Chemical interactions among milkweed plants (asclepiadaceae) and lepidopteran herbivores


Material Information

Chemical interactions among milkweed plants (asclepiadaceae) and lepidopteran herbivores
Physical Description:
xi, 147 leaves : ill. ; 28 cm.
Cohen, James A., 1952-
Publication Date:


Subjects / Keywords:
Asclepiadaceae   ( lcsh )
Milkweeds   ( lcsh )
Lepidoptera   ( lcsh )
bibliography   ( marcgt )
theses   ( marcgt )
non-fiction   ( marcgt )


Thesis (Ph. D.)--University of Florida, 1983.
Includes bibliographical references (leaves 134-146).
Statement of Responsibility:
by James A. Cohen.
General Note:
General Note:

Record Information

Source Institution:
University of Florida
Rights Management:
All applicable rights reserved by the source institution and holding location.
Resource Identifier:
aleph - 000496943
notis - ACR6170
oclc - 12060805
System ID:

Full Text








It is a pleasure to acknowledge the many friends and colleagues

whose aid and cooperation have been instrumental in carrying out this

work. My major professor, Lincoln P. Brower, has been a constant

source of stimulation, guidance, and encouragement. His contagious

enthusiasm for biological problems is truly inspirational. To my good

friend Lincoln I extend my sincerest thanks. The members of my

graduate and examining committees, Drs. H. Jane Brockmann, Richard A.

Kiltie, Frank Slansky Jr., and Norris H. Williams, have provided much

willing assistance during the research phase and many helpful

editorial suggestions, for which I am grateful. Drs. C. Covell, D.

Griffin, D. Habeck, J.N. Seiber, and S. Yu provided further advice and


I am also fortunate to have had the cooperation and aid of Dr.

Norm Leppla and the staff of the USDA Insect Attractants, Behavior,

and Basic Biology Research Laboratory. Drs. K.S. Brown and D.J.

Futuyma kindly provided unpublished manuscripts. The hospitality and

assistance of Dr. R. Rutowski in collecting queen butterflies is also

gratefully acknowledged. Assistance in the laboratory was kindly

provided by J. Frey, T. van Hook, and M. Hoggard. Figures were

prepared by P. Ibarra and W. Adams.

I also wish to thank the Department of Zoology and the Graduate

School of the University of Florida for generous financial support

during my graduate program. Numerous other graduate students and

post-doctoral associates acted as sounding-boards for my ideas and


provided an ideal atmosphere for pursuing these studies. Among these

I would especially like to thank J.B. Anderson, W.H. Calvert, S.B.

Malcolm, P.G. May, and N.E. Stamp. J.R. Lucas helped me greatly to

become computer-literate. He and S.A. Frank further instructed me in

the gentle art of statistics.

I reserve a separate paragraph to thank Craig S. Hieber (UVPD)

for engaging in countless late-night ravings with me and for an

infinite willingness to generate and explore all kinds of hypotheses.

I could not have asked for a finer office-mate, nor for a better


Finally, a word of thanks to my parents, Isaac and Marie Cohen,

for assistance in ways far too numerous to mention.


ACKNOWLEDGMENTS ...................... ................. o........... ii

LIST OF TABLES .................................. ........... ........ vi

LIST OF FIGURES .........................................*********** vii

ABSTRACT......................................................***********...... ix





INTRODUCTION.................................... ...... 1

Plant/Herbivore Coevolution........................ .4
Specialization and Generalization in Herbivory.......12
Chemical Defense of Milkweed Plants.................16
The Metabolic Cost of Cardenolide Ingestion.........22


Introduction.......................... *....*.*.....* 32
Methods.................... .. *.... *... .. ... ......... 34
Results .............................***. ........ 35
Discussion............... ............ .************......38

IMPLICATIONS ...........................................44

Introduction............................... ......********44
Methods ................ ................... ....... .. 47
Discussion.......... ......................*.****** 59

(CYCNIA TENERA; ARCTIIDAE) ............................74

Introduction........................... ** *******74
Methods....... ...... .. ... ... ..*** ****** *****.76
Results..... ..... ......*......* **.. .. .. ..* ***** ** 79
Discussion. .................... .. .. .. .. .. .. .84


Introduction ....................................... 91
Methods ............ .......... ..................... 93
Results ..... .. .......................... 98
Discussion......................* ...**.**.********* 103


General Discussion................................. 112
Conclusion................... ...... ...... ..... 130

LITERATURE CITED................................................ .. 134

BIOGRAPHICAL SKETCH...................... .......................... 147


Table 1. A survey of midstem leaf cardenolide concentrations
for six milkweed species in north-central Florida..................21

Table 2. Eggs, larvae, larval success, and cardenolide
concentrations for 10 Asclepias humistrata plants.................36

Table 3. Cardenolide concentrations for 7 matched sets of
partially eaten and uneaten Asclepias humistrata leaves............39

Table 4. Correlation coefficients for monarchs and plant
area, and for monarchs and cardenolide concentration..............40

Table 5. Right wing lengths, weights, fat, and cardenolide
contents of wild-caught queen and monarch butterflies from
three sites in Florida.................... .....................**53

Table 6. Spearman correlation coefficients for cardenolide
concentration vs. body size, weight, and fat contents of
wild-caught queen and monarch butterflies from Florida............58

Table 7. Body sizes, weights, fat and cardenolide contents
of queens and monarchs (Dominican Republic stock) reared in
the laboratory on the milkweed, Asclepias humistrata..............60

Table 8. Spearman correlation coefficients for cardenolide
concentration vs. body size, weight, and fat content of
queen and monarch butterflies (Dominican Republic stock)
reared in the laboratory on Asclepias humistrata..................61

Table 9. Means and standard deviations for the weight and
cardenolide contents of adult Cycnia tenera reared on the
milkweeds Asclepias humistrata and A. tuberosa................... 80

Table 10. Relative growth rate, relative consumption rate,
and efficiency of conversion of ingested matter for fourth
instar monarch butterfly larvae reared on four Asclepias
tuberosa-based diets incorporating different cardenolide
concentrations ....................................... .....* *...... 99

Table 11. Relative growth rate, relative consumption rate,
and efficiency of conversion of ingested matter for fifth
instar fall armyworm larvae reared on artificial diets
incorporating varying amounts of cardenolide.................... 101

Table 12. Additional data for fall armyworms reared on
artificial diets incorporating varying amounts of cardenolide....102

Table 13. Relative growth rate, relative consumption rate,
and efficiency of conversion of ingested matter for fifth
instar velvetbean caterpillars reared on artificial diets
incorporating varying amounts of cardenolide.....................104

Table 14. Additional data for velvetbean caterpillars
reared on artificial diets incorporating varying amounts
of cardenolide .............................. ....******* .......... 105



Figure 1. The number of monarch eggs and larvae of each
instar on 10 Asclepias humistrata plants, pooled over
11 observation days.............................................. 37

Figure 2. Frequency distributions of cardenolide
concentration for three populations of Florida queen
butterfly, and one population of monarchs.........................51

Figure 3. Frequency distributions of total cardenolide
per individual butterfly.......................................... 52

Figure 4. Thin layer chromatogram of wild-caught monarchs
and queens from Miami area........................................ 55

Figure 5. Chromatograms of adult monarchs and queens
reared in the laboratory on Asclepias humistrata .................63

Figure 6. An adult male Cycnia tenera and a late-instar larva.....77

Figure 7. TLC profiles of four adult Cycnia tenera reared
on either Asclepias humistrata or A. tuberosa....................83


Abstract of Dissertation Presented to the Graduate School
of the University of Florida in Partial Fulfillment
of the Requirements for the Degree of
Doctor of Philosophy



James A. Cohen

December, 1983

Chairman: Lincoln P. Brower
Major Department: Zoology

Milkweed plants are widely believed to be chemically defended

against insect herbivory by steroidal compounds known as cardenolides

(cardiac glycosides). Moreover, it has frequently been claimed that

milkweed-specializing insects have coevolved with their hostplants by

breaching a cardenolide-barrier over the course of evolutionary time.

However, there has been, until now, no evidence that ingested

cardenolides have any defensive properties against any invertebrate

herbivore. Thus, cardenolide-based models of coevolution among

milkweeds and insects have had little foundation. This study seeks to

provide such a foundation by comparing the ecological and

physiological effects of cardenolides upon lepidopteran species which

do or do not feed on milkweeds.

In accordance with evolutionary predictions, the results provide

little evidence that milkweed-specializing insects pay metabolic costs

of handling cardenolides. These plant chemicals appear to influence

neither oviposition nor larval development in wild monarch butterflies

(Danaus plexippus; Danaidae), nor to affect (even when in abnormally

high concentrations) consumption, growth, or food conversion

efficiency of fourth instar monarchs in the laboratory.

In contrast, the polyphagous fall armyworms (Spodoptera

frugiperda; Noctuidae) exhibited reduced growth, food consumption, and

food conversion efficiency in the fifth instar, even when fed

moderate, natural cardenolide concentrations. Although no such

negative effects were found for fifth instar velvetbean caterpillars

(Anticarsia gemmatalis; Noctuidae), a species that specializes on

cardenolide-free plants (i.e., legumes), younger instars of this

species were severely affected by cardenolides. This serves as a

warning to other researchers who frequently test only older instars

and may thereby generate misleading results.

This study is the first to demonstrate that ingested cardenolides

may provide a chemical barrier to herbivory by certain insects.

However, a detailed review of the literature provides little support

for frequent contentions that milkweed-specialists were once deterred

by, and subsequently breached, such a barrier. An alternative

coevolutionary model is considered which predicts (inter alia) that

certain danaid and ithomiid butterflies should utilize cardenolides as

hostplant recognition cues. Although my data do not confirm this

prediction for monarch butterflies, this latter model nevertheless

holds considerable promise as a prime example of plant/herbivore




A major thrust of modern ecology involves the study of

relationships between heterotrophic organisms and their food supplies.

Such interactions (e.g., predator vs. prey, host vs. parasite) occur

not only in ecological time but also in evolutionary time (e.g.,

Dawkins and Krebs, 1979; Thompson, 1982). Natural selection may be

presumed to act within populations to optimize simultaneously the

ability to acquire resources (e.g., Pyke et al., 1977; Krebs, 1978)

and the ability to avoid becoming the acquired resources of similarly

selected organisms of higher trophic levels (e.g., Edmunds, 1974;

Rhoades, 1979). Clearly, then, a major component of ecological

research must concern the study of the varied defenses utilized by

organisms to avoid being attacked by their enemies.

In this context, plants may be viewed as subject to the attacks

of both predators and parasites. Predation occurs when the plant is

killed outright by another organism(s) and parasitism occurs when, as

in most cases of herbivory, some of the plant's biomass is consumed,

lowering plant fitness, but the plant does not die in the process

(Gilbert, 1979; Price, 1980). The theory of plant/herbivore

coevolution (Ehrlich and Raven, 1964) has provided a theoretical

framework for the organization of a large and diverse body of

literature concerning the ability of plants to avoid enemy-attack, as

well as the counter-adaptations of attacking species. In essence,

plants have evolved a variety of defensive adaptations (e.g.,

morphological, phenological, and chemical traits) against predators

and parasites. In turn, such enemies have evolved means to circumvent

or counteract these defenses, leading to further elaboration of plant

defensive strategies, subsequent enemy offenses, and so on.

A key system, widely regarded as providing the classic, textbook

example of plant/herbivore coevolution, is that of milkweed plants

(Asclepiadaceae) and their specialized herbivores, most notoriously

the monarch butterfly (Danaus plexippus; Danaidae). Consider the

following excerpts from Harborne's (1983) text, Ecological

Biochemistry (emphases added):

What is now the classic example of plant-animal
coevolution in which secondary compounds have a
key role is the interaction between milkweeds,
monarch butterflies, and blue jays. [p. 87]

The milkweed produces several cardiac glycosides
within its tissues as a passive defense against
insect feeding. [p.87]

The monarch butterfly caterpillar learns to adapt
to these toxins. [p. 88]

production of the toxins continues to provide the
plant with protection not only from most insects
but also from all grazing animals. [p. 89]

The present dissertation is concerned precisely with this interaction,

for several aspects of Harborne's summary have never been

substantiated by data, despite their general acceptance among

evolutionary biologists. In particular, I address the question of

whether the cardiac glycosides (cardenolides) present in certain

milkweed plants may, in fact, offer protection against generalist-

feeding insects, but not against purportedly coevolved specialists

such as monarchs. The crucial question is: Do generalists pay "costs"

of ingesting cardenolides that specialists do not pay? This is the

first, basic step that must be established before one can entertain

claims that milkweed-specialists have breached a cardenolide-barrier

during the course of coevolution.

In the remainder of this chapter, I shall briefly outline the

theory of plant/herbivore coevolution and the related question of

chemical defense in plants. As it is supposed that a breaching of

such defenses leads to the evolution of feeding specialization (e.g.,

Ehrlich and Raven, 1964; Feeny, 1975), I will then contrast the two

strategies of specialization and generalization, with emphasis on

biochemical (detoxification and tolerance) differences. By way of

introduction to the specific subject matter of this dissertation, I

shall then review the chemical defense of milkweed plants in

particular, focusing on cardenolides, and finally summarize what is

currently known about the metabolic costs, to specialists or

generalists, of ingesting these purportedly defensive chemicals.

Plant/Herbivore Coevolution

Ehrlich and Raven's (1964) formulation of plant/herbivore

coevolution was derived primarily from considerations of larval

feeding specialization in the Lepidoptera, the most thoroughly studied

order of insect herbivores. These authors summarized the large body

of observations showing that certain lepidopteran taxa (frequently

subfamilies or tribes) were restricted to feeding upon specific plant

taxa, and suggested that these patterns of relationships were most

easily interpreted in terms of the allelochemistry of the plants.

Allelochemics are plant compounds apparently not directly required for

the primary metabolic functions of the plant but having a variety of

inhibitory or (in some cases) stimulatory effects upon potential or

actual herbivores, pathogens, or other plants (Fraenkel, 1959, 1969;

Whittaker and Feeny, 1971; Levin, 1976). In many cases the compounds

have potentially toxic or digestibility-reducing effects and are

presumed to have evolved in response to herbivore pressure. Ehrlich

and Raven (1964) suggested that certain plants, through mutation and

recombination, have elaborated novel chemicals which reduce plant

palatability or suitability as food for herbivores. Thus freed from

previous herbivore pressure, these plants entered a new "adaptive

zone" wherein phylogenetic radiation was possible. In the course of

evolution, certain herbivore groups, also through a fortuitous genetic

event, breached the plant's chemical defense and, in so doing, were

also able to enter (and radiate within) a new adaptive zone. Ehrlich

and Raven (1964:602) further suggested that the ability of an

herbivore to tolerate one kind of defensive chemistry reduces its

ability to tolerate other kinds of chemical defense. Thus, the

breaching of a chemical barrier is presumably followed by the rapid

evolution of herbivore feeding specialization for plants having

similar defensive chemistry (commonly confamilials). This hypothesis

is supported by the fact that when, in exceptional cases, oligophagous

lepidopterans feed on taxonomically disparate plant species, these are

frequently shown to share a defensive chemistry. In such cases the

herbivore may have evolved the ability to utilize the allelochemic for

hostplant recognition (Fraenkel, 1959; Smith, 1966; Rees, 1969; van

Emden, 1972; Nielsen, 1978).

Extending this general coevolutionary scenario, Brower and Brower

(1964) demonstrated that many of the lepidopteran families feeding on

toxic plant families are themselves unpalatable and aposematically

colored, while those feeding on relatively innocuous plants were

typically cryptic and palatable. This suggested that, following the

breaching of a plant's chemical defense, certain insects might not

only be able to tolerate the allelochemics but might store them in

their tissues for their own defensive purposes. These compounds may

even be incorporated into the tissues of some invertebrate predators

feeding upon allelochemic-storing herbivores (e.g., Malcolm, 1981).

Thus, defensive chemicals may pass through a food web, having

ramifications at several trophic levels.

Berenbaum (1983) has provided an example in which each step of

the evolutionary sequence postulated by Ehrlich and Raven (1964) can

be supported by either direct or circumstantial evidence. She makes

the following argument. (1) The production of phototoxic

furanocoumarins (the toxicity of which is induced only in the presence

of ultraviolet light) in numerous plant families could plausibly have

evolved as a fortuitous elaboration of pre-existing biosynthetic

pathways (e.g., lignin synthesis). (2) These compounds are more toxic

to certain herbivorous insects than are their precursors. (3) Within

the Umbelliferae, the one subfamily containing coumarins is far more

diverse in number of species than are two subfamilies that lack

coumarins. This is circumstantial evidence for an evolutionary

radiation following the evolution of the chemical defense. (4) Various

biochemical and behavioral means of circumventing the plant's defense

have evolved among insects. Most striking among these is the habit of

leaf-rolling among the Oecophoridae, lepidopterans that feed

exclusively on Umbelliferae. While feeding within rolled leaves, the

larva is shielded from ultraviolet light and consequent phototoxic

effects. (5) Genera of oecophorids and papilionids which are

associated with furanocoumarin-containing plants have diversified (in

species numbers) more extensively than have those which feed on

furanocoumarin-free hostplants. This is circumstantial evidence for

an evolutionary radiation following the breaching of the plant's

chemical defense.


Berenbaum's (1983) work provides one of the best-developed

examples of what has been called "diffuse coevolution" (Gilbert, 1975;

Fox, 1981; Futuyma and Slatkin, 1983), i.e., that resulting from the

simultaneous interaction of several plant and animal species. Some

workers (e.g., Janzen, 1980), however, prefer to restrict the term

(and concept) of plant/herbivore coevolution to instances in which a

close evolutionary sequence of reciprocal counter-adaptation between

one specific plant population and one specific herbivore population

("pairwise coevolution") can be demonstrated:

"Coevolution" may be usefully defined as an
evolutionary change in a trait of the individuals
in one population in response to a trait of the
individuals of a second population, followed by an
evolutionary response by the second population to
the change in the first. [Janzen, 1980:611]

By "evolutionary change" or "evolutionary response," Janzen refers to

genetic change. For example, specific alleles in wheat have been

identified which confer resistance to Hessian flies (Mayetiola

destructor), and each of which corresponds to a specific fly allele

for counter-resistance (Hatchett and Gallun, 1970). While this has

been cited as one of the clearer examples of coevolution (Futuyma,

1979; Thompson, 1982), it still does not meet Janzen's (1980) strict

criteria, since the frequencies of alleles for resistance in wheat

populations have been determined by agricultural geneticists rather

than by Hessian fly herbivory. In another well-studied case, Williams

and Gilbert (1981) have described the production by certain Passiflora

plant species of structures which mimic the eggs of, and inhibit

oviposition by, their specialized herbivores, the Heliconius

butterflies. These structures exploit the behavioral tendency of

Heliconius females not to oviposit on plants which harbor previously

laid eggs, since early-hatching larvae frequently cannibalize

subsequently laid eggs. It seems plausible that the evolution of

Heliconius feeding specialization on Passiflora, with its increased

threat to the plant's fitness, has in turn selected for the counter-

evolution of these defensive pseudo-eggs. Note that even here the

strict criterion for coevolution will only be met if the butterflies

are shown to have a specific counter-adaptation to circumvent the

plant's deception, such as the ability to discriminate mimetic eggs

from real ones. Although such discrimination has not been directly

observed (Gilbert, 1983), it may be inferred from the close degree of

resemblance of the mimetic eggs to real Heliconius eggs. Surely the

first "mimetic eggs" on Passiflora were crude bits of tissue and their

present resemblance to real eggs must have evolved gradually as the

butterflies evolved an improved ability to detect the difference (even

if they no longer can do so). This example of coevolution emphasizes

the importance of integrating direct observations in ecological time

with sound inferences about processes that may occur over evolutionary


While coevolution has proven to be, both theoretically and

empirically, an elusive and controversial concept (e.g., Jermy, 1976),

it has fostered a great deal of work on the biochemical interactions

of plants and their enemies, especially herbivores. A major

organizing theory attempting to explain the kinds of plant chemical

defenses evolved has been provided simultaneously by Feeny (1976) and

Rhoades and Cates (1976). This theory, now generally known as

apparencyy theory," assumes that ephemeral, early successional plant

species are more able than long-lived, prominent, late-successional

species to escape herbivory spatiotemporally. Such "unapparent"

species typically devote a large proportion of their energy and

nutrients to rapid growth, therefore presumably allocate less to

chemical defense and thus rely on what Feeny (1976) terms

"qualitative" chemical defenses. These are compounds which are

produced in low concentrations and have specific, highly toxic effects

against a broad range of herbivorous species. Examples of such

compounds include alkaloids, saponins, cyanogenic glycosides,

glucosinolates and non-protein amino acids (reviewed in Rosenthal and

Janzen, 1979). In contrast, long-lived, late-successional species are

less able to escape in ecological space and time, are "bound to be

found" (Feeny, 1976) by herbivores, and are therefore selected for a

greater investment in the amounts of chemicals allocated to defense.

Because some herbivores will eventually succeed evolutionarilyy) in

breaching any qualitative defense, these "apparent" plants should be

selected to contain chemicals which are sufficiently general in mode

of action as to be effective, in a dose-dependent manner, against both

polyphagous and oligophagous species. The best-known examples of such

a "quantitative" chemical defense are the tannins, a group of

polyphenol compounds which, at least in vitro, precipitate proteins

(both dietary and enzymal) and thereby inhibit digestion (e.g., Feeny,

1970; but see Bernays, 1981, and Zucker, 1983, for a growing

controversy regarding their in vivo effects).

Several criticisms can be raised against the theory of plant

apparency. First, the prediction that early-successional species

should invest in qualitative rather than quantitative defenses rests

on the assumption that qualitative defenses are less costly to produce

than quantitative ones, simply because they are effective at, and

therefore produced in, lower concentrations. However, the cost of

chemical defense cannot be assessed simply by comparing tissue

concentrations when different kinds of compounds are involved. It is

not necessarily true that it costs a plant as much to synthesize a

milligram of tannin as to produce a milligram of, say, cyanogenic

glycoside. Furthermore, some toxins (e.g., alkaloids) may be broken

down and resynthesized many times in the same plant (requiring

repeated investments of energy), while digestibility-reducers such as

tannins may be synthesized only once and retained for life (Fox,

1981). It is not yet known which of these two strategies is the more

costly (and this too will surely vary with the chemical nature of the

particular compounds involved). There may also be costs associated

with the internal transport of the compounds or the production of

special sites for their storage (Chew and Rodman, 1979). Thus, while

it is reasonable to assume that there is some cost to the plant for

chemical defense, the a priori assumption that the cost to ephemeral

species is necessarily less than that paid by persistent species is

not justified (Fox, 1981).

Second, one should also be wary of the dichotomy between

qualitative and quantitative defenses (Feeny, 1976). Toxins may well

have dose-dependent effects (e.g., glucosinolates: Chew and Rodman,

1979; sesquiterpenes: Langenheim et al., 1980) and purported

digestibility-reducers such as tannins may have little or no effect

upon adapted specialists (Fox and Macauley, 1977; Bernays et al.,


Third, apparency itself can be quite an elusive concept. While

Rhoades and Cates (1976) base their definition on ecological

longevity, growth form, seral stage, etc., Feeny (1976) includes the

host-finding ability of the herbivores as well. Rhoades (1979) has

argued that if host-finding adaptations are included as a component of

plant apparency, the entire concept of apparency is rendered circular.

This is because the intensity of selection for host-finding ability

should be greater for herbivores attacking rare or ephemeral plants

than for those attacking highly apparent plants. Once host-finding

ability is included, the apparencies (sensu Feeny) of the two plant

types converge. Thus, it is not clear that a short-lived herbaceous

plant is any less apparent to its specialized herbivores (which may

have sophisticated host-finding adaptations) than is an ecologically

more persistent plant species. Yet, in order to make predictions

about the kinds of defense to expect among various plant groups, we

need to have some relative measure of the selection pressure exerted

by the plants' enemies. Clearly, this can only be appreciated by

taking into account enemy host-finding abilities.

Given these problems with apparency theory, it may be best to

restrict its application to discussions of the defenses of related

plants facing similar guilds of enemies. If, for example, most plants

within a family utilize similar allelochemics for defense (Ehrlich and

Raven, 1964), then it may be safe to assume that those species

producing greater quantities of the allelochemic make greater

investments in chemical defense than do related species producing less

of the allelochemic. This is surely safer than assuming cost

differentials for qualitatively very different compounds in widely

divergent plant families (see above). Also, if the enemy guilds are

similar for the related plants, one might assume similar host-finding

abilities and then concentrate on making predictions about defensive

investments based on such considerations as growth habit (e.g., herbs

vs. shrubs), population density, and phenology. This approach will be

taken in a later section in discussing the chemical defense of

milkweed plants.

Specialization and Generalization in Herbivory

It has frequently been suggested that since generalist herbivores

use a broader range of resources than do specialists (by definition),

they are less susceptible to fluctuations in the availability of any

particular food species and are therefore "ecologically resilient"

(e.g., Levins, 1968; Schoener, 1970; MacArthur, 1972). They must,

however, cope with a broader variety of resource defenses and are

therefore thought to be less efficient at utilizing any one particular

resource type than are species which specialize on eating only one or

a few resource types (MacArthur, 1972; Feeny, 1975; Scriber and Feeny,

1979). One means by which generalist insects have adapted

biochemically to a broad range of dietary toxins is through the mixed-

function oxidase (MFO) system (Brattsten et al., 1977). This

microsomal system permits the rapid induction, upon contact with

specific toxins, of the specific oxidative enzymes required for their

detoxification. Krieger et al. (1971) demonstrated that the general

level of MFO activity was significantly greater in polyphagous

lepidopteran larvae than in oligophagous species.

Such biochemical adaptation is presumably effected at some

metabolic cost to the insect, i.e., requiring energy which could

instead be utilized for other fitness-promoting activities. Feeny

(1975) has conceptualized two kinds of biochemical metabolic costs

which may be incurred. The first is the cost of synthesizing

appropriate detoxification enzymes and any morphological structures

required for their storage. This is the "fixed cost" that is paid

regardless of the quantity of toxin encountered. The second kind of

cost is the "variable cost" of running the detoxification process

itself, the magnitude of which depends upon the quantity (and chemical

nature) of toxin to be handled. Feeny (1975) argues that while the

variable costs of detoxifying a particular compound are probably

similar for both specialists and generalists (since similar processes

would be required), the fixed costs should be lower for specialists

since they must construct and retain biochemical mechanisms for

dealing with a more limited range of toxic compounds. This leads to

the prediction that polyphagous species should utilize any particular

food type less efficiently than would oligophagous or monophagous


This prediction was confirmed by Auerbach and Strong (1981) who

found that specialist insects that feed on Heliconia imbricata leaves

had higher food utilization efficiencies than did more polyphagous

species also fed H. imbricata. Scriber (1979) also provided data that

support the prediction. He showed that larvae of the Lauraceae-

specializing butterfly, Papilio troilus, had food utilization

efficiencies 2 to 3 times greater than did the more polyphagous P.

glaucus when also fed a lauraceous diet. Other studies have failed to

confirm the prediction, however. For example, Futuyma and Wasserman

(1981) fed leaves of Prunus serotina to the tent caterpillars,

Malacosoma americanum (a specialist on Prunus) and M. disstria (a more

polyphagous species), finding no significant difference in their food

utilization efficiencies. [Fueling the debate further, studies by

Schroeder (1976, 1977) and Scriber and Feeny (1979) claim to show that

specialists are no more efficient food-converters than are

generalists. However, because these authors did not compare

efficiencies when the specialists and generalists were fed the same

plants, their observations are of limited value in the present


One reason why conflicting evidence may be found in studying

differences between specialists and generalists has been suggested by

Fox and Morrow (1981). These authors show that many species commonly

regarded as "generalists" nevertheless exhibit a considerable degree

of specialization at the population level. Because the evolution of

feeding adaptations would normally occur at the population, rather

than entire species, level, it is possible that such populations are

as biochemically and digestively specialized as are populations of

species which feed on a single food type throughout their ranges.

A second possible explanation for the contradictory evidence is

suggested by the work of Smiley (1978) who found that larvae of the

butterfly Heliconius erato (which is essentially monophagous on

Passiflora biflora) grew faster on P. biflora than did larvae of the

more oligophagous species H. cydno (which feeds on at least five

Passiflora species). However, another monophagous species, H.

melpomene, grew no faster on its hostplant (P. oerstedii) than did H.

cydno. Smiley (1978) suggests that H. melpomene has specialized on P.

oerstedii for ecological rather than biochemical or nutritional

reasons. These might include higher interspecific competition or

parasitism rates on other available host plant species. If such

"ecological monophagy" (Gilbert, 1979) is widespread (see also Singer,

1971), specialists may not necessarily exhibit greater food

utilization efficiencies than generalists.

The case of H. erato described above exemplifies a second type of

monophagy, in which specific digestive/biochemical adaptations for

coping with the primary and secondary chemistry of the particular host

species have apparently evolved. Gilbert (1979) terms this "coevolved

monophagy," but "evolved monophagy" might be a preferable term since

such herbivore adaptation may exist even when "coevolution" per se can

not be demonstrated.

While the basic distinction of evolved vs. ecological monophagy

is valid, it is probably best to regard these as opposite ends of an

evolutionary continuum. For example, if the selection pressures

encouraging ecological monophagy persist for sufficient time, then

there will presumably also be selection for an increase in the

efficiency with which the particular host type is utilized. Thus,

what begins as ecological monophagy may well result in evolved

monophagy. At any point in evolutionary time, we may expect to find

species (or populations) that are at different (and unknown) points

along this evolutionary path. This fact will certainly hamper our

ability to make predictions about the relative dietary efficiencies of

"specialists" and "generalists."

Chemical Defense of Milkweed Plants

The milkweed family, Asclepiadaceae (order Apocynales), provides

a good example of an early-successional, herbaceous plant group which

is fed upon by several oligophagous insect species. It is comprised

of some 1700 species of perennial herbs, vines, and shrubs (Willis,

1931). While the family is predominantly pantropical, 108 species

within the genus Asclepias alone have been described from North

America (Woodson, 1954). Milkweeds are renowned for their toxic

properties, based upon a class of C-23 steroids called cardenolides

(Kingsbury, 1964). These typically occur in glycosidal form in the

plants (but are not present in all species; Roeske et al., 1976) and

are therefore often referred to as "cardiac glycosides." The

designation, "cardiac," refers to the well-known effect of these

compounds on vertebrate heart tissue, acting through an inhibition of

the Na-K-ATPase system (Hoffman and Bigger, 1980). In therapeutic

dosages, cardenolides cause the heart to beat more slowly and

regularly, but in excessive dosages, they are lethal (Hoffman and

Bigger, 1980).

Although the toxic effects of cardenolides to many species of

domestic mammals are well-established (e.g., Detweiler, 1967), their

toxicity to invertebrate herbivores is less clear. This is a serious

gap in our knowledge because there exist several groups of herbivorous

insects that specialize on asclepiads (Duffey and Scudder, 1972;

Scudder and Duffey, 1972; Rothschild and Reichstein, 1976) and it has

frequently been inferred that these have "coevolved" with milkweed

plants after breaching the presumed cardenolide defense (see, e.g.,

Brower and Brower, 1964; Dixon et al., 1978; Harborne, 1983). Thus,

it is obviously crucial to know whether cardenolides are, in fact,

toxic to herbivorous insects. The evidence for this will be reviewed

in detail in a later section.

While qualitative variation in cardenolide contents, within and

among milkweed populations, has been noted (e.g., Brower et al.,

1982), little effort has been made to relate such variation to

ecological and evolutionary forces. Roeske et al. (1976) summarize

work on the individual cardenolides isolated and identified from

various milkweeds and show that some populations of Asclepias

curassavica and A. syriaca contain specific cardenolides not reported

from other populations (see also Malcolm, 1981). Also, in A.

eriocarpa, the relative proportions of the various leaf cardenolides

appear to vary between populations. The authors do not speculate on

the ecological significance of such differences. However, because the

toxicity of most allelochemics is positively related to their

lipophilicity (e.g., Harborne, 1983), this may suggest that those

populations containing more lipophilic cardenolides face more

resistant (more "highly coevolved"?) herbivores than do populations

having more hydrophilic (i.e., polar) compounds. This argument could

be extended to cover interspecific qualitative differences as well

(see below).

Quantitative variation in cardenolide concentration within

populations has also been described (e.g., Nelson et al., 1981; Brower

et al., 1982). One of the most variable species is Asclepias syriaca,

which may exhibit a 40-fold range of cardenolide concentration within

a single population (Roeske et al., 1976). Other populations of this

species are reportedly devoid of cardenolides altogether (e.g.,

Rothschild et al., 1975). The ecological significance of such

differences has never been studied and it would be instructive to

learn, for example, whether individuals or populations lacking or low

in cardenolide face a different set of herbivores than do those having

higher tissue concentrations. [It is possible that some plants

lacking cardenolides compensate for this by the storage of other

allelochemics. Although both saponins and alkaloids reportedly occur

in at least some milkweeds (see Brower and Brower, 1964), their

functions in these plants remain totally unstudied.]

Although nearly all milkweed species examined have leaf

cardenolide concentrations less than 1% of dry weight (Roeske et al.,

1976), interspecific variation should not be ignored. Several species

apparently lack cardenolides entirely, or contain undetectable amounts

(e.g., Gonolobus spp., Brower, 1969; A. viridiflora, Roeske et al.,

1976; Hoya, Stephanotis, Marsdenia, Tylophora, Cynanchum, Gymnema,

Rothschild and Marsh, 1978). The most complete tabulation of milkweed

cardenolide concentrations is provided by Roeske et al. (1976: Table

2) for 24 species, and well illustrates the wide interspecific

variation. The highest recorded cardenolide concentration in leaves

is 14.7% of dry weight for A. masonii from California (Roeske et al.,

1976; but this is based upon a single determination and should be


How can these differences be explained? Using apparency theory,

one might predict that the most apparent species would, other factors

being equal, tend to have the highest cardenolide concentrations,

whereas the least apparent would have the lowest. Roeske et al.

(1976), using Woodson's (1954) data on species distributions and

density, have found that those species having wide geographic

distributions and growing densely tend to have low cardenolide

concentrations, while those of more restricted distributions,

occurring in low density, have higher cardenolide concentrations.

These trends are opposite to those expected from apparency

considerations, since widespread, densely growing plants are

presumably more apparent than restricted, sparsely growing plants.

However, the abundance and density of milkweed plants have probably

changed drastically with the development of North American agriculture

(e.g., Fink and Brower, 1981), and so Woodson's (1954) biogeographic

(county record) information may no longer be valid for making

quantitative comparisons.

It is noteworthy that, of those milkweed species growing in

north-central Florida, only two of six studied produce leaf crops by

early spring (Cohen, unpubl. data), when large numbers of asclepiad-

specializing monarch butterflies re-enter eastern North America from

overwintering sites in Mexico (Urquhart and Urquhart, 1977). These

two species, Asclepias humistrata and A. viridis, tend to grow in

dense patches, and were located in each of the years 1981-1983 by the

migrants and utilized heavily as larval food sources (Cohen, pers.

observ.). Both species have relatively high leaf cardenolide

concentrations (Table 1). In contrast, the remaining four species (A.

tuberosa, A. amplexicaulis, A. tomentosa, and A. verticillata) grow

more sparsely, produce leaf crops too late for use by most migrant

monarchs, and have very low leaf cardenolide concentrations (Table 1).

While monarchs are surely not the only herbivores attacking milkweeds

in north-central Florida (see, e.g., Chapter IV), their massive

migrations into the area may provide a more acute risk than the more

regular, but lower level, herbivory by other species.

If the quantitative cardenolide differences within or among

milkweed species are indeed the result of differential risks to

herbivorous insects, it would suggest either that (1) some insects are

affected by cardenolides in a dose-dependent manner and, through

Table 1: A survey of midstem leaf cardenolide concentrations for six
milkweed species in north-central Florida. Data are expressed as
digitoxin equivalents (see Brower et al., 1972) as a percentage of dry
weight of leaf tissue.


Asclepias humistrata

A. viridis2
A. tuberosa

A. amplexicaulis4

A. tomentosa5

A. verticillata6

x (%)













1 Route 24, ca. 8 km W of Gainesville (Alachua Co.) city limit;
collected 4 May 1981.
2 ca. 1 km from Alt. Route 27, ca. 3 km W of junction Route 41,
3 Williston (Levy Co.); 18 April 1981.
Route 24, at E. city limit Bronson (Levy Co.); 2 July 1981.
Devil's Millhopper State Geological Site, NW 53 Ave, ca. 1 km W.
Gainesville city limit; 2 June 1981.
5 NW 62 St., ca. 1 km W of Gainesville city limit; 7 June 1981.
6Route 24, at E. city limit Bronson; 26 May 1981; leaves pooled from
two plants for a single cardenolide determination.

coevolution, have selected for higher defensive levels in their

hostplants, and/or (2) different insect species have different

threshold tolerances for cardenolides and those plant species having

greater concentrations may be adapted for defense against more

cardenolide-resistant herbivore species.

The Metabolic Cost of Cardenolide Ingestion

The two propositions above require that at least some herbivorous

insects be adversely affected by ingested cardenolides. Indeed, only

if it can be convincingly demonstrated that these compounds do present

a chemical barrier to herbivory by at least some species will it

become credible to think of milkweed-specialization by such

oligophagous groups as danaid butterflies (Brower, 1969; Brower et

al., 1975; Rothschild and Marsh, 1978) and lygaeid bugs (Scudder and

Duffey, 1972) as the possible result of a coevolutionary process

involving the breaching of such a barrier. Moreover, it should be

possible to demonstrate that milkweed-specialists have cardenolide-

tolerance capacities exceeding those of either polyphagous species or

species that specialize on cardenolide-free plant families.

Considerable attention has been paid to the question of whether

milkweed specialists suffer any negative physiological consequences

(which are here collectively called "metabolic costs") of ingesting,

metabolizing, detoxifying, or sequestering cardenolides. In contrast,

very little effort has been expended in testing the effects of these

compounds upon polyphagous species, despite the expectation that

generalists should be more likely than specialists to be affected by

toxins (see above; also Blau et al., 1978). Nevertheless, the first

suggestion of such a metabolic cost in an adapted specialist was made

by Brower et al. (1972), who found a progressive southward decrease in

the mean cardenolide concentration of migrating monarch butterflies

and suggested that those individuals having high cardenolide contents

may have had reduced viability or migratory ability relative to those

of lower cardenolide content. Rothschild et al. (1975) offered an

alternative hypothesis, suggesting that larvae developing in colder,

more northern locales would simply require a longer development time

than those in warmer, southern areas. As a result, they might ingest

more food and, consequently, more cardenolide. This suggestion was

later refuted by Dixon et al. (1978) who reared monarch larvae in the

laboratory at either 170 C or 280 C. Although those reared at 170

took longer to complete development, they consumed less food, and

incorporated into their tissues both the nutrients and the dietary

cardenolide more efficiently, than did larvae reared at 280. As a

consequence, rearing temperature had no significant effect upon final

cardenolide content of the insects.

A third possibility, not previously considered, is that the lower

cardenolide content of southern monarchs is due to a greater degree of

wing-scale loss during migration. It has been shown that the

cardenolide concentration in the wings is approximately twice that in

the remainder of the body (Brower and Glazier, 1975), and nearly 30

per cent of this wing cardenolide is localized in the wing-scales

(Nishio, 1980). It therefore seems possible that the further south a

monarch flies, the more scales it loses, and the lower its cardenolide

content becomes.

A second suggestion of metabolic costs of cardenolide ingestion

was provided by Brower and Moffitt (1974), who found a significant

negative correlation between the amounts of cardenolide stored by and

the sizes and dry weights of monarch females collected in

Massachusetts. The correlation approached significance for males

(which, on average, were 9 per cent lower than females in cardenolide

concentration) but disappeared entirely among both males and females

collected in California. On average, the California butterflies were

62 per cent lower in cardenolide concentration than were the

Massachusetts butterflies. It is thus possible that the purported

metabolic cost was seen only among the Massachusetts females because

only they contained sufficient concentrations to be affected.

Massachusetts males came next closest to exhibiting the effect,

perhaps because they contained the second highest cardenolide

concentrations, and California butterflies exhibited no apparent cost

because both sexes contained insufficient concentrations. Future

analyses of this problem should attempt to separate true geographic

and gender differences from cardenolide concentration differences with

which they may be correlated. It would also be valuable to learn

whether the apparent cardenolide/body size trade-off phenomenon

extends to other milkweed-feeding species, or is limited to monarchs.

Several recent studies have shown that monarchs (Erickson, 1973),

African queen butterflies (Danaus chrysippus; Smith, 1978) and

milkweed bugs (Oncopeltus fasciatus; Isman, 1977; Chaplin and Chaplin,

1981) tend to grow better when fed those plant species high in

cardenolide than when fed those low or lacking in cardenolide. These

authors (see also Blum, 1981) interpret such results as evidence

against a metabolic cost of cardenolide ingestion. The major fallacy

of such an interpretation is the confounding of cardenolide

differences among plant species with other nutritionally important

species differences. For example, other factors being equal, one

might expect those plant species high in nitrogen or water (two

important nutrients for insects; see Scriber and Slansky, 1981) to be

more desirable foodplants than those low in these nutrients. They

should therefore be under stronger selection for chemical (and other)

defense. Thus, we might expect a correlation between allelochemic

content and nutritional quality (e.g., furanocoumarins: Berenbaum,

1981), and the fact that some insects may grow better on high

cardenolide plants may be due to the correlated higher nutritional

quality of such plants, and may occur despite any metabolic costs that

may still be paid.

For this reason, a solution to the problem of metabolic costs

cannot be gained from studies in which several species of hostplant,

each having different cardenolide (and nutrient) contents, are fed to

insects. Rather, one must add purified cardenolide(s), preferably in

varying dosages, to diets of standardized nutritional quality (e.g.,

artificial diets). This has recently been done for monarchs fed

cardenolide-free milkweed leaves (Gonolobus rostratus) to which were

added increasing concentrations of the cardenolides calotropin,

uscharidin, uzarigenin, or digitoxigenin (Seiber et al., 1980). No

effects upon larval development time, food consumption, or body weight

were identified at any dose level for any of the four cardenolides.

The only suggestion of a metabolic cost was a dose-dependent melanism

occurring in larvae fed uzarigenin. In Lepidoptera, abnormal melanism

is a common indicator of stress such as crowding or starvation (Peters

and Barbosa, 1977). However, Seiber et al. (1980) housed their larvae

individually and showed that uzarigenin-fed larvae consumed as much

food as controls, so that crowding and starvation cannot explain the

result. The melanistic response occurred only at dosages 4 to 61

times greater than that found in the hostplant (A. curassavica) from

which the uzarigenin was isolated. Thus, while there may be a dose at

which a cost is paid, it is rarely (if ever) encountered in nature and

it is unlikely that the present levels of uzarigenin in A. curassavica

are attributable to herbivore pressure from monarchs.

Seiber et al. (1980) also found a quantitative regulation by

monarch larvae of the amount of cardenolide stored in their tissues.

Thus, when fed low-cardenolide diets, proportionally more cardenolide

is sequestered than when high-cardenolide diets are fed (see also

Brower et al., 1982). As a result, the variation in the total amount

of cardenolide sequestered is damped. Although the mechanism

underlying this regulation has not been investigated, one may wonder

why the insects do not simply store as much cardenolide as they

ingest, or do not excrete (or metabolize to noncardenolide compounds)

a constant fraction of the ingested cardenolide. Again, there is a

suggestion that there may be an upper limit on monarch cardenolide


Using in vitro preparations, Vaughan and Jungreis (1977) found

that monarch neuronal tissues were some 300 times less sensitive to

cardenolide- (ouabain-) induced inhibition of the Na-K-ATPase system

than were tissues of two polyphagous lepidopteran species (Hyalophora

cecropia and Manduca sexta). This study is important because it

represents the first serious attempt to ascertain whether a metabolic

cost may be paid by species not adapted as milkweed specialists. It

must be regarded with caution, however, because in vitro preparations

circumvent a good part of the potentially protective physiology of the

organisms. For example, it is possible that the digestive systems of

Hyalophora and Manduca prohibit assimilation of ingested cardenolides

into the hemolymph. If so, one would not expect the evolution of

protection at the neuronal level. Since injection bioassays are

equally unrealistic, this caveat also applies to the results of von

Euw et al. (1967) and Rafaeli-Bernstein and Mordue (1978) that

milkweed-feeding orthopteran species are less sensitive to injections

of cardenolide than are species that do not feed on milkweeds. What

remains to be done is a comparative study of the effects of ingested

cardenolide upon the growth and metabolism of both milkweed-adapted

and non-adapted herbivorous insects.

In this dissertation, I present the results of several studies

bearing on the question of the metabolic costs of cardenolide

ingestion in lepidopteran species that are either adapted or not

adapted to feeding on milkweed plants. The specialized milkweed-

feeders studied were the monarch butterfly (Danaus plexippus), the

queen (Danaus gilippus), and the dogbane tiger moth (Cycnia tenera;

Arctiidae). The fall armyworm (Spodoptera frugiperda; Noctuidae) was

studied as an example of a highly polyphagous herbivore. Finally, the

velvetbean caterpillar (Anticarsia gemmatalis; Noctuidae) was studied

as an example of a specialist species which feeds on plants lacking

cardenolide (in this case, only plants in the family Leguminosae are

eaten). The species selected represent a wide taxonomic range (i.e.,

both moths and butterflies).

In Chapter II, I discuss the hostplant selectivity and larval

success of monarch butterflies in relation to plant cardenolide

concentration. If there is, in this species, a metabolic cost of

feeding on high cardenolide diets, then ovipositing females might be

expected to have evolved a preference for lower cardenolide plants

over higher ones. Furthermore, if the costs accrue in the larval

stage, then the survival of larvae on high-cardenolide plants might be

reduced relative to that of larvae on lower-cardenolide plants.

Alternatively, if no significant cost exists, females might instead

behave in such a way as to maximize the opportunities of their

offspring to sequester defensive chemicals, i.e., higher cardenolide

plants should be preferred. Finally, if larvae on high-cardenolide

plants are better protected from parasites and predators than those on

low-cardenolide plants, a differential in larval survivorship should

be measurable.

Chapter III presents a comparative analysis of the cardenolide

contents of wild queen butterflies (D. gilippus berenice) collected

from three locations in Florida, and of monarchs from one of those

sites. Since it is known for other lepidopteran species that body

size is positively correlated with fecundity (Hinton, 1981), the

hypothesis that these danaid species pay fitness costs of cardenolide

ingestion would be supported if negative correlations between

butterfly cardenolide concentration and body size or weight were

found. In a separate experiment, also reported in Chapter III, these

two species were simultaneously reared in the laboratory on the

milkweed, Asclepias humistrata, and their cardenolide sequestration

abilities compared. Again, correlations between body size, weight,

and cardenolide content were sought.

It should be noted that even if such indications of metabolic

cost are found, this need not imply that the net benefit to the insect

is necessarily negative. Sequestered allelochemics may provide a

chemical defense that outweighs any costs paid to acquire them. As a

result, there may be a net gain in fitness. Nevertheless, a reduced

body size may be considered an investment made in order to achieve

this level of defense. It is assumed that, as adaptation to

allelochemics evolves, the magnitude of the investment required would

be reduced.

In addition to their relevance to the issue of metabolic costs,

the data on wild queens provide the first practical application of the

chromatographic cardenolide "fingerprinting" technique devised by

Brower et al. (1982) to identify the hostplants of wild-caught danaid

butterflies. The results may also shed some light on a long-standing

problem in the study of mimicry, viz., why the viceroy butterfly

(Limenitis archippus) in Florida mimics the queen instead of its usual

model, the monarch (e.g., Brower, 1958b).

In Chapter IV, I investigate the issue of cardenolide

sequestration in an oligophagous arctiid moth, Cycnia tenera. This is

an aposematic, apocynad/asclepiad-specializing species reported by

Rothschild et al. (1970) not to sequester cardenolides from its

hostplant (Asclepias syriaca in their study). This raises the

interesting question of whether, even among specialists, some species

do not sequester cardenolides because they are more sensitive to these

compounds than are sequestering species such as danaids.

Unfortunately, Rothschild et al. (1970) did not confirm that the A.

syriaca fed to the C. tenera larvae indeed contained cardenolides. It

is thus necessary first to either confirm or refute their findings,

using hostplants of known cardenolide contents. In this chapter, I

report on the growth, fat storage, and both qualitative and

quantitative aspects of the cardenolide contents of C. tenera reared

on the milkweeds, A. humistrata and A. tuberosa.

In Chapter V, I report the results of a series of gravimetric

experiments (Waldbauer, 1968) designed to test whether dietary

cardenolides per se affect consumption rates, growth rates, body size,

and other important fitness indices of larval Lepidoptera. I will

compare the metabolic responses to ingested cardenolide of a milkweed


specialist (monarchs), a polyphagous species (fall armyworms), and a

species specializing on cardenolide-negative plants (velvetbean


Chapter VI summarizes the major results presented in the

dissertation and discusses their relevance to the theories of

plant/herbivore coevolution and chemical defense.




Larvae of the monarch butterfly (Danaus plexippus'L.) feed upon

milkweed plants (Asclepiadaceae), from which they may ingest and

sequester plant defense chemicals known as cardiac glycosides or

cardenolides (Parsons, 1965; Reichstein et al., 1968; Brower, 1969;

Roeske et al., 1976). In turn, these chemicals are important in

defending the monarch against some vertebrate predators by making the

butterflies unpalatable (Brower and Brower, 1964; Brower, 1969).

Therefore, one might expect the presence, concentration, or particular

types of cardenolides present in individual milkweed plants to be

among the factors influencing a monarch female's decision to oviposit

*Published 1982 in Journal of the Kansas Entomological Society 55:343-


on a particular plant. Dixon et al. (1978) investigated this question

on an interspecific level by comparing the numbers of eggs laid on

greenhouse plants of three milkweed species differing in cardenolide

concentration. They reported an oviposition preference for the

species having the lowest cardenolide concentration (Asclepias

curassavica), although they concluded that factors such as plant age

and the presence of previously laid eggs and larvae are more important

determinants of oviposition. However, their study did not test the

independent effect of cardenolide concentration upon oviposition

preference for the following reasons: 1) milkweed species differ in

many ways other than cardenolide concentration, such as water and

nitrogen content (Erickson, 1973) or plant morphology (Woodson, 1954)

so that ovipositional preferences for a particular species cannot be

attributed to any one of these plant differences unless more

controlled studies are done; 2) the data on cardenolide concentrations

were not derived from the same individual plants upon which the

butterflies oviposited, and significant intraspecific differences in

milkweed cardenolide concentrations occur (Brower et al., 1982; and

this study); and 3) the study was carried out in captivity where

unnatural butterfly behavior commonly occurs (e.g., Dixon et al.,

1978:443, 448).

To avoid the above problems, I carried out field observations of

oviposition and larval success of wild monarchs on individual

Asclepias humistrata Walt. plants which were then analyzed for

variation in cardenolide concentration.


Ten plants of Asclepias humistrata (Asclepiadaceae), growing

along a sparse grassy roadside (Route 346) near Cross Creek, Alachua

Co., Florida, were marked for study. The total leaf area of each

plant was estimated as follows. A separate collection of 25 A.

humistrata leaves was first used in determining the relationship

between leaf length and leaf area. Lengths were measured to the

nearest 0.1 cm with a ruler and areas determined electronically (to

the nearest cm2) with a Li-Cor Model 3100 Area Meter. The regression

relationship was: area = 6.34 [length] 15.14, with an r = 0.92.

Because A. humistrata has opposite leaves of approximately equal size,

I then measured (in situ) the length of the least eaten of each pair

of leaves on the 10 marked plants. Their leaf areas were then

estimated using the regression formula, summed for each plant, and

then doubled to provide an estimate of the total plant leaf area.

I counted the total numbers of monarch eggs and larvae on each

plant on each of the following dates in April 1981: 4, 5, 6, 8, 10,

12, 14, 16, 18, 20, and 23. These total counts overestimate the

actual egg and larval numbers since each individual may be counted

more than once (i.e., on different dates). However, this problem

applies equally to all 10 plants and thus does not affect the

statistical analyses or conclusions. I therefore define larval

success as the total number of fifth instars per egg counted.

On the day final counts were made, I collected for cardenolide

analysis four midstem leaves from each of the nine surviving plants,

avoiding any partially eaten leaves. To assess whether this

collection of only uneaten leaves would bias the results (i.e., if

their cardenolide concentrations were not representative of the

plants' foliage in toto), two partially eaten and two uneaten midstem

leaves were also taken from each of seven adjacent plants. All leaves

were then sealed in plastic bags, placed on ice, and kept in a deep-

freeze until analysis in mid-July. Cardenolide concentrations

(expressed as micrograms of digitoxin equivalent per 0.1 g dry weight

of leaf) were determined by standard spectrophotometric methods (see

Brower et al., 1972, 1975).

Tests of association between variables were done with Spearman's

rank order correlation, followed by t-tests for significance.


The numbers of eggs and larvae, larval success rates, plant

sizes, and cardenolide concentrations for each plant are shown in

Table 2. Egg densities were also calculated in order to assess the

influence of cardenolide concentration, independent of plant size,

upon oviposition. It is evident from these data that a large

percentage of eggs (i=96%) failed to reach the fifth instar. Figure 1

shows that a constant (43%) proportion of the larvae appear to be

lost, presumably to mortality, during each instar.

Table 2. Eggs. larvae, larval success, and cardenolide concentrations
for 10 Asclepias humistrata plants.

total Egg
leaf Total density
No. No. area no. (eggs/
Plant stems leaves (cm2) eggs 100 cm')

1 1 12 362 8 2.21
2 2 18 322 21 6.52
3 1 11 300 7 2.33
4 1 8 202 27 13.37
5 1 8 122 5 4.10
6 11 77 1976 92 4.66
7 3 32 699 83 11.74
8 2 26 655 35 5.34
9 3 42 1250 24 1.92
10 7 64 1510 25 1.66
X 3 30 740 33 5.39
SD 3 24 630 31 4.12

Percent larval success (g. 0.1
Total number larvae (No. larvae egg) x 100 g dry
I 1I II IV V I II I1 IV V leaf)

0 50 25 13 0
2 24 38 14 14
0 43 0 0 0
0 15 0 0 0
0 60 0 0 0
7 58 38 18 11
3 25 12 6 2
0 43 20 3 0
2 17 42 21 21
3 84 80 68 44
2 42 26 14 9
2 49 44 35 29

0 438
10 205
0 560
0 602
0 415
8 279
4 374
0 496
8 601
12 *
4 441
20 140

Source: Cohen and Brower. 1982

* Not determined because plant died before termination of study.


M 100



Fig. 1. The number of monarch eggs and larvae of each instar
on 10 Asclepias humistrata plants, pooled over 11 observation
days. The least-squares regression line suggests a constant
43% mortality rate at each instar. (Source: Cohen and Brower,

Cardenolide concentrations within the plant population varied

approximately three-fold (coefficient of variation = 32%; Table 2).

Since no significant difference in the cardenolide concentrations of

uneaten and partially eaten leaves was found (Table 3), the analysis

of only uneaten leaves from the marked plants was not biased by this


Significantly more eggs were laid on larger plants, with the

consequence that egg densities did not vary with plant area (Table 4).

Fifth instar larval success was positively correlated with plant size.

However, the number of eggs laid per plant, egg densities, and larval

success rates all varied independently of the cardenolide

concentrations of the plants.


I found a positive correlation between the number of eggs on a

plant and its total leaf area. This is expected on purely statistical

grounds since larger plants present larger targets for oviposition. A

consequence of this relationship is that egg density does not vary

regularly with plant size (Tables 2 and 4). It may be surprising,

therefore, that larval success does correlate with plant size, since

this is not likely to be due to overcrowding and competition among

larvae for food or space. I suggest that larger plants, having more

overlapping stems and leaves, may offer greater protection against

Table 3. Cardenolide concentrations (ug/O.lg dry weight) for 7
matched sets of partially eaten (PE) and uneaten (UE) Asclepias
humistrata leaves.



PE 390 497 681 205 723 434 513 492 176

UE 395 639 613 231 211 807 438 476 221

t=0.15; P>0.05; Student's t-test, matched pairs design. Variances
are not significantly different: F-max=1.26; P>0.05. (Source: Cohen
and Brower, 1982)

Table 4. Correlation coefficients for monarchs and plant area, and
for monarchs and cardenolide concentration. Egg numbers and larval
success vary positively with plant area, but no correlations with
cardenolide concentration occur.



TOTAL NO. EGGS 0.66 0.04 -0.17 NS

EGG DENSITY -0.35 NS -0.20 NS

V LARVAL SUCCESS 0.69 0.03 -0.25 NS

Spearman rank order correlations; alpha=0.05.
* No cardenolide data available for plant 10 which died.
(Source: Cohen and Brower, 1982)

both biotic (e.g., parasites and predators) and physical (e.g., direct

sunlight and dessication) mortality factors. The curve in Fig. 1

suggests a constant proportional mortality rate. However, this is

based upon a mixed cohort of eggs and larvae of varying initial ages,

as well as some recruitment, summed over the 19 day period. Also,

unlike monarchs in California (Brower et al., 1982), these fifth

instars usually left their hostplants when ready to pupate, and thus a

few may have been missed. The data may therefore slightly

underestimate fifth instar larval success. Yet, despite these error

factors, it is remarkable that the data fit so well the logarithmic

relationship characteristic of Type II survivorship curves (Wilson and

Bossert, 1971).

No relationship between egg numbers or egg densities and the

cardenolide concentrations of the plants was found. This is in

contrast to the apparent oviposition preference for low-cardenolide

plant species reported by Dixon et al. (1978). Moreover, the

relatively high intraspecific variance in plant cardenolide

concentrations (Table 2), together with similar variability found by

Brower et al. (1982) for A. eriocarpa, reaffirms the necessity of

measuring concentrations in the particular plants for which

oviposition data are gathered, rather than generalizing for entire

species or even populations. Intraplant variation is also possible

(e.g., Rhoades and Cates, 1976). I have investigated the specific

hypothesis that intraindividual cardenolide differences exist between

uneaten and partially eaten leaves, but no such differences were found

(Table 3).

I measured plant sizes on 4 April but collected leaves for

analysis on 23 April. Plant sizes change over time as a function of

both herbivory and growth. Similarly, cardenolide concentrations may

vary with time (e.g., Nelson et al., 1981, for A. eriocarpa).

However, my conclusions are based on rank-order, rather than

parametric, correlations and thus require only the assumption that

size and cardenolide rankings of the plants did not change over the

course of the 19-day study. At least for plant size, this accords

with subjective observations of the plants.

Brower et al. (1982) demonstrated that monarchs reared on low-

cardenolide A. eriocarpa plants sequester proportionately more

cardenolide than do those reared on high-cardenolide plants. Thus,

the monarchs effectively regulate their own chemical concentrations,

with the resultant variation in butterfly concentrations considerably

less than that in the plants upon which they had fed. This regulatory

ability suggests that the actual cardenolide concentration of the

plant on which an egg is laid is, within certain limits, of little

significance in determining the ultimate amount of cardenolide in, and

hence probable degree of protection of, the butterflies. However,

this regulatory ability appears to break down when plant species of

extremely low cardenolide content are eaten (Brower, Seiber, and

Nelson, in prep.). Since A. humistrata is a relatively high-

cardenolide species (see Roeske et al., 1976), the possibility remains

that females would be more selective in choice of hostplant when

ovipositing on lower-cardenolide species.


The lack of correlation between plant cardenolide concentration

and monarch egg deposition patterns or larval success rates provides

evidence against the hypothesis that specialist herbivores pay

metabolic costs of feeding on cardenolide-containing plants. These

are, however, but two possible measures of such costs, utilizing only

one specialist insect species. In the next chapter, I will examine

(inter alia) the relationship between the amount of cardenolide

sequestered by two specialist species and their body sizes and

weights. Since the latter variables are correlated with fecundity in

most Lepidoptera (Hinton, 1981), this will constitute a further

(indirect) test of the metabolic cost hypothesis. Moreover, since it

is unlikely that all herbivores are equally well adapted to plant

allelochemics (perhaps reflecting different degrees of coevolution?),

the addition of a second species will help provide a comparative basis

for understanding the evolution of such adaptations.




The butterfly family Danaidae has figured prominently in the

development of the theory of plant/herbivore coevolution (Brower and

Brower, 1964; Ehrlich and Raven, 1964) and has provided one of the

best examples of the chemical defenses of insects against vertebrate

predators (Brower, 1969; Brower et al., 1982; Dixon et al., 1978;

Reichstein et al., 1968). However, much of our understanding of the

ecology and evolution of plant/danaid/predator interactions is based

upon only one species in this tropical family, the monarch butterfly

(Danaus plexippus L.; review in Brower, in press). Because this

species is known to be unique among danaids in certain other respects

(e.g., migration: Urquhart, 1960; lack of plant-derived sex

pheromones: Edgar et al., 1976; Boppre, 1978), one must question

whether monarchs are representative of the Danaidae with respect to

hostplant adaptation and chemical defense, or whether they represent

just one point (or range) within a broader spectrum of danaid

adaptations. If such a spectrum is found to exist, it may ultimately


become possible to reconstruct some of the evolutionary steps leading

to the close association between danaids and their highly toxic

milkweed hostplants (Asclepiadaceae).

Danaid species differ in the amounts of toxic cardenolides which

they sequester from their foodplants. For example, Brower et al.

(1975) showed that the cardenolide concentrations of laboratory-reared

African queen butterflies (Danaus chrysippus L.) were only about 30

per cent that of monarchs simultaneously reared on the same milkweed

hostplants (see also Brower et al., 1978; Rothschild et al., 1975).

The cardenolide concentrations of queen butterflies (D. gilippus

berenice Cramer) from Florida were, on average, about 75 per cent that

of monarchs reared in the laboratory on the same hostplants. Other

genera of danaids (e.g., Amauris, Euploea) appear to prefer as larval

hostplants those milkweed species lacking cardenolides (Rothschild and

Marsh, 1978).

Does this diversity in the storage of cardenolides reflect

differences among danaid species in their degree of adaptation to

these defensive plant allelochemics? Do queens, for example, store

less cardenolide than monarchs because they are "less coevolved" and

can tolerate these compounds less readily? Do the cardenolide

contents of various danaid species differ only quantitatively, as

described above, or are there also qualitative differences in the

particular set of hostplant cardenolides sequestered?

Here, I address these questions through a comparison of the

cardenolide contents of wild queen butterflies (D. gilippus berenice)

from three populations in Florida and of monarchs (Danaus plexippus)

from one of these populations. Following Brower and Moffitt (1974), I

will assess the "cost" of storing cardenolides by searching for

correlations between body size or weight and cardenolide concentration

of the butterflies. Since body size and weight are typically

correlated with fecundity in Lepidoptera (see Hinton, 1981), a

negative correlation with cardenolide concentration would suggest that

one component of fitness (i.e., fecundity) has been traded for another

(e.g., higher survival due to chemical defense from cardenolides).

While such a trade may well be of positive net value to the insect, it

nevertheless would require an investment or cost which, it is assumed,

should be lessened as adaptation to allelochemics evolves.

The subspecies of queen studied here is of further interest

because it is the apparent model for mimicry by the southern

subspecies of the viceroy butterfly (Limenitis archippus floridensis

Strecker), which throughout the remainder of North Anerica mimics the

monarch (Brower, 1958a,b; Klots, 1951; Remington, 1968). Thus, a

comparison of the chemical basis for defense in the queen and monarch

should aid in understanding both the selective rationale for the

switch in viceroy mimicry and the broader issues of plant/herbivore

coevolution mentioned above.


Wild queen butterflies were collected from three populations in

Florida, listed from north to south as follows: Lake Istokpoga,

Highlands County, 7 September 1981; Corkscrew Swamp Sanctuary, Collier

County, 6 September 1981; Miami, Dade County, 4 December 1981. In

addition, monarchs were collected from the Miami site where they were

sympatric with queens, and where only the milkweed Asclepias

curassavica grew abundantly. (However, a few A. incarnata plants were

also located.) These collections therefore permit a study of

population variation in cardenolide content of queens, as well as a

comparison of the relative value, with respect to cardenolides, of

monarchs and queens as models for viceroy mimicry. All butterflies

were placed on ice immediately following capture, killed by freezing,

and later dried for 16 h at 600 C. Dry weights were determined using

a Mettler AK-160 electronic balance. The right wing was removed with

forceps and the distance from the apex to the anterior notal process

was measured to the nearest 0.5 mm with hand calipers. Fat was

removed from the butterflies by petroleum ether extraction of each

entire insect for 1 h (methods in Walford and Brower, in prep.). This

procedure removes only negligible amounts of cardenolide from the

insect (Nelson and Brower, unpublished data; see also Nishio, 1980).

Lean weights were calculated by subtracting the weight of extracted

fat from the total dry weight of each insect.

Cardenolide content was determined by standard spectrophotometric

methods (Brower et al., 1972, 1975), with one modification. Soon

after beginning the spectroassay of queens from Corkscrew Swamp, it

became evident that many of the butterflies contained very little

cardenolide. In such cases it is frequently difficult to achieve a

stable absorbance reading. Consequently 0.3 ml of a 12.5 X 10- M

ethanolic solution of digitoxin was added to each butterfly extract in

the cuvette (replacing 0.3 ml of 95% ethanol; see Brower et al., 1972)

in order to artificially "boost" absorbance readings to a more stable

midrange. A pilot test demonstrated that the absorbance of the

"boost" digitoxin alone was 0.400 + 0.009 (X + SD; N=9). This mean

value was therefore subtracted from the total absorbance read for a

sample, the remainder being the absorbance due to the butterfly

extract alone.

In order to compare qualitatively the cardenolides present in

sympatric monarchs and queens from the Miami sample, those butterflies

containing a total of at least 30 ug equivalents of digitoxin (as

determined from the spectroassay) were subjected to a lead-acetate

clean-up procedure in preparation for thin-layer chromatography (TLC).

The procedure used was that described by Brower et al. (1982) with the

exception that the final solution was filtered through a Millipore

filter (Millipore Corp., Bedford, MA) rather than through a funnel of

glass wool and anhydrous sodium sulfate. This clean-up procedure

removes much of the interfering pigments and other noncardenolide

compounds from the butterfly samples. TLC was then performed and

plates developed four times in a chloroform : methanol : formamide

solvent system (90:6:1). Further details of the TLC procedure are

available in Brower et al. (1982).

In addition to the wild-caught butterflies, eggs and first instar

larvae of both monarchs and queens were collected from milkweed plants

growing in La Vega province, Dominican Republic, during July, 1981.

These were brought back to the laboratory and reared to maturity on an

exclusive diet of A. humistrata leaves, collected wild in the vicinity

of Gainesville, Florida. The leaves of this species contain

relatively high concentrations of cardenolide (see Chapter I). From

these hearings, 10 adult monarchs and 7 queens were compared for

quantitative (via spectrophotometry) and qualitative (via TLC)

differences in cardenolide storage. For comparison, two arctiid moth

species (Cycnia tenera, an apocynale specialist; and Estigmene acraea,

a highly polyphagous species; Tietz, 1972) were reared on A.

humistrata and chromatographed along with the danaids.


Wild-Caught Butterflies from Florida

The frequency distributions of cardenolide concentration and

total cardenolide per insect for the wild-caught butterflies are shown

in Figs. 2 and 3, respectively. Each distribution departs

significantly from normality (Shapiro-Wilk (W) tests; p < 0.01; Helwig

and Council, 1979). Thus, nonparametric statistical analyses were


Males and females did not differ significantly either in

cardenolide concentration or total content in any of the queen or

monarch populations studied (Table 5; Wilcoxon 2-sample tests, p>0.05

for all pairwise comparisons). Consequently, the data from both sexes

of each population were pooled for all subsequent analyses.

Cardenolide concentration varied significantly among the three

queen populations (Kruskal-Wallis H=31.60, df=2, p<0.0001), as did the

total cardenolide content per butterfly (H=30.03, df=2, p<0.0001).

Pairwise comparisons revealed that the queens from Lake Istokpoga had

significantly lower cardenolide concentrations than either those from

Corkscrew Swamp (Wilcoxon 2-sample test; Z=5.17, p<0.0001) or those

from Miami (Z=4.72, p<0.0001), but that the latter two populations did

not differ significantly from one another (p>0.05). Similar results

were found for total cardenolide content per butterfly.

Analysis of the Miami samples shows that monarchs had

significantly greater cardenolide concentrations (Wilcoxon 2-sample

test; Z=4.82, p<0.0001) and total contents (Z=6.12, p<0.0001) than the

sympatric queens. Chromatography reveals a single cardenolide profile

common to both species (Fig. 4). This consists of nine spots,

including a major one at the approximate Rf of digitoxin, and one of

higher Rf. A few individuals (of both species) exhibit all of these

spots plus an additional two faint spots of still-higher Rf (numbered

spots 10 and 11 in Fig. 4). This chromatogram is virtually identical

to that of monarchs reared in the laboratory on Asclepias curassavica

(Fig. 4 inset), providing strong evidence that this was the hostplant

utilized by the Miami monarchs and queens. To date, no other milkweed

o O
In .

in cc

Ec Q F o r!
0 0 0 0
N CD as
Q rn (\o W co o r

o to

1 -- e


O i UC

c cr

'H C
0. I 1 *

E 0 Z
3 0

it t L- : 3

F 4-0

OH 41

F 0D Z


31dVS 0 N3 3d;a_,I

S2 zlx v(n



o. Zlx j O
ci 0 zx
C ) IL () ZIX U) I


^ E < J O'


zlx U)

o LL

0 DW

- I-


- m
-n .-

- I -


a -



3ldVJVS d 0 IN3083d

0 0 0 0 0
U) v rn N -


'S 4N C4
-T -1

0l m-
vr r

r- co
Lf) LI)

m 10 r (

s r rC f


3 0


W )C


If -4



W r


l 0)

0 1-1




co a

.,-4 0O
C *H

( A
U l



0 Q
to o

1-1 E! w
tlr- ^>~
C 4-1

60 ^Q n

01 e c
<-1 ^
^3 T }-
tflC 0
E-i CO

l -N 0 0
-C4 CN C0











m to
. 0
F-4 v-,


0 +
CO --I


C0 -

C) .0

in \o



N- '.


-I 0

'0 -

> 0 60
Od H 8
Q 3M -


Q C4

Q -4









Fig. 4. Thin-layer chromatogram of wild-caught monarchs (M) and
queens (Q) from Miami area. Developed four times in chloroform:
methanol: formamide (90:6:1 v/v/v). The great concordance of all
profiles below spot #10 suggests that these animals had all fed
upon the same larval hostplant species. Note the agreement of
this profile with that of monarchs reared in the laboratory on
Asclepias curassavica (Inset), a milkweed growing abundantly in
the collection area. Since nearly all milkweed species studied
to date produce qualitatively different TLC profiles in danaids
(Brower et al., 1982, in press, and in preparation; see also Fig.
5), A. curassavica is the likely larval hostplant of the Miami
butterflies. (Inset from Brower, in press). For reference, a
1:1 (v/v) mixture of digitoxin (dig) and digitoxigenin (dgn) was
spotted in three channels. Numbers to the right of each channel
indicate the micrograms of cardenolide spotted, calculated prior
to lead-acetate clean-up.

S ..... I iii iI i : i Dll


1, -6

a*lig* ) t

bu .9Q

beaR ae .I
a ttPde*05

4* P.i



D. plexippus

D. plexippus

D.g. berenice

D. plexippus

D. plexippus

D.g. berenice

D. plexippus

D. plexippus

D. plexippus

D.g. berenice

D. plexippus
D. plexippus

A. curassavica

species is known to produce this particular chromatographic profile in

danaid butterflies (Brower et al., 1982, in press, and in


The wing lengths, dry and lean weights, and fat contents of the

butterflies are summarized in Table 5. The correlations between these

variables and cardenolide concentration are indicated in Table 6. For

both male and female monarchs there was a highly significant negative

correlation between wing length and cardenolide concentration (Table

6C). Moreover, males also showed negative correlations between

cardenolide concentration and both body weight and fat content. For

queens, none of the correlations was significant in either sex (all 3

populations pooled; Table 6A).

These statistical differences between monarchs and queens could

reflect true species differences. However, the monarchs were all

collected in the Miami region, while the queens for this analysis were

pooled from 3 areas. It is possible that an unknown geographic effect

is operating here, such that butterflies from the Miami region,

regardless of species, would show these negative correlations (e.g.,

due to a common hostplant). To test this, the queens from Miami were

also analyzed separately from the other 2 populations (Table 6B).

However, as before, no significant correlations emerged (p > 0.10 for

all tests). Thus, the observed species difference is not attributable

to a geographic difference.

It is possible that monarchs show negative correlations between

body size and cardenolide concentration, while queens do not, simply

because they store greater concentrations of these chemicals (i.e.,

queens may not store sufficient cardenolide to be adversely affected

by it). If this is the case, then the negative correlations should

vanish for that subset of monarchs having concentrations similar to

those of queens. To test this, I analyzed only those monarchs (sexes

pooled) having cardenolide concentrations equal to, or less than that

of, the most highly-concentrated queen (i.e., 226 ug/0.lg). In this

case, significant negative correlations between cardenolide

concentration and wing length, dry weight and lean weight still

occurred (Table 6D). Thus, it appears that the difference between

monarchs and queens is not due merely to geographic or cardenolide

concentration differences. It is also not due to differences in

larval hostplant species, since Fig. 4 demonstrates that both species

in Miami had most likely developed on A. curassavica. Rather, the

negative correlations appear to represent inherent species


Laboratory-Reared Butterflies (Dominican Republic Stock)

When monarchs were reared in the laboratory on Asclepias

humistrata, the females developed significantly higher cardenolide

concentrations than did males (Wilcoxon 2-sample test; Z=2.02, p<0.05;

Table 7). No such sex difference was evident for queens reared under

identical conditions (Z=0.53; p>0.50). As in the wild-caught samples,

only the monarchs showed significant (again, negative) correlations

Table 6. Spearman correlation coefficients for cardenolide
concentration vs. body size, weight, and fat content of wild-caught
queen and monarch butterflies from Florida. The data for queens are
first shown for all three populations pooled (A), and then for the
Miami sample separately (B). Monarchs were first analyzed using the
entire data set (C) and then by truncating the set such that only
individuals having cardenolide concentrations equal to, or less than,
that of the most concentrated queen (226 ug/0.lg) were included (D).
See text for explanation.




(all popu-

(Miami only)

Both 119 -0.004

M 61 -0.13

F 58 0.03

Both 43 -0.17

M 26 -0.25

F 17 -0.20



less than
226 ug/0.lg)

Both 48 -0.53*** -0.22



M 25 -0.55** -0.44* -0.39

F 23 -0.50*

Both 35 -0.40*

M 18 -0.36

F 17





-0.33* -0.43** -0.18

-0.46* -0.49* -0.36




* P < 0.05; ** P < 0.01;



















*** P < 0.0001.

between cardenolide concentration and other size and weight parameters

(Table 8).

Thin-layer chromatography (Fig. 5) demonstrates that the two

danaid species (as well as two arctiid moth species) sequestered

virtually identical sets of cardenolides from A. humistrata. This TLC

profile is clearly distinguishable from that of butterflies reared on

A. curassavica (cf. Fig. 4 inset).


Body Size and Cardenolide Content

Brower and Moffitt (1974) reported a negative correlation between

the body weight and cardenolide concentration of female monarchs from

Massachusetts, and suggested that these individuals may have suffered

a "metabolic cost," in terms of growth, of sequestering cardenolides

for defense (see also Brower et al., 1972). Such negative

correlations were not found for males from Massachusetts (which were 9

per cent lower than females in mean cardenolide concentration), or in

either sex collected in California (which were 62 per cent lower in

mean concentration than the Massachusetts females). However, these

data confound sexual and geographic differences with correlated

cardenolide concentration differences (see Chapter I). Here, I have

shown that negative correlations between cardenolide concentration and

various size and weight parameters, which occur in monarchs, but not

Q -d
C03 ^

L(n r- n

rn l Lr
0'. 1.
o i~. Lf
CMJ (M ^

00C -a, co O -r
n Lin en o -D --T o
-4 --4 ^
00 OO D 01, Lr 0

IX so 3- o oo o i
rD -I 00 00 00 in
cn cr C) 'IT CY Lf)


0 -H
) I

C )



0 0
0 0
0 &1



a L
3 u


c c


0 0
C 0
OJ .

-4 C4

0 0 %D 00

03 m

' 0-.
- 0T" *-4



r- c~ -~ 0



I- E-

0 w



0 i
U i s

0 C
0j3 0

-4 0






Table 8. Spearman correlation coefficients for cardenolide
concentration vs. body size, weight, and fat content of queen and
monarch butterflies (Dominican Republic stock; both sexes pooled)
reared in the laboratory on Asclepias humistrata.





7 0.33


0.25 -0.57

MONARCHS 10 -0.63** -0.56* 0.05 -0.58*

* 0.05 < P < 0.10;

** P < 0.05.

Fig. 5. Chromatograms of adult monarchs and queens reared in the
laboratory on Asclepias humistrata (developed as in Fig. 4).
Note that the profiles of the two species are virtually
identical, yet differ from those shown in Fig. 3. All
butterflies were collected as eggs or first instars in the
Dominican Republic, except for the monarch labelled "FL," which
was collected wild in the egg stage in Florida and shows an
identical "fingerprint" pattern to those of Dominican Republic
stock. For comparison, extracts of 6 dogbane tiger moths (Cycnia
tenera) and 1 saltmarsh moth (Estigmene acraea), reared
simultaneously on A. humistrata, were spotted on the same silica
plate. Also spotted are an ethanolic leaf extract of A.
humistrata and a 1:1 (v/v) mixture of digitoxin (dig) and
digitoxigenin (dgn). Numbers to the right of each channel
indicate the micrograms of cardenolide spotted, calculated prior
to lead acetate clean-up.


, )

0 .


e,1 *

-, ^ 4*

.igD I, ,

C. tenera
C. tenera

C. tenera

D. gilippus

D. plexippus
D. plexippus
A. humistrata
D. plexippus

D. gilippus

E. acraea

C. tenera

C. tenera

C. tenera
* Digitoxin/

,t910 f N '3VA1I

in queens, are independent of geographic, sexual, foodplant, or

correlated concentration differences.

While these negative correlations might well represent "metabolic

costs" of cardenolide ingestion, as suggested by Brower and Glazier

(1975), there is no direct evidence of a causal connection between

cardenolide differences and body size differences. Seiber et al.

(1980) found no effect of ingested digitoxin (added to a controlled

diet) upon the development time, food consumption, or body weight of

fourth instar monarch larvae (see also Chapter V). However, if it is

true that the negative correlations do reflect metabolic costs, then

we must inquire why such costs should be paid by monarchs and not by

queens. One possible explanation can be found in the different

population structures of these two species. Monarchs are migratory,

with most individuals in eastern North America flying to a few

restricted areas in Mexico to overwinter (Urquhart and Urquhart,

1977). Mating occurs during the remigration back to North America

each spring and most probably results in many matings between

individuals which developed as larvae on different species of

milkweed, some of which lack cardenolide (see Roeske et al., 1976).

Many of these matings would therefore involve butterflies that had not

been subjected to selection for allelochemic-tolerance. This would

tend to recombine any evolving gene complexes for cost-free adaptation

to allelochemics. In contrast, queens are fundamentally non-migratory

(Young, 1982), making only limited regional movements (Brower, 1961;

Burns, 1983). This greater degree of philopatry should more often

result in inbred matings among individuals that fed, as larvae, on the

same hostplant species. In areas where high-cardenolide plants

predominate, this should facilitate the evolution of a more rapid,

fine-tuned adaptation to hostplant allelochemics.

Sexual Differences in Cardenolide Concentration

Previous studies of monarch butterflies have found that females

have higher cardenolide concentrations than males (Brower and Glazier,

1975; Brower and Moffitt, 1974; Brower et al., 1972). This was again

demonstrated here for the laboratory-reared monarchs. Brower and

Glazier (1975) suggested that such a dimorphism might reflect

relatively stronger selection for chemical defense in females which

must spend considerable time exposed to potential enemies while

searching for oviposition plants. However, males also incur certain

risks in searching for females, and it is not clear how these risks to

males are to be compared with risks to females when predicting

differences in defensive strategies. In any event, such risks would

presumably apply to the congeneric queen butterfly as well. The lack

of sexual difference in cardenolide content of queens shown here

therefore suggests that the "differential risk" hypothesis is not

sufficient to explain the dimorphism previously observed in monarchs.

Another possible explanation is that female monarchs store more

cardenolide than males as a means of providing for the defense of

their eggs (Brower et al., 1982). Thomashaw (in Brower, in press)

reported that each egg of a female monarch reared on A. curassavica

contained, on average, 0.97 ug of cardenolide. Since a female may lay

from 100 (Erickson, 1973) to 400 (Urquhart, 1960) eggs, these could

collectively contain as much as 97 to 388 ug of cardenolide. Adult

monarchs reared on A. curassavica contain an average of 670 ug of

cardenolide (Roeske et al., 1976). Thus, the amount placed by females

into eggs constitutes a substantial proportion (14% to 58%) of this

total. One might therefore expect cardenolides to be effective

predator- or parasite-deterrents in monarch eggs but this has not yet

been tested. The lack of sexual difference in cardenolides of queens

may suggest that females of this species are not strongly selected to

store these compounds for egg-defense, and leads to the prediction

that they should allocate proportionally less cardenolide to eggs than

do monarchs. Further work should be directed at this issue.

Interestingly, while laboratory-reared monarch (but, again, not

queen) females did have significantly higher cardenolide

concentrations than males, this was not observed in field-collections

from the Miami area. This may suggest that some of the wild-caught

females in the Miami sample had already laid some of their

cardenolide-rich eggs, thereby reducing their (initially greater)

cardenolide loads to a level similar to that of males. (This

suggestion is consistent with Dixon et al.'s (1978) laboratory finding

that female monarchs that had laid all of their eggs were less emetic

when force-fed to pigeons than were freshly-eclosed females.) Since

females would likely oviposit at varying rates, this should lead to a

greater variation in cardenolide concentration of females, relative to

males. Indeed, females do tend to have higher variances than males

for this trait (see Table 5; F-max = 2.12, 0.05 < p < 0.10; Sokal and

Rohlf, 1981).

Cardenolide Variability and its Implications

The great intra- and interspecific uniformity in qualitative

cardenolide profiles of the Miami butterflies (see Fig. 4), suggests

that a single hostplant species had been utilized by larvae of both

species, but that there is individual variation in storage of certain

compounds, especially those of highest Rf value. That these TLC

profiles are virtually indistinguishable from those of monarchs reared

in the laboratory on A. curassavica (Fig. 4 inset) strongly suggests

that this was the hostplant species utilized by both danaid species in

the Miami sample (cf. Fig. 5 for A. humistrata-reared butterflies).

This represents the first practical application of the cardenolide

"fingerprinting technique" (Brower et al., 1982) to identify the

hostplants utilized, as larvae, by wild-caught danaid butterflies. It

also demonstrates the potential of the technique as an aid in

understanding the natural history and migration patterns of danaids.

Since the monarchs studied were collected in Miami in December and

developed on A. curassavica (an introduced milkweed with a North

American distribution restricted to southern states; Woodson, 1954),

we may conclude that they were not merely migrant butterflies from

northern states that had "become trapped" in peninsular Florida en

route to Mexico. Rather, this is strong evidence that monarchs breed

in south Florida during the winter months.

A large percentage of each queen sample consisted of butterflies

containing no measurable cardenolide (57% in Lake Istokpoga, 17% in

Corkscrew Swamp, and 21% in Miami). Since there were no significant

sex differences, such intrapopulational variability may instead

reflect localized differences in hostplant species availability,

intraspecific variation in plant cardenolide content (see, e.g.,

Nelson et al., 1981), or individual butterfly differences in

cardenolide sequestration. Whatever its origin, such variability

suggests the existence of a cardenolide-based palatability spectrum

for queens, similar to that previously described for monarchs (Brower,

1969; Brower et al., 1968; Brower and Moffitt, 1974).

This result has important implications for understanding the

southern viceroy's apparent switch from mimicking the monarch (as it

does elsewhere in its range), to mimicking the queen in Florida (see

Chapter I). If queens had been found to contain, on average, either

more cardenolide than monarchs, or a different and potentially more

potent (e.g., more emetic; Brower, 1969) set of cardenolides, then

such a mimetic switch might be easily understood. However, the Miami

queens clearly contained lower cardenolide concentrations and total

amounts than did the sympatric monarchs. Since the two species had

virtually identical cardenolide "fingerprints," it cannot be argued

that queens stored a more noxious array of cardenolides than did

monarchs and were therefore more emetic even at lower concentrations.

Thus, when fed on the same plants, monarchs and queens store the same

cardenolides but monarchs concentrate these to a greater extent than

do queens. This conclusion is further supported by laboratory

hearings of the two species on Asclepias humistrata (Table 7 and Fig.

5). Moreover, Brower et al. (1975) have shown that, in order for an

A. curassavica-reared monarch (sexes pooled) to be emetic to an 85 g

blue jay on 50% of test trials, it must contain at least 76 ug of

cardenolide. Of the Miami butterflies analyzed here, 85% of monarchs

met this criterion of unpalatability, while only 30% of queens did so.

It therefore seems that, at least with respect to cardenolides, queens

are poorer models for viceroy mimicry than are monarchs.

Why then should the viceroy have abandoned its usual model in

favor of the queen? Since monarchs are migratory, "pulsing" through

Florida in large numbers only in the spring and fall (Urquhart and

Urquhart, 1976; Brower, Malcolm and Cockrell, unpubl. data), while

queens are more sedentary, the latter species would be

spatiotemporally more "available" than monarchs to act as models in

Florida. A theoretical model developed by Pough et al. (1973) showed

that mildly noxious species could serve as suitable models for mimicry

if they occurred in sufficient abundance. This may provide the

explanation for the switch in viceroy mimicry. If resident, Florida

monarch populations have been stable and predictable in their current

locations for sufficient time, then reversals in the trend of mimicry

might be expected, such that viceroys in those areas should tend to be

more "monarch-like" than those elsewhere in the state. A detailed

geographic analysis of wing patterns is needed to test this


Alternatively, queens might, in fact, be superior models to

monarchs, not because of their cardenolide content but, rather, due to

sequestered pyrrolizidine alkaloids (PA's) that adults ingest from

certain withering plants (Edgar, 1975; Edgar et al., 1979). While

both sexes typically store these compounds as adults, male queens

employ them further as precursors of their sex pheromone (Meinwald et

al., 1969; Pliske and Eisner, 1969). Male and female monarchs are

also somewhat attracted to PA sources and may store the alkaloids but

males apparently do not use them as pheromone precursors (Edgar et

al., 1976). While there is as yet no experimental verification of PA-

based defense in danaid butterflies, K.S. Brown (unpublished

manuscript) reports that certain neotropical spiders will release

ithomiid butterflies from their webs unharmed if they contain PA's.

The dependence of male queens (and not monarchs) on PA's for sexual

competence suggests that, on average, queens may contain more of these

compounds than monarchs, and therefore possibly serve as better models

for viceroy mimicry. A comparative study of the PA concentrations of

wild-caught monarchs and queens would shed further light on this

intriguing problem.

Plant/Herbivore Coevolution

With respect to the issues of plant/herbivore coevolution

presented in the Introduction, this study leads to the following

provisional conclusions: (1) There is no evidence that the lower

cardenolide concentrations sequestered by queens relative to monarchs

reflects a poorer underlying tolerance for these allelochemics. On

the contrary, since only monarchs demonstrate significant negative

correlations between body size and cardenolide concentration, it might

be argued that monarchs are less adapted than queens for handling

cardenolides. However, a causative connection between cardenolide

sequestration and body size has not been established (see also Chapter

V); (2) Despite quantitative differences between monarchs and queens

in cardenolide storage, both species appear to sequester the same

individual cardenolide compounds from their hostplants. Indeed, when

reared on the milkweed A. humistrata, even dogbane tiger moths (Cycnia

tenera; Arctiidae), also apocynad/asclepiad-specialists, produced the

same characteristic TLC profile (Fig. 5) as did the polyphagous

arctiid moth, Estigmene acraea (although only in trace amounts; Fig.

5). Moreover, Marty (1983) has shown that gut homogenates of both

monarch and E. acraea larvae are capable of effecting a similar

enzymatic transformation of one milkweed cardenolide, uscharidin, to

two more polar metabolites (calactin and calotropin). Such similarity

among taxonomically disparate Lepidoptera, whether oligo- or

polyphagous, suggests that there may exist only a single qualitative

route of cardenolide processing in this insect order but that further

evolution may involve a quantitative increase in tissue cardenolide

concentration in accordance with the defensive requirements of each

species. However, the basic biochemical processes shared by all these

species may represent a common preadaptation to feeding on milkweeds

or other cardenolide-containing plants.

In summary, this chapter has demonstrated that although monarch

butterflies are the most thoroughly-studied danaids, they may not be

representative of their family in respect to adaptation to hostplant

allelochemics. A better understanding of danaid/milkweed

relationships will therefore depend upon obtaining a broader

comparative data base to document differential species responses to

ingested or sequestered cardenolide. Here it was shown that although

monarchs and queens sequester the same qualitative set of cardenolides

from milkweed plants, monarchs concentrate these in their tissues to a

greater extent than do queens. Furthermore, adult monarchs exhibit

significant negative correlations between the concentration of

sequestered cardenolide and both body size and weight, whereas queens

show no such correlations. Although this may suggest a metabolic cost

paid by monarchs and not by queens (and perhaps attributable to

different life-history characteristics; see above), there is, as yet,

no direct evidence of a causal relationship between cardenolide

differences and body size or weight differences. This question will

be explored more directly in Chapter V, following a consideration in

Chapter IV of cardenolide sequestration in a third milkweed-

specializing insect, the dogbane tiger moth (Cycnia tenera). This


species was previously reported not to sequester hostplant

cardenolides (Rothschild et al., 1970). This raises the interesting

question of whether, even among specialists, some species do not

sequester cardenolides because they are more sensitive to these

compounds than are sequestering species. If so, then species such as

C. tenera might represent an intermediate stage in the evolution of





Immature stages of monarch butterflies (Danaus plexippus L.;

Parsons, 1965; Reichstein et al., 1968; Brower, 1969) and milkweed

bugs (Hemiptera: Lygaeidae; e.g. Scudder and Duffey, 1972; Isman et

al., 1977a,b) are well-known for their ability to sequester plant

defense chemicals known as cardenolides (cardiac glycosides) from

their milkweed hostplants (Asclepiadaceae). Moreover, laboratory

studies have shown that these stored chemicals protect the insects

from some vertebrate predators (Brower and Brower, 1964; Brower, 1969;

Rothschild and Kellett, 1972). Several other insect species feed at

least occasionally on milkweed plants (Wilbur, 1976; Price and

Willson, 1979), and some of these have been assayed for the presence

or absence of cardenolides (e.g., Duffey and Scudder, 1972; Rothschild

and Reichstein, 1976). However, for relatively few species has it

*Published 1983 in Journal of Chemical Ecology 9:521-532. Reprinted

with permission.

been experimentally demonstrated that the cardenolides found in wild

and/or lab-reared insects are indeed derived from the hostplants

(allochthonous) rather than synthesized by the animals de novo

autochthonouss), as occurs, for example, in some chrysomelid beetles

(Pasteels and Daloze, 1977; Daloze and Pasteels, 1979). One may

distinguish these two possibilities by simultaneous hearings of

insects on cardenolide-rich and cardenolide-poor diets (e.g., Brower

et al., 1967; Rothschild et al., 1978). Differences in the

cardenolide contents of insects reared on such foods may then be

attributed to these dietary differences. This approach was taken in

the present study of the dogbane tiger moth, Cycnia tenera Huebner

(Arctiidae), a species previously reported not to sequester hostplant

cardenolides (Rothschild et al., 1970). Since a related species (C.

inopinatus), which is also a specialist feeder on asclepiadaceous and

apocynaceous plants (Tietz, 1972; Nishio, 1980), has since been found

to store these chemicals (Nishio, 1980), a reevaluation of C. tenera

seemed desirable. Among other arctiids, two species (Arctia caja and

Euchaetias antica) are known to sequester hostplant cardenolides (see

Rothschild and Reichstein, 1976). While three others (Euchaetias

egle, Digama aganais, D. sinuosa) were reportedly devoid of

cardenolides, it remains unclear whether this was due to storage

inability or perhaps to feeding on milkweeds low, or lacking in these

compounds. Sufficient intraspecific and interspecific diversity of

milkweed cardenolide contents exists (e.g., Roeske et al., 1976;

Nelson et al., 1981; Seiber et al., in press), so that it is possible

for an herbivore to be a milkweed specialist without being a

cardenolide specialist (see also Rothschild et al., 1970). However,

if C. tenera are found to sequester cardenolides then the hypothesis

that they pay metabolic costs of doing so may be evaluated by

searching for correlations between the concentration of sequestered

cardenolide and body weight (see Chapter III for rationale).

Cycnia tenera females lay clutches of 50 to 100 eggs (personal

observations), and larvae feed gregariously on their hostplants (we

have frequently seen groups of 5-7 larvae on a single A. humistrata

leaf). If the female has mated only once prior to oviposition, these

larvae will be full-siblings whereas, if she has mated more than once,

they will be at least half-siblings. This high degree of relatedness

permits the evolution of unpalatability and aposematism via kin

selection (Hamilton, 1964; Harvey et al., 1982; Brower, in press).

Thus, if cardenolides are, in fact, sequestered by the larvae and

retained into adulthood (both stages being aposematic; see Fig. 6),

then they may contribute importantly to the chemical defense and

evolution of this species.


In north-central Florida, Cycnia tenera larvae are commonly found

on Asclepias humistrata, a milkweed species having a relatively high

cardenolide concentration in its leaves (Nishio, 1980; cf. Roeske et

Fig. 6. An adult male Cvcnia tenera and a late-instar larva. The
taxonomy of this genus has been disputed. Forbes (1960) considered
C. tenera as distinct from C. inopinatus, while Kimball (1965) merged
both species under the latter name. More recent treatments (e.g.,
Hodges, in press) maintain separate status for the two species.
Voucher specimens of larvae and adults used in this study have been
deposited in the Florida State Collection of Arthropods. Doyle Conner
Building, Gainesville. FL. (Photos by L.P. Brower)

(Source: Cohen and Brower, 1983)


al., 1976; see also Chapters I and II). A clutch of 54 C. tenera eggs

was collected on 12 May 1981 from an A. humistrata plant growing wild

approximately 3 Km W of Gainesville (Alachua County), FL. The eggs

were brought to the laboratory and, upon hatching, each larva was

placed in an individual 250 cc closed (vented) plastic container and

reared to eclosion at 23+10C. Of these, 7 were reared through to

adulthood on a total diet of A. humistrata leaves, while 12 fed only

on A. tuberosa, a local species known to contain only very slight

amounts of cardenolide (Chapter I; also Roeske et al., 1976). All

adult moths were killed by freezing between 12 and 24 hours after

eclosion and remained frozen until chemical analysis in March, 1982.

Prior to analysis, the moths were dried at 60 C for 16h in a

forced-draft oven. Dry weights were determined to the nearest 0.1 mg

using a Mettler AK 160 electronic balance. Fat was then removed from

each insect by petroleum ether extraction of the dried material for 30

min in a 350C shaker bath (methods in Walford and Brower, in

preparation). Fat extraction removes only negligible amounts of

milkweed cardenolide from the insect material (C. Nelson and L.P.

Brower, unpublished data; see also Nishio, 1980).

Cardenolides were extracted from the fat-free material and

concentrations determined (as microgram equivalents of digitoxin)

using spectrophotometric procedures described in Brower et al. (1972,

1975). A lead-acetate clean-up procedure (Brower et al., 1982) was

then used to remove interfering pigments and other noncardenolide

compounds from the extract remaining after spectroassay. Thin-layer

chromatography (TLC), employing an ethyl acetate methanol solvent

system (97:3 by volume; see Brower et al., 1982), was used to

visualize the cardenolides present in each hostplant species and in

two randomly selected moths that had developed on each species.


The mean dry and lean weights, cardenolide concentrations and

total cardenolide contents of C. tenera adults reared on A. humistrata

and A. tuberosa are shown in Table 9. Two-way analyses of variance

demonstrate that females reach heavier dry weights than do males,

regardless of foodplant eaten (F=32.5, df=1,15, p<0.001), but that

both sexes are heavier when reared on A. humistrata than on A.

tuberosa (F=40.9, df=1,15, p<0.001). The interaction of sex and

foodplant was not significant (p>0.50). The results for lean weights

are similar, with females heavier than males (F=16.2, df=1,15,

p<0.005) and both sexes tending to be heavier on A. humistrata than on

A. tuberosa (F=5.6, df=1,15, 0.1>p>.05). Females contain slightly

more fat than do males (F=4.6, df=l,15, O.l>p>.05) and both sexes

store more fat when fed A. humistrata than when fed A. tuberosa

(F=21.2, df=1,15, p<0.001). Overall, the fat content of the adult

moths constitutes from 45 to 60 per cent of the total dry weight.

0 CO' 1
o i-4

H 0 E-'

a *

30 C0 C 0
c 1 in E-4 w M X y

w -00
Cb -W C C C

4C. -, E 1 0c

E* *
to cn Z

0 E-4 U 0

Lo ~~~E--q~ ~
r w rrZ \oN

U0 04 IM t

tvi il E-4
>* 00
v -W E >4 C4 C; .4


4- W W CC \ M I
4-> co) co cc co
r 4iX 4 o

E-4 0 4-1 n

o- 4J r: F O

Clearly, both sexes sequestered higher concentrations and total

amounts of cardenolide from A. humistrata than from A. tuberosa

(F=76.4 for concentration; F=50.0 for total amount; df=1,15 and

p<0.0001 for both tests). However, the sexes did not differ

significantly either in concentration (F=1.2, df=1,15, p>0.2) or in

total cardenolide content (F=2.4, df=1,15, p>0.2). No significant

interaction effects were identified (p>0.2). When both sexes are

considered together, moths reared on A. humistrata had, on average, 10

times higher concentrations, and contained 15 times more total

cardenolide, than did moths reared on A. tuberosa.

Spearman rank order correlations demonstrate no significant

relationship between the concentration of sequestered cardenolide and

either dry weight (r=-0.28; P>0.50) or lean weight (r=-0.43; P>0.30).

This analysis was limited to those moths reared on A. humistrata (n=7;

sexes pooled) in order not to confound cardenolide-differences with

other plant species differences (e.g., in nutritional quality; see


Thin-layer chromatography (Fig. 7) clearly confirms that

sequestration of hostplant cardenolides occurred, since individuals

reared on A. humistrata contain virtually every cardenolide visualized

in their hostplants. However, the chromatogram of A. tuberosa, a very

low-cardenolide species, consists only of a faint band ranging from

approximately Rf = 0.1-0.5. Much of this band is pink (i.e.,

noncardenolide interference) rather than blue (cardenolide) and it is

possible that these interfering compounds may have contributed to the

Fig. 7. TLC profiles of four adult Cycnia tenera reared on
either Asclepias humistrata (29 ) or A. tuberosa ( dc, 1 ).
For reference, a mixture of commercially procured digitoxin (dig;
Rf = 0.31) and digitoxigenin (dgn; Rf = 0.60) was spotted in
three channels. The amounts of cardenolide (ug equivalent to
digitoxin) spotted in each channel are shown at the bottom of the
plate. (These are calculated prior to lead-acetate clean-up, a
procedure which sometimes reduces the amount actually spotted.
Thus, the female in channel 3 has a lighter trace than that in
channel 2, presumably having lost disproportionately more
cardenolide during clean-up.) For A. humistrata there is a very
close correspondence between the cardenolides present in the leaf
tissue and those in the moths reared thereupon. However, A.
tuberosa has a very low cardenolide concentration and only faint
spots are produced from the leaf aad moth extracts. In the
original color prints, all cardenolides appear as blue spots.
However, much of the A. tuberosa leaf and moth chromatographs
appears pink in the color original, except for faint blue
cardenolide regions indicated by broken lines. No such
interference occurs in the chromatographs of A. humistrata, the
moths reared thereupon, and the commercial standards. (Source:
Cohen and Brower, 1983)

& umitrata A. tubro

Moth Plant Plant Moth

don don dgn

E4 *dig die
8" i1

,i o I s t
g o crdenol spott
,ug of cardenolide spotted

absorbances recorded for the moths reared on this species, thereby

overestimating cardenolide concentrations. If this is so, then the

true difference in cardenolide contents between moths reared on the

two hostplants may be even greater than reported above.


These data demonstrate the sequestration of hostplant

cardenolides by Cycnia tenera larvae and the subsequent retention of

these chemicals into adulthood. Individuals reared on Asclepias

humistrata, a cardenolide-rich milkweed species, attained much higher

concentrations and total amounts of cardenolide than did those reared

on a cardenolide-poor diet of A. tuberosa. Moreover, TLC analysis

demonstrates a close correspondence between the individual

cardenolides of A. humistrata and those of the moths reared thereupon.

If all of the cardenolide present in C. tenera were of autochthonous

origin, moths reared on both milkweed species would be expected to

contain similar amounts and kinds of cardenolide. Clearly, they do

not. However, the possibility that the small amount of cardenolide

present in A. tuberosa-reared moths may, at least in part, be of

authochthonous origin cannot be excluded.

Both sexes reached higher fat and lean weights when reared on the

higher cardenolide species, A. humistrata. Similar results for Danaus

plexippus (Erickson, 1973), D. chrysippus (Smith, 1978), and

Oncopeltus fasciatus (Isman, 1977; Chaplin and Chaplin, 1981) have

been interpreted as evidence against a metabolic cost of handling

hostplant cardenolides (see also Blum, 1981). However, this inference

is problematical because milkweed species differ in many ways other

than in cardenolide content (e.g., nitrogen and water content,

Erickson, 1973; leaf shape, growth form, texture and pubescence,

Woodson, 1954). Such differences contribute to the overall

suitability and quality of these plants as insect food (Scriber and

Slansky, 1981). Asclepias tuberosa is a relatively poor food source

due to low nitrogen and water content (Erickson, 1973) and is

infrequently used by herbivores in north-central Florida (Cohen and

Brower, unpublished observations), or elsewhere (e.g., Wilbur, 1976;

Price and Willson, 1979). It is to be expected that such poor

foodplant species will be under weaker selection for antiherbivore

adaptations than will species of higher nutritive value, and might

therefore have relatively lower cardenolide contents. Hence, it is

possible that herbivores develop more successfully on relatively-high

cardenolide species, such as A. humistrata, not because there is no

cost of handling these chemicals, but because the higher food quality

is great enough to offset any such costs that might exist.

This complication explains why correlations between insect

cardenolide concentration and body weight were sought only for those

individuals reared on A. humistrata. Clearly, had the A. tuberosa-

reared moths been included in the analysis, a positive correlation

would result. However, this would most likely be due to the

nutritional differences between the two plant species, rather than

their allelochemic differences. That the A. humistrata-reared moths

exhibited no significant correlations between cardenolide

concentration and body weight provides further (indirect) evidence

that milkweed-specializing lepidopterans pay little, if any, cost of

feeding on cardenolides.

It is interesting that the extracted fat from C. tenera adults

may constitute from 45 to 60% of the total dry weight (Table 9). This

is sharply higher than the 5 to 20% common for freshly closed danaid

butterflies (Beall, 1948) and is similar to the levels found in

migrating monarch butterflies just prior to overwintering in Mexico

(Walford and Brower, in prep.). Moths of a related arctiid genus,

Euchaetias, do not feed as adults (Forbes, 1960; Schroeder, 1977) and

lifetime energy stores must be accumulated by the larvae prior to

pupation. However, unlike Euchaetias, adult Cycnia have well-

developed probosces (Forbes, 1960), and presumably feed. Thus their

apparently high fat storage remains enigmatic. While this result is

not directly relevant to the issue of cardenolide-defense, it does

raise an intriguing question and it should prove interesting to

compare larval fat storage between those arctiid species which have a

feeding adult stage and those which do not.

The results on cardenolide sequestration are in contrast to those

of Rothschild et al. (1970), in which no cardenolides were found in C.

tenera reared on Asclepias syriaca. That plant species, however, is

extremely variable in cardenolide content (see Roeske et al., 1976),

and certain strains are reportedly devoid of cardenolides (Rothschild

et al., 1975). It is thus possible that the particular plants fed to

the larvae in the experiments of these authors were of insufficient

cardenolide content for larval sequestration to occur. Alternatively,

intraspecific geographic differences in sequestration abilities in the

moths may be indicated. Such differences have been hypothesized for

the various geographic races of the African queen butterfly, Danaus

chrysippus (Rothschild, et al., 1975) but have not yet been

definitively established for any species. A third possibility is that

C. tenera is not adapted to sequester the particular array of

cardenolides found in A. syriaca. Indeed, of 6 milkweed species

studied by Price and Willson (1979) in central Illinois, A. syriaca

was one of only two never utilized by C. tenera as a larval foodplant.

However, there is no evidence that this apparent rejection was

determined in any way by the cardenolides of A. syriaca.

Cycnia tenera adults are conspicuously colored, with off-white

wings and black-spotted, yellow abdomens (Fig. 6). The bitter-tasting

cardenolides they contain probably impart to them a noxious quality,

as is true in the monarch butterfly (Brower, 1969; Brower and Moffitt,

1974). Moreover, they produce both audible and ultrasonic sounds

(Fullard, 1977) which may serve to warn bats and other potential

predators of the unpalatability of the moths, thus serving an

aposematic function (e.g., Dunning and Roeder, 1965).

Larvae, too, are aposematic, with bright orange bodies and

contrastingly-dark tufts of setae along the dorsum (Fig. 6).

Interestingly, the setae of a closely related species, C. inopinatus,

are virtually devoid of cardenolides, while the underlying larval

cuticle is rich in these compounds (Nishio, 1980). This suggests the

potential operation of three separate lines of defense in larval

Cycnia. The setae may have a sensory function, permitting the larvae

to recognize the approach of a predator or parasite. Larvae respond

to tactile stimulation by dropping to the ground and curling the body

(unpublished observations). This behavior has the effect of exposing

the dorsal setae maximally. Should the larva nevertheless be found by

a vertebrate predator, a second, mechanical line of defense may come

into play: naive predators may attempt to eat a larva, but release it

unharmed when the mouth is irritated by the hairs. The setae may then

become an aposematic signal, preventing further attacks by experienced

predators. The evolution of such sensory and mechanical defense is

not problematic since larvae having such setae would presumably be

more likely than those lacking them to survive an attack. If,

however, these defenses should fail, the cardenolides present in the

larval tissues would provide an unpalatable, and possibly emetic,

experience (see, e.g., Brower, 1969), leading to later avoidance of

further larvae encountered. Indeed, it is possible that the setae,

once ingested, may irritate the gastro-intestinal lining (Bisset et

al., 1960; Frazer, 1965) and thereby facilitate absorption of

cardenolides, i.e., an interaction of mechanical and chemical

defenses! The larva would die in the process, however, and thus its

genes for cardenolide sequestration would not be propagated in the

population. Kin-selection (see Hamilton, 1964) would be required for

the evolution of such chemical defense and seems plausible in this

species, since eggs are laid in clutches and the larvae feed

gregariously on their hostplants. Thus, kin-groups of at least half-

siblings, and possibly full-siblings, feed together on the same plant.

If a predator were to sample and kill one or a few of these aposematic

larvae before learning to avoid them, the remaining siblings would be

spared and shared genes both for unpalatability and aposematism would

continue to spread within the population.

My work on cardenolides in C. tenera suggests that these compounds

may provide at least a partial basis for an underlying unpalatability

of both larvae and adults. However, this does not preclude the

possibility of other chemical defenses. Parsons and Rothschild (in

Rothschild et al., 1970) have noted the presence of histamine-like

and/or acetylcholine-like substances in C. tenera, although the manner

in which these compounds function in nature remains largely

unexplored. Moreover, several species of arctiid moths (including

Cycnia mendica; Rothschild, 1973) are known to sequester pyrrolizidine

alkaloids (PA's) from larval hostplants (although these have not yet

been identified from Asclepiadaceae) or, for those species with

feeding adult stages, from decomposing leaves (Rothschild et al.,

1979). Finally, it is possible that other noxious chemicals in these

moths may be of autochthonous origin, rather than derived from plant

sources (Rothschild et al., 1970, 1979). The relative contributions

and possible interactions of cardenolides, biogenic amines, PA's,