifth molt was very temperature sensitive between 12 and 18°
(010 of Molt5 = 5.38 and 5.92 for Alaska and Michigan respectively), temperature
insensitive from 18 to 24° (010 = 1.02 and 1.21), and moderately temperature
sensitive from 24 to 30° (Q10 = 1.70 and 2.92). The fifth molt required a
surprisingly long time (up to 18 d for Michigan larvae at 12°), but this did not
appear to be an artifact Of the way it was estimated. The interval from when late
fifth instars cleared their gut to when they completed ecdysis to enter the pupal
stage averaged 55% as long as MoltS (range of 41 to 69% in eight treatments).
This interval must be an underestimate of the contribution of the fifth molt to

total development time. In other experiments (at 24°), repeated weighings of

 

60

DURATION OF FOURTH MOLT

 

 

 

10‘
iSE
8-1
’07
> 6‘
(U
3
LU ..
E 4
p—
2-
O I r T I

TEMPERATURE (C°)

Figure 14: The duration of the fourth molt (Molt4, fourth instar to fifth instar) in
P. canadensis from Alaska (solid line) and Michigan (dashed line) as a function of
temperature. Experiments included larvae reared on Populus balsamzfera and P.
tremuloides in 1988 and P. tremuloides in 1989. Some standard errors are
obscured by the data point. See Table 9 for corresponding ANOVAs.

 

61

Table 9: Summary of ANOVAs comparing the duration of the fourth
molt (Molt4) in P. canadensis from Alaska and Michigan on three
hosts at {our temperatures. Corresponds to data in Fig. 14.

 

F statisticsa

 

 

Source dlb 12° 18° 24° 30°

Population 1 753' 7.81" 1.23 4.62.
Host 2 0.18 155 0.70 251
Pop. x Host 2 2.03 0.12 0.25 1.46

Family(Pop.) 19-25 0.35 0.53 0.96 1.51

 

3‘ The F test denominator for Population was MSFamin(Pop.); other
terms were tested over MScmr.

b' Error degrees of freedom equalled 10-37.

'P < 0.05; " P < 0.01; P < 0.0001

 

 

62

DURATION OF FIFTH MOLT

 

20-
. % :SE
15~
’0?
>
(U .
E 10-
LIJ .
g .
1— 1
5—1
0 I l r I

 

 

12 18 24 30

TEMPERATURE (C°)

Figure 15: The duration of the fifth molt (Molts, fifth instar to pupa) in P.
canadensis from Alaska (solid line) and Michigan (dashed line) as a function of
temperature. Larvae were reared on P. tremuloides in 1989. Some standard
errors are obscured by the data point. See Table 10 for corresponding ANOVA.

 

 

 

 

63

Table 10: ANOVA comparing the duration of the fifth molt in P.
canadensis larvae from Alaska and Michigan at four temperatures (square
root transformed data). Corresponds to data in Fig. 15.

 

 

SOURCE DF MS. 103 5“
Temperature 3 11565 189.15'"
Population 1 1313 20.60".
Temp. x Pop. 3 502 8.21."
Family(Pop.) 17 64 1.04
Error 64 61

 

 

3‘ The F test denominator for Population was MSFamily(Pop.); all others
were tested over MScmr.

' P < 0.05; " P < 0.01; P < 0.0001

 

64
larvae throughout the fifth instar indicated nearly exponential growth for the first
several days, then 1-2 days longer when larvae continued to feed but grew very
little (unpublished data). My estimates of Molt5 (Figure 15) attribute the
dramatically slowed growth in this stage to the early physiological demands of

molt and preparations for metamorphosis to the pupal stage.

Pupal mass

Michigan caterpillars produced significantly larger pupae than Alaskan
caterpillars (Figure 16; F188 = 11.89, Table 11). There were also strong main
effects Of temperature and sex on pupal mass (Table 11). Pupal mass averaged
25% less at 12° than at 18°, and female pupae averaged 9% larger than male
pupae (Figure 9). Population differences were less pronounced in males and at

low temperatures.

Survival

Populations did not differ in their survival across temperatures or at any
one temperature (Table 12; chi-square for population effects = 0.12, P = 0.73, df
= 1; chi-square for population x temperature interaction = 2.28, df = 3, P =
0.52). However, survival was influenced by temperature (chi-square = 39.77, df
= 3, P = 0.0001). Survival to the reproductive adult stage was'only about 25%
among animals reared through the final two instars at 12°C, primarily due to low

pupal survival (Table 12). Overall survival was highest at 18° (average of 91%),

and intermediate at 24 and 30° (average of 60%).

 

 

 

1°00? MALES

900
8001
4 AK
700:

600: iSE

 

500 T I I 5

100°? FEMALES

PUPAL MASS (mg)

900:
800:
700:

600 :

 

 

500‘ 1 1 .
12 18 24 30

TEMPERATURE (C°)

Figure 16: Pupal mass of male and female P. canadensis from Alaska (solid line)
and Michigan (dashed line) as a function of temperature. Larvae were reared on
P. tremuloides in 1989. See Table 11 for corresponding ANOVA.

 

 

 

66

Table 11: ANOVA comparing the pupal mass of male and female P.
canadensis from Alaska and Michigan reared at four temperatures.
Corresponds to data in Fig. 16.

 

 

Source df MS - 10'1 F“
Temperature 3 19372 36.12."
Population 1 6375 1189'"
Sex 1 9213 1718'"
Temp. x Pop. 3 666 1.24
Temp. x Sex 3 1140 2.13
Pop. x Sex 1 250 0.47
Temp. x Pop. x Sex 3 458 0.85
Error 88 536 0.85

 

'P < 0.05; " P < 0.01; P < 0.0001

 

67

 

 

 

 

l I
«9:. Ame—w Akcvw Axum.» $8 memo 8&8 SSWN H.123 120%
25p 858 8&3 §mw o\eco_ 0&9: $89.. 8&3 133:; 18:5
993 smug 850w Skim $50 enema 8&3 «with 1333.0. 7.551—
NN mm 3 _N E w— 0 MH OLE—E 3::—
:z v_< :2 x< :2 ¥< =2 M<
cc». evm 0*: 0.2

 

45.35:; GEE saw its .3 8.60.5 2: fl FEE; 3:22:50 .wEEm

9:33.... 2: 2.3:. mm 392:9 van 32am? .3 OE a BEE; :2: 395 be 03.58% 2: mm 733:; 395
€332. 3.33.383. =2: E25388 2: 9.1::ch 23:; BE: 558 3:6 he uwfiaoueom of mm fififizm
13.5-— .oxi E 35.52.53 53.. E 322 59:32 new 87d? 50.: 22233523 .m “O 135:5 ”Q 233.

 

68

Development time

To develop from early fourth instars to pupae, P. canadensis in 1989
required 9.5 to 58 d depending on temperature and genotype (Figure 17).
Michigan larvae required 4.9 times longer at 12° than at 30° (mean t SE = 57.5
t 2.4 versus 11.7 i 0.6 (1). Alaskan larvae were somewhat less temperature
sensitive (required 4.0 times longer at 12° compared to 30°: 37.7 t 2.6 versus 9.5
i 0.3 d), and required 19-34% less time overall (larger percentage differences at
low temperatures) (Figure 17). The temperature sensitivity of development time
would have been even more pronounced if fifth instars at 12° had grown as large
as at warmer temperatures (Figure 9). The fourth and fifth molts together
accounted for 35 to 51% of development time (Figure 17). In both populations,
molt made up the largest fraction of development time at 12°. The fourth instar
required less time for Alaskan larvae than Michigan larvae partly because of
higher relative growth rates (Figure 10) and partly because they began the instar
at a slightly larger size (W4) 1 SD = 112 t 54 versus 86 t 42 mg; P = 0.006)
(our estimates of time in the fourth instar assume that larvae began the instar at
90% of W”). The mass of larvae at the start of the fifth instar did not differ
between populations (W5) : SD = 449 t 151 versus 492 t 170 mg for Alaska
and Michigan respectively; P = 0.21), and only modestly (if at all) across

temperatures (F394 = 2.54, P = 0.061, maximum difference among 4 temperatures

= 14%).

 

69

60

 

50 [:1 MOLTING
GROWTH

 

 

 

P. canadensis AK

 

 

 

  
              
    

’0?

>

m

3

E

: 5°“:

< :

CI: 50:

D l

c: 3 .

40“: P. canadenSIs MI

  

...........................

IN S Fl

   

awn-m

 

1 2 1 8 2 4 3 0

TEMPERATURE (C°)

 

 

 

Figure 17: Development time from early fourth instars to pupae of P. canadensis
from Alaska (upper) and Michigan (lower) as a function of temperature.
Durations of the fourth instar, the fourth molt, the fifth instar, and the fifth molt,
are indicated.

 

70

Basking success

I found only rare circumstances that allowed P. canadensis larvae in Alaska
to elevate their body temperature much above ambient. Larvae were deliberately
placed in high radiation microsites (maximally exposed south-facing branches)
within a high-radiation site (a south-facing forest edge on a south-facing slope),
yet in 145 caterpillar observations on 8 sunny days (7-12 and 18-20 August 1989, I
never found a larva more than 45° above ambient air temperature (mean : SD
= 0.9 : 13°C). Generally the basking potential (measured by black body
temperatures) was not very high (maximum of 11.0 °, mean : SD = 2.7 : 25°)
because of breezes and/or shading from nearby leaves. Larvae seldom moved.
Most larvae remained within 20 cm of where I placed them throughout the
experiment (3-5 (1). I did not observe larvae moving to high radiation microsites
even though such sites were often only cm away. On 22-23 August, in an attempt
to maximize basking potential, I placed larvae individually on freshly cut branches
(ca. 50 cm high) positioned on a lawn in full sunlight but protected from the
breezes. Under these conditions, larvae were able to elevate body temperatures
as much as 10-14°C above air temperatures. Across a range of environmental
conditions, the basking success of larvae was about 50% of basking potential as
measured by black body temperatures (Figure 18). There was no evidence for
population differences in basking success (Figure 18). The body temperatures of
Alaskan larvae did not differ from that of Michigan larvae when exposed to the
same field conditions (Figure 19). ‘

Growth rates of larvae in the field and laboratory were indistinguishable

when fed the same foliage and subjected to the same mean daily air temperature

 

 

71

success = -o.24 + 0.53-POTENTIAL, r2 = 0.82

 

 

 

 

 

 

 

 

 

 

 

14 .
9 AK
V F10“ AK .:/
33 a ‘ “M' . -
ol—< °° .0 I’ll o
D ' 6" . . I I O
U) < ’I30 0.
0 E 293° 69" 9 °' '
E S .. .8 ,0" 0° 0
éh % .’/ C. .0:
g 2“ 00 $0 (2) 0
cl . /
a:
I o
-2 . . . , - . . , . . . , . . fl, . . . T .
-2 2 6 10 14 18

BASKING POTENTIAL (C°)
TBLACK " TAMBIENT

Figure 18: Basking success as a function of basking potential in fifth instar P.
canadensis from Alaska (closed circles) and Michigan (open circles). Basking
success equals the difference between larval temperature and ambient air
temperature; basking potential equals the difference between the temperature of
a high absorption "black body" and ambient air temperature. The heavy solid line
shows the points of equality between success and potential. The pooled
regression function is indicated (regressions fit to each population separately did
not differ: F2,112 = 1.12, P > 0.20). Note that most points to the right of 6°C on
the x axis represent observations of larvae on detached branches placed in full
sunlight and protected from the wind.

 

72

 

 

 

 

30: .0
A o o
2) o
%m 25‘
< .
Lu>
ED:
:< ‘
l-_l 20'
33:2 J O
LU< 1
of) 4
5(15:
,__1
<
10....,.-..,.-.,....,.
10 15 20 25 30

TEMPERATURE OF
MICHIGAN LARVAE (C°)

Figure 19: The temperatures of Alaskan and Michigan larvae when allowed to
move and bask naturally under field conditions near Fairbanks, AK. Each point
is the mean of 5-10 larvae from each population. Standard deviations ranged
from 05°C at low temperatures to 4°C at high temperatures.

 

73
(RGR : SE = 0.157 x 0.006 versus 0.143 x 0.014 mg-mg'l-d'l for field and
laboratory larvae respectively; t12 = 0.79, P > 0.20). Over the three days of this
experiment, daily minimum and maximum air temperatures in the field were: 10.5
and 24°; 10 and 20°; 8 and 17°. Larvae in the laboratory were held at constant

temperatures of 17° on day 1, 15° on day 2, and 125° on day 3.

Discussion

Adaptive modifications of temperature physiology

P. canadensis populations from Alaska and Michigan have diverged in their
larval temperature physiology (Figures 1012, 14-15). Alaskan larvae are capable
of growing and molting more rapidly at the low temperatures (Figures 8-9) they
typically encounter. lntraspecific differentiation in insect temperature responses
has also been reported in milkweed bugs (Baldwin and Dingle 1986), Drosophila
larvae (Barnes et al. 1989), and Colorado potato beetles (Tauber et al. 1988). No
differences were found in the diamondback moth (Sarnthoy et a1. 1989).

Adaptation of P. canadensis to short cool summers has been accomplished
in part through a general elevation of metabolic activity at all temperatures.
Average daily metabolic rates were 36% higher in the Alaskan population (Figure
12). Consequently, Alaskan larvae had very similar metabolic rates at their
average environmental temperature (0.268 mg-mg‘1 - (1’1 at 14.4°) as Michigan
larvae at theirs (0.261 mg- mg'1 - d‘1 at 188°) (linear interpolation of ADMR data
in Figure 12). Probably as a result of this "metabolic compensation" (Scholander
et al., 1953), Alaskan caterpillars were able to consume more, grow faster, and

molt more quickly at low temperatures than Michigan caterpillars. The metabolic

74
rate of diapausing pupae was similarly elevated in the Alaskan population (35%
higher rate of CO2 production than the Michigan population; Kukal et a1. 1992).

There was limited evidence for concomitant reductions in larval growth
performance at high temperatures. Alaskan fourth instars feeding on Populus
tremuloides in 1988 grew only half as fast at 30" as Michigan fourth instars, even
though they grew 34% faster at 12° (Figure 10). The low growth rate of Alaskan
larvae at 30° was due to high respiration rates (Figure 11), suggesting an
energetic tradeoff between adaptation to low temperatures and performance at
high temperatures. However, such tradeoffs were far less apparent than I would
have predicted.

In 1989, Alaskan larvae grew faster (Figure 10) and survived as well
(Table 12) at 30° as Michigan larvae, even though the Alaskan population never
encounters sustained temperatures that warm (Figure 9). The apparent "cost" of
low temperature adaptation (high metabolic rates) was still evident in these trials
(Figure 12), but high consumption rates with equivalent digestive efficiency
allowed the Alaskan larvae to grow faster nonetheless. As might be expected
given their high maintenance metabolism, Alaskan larvae were more sensitive to
host quality than Michigan larvae, especially at high temperatures. The variance
among experiments in mean relative growth rate at 30° was 10-33 times higher for
Alaskan compared to Michigan larvae (fourth instar SD = 0.177 versus 0.056 for
Alaska and Michigan respectively; fifth instar SD = 0.097 versus 0.017; data from
Figure 10). Yet this represents only a limited spectrum within generally high
quality hosts; Populus balsamifera and P. tremuloides ranked one and two in a

comparison of fifth instar growth rates across nine host species (Chapter 1, Figure

75
4). Alaskan caterpillars, as a result of their elevated metabolism, may be less
able to exploit hosts of low nutritional quality. Shallow growth-temperature
responses, such as on Betula resim'fera (Figure 13), are expected to be more
common as a consequence of low temperature adaptation. Even on high quality
hosts, the growth rates of Alaskan larvae were less temperature sensitive than
those of Michigan larvae: fifth instar 010’s of 1.99-2.16 versus 2.27-2.58 (Figure 10,
12-30°).

In general, biochemical adaptation to low temperatures can occur through
two routes (Hochachka 1973, pp. 212-270): (1) increased concentrations of
enzymes, which increase reaction rates at all temperatures, or (2) qualitative
changes in enzyme structure that result in higher substrate affinity at low
temperatures (changes in the response of K,” to temperature). The generally
elevated metabolic rate of Alaskan P. canadensis and other high latitude animals
(Scholander et a1. 1953, Hochachka 1973, Block 1990) is an expected correlate of
increased rates of enzyme production and turnover. Changes in enzyme structure
that enhance metabolic performance at low temperatures are expected to
compromise performance at high temperatures (Powers 1987), which may explain
the temperature x population interaction in the duration of the fifth molt (Figure
15, Table 10). That a similar pattern was not seen in the fourth molt (where
Alaskan larvae molted faster at 30° as well as 12°; Figure 14, Table 9), suggests
changes of the Km-temperature function in enzymes uniquely associated with the
larva-to-pupa metamorphosis (as opposed to molt per se). In general, there was
little evidence for shifts in the temperature optima of physiological systems.

Consumption rates, assimilation efficiencies, molting rates, survival, and usually

76
growth rates, remained as high or higher in the Alaskan population, even at the
ecologically extreme temperature of 30°. Distinct temperature optima in these
same processes are often apparent over only a few degrees centigrade in aquatic
insects (Sweeney and Vannote 1978, Grafius and Anderson 1979, Vannote and
Sweeney 1980), presumably because their thermal environment is much less
variable. Qualitative changes in enzyme structure may typically be of less

importance in the thermal adaptation of terrestrial compared to aquatic insects.

Conserved attributes

Although Alaskan larvae had higher development rates at 12° than their
Michigan counterparts, there was little evidence for changes in developmental
thresholds. The minimum temperature at which a biological process occurs (To)
can be estimated by linear extrapolation from the temperature response of
developmental rate (Arnold 1959). In four experiments, the temperature
responses of relative growth rate were sufficiently linear to allow reasonable
extrapolation (fourth and fifth instars on Populus balsamifera in 1988 and P.
tremuloides in 1989; Figure 10). Estimates of T t SD = 6.8 t 0.5 and 7.6 t 1.9
for Alaskan and Michigan caterpillars respectively (t7 = 0.80, P > 0.20, paired
t-test). Developmental thresholds for molting were also similar between
populations (Molt4 T0 = 96° and 95° for Alaska and Michigan; Molts rates were
too non-linear for extrapolation), but tended to be higher than thresholds for
growth (see also Ayres and MacLean 1987). Estimates of T0 for Papilio glaucus,

a southern sister species, range from 8-12° (Scriber and Lederhouse 1983, Ritland

and Scriber 1985, Grossmueller and Lederhouse 1985). Developmental

 

 

77
thresholds of swallowtail caterpillars do not differ very much from Florida to
Alaska, even across species.

The probability of P. canadensis successfully completing development in
Alaska might be enhanced if larvae could regularly elevate body temperatures
through basking. Lepidopteran larvae can attain temperatures 5-20° above
ambient by adjusting posture and orientation, exploiting thermal heterogeneity
within the environment, and minimizing convective heat losses (Casey 1976, Casey
et a1. 1988, Kukal et al. 1988, Weiss et al. 1988). P. canadensis and P. glaucus
spend most of their time sitting on silken mats that they construct on the adaxial
surface of individual leaves. Besides providing a secure footing, these mats can
retard convective heat loss and contribute to radiant energy gain by turning the
leaf into a parabolic reflector (Grossmueller and Lederhouse 1985). P. glaucus
larvae also exhibit positive phototaxy within the host canopy which tends to put
them in high-radiation microsites. I hypothesized that these behaviors would be
particularly well developed in Alaskan caterpillars, but in fact, basking efficiency
(temperature elevation as a fraction of basking potential), and average caterpillar
temperatures were no different in the two populations (Figures 18-19). The
opportunities for basking appear to be are very rare for Alaskan swallowtails
under natural conditions, apparently because of generally low light intensity at

high latitudes.

Egg size and adult size
As a consequence of different provisioning strategies by their mothers,

Alaskan caterpillars begin larval development 36% larger than Michigan

78
caterpillars (neonate mass = 1.33 versus 0.98 mg, Chapter 1), and because larval
growth is nearly exponential, this relative advantage is retained throughout larval
development (Figure 3). Given equal growth rates, Alaskan P. canadensis could
attain the same pupal mass with less larval development time due solely to the
difference in hatching mass.

Alaskan P. canadensis further shorten development time by terminating
feeding and initiating pupation at a smaller mass. If we disregard the 12°
treatment, where pupae were very small and survival very low, female Alaskan
pupae averaged 814 mg compared to 899 mg for Michigan pupae (9% difference;
Figure 9). In a different experiment (Figures 6-7), Alaskan pupae averaged 8%
smaller. Field captured butterflies reveal the same pattern. Female forewing
length averaged 43.7 .t 0.80 mm across 3 summers in Alaska and 46.6 i 0.88 mm
across 7 summers in Michigan (means x SD of yearly means, each based on 10-20
butterflies). Pupal mass is related to forewing length as Mpupa = -528 +
31.05 - WING (r2 = 0.55, unpublished data), which indicates that wild female
pupae averaged 829 mg in Alaska versus 919 mg in Michigan (10% smaller in
Alaska).

Comparing the ecological worth of temperature adaptations

Four differences between Alaskan and Michigan populations of P.
canadensis are interpretable as adaptations to short cool subarctic summers:
increased egg mass, reduced adult size, enhanced molting abilities at low
temperatures, and enhanced growth rates at low temperatures. To evaluate the

contribution of these hypothesized adaptations, individually and in toto, to

79
swallowtail fitness in an Alaskan environment, I incorporated them into a
temperature-driven development model that input 48 years of daily temperature
records and predicted the proportion of swallowtails that would successfully
complete development each year.

Parameter values are summarized in Appendix 2. The model evaluated
five cohorts in each year, corresponding to phenologically early, average, and late
eggs (neonate hatch = 240, 280, 320, 360, and 400 degree days). The mass of
hatching neonates was set at 0.98 or 1.33 mg representing Michigan and Alaskan
phenotypes respectively. The size threshold for female pupation was set at 899 or
814 mg fresh pupal mass (Figure 16). Molting rates for the penultimate and final
molt were calculated for each population as a function of temperature, using
linear interpolation between measured temperatures and from 12° to a
developmental threshold of 95° (based on data in Figures 14-15). Growth rates
were estimated under two scenarios of host quality: a "good host" (based on
Populus tremuloides in 1988) and an "excellent host" (based on P. tremuloides in
1989). Relative growth rates during the penultimate (RGR4) and final instars
(RGRS) were calculated for each population as a function of temperature, using
linear interpolation between measured temperatures and from 12° to a
developmental threshold of 725° (based on Populus tremuloides data in Figures
10).

Two host quality scenarios were used to model growth during instars 1-3
(Stage 1), but the same functions were used for all swallowtail phenotypes.
Growth rate at 24° during instars 1-3 was set at 0.270 mg-mg‘1 - d’1 for the good

host scenario (= growth rate for this stage on Populus tremuloides in 1988; Figure

80
3) and 0.310 mg- mg'1 - d’1 for the excellent host scenario (= highest growth rate
among nine hosts in 1988; Figure 3). Stage 1 temperature responses assume the
same 0105 as fourth instar larvae feeding on Populus tremuloides in 1988 (good
host) or 1989 (excellent host).

Each cohort in each year was represented by hypothetical neonates that
hatched at a specified degree day, then developed through five growth stages, at
rates dictated by prevailing temperatures and stage-specific temperature
responses, until they reached the overwintering pupal stage or the season ended.
When they were not molting, larval mass was incremented daily as
M”, = M,-eRGR" where M, was the mass at the start of the day and relative
growth rate (RGR) was estimated as a function of temperature. When the
average daily temperature was less than the developmental threshold, but the
maximum was not, development was based on proportion of the day when
temperatures were above the threshold (sine function follows Watanabe 1978).
Larvae grew from their hatching mass to 99 mg (Stage 1), then to 369 mg at
fourth instar growth rates (Stage 2). Then growth ceased until completion of the
fourth molt (Stage 3). With their mass depreciated by exuvial and respiratory
losses during molt, larvae resumed growth at fifth instar rates until they reached a
size threshold for pupation (Stage 4). Successful cohorts completed feeding prior
to leaf senescence (15 September) and became cold-tolerant diapausing pupae
(completed Stage 5) before the onset of winter (1 October).

Under the good host scenario, the Michigan phenotype was predicted to go
extinct (no cohorts completed development) in 31 of 48 years under the Alaskan

climate (Figure 20). Changing egg size to that of the Alaskan phenotype

81
GOOD HOST EXCELLENT HOST

  
   
      

MICHIGAN
PHENOTYPE

xg-‘flemww '7 3
I. .44. 1 ' '

~sz42%: ’ -

 
 
   
 
 
 
 
 
 
 
    

ALASKAN
EGG MASS

 

 

a) ALASKAN
I: 20 PUPATION
a 10 . THRESHOLD
>-
u. . 2
O 0
a:
m ALASKAN
m 20 MOLTING
E
D 10
z
0
ALASKAN cRoer RATES
20

 

o w . v v v

ALASKAN PHENOTYPE

 

   

NUMBER OF SUCCESSFUL COHORTS

Figure 20: The predicted developmental success of various P. canadensis
phenotypes under two scenarios of host quality during 48 seasons in interior
Alaska. Possible developmental success in each year ranged from 0 successful
cohorts (extinction) to 5 successful cohorts (no mortality due to incomplete
development). Under each host quality scenario, I began with the Michigan
phenotype (top), then changed egg mass, pupation threshold, molting rates, and
growth rates (one at a time) to the condition of the Alaskan phenotype. Bottom
figures show results when the four Alaskan adaptations were combined.

 

82
eliminated 6 extinctions, changes in molting physiology and pupation threshold
each eliminated 3 extinctions, and changes in fourth and fifth instar growth rates
eliminated 19 extinctions. The Alaskan phenotype, which incorporates all these
changes, was predicted to go extinct in only 6 of 48 years. Under the excellent
host scenario, extinctions were rare, but even the Alaskan phenotype still had
failed cohorts in 20 of 48 years (Figure 20). This indicates that climate continues
to exert hard selection (sensu Wallace 1968) even on the adapted phenotype
under the best available conditions. Estimating the relative size of the cohorts
allowed an assessment of this selection.

On a good host, only 14% of the Michigan phenotypes, compared to 42%
of the Alaskan phenotypes, were predicted to reach pupation in the time
available (Table 13). Thus the estimated fitness of the Alaskan phenotype was
about 3.0 times higher (42/ 14). On an excellent host, larval success was higher
(65% versus 89%), but selection against the Michigan phenotype would still be
very strong (relative fitness of the Alaskan population = 1.39). Under both host
quality scenarios, changes in growth rate made the largest contribution to
improved fitness. Changes in egg mass made the second largest contribution (on
a good host, increased egg mass resulted in a fitness of 1.21 relative to the
Michigan phenotype). Changes in the size at pupation had the smallest effect on
fitness. All the hypothesized adaptations of the Alaskan population contributed

to successful development during Alaskan summers.

83

Table 13: A comparison of the value in enhanced developmental success of adaptations
exhibited by Alaskan P. canadensis. Percent larval success indicates the average proportion of
larvae predicted to complete development during 48 seasons in Alaska (based on data in
Figure 2D assuming cohort sizes of 12%, 32%, 27%, 18%, and 11% for cohorts 1-5°). Alaskan
attributes were introduced into the Michigan phenotype one at a time (eg size, pupation size,
molting rates, and growth rates) and then simultaneously (Alaskan phenotype).

 

 

 

 

 

 

 

 

Good Host Excellent Host

Percent Number Fitness Percent Number Fitness

larval of relative larval of relative

Phenotype success Extinctionsb to MIc success Extinctionsb to MIC
Michigan 14 31 1.00 65 1 1.00
Ml with AK egg size 17 25 1.21 68 1 1.05
Ml with AK pupation size 15 28 1.06 67 1 1.03
MI with AK molting rates 16 28 1.12 69 1 1.06
Ml with AK growth rates 29 12 2.03 84 0 1.29
Alaskan 42 6 3.00 89 0 1.37

 

‘ Cohort sizes based on the emergence phenology of female butterflies in Alaska (median =

164 and 167 degree days in two years, 10 to 90% cumulative emergence = 137 to 209
degree days), and assuming: a butterfly mortality rate of 11% per 10 degree days; constant
egg production per degree day among living butterflies; and 90 degree days from oviposition

to hatch (unpublished data).

b‘ Number of times in 48 years when even the earliest cohort failed to complete development.

c' Average larval success relative to the Michigan phenotype.

 

fl

84

Tradeofl‘s

Because smaller pupae produce smaller less fecund adults, there is clearly
a tradeoff between the probability of completing development and fitness given
successful development. Two days after adult eclosion, an Alaskan butterfly that
weighed 812 mg as a pupa has matured 21 eggs compared to 26 eggs for the
butterfly resulting from an 899 mg pupa (Lederhouse and Ayres, unpublished
data). Thus the fitness cost of pupating smaller is about 20%, compared to an
apparent benefit of only 3-6% (Table 13). The actual benefits of pupating
smaller are underestimated if there is appreciable larval mortality due to
predation risks. In Alaska, we have recorded fifth instar mortality rates of 27%/d
in 1989, 2%/d in 1990, and 5%/d in 1991. Under 1989 temperatures, lowering
the pupation threshold shortened the fifth instar of the middle cohort from 14.2
to 13.4 d. Given mortality rates of 27%/d, the probability of surviving the fifth
instar was 1.28 times higher with the lower pupation threshold; daily mortality
rates of 5% suggest an advantage of only 1.04. Average temperatures and
predation risks seem inadequate to explain the reduced fecundity that
accompanies lower pupation thresholds in Alaskan P. canadensis. I suggest that
the occasional extreme conditions (e.g., a sequence of three unusually cold
summers during 1947-1949, or an outbreak of predatory wasps in 1988-1989) exert
disproportionately strong selection, especially on populations that rely on an
annual life history (Wigley 1985).

By virtue of producing smaller eggs, Michigan butterflies can mature about

1.59 times as many eggs per day as equal sized Alaskan butterflies (Lederhouse

and Ayres, unpublished data). Compare this cost to an estimated benefit of 1.21

 

85
based on average summer temperatures in Alaska (Table 13). The actual costs of
producing larger eggs may not be as great if summer temperatures frequently
limit flight more than they limit vitellogenesis (if oviposition rate is not limited by
egg production), but this apparent discrepancy between costs and benefits again
argues for the selective strength of occasional extreme conditions. Increased egg

size eliminated 6 extinctions in 48 years on good hosts (Table 13).

Model Validation

There were no obvious discrepancies between reality and results from the
development model. During 1987-1990, the predicted dates of cohort initiation
overlapped and followed the actual times of butterfly flight, and matched the
times when eggs were hatching in the field. In 1988, two wild larvae with known
hatching dates (4 July and 8 July) survived until they were collected as fifth
instars on 3 August (mass of 936 and 696 mg respectively). The actual
development of both larvae fell between model predictions under the good host
and excellent host scenarios (predicted masses = 97 and 1166 mg for the 936 mg
larva). Careful validation requires more such comparisons, but few larvae survive
that long in the field. Serious inaccuracies could arise if larval temperatures are
typically higher than ambient due to basking (Lamb and Gerber 1985), if foraging
larvae tend to select high quality leaves (Schultz 1983), or if growth at naturally
fluctuating temperatures differs from growth at a constant temperature (Taylor
and Shields 1990). Congruence between field and laboratory growth rates (see

Results) suggest that none of these introduce serious errors for P. canadensis in

interior Alaska. Some parameter estimates, particularly early instar temperature

 

86
responses (Stage 1) are not as robust as would be desirable, but this has little

effect on comparisons among phenotypes (Table 13).

Northern distribution limits of P. canadensis

There is a dramatic effect of host quality on developmental success in
Alaska: average larval success of 42% versus 89% for Alaskan larvae under good
host versus excellent host scenarios (Table 13). Many potential hosts are worse
than our good host scenario, and few are better. Growth performance on
Populus tremuloides in 1988, the basis for the good host scenario, was among the
highest of nine host species tested (Figures 4, 6). Our excellent host scenario was
based on the same tree species in 1989, suggesting that such high quality foliage
may not be available in all seasons. Interior Alaska contains many tree species
on which P. canadensis can potentially produce viable adults (laboratory
experiments in Chapter 1), but on which growth rates are too low to allow
successful development of any but the earliest larvae in the warmest years.
Consequently, climate may select for more discriminating oviposition behavior in
Alaskan populations of P. canadensis. Species of Salix and Betula that are
suitable hosts for P. canadensis given adequate degree days extend throughout the
circumpolar regions well beyond the occurrence of P. canadensis. I hypothesize
that northern distribution limits are a joint function of climate and host quality.
As summers become shorter, the number of potential hosts that can be realized

hosts becomes increasingly restricted until even the best hosts no longer allow

development. If this is a general scenario, insect distributions may respond to

 

87
climate change (Mitchell et al. 1989) almost immediately, without requiring
changes in host distribution.

Regional specialization allows P. canadensis to maintain a broader climatic
distribution than would otherwise be possible. It scents unlikely that a population
with Michigan attributes could be sustained in the Fairbanks area (Figure 20,
Table 13). In spite of geographic differentiation, many larvae in many years
apparently still fail to complete development during the short subarctic summers
(estimated mortality of 58% on a good host). Evolutionary divergence in the
thermal physiology of P. canadensis is not trivial, but Alaskan swallowtails do not
approach the low temperature capabilities of truly high-latitude herbivores such
as the tenthredinid sawfly, Dineura virididorsata (Matsuki and MacLean 1990), or
the geometrid caterpillar, Epirrita aurumnata (Ayres and Maclean 1987), which
are capable of feeding, growing, and molting at temperatures just above 0°C.
Further adaptation in P. canadensis may be limited by gene flow from southern
regions or by even more fundamental constraints arising from their common

ancestry with the basically tropical Papilioninae (Scriber 1973, Scriber et al. 1991,

Kukal et al. 1992).

 

SUMMARY AND CONCLUSIONS

Important physiological attributes of P. canadensis appear to differ
markedly in their evolutionary lability (Table 14). Neonate mass and the
duration of the fifth molt differed by more than 3 standard deviations between
populations, while many other traits differed by less than 0.5 standard deviations.
Although they seemed about equally important a priori, differences in summer
temperature regime appeared to be a more potent force in local adaptation
(divergence of > 18D in 5 of 8 traits) than differences in host community
(divergence of < 18D in 7 of 7 traits).

The ability to consume, detoxify, assimilate, and convert chemically diverse
plant tissue is highly conserved within P. canadensis (Chapter 1). There was no
detectable variation in host-use abilities between Michigan and Alaskan
populations even though there is little overlap between their host communities.
Michigan caterpillars grew just as well as Alaskan caterpillars on Alaskan hosts,
and no better on Michigan hosts. There was also no evidence of host specialists
within populations. Consequently, the genetic mosaic hypothesis was rejected
with respect to detoxification systems and nutritional physiology. The explanation
for extraordinary polyphagous abilities in P. canadensis lies not within the
population structure of the species, but within the genome of any individual

within the species (super-genotype hypothesis). However, there is a marked

 

 

 

 

 

89

Table 14: Summary of geographic differentiation between P. canadensis from Alaska and Michigan.
Traits are organized by the environmental factor hypothesized to select for divergence in the Alaskan
population (low summer temperatures, or unique phytochemistry of their hosts). Change is expressed in
terms of percentage, standard deviations, and statistical significance.

 

Change from
Selective force Alaskan phenotype Michigan phenotype

 

 

Trait Mean : SDa % SD units Pb

 

 

Low summer temperatures

 

Neonate mass 133 t 0.10 mg 36 3.20 ‘“
Pupal mass 814 t 73 mg -9 -1.16 ‘”
Molt4 at 12° 658 t 2.78 d -19 —0.56 ‘
Molts at 12° 11.83 t 1.83 d -33 -3.27 ‘"
4th instar growth at 12° 0.152 x 0.036 mg-mg‘l -d'1 33 1.06 W
5th instar growth at 12" 0.119 t 0.025 mg: mg'1 -d'1 57 1.72 “
Apparent digestibility at 12° 0.41 t 0.04 mg/mg 8 0.75 NS
ECD (Growth/Assimilation) at 12° 0.38 1 0.07 mg/mg 23 1.00 NS
Nitrogen use efficiency at 12° 0.71 t 0.09 mg/mg -9 —0.78 NS
Survival at 12° 25 t % -7 NS
Basking efficiency at 12° 0.54 : C°/C° 6 i NS
Host phytochemistryc
lst instar growth 0.517 1 0.160 mg- mg'1 - d'1 12 0.34 +
Middle instar growth 0.368 : 0.022 mg- mg‘1 - tr1 0 0.03 NS
5th instar growth 0.184 t 0.050 mg- mg‘1 - d'1 -4 -017 NS
Consumption 1.65 t 0.24 mg- mg'1 - d'1 4 0.26 NS
Apparent digestibility 0.30 t 0.05 mg/mg -5 0.31 NS
ECD (Growth/Assimilation) 0.39 t 0.11 mg/mg -2 -0.06 NS
Nitrogen use efficiency 0.55 t 0.11 mg/mg -10 -0.55 NS
Survival 59 t % 3 NS
a. MSon

b- + 0.10 > P > 0.05; t P < 0.05; " P < 0.01; P < 0.001

Performance is compared on 7 Alaskan host species not naturally encountered by the Michigan
population.

 

90

discontinuity in host-use abilities between P. canadensis and its presumed
progenitor, P. glaucus. P. canadensis grows faster and survives better than P.
glaucus on a broad spectrum of boreal host species, and fares substantially worse
on at least two southern tree species that are important hosts of P. glaucus. I
hypothesize that the nutritional physiology of P. canademis diverged rapidly from
its ancestral type at the time of speciation and has been little changed since. This n
is consistent with models of stepwise evolution (Ehrlich and Raven 1964) and

punctuated equilibria (Eldredge and Gould 1972). Even strong natural selection

 

may be inadequate to alter the nutritional physiology of herbivores. Changes in
boreal tree communities that are expected to accompany global change are
unlikely to lead to changes in the nutritional physiology of P. canadensis.

In contrast, there was conspicuous divergence in many attributes related to
the temperature physiology of P. canadensis (Chapter 2; Table 14). Alaskan
caterpillars consume more, grow faster, and molt quicker than their Michigan
counterparts at low temperatures typical of the Alaskan environment. Changes in
reproductive allocation (Alaskan females lay fewer, larger eggs) and adult size
(Alaskan larvae initiate pupation at a smaller size) represent additional adaptive
responses to cool subarctic summers. Evaluation of fitness tradeoffs in egg size
and adult size suggest that the occasional extreme years are more important than
the average years in determining the optimal phenotype. Analysis of 48 years of
Alaskan weather records indicates that P. canadensis with the attributes of the
Michigan p0pulation could not occur as far north as Fairbanks. Regional
adaptation of temperature physiology allows P. canadensis to maintain a broader

geographic distribution than would otherwise be possible. In spite of low

91
temperature adaptations, the northern distribution of P. canadensis is probably
still constrained by low summer temperatures. Tree species that are suitable
hosts given sufficient degree days occur throughout the Holarctic, well beyond the
northern limits of P. canadensis. Consequently, climatic warming would be
predicted to yield ecological changes (e.g. rapid northern distribution extensions)
as well as evolutionary changes (e.g.ilarger adults, smaller eggs, and lower
metabolic rates in the Fairbanks population).

Since its origins as a discipline, a central problem of evolutionary biology
has been to appreciate the potential of natural selection while understanding its
constraints (Darwin 1859). Natural history studies have long indicated that
attributes such as host-use abilities and temperature responses differ between taxa
inhabiting different environments; these patterns suggest that the traits are
ecologically relevant, but tell us little about their evolutionary lability. These
traits could be of such physiological and genetic complexity that major alterations
only come about in the evolution of new genera or families, or they could be very
plastic characteristics that are critical in local adaptation. The first scenario
emphasizes constraint and extinction (the physiology of the organism restricts its
distribution), while the second emphasizes adaptive potential. Patterns of
punctuated equilibria, as revealed by the fossil record for some morphological
traits (Eldredge and Gould 1973), have given new legitimacy to arguments for
evolutionary constraint within species. However, we cannot easily generalize from
whorl patterns in snail shells to the physiological and behavioral attributes that
interest ecologists. Even in the absence of a fossil record, the increasingly

sophisticated tools of comparative biology offer a means of discriminating

92
between conserved and labile traits (Feder 1987, Huey 1987). My results suggest
that swallowtail temperature responses are relatively labile compared to host-use
abilities (Table 14). Analysis of variation within and between other taxa will test
the generality of these patterns, suggest other generalizations, and allow
inferences about the evolutionary history of key ecological traits. In combination
with better knowledge of gene flow and the genetic architecture of traits, this
research program should clarify the role of adaptation and constraint in the

interactions of organisms with their environment.

APPENDICES

 

 

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.: 1991-10
Title of thesis or dissertation (or other research projects):
Adaptation and constraint in Papilio canadensis:

geographit: var*iati()n in nutritional physiology
and temperature responses

 

 

Museum(s) where deposited and abbreviations for table on following sheets:
Entomology Museum, Michigan State University (MSU)

Other Museums:

Investigator's Name (3) (typed)
Matthew P. Ayres

 

 

Dang 14 November 1991

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

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

dissertation. -
Included as Appendix 1 in c0pies of thesis or dissertation.

Museum(s) files.
Research project files.

Original:

Copies:

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

93

 

APPENDIX 1.1

Voucher Specimen Data

1 Pages

of

Page 1

23% 83

 

 

 

puma peumpso
8% 38% .fl remap? sexy \
\.E:mmsz kmoHoE0ucm

 

zuwmpo>wc3 oumum cmmmcowz ecu Cm uwwoaop
new mcmeumam pmumwa m>opm ecu pm>wmuwm
oHermH .oz nozoso>

mmca< .m gospumz

Apoaxuv AmvoEmZ m.u0uwmmumo>CH

Axpmmmoomc mm wuoozm Hmcomumppm mmbv

 

mmcx< .m .z
mmmflr>HroH
Hm omm 3mem zome

 

 

 

 

 

 

 

 

 

 

 

 

 

3m: 0 H .09 owpmmoo “Hz a w a mwmcmumcmo OTFEQQQ
mwcx< .m .2
mwmfirH>rmH
2m: CA N mxcmncrmm Hx< a w m mwmcmomcmo OWFVQmm
400+ pmuflmcaoc mew mew: mo pwuowHHCe coxmu umzuo no mwfluwam
m e k r m w e m % mcmeuoam no» muse Hmflmd
erodellapVS
sepehuupmrg
uhettddUV.aoo
M w d 1 o A A P N L E .
“Cc LOLEzz

94

 

Appendix 2

First instar growth rates of P. canadensis from Alaska, P. canadensis from Michigan, and P. glaucus
from Ohio, tested on tree species from Alaska, northern Michigan, and Southern Michigan. Except
as noted, SE is based on the number of families, and N equals the number of families.

 

First instar growth rate (mg - mg‘l - d'l)

 

 

 

 

 

P. canadensis AK P. canadensis WI P. glaucus OH
Mean SE N Mean SE N Mean SE N
Alaskan hosts — — — - ——
Populustremuloides Michx. 0.462 0.030 7° 0.409 0.052 7° 0.220 0.015 9b
Populus balsamrfera L. 0.653 0.023 8° 0.598 0.026 8° 0.418 0.024 9b
Salix alaxensis (Anderss.) Cov. 0.469 0.030 8° 0.442 0.033 7° 0.393 0.024 9b
Salix novae-angliae Anderss. 0.724 0.023 7° 0.651 0.044 7° 0.398 0.015 9b
Salix bebbiana Sarg. 0.435 0.043 8° 0.371 0.035 8°I 0.307 0.020 9b
58sz glauca L. 0.566 0.036 8ll 0.403 0.030 73 0.336 0.017 9b
Betula resimfera Britton 0.408 0.020 7° 0.381 0.033 7° 0.331 0.031 10b
Alnus crispa (Alt.) Pursh 0.454 0.071 7° 0.400 0.049 7° 0.177 0.022 9b
Alnustenurfolia Nutt. 0.560 0.020 9° 0.541 0.026 7° 0.347 0.029 10b
Northern Michigan hosts
Tilia americana L. 0.336 0.032 8° 0.356 0.017 8°
Prunus virginiana L. 0.450 0.042 8° 0.374 0.054 8°
Prunus serotina Ehrh. 0.464 0.050 7° 0.359 0.031 6°
Populus grandidentata Michx 0.477 0.030 9° 0.460 0.035 8°
Fraxinus americana L. 0.373 0.026 6° 0.447 0.026 6°
Betula papynfera Marsh. 0.442 0.026 7° 0.448 0.020 6°
Betula alleghaniensis Britton 0.162 0.029 8° 0.233 0.020 8°
Betula nigra L. 0.598 0.062 8° 0.623 0.031 7c
Alnus rugosa (Du Roi) Spreng. 0.461 0.045 7° 0.394 0.059 6°
Southern Michigan hosts
Lin'odendron tulipifera L. 0.188 0.056 7d 0299 0.084 9dc 0.758 0.046 7d
Ptelea Infoliata L. 0.550 0.047 8d 0.416 0.077 9dc 0.649 0.044 6d
771m americana L. 0.267 0.041 7d 0.296 0.039 4d
anus virginiana L. 0.340 0.030 8d 0460 0.035 5d
Prunur serotina Ehrh. 0.384 0.055 9d 0.363 0.032 6d
Fraxinus amen'cana L. 0.405 0.041 7°I 0.388 0.096 5d

 

a Foliage collected on 28 June 1988 from University of Alaska Fairbanks arboretum.

b Foliage collected on 18 August 1989 from University of Alaska Fairbanks arboretum.
° Foliage collected on 5 July 1989 from Dunn County, Wisconsin.

d Foliage collected on 18 June 1988 from Ingham County, Michigan.

° All larvae from a single family; N = number of larvae.

95

 

Appendix 3

Summary of parameters used in P. canadensis development model. M1 = Michigan; AK = Alaska.

 

A. Phenology

First day of season to begin accumulating thermal sums: 1 April.

Degree days (10°C base) of neonate hatch for cohorts 1-5: 240, 280, 321), 360, 400.
Last day of permissable growth (Stages 1-4): 14 September.

Last day of permissable prepupation (Stage 5): 30 September.

B. Mass of hatching neonates
Ml phenotype: 0.98 mg.
AK phenotype: 1.33 mg.

C. Stage 1, lnstars 1-3

Relative growth rates at 12, 18, 24, and 30°C
Good host: 0.083, 0.116, 0.270, 0.280 mg'mg'l°d'1.
Excellent host: 0.099, 0.223, 0.310, 0.416 mg- mg'1 ' d'l.

D. Stage 2, lnstar 4

Initial mass: 99 mg.

Relative growth rates at 12, 18, 24, and 30°C
Ml phenotype, good host: 0104, 0.161, 0.449, 0.533 mg'mg‘1 -d'1.
AK phenotype, good host: 0.139, 0.174, 0.334, 0.279 mg°mg'1'd'1.
Ml phenotype, excellent host: 0.109, 0.268, 0.350, 0.456 mg'mg'l-d'l.
AK phenotype, excellent host: 0.146, 0.305, 0.448, 0.617 mg- mg'1 ' d'l.

Final mass: 369 mg.

E. Stage 3, Molt 4 (lnstar 4 to lnstar 5)

Molting rates at 12, 18, 24, and 30°C
Ml phenotype: 12.3, 32.1, 52.6, 84.8 percent/day.
AK phenotype: 15.2, 48.1, 60.2, 116.3 percent/day.

F. Stage 4, lnstar 5

Initial mass: 348 mg.
Relative growth rates at 12, 18, 24, and 30°C
Ml phenotype, good host: 0.067, 0.127, 0.382, 0.369 mg°mg'1‘d'l.
AK phenotype, good host: 0.099, 0.181, 0.345, 0.341 mg mg‘1 ~ d'l.
Ml phenotype, excellent host : 0.069, 0.161, 0.255, 0.375 mg mg'1 - d’l.
AK phenotype, excellent host: 0.142, 0.214, 0.317, 0.532 mg mg’1 ' d'l.
Final mass
Ml phenotype: 1798 mg (corresponds to female pupal mass of 899 mg).
AK phenotype: 1628 mg (corresponds to female pupal mass of 814 mg).

C. Stage 5, Molt 5 (lnstar S to Pupa)

Molting rates at 12, 18, 24, and 30°C
Ml phenotype: 5.61, 16.3, 18.3, 34.8 percent/day.
AK phenotype: 8.45, 23.2, 23.5, 32.3 percent/day.

H. Minimum developmental thresholds
Growth (Stages 1, 2, and 4): 725°C.
Molting (Stages 3 and 5): 950°C.

 

96

BIBLIOGRAPHY

 

BIBLIOGRAPHY

Alstad, D. N., and K. W. Corbin. 1990. Scale insect allozyme differentiation
within and between host trees. Evolutionary Ecology 4:43-56.

Arnold, C. Y. 1959. The determination and significance of the base temperature
in a linear heat unit system. Am. Soc. Hort. Sci. 74:430-445.

Ayres, M. P., and D. L Thomas. 1990. Alternative formulations of the
mixed-model ANOVA applied to quantitative genetics. Evolution 44:221-226.

Ayres, M. P. and S. F. MacLean, Jr. 1987. Molt as a component of insect
development: Galerucella sagittariae (Chrysomelidae) and Epim'ta autumnata
(Geometridae). Oikos 48:273-279.

Ayres, M. P., J. Suomela, and S. F. Maclean, Jr. 1987. Growth performance of
Epim’ta autumnata (Lepidoptera: Geometridae) on mountain birch: trees,
broods, and tree it brood interactions. Oecologia 74:450-457.

 

Baldwin, J. D., and H. Dingle. 1986. Geographic variation in the effects of
temperature on life-history traits in the large milkweed bug Oncopeltus
fasciatus. Oecologia 69:64-71.

Barnes, P. T., B. Holland, and V. Courreges. 1989. Genotype-by-environment
and epistatic interactions in Drosophila melanogaster: the effects of Gpdh
allozymes, genetic background, and rearing temperature on larval
development time and viability. Genetics 122:859-868.

Berger, T. A. 1986. Habitat use and reproductive ecology of the. eastern tiger
swallowtail, Papilio glaucus L. Dissertation. Rutgers Univer51ty, Newark, NJ.

Bernays, E. A., and S. J. Simpson. 1982. Control of food intake. Advances in
Insect Physiology 16:59-118.

Blau, W. 8., and P. Feeny. 1983. Divergence in larval responses to food plants
between temperate and tropical populations of the black swallowtail butterfly.

Ecological Entomology 8:249-257.

Block, W. 1990. Cold tolerance of insects and other arthropods. Phil. Trans. R.
Soc. Lond. B 326:613-633.

97

 

98

Bowers, M. D. 1986. Population differences in larval hostplant use in the
checkerspot butterfly, Euphydryas chalcedona. Entomol. exp. appl. 40:61-69.

Brues, C. T. 1924. The specificity of food-plants in the evolution of
phytophagous insects. Am. Nat. 58:127-144.

Bryant, J. P. 1981. Phytochemical deterrance of snowshoe hare browsing by
adventitious shoots of four Alaskan trees. Science 213:889-890.

Casey, T. M. 1976. Physiological responses to temperature of caterpillars of a
desert population of Manduca sexta. Comp. Biochem. Physiol. A 57:53-58.

Casey, T. M., B. Joos, T. D. Fitzgerald, M. E. Yurling, and P. A. Young. 1988.
Synchronized group foraging, thermoregulation, and growth of eastern tent
caterpillars in relation to microclimate. Phys. Zoo]. 61:372-377.

Chapman, R. F., and A. Joern. 1990. Biology of grasshoppers. Wiley, New
York.

Cheverud, J. M., M. M. Dow, and W. Luetenegger. 1985. The quantitative
assessment of phylogenetic constraints in comparative analyses: sexual
dimorphism in body weight among primates. Evolution 39:1335-1351.

Clausen, T. P., J. P. Bryant, and P. B. Reichardt. 1986. Defense of
winter-dormant green alder against snowshoe hares. J. Chem. Ecol.
12:2117-2131.

Clausen, T. P., F. D. Pruenza, E. A. Burritt, P. B. Riechardt, and J. P. Bryant.
1990. Ecological implications of condensed tannin structure: a case study. J.
Chem. Ecol. 16:2381-2392.

Darwin, C. 1859. The orign of species. John Murray, London.

Denno, R. F., and M. S. McClure. 1983. Variable plants and herbivores in
natural and managed systems. Academic Press, New York.

Dethier, V. G. 1954. Evolution of feeding preferences in phytophagous insects.
Evolution 8:33-54.

Diehl, S. R., and G. L. Bush. 1984. An evolutionary and applied perspective of
insect biotypes. Ann. Rev. Ent. 29:471-504.

Dobzhansky, T. 1955. A review of some fundamental concepts and problems of
population genetics. Cold Spring Harbor Group Symp. Quant. Biol. 2021-15.

Eanes, W. P., and R. K. Koehn. 1978. An analysis of genetic structure in the
monarch butterfly, Danaus plexippus L. Evolution 32:784-797.

 

99

Edmunds, G. P., Jr. and D. N. Alstad. 1978. Coevolution in insect herbivores
and conifers. Science 199:941-945.

Ehrlich, P. R. and P. H. Raven. 1964. Butterflies and plants: a study in
coevolution. Evolution 18:586-608.

Ehrlich, P. R., R. R. White, M. C. Singer, S. W. McKechnie, and L. E. Gilbert.
1975. Checkerspot butterflies: a historical perspective. Science 188:221-228.

Eldredge, N., and S. J. Gould. 1972. Punctuated equilibria: an alternative to
phyletic gradualism. Pages 88-115 in T. J. M. Schopf (ed.), Models in
Paleobiology, Freeman, Cooper and Company, San Franscisco.

Endler, J. A. 1977. Geographic variation, speciation, and clines. Princeton
University Press, Princeton, NJ.

Feder, M. E. 1987. The analysis of physiological diversity: the prospects for
pattern documentation and general questions in ecological physiology. Pages
38-70 in M. E. Feder, A. F. Bennett, W. W. Burggren, and R. B. Huey,
editors. New directions in physiological ecology. Cambridge University
Press, Cambridge.

Feeny, RP. 1976. Plant apparency and chemical defense. Rec. Adv. Phytochem.
:1-40.

Fox, L. R. and P. A. Morrow. 1981. Specialization: Species property or local
phenomenon? Science 211:887-893.

Fritz, R. S., and P. W. Price. 1990. A field test of interspecific competition on
oviposition of gall-forming sawflies on willow. Ecology 71299-106.

Fritz, R. S., C. F. Sachi, and P. W. Price. 1986. Competition versus host plant
phenotype and species composition: willow sawflies. Ecology 67:1608-1618.

Futuyma, D. J., R. P. Cort, and I. Noordwijk. 1984. Adaptation to host plants in
the fall cankerworm (Alsophila pometan'a) and its bearing on the evolution of
host affiliation in phytophagous insects. Am. Nat. 123:287-296.

Futuyma, D. J., and G. Moreno. 1988. The evolution of ecological specialization.
Ann. Rev. Ecol. Syst. 19:207-233.

Futuyma, D. J. and S. C. Peterson. 1985. Genetic variation in the use of
resources by insects. Ann. Rev. Entomol. 30:217-238.

Futuyma, D. J., and T. E. Philippi. 1987. Genetic variation and covariation in
responses to host plants by Alsophila pometan'a (Lepldoptera: Geometridae).
Evolution 41:269-279.

 

 

100

Gordon, G. T. 1968. Quantitative aspects of insect nutrition. Am. Zool.
8:131-138.

Grafius, E., and N. H. Anderson. 1979. Population dynamics, bioenergetics, and
role of Lepidostoma quercina Ross (Trichoptera: Lepidostomatidae) in an
Oregon woodland stream. Ecology 60:433—441.

Grossmueller, D. W. and R. C. Lederhouse. 1985. Oviposition site selection: an
aid to rapid growth and development in the tiger swallowtail butterfly, Papilio
glaucus. Oecologia 66:68-73.

Guelz, P. G., E. Mueller, and B. Moog. 1988. Epicuticular leaf waxes of Tilia
tomentosa Moench. and Tilia europaea L Tiliaceae. Z. Naturforsch. Sect. C
Biosci. 43:173-176.

Hagen, R. H. 1989. Population structure and differential host use in hybridizing
subspecies of Papilio glaucus (Lepdidoptera: Papilionidae). Evolution
44:1914-1930.

Hagen, R. H. 1990. Population structure and differential host use in hybridizing
subspecies of Papilio glaucus (Lepidoptera: Papilionidae). Evolution
44:1914-1930.

Hagen, R. H., R. C. Lederhouse, J. L. Bossart, and J. M. Scriber. 1991. Papilio
canadensis and P. glaucus are distinct species. J. Lep. Soc. in press.

Hagen, R. H., and J. M. Scriber. 1989. Sex-linked diapause, color, and allozyme
loci in Papilio glaucus: linkage analysis and significance in a hybrid zone. J.
Heredity 80:179-185.

Hagen, R. H., and J. M. Scriber. 1991. Systematics of the Papilio glaucus and P.
troilus species groups (Lepidoptera: Palilionidae): inferences from allozymes.
Ann. Ent. Soc. Am. 84:380-395.

Haldane, J. B. S. 1931. A mathematical theory of natural and artificial selection.
Part 8. Proc. Cambridge Philosophical Society 27:137-142.

Hare, D. J., and G. G. Kennedy. 1986. Genetic variation in plant-insect
associations: survival of Leptinotarsa decemlineata populations on Solanum
carolinense. Evolution 40:1031-1043.

Haukioja, E. and S. Hanhimaki. 1985. Rapid wound-induced resistance in white
birch (Betula pubescens) foliage to the geometrid Epim'ta autumnata: a
comparison of trees and moths within and outside the outbreak range of the
moth. Oecologia 65:223-228.

 

 

101

Haukioja, E. and S. Neuvonen. 1985. Induced long-term resistance of birch
foliage against defoliators: defensive or incidental? Ecology 66:1303-1308.

Hedin, P. A., W. L. Parrott, J. N. Jenkins, J. E. Mulrooney, and J. J. Menn. 1988.
Elucidating mechanisms of tobacco budworm resistance to allelochemicals by
dietary tests with insecticide synergists. Pest Biochem. Physiol. 32:55-61.

Herms, D. A. 1991. Variation in the resource allocation patterns of paper birch:
evidence for physiological tradeoffs among growth, reproduction, and defense.
Dissertation. Michigan State University, East Lansing, MI.

Hickok, L G., and J. C. Anway. 1972. A morphological and chemical analysis of
geographical variation in Tilia of eastern North America. Brittonia 24:2-8.

Hochachka, P. W., and G. N. Somero. 1973. Strategies of biochemical
adaptation. W. B. Saunders Co., Philadelphia, PE.

Horsley, S. B., and J. Meinwald. 1981. Glucose 1 benzoate and prunasin from
Prunus serotina. Phytochemistry 20:1127-1128.

Horton, D. R., J. L. Capinera, and P. L. Chapman. 1988. Local differences in
host use by two populations of the colorado potato beetle. Ecology
69:823-831.

Hsiao, T. H. 1978. Host plant adaptations among geographic populations of the
Colorado potato beetle. Ent. Exp. Appl. 24:237-247.

Huey, R. B. 1987. Phyologeny, history, and the comparative method. Pages
76-101 in M. E. Feder, A. F. Bennett, W. W. Burggren, and R. B. Huey,
editors. New directions in ecological physiology, Cambridge University Press,
Cambridge, UK

Hurlbert, S. H. 1984. Pseudoreplication and the design of ecological field
experiments. Ecological Monographs 54:187-211.

Jaenike, J. 1990. Host specialization in phytophagous insects. Ann. Rev. Ecol.
Syst. 21:243-273.

James, A. C., J. Jakubczak, M. P. Riley, and J. Jaenike. 1988. On the causes of
monaphagy in Drosophila quinan'a. Evolution 42:626-630.

Kacser, H., and J. A. Burns. 1981. The molecular basis of dominance. Genetics
97:639-666.

Karban, R. 1986. Interspecific competition between folivorous insects on
En'geron glaucus. Ecology 67:1063-1072.

102

Kowalczyk, B., and W. Olechnowicz. 1988. Study of Fraxinus excelsior L. leaves.
I. Phenolic acids and flavonoids. Herba Polonica 34:7-14.

Kukal, 0., M. P. Ayres, and J. M. Scriber. 1992. Cold tolerance of pupae in
relation to the distribution of swallowtail butterflies. Can. J. Zool., in press.

Kukal, O., B. Heinrich, and J. G. Duman. 1988. Behavioral thermoregulation in
the freeze-tolerant arctic caterpillar, Gynaeophora groenlandica. J. Exper.
Biol. 138:181-193.

Lamb, R. J., and G. H. Gerber. 1985. Effects of temperature on the
development, growth, and survival of larvae and pupae of a north-temperate
chrysomelid beetle. Oecologia 67:8-18.

Lawton, J. H., and D. R. Strong, Jr. 1981. Community patterns and competition
in folivorous insects. Am. Nat. 118:317-338.

Lederhouse, R. C. 1982. Factors affecting equal catchability in two swallowtail
butterflies, Papilio polyxenes, and P. glaucus. Ecol. Entom. 72379-383.

Lederhouse, R. C., M. P. Ayres, J. K. Nitao, and J. M. Scriber. 1992.
Differential use of lauraceous hosts by swallowtail butterflies, Papilio troilus
and P. palamedes (Papilionidae). Oikos, in press.

Lederhouse, R. C., M. P. Ayres, and J. M. Scriber. 1990. Adult diet affects male
virility in Papilio glaucus. Functional Ecology 4:743-751.

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

Lindroth, R.L 1989a. Biochemical detoxification: mechanism of differential tiger
swallowtail tolerance to phenolic glycosides. Oecologia 81:219-224.

Lindroth, R.L. 1989b. Host plant alteration of detoxification activity in Papilio
glaucus glaucus. Entomol. exp. appl. 50:29-35.

Lindroth, R. L., M. T. S. Hsia, and J. M. Scriber. 1987. Seasonal patterns in the
phytochemistry of three Populus species. Biochemical Systematics and
Ecology 15:681-686.

MacLean, S. F., Jr. 1983. Life cycles and the distribution of psyllids
(Homoptera) in arctic and subarctic Alaska. Oikos 40:445-451.

Malcolm, S. B., and L. P. Brower. 1989. Evolutionary and ecological
implications of cardenolide sequestration in the monarch butterfly.
Experientia 45:284-294.

 

103

Matsuki, M., and S. F. MacLean, Jr. 1990. The effect of temperature on the
molt of Dineura viridorsata (Hymenoptera, Tenthredinidae). Rep. Kevo
Subarctic Res. Stat. 21:21-25.

Mattson, WJ. 1980. Herbivory in relation to plant nitrogen content. Ann. Rev.
Ecol. Syst. 11:119-161.

Mayr, E. 1963. Animal species and evolution. Harvard University Press,
Cambridge, MA.

McClure, M. S. 1989. Importance of weather to the distribution and abundance
of introduced adelgid and scale insects. Agricultural and Forest Meteorology
47:291-302.

Mitchell, J. F. B., C. A. Senior, and W. J. Ingram. 1989. C02: a missing
feedback? Nature 341:132-134.

Moran, N. 1981. Intraspecific variability in herbivore performance and host
quality: a field study of Uroleucon caligatum (Homoptera: Aphididae) and its
Solidago hosts (Asteraceae). Ecol. Entomol. 6:301-306.

Ng, D. 1988. A novel level of interactions in plant-insect systems. Nature
334:611-613.

Nitao, J. K., M. P. Ayres, R. C. Lederhouse, and J. M. Scriber. 1991. Larval
adaptation to Lauraceae hosts: geographic divergence in the spicebush
swallowtail. Ecology 72:1428-1435.

Pagel, M. D., and P. H. Harvey. 1988. Recent developments in the analysis of
comparative data. Quart. Rev. Biol. 63:413-440.

Palo, R. T. 1984. Distribution of birch (Betula spp.), willow (Salix spp.), and
poplar (Populus spp.) secondary metabolites and their potential role as
chemical defense against herbivores. J. Chem. Ecol. 10:499-520.

Pawlowska, I... 1983. Flavonoids from the leaves of some American species of
the genus Betula. Acta Soc. Bot. Pol. 52:295-300.

Powers, D. A. 1987. A multidisciplinary approach to the study of genetic
variation within species. Pages 102-130 in M. E. Peder, A. F. Bennett, W. W.
Burggren, and R. B. Huey, eds. New Directions in Physiological Ecology.
Cambridge University Press, Cambridge.

Rausher, M. D. 1982. Population differentiation in Euphydryas editha butterflies:
larval adaptation to different hosts. Evolution 36:581-590.

104

Rausher, M. D. 1984. Tradeoffs in performance on different hosts: evidence
from within and between site variation in the beetle, Deloyala guttata.
Evolution 38:582-595.

Reichardt, P. B., J. P. Bryant, B. R. Mattes, T. P. Clausen, F. S. Chapin III, and
M. Meyer. 1990. Winter chemical defense of Alaskan balsam poplar against
snowshoe hares. J. Chem. Ecol. 16:1941-1960.

Reilly, C. C., J. H. Edwards, and W. R. Okie. 1986. Isolation and
characterization of endogenous inhibitors of nitrate reductase of peach Prunus
penica and their distribution in the genus Prunus. J. Plant Nutr. 9:1335-1351.

Rhoades, DE, and R.G. Cates. 1976. Toward a general theory of plant
antiherbivore chemistry. Rec. Adv. Phytochem. 10:168-213.

Ritland, D. B. and J. M. Scriber. 1985. Larval developmental rates of three
putative subspecies of tiger swallowtail butterflies, Papilio glaucus, and their
hybrids in relation to temperature. Oecologia 65:185-193.

Sarnthoy, 0., P. Keinmeesuke, N. Sinchaisri, and F. Nakasuji. 1989.
Development and reproductive rate of the diamondback moth Plutella
xylostella from Thailand. Appl. Entomol. 2001. 24:202-208.

SAS Institute Inc. 1985. SAS User’s Guide: Statistics, Version 5 Edition. SAS
Institute Inc., Cary, NC.

Scholander, P. F., W. Flagg, V. Walters, and L. Irving. 1953. Climatic adaptation
in arctic and tropical poikilotherms. Physiological Zoology 26:67-92.

Schultz, J. C. 1983. Impact of variable plant defensive chemistry on susceptibility
of insects to natural enemies. Am. Chem. Soc. Symp. Ser. 208:37-54.

Scriber, J. M. 1973. Latitudinal gradients in larval specialization of the world
Papilionidae (Lepidoptera). Psyche 80:355-373.

Scriber, J. M. 1983. Evolution of feeding specialization, physiological efficiency,
and host races in selected Papilionidae and Saturniidae. Pages 373-412 in
R.F. Demo and MS. McClure (eds.), Variable plants and herbivores in
natural and managed systems. Academic Press, New York.

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

Scriber, J. M. 1986. Origins of the regional feeding abilities in the tiger
swallowtail butterfly: ecological monophagy and the Papilio glaucus australis
subspecies in Florida. Oecologia 71:94-103.

105

Scriber, J. M. 1988. Tale of the tiger: Beringial biogeography, binomial
classification, and breakfast choices in the Papilio glaucus complex of
butterflies. Pages 241-301 in K. C. Spencer (Ed.) Chemical Mediation of
Coevolution, Academic Press, San Diego, CA.

Scriber, J. M., and M. P. Ayres. 1990. New foodplant and oviposition records for
the tiger swallowtail butterfly, Papilio glaucus canadensis in Alaska and
Michigan. Great Lakes Entomol. 23:145-147.

Scriber, J. M. and R. C. Lederhouse. 1983. Temperature as a factor in the
development and feeding ecology of tiger swallowtail caterpillars, Papilio
glaucus (Lepidoptera). Oikos 40295-102.

Scriber, J. M., R. C. Lederhouse, and R. H. Hagen. 1991. Foodplants and
evolution within Papilio glaucus and Papilio troilus species groups
(Lepidoptera: Papilionidae). Pages 341-373 in P. W. Price, T. M. Lewinsohn,
G. W. Fernandes, and W. W. Benson. Herbivory: tropical and temperate
perspectives. John Wiley, NY.

Scriber, J. M. and F. Slansky Jr. 1981. The nutritional ecology of immature
insects. Ann. Rev. Entomol. 26:183-211.

Service, P. 1984. Genotypic interactions in an aphid-host plant relationship:
Uroleucon rudbeckiae and Rudbeckia laciniata. Oecologia 61:271-276.

Stearns, S. C. 1980. A new view of life-history evolution. Oikos 35:266-281.

Sved, J. A., and 0. Mayo. 1970. The evolution of dominance. Pages 289-316 in
K. Kojima (ed.), Mathematical topics in population genetics, Springer-Verlag,
New York. .

Sweeney, B. W., and R. L. Vannote. 1978. Size variation and the distribution of
hemimetabolous aquatic insects: two thermal equilibrium hypotheses. Science
200:444-446.

Tauber, C. A., M. J. Tauber, B. Gollands, R. J. Wright, and J. J. Obrycki. 1988.
Preimaginal development and reproductive responses to temperature in two
populations of the Colorado potato beetle (Coleoptera: Chrysomelidae).
Ann. Entomol. Soc. Am. 81:755-763.

Tavormina, S. J. 1982. Sympatric genetic divergence in the leaf-mining insect
Liriomyza brassicae (Diptera: Agromyzidae). Evolution 36:523-534.

Taylor, F. 1981. Ecology and evolution of physiological time in insects. Am. Nat.
117:1-23.

 

 

 

106

Taylor, P. S., and E. J. Shields. 1990. Development of the armyworm
(Lepidoptera: Noctuidae) under fluctuating temperature regimes. Environ.
Entomol. 19:1422-1431.

Thompson, J. N. 1988. Variation in interspecific interactions. Ann. Rev. Ecol.
Syst. 19:65-87.

Thorpe, W. H. 1930. Biological races in insects and allied groups. Biol. Rev.
5:177-212.

Vannote, R. L, and B. W. Sweeney. 1980. Geographic analyis of thermal
equilibria: a conceptual model for evaluating the effect of natural and
modified thermal regimes on aquatic insect communities. Am. Nat.
115 :667-695.

Via, S. 1984a. The quantitative genetics of polyphagy in an insect herbivore. I.
genotype-environment interaction in larval performance on different host
species. Evolution 38:881-895.

Via, S. 1984b. The quantitative genetics of polyphagy in an insect herbivore. II.
Genetic correlations in larval performance within and among host plants.
Evolution 38:896-905.

Via, S. 1990. Ecological genetics and host adaptation in herbivorous insects: the
experimental study of evolution in natural and agricultural systems. Ann.
Rev. Entomol. 35:421-446.

Waldbauer, G. P. 1968. The consumption and utilization of food by insects.
Advances in Insect Physiology 5:229-288.

Wallace, B. 1968. Polymorphism, population size, and genetic load. Pages
87-108 in R. C. Lewontin, editor. Population biology and evolution. Syracuse
University Press, Syracuse, NY. .

Wanntorp, H.-E., D. R. Brooks, T. Nilsson, S. Nylin, F. Ronquist, S. C. Stearns,
and N. Wedell. 1990. Phylogenetic approaches to ecology. Oikos
57:119-132.

Watanabe, N. 1978. An improved method for computing heat accumulation
from daily maximum and minimum temperatures. Appl. Ent. 2001. 13:44-46.

Webb, T., III, E. J. Cushing, and H. E. Wright, Jr. 1983. Holocene changes in
the vegetation of the Midwest. Pages 142-165 in H. E. Wright, Jr., editor.
Late quartemary environments of the United States, Vol. 2, The Holocene.
Unversity of Minnesota Press, Minneapolis, MN.

Weber, G. 1985. Genetic variability in host plant adaptation of the green peach
aphid, Myzus persicae. Entomol. exp. appl. 38:49-56.

 

 

107

Weiss, S. B., D. D. Murphy, and R. R. White. 1988. Sun, slope, and butterflies:
topographic determinants of habitat quality for Euphydryas editha. Ecology
69:1486-1496.

Wigley, T. M. L 1985. Impact of extreme events. Nature 316:106-107.

Wright, S. 1977. Evolution and the genetics of populations, Vol. 3:
Experimental results and evolutionary deductions. University of Chicago
Press, Chicago.

 

 

.‘|

 

 

 

Z

til-tint!”

w m .nrv‘

 

,

Winn...

       

.1-

 

HICHIGQN STQTE UNIV LIBRQR IE

919119 11 1111199919

9630089