|
(ISSN 0015-4040)
FLORIDA ENTOMOLOGIST
(An International Journal for the Americas)
Volume 67, No. 1 March, 1984
TABLE OF CONTENTS
The 67th Annual Meeting of the Florida Entomological Society; Second
Announcement and Call for Papers
SYMPOSIUM: INSECT BEHAVIORAL ECOLOGY-'83
LLOYD, J. E.-Gourmet Insect Behavioral Ecology: Stalking the Wild
Speculation --..--....--........-----........ --...----...............---..........------ 1
CHARNOV, E. L., AND S. W. SKINNER-Evolution of Host Selection and
Clutch-Size in Parasitoid Wasps ___---.....-....-..---------..----- 5
ANGELO, M. J., AND F. SLANSKY, JR.-Body Building Insects: Trade-
Offs in Resource Allocation with Particular Reference to Migra-
tory Species .--..----..----_ --.------------.. -.--.-..---.--------_ 22
BARFIELD, C. S., AND R. J. O'NEIL--Is an Ecological Understanding a
Prerequisite for Pest Management? --------...---....-----..-..---.---- 42
ALTMANN, J.-Observational Sampling Methods for Insect Behavioral
Ecology ..--.......-- .......-----. -------....- -......-.. ------- ...__ .. 50
SIVINSKI, J.-The Behavioral Ecology of Vermin -...--....-...............--- .. 57
McNAB, B. K.-Energetics: The Behavioral and Ecological Conse-
quences of Body Size ....--------......................--------........... 68
THORNHILL, R.-Scientific Methodology in Entomology ......................... 74
CROCKER, R. L., AND R. B. SKINNER-Boolean Model of the Courtship
and Agonistic Behavior of Hentzia palmarum (Araneae: Salti-
cidae) ___ ---..---------------........................... ............-_ 97
MUCHMORE, W. B.-Pseudoscorpions from Florida and the Caribbean
Area. 12. Antillochernes, A New Genus with Setae on the Pleural
Membranes (Chernetidae) --------..........---........--...........----. 106
MUCHMORE, W. B.-Pseudoscorpions from Florida and the Caribbean
Area. 13. New Species of Tyrannachthonius and Paralioch-
thonius from the Bahamas with Discussion of the Genera
(Chthoniidae) ----.....--.......-....---- ........ ..-. ..........--....--... .. ----119
SIVINSKI, J.-Effect of Sexual Experience on Male Mating Success in
a Lek Forming Tephritid Anastrepha suspense (Loew) --....... 126
JACKSON, D. M., F. C. TINGLE, AND E. R. MITCHELL-Survey of
Heliothis spp. Larvae Found on Florida Beggarweed and Post-
harvest Tobacco in Florida -------------...---.................. ... 130
Published by The Florida Entomological Society
FLORIDA ENTOMOLOGICAL SOCIETY
OFFICERS FOR 1983-84
President --...--- .........--- .......-..----..--...--.--------... ............ C. W. McCoy
President-Elect ...........--..--...---- .....-............ ------------------M. L. Wright, Jr.
Vice-President .............-......_.....~.--........ --.----- ......-..... J. A. Reinert
Secretary .-- ----............--............... ............ ..... ...--.. . ................ .. D. F. W illiams
Treasurer ........ -----.................-..........-... ........... ...... A. C. Knapp
J. R. Cassani
J. L. Knapp
D. C. Herzog
Other Members of the Executive Committee ... K. Lee
C. A. Morris
W. L. Peters
C. A. Musgrave Sutherland
PUBLICATIONS COMMITTEE
Editor .....--...-...----------......... ---.-.......-......- C. A. Musgrave Sutherland
Associate Editors .........................-......-......-......-..-............-------- ...... D. C. Herzog
F. W. Howard
M. D. Hubbard
J. R. McLaughlin
J. B. Heppner
H. V. Weems, Jr.
Business Manager .........---- ........-- ...............-----................. A. C. Knapp
FLORIDA ENTOMOLOGIST is issued quarterly-March, June, September,
and December. Subscription price to non-members is $20.00 per year in
advance, $5.00 per copy. Membership in the Florida Entomological Society,
including subscription to Florida Entomologist, is $15 per year for regular
membership and $5 per year for students. Inquires regarding membership
and subscriptions should be addressed to the Business Manager, P. O. Box
7326, Winter Haven, FL 33883-7326. Florida Entomologist is entered as
second class matter at the Post Office in DeLeon Springs and Gainesville, FL.
Authors should consult "Instructions to Authors" on the inside cover of
all recent issues while preparing manuscripts or notes. When submitting a
paper or note to the Editor, please send the original manuscript, original
figures and tables, and 3 copies of the entire paper. Include an abstract and
title in Spanish, if possible. Upon receipt, manuscripts and notes are ac-
knowledged by the Editor and assigned to an appropriate Associate Editor
who will make every effort to recruit peer reviewers not employed by the
same agency or institution as the authors (s). Reviews from individuals
working out-of-state or in nearby countries (e.g. Canada, Mexico, and others)
will be obtained where possible.
Manuscripts and other editorial matter should be sent to the Editor,
C. A. Musgrave Sutherland, 4849 Del Rey Blvd., Las Cruces, NM 88001.
June 1, 1984
THE 67th ANNUAL MEETING OF THE
FLORIDA ENTOMOLOGICAL SOCIETY
FIRST ANNOUNCEMENT
The Florida Entomological Society will hold its 67th Annual meeting on
24-27 July 1984 at the Holiday Inn, 6515 International Drive, Orlando FL
32809; telephone-1-(305)-351-3500. Room rates will be $58.00, for single,
double, triple, or quadruple.
Questions concerning the local arrangements should be directed to:
FREDERICK L. PETITT, Chairman
Local Arrangements Committee
Florida Entomological Society
Walt Disney World-Epcot Center-The Land
P.O. Box 40
Lake Buena Vista, Florida 32830 USA
Phone: 1-(305)-827-7256
To present a paper, the tear out sheet must be postmarked and sent no
later than 15 MAY 1984, to:
JAMES A. REINERT, Program Chairman
Ft. Lauderdale Research and Education Center
University of Florida
3205 S.W. College Avenue
Ft. Lauderdale, Florida 33314 USA
Eight minutes will be allotted for presentation of oral papers, with 2
minutes for discussion. In addition, there will be a separate session for
members who may elect to present a Project (or Poster) Exhibit.
The 3 oral student papers judged to be the best on content and delivery
will be awarded monetary prizes during the meeting. Student authors must
be Florida Entomological Society Members and must be registered for the
meeting. Awards will be $125.00, 75.00 and 50.00.
The 3 student display presentations judged to be the best on content and
preparation will also be awarded monetary prizes during the meeting.
Student authors must be Florida Entomological Society Members and must
be registered for the meeting. Awards will be $125.00, 75.00 and 50.00.
Registration Schedulel for Annual Meeting:
Preregistration Registration On Site
Full & Sustaining Members $35.00 $40.00
Student not in Student Contest 18.00 20.00
Student in Student Contest 13.00 15.00
Each Extra Banquet Ticket 10.00 10.00
'Each fee includes one banquet ticket.
PAPER SUBMISSION
Deadline:
15 May 1984
SLIDE POLICY FOR ANNUAL MEETINGS
The following slide policy will govern slide presentations at the Annual
Meetings. Only Kodak Carousel projectors for 2 x 2 slides will be available.
However motion picture projectors will be available by special request to
the Local Arrangements Chairman prior to the date of the meeting.
Authors should keep slides simple, concise, and uncluttered with no more
than 7 lines of type on a rectange 2 units high by 3 units wide. All printed
information should be readable to an audience of 300 persons.
A previewing room will be designated for author's use. A projectionist
will be available in the previewing room at least one hour before each session.
Authors are expected to give the projectionist their slides in the previewing
room prior to each session. Slides will be returned to the authors after each
session in the meeting room.
Authors are expected to organize their slides in proper order in their
personal standard Kodak Carousel slide tray (no substitution, please). Only
a few slide trays will be available in the previewing room from the projec-
tionist for hardship cases. Slides in the tray should be in correct order start-
ing with slot #1 of the tray and positioned correctly (position of slides to go
into tray: 1. upside down, and 2. lettering readable from this position upside
down and from right to left). A piece of masking tape should be placed on
the slide tray by the author and the following information should be written
on the tape: 1. author's name, 2. session date, and 3. presentation time.
^ g^
4
a) a
30 a
^ '0 ,c
o Q o
a)i-
0
.)Fl
m etU~ a, e
~c3 CTS-
00
.>- .
O0
S0
or-l
ed
M
OF:
Qlm
^*rt
DE] E
r=j
0
0
CO
bo
t4-
aR
I
Insect Behavioral Ecology-'83 Lloyd
GOURMET INSECT BEHAVIORAL ECOLOGY:
STALKING THE WILD SPECULATION
JAMES E. LLOYD*
"Bold ideas, unjustifiable anticipations, and speculative thought, are
our only means for interpreting nature: our only organon, our only
instrument, for grasping her."
(Karl Popper, quoted in Beveridge 1980)
Insect Behavioral Ecology is a new and thriving discipline within
Entomology. What is it that is new, and different from S. W. Frost's ad-
mirable insect natural history or bionomics? (Lloyd 1980). Whereas the
discipline known as exobiology-the study of life on other planets and in
outer space-has been called a science without a subject, insect bionomics
might have been called a subject without a science-science is far more
than the collection of facts and the arrangement of them into convenient
but more or less arbitrary patterns. An historian of Newton said that "The
naturalist is indeed a trained observer, but his observations differ from those
of a gamekeeper only in degree, not in kind; his sole esoteric qualification
is familiarity with systematic nomenclature." Actually this was a case of
science snobbery, the subject Mayr (1982) was discussing when he passed
this quotation along, but it helps me get to my point. It is Darwin's theory
(and improvements on it) that is primarily responsible for raising our study
of insects above what has gone before, and that gives us an "esoteric qualifica-
tion." Now if used creatively, more than merely giving insight to explain the
adventures of favorite animals and old problems, it can lead to the discovery
of new and useful biological questions and generalizations at several levels
in the explanation hierarchy. You would be asked, and then badgered, were
you fortunate enough to be in one of Dick Alexander's classes, "What are the
10, exactly 10, no more no fewer, what are the 10 most important concepts
in [some subject area]." Well then, it is 18 past Williams (1966), do you
know where your conceptual progeny are? Do you know how to make an
innovation or tickle an inspiration out of the right side of your brain?
"The creative mind is able, as Schopenhauer has stated it, 'to think
something that nobody has thought yet, while looking at something
that everybody sees." Imagination, thus, is ultimately the most im-
portant prerequisite of scientific progress."
(Ernst Mayr 1982)
The process of going from original observation, to disciplined observa-
tion (see Altmann, this Symposium), to testing and "proof" requires
imagination. Imagination is the critical and certainly the most intellectual
element of the hypothetico-deductive method (see Ghiselin 1969, Mayr 1982,
and Thornhill, this Symposium.) Some people have it. Perhaps some
don't. Some seem to have more of it than others, but can be helped with
training and practice (Adams 1979, Beveridge 1980, Daitzman 1980). The
first step is to ignore those who remain of the opinion that speculation
*Professor in the Dept. of Entomology and Nematology, Univ. of Florida, Gainesville
32611.
Florida Entomologist 67(1)
("trivial and stupefying natural philosophy," H. Kolbe, see Van't Hoff
1967) should be bridled or eschewed entirely, for they will only be helpers
at the nest-show me a biologist who can't appreciate a good just-possibly-
so story and I'll show you a technician (see Mayr 1983). A colleague re-
cently sent me a paper in which, after presenting pages of data, he sub-
missively and apologetically tendered the notion that there was "scope for
profitable speculation in some instances." However, he noted that functional
interpretations were a minefield for the incautious! Who is going to get
blown up by whose mines for trying to narrow the search for meaning
in the obese hog of Baconian trivia that fattens at granting agency
troughs?, but, alas, I see I have answered my own question. Even the most
unimaginative know or feel that an idea that explains their technical out-
put makes them eunuchs (are we all merely Darwin's retainers?), so per-
haps they favor a system that cuts off the potential seminal contributions of
others. I think we have all sensed more than once that the castration of
imagination is part of the rite of passage in "peer" review. But, when an
idea is published everybody can think about it, develop it, juxtapose it
with others, store it for future use or reference, use it to get to a better
alternative, or constructively shoot it down. When an idea is censored all
are poorer, except the reviewers and they are anonymous-and regardless
of what you would prefer to believe, ideas are pirated from reviewed manu-
scripts and research proposals (more on this later; see Broad and Wade
1983). "Wild speculation"? From my experience this means that the
accuser is too unimaginative, short-sighted, or unknowledgeable to under-
stand it, where it might lead, or how to profit from it (e.g. see Van't Hoff
1967). I will eventually illustrate this in detail in the case history of a re-
viewer who referred to a suggested (shared) working hypothesis that grew
out of and explained an aggregate of more than 60 years of observation
as "wild speculation," though he himself had just proposed an explanation
that was not merely pre-Darwinian, but, in fact, pre-Lamarckian and in-
voked a species spirituality. Now that's his kind of wild. Mine is the spice
that completes the process of turning the study of insects into an epicurean
delight and a scientific accomplishment.
Beveridge (1980) discussed several ways to put your full brain to work,
in the search for innovation, new concepts, generalizations, and theories, and
I cannot possibly do him justice here. (I enthusiastically recommend
Beveridge's book-use it for a seminar text, and a source of numerous
references on the subject.) The obvious first way is critical thinking. This
is disciplined thought that is directed by one's consciousness. It follows
logical pathways and stops lines of thought that are inconsistent with
known facts and accepted theories. This is the supposed basic modus
operandi of science, and all of us are trained in it and inhibited by it to
some degree. Another is imaginative thinking. This is not usually con-
sciously directed and is often subjective, and is what psychologists call
associative flow. It is a way of avoiding the use of words, which are the
material of critical thinking but a hindrance to imaginative thinking.
Visual not verbal symbols are used. For example, you might imagine your-
self to be a parasite looking for a host, or a sperm in a race in the female
reproductive trick. Who are you to argue or chdrtle-Einstein did it!
And then there is wild thinking. Wild thinking goes the next step,
and breaks out of the restrictions imposed by the limits of ingrained in-
March, 1984
Insect Behavioral Ecology-'83 Lloyd
hibitions that will still be present in the unaided imagination of the pre-
vious method. In wild thinking, aids to the imagination are used. For
example, you might go to the hardware store and look at every item on the
shelves and force yourself to figure out how each could be used to solve the
problem, or could be analogous to part of the biological mechanism you
can't understand. (I suppose if you are researching mating biology you
should visit the "Sex Toys" store, reported to be on Bourbon Street, New
Orleans). To get in condition for this sort of thing, try what is called
mental jogging (Daitzman 1980). For example, as your assignment, list
seven changes that would or could be made at "Wendy's (Where's the
beef?)" fast food restaurants if humans had hypognathous heads, piercing
mouth-parts, and parthenogenesis. But even this will not be enough, for re-
member what J. B. S. Haldane said about Nature-its not only queerer than
we suppose, it's queerer than we can suppose.
"Everything about which I thought or read was made to bear directly
on what I had seen or was likely to see; and this habit of mind was
continued during the five years of the voyage. I feel sure that it was
this training which has enabled me to do whatever I have done in
science."
(Charles Darwin, ca. 1876, in Barlow 1969)
A less avant-garde form of inspirational activity makes use of the fact
that a fertile union is often made at the interface of two sciences, subjects,
or different disciplines within a science. A classic example is that Gregor
Mendel not only knew his peas but his mathematics as well. In these Sym-
posia we are trying to have some papers that are right on insect behavioral
ecology (wherever that is, Lloyd 1980), and others that run to the side, at
another level, or across the grain, and that might make a neural bridge in
the gray matter of someone who hears or reads the proceedings. Toward
that connection, let us now begin with the papers at hand.
But first, I thank the Executive and Program Committees of the Society
for their enthusiastic support and cooperation and I also thank the follow-
ing individuals for reviewing and making helpful suggestions on manu-
scripts and for technical assistance: John Alcock, Tim Forrest, Barbara
Hollien, Dick Johnston, Ngo Dong, John Sivinski, John Strayer, Tom Walker,
Susan Wineriter, Dan Wojcik, and Lewis Wright. Symposium-'83 was made
possible by contributions from Carl Barfield, IFAS through the efforts of
Dan Shankland, and the Society.
Symposium-'84 will be held at the annual FES meeting at Orlando in
July 1984. Florida Agricultural Experiments Station Journal Series No. 5473.
REFERENCES CITED
ADAMS, J. L. 1979. Conceptual blockbusting. W. W. Norton and Co., New
York.
BARLOW, N. 1969. The autobiography of Charles Darwin (1808-1882).
W. W. Norton and Co., New York.
BEVERIDGE, W. I. B. 1980. Seeds of discovery. W. W. Norton and Co., New
York.
BROAD, W., AND N. WADE. 1982. Betrayers of the truth. Simon and Schuster,
Inc., New York.
DAITZMAN, R. J. 1980. Mental jogging. Richard Marek Publ., New York.
Florida Entomologist 67(1)
GHISELIN, M. T. 1969. The triumph of the Darwinian method. Univ. of
California Press, Berkeley.
LLOYD, J. E. 1980. Insect behavioral ecology: Coming of age in bionomics
or compleat biologists have revolutions too. Florida Ent. 63: 1-4.
MAYR, E. 1982. The growth of biological thought. Belknap Press of Harvard
Univ. Press, Cambridge, MA.
MAYR, E. 1983. How to carry out the adaptationists program? American
Nat. 121: 324-334.
VAN'T HOFF, J. H. 1967. Imagination in science. (G. F. Springer translation
of 1877 inaugural address). Springer-Verlag, Berlin.
WILLIAMs, G. C. 1966. Adaptation and natural selection. A critique of some
current evolutionary thought. Princeton Univ. Press, Princeton, NJ.
Symposium Participants. Back row, 1 to r: Awinash Bhatkar, Carl
Barfield, John Sivinski, Franz Huber, Randy Thornhill. Front row: Howard
Seliger, Mary Jane Angelo, Brian McNab, Eric Charnov, Jim Lloyd. Photo-
graphed by Frank Mead, FADCS-DPI. 11 August 1983, Clearwater Beach,
Florida.
March, 1984
Insect Behavioral Ecology-'83 Charnov and Skinner
EVOLUTION OF HOST SELECTION AND CLUTCH SIZE
IN PARASITOID WASPS
ERIC L. CHARNOV AND SAMUEL W. SKINNER*
SYNOPSIS
This paper discusses the evolution of host selection, including super-
parasitism, and the evolution of clutch size in non-solitary parasitoids. First,
we review the natural selection (as opposed to proximate mechanism) ap-
proach to understanding life-histories. We discuss this approach specifically
in reference to parasitoids, reviewing previous work. Then we build a simple
or first order model for a primary parasitoid's clutch size; this is tested
against lab/field data for two parasitoids (Trichogramma, Nasonia). The
model fails to account for the data, but the failure is in a very particular
way. This leads us to revise the theory, adding factors left out of the first
order approach. This new theory building is not ad hoc as the factors added
are very natural and general considerations. We then consider the problem
of superparasitism, the two key questions being, when should it be expected
to happen, and what clutch size ought the superparasite have? Finally, we
make some general remarks on the natural selection approach to evolutionary
ecology.
INTRODUCTION
All organisms pass through a series of physiological and behavioral
stages over the courses of their lifetimes. Sometimes these stages of life
are reflected in profound changes in the form: for instance, a butterfly
begins life as an egg, hatches into a larva, and goes through a pupal stage
before emerging as an adult. Other creatures undergo less drastic physical
changes.
The study of life histories takes as its subject these general patterns in
the lives of individual organisms. It is a powerful tool with which to under-
stand the compromises that organisms have evolved to deal with the many
physical and biological factors that affect them during their lives.
Most of the important concepts in the study of life histories can be
framed in terms of trade-offs. For example, Pacific salmon of the genus
Oncorhynchus spend two to three years of rapid growth at sea and then
travel up a river or estuary to spawn, after which they die. Trout, of the
genus Salmo, are closely related to salmon and also come in from the sea to
spawn in many of the same streams. But trout do not die after a single
spawning; they return to the sea and breed again the next year. Assuming
that these two patterns were produced by natural selection operating upon
the reproductive advantages to an individual organism, why should any fish
die after one spawning? Why not return, as trout do, to spawn many times?
*Eric L. Charnov first began working on evolutionary ecology when he was a graduate
student with Gordon Orians at the University of Washington, in the early 1970's. His
personal discovery of natural selection constitutes the only religious experience of his life.
Currently at the University of Utah, he is a Professor in the departments of Biology, Psy-
chology and Anthropology. His current research concerns human social development, human
hunter-gatherer ecology; and sex allocation in plants and animals especially barnacles and
coral reef fish.
Samuel Skinner completed a Ph.D. in Biology at the University of Utah in 1983. Presently
he is an NIH Postdoctoral Fellow with James F. Crow at the University of Wisconsin. His
interests include sex ratio and clutch size evolution. Current addresses: Charnov, Dept. of
Biology, Univ. of Utah, Salt Lake City 84112. Skinner, Laboratory of Genetics, Univ. of
Wisconsin, Madison 53706.
Florida, Entomologist 67(1)
In this case, there is a trade-off which relates to the fact that a fish
must reduce the number of eggs it produces during a single spawning if
it is to have enough energy to return to the sea and survive until the next
spawning season. Trout produce fewer eggs each year than they would if
they invested all their energy in reproduction and died soon after spawning.
With salmon, this trade-off is made differently-they put all of their energy
into reproduction and in this way increase the chances that more of their
offspring will survive. Just why the tradeoff is resolved differently in trout
versus salmon is unclear.
The concept of fitness-an organism's genetic contribution to future
generations-is central to analyses of such problems. Almost all current
models use as a criterion for fitness the reproductive success of an in-
dividual, that is, the number of its offspring that survive to reproduce. But
reproductive success is in turn determined by several different factors.
The general problem was first clearly posed by Darwin in 1871:
"Thus the fertility of each species will tend to increase, from the more
fertile pairs producing a larger number of offspring, and these from
their mere number will have the best chance of surviving, and will
transmit their tendency to greater fertility. The only check to a continued
augmentation of fertility in each organism seems to be either the ex-
penditure of more power and the greater risks run by the parents that
produce a more numerous progeny, or the contingency of very numerous
eggs and young being produced of smaller size, or less vigorous, or
subsequently not so well nurtured. To strike a balance in any case
between the disadvantages which follow from the production of a
numerous progeny, and the advantages (such as the escape of at least
some individuals from various dangers) is quite beyond our power of
judgment."
The balance spoken of by Darwin is the basis of current life history
theory. For example, there is often a trade-off between a parent's re-
production and its survival; helping offspring to survive reduces the chance
that the parent will survive to reproduce again. A simplified representation
of this trade-off is shown in Figure 1. The exact form of the trade-off curve
depends on the species and its particular environment. To determine this
form, one must know the extent to which producing more eggs or providing
more care for offspring reduces the parent's chances of reproducing in
the future. Given such a trade-off, the theory of natural selection allows us
to calculate the point on the trade-off relation which provides greatest
reproductive fitness (illustrated on the curve).
For parasitoid wasps (and many other insects), the production of off-
spring involves at least two general components: host selection (including
the habitat searched for hosts) and determination of clutch size. For this
discussion, we will generally ignore a third important component, the sex
ratio (reviewed for parasitoids in Charnov (1982) and discussed in a pre-
vious symposium by Frank (1983)). The parasitoid and general insect
literature shows two rather different approaches to host selection problems.
The first, well illustrated by the recent reviews of Vinson (1976) and Vinson
and Iwantsch (1980), is the proximate mechanism approach. Clearly off-
spring production involves the factors of (1) habitat selection (where does
the female parasitoid search for hosts?), (2) the detailed search for hosts,
March, 1984
Insect Behavioral Ecology-'83 Charnov and Skinner
Cl
0 \ \
R \
U.,
O \
01
Ui-
ADULT SURVIVAL
Fig. 1. An assumed tradeoff between the mother's yearly survival and the
survival to adulthood of the single offspring born each year leads to an
intermediate for both being favored by selection. Population genetic tech-
niques (for example, Charnov 1982) show that the equilibrium is at point b,
where a line of slope minus two is just tangent to the tradeoff curve.
(3) the acceptance or rejection of "hosts" found and (4) the suitability of
the chosen hosts for offspring growth and survival. All of these must be
mediated by various cues (chemical, tactical, etc.) and a great deal of re-
search has been devoted to sorting them out. Thus, one attempts to under-
stand in great detail the immediate or proximate mechanisms involved in
the process of offspring production. This approach is rather different from
the natural selection approach where we calculate the fitness-maximizing
alternative from among those possible in a trade-off situation. This natural
selection or proscriptive approach says what the beast ought to do, not
how it arranges to do it. Of course, both approaches are useful and interest-
ing; it's just that they ask different questions. That is the two approaches
are not alternatives, but are complementary.
The natural selection approach is the newer of the two, although sex
ratio decisions (Charnov 1982) enjoy a large literature. Let us briefly
mention previous work, with parasitoids or related (life-history-wise)
Florida Entomologist 67(1)
insects. Two important early papers are Klomp and Teernik (1967), on
optimal clutch size in Trichogramma, and Mitchell (1975) for the same
applied to a bean weevil. More recent work on Trichogramma applies precise
adaptational models to clutch size and sex ratio (Charnov 1982, Waage and
Ming 1983, Waage and Lane 1983). Green (1982) has similarly considered
the coevolution of host choice and sex ratio for solitary parasitoids. Chew
and Bobbins (1983) have reviewed oviposition decisions in butterflies from
both the proximate and natural selection viewpoints. Weis, Price and Lynch
(1983) have modeled clutch size for a gall making Dipteran. Finally, Skinner
(1983a) and Parker and Courtney (1983) have discussed some general
adaptational models for several aspects of insect oviposition. Our work
discussed here is in the spirit of these last two papers.
LACK'S THEORY FOR BIRD CLUTCH SIZE . APPLIED TO WASPS
In the 1940's, the great British ecologist, David Lack, became interested
in the factors affecting clutch size in birds. Why did some sea birds attempt
to rear but a single offspring, while some titmice produced a dozen? And
why did clutch size within a species alter from year to year or with lati-
tude? His approach to the determinants of clutch size consisted of asking
the ultimate question of the consequences on the parents' fitness of rearing
a clutch of a particular size. The fundamental idea was quite simple. Sup-
pose that the survivorship to adulthood of each offspring declined with in-
creasing clutch size (perhaps due to less food available for each child), as
shown in Figure 2. This figure is simply illustrative, as the survival decline
may well not be linear. Disregarding clutch size influences on parental
survival, this survival decline would mean that some intermediate clutch
size would be the value which resulted in the largest number of surviving
offspring, a fairly good measure of parental fitness. As a first order theory,
we might ask if the clutch size of a primary parasitoid, for each single host,
obeys a rule like Lack's hypothesis. Do female parasitoids lay a clutch which
maximizes the number of offspring surviving to adulthood?
The application of these ideas to parasitoids is, however, a bit more
complicated than for birds, for 'one simple reason. Birds have determinate
growth and same-sex adults of a given species are roughly all the same
size. Insects have no such constraints and size variation among reproductive
adults may be several-fold. Thus immatures growing up in a crowded host
may simply emerge as small adults. Thus crowding may cause immature
survival to decline and/or adult size to decrease. Adult body size may have
large effects on adult fitness, altering both life span and egg production (cf.
Charnov et al. 1981). Figure 3 shows data from the parasitoid literature
which illustrate these notions. In 3a larval crowding (for a bruchid weevil
attacking beans) results in a linear decline in survival. In 3b, data from
Salt's (1940) work on the parasitoid Trichogramma shows how larval crowd-
ing lowers the resulting adult size. Figure 3c is a hypothetical relation
between female adult size and lifetime egg production, a measure of lifetime
fitness for an offspring. If these data were available for one parasitoid in a
specified host, we could easily calculate the total fitness through offspring
(here restricting our calculations to female offspring) a mother would
realize through a given clutch size. This value (W,) would be the product of:
March, 1984
Insect Behavioral Ecology-'83 Charnov and Skinner
"-.0
a
o
T I
0 Cx
0
survival to adulthood of each offspring declines, resulting in an intermediate
clutch size which maximizes the number of surviving young. All else equal,
C 0
\
size) (1) C
two parasitoids.
A. Trichogramma embryophagum
aI
Ic I-
SClutch si e e e eee
Fig. 2. Lack's hypothesis for clutch size. As clutch size increases, the
survival to adulthood of each offspring declines, resulting in an intermediate
clutch size which maximizes the number of surviving young. All else equal,
this is the size favored by selection (Charnov and Krebs 1975).
Wd = (clutch size)x(proportion of offspring surviving to adulthood) x
(size of offspring) x (lifetime egg production for offspring of this
size) (1)
Under Lack's hypothesis applied to parasitoids, natural selection would
favor the clutch size which maximizes Wf. Let us now apply this theory to
two parasitoids.
A. Trichogramma embryophagum
Clutch size and sex ratio (from a natural selection view) have been
studied in several species of egg parasitoids of the genus Trichogramma
(Charnov 1982, Waage and Ming 1983, Waage and Lane 1983, Klomp and
Teerink 1967). Here we briefly review the classic work of Klomp and
Teerink (1967). For the species T. embryophagum, they studied clutch size
in hosts of a wide size range (100-fold range, although five of the six host
species were in a 10-fold size range). As expected, the wasp's clutch size
increased with host size. For three of the hosts (two large, one small) they
determined the effect of clutch size on offspring survival and final adult size.
Florida Entomologist 67(1)
-a
March, 1984
- K
-\
1 2 3 4
Eggs per bean
6.0 I-
4.0 -
I I I I I 1 *-* I
2 4 6 8
Progeny per host
Adult female size
Fig. 3. Clutch size tradeoffs for parasitic insects. a. A bruchid weevil
attacking beans; immature survival declines with clutch size (data from
Mitchell 1975). b. The wasp Trichogramma; in crowded hosts, the adults
emerge at a smaller size (data from Salt 1940). c. A hypothetical relation-
ship between adult size at emergence and the adult's lifetime fecundity.
These 3 relations allow one to calculate mom's fitness (through offspring
from a single host) for a specified clutch size, and thus to derive the insect
parasitoid analogue of Figure 2.
0.6
0
S0.4
Ix
0.2
Insect Behavioral Ecology-'83 Charnov and Skinner 11
In Figure 4a, we show the survival effect for the small and one large host
type. Both show declines, but as expected the smaller host goes down fastest.
Figure 4b shows a similar plot for clutch size versus resulting adult size.
Adult size clearly declines with clutch size. Under lab conditions they also
determined the lifetime fecundity for female wasps as a function of female
size. Figure 4c shows the resulting relation. While the data showed a fair
bit of scatter (and the lab may not represent the field), as a first approxima-
tion we may take Figure 4c to translate female size into lifetime fecundity.
Figures 4a-c provide sufficient data to calculate W, versus clutch size for
two host sizes (actually three since we omitted from Figure 4 the other large
host). Figure 5 shows such a plot, with the hosts labeled. The most productive
clutch sizes for the three hosts (in increasing order) are four, seven, and
nine eggs respectively. From the lab data on oviposition, the parasitoid laid
the following ranges of eggs in these three host species: 1-2, 5-8, 5-8. As can
be seen, in general the parasitoid laid a smaller clutch than that predicted by
Lack's hypothesis.
B. Nasonia vitripennis
Nasonia vitripennis is parasitic on the pupae of numerous cyclorraphous
Diptera, particularly in the families Calliphoridae and Sarcophagidae
(Whiting 1967). The ones attacked are largely carrion feeding flies, al-
though one genus utilized by Nasonia (Protocalliphora) is parasitic on
nestling birds. The wasp has been the subject of extensive laboratory re-
search and its general biology is well known (see reviews by Whiting 1967
and Cassidy 1975). The treatment of the data presented here is from
Skinner (1983b).
Nasonia attacks several fly species, which in our study sites range 12-
fold in body volume. Phormia regina is the most abundant host. Under less
than severe larval crowding, Nasonia shows no increased larval mortality.
However, the size of emerging adults declines with crowding. Figure 6a shows
data on this for 4 host 'sizes of Phormia. For each of the four there is a
significant linear decline in wasp size with increasing clutch size. Figure 6b
shows lab data on the number of oocytes in a female as a function of her
head width. There is an almost perfect linear regression which suggests
that females with heads of width .5 mm or less will have zero fecundity. Un-
fortunately, we have no data relating oocyte number to lifetime female
fitness; thus here we will make the simple assumption that oocyte number
is proportional to lifetime fitness. If we combine this with the last two
figures, we can calculate Wf versus clutch size for the four host sizes. These
calculations are shown in increasing order (a to d) in Figure 7. Now . .
how do the predicted "most productive clutches" (as a function of host
size) compare to the clutches found in nature? Figure 8 shows field data for
Nasonia clutch size versus host volume. The associate the "Lack clutch
size" with its appropriate host volume (for the four host volumes of Figure
6a). Again, most all the clutches found in nature are smaller than those
predicted by the Lack hypothesis. Indeed, here the hypothesis predicts rather
well the maximum clutch size observed.
WHY DOES THE HYPOTHESIS FAIL?
Rather than predicting the average clutch size, the Lack hypothesis ap-
12 Florida Entomologist 67(1) March, 1984
100 -0-A--A-A--A-A-A- -
SAa\
75
0 \
S50- 0,0
25- 00
5 O
a..
4 8 12 16
Eggs laid per host
13 -
Sb
oO 11 -
)9 -A A A
O ,. ,.O O-
7 O-
4 8 12 16
Eggs laid per host
300
C
:5 200
100
I I I
6 8 10 12
Adult head width
Fig. 4. Progeny fitness for the wasp Trichogramma. a. Immature survival
as a function of clutch size for a large (A) and small (0) host species. See
Figure 5 for host key.
b. Emerging adult ( 9 ) size versus clutch size for the same two hosts.
c. Lifetime female fecundity (total number of eggs produced) versus female
size in laboratory experiments. (Data from Klomp and Teerink 1967)
Insect Behavioral Ecology-'83 Charnov and Skinner 13
5
a4<
C
/ 3
/
S0 *- Ellopia
S0O A Bupalus
L.1 Anagasta
I I I I I I
2 4 6 8 10 12
Number of eggs laid
Fig. 5. Data from Figure 4 allow us to plot the mother's fitness (through
offspring on a host) versus clutch size. Shown here are plots for 3 host
species (2 of them from Fig. 4). See text for further discussion.
pears, in the case of both species, to be predicting clutch sizes just a bit
larger than typically observed. The hypothesis fails. But why? We could at
this point simply give up the natural selection approach to clutch size but
such a decision seems very premature. There are at least three very natural
life history factors which we ignored in equation 1, either in the original
hypothesis or in its application to the wasps. Actually, we (i.e., we two
authors) did not ignore these factors in our theory making (we built a
large range of selection models at the beginning of the research), but for
this exposition chose to present the simplest model and use its confrontation
with data to suggest what to do next. The three factors we have so far
ignored are:
(1) A possible negative correlation between increasing clutch size and the
mother's adult survival (or a negative relation between the rate of egg
production and mother's survival).
(2) The possibility that our measure of offspring fitness (i.e., their life-
time egg production) is incomplete.
(3) The possibility that offspring production per host (that is, the mother's
fitness through the offspring she produces in each host) is an incomplete
measure of maternal fitness.
Factor 1: It is interesting to note that several years of applying Lack's
hypothesis to birds showed a pattern similar to the wasps; the clutch sizes
Florida Entomologist
Volume: 15.00-19.99
n=9
67(1)
*
* *
March, 1984
Volume: 20.00-24.99
n=10
:*
00
I I I I I I I I
Volume: 30.00-34.99 Volume: 35.00-39.99
n=10 n=7
- *
*
0
*
*0
SI
10 20 30 40 10 20 30 40
Number in brood
* ~
0,000
m
Female head width (mm X 10)
Fig. 6. a. In the wasp Nasonia, increased larval crowding typically results
in smaller emerging adults, illustrated here for four host sizes of the fly
Phormia. Immature survival did not decline with increasing clutch size. b.
Larger females have more oocytes and presumably greater lifetime fitness
(data from O'Neill and Skinner).
8 a
40 --
20-
4.5
Insect Behavioral Ecology-'83 Charnov and Skinner
20 -
co C
10 -
LLI I
10 20 30
Number of eggs on host
Fig. 7. Relative Nasonia adult fitness (through offspring production) as a
function of clutch size for the four host sizes of Figure 6 (under the as-
sumption that oocyte production is proportional to lifetime egg production).
observed in nature were' generally smaller than the theory predicted. A
decade ago Charnov and Krebs (1974) suggested that this might still
follow from the natural selection model if adult yearly survival decreased
with increasing clutch size. While supporting data are scant, the theoretical
effect is real. There are several ways in which "egg production rate" might
be negatively related to the parents' survival rate. Since natural selection
is concerned with "lifetime adult fitness", such a correlation would mean
that "fitness" would often be maximized at a smaller than Lackian clutch
size (the math details will be published elsewhere).
Consider (2) : in the Nasonia calculations we assumed that lifetime off-
spring (daughter) fitness was proportional to oocyte production. This is the
same as assuming that each female lays the same proportion of her potential
brood. Clearly, this need not be the case; a likely possibility is that larger
wasps have relatively longer lifespans and thus lay relatively more eggs. If
so, then small wasps are penalized more than the proportional assumption
allows; then the "true" Lack clutch size is smaller than our calculated one.
The same might apply to the Trichogramma data if the field is harsher on
small females than the lab tests suggest (see also Waage and Ming 1983).
Now consider (3): Begin with a simple question-which of these two
situations produces greater maternal fitness? (Assume offspring survival and
size are thesame for both.): Jdt -
Florida Entomologist 67(1)
40
n=190
O-1
-2
*-3
D) 30 0
NO
01 oo0 S
020 *
M 0
0:. :. *
o **
** :* e
0 *:*** ** **
I ?*
10 20 30 40 50
Estimated puparial volume
Fig. 8. Field clutch size data for Nasonia (Phormia hosts). The dotted
line, through the E points, shows the maximum observed clutch sizes, as a
function of host volume. The four stars (*) are the calculated "Lack clutch
size" from the two previous figures. Note that the Lack hypothesis predicts
fairly well the largest clutches, but certainly not the average clutch sizes
(see Skinner (1983b) for more details).
a) 20 eggs per host, 40 minutes to find and handle each host
b) 30 eggs per host, 65 minutes to find and handle each host
Clearly, b provides greater maternal fitness under the Lack hypothesis (that
is, greater fitness through the offspring produced in each host), but (a)
provides greater maternal fitness per unit tim since each host requires
relatively less time to find and attack. The latter can be written as:
Total fitness through offspring/host
(2)
Total time per host
We have studied several population genetic models on these clutch size
problems and this latter rate measure of fitness is usually a better indication
of the direction of natural selection. (See also Parker and Courtney (1983)
and Skinner (1983a).) And this rate measure applies even if the other
two considerations do not (indeed its use often depends on the assumption
that clutch size and mother's survival are uncorrelated). We now consider
the implications of this alteration of the Lack hypothesis.
ON USE OF THE RATE DEFINITION OF FITNESS
To use the rate definition of fitness, we must first translate clutch size into
March, 1984
Insect Behavioral Ecology-'83 Charnov and Skinner 17
time. Upon encountering a host at least two sorts of time might be involved in
producing a particular clutch size: time to mature the eggs and time to lay
them. The first of these may be related to the general nutritional condition
of the mother, or perhaps the time since the last oviposition. Considering
both times (called here oviposition time) should allow us to associate with a
given clutch size a particular time to get it into a host just discovered. If
this is known, then the clutch size or X-axis of Figures 2, 5 or 7 can simply
be rescaled into oviposition time. We have done this, for a hypothetical
example, in Figure 9a. Such a plot gives us the mother's fitness through
the offspring produced on a host as a function of time at the host. But the
total time per host includes the search time between hosts, and this is also
indicated on the graph. For simplicity, we limit this discussion to a single
kind of host. Note that the "Lack clutch size" is the time which gives the
greatest maternal fitness per host; i.e., the peak of the curve. However,
the rate hypothesis says that selection favors the clutch size which maxi-
mizes the fitness gained per unit time (Equation 2). To find this, note the
following on Figure 9a. We graph the time between hosts (the search time)
increasing to the left of the oviposition curve. Now, if we draw a line from
the search time (ts), the line has a slope which is in units of Wf/(ts + t),
which is equivalent to Equation 2. Thus, we can find the optimal clutch size
(here, optimal oviposition time) by finding the line through the specified
search time which intercepts the Wf curve at the highest point possible. This
is illustrated in 9a. In 9b we illustrate the effect of increasing search time,
while holding the W, curve (the oviposition relation) constant. Here the
predicted clutch increases in size (from t2 to t,). We note here that only if
search time is large is the predicted clutch size as big as Lack's hypothesis
would predict. Otherwise, the clutch size favored by natural selection is
smaller than the Lack size. This is exactly what the parasitoid data show-
the Lack size is an outer bound and most clutches are smaller. This rate
hypothesis makes two other sorts of predictions. First is that shown in 9b.
Increasing search time should increase the clutch size. The second is that
the oviposition relation may be sensitive to the rate at which eggs can be
matured. It includes at the minimum the time to lay the eggs; if this is all
it includes, then the faster they are laid the steeper the Wf curve rises, and
the closer the optimum clutch is to Lack's. By manipulating maternal nu-
trition, one ought to be able to alter the oviposition relation, and see if the
clutch laid alters accordingly. Unfortunately, we know of no data which
bear on these issues. (Nor do we know of any data which bear on whether
maternal survival is itself negatively related to the egg production rate
(Factor 1). Such a correlation would similarly favor a smaller clutch size, as
previously discussed.)
We now turn to a second general question for parasitoids, that of which
hosts to attack in the first place.
HOST SELECTION
A vast literature documents that parasitoids of a given species "prefer"
some host types to others. It seems reasonable to conceptualize this choice
in the same fitness terms as the clutch size problem. This is not a new
suggestion; models of optimal foraging (patch choice, food item choice, etc.)
have long been used to view insect (including parasitoids) diet choice
18
CO
CO
(D
Ca
a)
75
Eu
E
(a)
Florida Entomologist 67(1)
I1=U L2-U L20pt liopt
Search I Oviposition
time (ts) time (t) I
Fig. 9. Natural selection should favor the maximum rate of fitness pro-
duction through offspring (all else equal), rather than the maximum per
host (Lack's hypothesis), a. This figure shows how to calculate that clutch
size, as a function of the search time between hosts (ts), the oviposition
time once a host is encountered (t) and the parent's fitness through off-
spring (Wf) ; related to clutch size, thus to the time (t) to deposit a clutch
of a given size). The clutch size-(= oviposition time) favored under Lack's
hypothesis is at the peak of the curve. b. If the search time between hosts in-
creases (t2 -> t,), the most fit clutch size increases. The limiting or largest
clutch is Lack's.
(e.g., Waage 1979, Green 1982, Charnov 1976a,b, Charnov and Orians 1973).
We will not review this literature here but instead will view one new
problem: superparasitism. Superparasitism, where a host already attacked
by a member of species Z is then further attacked by a second individual
of Z, is widely described in the parasitoid literature as a "screw up", a
mistake on the part of the second individual (van Lenteren 1981). Recently,
Alphen and Nell (1982) suggested that superparasitism may sometimes be
an advantageous (adaptive) response for the second individual. Inde-
pendently, we reached the same conclusion through treating the host popu-
lation as two separate kinds of prey (unattacked and attacked) and simply
asking when the mother should include the poorer, already attacked hosts in
those she attacks. In general, the answer to this optimal foraging question
turns on factors such as how abundant the good hosts are, how easy eggs
are to produce, how poor the already attacked hosts are for offspring pro-
March, 1984
Insect Behavioral Ecology-'83 Charnov and Skinner 19
duction, and so forth. In Figure 10, we show graphically how to answer the
question of whether superparasitism ought to occur or not, using our pre-
viously defined oviposition relation. If superparasitism happens, there is the
further question of the respective clutch sizes for the 1st and 2nd parasite.
Such clutch size questions are beyond the technical scope of this paper and
the reader is referred to the work of Parker and Courtney (1983).
IN CONCLUSION
We close this paper with the simple observation that our natural se-
lection perspective has focused attention upon specific variables (e.g., search
time, maternal survival, offspring survival and size related to brood host)
seemingly important in understanding host utilization in parasitoids; vari-
ables which become part of a comprehensive and integrated approach to the
problem. Of importance to us is not the present lack of answers, but the
host of new questions, ones which readily suggest tests in lab and field. The
natural selection approach also raises questions about the nature of the
proximate mechanisms involved with the adaptations.
ACKNOWLEDGEMENTS
We thank D. Temme and G. Jeppesen for help with the experiments, and
0)
3 AP
ts=O
Search Oviposition
time (ts) time (t)
Fig. 10. When is it advantageous to superparasitize? The upper curve is
a non-parasitized host, and the tangent argument from Fig. 9 shows the
optimal oviposition time (in the absence of superparasitism). The AP curve
shows the mothers fitness gain for attacking an "already parasitized" host.
The dotted line is parallel to the upper tangent line. If it passes through the
AP curve, as it does here, superparasitism is favored. (Argument from
Charnov 1976a,b.). Optimal clutch sizes under superparasitism are treated
in Parker and Courtney (1983).
Florida Entomologist 67(1)
NSF for paying the bills. J. Lloyd, J. Endler, J. Alcock and R. Thornhill
read and improved the paper. The mistakes still left are the fault of Jim
Bull, as always. Maurine Vaughan expertly typed the paper and helped keep
us sane. Sine and Cosine helped give meaning to ELC's life.
LITERATURE CITED
ALPHEN, J. T. M. VAN AND H. W. NELL. 1982. Superparasitism and host dis-
crimination by Asobara tabida (Braconidae: Alysiinae), a larval
parasitoid of Drosophilidae. Netherlands J. Zool. 32: 232-60.
CASSIDY, J. D. 1975. The parasitoid wasps Habrobracon and Mormoniella. In
R. C. King (ed.) Handbook of Genetics, 3: 173-203, Plenum Press.
CHARNOV, E. L. 1976a. Optimal foraging: attack strategy of a mantid.
American Natur. 110: 141-51.
1976b. Optimal foraging: the marginal value theorem. Theor. Pop.
Biol. 9: 129-136.
S1982. The Theory of Sex Allocation. Princeton Univ. Press, 355 pp.
-- AND G. H. ORIANS. 1973. Optimal foraging: some theoretical ex-
plorations. (Book published by the authors, 160 pp.)
AND J. R. KREBS. 1974. On clutch size and fitness. Ibis 116: 217-19.
--- R. L. LOS-DENHARTOGH, W. T. JONES AND J. VAN DEN ASSEM. 1981.
Sex ratio evolution in a variable environment. Nature 289: 27-33.
CHEW, F. S. AND R. K. ROBBINS 1983. Egg laying in butterflies. In Biology
of Butterflies, Symp. R. Ent. Soc. London 11: (in press).
DARWIN, C. 1871. The Descent of Man and Selection in Relation to Sex.
John Murray, London.
FRANK, S. A. 1983. A hierarchical view of sex ratio patterns. The Florida
Entomologist 66: 42-75.
GREEN, R. F. 1982. Optimal foraging and sex ratio in parasitic wasps. J.
Theor. Biol. 95: 43-48.
KLOMP, H. AND B. J. TEERINK. 1967. The significance of oviposition rates in
the egg parasite, Trichogramma embryophagum. Arch. Neerl. Zool.
17: 350-75.
MITCHELL, R. 1975. The evolution of oviposition tactics in the bean weevil,
Callosobruchus maculatus. Ecology 56: 696-702.
PARKER, G. A. AND S. P. COURTNEY. 1983. Models of clutch size in insect
oviposition. (Submitted: Theor. Pop. Biol.)
SALT, G. 1940. Experimental studies in insect parasitism. VII. The effects
of different hosts on the parasite Trichogramma evanescens. Proc. R.
Ent. Soc. London (A) 15: 81-95.
SKINNER, S. W. 1983a. Clutch size as an optimal foraging problem for
insect parasitoids. (Submitted: Oecol.)
SKINNER, S. W. 1983b. Clutch size and sex ratio ecology of a parasitoid
wasp, Nasonia vitripennis. (Submitted: J. Anim. Ecol.)
VAN LENTEREN, J. C. 1981. Host discrimination in parasitoids. In: The role
of semiochemicals in insect behavior. (Jones, Lewis and Nordlund,
eds.) Wiley and Sons.
Vinson, S. V. 1976. Host selection by insect parasitoids. Annu. Rev. Ent.
21: 109-33.
VINSON, S. V. AND G. F. IWANTSCH. 1980. Host suitability for insect parasi-
toids. Annu. Rev. Ent. 25: 397-419.
WAAGE, J. K. 1979. Foraging for patchily-distributed hosts by the parasi-
toid, Nemeritis canescens. J. Anim. Ecol. 48: 353-371.
--- AND J. A. LANE. 1983. The reproductive strategy of a parasitic wasp.
II. Sex allocation and local mate competition in Trichogramma evanes-
cens. (in press-J. Anim. Ecol.)
March, 1984
Insect Behavioral Ecology-'83 Charnov and Skinner 21
--- AND N. S. MING. 1983. The reproductive strategy of a parasitic wasp.
I. Optimal progeny and sex allocation in Trichogramma evanescens.
(in press-J. Anim. Ecol.)
WEIS, A. E., PRICE, P. W. AND M. LYNCH. 1983. Selective pressures on clutch
size in the gall maker Asteromyia carbonifera. (Ecology, in press).
WHITING, A. R. 1967. The biology of the parasitic wasp Mormoniella
vitripennis. Quart. Rev. Biol. 42: 333-406.
Florida Entomologist 67(1)
BODY BUILDING BY INSECTS: TRADE-OFFS IN
RESOURCE ALLOCATION WITH PARTICULAR
REFERENCE TO MIGRATORY SPECIES
MARY JANE ANGELO AND FRANK SLANSKY JR.*
SYNOPSIS
Individuals of many species of insects inhabiting transient habitats ex-
hibit migratory behavior and thereby escape deteriorating environmental
conditions and colonize sites with favorable conditions. Limited food avail-
ability is one cue that may serve to induce migration, either by stimulating
the adult to undertake migration, and/or by stimulating the larva to alter its
resource allocation to "build" a migratory-form adult. Such changes in re-
source allocation often involve changes in lipid storage, body size, proportion-
ing of wing size and body weight, and reproduction.
Detailed study of four species of presumed migratory noctuid moths indi-
cated that the larvae retain the ability, when starved from various days in
the last larval stadium, to pupate and metamorphose into adults. These
adults have reduced body weights (from 12 to 24% of the weight of adults
from fully fed larvae), and significantly lower wing loading ratios (i.e.,
body weight/wing area) than would be expected based on the predicted
allometric relationship between body weight and wing area. We propose that
this altered allocation of food to body weight and wing area is an adaptive
response producing individuals with low wing loading ratios that pre-
sumably exhibit less energetically costly flight. This hypothesis is consistent
with principles of flight energetic.
"The earth-bound early stages built enormous digestive tracts and
hauled them around on caterpillar treads. Later in the life-history
these assets could be liquidated and reinvested in the construction
of an essentially new organism-a flying-machine devoted to sex."
(C. M. Williams 1958)
". .. the central problem of evolutionary biology: to provide a
general explanation for the design of organisms." (S. C. Stearns 1982)
INTRODUCTION
The liquidation and reinvestment of "assets" by insects during meta-
morphosis, mentioned in Williams' quote above, involves the allocation of
acquired resources to building and maintaining the adult body. In addition,
energy and nutrients are necessary for adult activities, including defense,
dispersal and reproduction (Townsend and Calow 1981). We are in particu-
lar interested in the trade-offs of resource allocation among body-building,
dispersal and reproduction.
Differences in resource allocation may be evident both among and within
*Mary Jane Angelo is a graduate student in the Department of Entomology and Nema-
tology, University of Florida. Her research interests include resource allocation by insects,
especially migratory species. Frank Slansky Jr. is an assistant professor in the same De-
partment. His research interest is in the nutritional ecology of larval and adult insects within
both basic and applied contexts.
Current address: Department of Entomology and Nematology, University of Florida,
Gainesville, Florida 32611. Florida Agricultural Experiment Station Journal Series No. 5198.
March, 1984
Insect Behavioral Ecology-'83 Angelo and Slansky, Jr. 23
species. For example, an insect with a high-powered, hovering type of flight
exhibits differently proportioned body and wings compared to an insect with
a low-powered, gliding type of flight (Kammer and Heinrich 1978, Casey
1981a). Within a species, resource allocation may vary between generations.
For example, individuals of certain seasonally migratory species respond to
changing environmental conditions (e.g., photoperiod, food quality and
quantity, and larval density) by altering their allocation of resources to
body and wings, with consequent effects on fecundity (see below).
It is one goal of nutritional ecology to achieve an understanding of the
patterns of resource allocation by species (Slansky 1982a, 1982b; see also
Stearns 1982). In this paper we examine resource allocation to body
building in insects, with particular reference to migratory species. Using
data from the literature and our own unpublished data, we investigate the
trade-offs among body weight, lipid content, wing size and reproduction. Our
discussion is framed within the contexts of alteration of resource allocation
in changing environments and the principles of flight energetic. In
addition, we identify areas where more research is needed.
RESOURCE ALLOCATION TO BODY SIZE, PROPORTIONS AND COMPOSITION
Within a given environment, there will be some optimal adult body size
(or sizes) for the individuals of a particular species that yields their maximal
fitness. Selective forces acting at different times during an insect's life
determine the evolution of optimal adult body size (Roff 1981, Ricklefs
1982). Such forces involve an individual's size-dependent relationships with
food (Enders 1976, Wasserman and Mitter 1978, Mattson 1980), enemies
(Hespenheide 1973, Enders 1975), competitors (Pearson and Stemberger
1980, Eberhard 1982) and the physical environment (Sweeney and Vannote
1978, Roff 1981), as well as the size-dependency of life history variables
such as developmental time, fecundity and dispersal behavior (Dingle et al.
1980, Derr et al. 1981, Hinton 1981, Roff 1981, Ricklefs 1982, Stearns 1982).
Resource allocation to body building involves not only achieving a certain
optimal size, but also, within the limits of that size, partitioning the re-
sources among the various organs, structures and biochemical components
of the body such that the insect will achieve its best performance (Calow
1977). One important feature of resource partitioning that is responsible
for the different proportions of the various organs and structures to the
body as a whole is allometric growth, whereby some tissues in the body
grow at faster rates than others (Huxley 1972). Concentrations of the
various biochemical components of the body (i.e., proteins, lipids, carbo-
hydrates, amino acids, salts, etc.), also important to insect performance,
result from the complex relationships of assimilation, anabolism, catabol-
ism and excretion (Wigglesworth 1965, Gordon 1972, Rockstein 1978).
ADAPTIVE ALTERATION IN RESOURCE ALLOCATION: MIGRATORY INSECTS
Individuals of many species in habitats that can become adverse for re-
production or survival exhibit migratory behavior (Southwood 1977, Welling-
ton 1980). These individuals can thereby escape deteriorating habitats and
colonize favorable sites (Dingle 1978). The decision to migrate may be
made either during the adult or larval stage.
Adult insects may initiate a migratory response based on their evaluation
Florida Entomologist 67(1)
of environmental cues such as photoperiod, temperature, and food quality
and quantity (see Slansky 1982a). Induction of reproductive diapause
commonly is associated with the decision to migrate. The integration of re-
productive and flight behaviors occurs at the neurohormonal level (DeWilde
and DeLoof 1973, Rankin 1978) and involves the investment of resources
in the flight system and its metabolism rather than in reproduction (Slansky
1980, Heinrich 1981).
The flight system is energetically costly to maintain and operate: meta-
bolic rate during flight may be increased 50-100 times that at rest, and over
50% of an insect's resting metabolism may result from the metabolism of the
resting flight muscles (Kammer and Heinrich 1978, Heinrich 1981). How-
ever, the extent to which metabolic costs of flight may divert energy
away from other behaviors, and thus reduce reproduction and survival
probability is not clear. Significant negative effects of flight duration on
reproduction and survival have been found in tethered flight studies of
only some of the species studied (for review see Slansky and Scriber
1984). Perhaps tethered flight techniques do not put as great a metabolic
demand on a flying insect as does free flight. In addition, although there
are few data available, insects may exhibit adaptations that reduce the
potential negative effects of flight. For example, adequate storage of lipid
reserves prior to long flight may satisfy the energetic demands of the flight
muscles, such that sufficient energy stores remain for reproduction and
survival; also, post-flight increases in food consumption may occur (see
Slansky 1980). Reduction of metabolic costs may result from histolysis of
flight in muscles in some situations where the flight system is no longer
needed by the insect, such as after the mating flight in termites and when
certain scolytid bark beetles have dispersed to and colonized a host tree
(Atkins and Farris 1962, Johnson 1973; see also Solbreck and Pehrson 1979).
Whereas adults of some species may themselves decide to migrate, adults
of other species may, under certain circumstances, produce offspring
destined to exhibit migratory tendencies as adults. Larval (or nymphal) re-
sponse to environmental cues also may influence adult migratory behavior.
Because these responses occur before the adult stage, differential allocation
of food during the larval and pupal stages may result in migratory forms
being produced that differ from non-migratory adults in the size, proportions
and biochemical composition of their body, as well as in other aspects of
their biology (Johnson 1976, Wellington 1980; Table 1). Among the most
dramatic examples of this phenomenon are aphids and locusts (see Dingle
1980).
Several species of aphids display alary polymorphism associated with an
alteration between sexual and parthenogenetic generations (Dixon 1973).
Production of the alate (winged) form may occur in response to dietary
influence (Mittler and Sutherland 1969) or population density (Dixon 1973)
and may depend upon the population age structure (MacKay and Lamb
1979). Differences in food consumption and fecundity between alate and
apterous forms (Table 1) indicate further differences in resource allocation.
In migratory locusts, individuals of the gregarious phase, produced in re-
sponse to increased density, differ from solitary phase individuals in
morphology and reproductive and migratory behavior. They are migratory
throughout their lives, first as nymphs "marching" across the terrain en
March, 1984
Insect Behavioral Ecology-'83 Angelo and Slansky, Jr. 2
LO
a:d
-E-
0_ t-
sl ^ S
Z 0 0
zz
(n 00V
5i r^ t-
0 U2
I i g i -B ?
-- 3
m 00
" n -
z ZO a) CdC
0 "~~, .~ .d
4 0 .-
ZZ~~ 0Q)~>
-~ .lF- 0 *~ r E
SV bl w4-
1-o m 0 0
a) ) k)C
W S ; k l 1 F
00
-4 > V m it mtI |
4-1 m > E Cd Ce Cd
I HC III | tl Id X ii I
C) Q 4a d a ) C O
Cd (2) "V 0 4 9.1 d -4
-i I, 1. .^ 1 1
M Cd m
d M C1. bo'? 4Js* s s^ ^ l
03 k V r| +- U2 .
!-s 1CilC)i sil
1|-
U)
Z
m P40
S04 CO U) U2
-) w-. W 0 o () 4
Z 4Q M '0ga ^
C) 0 )
04 4 10 ~ C
;L2 0 0 M 4.
1 0 C) C 0 -1
0z C)
Z Q Q Q
OCO
14) ;tF
P4 d M%)gZ
VQ k t, q g
oz
o ~ 4 41 P. 0'
-ell
Florida Entomologist 67(1)
ia
tb
05C
1-
Gb
+3
cM.
'-4
d
bk
^0
to
0
Z2
Cd
'-4
bC
0
be
.s >
.5>.
a )
0
wf
>be
gg
p3
'.b
*S-4
uiS
C)
so
F-
0) C
r
O0)
Co
Gb
ci
r.
ci
Cd
C1
ci
F>
"0
0
0
^a
+j
'-'c
b
4O
9e
.4-4
5a2
S ^
m 'S
A0
C
00
0
ra,
01
r=
*O
beH
be'
a
+3
C3)
0)
.- +3
f--4 0,
0) |
"c a
rdbeD
Gb
to
Gb
+3
b0
0
CO
F-'
0
0be
711
>S0
-o
a)
*--4 F-
C)^
Gb
to
Gl
v
be
0
,a
-C
enl
F
bo ba
cd b
30
?E13
0ci
o.a
SB$
'o
<1 &
Ss
beb
*c
'.4
CO 3
+3-
*0 0
March, 1984
Bi
>>
'0
S
S
be
0
+3
E
a}
bo
+3k
-'0
ai
Cl)
FI
43
CC
cd
CQ)
43
~bo
o
a
,+-
Zs
0
a,
Q)
4
4'
-3
SP
C)
0
0u-
CdC
E
U2
4
as
~a^
'.4
Oi
ag
&43
0>
E3 O
0
cc
'a'
21
o
r^
tCQ i
C
i0
+a
CO
r'i
0
z
0)
0;
C.)
) )
V
-S
R
a,
'.4
k
u
w
i-C
0
0)Oh
'.4
0
"a
*M
a a
bo
O-
f-
bo
43
I-
0
ci
ci
ci
H
El
bo
0a '
p~ci 0
443 S
of0
F-iEO
Btw P
i a a
8 ^s
wCM
^^-
-ed
cii
-0
0 0
.)
43
Insect Behavioral Ecology-'83 Angelo and Slansky, Jr. 27
masse, and then as adults forming large swarms that are carried for long
distances by prevailing winds (see Rainey 1978).
Differences in lipid content, wing size, body weight and fecundity be-
tween presumed migratory and non-migratory individuals also occur in
*Coleoptera, Diptera, Lepidoptera and other insects (Table 1). Although
these differences are generally less extreme than the examples cited pre-
viously, they nonetheless similarly indicate that alterations in the con-
sumption and allocation of food have occurred during the "building" of the
various adult forms.
Insect flight involves different channels of power output or cost (Figure
1). In simplified terms, the energy actually utilized for flight provides the
mechanical power output; the remaining power input is lost through muscle
inefficiency as heat. Of the mechanical power output, some is used by the
insect to accelerate and decelerate its wing mass (i.e., inertial power)
whereas the remainder is used to do work on the surrounding air (i.e.,
aerodynamic power). The three aerodynamic power output components of a
flying insect are profile power, which overcomes the drag on the surface
area of the wings, induced power, which is used to accelerate air across
the wings at a velocity sufficient to overcome the force of gravity and
parasite power, which is necessary to overcome the drag on the body (Figure
1; Casey 1981a, 1981b). The design of the adult, in particular the ratio
Fig. 1. Energy demands for power in flying insects (modified from Casey
(1981b); see text for explanation).
Florida Entomologist 67(1)
March, 1984
between body weight and wing area (i.e., wing loading ratio), along with
wing beat frequency and flight speed, in large part determine these power
output demands.
A low wing loading ratio (i.e., large wing area relative to body weight),
when coupled with a relatively slow wing beat frequency, would tend to
minimize the power output demands of flight. Many insects presumed to be
migratory exhibit low wing loading ratios compared to non-migratory in-
dividuals (Table 1), and many of these species seem to utilize air thermals
and upper air currents to aid in their dispersal, thus presumably reducing
the power output demands of their flight (Johnson 1969, Danthanarayana
1976, Gibo 1981).
RESOURCE ALLOCATION IN "LESS THAN IDEAL" ENVIRONMENTS
Associated with the strongly determinant relationship between an in-
dividual's body size, proportions and composition (i.e., its body state), and
its fitness, are adaptations that may allow the insect to attain its optimal
body state under less than ideal environmental conditions. For example, a
larva may alter its food intake in response to changes in the nutrient
content of its food, consuming more of a food with a reduced nutrient
content and thereby achieving a body weight and composition similar to
that of a larva consuming a more nutrient-rich food (Slansky and Feeny
1977). Increased catabolism and excretion also may be exhibited in re-
sponse to a nutrient imbalance in the food (Gordon 1972, Horie and Ino-
kuchi 1978, Rockstein 1978). Another food-related response involves the
induction of detoxication enzymes upon consumption of potentially toxic
allelochemicals (Brattsten 1979). These compensatory responses may incur
various costs that could reduce fitness (e.g. energetic costs of increased
feeding, catabolism and detoxication, and increased exposure to predators
and parasitoids while feeding). Thus, the "success" of a compensatory re-
sponse will be determined by the extent to which its benefits outweigh its
costs. Compensatory success, when defined as the degree to which the actual
body state approaches the optimal body state) varies among individuals,
situations and species, and may reflect different adaptive strategies (Scriber
and Slansky 1981, Slansky 1982b). However, more research is necessary
to determine the benefits and costs of compensatory responses and the
extent to which compensatory ability reflects adaptive strategies.
The frequent occurrence of prolonged larval development (in some cases
including an increase in the number of instars) under conditions of "poor"
food quality or reduced food quantity (for references see Scriber and
Slansky 1981) may be another response by which larvae attempt to get as
close as possible to their optimal body state with minimal costs. If feeding
conditions improve within a relatively short period of time, then the larva
may be able to closely approach its optimal body state with only a short
delay in the timing of pupation. However, the success of this response may
be tempered, for example, by the subsequent delay in onset of reproduction
resulting from the prolonged development. Thus in certain circumstances,
rather than exhibiting a compensatory response, larvae may be "making
the best of a bad situation", prolonging their development in order to
achieve not their optimal state but their minimal state necessary to survive
metamorphosis and produce a reproductively competent adult.
Minimal-weight-values, below which a starved immature insect does not
Insect Behavioral Ecology-'83 Angelo and Slansky, Jr. 29
undergo successful metamorphosis to an adult, vary considerably among
species (Table 2), but similar to the situation of the compensatory abilities
discussed above, we have little understanding of the ecological and evolution-
ary relationships between minimal-weight-values and adaptive strategies.
Questions such as "Are adults at the minimal weight reproductively compe-
tent?" and "Have species subject to frequent food limitation (e.g., because
of the ephemeral nature of their food in relation to the duration of the
larval stage) evolved as an adaptation minimal-weight-values that are a
low percentage of their presumed optimal weights (e.g., Drosophila
m.elanogaster, Table 2, and certain species of tree-boring Coleoptera; see
Andersen and Nilssen 1983) ?", remain to be answered.
A further adaptation involves the utilization of environmental cues by
insects to facilitate their avoidance of or escape from a deteriorating en-
vironment. Poor food quality and starvation, as well as other environmental
factors, may serve as cues to larvae of many species to produce migrant-
form adults (see above) and as cues to adults to stimulate migratory flight
behavior (Dingle 1968, 1978, Mittler and Sutherland 1969, Elsey 1974,
Sanders and Lucuik 1975, Solbreck and Pehrson 1979, Duelli 1980). These
inductory responses (Slansky 1982a) involve changes in resource allocation
that alter the optimal body state of the non-migratory form to produce the
optimal body state of the migratory form. As discussed above, this
commonly involves changes in body size and weight, and in wing size.
LARVAL STARVATION AND RESOURCE ALLOCATION IN NOCTUID MOTHS
Assuming that changes in body weight and wing area with body size
follow some theoretical relationship, then significant differences from this
relationship should indicate that resources are being differentially allocated
to body weight and wing area at different body sizes. By relating this
differential allocation of resources to the flight behavior and other aspects
of a species' lifestyle, one can infer whether this differential allocation
has adaptive value.
We investigated resource allocation in four presumed migratory species
of noctuid moths in response to larval starvation (Angelo and Slansky in
prep.). Larvae were experimentally starved from various days of their last
instar, and measurements were made of the body weight, wing area and
lipid content of newly emerged adults. Because area is the square and
weight is the cube of linear dimensions, a log-log plot of these as length
changes yields a straight line with a slope of 0.67 (indicating that area
and weight do not change in proportion to each other; if they changed pro-
portionately, the slope would equal 1). We compared the slope values be-
tween body weight and wing area for the four species with the 0.67 value,
as well as among themselves, in an effort to demonstrate differential alloca-
tion of resources during starvation. Furthermore, we related the observed
changes to aspects of these species' lifestyles.
Use of such theoretical allometric relationships is common in com-
parisons of organism performance (e.g. metabolic rate) across a range of
species with different body sizes; both the slope of the line and the extent
to which the performance value for a particular species lies off the line have
been used to draw conclusions of biological significance (Schmidt-Nielson
1970, Blueweiss et al. 1978, Greenstone and Bennett 1980). However, there
seems to have been considerably less use of comparisons between predicted
Florida Entomologist 67(1)
TABLE 2. THE MINIMAL WEIGHT AT WHICH LARVAE WILL PUPATE WHEN SUB-
JECTED TO STARVATION, EXPRESSED AS A PERCENTAGE OF THE
MAXIMAL WEIGHT ATTAINED WHEN FULLY FED. WHEN SUBJECTED
TO STARVATION BELOW THIS MINIMAL WEIGHT, LARVAE STARVE TO
DEATH. (REVISED FROM SLANSKY AND SCRIBER, 1984).
Minimal
weight
(% of
Orders/Species Sex maximal) References
Diptera
Drosophila melanogaster
Hemiptera
Oncopeltus cingulifer
O. fasciatus
Rhodnius prolixus
Lepidoptera
Anticarsia gemmatalis
Bombyx mori
Choristoneura conflictana
Danaus chrysippus
Galleria mellonella
Heliothis zea
Lymantria dispar
Manduca sexta
Plathypena scabra
Pseudaletia unipuncta
Spodoptera frugiperda
12%1
12%1
19%'
18%'
44%"
38%2
51%'
53%2
25-43%'
9 18%'
& 12%1
60%'
9 32%2
23%'
36%4
9 14%1
& 23%1
25%2
45%4
50-60%2
36-47%2
26%1
9 19%'
24%'
19%'
S. latifascia
Beadle et al. (1938)
Bakker (1959)
Blakley and Goodner (1978)
Nijhout (1979)
Friend et al. (1965)
M. J. Angelo and F. Slansky
(unpubl.)
Lees (1955)
Beckwith (1970)
Mathavan and
Muthukrishnan (1976)
Lees (1955)
M. J. Angelo and F. Slansky
(unpubl.)
Kopec (1924)
Lees (1955)
Nijhout (1975)
Higgins and Pedigo (1979)
Mukerji and Guppy (1970)
M. J. Angelo and F. Slansky
(unpubl.)
M. J. Angelo and F. Slansky
(unpubl.)
S 20%'
'Dry weight.
'Fresh weight.
'Minimal blood meal promoting molting in various instars.
4Time spent feeding in last instar.
and actual allometric relationships in interpreting differences in per-
formance values among the individuals within a species (see, for example
Casey 1976), and as a consequence, the significance of deviations of actual
data from "expected" relationships is not as well understood. Nonetheless, we
believe our logic is a good "first step" in interpreting the significance of
March, 1984
Insect Behavioral Ecology-'83 Angelo and Slansky, Jr. 31
differential changes in body weight and wing area among the individuals
within a species. In addition, we believe that between-species' comparisons
of regression slopes for within-species' allometric relationships (such as
that between wing area and body weight) can suggest meaningful differences
between species (see, for example, Casey's (1976) study on sphinx moths)
in how they respond to changes in their environment, as indicated in the
following discussion.
The first species studied, the velvetbean caterpillar (VBC) Anticarsia
gemmatalis is a presumed migratory insect that appears to fly northward
each summer from overwintering sites in south Florida and elsewhere
(Buschman et al. 1977, Greene 1979). When starved from various days in
the last stadium, larvae of the VBC produce adults that are lighter in weight
and have reduced wing area compared with fully fed larvae; the slope for
these data (0.52) is significantly different from the theoretical value (Figure
2). Thus, as body weight is decreased due to starvation, moths are produced
with significantly lower wing loading ratios than would be expected if the
theoretical relationship is applicable. Smaller moths carry less weight per
unit wing area than do larger moths, and the energetic cost of their flight
is presumably reduced (Casey 1981a, Casey and Joos 1983; also see above).
6-
*I .* **< /
4-
i S
E m-=.52 r
3-
* ^ e% m=.67
0 r=.87, P<0.001
.c m:P
2I-- -- ---------------- ---- ---|--
I I I
10 20 30 40 50 60
Adult dry weight [mg] log scale
Fig. 2. Relationship between log wing area and log dry body weight.
Dashed line: expected relationship based on area = the square and mass =
the cube of linear dimensions. Solid line: actual relationship found for adult
velvetbean caterpillar moths (both sexes) subjected to different durations
of starvation during the last larval instar; regression is significant (p<.001).
The slope of the VBC line (m = 0.52) is significantly different from the
slope of the expected line (m = 0.67; t test, p < .0001), indicating that as
body weight is reduced a relatively larger wing area is exhibited. Thus,
smaller VBC moths have a lower than expected wing loading ratio
(WLR = body weight/wing area) (from Angelo and Slansky in prep.).
32 Florida Entomologist 67 (1) March, 1984
Adult VBC, and many other species of insects, are presumably "passive"
migrators, which, although maintaining active flight, are carried for long
distances by upper air currents (see Rainey 1978, Rabb and Kennedy 1979,
Walker 1980); a lower wing loading ratio may facilitate this type of move-
ment (see above).
Migratory ability of VBC may have evolved to allow escape from
seasonally unfavorable and/or food-depleted habitats, and colonization of
seasonally favorable habitats containing suitable food. This ability to
colonize and rapidly exploit new food sources undoubtedly contributes to its
current status as a severe pest of soybean fields (see Barfield and O'Neil,
this Symposium). In soybean fields (and perhaps when feeding on wild
foodplants as well), starvation of VBC may frequently occur, especially due
to high larval densities at certain times of the year (Herzog and Todd 1980,
Linker 1980). If a relatively low minimal weight is an adaptation to fre-
quent starvation as discussed above, then the low minimal-weight-values
for VBC (Table 2) would further suggest the occurrence of frequent starva-
tion in the field. If this is the case, then the lower than predicted wing
loading ratio of the moths produced by starved larvae, involving a change
in the allocation of food to body and wings, may be an adaptive response
facilitating less costly flight in search of new larval foodplants.
Consistent with the reasoning above are the results from a similar
starvation experiment with three other species of noctuid moths: the fall
armyworm (FAW), Spodoptera frugiperda; the corn earworm (CEW),
Heliothis zea; and another armyworm (SPLAT), Spodoptera latifascia.
The FAW apparently has no diapause mechanism to allow survival during
extended periods of low temperature, and therefore in the United States it
is restricted to overwintering (with continuous generations) in subtropical
areas of Florida and Texas (Mitchell 1979). Each spring and summer,
FAW adults disperse throughout much of the United States as far north
as Canada (Luginbill 1928). Adults of the CEW are also capable of dis-
persing over long distances (Phillips 1979). However, this species has a
pupal diapause enabling it to survive cold winter temperatures throughout
much of North America as far north as Canada (Hardwick 1965). Virtually
nothing is known about the life history of SPLAT (Kovitvadhi 1969).
The wing area to body weight relationships for VBC, FAW and SPLAT
yield very small slopes (0.35, 0.22 and 0.22, respectively; Figure 3), con-
sistent with the hypothesis that these migratory moths should achieve
lower than theoretically predicted wing loading ratios at reduced body
weights to facilitate less costly flight in search of new larval foodplants.
The slope for CEW (0.59; Figure 3) also is smaller than the theoretical
slope of 0.67, but it is substantially larger than that of the other species,
suggesting that the CEW has evolved a somewhat different strategy of re-
source allocation than the other three species. This is further indicated by
the substantially different slope of the relationship between % lipid and
dry weight for the CEW (0.25) compared to those of the VBC and FAW,
which are similar (0.52 and 0.57, respectively; Figure 4). The slope for
SPLAT (0.15; Figure 4) is surprisingly more similar to that of the CEW,
from which it differs in the wing area to body weight relationship, than to
those of the VBC and FAW, to which it is very similar in the wing area
to body weight relationship (Figure 3). Like the VBC, these three species
exhibit relatively low minimal weights (Table 2).
Insect Behavioral Ecology-'83 Angelo and Slansky, Jr.
1000 FEMALES
500-
400- SPLAT m=.22_-
VBC m=.35 -
300- CEW m.5
CEW m=.59
FAW m=.22
20 30 40 50
100 200 300 400
ADULT WET WEIGHT (mg)-Log Scale
Fig. 3. Relationship between log wing area and log fresh weight for
four species of noctuid moths subjected to different durations of starvation
during the last larval instar. VBC = velvetbean caterpillar (n = 24),
FAW = fall armyworm (n = 27), CEW = corn earworm (n = 18) and
SPLAT = the armyworm Spodoptera latifascia (n = 25). Linear re-
gression lines and slopes (m) are presented for females only; males were
similar. Analysis of covariance indicated significant differences between
some of the slopes (from Angelo and Slansky in prep.).
FEMALES
CEW m=.25
FAW m=.57
50 100
DRY WEIGHT (mg)
Fig. 4. Relationship between % lipid (dry weight) and dry body weight
for four species of noctuid moths (see Figure 3 for explanation of abbrevi-
ations). Linear regression lines and slopes (m) are presented for females
only; males were similar. VBC (n = 31), FAW (n = 28), CEW (n = 18)
and SPLAT (n = 25). Analysis of covariance indicated significant differ-
ences between some of the slopes (from Angelo and Slansky in prep.).
I I r I I
Florida Entomologist 67(1)
CEW (and SPLAT) thus exhibit less reduction in lipid when starved as
larvae compared with VBC and FAW. Perhaps it is important for larvae
of the CEW to maintain a relatively high lipid content (i.e., a relatively high
metabolic fuel reserve) throughout their last instar in the event an over-
wintering pupal diapause is required. For VBC and FAW, however, which
apparently lack an overwintering diapause, to be able to produce adults
with reduced wing loading ratios, facilitating less costly flight in search
of more favorable habitats, seems to take precedence over maintaining their
lipid content. When starved, individuals of SPLAT substantially reduce
their wing loading ratio and maintain a relatively high lipid content, but
how this relates to their biology in the field is unknown.
In addition, the maintenance of a relatively high lipid content by CEW
may provide it with the metabolic fuel source for a more highly powered
migratory flight. The relatively high wing loading ratio and steeper slope
for the wing area to body weight relationship for CEW compared with the
other three species (Figure 3) could indicate that CEW does undergo a
more highly powered flight. Although more energetically costly per unit
time, perhaps the high wing loading ratio of a high-powered flier would
allow greater flight speed with the cost of transport (energy cost per unit
distance travelled) actually being similar to that of a more passive flier
with a lower wing loading ratio. Because cost of transport is dependent on
both the metabolic cost and the speed of flight, it would seem possible that
a migratory insect could reduce its cost of transport by reducing the former
or increasing the latter. Use of upper air currents by many migratory
insects (see above) may increase their speed of flight with little or no
increase in metabolic cost, thus reducing their cost of transport.
Much more research is needed to critically evaluate these hypotheses.
Laboratory studies indicate apparent differences in flight behavior among
these noctuid moths, with VBC and CEW exhibiting greater degrees of
nocturnal activity than FAW (Leppla et al. 1979); adults of SPLAT also
seem relatively sedentary in the laboratory (N. C. Leppla, personal comm.).
However, the extent to which these species differ in their flight behavior,
both when searching within a habitat for adult food, mates and larval food-
plants and when migrating between habitats remains to be determined. The
similarity in body weight between the small-sized adult VBC produced in
our experiments by larval starvation and the smallest field caught specimens
(Angelo and Slansky in prep.) suggests that we are dealing with adults
within the natural size range in our experiments and that the small adults
are at least flight-worthy. However, whether these small-sized adults are
reproductively competent, and whether they exhibit migratory flight are
interesting questions that remain to be answered. In addition, we lack de-
tailed information on how flight behavior, metabolic costs of flight, longevity
and reproduction vary as a function of body size, and on the impact of
flight behavior on fitness.
CONCLUSIONS
The way in which the individuals of a species allocate their resources
is intimately associated with the evolved lifestyle of the species. Some
of the more evident differences in resource allocation are manifested in the
variety of sizes, shapes and colors of different species of insects. Differences
in the relative abundance of different species also are often very obvious
March, 1984
Insect Behavioral Ecology-'83 Angelo and Slansky, Jr. 35
(e.g. the extremely numerous larvae of a particular moth species defoliating
much of a crop versus that "rare" species still absent from a devoted lepi-
dopterist's collection) and may reflect interspecific differences in resource
allocation to production of offspring and defense from enemies.
Less obvious but nonetheless important differences in resource alloca-
tion appear in the different body compositions, proportions of body parts,
and behaviors of species. As research progresses, we are finding that
insects differ in how they respond, through changes in resource allocation,
to changes in environmental conditions. Two significant features of the en-
vironment that frequently exhibit changes are the quality and quantity of
food. Obtaining a sufficient quantity of adequate quality food seems to be
a common dilemma among animals (White 1978), and for insects flight is a
common means of dispersal in search of food. Some species exhibit dramatic
shifts between wingless and winged forms depending on the environment;
individuals of other species are always winged, but they may exhibit altera-
tions in the relationship between body weight and wing area (i.e., wing
loading ratio), among other changes.
Of the four species of noctuid moths that we studied, all altered the
allocation of their resources under larval starvation to produce adults with
relatively larger wings and low wing loading ratios, even lower than the
presumed theoretical prediction (i.e., the slopes of the relationship between
wing area and body weight were all less than 0.67). This response seems
to have adaptive value, associated with the apparent "passive" mode of long
distance dispersal of these species, whereby individuals are blown along by
upper air currents while maintaining active flight.
In contrast to the above species, some animals with a more energy-
demanding flight, such as hummingbirds and euglossine bees, exhibit
relatively small wings and high wing loading ratios, even at low body
weights. This is indicated by values for the slope of the relationship between
wing area and body weight that are greater than 0.67 (the slope for
hummingbirds is 0.75 (Greenewalt 1962) and for euglossine bees is about
one (Casey et al. 1984)). For animals like these with high-powered flight
(i.e., relatively high wing beat frequency), a strategy of producing relatively
larger wings at smaller body weights (i.e., reducing wing loading ratio like
the noctuid moths studied here), would undoubtedly increase, rather than
decrease, the energy cost of their style of flight due in part to the increase
in drag over the surface of the wings as they became larger (see Kammer
and Heinrich 1978, Casey 1981a).
Support for this hypothesis of increased energy requirements associated
with larger wings in species with a high-energy-demanding mode of flight
comes from data for two species of sphinx moths, Manduca sexta and
Hyles lineata. Individuals of these species exhibit a very energy-demanding
mode of flight, both because they have relatively high wing loading ratios
(ranging from approximately 5 to 80 times those found for the noctuid
moths in our study) and because they exhibit hovering flight (Casey 1976).
The slope of the relationship between log wing area and log body weight
for these two species is negative, indicating that individuals lighter in
weight have larger wings (on an absolute scale) than heavier individuals.
Associated with this is a greater weight-specific power requirement for
the moths that are lighter in weight (Casey 1976). Why these species pro-
duce larger (rather than smaller) wings at reduced body sizes is not clear.
Florida Entomologist 67(1)
Before the adaptive significance of the various alterations in resource
allocation seen among species in response to environmental changes can
be understood, more information on the consequences of such alterations is
required. Thus, questions pertaining to the significance of interspecific
differences in ability to achieve optimal weights, in minimal weight values,
in changes in wing loading ratio and so forth must be answered within
the context of the species' lifestyle in nature, in terms of trade-offs be-
tween reproduction, dispersal and mortality. This is a difficult task requiring
detailed behavioral observations and quantitative measurements; it is
clear that considerable research in nutritional ecology remains to be done.
In addition to the goal of understanding the evolution of species' life-
styles, there is a pragmatic side to research in nutritional ecology. If the
factors influencing the behavior (including resource allocation) of insect
pests, both in terms of their relative impact and their mechanisms of action
are understood, then better methods of altering the behavior of pest species
to their detriment may be devised (see Barfield and Stimac 1980). For
example, many insect pests apparently exhibit long distance migration, as
well as dispersal between and within crops (Stinner et al. 1983). Field-
level control measures often overlook this fact; successful pest management
must include a broader view of pest behavior, in this case attempting to
identify "source areas" and influence the pest in these before it moves into
the crop fields (see Barfield and O'Neil, this Symposium). Thus it can be
seen that research on resource allocation is a valuable approach to under-
standing flight behavior and other features of insect lifestyles that have a
significant bearing on crop production.
ACKNOWLEDGEMENTS
We thank Jim Lloyd for inviting us to participate in this Symposium
and for his thorough editorial comments, Jack Rye, USDA Insect Attractants
Lab., for supplying artificial diet and livestock of three noctuid moth species,
Reed Pedlow for graphical assistance, Greg Piepel for statistical advice, and
Bob Haack for bringing to our attention the important paper by Andersen
and Nilssen. We are grateful to Carl Barfield, Tim Casey, Norm Leppla and
Tom Walker for their comments on an early draft of this paper.
LITERATURE CITED
AIDLEY, D. J., AND M. LUBEGA. 1979. Variation in wing length of the
African armyworm, Spodoptera exempt in East Africa during 1973-
74. J. Appl. Ecol. 16: 653-62.
ANDERSEN, J., AND A. C. NILSSEN. 1983. Intrapopulation size variation of
free-living and tree-boring Coleoptera. Canadian Ent. 115: 1453-64.
ATKINS, M. D. 1966. Laboratory studies on the behavior of the Douglas-fir
beetle, Dendroctonus pseudotsugae Hopk. Canadian Ent. 98: 953-91.
- AND S. H. FARRIS. 1962. A contribution to the knowledge of flight
muscle changes in the Scolytidae (Coleoptera). Canadian Ent. 94:
25-32.
BAKKER, K. 1959. Feeding period, growth, and pupation in larvae of
Drosophila melanogaster. Ent. Exp. and Appl. 2: 171-86.
BARFIELD, C. S., AND J. L. STIMAC. 1980. Pest management: an entomologi-
cal perspective. BioScience 30: 683-9.
BEADLE, G. W., E. L. TATUM, AND C. W. CLANCY. 1938. Food level in relation
March, 1984
Insect Behavioral Ecology-'83 Angelo and Slansky, Jr. 37
to rate of development and eye pigmentation in Drosophila melano-
gaster. Biol. Bull. 75: 447-62.
BECKWITH, R. C. 1970. Influence of host on larval survival and adult
fecundity of Choristoneura conflictana (Lepidoptera: Tortricidae).
Canadian Ent. 102: 1474-80.
BLAKLEY, N., AND S. R. GOODNER. 1978. Size-dependent timing of meta-
morphosis in milkweed bugs (Oncopeltus) and its life history implica-
tions. Biol. Bull. 155: 499-510.
BLUEWEISS, L., H. Fox, V. KUDZMA, D. NAKASHIMA, R. PETERS, AND S. SAMS.
1978. Relationships between body size and some life history pa-
rameters. Oecologia 37: 257-72.
BRATTSTEN, L. 1979. Biochemical defense mechanisms in herbivores against
plant allelochemicals. Pages 199-270 In G. A. Rosenthal and D. H.
Janzen, Eds. Herbivores: Their interaction with secondary plant
metabolites. Academic Press, New York.
BUSCHMAN, L. L., W. H. WHITCOMB, T. M. NEAL, AND D. L. MAYS. 1977.
Winter survival and hosts of the velvetbean caterpillar in Florida.
Florida Ent. 60: 267-73.
CALOW, P. 1977. Ecology, evolution and energetic: a study in metabolic
adaptation. Adv. Ecol. Res. 10: 1-62.
CASEY, T. M. 1976. Flight energetic of sphinx moths: power input
during hovering flight. J. Exp. Biol. 64: 529-43.
1981a. Insect flight energetic. Pages 419-52 In C. F. Herreid II
and C. R. Fourtner, Eds. Locomotion and energetic in arthropods.
Plenum Publishing Corp., New York.
1981b. A comparison of mechanical and energetic estimates of flight
cost for hovering sphinx moths. J. Exp. Biol. 91: 117-29.
AND B. A. Joos. 1983. Morphometrics, conductance, thoracic
temperature and flight energetic of noctuid and geometrid moths.
Physiol. Zool. 56: 160-73.
---, M. L. MAY, AND K. R. MORGAN. 1984. Flight energetic of euglos-
sine bees in relation to morphology and wing stroke frequency. J.
Exp. Biol. (in press).
DANTHANARAYANA, W. 1976. Environmentally cued size variation in the
light-brown apple moth, Epiphyus postvittana (Walk.) (Tortricidae),
and its adaptive value in dispersal. Oecologia 26: 121-32.
DERR, J. A., B. ALDEN, AND H. DINGLE. 1981. Insect life histories in relation
to migration, body size, and host plant array: a comparative study of
Dysdercus. J. Animal Ecol. 50: 181-93.
DEWILDE, J., AND A. DELOOF. 1973. Reproduction-endocrine control.
Pages 97-157 In M. Rockstein, Ed. The physiology of Insecta. Vol. 1,
2nd ed. Academic Press, New York.
DINGLE, H. 1968. The influence of environment and heredity on flight
activity in the milkweed bug Oncopeltus. J. Exp. Biol. 48: 175-84.
-- (Ed.). 1978. Evolution of insect migration and diapause. Springer-
Verlag, New York.
1980. Ecology and evolution of migration. Pages 1-101 In S. A.
Gauthreaux, Jr., Ed. Animal migration, orientation, and navigation.
Academic Press, New York.
---, N. R. BLAKLEY, AND E. R. MILLER. 1980. Variation in body size
and flight performance in milkweed bugs (Oncopeltus). Evolution 34:
371-85.
DIXON, A. F. G. 1972. Fecundity of brachypterous and macropterous alatae
in Drepanosiphon dixoni (Callaphididae, Aphididae). Ent. Exp. and
Appl. 15: 335-40.
.1973. Biology of aphids. Edward Arnold, London.
Florida Entomologist 67(1)
DUELLI, P. 1980. Adaptive dispersal and appetitive flight in the green lace-
bug, Chrysopa carnea. Ecol. Ent. 5: 213-20.
EBERHARD, W. G. 1982. Beetle horn dimorphism: making the best of a bad
lot. American Nat. 119: 420-6.
ELSEY, K. D. 1974. Jalysus spinosus: effect of age, starvation, host plant,
and photoperiod on flight activity. Environ. Ent. 3: 653-5.
ENDERS, F. 1975. The influence of hunting manner on prey size, particularly
in spiders with long attack distances (Araneidae, Linyphidae and
Salticidae). American Nat. 109: 737-63.
1976. Size, food-finding, and Dyar's constant. Environ. Ent. 5: 1-10.
FRIEND, W. G., C. T. H. CHOY, AND E. CARTWRIGHT. 1965. The effect of
nutrient intake on the development and the egg production of Rhodnius
prolixus Stahl (Hemiptera: Reduviidae). Canadian J. Zool. 43: 891-
904.
GIBO, D. L. 1981. Altitudes attained by migrating monarch butterflies,
Danaus p. plexippus (Lepidoptera: Danaidae), as reported by glider
pilots. Canadian J. Zool. 59: 571-2.
GORDON, H. T. 1972. Interpretations of insect quantitative nutrition. Pages
73-105 In J. G. Rodriguez, Ed. Insect and mite nutrition: significance
and implications in ecology and pest management. Elsevier-North
Holland, Amsterdam.
GREENE, G. L. 1979. Evidence for migration of the velvetbean caterpillar.
Pages 406-8 In R. L. Rabb and G. G. Kennedy, Eds. Movement of
highly mobile insects: concepts and methodology in research. North
Carolina State Univ., Raleigh.
GREENEWALT, C. H. 1962. Dimensional relationships for flying animals.
Smithsonian Misc. Collect. 144(2): 1-46.
GREENSTONE, M. H., AND A. F. BENNETT. 1980. Foraging strategy and
metabolic rate in spiders. Ecology 61: 1255-9.
HARDWICK, D. F. 1965. The corn earworm complex. Mem. Ent. Soc. Canada
40: 1-247.
HEINRICH, B. 1981. Ecological and evolutionary perspectives. Pages 235-302
In B. Heinrich, Ed. Insect thermoregulation. Wiley, New York.
HERZOG, D. C., AND J. W. TODD. 1980. Sampling velvetbean caterpillar in
soybean. Pages 107-40 In M. Kogan and D. Herzog, Eds. Sampling
methods in soybean entomology. Springer-Verlag, New York.
HESPENHEIDE, H. A. 1973. Ecological inferences from morphological data.
Annu. Rev. Ecol. Syst. 4: 213-29.
HIGGINS, R. A. AND L. P. PEDIGO. 1979. Evaluation of guazatine triacetate as
an antifeedant/feeding deterrent for the green cloverworm on soy-
beans. J. Econ. Ent. 72: 680-6.
HINTON, H. E. 1981. Biology of insect eggs. Vol. 1. Pergammon Press,
Oxford.
HOLMES, E. A., E. M. PETERSON, AND C. A. PETTI. 1979. Physiological
characterization of the migratory milkweed bug, Oncopeltus fasciatus.
Proc. Indiana Acad. Sci. 88: 223-7.
HORIE, Y., AND T. INOKUCHI. 1978. Protein synthesis and uric acid excretion
in the absence of essential amino acids in the silkworm, Bombyx mori.
Insect Biochem. 8: 251-4.
HUXLEY, J. S. 1972. Problems of relative growth. 2nd ed. Dover Publ., Inc.,
New York.
JOHNSON, C. G. 1969. Migration and dispersal of insects by flight. Methuen,
London.
1973. Insect migration: aspects of its physiology. Pages 279-334 In
M. Rockstein, Ed. The physiology of Insecta. Vol. 3, 2nd ed. Academic
Press, New York.
March, 1984
Insect Behavioral Ecology-'83 Angelo and Slansky, Jr. 39
.1976. Lability of the flight system: a context for functional
adaptation. Pages 217-33 In H. C. Rainey, Ed. Insect flight. Symp. R.
Ent. Soc. London, Vol. 7.
KAMMER, A. E., AND B. HEINRICH. 1978. Insect flight metabolism. Adv.
Insect Physiol. 13: 133-228.
KOPEC, S. 1924. Studies on the influence of inanition on the development
and the duration of life in insects. Biol. Bull. 46: 1-21.
KOVITVADHI, K. 1969. Feeding preference of Prodenia species on sweet
potato varieties. Ph.D. disser., University of Florida, Gainesville.
LEES, A. D. 1955. Th physiology of diapause in arthropods. University
Press, Cambridge.
LEPPLA, N. C., E. W. HAMILTON, R. H. GUY, AND F. L. LEE. 1979. Circadian
rhythms of locomotion in six noctuid species. Ann. Ent. Soc. America
72: 209-15.
LINKER, H. M. 1980. An analysis of seasonal abundance and sampling pro-
cedures for the major defoliating Lepidoptera in peanuts and soy-
beans in north Florida. Ph.D. dissertation. Univ. Florida, Gainesville.
LONG, D. B. 1959. Observations on adult weight and wing area in Plusia
gamma L. and Pieris brassicae L. in relation to larval population
density. Ent. Exp. and Appl. 2: 241-8.
LUGINBILL, P. 1928. The fall armyworm. USDA Tech. Bull. No. 34, 92 pp.
MACKAY, P. A., AND W. G. WELLINGTON. 1975. A comparison of the repro-
ductive patterns of apterous and alate virginoparous Acyrthosiphon
pisum (Homoptera: Aphididae). Canadian Ent. 107: 1161-6.
AND R. J. LAMB. 1979. Migratory tendency in aging populations of
the pea aphid, Acyrthosiphon pisum. Oecologia 39: 301-8.
MATHAVAN, S., AND J. MUTHUKRISHNAN. 1976. Effect of ration levels and
restriction of feeding durations on food utilization in Danaus chrysip-
pus. Ent. Exp. and Appl. 19: 155-62.
MATTHEE, J. J. 1945. Biochemical differences between the solitary and
gregarious phases of locusts and noctuids. Bull. Ent. Res. 36: 343-71.
MATTSON, W. J., JR. 1980. Herbivory in relation to plant nitrogen content.
Annu. Rev. Ecol. Syst. 11: 119-61.
MITCHELL, E. R. 1979. Migration by Spodoptera exigua and S. frugiperda-
North American style. Pages 386-93 In R. L. Rabb and G. G. Kennedy,
Eds. Movement of highly mobile insects: concepts and methodology
in research. North Carolina State University, Raleigh.
MITTLER, T. E., AND O. R. W. SUTHERLAND. 1969. Dietary influences on aphid
polymorphism. Ent. Exp. and Appl. 12: 703-13.
MUKERJI, M. K., AND J. C. GUPPY. 1970. A quantitative study of food con-
sumption and growth in Pseudaletia unipuncta (Lepidoptera: Noctui-
dae). Canadian Ent. 102: 1179-88.
NAYAR, J. K., AND D. M. SAUERMAN, JR. 1969. Flight behavior and phase
polymorphism in the mosquito Aedes taeniorhynchus. Ent. Exp. and
Appl. 12: 365-75.
NIJHOUT, H. F. 1975. A threshold size for metamorphosis in the tobacco
hornworm, Manduca sexta (L.). Biol. Bull. 149: 214-25.
1979. Stretch-induced molting in Oncopeltus fasciatus. J. Insect
Physiol. 25: 277-81.
NORRIS, M. J. 1950. Reproduction in the African migratory locust (Locusta
migratoria migratorioides R. & F.) in relation to density and
phase. Anti-locust Bull. 6:1-10.
ONO, T., AND F. NAKASUJI. 1980. Comparison of flight activity and ovi-
position characteristics of the seasonal forms of a migratory skipper
butterfly, Parnara guttata guttata. Kontyu, Tokyo 48: 226-33.
PEARSON, D. L., AND S. L. STEMBERGER. 1980. Competition, body size and
Florida Entomologist 67(1)
the relative energy balance of adult tiger beetles (Coleoptera: Cicinde-
lidae). American Midl. Nat. 104: 373-7.
PHILLIPS, J. R. 1979. Migration of the bollworm, Heliothis zea (Boddie).
Pages 409-11 In R. L. Rabb and G. G. Kennedy, Eds. Movement of
highly mobile insects: concepts and methodology in research. North
Carolina State University, Raleigh.
RABB, R. L., AND G. G. KENNEDY (Eds.). 1979. Movement of highly mobile
insects: concepts and methodology in research. North Carolina State
University, Raleigh.
RACCAH, B., AND A. S. TAHORI. 1971. Wing dimorphism influencing re-
sistance or toxicity tests and food uptake in Myzus persicae. Ent. Exp.
and Appl. 14: 310-4.
RAINEY, R. C. 1978. The evolution and ecology of flight: the "oceano-
graphic" approach. Pages 33-50 In H. Dingle, Ed. Evolution of insect
migration and diapause. Springer-Verlag, New York.
RANKIN, M. A. 1978. Hormonal control of insect migratory behaviour.
Pages 5-32 In H. Dingle, Ed. Evolution of insect migration and
diapause. Springer-Verlag, New York.
RICKLEFS, R. E. 1982. A comment on the optimization of body size in
Drosophila according to Roff's life history model. American Nat.
120: 686-8.
ROCKSTEIN, M. (Ed.). 1978. Biochemistry of insects. Academic Press, New
York.
ROFF, D. 1977. Dispersal in dipterans: its costs and consequences. J.
Animal Ecol. 46: 443-56.
1981. On being the right size. American Nat. 118: 405-22.
ROSE, D. J. W. 1972. Dispersal and quality in populations of Cicadulina
species (Cicadellidae). J. Animal Ecol. 41: 589-609.
SANDERS, C. J., AND C. S. LUCUIK. 1975. Effects of photoperiod and size
on flight activity and oviposition in the eastern spruce budworm
(Lepidoptera: Tortricidae). Canadian Ent. 107: 1289-99.
SCHMIDT-NIELSON, K. 1970. Energy metabolism, body size, and problems
of scaling. Feder. Proc. 29: 1524-32.
SCRIBER, J. M., AND F.' SLANSKY, JR. 1981. The nutritional ecology of im-
mature insects. Annu. Rev. Ent. 26: 183-211.
SLANSKY, F., JR. 1980. Food consumption and reproduction as affected by
tethered flight in female milkweed bugs (Oncopeltus fasciatus). Ent.
Exp. and Appl. 28: 277-86.
1982a. Insect nutrition: an adaptationist's perspective. Florida Ent.
65: 45-71.
1982b. Toward a nutritional ecology of insects. Proc. 5th Int. Symp.
Insect-Plant Relationships, Wageningen. Pudoc, Wageningen.
AND P. FEENY. 1977. Stabilization of the rate of nitrogen accumula-
tion by larvae of the cabbage butterfly on wild and cultivated food
plants. Ecol. Monogr. 47: 209-28.
.AND J. M. SCRIBER. 1984. Food consumption and utilization.
Chapter 3, Vol. 4 In G. A. Kerkut and L. I. Gilbert, Eds. Compre-
hensive insect physiology, biochemistry and pharmacology. Pergamon
Press, Oxford. (In press).
SOLBRECK, C., AND I. PEHRSON. 1979. Relations between environment, migra-
tion and reproduction in a seed bug, Neacoryphus bicrusis (Say)
(Heteroptera: Lygaeidae). Oecologia 43: 51-62.
SOUTHWOOD, T. R. E. 1977. Habitat, the templet for ecological strategies. J.
Anim. Ecol. 46: 337-65.
STEARNS, S. C. 1982. The role of development in the evolution of life
histories. Pages 237-58 In J. T. Bonner, Ed. Evolution and develop-
March, 1984
Insect Behavioral Ecology-'83 Angelo and Slansky, Jr. 41
ment. Springer-Verlag, Berlin.
STINNER, R. E., C. S. BARFIELD, J. L. STIMAC, AND L. DOHSE. 1983. Dis-
persal and movement of insect pests. Annu. Rev. Ent. 28: 319-36.
SWEENEY, B. W., AND R. L. VANNOTE. 1978. Size variation and the distribu-
tion of hemimetabolous aquatic insects: two thermal equilibrium
hypotheses. Science 200: 444-6.
TANAKA, S. 1976. Wing polymorphism, egg production and adult longevity
in Pteronemobius taprobanesis Walker (Orthoptera, Gryllidae).
Kontyu, Tokyo 44: 327-33.
TOWNSEND, C. R., AND P. CALOW (Eds.). 1981. Physiological ecology.
Sinauer Associates, Sunderland, Massachusetts.
WALKER, T. J. 1980. Migrating Lepidoptera: are butterflies better than
moths? Florida Ent. 63: 79-98.
WASSERMAN, S. S., AND C. MITTER. 1978. The relationship of body size to
breadth of diet in some Lepidoptera. Ecol. Ent. 3: 155-60.
WELLINGTON, W. G. 1980. Dispersal and population change. Pages 11-24
In A. A. Berryman and L. Safranyik, Eds. Dispersal of forest insects:
evaluation, theory and management implications. Proc. 2nd IUFRO
Conf. Washington State Univ., Pullman.
WHITE, T. R. C. 1978. The importance of a relative shortage of food in
animal ecology. Oecologia 33: 71-86.
WIGGLESWORTH, V. B. 1965. The principles of insect physiology. 6th ed.
Methuen, London.
WILLIAMS, C. M. 1958. Hormonal regulation of insect metamorphosis.
Pages 794-806 In W. D. McElroy and B. Glass, Eds. A symposium
on the chemical basis of development. John Hopkins Press, Baltimore.
YAMADA, H., AND K. UMEYA. 1972. Seasonal changes in wing length and
fecundity of the diamond-back moth, Plutella xylostella (L.). Japan-
ese J. Ent. Zool. 16: 180-6.
Florida Entomologist 67(1)
IS AN ECOLOGICAL UNDERSTANDING A
PREREQUISITE FOR PEST MANAGEMENT?
CARL S. BARFIELD AND ROBERT J. O'NEIL*
SYNOPSIS
Comparison between the principles and real world practices of integrated
pest management (IPM) reveals severe discrepancies. The cotton agri-
cultural system is a problem with a long and sad history, and illustrates
what not to do. But, examination of IPM relative to the mobile pests of
polycultures in the southeastern USA would suggest that "here we go
again?"
Integrated pest management (IPM) is the current paradigm for dealing
with pests; thus, it is prudent to inquire whether IPM will have a higher
probability of long term success than other approaches which have failed.
This paper is an attempt to compare the principles of IPM to the current
control practices and to examine management approaches in two cropping
systems. Primary objectives are to show that agriculturists are far from
implementing programs that will solve pest problems and that such solutions
will arise only from an ecologically sound foundation.
PRINCIPLES OF IPM
The concept of integrated pest management (IPM) is well documented
(e.g., Rabb and Gutherie 1970, Metcalf and Luckmann 1975, Apple and
Smith 1976, Smith et al. 1976, Smith and Pimentel 1978, Bottrell 1979, Bar-
field and Stimac 1980); however, Bottrell (1979) appears to be the first to
state explicitly the principles underlying IPM.
1. POTENTIALLY HARMFUL SPECIES WILL CONTINUE TO
EXIST AT TOLERABLE LEVELS OF ABUNDANCE. The objec-
tive of IPM is to lower pest populations below economically important
levels; eradication is not the objective.
2. THE ECOSYSTEM IS THE MANAGEMENT UNIT. The bound-
aries of and the couplings among components of the system must be
identified before design and implementation of an IPM program.
3. THE USE OF NATURAL ENEMIES IS MAXIMIZED. An under-
standing of how natural enemies work in the system must be acquired
so that optimal use can be made of their impact on target pest
populations.
4. ANY CONTROL PROCEDURE MAY PRODUCE UNEXPECTED
AND UNDESIRABLE CONSEQUENCES. An ecologically based
management strategy is less likely to result in "negative effects"
within the system being managed.
*Carl S. Barfield is an Associate Professor in the Department of Entomology and Nema-
tology, University of Florida. He teaches a course entitled UNDERSTANDING AND
IMPLEMENTING PEST MANAGEMENT STRATEGIES IN AGRICULTURAL SYSTEMS
where a strong emphasis is placed on both the theory and practice of IPM. His research has
concentrated on biology & dynamics of mobile noctuid moths. Robert J. O'Neil is a Univ.
of Florida doctoral candidate and is studying the impact of a complex of predaceous organ-
isms on the dynamics of a noctuid pest of soybeans. Addresses for both: 3103 McCarty Hall,
Univ. of Florida Gainesville, FL 32611. Fla. Ag. Exp. Sta. J. Ser. No. 5484.
March, 1984
Insect Behavioral Ecology-'83 Barfield and O'Neil
5. AN INTERDISCIPLINARY SYSTEMS APPROACH IS ES-
SENTIAL. The assumption is that information collected by various
scientists can and will be integrated.
Implicit in Bottrell's principles is the concept of monitoring. Design or
evaluation of an IPM program demands monitoring of relevant aspects of
the system. Thus, any IPM program must have well defined and utilized
monitoring schemes.
An examination of currently used IPM programs for a variety of crops
and pests (see Bottrell 1979, Barfield and Stimac 1980, Huffaker 1980,
Flint and van den Bosch 1981 for general reviews) reveals that instead of
IPM programs having the above six characteristics, most existing programs
have the following:
1. There is virtually no appreciation for the boundaries and character-
istics of the system being managed. Target pests are dealt with as if
sessile, and individual fields as though they are independent of the
agroecosystem.
2. Mortality from natural enemies is poorly understood, even totally
ignored.
3. Most "IPM programs" are not integrated, but are actually four uni-
lateral efforts-one each for weeds, insects, pathogens, and nematodes.
Farmers must do the integrating, if any is to occur. Potentially useful
integrative tools (e.g., systems models) have not been incorporated
into the mainstream of agricultural thinking.
4. The level at which a pest population is considered to be economically
important is usually considered static, not a function of changes in
the system being managed.
5. Monitoring is not often a part of field activities. Sampling plans are
often inadequate and do not allow estimation of pest population levels
with precision or accuracy.
While some programs have been deemed successful, others have not (see
Barfield and Stimac 1980). Assessing why a program fails is often difficult.
To evaluate the applicability of IPM as an approach, we must first delineate
"true" IPM programs from the plethora of programs that are called IPM.
Determining whether IPM will lead to more solutions to pest problems than
other approaches requires a historical perspective. For illustration, two
approaches to boll weevil (Anthonomus grandis) management are compared,
with respect to their spatial and temporal utility. The amount of ecological
information incorporated into particular programs is the primary focus.
The thesis is that ecologically based programs will be effective, but pro-
grams that ignore significant components of pest ecology will not. Following
the boll weevil example, a description of the complex of noctuids typical of
polycultural systems in the southeastern USA is given. This second system
forces the question "are the same mistakes still occurring?"
THE COTTON WEEVIL: A MODEL FOR IPM
After entering the United States (ca. 1892) and spreading throughout
most of the range of its cotton host plant, the boll weevil caused radical
changes in the way cotton was cultivated (Adkisson and Bottrell 1977,
Bottrell 1983). The numerous approaches that were attempted to manage
the weevil mirrored developments in Economic Entomology in the 20th
Florida, Entomologist 67(1)
century (see Perkins 1982). To a varying degree, weevil management relied
on a knowledge of pest biology and ecology. Two management approaches
can be contrasted for their long term utility in space and time (robustneess).
PRE-INSECTICIDE ERA. Since little was known about the boll weevil
prior to its introduction, early workers had to investigate weevil ecology
before they could develop even a preliminary management program. Signifi-
cant constituents of early management programs were strategies that maxi-
mized within- and between-season mortality (Adkisson and Bottrell 1977,
Bottrell 1983). Within seasons mortality was identified primarily to be a
function of two components: (1) natural enemies and (2) host plant re-
sponses to weevil infestation (Hunter and Hinds 1904; 1905, Pierce et al.
1912, Fenton and Dunnam 1929). Natural enemies were studied extensively
(see Pierce 1908, Pierce et al. 1912). Of particular interest was the inter-
action between host plant and insect parasitoids. Since weevil immatures
(eggs to pupae) developed inside the cotton floral buds (weevils also attack
fruits, but prefer buds-called "squares"), parasitoids had to search buds
for suitable hosts. In response to weevil attack, the plant abscised infested
buds. Most buds fell to the soil surface, but some remained on the plant.
These were referred to as "hanging squares". It had long been noted that
parasitism rates in hanging squares consistently were higher than in fallen
squares (Hunter and Hinds 1904; 1905, Pierce et al. 1912, Fenton and
Dunnam 1929). To maximize parasite efficacy, Pierce et al. (1912) suggested
that farmers refrain from destroying hanging squares and that a "hang-
ing square" variety of cotton should be developed and used in production.
Abscission of infested squares also played an important role in weevil
mortality. Squares that fell between rows were sunlit and dried more
rapidly than those that fell in plant shade. Immature weevil mortality was
found to be affected significantly by the location and subsequent drying
time of fallen squares (Hunter and Hinds 1904; 1905, Fenton and Dunnam
1929, Folsom 1932). To take advantage of this source of weevil mortality,
some authors suggested varying cotton row spacing (Mally 1901, Cook 1932).
Perhaps the most significant component of weevil ecology was identified
to be the survivorship of overwintering weevils (Hunter and Hinds 1904;
1905, Pierce et al. 1912). Weevil adults overwinter in and around cotton
fields under leaf litter and crop residue. To maximize adult mortality, two
suggestions were made: (1) shorten the growing season to increase the
time in overwintering sites and decrease the time suitable host material
was available, and (2) destroy crop residue and other overwintering habitats
(see Cook 1932).
Growers following these ecologically-based recommendations were able
to produce an economically viable cotton crop for over 40 years (Adkisson
and Bottrell 1977, Bottrell 1983). Although insecticides were available
(e.g., calcium arsenate), they were used sparingly (Isley 1926, Folsom
1932). The advent of inexpensive, effective synthetic insecticides (e.g., DDT)
led to major changes in boll weevil management-changes that ultimately
led to disaster.
INSECTICIDE ERA. Following World War II, the incorporation of syn-
thetic insecticides into cotton crop protection schemes led to a dramatic
change in boll weevil management. The decimation of weevil populations
following insecticide application allowed growers to maximize yields while
minimizing damage (Newsom 1970, Reynolds et al. 1975). Preventative ap-
March, 1984
Insect Behavioral Ecology-'83 Barfield and O'Neil
plications and "calendar sprays" (i.e., "ever so often, need it or not") were
used widely, eliminating the "need" to determine whether weevil densities
were above economically damaging levels (Adkisson and Bottrell 1977). With
the widespread acceptance of a management plan based solely on insecti-
cides, research on weevil ecology was de-emphasized. Eventually, this over-
reliance on insecticides led to widespread ecological perturbations and near
economic collapse of the cotton agricultural system (Adkisson and Bottrell
1977).
Weevil resistance to organochlorine insecticides in the 1950's was
followed quickly by analogous resistance to other compounds by both the
weevil and other cotton pests (Newsom 1970, Adkisson and Bottrell 1979),
and growers soon found that increased application rates could not provide
needed control. Organisms not formerly pests became pests, and existing
pests got worse. Cotton agriculture was out of control (Newsom 1970, Ad-
kisson and Bottrell 1977), and several major cotton growing regions
faced economic ruin.
What emerged from this disaster was an approach called IPM. IPM
largely adopted the recommendations made by workers in the pre-insecticide
era. Focus was again on maximizing overwintering weevil mortality, shorten-
ing the cotton growing season and judicious insecticide use (Bottrell 1983).
Weevil management was fortunate at least to have had an ecological
template. However, examination of how much remains to be learned about
boll weevil shows that agriculturists are far from implementing a com-
plete solution to the boll weevil problem.
Although early workers identified the importance of the plant's ab-
scission of infested squares, just how this affects weevil dynamics or whether
immature parasitoids also suffer from square drying mortality remains to
be learned. Given the apparent increase in parasitization in handing
squares, should parasitoids that prefer to search in this region of the
plant's environment be released? There are many such questions. In ad-
dition, major elements of weevil biology and ecology are still being
discovered. For example, despite intensive studies on adult overwintering,
it was 1959 before diapause in adult weevils was discovered (Brazzell and
Newsom 1959). By 1968, only four exotic natural enemies (see Clausen
1978) had been released against the weevil, a species which itself was intro-
duced. As late as 1975, major alternative host plants were being found (Cross
et al. 1975). In 1979, a closely related weevil (Anthonomus hunteri, and a
potential clue to natural enemies, was described from the boll weevil's
Central American aboriginal home (Burke and Cate 1979).
What has been shown here for the boll weevil is true of many other
pests. Now, examination of a second system will reveal whether today's
workers have profited from yesterday's experiences.
THE NOCTUIDAE
The southeastern USA contains a mosaic of agricultural production
systems. Within these, a complex of pest organisms exists that appears to
reinvade crops annually through migration and/or dispersal. Of particular
concern in recent times is a complex of moths (mostly Noctuidae) that is
suspected to overwinter in more southern latitudes and move northward
each spring and summer. A number of recent reviews and symposia have
addressed these insects and what is and is not known about them (e.g.,
Florida Entomologist 67(1)
March, 1984
Rabb and Kennedy 1979, Stinner et al. 1983). Others have reviewed and
discussed the phenomenon of movement in great detail (Johnson 1969, Baker
1978, Gauthreaux 1980), theoretical evolutionary problems and selection
models (Walker 1980), and movement of pests relative to the structure of
agricultural systems (Stinner et al. 1983, Rabb and Stinner 1978, Johnson
et al. 1975, Stimac and Barfield 1979). Inability to forecast when and where
these mobile moths will occur and the reasons why also have been con-
fronted (Barfield et al. 1980).
Current IPM strategies against these pests are similar to those for less
mobile organisms-an individual farmer's field is scouted and the pest
population treated (primarily chemically) when damaging density levels
are suspected. These fields are treated as "islands," and there is virtually
no consideration of the significance of the processes determining the timing
and rate of influx. Though most agriculturists recognize that pest popu-
lation levels are related to both the timing and magnitude of immigration,
few measure influx rates and evaluate quantitatively the consequences of
those influxes (see Rabb and Kennedy 1979). Research extended from the
soybean plant growth model (see Wilkerson et al. 1983) is an exception to
this generalization. Unless an understanding of the role movement plays in
moth dynamics and "pest status" is acquired, agriculturists can not design
robust management strategies against pests such as these (e.g., Stinner et al.
1983, Rabb and Stinner 1978, Barfield 1983, Barfield et al. 1980, Rabb and
Kennedy 1979).
"IPM" programs against these noctuids could, over the long term, prove
to be just as unstable as boll weevil management when it abandoned an
ecologically based approach. Since the general attitude of the agricultural
community currently is for "judicious use of pesticides," the polycultural
systems of the southeastern USA may not suffer the catastrophe seen in
cotton; however, that is not the point. The thesis is that what IPM needs
is robustness-it needs to be based on ecological understanding sufficient
to adapt to the dynamic nature of the system being managed and to work
in space and time. For a proper noctuid IPM program, the following must
be known (see also Stinner et al. 1983):
1. seasonal patterns of appearance and geographical distribution of both
immature and adult stages
2. overwintering (quiescence or continuous breeding) habitats and as-
sociated environments
3. methods for differentiating local from migrant populations
4. weather patterns in a fashion meaningful for interpreting moth dis-
placement trajectories and flight behaviors
5. physiological and behavioral attributes conducive to initiating, main-
taining, and terminating non-trivial flight
6. relationships between relative density estimators (e.g., light traps)
and absolute densities occurring in particular crops
These six investigative areas will yield information that is crucial for
understanding the role movement plays in the occurrence of moths in space
and time (see Stimac and Barfield 1979). In addition, there is need to be
able to evaluate whether specific influxes cause economic damage (see
Barfield et al. 1980). The agricultural community appears to be a long way
from a sound ecological understanding of these mobile noctuids (see Rabb
Insect Behavioral Ecology-'83 Barfield and O'Neil
and Kennedy 1979), hence a long way from implementing an IPM program
against them.
THE REAL MESSAGE
Long term solutions to pest problems must have sound and broad
ecological bases. Boll weevil history offers a dramatic example of what can
happen with a unilateral approach that ignores the ecology of the system.
Will the Noctuidae be a repeat? Not if ecological understanding of the
system is a prerequisite for the design of IPM programs. This, of course,
means a re-orientation of experimental emphases and academic education
and training for crop protection practitioners (see Barfield and Jones
1979, Barfield and Stimac 1980, Strayer et al. 1983).
LITERATURE CITED
ADKISSON, P. L., AND D. G. BOTTRELL. 1977. Cotton insect pest management.
Annu. Rev. Entomol. 22: 451-81.
APPLE, J. L., AND R. F. SMITH (eds.). 1976. Integrated pest management.
Plenum Press (New York). 200 p.
BAKER, R. R. 1978. The evolutionary ecology of animal migration. Holmes
& Meier Pub., Inc. (New York). 1012 p.
BARFIELD, C. S. 1983. The role of movement in the dynamics of highly
mobile organisms. Proc. 12th Annual Conf. of Illinois Dept. Energy &
Nat. Resour., Champaign, IL, 13-4 Sept. 1983. (in press).
BARFIELD, C. S., AND J. L. STIMAC. 1980. Pest Management: an entomologi-
cal perspective. Bioscience 30: 683-9.
BARFIELD, C. S., J. L. STIMAC, AND M. A. KELLER. 1980. State-of-the-art for
predicting damaging infestations of fall armyworm. Fla. Entomol.
63: 363-75.
BARFIELD, C. S. AND J. W. JONES. 1979. Research needs for modeling pest
management systems involving defoliators in agronomic crop systems.
V\a. Entomol. 2:' 9%-114.
BOTTRELL, D. G. 1979. Integrated pest management. Rpt. for Council
Environ. Qual., U. S. Gov't Print. Office No. 041-011-00049-1, Washing-
ton, D.C. 120 p.
BOTTRELL, D. G. 1983. The ecological basis for boll weevil (Anthonomus
grandis Boheman) management. Agric. Environ. (in press).
BRAZZEL, J. R., AND L. D. NEWSOM. 1959. Diapause in Anthonomus grandis
Boh. J. Econ. Entomol. 52: 603-11.
BURKE, H. R., AND J. R. CATE. 1979. A new species of Mexican Anthonomus
related to the boll weevil (Coleoptera: Curculionidae). Ann. Entomol.
Soc. Am. 72: 189-92.
CLAUSEN, E. P. (ed.). 1978. Introduced parasites and predators of arthro-
pod pests: a world review. USDA/ARS Handbook 480. 545 p.
COOK, O. J. 1932. Common errors in cotton production. USDA Farmer's
Bull. 1686. 24 p.
CROSS, W. H., M. J. LUKERFAHR, P. A. FRYXELL, AND H. R. BURKE. 1975.
Host plants of the boll weevil. Environ. Entomol. 4: 19-26.
FENTON, F. A., AND E. W. DUNNAM. 1929. Biology of the cotton boll weevil
at Florence, S.C. USDA Tech. Bull. 112. 76 p.
FLINT, M. L., AND R. VAN DEN BOSCH. 1981. Introduction to pest manage-
ment. Plenum Press (New York). 240 p.
FOLSOM, J. W. 1932. Insect enemies of the cotton plant. USDA Farmer's
Bull. 1688. 28 p.
Florida Entomologist 67(1)
GAUTHREAUX, S. A., JR. (ed.). 1980. Animal migration, orientation, and
navigation. Academic Press (New York). 387 p.
HAFFAKER, C. B. (ed.). 1980. New technology of pest control. John Wiley &
Sons (New York). 500 p.
HUNTER, W. D., AND W. E. HINDS. 1904. The mexican cotton boll weevil.
USDA Div. Entomol. Bull. 45. 116 p.
HUNTER, W. D., AND W. E. HINDS. 1905. The mexican cotton boll weevil.
USDA Bur. Ent. Bull. 51. 81 p.
ISLEY, D. 1926. Early summer dispersion of boll weevil with special refer-
ence to dusting. Ark. Agr. Exp. Stn. Bull. 204. 17 p.
JOHNSON, C. G. 1969. Migration and dispersal of insects by flight. Methuen
& Co. Itd. (London). 763 p.
JOHNSON, M. W., R. E. STINNER, AND R. L. RABB. 1975. Ovipositional re-
sponse of Heliothis zea (Boddie) to its major hosts in North Carolina.
Environ. Entomol. 4: 291-7.
MALLY, F. W. 1901. The mexican cotton boll weevil. USDA Farmer's Bull.
130. 29 p.
METCALF, R. L., AND W. H. LUCKMANN (eds). 1975. Introduction to insect
pest management. John Wiley & Sons (New York). 587 p.
NEWSOM, L. D. 1970. The end of an era and future prospects for insect
control. Proc. Tall Timbers Conf. Ecol. Anim. Control Habitat Mgmt.
2: 117-36.
PERKINS, J. H. 1982. Insects, experts and the insecticide crisis: the quest
for new pest management strategies. Plenum Press (New York).
304 p.
PIERCE, W. D. 1908. Studies of parasites of the boll weevil. USDA Bur.
Ent. Bull. 73. 63 p.
PIERCE, W. D., R. A. CUSHMAN, C. E. HINDS, AND W. D. HUNTER. 1912. The
insect enemies of the cotton boll weevil. USDA Bur. Ent. Bull. 100.
99 p.
RABB, R. L., AND F. E. GUTHRIE (eds). 1970. Concepts of pest management.
Proc. of Conf., N.C. State Univ., Raleigh, N.C., 25-7 March, 1970.
N.C. State Univ. Press. 242 p.
RABB, R. L., AND G. G.. KENNEDY (eds). 1979. Movement of highly mobile
insects: concepts and methodology in research. Proc. Conf., Raleigh,
N.C., 9-11 Apr. 1979. N.C. State Univ. Press. 456 p.
RABB, R. L., AND R. E. STINNER. 1978. The role of insect dispersal and
migration in populations processes, pp. 3-16. IN Radar, insect popu-
lation ecology, and pest management. [C. R. Vaughn, W. Wolf, and W.
Klassen, eds.] NASA Conf. Pub. 2070. 249 p.
REYNOLDS, H. T., P. L. ADKISSON, AND R. F. SMITH. 1975. Cotton insect
pest management, pp. 379-443. IN Introd. to insect pest management
[R. L. Metcalf and W. H. Luckmann, eds.] John Wiley & Sons (New
York). 58 p.
SMITH, R. F., J. L. APPLE, AND D. G. BOTTRELL. 1976. The origins of inte-
grated pest management concepts for agricultural crops, pp. 1-16.
IN Integrated pest management [J. L. Apple and R. F. Smith, eds.].
Plenum Press (New York). 200 p.
SMITH, E. H., AND D. PIMENTEL (eds.). 1978. Pest control strategies.
Academic Press (New York). 334 p.
STIMAC, J. L., AND C. S. BARFIELD. 1979. Systems approach to pest manage-
ment in soybeans, pp. 249-59. IN Proc. World Soybean Conf. II,
Raleigh, N.C. [F. T. Corbin, ed.]. Westview Press (Boulder, CO).
897 p.
STINNER, R. E., C. S. BARFIELD, J. L. STIMAC, AND L. DOHSE. 1983. Dispersal
and movement of insect pests. Annu. Rev. Entomol. 28: 319-35.
STRAYER, J. R., C. S. BARFIELD, AND R. WILKERSON. 1983. New directions in
March, 1984
Insect Behavioral Ecology-'83 Barfield and O'Neil 49
post-baccalaureate training for pest management and plant pro-
tection. NACTA J., June, 1983, pp. 8-13.
WALKER, T. J. 1980. Migrating Lepidoptera: are butterflies better than
moths? Fla. Entomol. 63: 79-98.
WILKERSON, G. G., J. W. JONES, K. J. BOOTE, K. T. INGRAM, AND J. W.
MISHOE. 1983. Modeling soybean growth for crop management. Trans.
ASAE 26: 63-73.
Florida Entomologist 67(1)
March, 1984
OBSERVATIONAL SAMPLING METHODS FOR
INSECT BEHAVIORAL ECOLOGY
JEANNE ALTMANN*
SYNOPSIS
Researchers who want to retain relevant behavioral and ecological
settings and at the same time conduct scientifically rigorous studies have
increasingly turned to non-experimental research design. By understanding
the basis for wise data collecting decisions, and their consequences, the ob-
server can choose appropriate methods. Using illustrations from insect be-
havior, this paper discusses some of the basic principles involved and two
of the most useful ways of collecting behavioral data.
INTRODUCTION
One of the most important needs in observational research is for sensi-
tive, non-destructive methods of studying social processes (Barker 1963). A
major way this need can be met is through use of what Schneirla (1950)
called observation-selective (as opposed to manipulative) controls, including
systematic behavioral sampling methods. Many experienced students of
naturalistic behavior will be familiar with a number of well-established
methods for observing and sampling the responses that they want to study
systematically. What may be unfamiliar is the implicit rationale behind these
observational sampling methods and the fact that for a particular research
question, only particular sampling methods can provide the answers. I shall
discuss below several behavioral sampling methods and their uses, singly
and in combination. First, however, let us briefly consider those steps in
research design that should precede the choice of sampling methods.
As is the case for most tasks, different jobs call for different tools. So,
an unambiguous formulation of a behavioral question is critical to the
choice of an appropriate sampling technique. For behavioral problems
the differences among various questions often hinge on whether the dura-
tion of the behavior is significant, i.e. on whether the behavior can be
considered as a momentary event, a "happening" of negligible or inconse-
quential duration, or whether it should be considered as an enduring state.
In the latter case the onset and offset, or at least the duration of the be-
havior, is important, and, therefore should be sampled in an appropriate
way. For example, it may be that some insect obtains about the same amount
of energy from each flower it feeds from, so that we need only count fre-
quency of visits to flowers per unit time in order to estimate relative
energy intake. However, in the same insect, it may be that the expenditure
of energy is a function of time spent in transit between food plants and that
these are variable, in which case it is not the frequency of moves but rather
the duration or cumulative durations that we need to measure.
Consider another example: a male wasp's reproductive success might be
a function of the number of different mates he has, or it might depend on
*Jeanne Altmann is a Research Associate in the Department of Biology, University of
Chicago, Chicago, 60637. Her research focuses on non-experimental research design and
on the ontogeny, ecology, and evolution of family relationships. Her fieldwork is with
baboons in Kenya. Current address: Department of Biology, University of Chicago, 940 East
57th Street, Chicago, Illinois 60637.
Insect Behavioral Ecology-'83 Altmann
the length of time he spends guarding a female after mating. In the former
case we would need a sampling method that provides unbiased estimates of
rates or relative frequencies; in the latter, we need unbiased estimates of
duration or time spent. In another situation, reproductive success may be a
function of sperm count which, in turn, may be a function of the length of
time since the male's previous mating. In that case we would need a
continuous record that included the interval between occurrences of be-
haviors.
Many sampling methods are suitable for answering only frequency-based
questions, others, only for questions based on time spent in an activity,
and a few provide data for unbiased estimates of behavior durations or inter-
behavior durations. It is therefore important either to know ahead of time
which of these is the important parameter or, if this is not possible, to
choose one or more techniques that will allow the observer to gather both
kinds of information. Sometimes a pilot study will enable the researcher to
decide that only one of these kinds of information is needed, in which case
efficiency of data collection and analysis is often greatly enhanced.
In the first stage of formulating a research question about naturalistic
behavior, one usually uses informal, ad libitum observations to gain a
general understanding of the situation and to avoid totally meaningless and
inappropriate research directions. Burk & Calkins (1983) point out the
importance of this stage in the research on medflies. These early observations
may even be sufficiently informative to allow choice of the important pa-
rameters, such as frequency (including relative or conditional frequency),
interval duration, or time spent, as discussed above.
The background information that we have about our subjects may
enable us to make other important sampling decisions. If, for example, we
are interested in mating behavior and we know that it is somewhat variable,
we might rotate our observations among a number of (preferably identifi-
able) adults in order to evaluate individual differences or to at least be sure
that the data were from -a representative sample of adults. Likewise, if we
cannot rule out variability among nests in the behavior we measure, we
might decide to rotate among a sample of nests in a systematic way. De-
cisions on whether to sample during all times of the day, week, month, season,
will also need to be made, based on specificity of the problem, assumptions
about variability within and between classes, and the extent to which the
results will be generalized. All too often, such sampling decisions are
made without the investigator realizing that he or she has made a choice.
Of course, one does not thereby escape the consequences of the choice. To the
extent that decisions become explicit we shall improve the chances that they
are the most appropriate to our goals.
Much of traditional fieldwork never went beyond simple natural ob-
servations, or what I have called ad libitum sampling, and the findings were,
consequently, subject to the biases caused by attention-attracting, dramatic
events, by the individual biases of each observer, and so on. For reliable and
valid results, however, we must turn to more systematic techniques. Two of
the most useful and most commonly used ones will be discussed below. For
an analysis of others in the literature, the reader is referred to Altmann
(1974), which also contains a more in-depth treatment of some of the topics
mentioned only briefly here. More complete recent algebraic treatment of
Florida Entomologist 67(1)
some techniques can be found in Kraemer (1979) and in Griffin & Adams
(1983).
FOCAL-ANIMAL (CONTINUOUS) SAMPLING
I coined the term "focal-animal sampling" to refer to any sampling
method in which (i) all occurrences of specified behaviors or interactions
of an individual are recorded during each sample period, and (ii) a record
is made of the length of each sample period and the amount of time during
the sample period that the subject is actually adequately in view for record-
ing those behaviors. Under some conditions and at least for some behaviors,
one may reasonably assume that a complete record is obtained, not only of the
focal animal's actions, but also of behaviors directed to it by others. Under
other circumstances, it may be possible to record all acts by the focal in-
dividual, but not all those directed toward it by others (e.g. silent threats
performed at moderate distances). Several factors will affect the extent to
which the observers can reliably detect and record behaviors directed toward
the focal subject. These include the density of animals, the frequency and
attention-getting nature of the behaviors, and the number of behavior cate-
gories. It is critical that complete records of the target behaviors are ob-
tained during the observation period. Thus, it is important that the observer
not try to record more information than can be done consistently, even at
the busiest times, or else much of the advantage of systematic sampling will
be lost.
The importance of several temporal aspects of behavior has been in-
creasingly recognized. Fortunately, the development of field-portable,
electronic data recording devices, as well as good inexpensive digital time-
pieces, have greatly increased field workers' abilities to collect time-based
data and have eliminated the subsequent chore of hand-entering the data for
computer analysis. If the duration of activities or the duration of periods
between activities is important, these data can readily be included in a focal
animal sampling scheme. In the desire to sample more animals in any day,
observers will sometimes make each individual sample short, say five or ten
minutes. One should be aware that if sampling intervals are short relative
to the durations to be estimated, much poorer estimates will result. In
addition, one always loses some amount of time to searching and to sample
initiation between individuals; the ratio of this lost time to sample time may
be appreciable if short sample times are used. These factors will have to be
taken into account in making wise decisions about sample lengths. As always,
research design decisions involve compromises.
Focal sampling will be the method of choice, perhaps the only appropri-
ate method, for many problems. However, it is a very labor-intensive
method, and it is not suitable for most questions about behavioral synchrony
among individuals unless several observers are collecting data simultaneously
or if the synchrony questions involve only animals that are spatially very
close and thus simultaneously observable. Consequently, it is useful to con-
sider a complementary technique, instantaneous sampling.
POINT (INSTANTANEOUS) AND SCAN SAMPLING
Point sampling is a technique in which the observer records, at pre-
selected moments in time, an animal's ongoing activity. As such, it is a
March, 1984
Insect Behavioral Ecology-'83 Altmann
sample of behavioral states, not events. It is particularly useful in studies
of activity time budgets but not in answering questions based on frequency
of occurrence. Instantaneous sampling can be used to obtain data from a
large number of individuals, by observing each in turn. Moreover, if the
behavior of each group member is sampled successively within a very short
time period, the record approaches a simultaneous sample on all individuals,
which we refer to as scan sampling. If such sampling is done frequently,
data are obtained on the time distribution of behavioral states among a
whole group of animals. Data on behavioral synchrony are thereby obtained,
which are virtually impossible to obtain by almost any other sampling
technique.
In an ideal instantaneous sample, each individual's state is instantly
noted. In practice, however, the observation and classification of a state
takes time, particularly if one is moving or scanning from one individual to
another. The observer should try to scan each individual for the same brief
period of time, for otherwise a scan sample is equivalent to a series of
short focal-animal samples of variable and unknown durations-ones that
may be inadvertently biased by the different characteristics of the various
activities being recorded. In order to keep sampling time consistent and
brief, and to score reliably all items in one's catalog, the categories that
are recorded in this type of sampling should be easily and quickly dis-
tinguishable. For some animals this has meant that the technique is more
suited to studies of non-cocial rather than social behavior or to studies in
which social behaviors can be lumped into a few easily and rapidly identified
categories.
If animals are close together and little time is lost in sampling one
after another, scans offer an efficient means of gathering data on all group
members over a short time span. The absence of information on occurrence,
on durations, and on temporal patterning, will rule out scan sampling for
some questions. The observer also will need to be sensitive to the effects of the
order of individuals to observe, the frequency of the instantaneous samples
(which will affect their independence), and the distinguishability of the be-
haviors, as discussed above. As is the case for planning any research design,
there is no substitute for knowing one's animals and for having well-defined
questions.
USE OF COMBINATIONS OF SAMPLING METHODS
Sampling decisions almost always involve tradeoffs. Consequently, it is
particularly useful to consider the possibility of using combinations of
techniques to capitalize on the advantages of each. Often one can thereby
greatly increase the information gained in a study. I shall describe examples
of how this can be accomplished with ad libitum, instantaneous, and focal
sampling to give a sense of the possibilities that await the researcher's
creativity.
For many studies, focal-animal (continuous) sampling proves to be the
main technique of choice because it provides data for frequency-based
questions, for bout and inter-bout durations, time budgets, and so on. How-
ever, under most observational conditions, one is restricted to collecting data
on a single animal at a time. Moreover, it is sometimes difficult, even with
modern electronic event recorders, to collect very detailed data on a large
number of behavior types. As a consequence of this second limitation, one
Florida Entomologist 67(1)
March, 1984
might decide for some activities, to record only their occurrence, not their
duration, if the latter information were not critical. However, for some other
activity, say foraging, the more important information might be how much
time is spent in the activity, or whether foraging is ongoing when a second
activity, say predation, occurs. In this latter case, we would take instantane-
ous samples, perhaps at five-minute intervals, of the presence or absence
of foraging, and do so again when predation occurs. Then we can answer the
question of whether predation occurs disproportionately during foraging:
an affirmative answer to this question would suggest that there are costs as
well as benefits to foraging, and that the costs are not just the energetic
ones involved in obtaining food.
Questions about synchrony of activities in a group can be addressed by
taking scan samples of the group at the beginning and/or end of a focal
sample. Questions of subgroups or social affinities in a group can be ad-
dressed by instantaneous samples of nearest neighbors or all neighbors
within a specified distance at fixed intervals within a focal sample during
which details of social interaction are being recorded continuously. Then
one has the baseline data for asking whether the distribution of interaction
partners is predictable from the amount of time that individuals spend near
each other. Obtaining the latter from continuous records would have been
much more laborious to gather and such records are not needed to answer
this particular question.
While conducting focal samples, one is sometimes able to record some
additional observations on an ad libitum basis. While such data will not be
useful for any questions that require unbiased frequency estimates, they can
be useful for some other purposes. Their heuristic value has already been
mentioned. In addition, there are situations in which one is primarily con-
cerned with the asymmetry of some relationship within a pair of individuals,
such as who grooms whom, and for which one can reasonably assume that
the ad libitum observations are not biased toward any direction of this asym-
metry. The direction of food exchanges in insects may be of this sort, as
are agonistic/dominance interactions in some species. Thus, although inter-
action matrices constructed from ad libitum data cannot be used to answer
the question of who fights more often or to whom does an animal direct
most of its aggression (because those questions require data that are un-
biased in ways that are seldom the case with ad lib observations and that
require the focal sample data), they can answer certain questions: for
example, when animals A and B fight (or exchange food) with each other,
which member of the pair is more likely to be the recipient of the aggression
(or food)? Because much larger data sets are likely to accrue from such
a combination of ad libitum sampling and focal sampling, with the data from
each carefully separated, of course, many observers have found it a fruitful
combination.
Finally, the student of insect behavior may occasionally find that one
of the more specialized techniques is the best method for a particular
problem. In sequence sampling, for example, the focus of observation is a
particular interaction that can be identified even if the initial participants
discontinue the activity and others join. A primate play group can be of
this sort, as might insect nest-building, the fate of some food item that is
passed among and processed by many individuals in succession, and so on.
Another technique is one-zero sampling, in which the observer records
Insect Behavioral Ecology-'83 Altmann
the occurrence (or existence) of an activity within the sampling period, and
not the frequency, duration, or any other such information about the
activity. This sampling method was for a time one of the most commonly
used in some areas of behavioral research and as a consequence much at-
tention was devoted to attempts to relate these scores to measures of oc-
currence and time spent (e.g. Simpson & Simpson 1977, Rhine & Flanigon
1978). Although the technique is not a good one for estimates of those im-
portant parameters, or for any questions based on those parameters, just
the presence or absence of a behavior may itself be important in some
situations, in which case, one-zero sampling would be the appropriate method
to use. Thus, the main determinant of a potential foundresses' success may
be whether she mates at all (and not how often) within a fixed time after
she leaves her old nest. This is a one-zero situation. Or an animal's survival
may depend on whether it finds any food within a single day, the amount
being fairly irrelevant or constant. This, too, would be a one-zero biological
situation and would call for a one-zero sampling technique in which the
sampling unit of time should be the biologically-important unit, not the
arbitrary times units previously used for this technique. The point, as
always, is to choose the technique that is most appropriate to the question.
CONCLUSION
It is in field situations that biological problems usually arise and it is
to just such situations that we return to test our answers. This is often most
immediately clear in applied areas of science where the real-world tests are
eagerly awaited and where the successes and failures may be very much on
view. These applied areas, perhaps even more than others, require a thorough
blend of observational work, often in the field, and solid scientific method-
ology. They have much to gain from the increasing recognition that neither
laboratory nor field, experimental nor non-interventive methodologies, has
a monopoly on scientifically rigorous ways to answer questions (Schneirla
1950, Altmann 1974, Mertz & McCauley 1980; see Thornhill, this sym-
posium).
A primary function of any research design is to maximize the validity
of conclusions, that is, to minimize the number of alternative hypotheses
that are consistent with the data. Having done so, actions that are taken
based on these conclusions have a greater chance of being successful. Ex-
perimental methods and manipulative control of variables can sometimes
contribute to this goal, and are the approaches most commonly pursued to
this end. They are not the only means however, and often not the best
methods. Experimental research design, and the appropriate statistical
techniques for each design, have been the objects of extensive investigation.
While some of the results can be applied directly to non-experimental de-
sign, much of it cannot. The study of quasi-experimental and non-experi-
mental design (e.g. Webb et al. 1966) is increasingly recognized as an im-
portant one (see, e.g. Slater 1978, Lehner 1979) that now even receives
treatment in introductory texts, but is still in its infancy. At times this may
be frustrating to the researcher, but it is also challenging, and exciting as
one plays the game "Eliminate the Alternatives!"
56 Florida Entomologist 67(1) March, 1984
LITERATURE CITED
ALTMANN, J. 1974. Observational study of behavior: sampling methods. Be-
haviour 49: 227-67.
BARKER, R. G. 1963. The stream of behavior as an empirical problem. Pages
1-22 in R. G. Barker, ed. The stream of behavior. Appleton-Century-
Crofts, New York.
BURK, T., AND C. 0. CALKINS. 1982. Medfly mating behavior and control
strategies. Florida Ent. 66: 3-18.
GRIFFIN, B., AND R. ADAMS. 1983. A parametric model for estimating
prevalence, incidence, and mean bout duration from point sampling.
American J. Primat. 4: 261-71.
KRAEMER, H. C. 1979. One-zero sampling in the study of primate behavior.
Primates 202: 237-44.
LEHNER, P. N. 1979. Handbook of ethological methods. Garland STPM, New
York.
MERTZ, D. B., AND D. E. MCCAULEY. 1980. The domain of laboratory ecology.
Synthese 43: 95-110.
RHINE, R. J., AND M. FLANIGON. 1978. An empirical comparison of one-zero,
focal-animal, and instantaneous methods of sampling spontaneous
primate social behavior. Primates 19: 353-61.
SCHNEIRLA, T. C. 1950. The relationship between observation and experi-
mentation in the field study of behavior. Ann. N. Y. Acad. Sciences
51: 1022-44.
SIMPSON, M. J. A., AND A. E. SIMPSON. 1977. One-zero and scan methods for
sampling behavior. Anim. Behav. 25: 726-31.
SLATER, P. J. B. 1978. Data collection. P. W. Colgan, ed. Quantitative
ethology. Wiley, New York.
WEBB, E. J., D. T. CAMPBELL, R. D. SCHWARTZ, AND L. SECHREST. 1966. Un-
obtrusive measures: nonreactive research in the social sciences. Rand
McNally, Chicago.
Insect Behavioral Ecology-'83 Sivinski
THE BEHAVIORAL ECOLOGY OF VERMIN
J. SIVINSKI*t
SYNOPSIS
Some topics in evolutionary biology might be of interest to medical
entomologists, particularly those concerned with the dispersal of ectopara-
sites. These include:
Sexual selection-Male competition and female mate choice may influence
the propensity to disperse, which hosts are chosen, and where parasites are
located on an animal. Phoretic flies on beetles can serve as models for verte-
brates and their vermin and illustrate some possible sexually selected
patterns of distribution.
The maintenance of sex-There is a cost to sexuality best described as
the cost of producing males. Plant, but not animal, ectoparasites commonly
mitigate this cost through cyclic parthenogenesis. It is suggested that pheno-
typic variability produced by the immune system of vertebrates may select
for genetically heterogeneous offspring, i.e. sexual reproduction.
The extended phenotype-The notion that "gene" activity may extend
into the form and behavior of a symbiont suggests that symptoms of in-
fection or infestation should be considered from the perspective of both the
host and the parasite. Vertebrate pathogens may influence the movement
of ectoparasitic vectors.
INTRODUCTION
I am not a medical entomologist and I do realize there is a certain pre-
sumption in writing on topics outside one's field of study. However, the
perspective of an outsider can sometimes include features overlooked by
the specialist. With this somewhat arrogant apology in mind, I will present
some behavioral and evolutionary themes that might be of interest to
medical entomologists. These topics are sexual selection, the maintenance
of sex and the notion of the extended phenotype, particularly as they pertain
to ectoparasitic insects and what seems to me to be a central problem in
medical entomology, the dispersal of insects both between and over the
surfaces of host animals.
SEXUAL SELECTION
Sexual selection results from the difference in male and female invest-
ment in offspring (Trivers 1972, Thornhill 1980). At its simplest, females
invest in large gametes and their reproductive success is limited by their
ability to make eggs and obtain the highest quality paternal genes for their
offspring. Males make cheap gametes and their reproduction is limited by
their access to females. Mate competition among low-investing males creates
intrasexual pressures that commonly result in the evolution of fierce, fast, or
sneaky males. Female choice of mates generates intersexual selection that
*A postdoctoral fellow employed through a cooperative agreement between the Insect
Attractants, Behavior, and Basic Biology Research Laboratory, Agricultural Research
Service, USDA, P. O. Box 14565. Gainesville, FL 32604 and the Department of Entomology
and Nematology, University of Florida.
tMention of a commercial or proprietary product does not constitute an endorsement
by the USDA.
Florida Entomologist 67(1)
can favor, among other things, male advertisement and greater female
powers of discrimination.1
Intra- and intersexual selection influence the movement of ectoparasites.
As a simple example, it can behoove males but not necessarily females to go
from host to host to find as many mates as possible.2 Among ectoparasites
this is reflected by the winged males but flightless females of the bat infest-
ing fly Ascodipteron spp. (see Hackman 1964).
Things get more complicated. In Carnus hemapterus, an acalypterate fly
that feeds on the skin secretions of nestling birds, all females and two-thirds
of the males shed their wings (Capelle and Whitworth 1973). Apparently
among male C. hemapterus there are both searchers and stayers. Reasons
for the difference are unknown but might include the probability of sharing
the nest with females and how intimidating the local rivals are.
When the possibility of females foraging for the best possible mates
is added, complexity is compounded. The tiny dung fly Borborillus frigi-
pennis (Sphaeroceridae) is a kleptoparasite that lays its eggs in the dung
stores of scarab beetles (Sivinski 1983). It bears a number of parallels to
certain lice and fleas and the symbionts can serve as a kind of scale model
of a vertebrate and its vermin. A fly often stays on a single host for 30%
of its adult life, for it rides on bettles both underground and in the air,
and important to any extrapolation, it mates upon the host. Like Carnus, B.
frigipennis males are stayers or searchers. This is not obvious from wing
polymorphisms but is seen in the way flies behave when they mount un-
occupied scarabs. Some hop on and ride, others scurry over beetles, dis-
mount and await another that they search in turn. However, in B. frigi-
pennis female as well as male dispersion appears to be influenced by sexual
motives.
The mean male/female sex ratio of beetle-back fly groups rises and then
falls with the increasing size of the group. This pattern is actually due to a
female's preference for sparsely and densely inhabited beetles (Sivinski, un-
published data). Females may like empty beetles because of the lower levels
of competition their larvae are likely to face, and there is some evidence
that females come to bigger groups in order to mate with the most competi-
tive of a large sample of males (see Sivinski 1984). If so, these beetle-back
aggregations are similar to what Richard Alexander calls resource based
leks (Alexander 1975, Lloyd 1979). That is, females prefer certain resources
not only because of their quality as food or oviposition sites, but also be-
cause of the sexual opportunities offered by associated males.
Among the more vagile ectoparasites, such as Hippoboscidae and Strebli-
dae, perhaps further study will add sexual partners already on an animal
to the list of qualities such as hair size, molting pattern, health, grooming,
and body temperature that influence the suitability of hosts. If so, this will
probably not be a universal criterion for host choice. Some species may
typically occur in such large numbers that the differences among animal
borne populations will be trivial. That is, if the sample of males on each
host is very large then the between-host variance in male quality is apt to
be low and animals will present a similar set of sexual partners. For
example, such lack of between-site variance might explain the failure of
yellow dung fly, Scatophaga stercoraria, females to choose dung pat ovi-
position sites on the basis of resident males (see the results of Borgia 1979).
The distribution on an animal's surface, as well as dispersal among
March, 1984
Insect Behavioral Ecology-'83 Sivinski
animals, can be sexually selected. A B. frigipennis perched on a beetle's
horn is usually a male. Horns may be good lookouts from which to search
for females. Among ectoparasites there are patterns of distribution where
adults are separated by sex, or less widely dispersed than their immatures.
Some of these may be reflections of mating strategies (see Table 1). For
instance the biting louse of cattle (Haematopinus eurysternus) has an in-
triguing distribution. Females are broken up into adult and nymphal
clusters. Adults in groups of up to 60 oviposit together on the back of the
neck. Males occur on the sides of the neck, or are associated singly or in
small groups with nymphal clusters (Craufurd-Benson 1941). There may be
a complex sexual environment on backs of cattle where some males obtain
mates by waiting for the immatures they guard to grow up, while others
condense on mating grounds that are near but not in breeding sites, either
to intercept migrating females or to form leks where females shop for the
most attractive mates.3
THE MAINTENANCE OF SEX
While considering sex, it is worth noting its surprising ubiquity among
very intimate insect ectoparasites, particularly lice. There is a cost to
sexuality best described as the cost of producing males; generally, a lineage
would increase twice as fast if it consisted of parthenogenetic females.
Therefore, there must be a greater than 2-fold reproductive advantage in
producing genetically variable offspring for sex to resist replacement by
asexual mutants (see Bell 1982 for a lengthy discussion). Most conjectural
TABLE 1. EXAMPLES OF ECTOPARASITE DISTRIBUTIONS THAT MAY REFLECT
MATING SYSTEMS.
Order
Insect
Comments and References
Dermaptera
Hemimerius talpoides
Hemiptera
Cimicidae
Phthiraptera
Haematopinus
eurysternus and
Damalinia bovis
Diptera
Melophagus spp.
Mystacinobia
zelandica
Joblingia
schmidti and
Trichobius yunkeri
Siphonaptera
Echidnophaga
gallinacea
Adults more anterior on backs of rats
(Ashfbrd 1970).
Aggregations of cimicids off hosts-bug trains
(e.g., Lee 1955, Overal and Wingate 1976,
Cheng 1973).
Adult and nymphas clusters on cattle
(Craufurd-Benson 1941).
Males congregate on hind parts of sheep
(Graham and Taylor 1941).
Communal oviposition and adult clusters off
bat-hosts (Holloway 1976).
Swarm in bat caves (Wenzel et al. 1966).
Males on body, females on heads of
chickens (Suter 1964).
Florida Entomologist 67(1)
advantages to sex suppose environmental heterogeneity selects against
genetically homogeneous clones, and one of the most important sources of
environmental heterogeneity is biotic, the unpredictable amalgam of preda-
tors, competitors, and symbionts (e.g., Hamilton et al. 1981). Relationships
with hosts are certainly intense and it is easy to see why colonizing ecto-
parasites might be sexual products. But it is less clear why sex should
continue once a comparable, relatively unutilized animal has been reached. It
would seem that cattle backs should be more biotically homogeneous than
the pasture in which they stand. There are very few records of predation on
lice by animals other than the host, and little opportunity for, or direct evi-
dence of competition (see however, Wenzel and Tipton 1966, and Hopkins
1949). Actually a few species of biting lice do what is expected of them;
males become progressively more rare after an animal is colonized (all are
Ischnocera, see citations in Marshall 1981). This type of reproduction is
much more common, however, among the ectoparasites of plants such as
aphids and scales (see Price 1980 for a discussion of parthenogenic para-
sites).
It is reasonable to ask how trees and cows differ as sexual substrates?
Could the genetic scrambling of animal ectoparasites be an attempt to keep
up with changes in host phenotype, a pace of change that is not matched in
long-lived plants? This is not to say that plants do not respond to their
parasites, but that animals defend themselves in ways that plants do not, such
as by learning and employing a sophisticated immune system (see Smith
1983 for an example of plant response). It is well known that animals differ
in susceptibility to infestation. For example, the body louse, Pediculus
humanus, has been known to do well on one person, while refusing to feed
on his brother (Riley and Johannsen 1938). The head louse, Pediculus capitis,
while specific to humans, prefers women to men and European strains do
not survive on blacks (citations in Marshall 1981). My impression is that
the causes of such variances are not always well understood (e.g., Nelson
et al. 1977). Certain individuals may carry compounds in their blood that
are toxic to endosymbionts or, as in the case of the generally undrinkable
blood of guinea pigs, have haemoglobins that crystalize and rupture the
gut (see Krynski et al. 1952, Nelson et al. 1975); or perhaps they carry
psychoactive compounds like a turn of the century French sailor whose blood
caused body lice to fight each other with "apparently vicious intent" (Foot
[1920] in Riley and Johannson [1938]; note that Polybia wasps will not sting a
hand covered with underarm perspiration, suggesting "secondary chemicals"
that protect humans against insects other than ectoparasites; see Young
1978). Some forms of defense are known to be acquired. A rise and then
a fall in the number of lice, mites, and keds on an animal is a fairly common
pattern (e.g., Nelson et al. 1977). For example, local vasoconstriction that
leaves insects unable to feed is a major form of such an acquired resistance.
This and other less understood reactions are apparently regulated by the
immune system.
It is tempting to think that a louse never bites the same animal twice;
that the defensive physiology of the host is sufficiently labile to force
migration/sexual recombination or to select for genetic shuffling in the
parasites that stay.
March, 1984
Insect Behavioral Ecology-'83 Sivinski
THE EXTENDED PHENOTYPE
A third area of evolutionary thought with implications for medical
entomology is the notion of the extended phenotype recently formalized by
Richard Dawkins (1982). Its basic concept is that the expression of a genetic
program commonly extends beyond the body walls, the traditional limits of
gene activity. Few would argue that caddisfly cases and termite mounds have
evolved through the differential reproduction of genes the constructions
themselves do not contain, but these are only the most obvious extrusions of
gene activity through an animals "skin". Dawkins (1982) has emphasized
the possibility that animals have evolved means to physically or psycho-
logically control each other and that nature may be a tangle of manipula-
tive forces stretched among incompletely autonomous genomes. An outcome
of extended phenotype thinking is heightened doubts about whose genes are
controlling whose body in cases of parasitism. That is, do "symptoms"
benefit the infector or the infected.4
Consider the dispersal of animal diseases. Only a few microbes seem to
take a direct hand in their own contagion. For instance certain bacterial
pathogens of arthropods luminesce and probably attract new victims or
vectors to themselves (see Harvey 1952, also Sivinski 1981, 1982). But
usually microbes would be best served by subverting their larger and more
complex hosts into spreading them around (Holmes and Bethel 1972, Ewald
1980, Dawkins 1982).
Such manipulation does not have to entail prodigious intellectual or
physical feats. One need only reflect on who benefits from sneezes associated
with cold virus or the biting of rabid dogs.5 An example closer to our theme
is that of tse-tse flies, which when infected with Trypanosoma brucei, feed
more often and more voraciously (Jenni et al. 1980). The trypanosomes are
associated with mechanoreceptors in the labrum that function in a feedback
loop to restrict probing. Probing is essential for transmission of the trypano-
some. A "gene" that changes the site of infection in a vector thus can be
better dispersed by undermining a fly's ability to determine how much
biting is enough.
Does anything like this influence the movement of ectoparasites? The
only case I know of is where a nematode that infects the flea Spilopsyllus
cuniculi, apparently causing its victim to remain and mate on doe rabbits
rather than moving onto their litters (Rothchild 1969). The purpose, if any,
is obscure. But one can imagine obviously functional changes in the be-
havior of ectoparasitic vectors that might be worth searching for. As an
example, it has been noted many times that some fleas and lice leave sick or
disturbed hosts (see citations Marshall 1981). A pathogen that causes
mild disease and could lower thresholds to cues ectoparasites use to monitor
host health (e.g., body temperature), would be able to hijack a flea out of
an environment eroded by antibodies. I know of no evidence for such hijack-
ing, but it is worth noting that sucking lice that might be vectors sometimes
appear to leave sick hosts more rapidly than certain biting lice that can-
not be vectors (again see citations Marshall 1981). Be that as it may, as a
general principle, it should be useful to keep in mind the options open to the
protagonists in diseases.
Florida Entomologist 67(1)
APPENDIX
1-While the major concern here is the sexual selection of parasite be-
havior, parasites themselves might provide opportunities for sexual selection
to occur in their hosts. Hamilton and Zuk (1982) have argued that displays
of male vigor and ability to grow and maintain extravaganzas of feathers or
fur could be advertisements of resistance to parasites. Such a scheme is an
improvement over traditional "good gene" models since coevolution of para-
site and host might generate genetic variance that intersexual selection
could not exhaust.
2-A related problem among the more vagile bloodsucking "micropreda-
tors" of veterinary importance is whether males should search for females
on or near hosts. It seems curious for instance that one can be surrounded
by large numbers of sanguinary female mosquitoes, but that no male
mosquitoes are overhead taking advantage of the concentration. There are
at least two determinants of male search strategy: 1) where are females
most likely to be concentrated, i.e. encountered (see Sivinski and Stowe
1980), and 2) the value of females in the different locales they inhabit-in a
monogamous species that normally mates upon emergence, subsequent con-
centrations on hosts are sexually useless to searching males. Parenthetically,
where females mate more than once and the last ejaculate fertilizes most of
the eggs, copulations just before oviposition are most valuable, and females
aggregated around a host may not be attractive if the host is widely sepa-
rated from oviposition sites (see Thornhill and Alcock 1983).
There are some data from the Diptera with which to test the later of these
determinants; i.e., the principle of changing female value over space and
time. Mosquitoes have distinct feeding and oviposition sites, and females
generally copulate only once (Gillett 1972). Male mosquitoes would be pre-
dicted to concentrate their mating efforts at emergence sites, and as expected,
males are only rarely found in the vicinity of hosts. However, I have found 13
species where males are located near hosts (the bizarre kleptoparasite of
ants Malaya leei, Miyagi 1981; Mansonia sp., McIver et al. 1980; Eretma-
podies chrysogaster, Gillett 1972; Aedes aegypti, Hartberg 1971; A. albopic-
tus, Basio et al. 1976; A. dominicii, Bates 1949; A. furcifer, Jupp 1978; A.
pseudoscutellaris, Horsfall 1955; A. scutellaris, Forbes and Horsfall 1946; A.
triseriatus, Loor and DeFoliart 1970; A. varipalpus, Lee 1971; A. vittatus,
Reeves 1951; A. diantaeus, Horsfall 1955). These exceptions are of interest
because 11 out of the 12 haematophagous species develop in small con-
tainers, principally rot holes (compared to only 59% of 409 species of Aedes
in Horsfall 1955). Such small, ephemeral, widely dispersed development sites
may make it difficult for males to search for emerging mates. If so, they
are "forced" to locate older females, ones less likely to be receptive, in the
vicinity of hosts (note that several Aedes species are found both near animals
and in the vicinity of their own emergence site, suggesting a dual sexual
strategy).
A similar case occurs in horn fly, Haematobia irritans. Females mate
once, both sexes are on cattle, and larvae develop in the ephemeral and
dispersed medium of cattle dung (see Bruce 1964). In a close relative, the
moose fly, Lyperosiops alcis, copulatory frequency is unknown but males are
associated with hosts, and females have been found ovipositing on feces
several inches up the rectum of freshly killed moose (Snow 1891).
A possibly contrary system to the proposal that males prefer to patrol
emergence sites in female monogamous species and oviposition sites in
polyandrous ones, is the multiple mating and male host occupancy of certain
psychodids whose females feed on reptile blood (notably Phlebotomus
vexator; see Chaniotis 1967). Larval substrates are undetermined but may
consist of host feces. If so, the difficulty is mitigated since valuable last
March, 1984
Insect Behavioral Ecology-'83 Sivinski 63
matings before oviposition could be obtained by males stationed on the fe-
male. Males of ceratopogonid Culicoides utahensis wait in the ears of rabbits
for feeding females (Downes 1969). Again, it is possible that larvae develop
in the litter of the host's burrow. Likewise, in "bobos" (Paraleucopis mexi-
cana) a chamaemyiid that laps fluids from the eyes of birds and reptiles,
both sexes are found about hosts and larvae are thought to develop in the
litter of birds' nests (Smith 1981).
Multiple inseminations and near-host male aggregations also occur in
tse-tse (Glossina spp: Muscidae) (citations in Mulligan 1970, Tobe and
Langley 1978). Peculiarities of tse-tse reproduction remove any difficulty.
Unlike most Diptera, Glossina spp. are viviparous, so that zygote formation
occurs long before deposition of offspring.
Male host occupancy occasionally occurs in other vertebrate-associated
fly taxa. Lack of information on mating behavior and/or oviposition sites
precludes analysis in these species: Culicoides nebeculosus (Ceratopogoni-
dae) (Downes 1955); C. variipenis (Jones et al. 1977); Lutzomyia vexatrix
(Psychodidae) (Chaniotis 1967); Tabanus auropuntatus, Haemoptopota
sewelli, H. pluvialis (Tababanidae) (Bailey 1948); and Wilhelmia equina
(Simuliidae) (Wenk and Schlorer 1963).
3The idea of lekking is particularly appealing when thinking about ecto-
parasites on large animals. Like lake-breeding mosquitoes or highly poly-
phagous tephritids, many ectoparasite populations look like they can be
widely distributed over an extensive and fairly uniform "resource surface"
(see Burk 1981 and Sullivan 1981 for discussions of the relationship between
resource concentration and mating systems). An effect of uniformity can be
unpredictability in locating sexual partners, and the result of this can be the
evolution of true or nonresource-based leks where males aggregate and
signal from an arena devoid of any special resources other than the males
themselves (see Bradbury 1981 for recent consideration of lek evolution).
4As an illustration with some preliminary data, consider galls, "subcu-
taneous" parasites of plants. Galls are formed around a number of organ-
isms including cynipid wasps and cecidomyid flies. One of their striking
qualities is the breadth of their structural complexity and color. They range
from green warts to objects that rival flowers. If color is treated as a
symptom of infestation thd extended phenotype question is: in whose inter-
est, the plant or the insect, is the color produced?
To expand the metaphor of the flower, could these colors attract
Hymenoptera, parasitic ones as opposed to pollinators? Up to 70% of galls
are commonly parasitized (Russo 1979). Color and shape could be a flag
that a plant raises over an infection to attract the macroscopic equivalent
of an antibody. Gall former of course would try to strike the colors down.
Are some colors attractive to parasites? Yes. Catches of parasitoids are
higher on gall-sized yellow balls (Tack-Traps covered and hung in trees)
than green or red ones (green 29 parasitoids, 9% of catch; red 21 parasites
7.6% of catch; yellow 67 parasites, 14.75 of catch; yellow > red, green p
< 0.05) (Sivinski, unpublished data).
Is yellow a common color on galls? Yes, of 525 oak-leaf galls, 20% are
yellow or have yellowish tints at some point in their development. (This and
following color data collected from the keys and descriptions of Felt 1918
and Russo 1979.)
Is there any reason to think that the presence of yellow is a signal?
Perhaps. If we compare oak-root galls hidden underground to those visible
on twigs and branches, a higher proportion of those that can be seen have
the proported signal color (18% of 165 branch galls vs. 7% of 14 root galls).
Is color more common in situations where selection for parasite removal
is strongest? Perhaps. Many gall-forming insects are weak fliers and poor
dispersers, so perennial plants run the risk of reinfection by the offspring
Florida Entomologist 67(1)
March, 1984
of the previous seasons' parasites. Annuals however are less likely to
survive their gall former and in the absence of younger relatives to protect,
should invest less in disinfecting. Yellow seems more common on perennial
galls of compositoid plants (0 of 13 annuals and 12 of 107 perennials).
On the other hand, gall former could benefit from gaudy houses. Many
galls have defensive attributes such as spines or tannic acid levels of up to
65% that could be advertised by bright warning colors. It is interesting that
galls formed by bacteria, fungi and mites, and presumably immune at least
to parasitoids, are sometimes brightly colored.
Both or neither of these explanations may be correct, but they should
illustrate the possibility that the appearance or behavior of an animal may
not be the work of its own genes and that "symptoms" deserve being con-
sidered from both perspectives.
5The ability of a sick host to defend itself against vectors is considerably
curtailed. Mosquitoes given a choice of feeding on a healthy or malarious
mouse almost always suck from the infected animal (Day et al. 1983). It
might be interesting to look for any differences in the ability of animals
to discourage vermin when ill with diseases transmitted by and without
vectors. Would the former be more listless, less able to brush away a fly?
ACKNOWLEDGMENTS
D. Bieman, T. Forrest, N. Leppla and J. E. Lloyd made helpful criticisms
of the manuscript that was professionally prepared by E. Turner.
LITERATURE CITED
ALEXANDER, R. 1975. Natural selection and specialized chorusing behavior
in acoustical insects. Pages 35-77 in D. Pimentel ed. Insects, science
and society. Academic Press, New York.
ANDERSON, J. R. 1974. Symposium on reproduction of arthropods of medical
and veterinary importance. II. Meeting of the sexes. J. Med. Ent. 11:
7-19.
ASHFORD, R. W. 1970.' Observations on the biology of Hemimerus talpoides
(Insecta: Dermaptera). J. Zool. London 162: 413-18.
BAILEY, N. S. 1948. The hovering and mating of tabanidae: a review of
the literature with some original observations. Ann. Ent. Soc.
America 41: 403-12.
BASIO, R. G., M. S. CHANG, C. E. GAJUDO, AND K. V. MENON. 1976. Notes on
the swarming and mating of Aedes albopictus (Skuse) in West
Malaysia. Philippine Ent. 3: 241-45.
BATES, M. 1949. The natural history of mosquitoes. The MacMillan Co.,
New York.
BELL, G. 1982. The masterpiece of nature. University of California Press,
Berkeley.
BORGIA, G. 1979. Sexual selection and the evolution of mating systems. Pages
19-80 in M. S. Blum and N. A. Blum eds. Sexual selection and repro-
ductive competition in insects. Academic Press, New York.
BRADBURY, J. W. 1981. The evolution of leks. Pages 138-69 in R. D. Alexand-
er and D. W. Tinkle eds. Natural selection and social behavior. Black-
well Scientific Publications, Oxford.
BRUCE, W. G. 1964. The history and biology of the horn fly Haematobia
irritans (L.): with comments on control. North Carolina Agric. Exp.
Stn. Tech. Bull. 157.
BURK, T. 1981. Signaling and sex in acalyptrate flies. Florida Ent. 64: 30-
43.
Insect Behavioral Ecology-'83 Sivinski
CAPELLE, K. J., AND T. L. WHITWORTH. 1973. The distribution and avian
hosts of Carnus hemapterus (Diptera: Milichiidae) in North America.
J. Med. Ent. 10: 525-26.
CHANIOTIS, B. N. 1967. The biology of California Phlebotomus (Diptera:
Psychodidae) under laboratory conditions. J. Med. Ent. 4: 221-33.
CHENG, T. C. 1973. General parasitology. Academic Press, New York.
CRAUFURD-BENSON, H. J. 1941. The cattle lice of Great Britain. Parasi-
tology 33: 331-58.
DAWKINS, R. 1982. The extended phenotype. W. H. Freeman and Co.,
Oxford.
DAY, J. F., K. M. EBERT, AND J. D. EDMAN. 1983. Feeding patterns of mos-
quitoes (Diptera: Culicidae) simultaneously exposed to malarious and
healthy mice, including a method for separate blood meals from con-
specific hosts. J. Med. Ent. 20: 120-7.
DOWNES, J. A. 1955. Observations on the swarming flight and mating of
Culicoides (Diptera: Ceratopogonidae). Trans. Royal Ent. Soc.
London 106: 213-36.
1969. The swarming and mating flight of Diptera. Annu. Rev. Ent.
14: 271-98.
EWALD, P. W. 1980. Evolutionary biology and the treatment of signs and
symptoms of infectious disease. J. Theor. Biol. 86: 169-76.
FELT, E. J. 1918. Key to American insect galls. New York University
Press, New York.
FORBES, J., AND W. R. HORSFALL. 1946. Biology of a pest mosquito common
in New Guinea. Ann. Ent. Soc. America 39: 602-6.
GILLETT, J. D. 1972. The mosquito. Doubleday & Co., Inc. New York.
GRAHAM, N. P. H., AND K. L. TAYLOR. 1941. Studies on some ectoparasites
of sheep and their control I. Observations on the bionomics of the sheep
ked (Melophagus ovinus). C.S.I.R.O., Australia Pamphlet 108, 9-26.
HACKMAN, W. 1964. On reduction and loss of wings in Diptera. Nat Ent.
44: 73-93.
HAMILTON, W. D., P. A. HENDERSON, AND N. A. MORAN. 1981. Fluctuation
of environment and coevolved antagonist polymorphism as factors in
the maintenance of sex. Pages 363-81 in R. D. Alexander and D. W.
Tinkls eds. Natural selection and social behavior. Chiron Press, New
York.
ZUK, M. 1982. Heritable true fitness and bright birds: a role for
parasites? Science 218: 384-7.
HARTBERG, W. K. 1971. Observations on the mating behavior of Aedes
aegypti in nature. Bull. World Health Orgn. 45: 847-50.
HARVEY, E. N. 1952. Bioluminescence. Academic Press, New York.
HOLLOWAY, B. A. 1976. A new bat-fly family from New Zealand (Diptera:
Mystacinobiidae). New Zealand J. Zool. 3: 279-301.
HOLMES, J. C., AND W. M. BETHEL. 1972. Molification of intermediate host
behavior by parasites. Pages 123-49 in E. V. Canning and C. A. Wright
eds. Behavioral aspects of parasite transmission. Academic Press,
London.
HOPKINS, G. H. E. 1949. The host associations of the lice of mammals.
Proc. Zool. Soc. London 119: 387-604.
HORSFALL, W. R. 1955. Mosquitoes. Ronald Press Co., New York.
HUTSON, A. M. 1978. Associations with vertebrates, their nests, roosts and
burrows. Pages 143-51 in A. Stubbs and P. Chandler eds. A dipterist's
handbood. Amat. Ent. Soc., Middlesex, United Kingdom.
JENNI, L., D. H. MOLYNEUX, J. L. LIVESEX, AND R. GALUN. 1980. Feeding
behavior of tse-tse flies infected with salivarian trypanosomes. Nature
283: 383-85.
JONES, R. H., R. O. HAYES, H. W. POTTER, JR., AND D. B. FRANCY. 1977. A
Florida Entomologist 67(1)
survey of biting flies attacking equines in three states of the
southwestern United States. J. Med. Ent. 14: 441-47.
JUPP, P. G. 1978. A trap to collect mosquitoes and flies attracted to
monkeys and baboons. Mosq. News 38: 288-9.
KRYNSKI, S., A. KUCHTA, AND E. BECLA. 1952. Research on the nature of
the noxious action of guinea-pig blood on the body-louse. Bull. Inst.
Mar. Med. Gdansk 4: 104-7.
LEE, R. D. 1955. The biology of the Mexican chicken bug Haematosiphon
inodorus (Dugg's) (Hemiptera: Cimicidae). Pan-Pacific Ent. 31:
47-61.
LEE, D. 1971. The role of the mosquito, Aedes sierrenis, in the epizoology
of the deer body worm, Setoria yehi. Ph.D. Dissertations, University of
California, Berkeley.
LLOYD, J. E. 1979. Mating behavior and natural selection. Florida Ent.
62: 17-34.
LOOR, K. A., AND G. R. DEFOLIART. 1970. Field observation on the biology
of Aedes triseriatus. Mosquito News 30: 60-4.
MARSHALL, A. G. 1981. The ecology of ectoparasitic insects. Academic Press,
London.
MCIVER, S. B., T. J. WILKES, AND M. T. GILLIES. 1980. Attraction to
mammals of male Masonia (Mansonioides) (Diptera: Cullicidae).
Bull. Ent. Res. 70: 11-16.
MIYAGI, I. 1981. Malaya leei (Whorton) feeding on ants in Papua New
Guina (Diptera: Culicidae). Jap. J. Sanit. Zool. 32: 332-3.
MULLIGAN, H. W. 1970. The African trypanosomeases. G. Allen and Uni-
versity, Ltd. London.
NELSON, W. A., J. E. KEIRANS, J. F. BELL, AND C. M. CLIFFORD. 1975. Host-
ectoparasite relationships. J. Med. Ent. 12: 143-66.
J. F. BELL, C. N. CLIFFORD, AND J. E. KEIRANS. 1977. Interactions
of ectoparasites and their hosts. J. Med. Ent. 13: 389-428.
OVERAL, W. L., AND L. R. WINGATE. 1976. The biology of the batbug Strietici-
mex antennatus (Hemiptera: Cimicidae) in South Africa. Ann. Nat.
Mus. 22: 821-8.
PRICE, P. W. 1980. Evolutionary biology of parasites. Princeton University
Press, Princeton, New Jersey.
REEVES, W. C. 1951. Field studies on carbon dioxide as a possible host
stimulant to mosquitoes: Proc. Soc. Exp. Bio. Med. 77: 64-6.
RILEY, W. A., AND O. A. JOHANNSEN. 1938. Medical entomology. McGraw-
Hill Book Co., Inc., New York.
ROTHSCHILD, M. 1969. Notes on fleas: with the first record of a mermithid
nematode from the order. Proc. Br. Ent. Nat. Hist. Soc. 1: 1-8.
Russo, R. A. 1979. Plant galls of the California region. The Boxwood
Press, Pacific Grove, California.
SIVINSKI, J. 1981. Arthropods attracted to luminous fungi. Psyche 88:
383-90.
1982. Prey attraction by luminous larvae of the fungus gnat Orfelia
fultoni. Ecol. Ent. 7: 443-6.
1983. The natural history of a phoretic sphaerocerid fauna. Ecol.
Ent. (in press).
1984. Eexual conflict and choice in a phoretic fly, Borborillus frigi-
pennis (Sphaeroceridae). Ann. Ent. Soc. America (in press).
AND M. STOWE. 1980. A kleptoparasitic cecidomyiid and other flies
associated with spiders. Psyche 87: 337-48.
SMITH, N. G. 1983. Host plant toxicity and migration in the day-flying
moth Urania. Florida Ent. 66: 76-85.
SMITH, R. L. 1981. The trouble with "bobos", Paraleucopis mexicana Steys-
kal, at King Bay, Sonora, Mexico (Diptera: Chainaemyiidae). Proc.
March, 1984
Insect Behavioral Ecology-'83 Sivinski 67
Ent. Soc. Washington 83: 406-12.
SNOW, W. A. 1891. The moose fly-a new Haematobia. Canadian Ent. 23:
87-9.
SULLIVAN, R. 1981. Insect swarming and mating. Florida Ent. 64: 44-65.
SUTER, P. R. 1964. Biologie von Echidnophaga gallinacea (Westw.) und
vergleich mit andern verhaltenstypen bei flbhen. Acta Trop. 21:
193-238.
THORNHILL, R. 1980. Competitive, charming males and choosy females:
was Darwin correct? Florida Ent. 63: 5-30.
AND J. ALCOCK. 1983. The evolution of insect mating systems.
Harvard University Press, Cambridge, Massachusetts.
TOBE, S. S., AND P. A. LANGLEY. 1978. Reproductive physiology of Glossina.
Annu. Rev. Ent. 23: 283-308.
TRIVERS, R. 1972. Parental investment and sexual selection. Pages 136-79
in B. Campbell ed. Sexual selection and the descent of man, 1971-1971.
Aldine-Atherton, Chicago.
WENK, P., AND G. SCHLORER. 1963. Wirtsorientierung und Kopulation bei
blutsaugenden Simuliiden (Diptera). Z. Tropenmed. Parasitol. 14:
177-91.
WENZEL, R. L., AND TIPTON, V. J. 1966. Some relationships between mammal
hosts and their ectoparasites. Pages 677-723 in R. L. Wenzel and V. J.
Tipton eds. Ectoparasites of Panama. Field Mus. Nat. History,
Chicago.
YOUNG, A. M. 1978. A human sweat-mediated defense against multiple
attacks by the wasp Polybia diguetana in northwestern Costa Rica.
Biotropica 10: 73-4.
Florida Entomologist 67(1)
ENERGETIC: THE BEHAVIORAL AND ECOLOGICAL
CONSEQUENCES OF BODY SIZE
BRIAN K. McNAB*
SYNOPSIS
The significance of body size for animal energetic is demonstrated by the
seasonal behavior of the monarch butterfly and the fin whale. Both species
in summer live in environments that are unacceptably harsh in winter;
consequently, the butterfly and whale migrate in winter to hospitable en-
vironments at lower latitudes, which nevertheless are characterized by
limited food supplies. As a result, both species in winter principally rely
on stored lipids as their source of energy. The difference between the
butterfly and whale in the time period over which starvation is tolerated
can be accounted for by their differences in body mass and in their level of
energy expenditure. The huge difference in body size between butterflies and
whales does not necessarily mean that the solutions to environmentally im-
posed problems are different.
The one most important characteristic of animals is, above all, body size.
It influences everything from the means of temperature regulation, type and
amount of food consumed, and rate of locomotion to longevity, potential
predators, and rate of reproduction. Size may be measured in many ways,
including body length, wing spread, body volume, and body mass. Body
mass is preferentially used as a measure of size in physiological studies be-
cause most functions depend on the amount of material reacting with the
environment. For example, the rate of energy expenditure in animals
normally is compared to their masses, and such a comparison has shown
(Figure 1) that an individual's rate of metabolism (i.e., the total rate)
generally increases with body mass raised (approximately) to the 3/4 power
(i.e., mO.75). This pattern occurs both in ectotherms, i.e., in animals that
have body temperatures determined by ambient conditions, and in endo-
therms, animals that have temperatures that are determined mainly by
high rates of chemical heat production (Hemmingsen 1960). It is some-
what disconcerting to note, however, that no adequate explanation for the
mo.07 proportionality has ever been given.
At any particular body size, endotherms have rates of metabolism that are
about 9 times those of ectotherms, assuming that endotherms have a body
temperature of about 39 C and that ectotherms have a temperature of about
200C, although endotherms can reduce energy expenditure somewhat by
having an effective insulation. When body temperature in ectotherms is
lower than 100C, the metabolism-mass curve, although parallel to the
endotherm and 200C curves, is lower still (Figure 1): at 10C rate of
metabolism is only about 4% of that expected in endotherms. Another way
of stating that total rates of metabolism are proportional to m075 is to note
that mass-specific rates of metabolism are proportional to m-0.25, which
*Brian K. McNab is Professor in the Department of Zoology at the University of
Florida. His interests are principally in the physiological ecology of vertebrates. Much of
his recent work has examined the ecological factors that influence the level of energy ex-
penditure in mammals and has attempted to place this information into an evolutionary
context. Current Address: Department of Zoology, University of Florida Gainesville, FL
32611.
March, 1984
Insect Behavioral Ecology-'83 McNab
1 1 I I I I I I I
I I
(D
I I
<*
I I
I I
o
I I r I I I
V4S1108V13VI 1 d 31VHl OLDO0
CO
Ft
0
4-
69
o
E-4
0o .
0
00 0
a-
0 0
W 0a
0 0C
I-~t
C1 1
0 3
3 P4
0 /
a~ g '
O u^
1^~
o o .
CTU'-1^.
'S o ^=
EQ|
ed Eoo
*^ w f
bSS^S
*r c M *
Co
-<-
-o
O
-I
-0
Florida Entomologist 67(1)
simply means that small animals have higher rates of metabolism per
gram than large animals. As convenient as mass-specific units are, the
ecologically relevant rates of energy expenditure are the total rates, because
they describe the rate at which energy is used by an individual and the
rate at which food must be harvested in the environment.
The only animals that had been considered to be endothermic, until
recently, were birds and mammals. All animals other than these vertebrates
were thought to be ectothermic. Within the last 20 years, however, many
other animals have been shown to be endothermic to some degree. Among
large vertebrates, some sharks and tunas are known to be endothermic
(Barrett and Hester 1964, Carey and Teal 1969a,b), as are female pythons
while incubating their eggs (Hutchison et al. 1966, Van Mierop and Barnard
1978); the leatherback turtle (Dermochelys) may also be endothermic (Frair
et al. 1972). Among small animals, various insects, including some moths,
hymenopterans, dragonflies, and beetles, are endothermic, at least during
periods of activity (Heinrich 1974, May 1979). Periodic endothermy is of
special interest because it is also found in mammals and birds that weigh
less than 10 g, such as small mice and insectivorous bats, and all humming-
birds. In other words, the existence of continuous endothermy, itself, de-
pends on body size.
At a small mass, animals, independent of taxonomy, can variously be
continuously endothermic, discontinuously (periodically) endothermic, or
completely ectothermic, depending on the level at which energy is expended
(McNab 1983): rate of metabolism must be high to insure effective endo-
thermy, ectothermy (as stated) is associated with low rates of metabolism,
and discontinuous endothermy is characterized by intermediate rates (Figure
1). The transition between ectothermy, at one extreme, and continuous
endothermy, at the other, also varies with body mass: compared to the
standard mammalian relation, small animals must have high rates of
metabolism to remain continuously endothermic, but animals weighing more
than 70 g may have a 'low rate of metabolism without sacrificing continuous
endothermy. These observations mean that another relationship between
total rate of metabolism and body mass can be described, and it defines the
boundary between continuous and periodic endothermy. This relationship,
the so-called boundary curve for endothermy, is proportional to mo.33 (Figure
1); it is derived empirically from measurements on rate of metabolism in
relation to body mass in those species of mammals, birds, fish, and snakes
in which there is continuous endothermy.
At large masses, the resting rate of metabolism of an endotherm may
be less than that of an ectotherm. For example, the mean boundary curve
intersects the 20C ectotherm curve at about 18 kg and the 100C ecto-
therm curve at about 180 kg (Figure 1). These observations mean that the
distinction between endothermy and ectothermy is not simply related to the
level of energy expenditure. Unfortunately, there are few measurements of
energetic in vertebrates at masses greater than a few kilograms, so the
boundary between these states at large masses is unclear, and needs to be
explored, especially in species with an intermediate form of thermal be-
havior. Nevertheless, the existence of the boundary curve at large masses
is shown in the behavior of the Indian python (Python molurus) : it raises
its rate of metabolism sufficiently to exceed the boundary curve (Figure 1),
March, 1984
Insect Behavioral Ecology-'83 McNab
but it does not need "mammalian" rates of metabolism to assure effective
endothermy (McNab 1983).
The impact of body mass on rate of energy expenditure can be shown by
comparing the seasonal energetic of multicellular animals at the two ends
of the size spectrum: a monarch butterfly (Danaus plexippus), which may
weigh only 0.5 g, and a fin whale (Balaenoptera physalus), which may weigh
up to 50,000,000 g. In spite of this great dissimilarity in mass, these species
both migrate to wintering grounds on which they have a restricted food
supply and intake. Monarchs winter in coastal California and in the high-
lands of central Mexico, while fin whales winter in tropical waters. There is
evidence that neither the whale (Brodie 1975) nor the monarch (Chaplin
and Wells 1982) feed much, or at all, on their wintering grounds. It is of
interest here to compare the periods of time that these species live without
eating, or at least with a highly restricted food intake, to see to what
extent the observed periods can be quantitatively accounted for by comparing
energy expenditures, as derived from the differences in their masses.
The amount of time that an animal can live without feeding is pro-
portional to the ratio of the size of the energy store divided by the rate
at which the store is consumed (Morrison 1960). If the size of the energy
store is proportional to body mass, the time period for starvation would be
proportional to mo25 (= ml.00/m0o-7). For example, if two animals are
compared, one the size of a monarch butterfly and the other the size of a
fin whale, and if both have rates of metabolism that fall on the same metabol-
ism-mass curve, then the ratio of time periods would be proportional to
whale mass 0.26 = (10s)0.25 = 100:1. Fin whales may not feed
monarch mass)
(much) for the half-year that they spend in tropical waters, which suggests
that an animal the size of a monarch would be expected to tolerate starva-
tion for about 1.8 (= 180/100) days, but only if it conformed in rate of
metabolism to the endotherm curve. If, however, a monarch were a con-
tinuous endotherm, it would have to follow the boundary curve for endo-
thermy (see Figure 1), which at' 0.5 g would raise the resting rate of
metabolism by a factor of 8.4. That is, a continuous endotherm the size of a
monarch could tolerate 1.8/8.4 = 0.21 days, or about 5 hours, by burning its
fat stores. This graphically illustrates why animals the size of an insect can-
not afford continuous endothermy. Actually, monarchs are not endothermic;
their body temperature equals air temperature, as long as they are not ex-
posed to the sun. In coastal California, monarchs face cool, cloudy weather
in winter; mean body temperature is only about 100C (Chaplin and Wells
1982). Measurements of oxygen consumption indicate that monarchs at
100C have resting rates that are only about 3% of the value expected from
the endotherm curve at 0.5 g. Consequently, the starvation time expected
in monarchs is about 1.8/0.03 = 60 days. Chaplin and Wells estimate that
monarchs in coastal California, given their fat stores, can tolerate starva-
tion for about 60 days, a close agreement, seeing that here the estimate is
derived from an extrapolation of the energetic of whales!
The difference in time period for starvation between monarch butter-
flies and fin whales, as absurd as this comparison might seem at first glance,
can be accounted for by differences in body size, thermal behavior, and level
of energy expenditure. In spite of all of these differences, both species
Florida Entomologist 67(1)
respond to the shortage of food and to seasonally harsh environments-such
as cold temperatures in western North America (in the case of monarchs)
and cold seas and winter storms (in the case of fin whales)-by the use
of a similar response, namely, migration to benign environments, which
nevertheless are characterized by a shortage of food and require the use
of stored energy resources. Given their fixed energy reserves (dictated
as they are by body size), these species reduce their rates of energy expendi-
ture to extend the period over which starvation can be tolerated. In the
endothermic fin whale, energy expenditure is reduced by living in warm
water, that is, by retreating to the tropics. Time of starvation is extended
to great lengths by a large body size, as is required to permit a polar-tropical
migration on an annual cycle. Monarchs, being ectothermic, have to walk a
thermal tightrope, because they have a reduced control over body tempera-
ture: they must avoid freezing temperatures, and must avoid warm
temperatures that raise rate of metabolism, thereby reducing starvation
time. It therefore is significant that wintering monarchs congregate in
coastal California, where the environment in winter is cool and damp, and
in the mountains of central Mexico, where monarchs cluster in forests at an
elevation of 3100 m. Ambient temperatures encountered in Mexico generally
fall between 3 and 120C (L. Brower, personal communication), tempera-
tures that are strikingly similar to those found in winter in coastal Cali-
fornia.
This analysis suggests that many of the obvious differences between
butterflies and whales are related to a striking difference in body mass, and
when this difference is taken into consideration, a commonality is seen in
their biology. Insects, by concentrating at the small end of the size spectrum,
tend to be ectothermic, or if they are endothermic, are so only on a periodic
basis. But as has been seen in the case of monarch butterflies, a small body
size does not mean that the ecological problems faced, or even some of the
solutions used, are necessarily different from those of large endotherms.
ACKNOWLEDGEMENTS
I thank Jim Lloyd for the invitation to speak to entomologists on ener-
getics from a vertebrate perspective. He improved the manuscript and
tolerated my delays. Lincoln Brower kindly informed me of the environ-
mental temperatures at the Mexican highland localities at the time of
monarch residence.
LITERATURE CITED
BARRETT, I., AND F. J. HESTER. 1964. Body temperature of yellowfin and
skipjack tunas in relation to sea surface temperature. Nature, London
203: 96-97.
BRODIE, P. F. 1975. Cetacean energetic, an overview of intraspecific size
variation. Ecology 56: 152-61.
CAREY, F. G., AND J. M. TEAL. 1969a. Mako and porbeagle: warm-bodied
sharks. Comp. Biochem. Physiol. 28: 199-204.
1969b. Regulation of body temperature by the bluefin tuna. Comp.
Biochem. Physiol. 28: 205-13.
CHAPLIN, S. B., AND P. H. WELLS. 1982. Energy reserves and metabolic ex-
penditures of monarch butterflies overwintering in southern Cali-
fornia. Ecol. Ent. 7: 249-56.
March, 1984
Insect Behavioral Ecology-'83 McNab 73
FRAIR, W., R. G. ACKMAN, AND N. MROSOVSKY. 1972. Body temperatures of
Dermochelys coriacea: warm turtle from cold water. Science 177:
791-93.
HEINRICH, B. 1974. Thermoregulation in endothermic insects. Science 185:
747-56.
HUTCHISON, V. H., H. G. DOWLING, AND A. VINEGAR. 1966. Thermoregula-
tion in a brooding female Indian python, Python molurus bivittatus.
Science 151: 694-96.
KLEIBER, M. 1932. Body size and metabolism. Hilgardia 6: 315-53.
MAY, M. L. 1979. Insect thermoregulation. Ann. Rev. Ent. 24: 313-49.
McNAB, B. K. 1983. Energetics, body size, and the limits to endothermy. J.
Zool., London 199: 1-29.
MORRISON, P. R. 1960. Some interrelations between weight and hibernation
function. Bull. Museum Comp. Zool., Harvard 124: 75-91.
VAN MIEROP, L. H. S., AND S. M. BARNARD. 1978. Further observations on
thermoregulation in the brooding female Python molurus bivittatus
(Serpentes: Boidae). Copeia 1978: 615-21.
Florida Entomologist 67(1)
March, 1984
SCIENTIFIC METHODOLOGY IN ENTOMOLOGY
RANDY THORNHILL*
SYNOPSIS
Considerable research on insects is not directed by evolutionary theory.
Fortunately, this is changing. The staggering diversity of insects will be
rendered intelligible only by the explanatory and predictive power of the
theory of evolution by natural selection. As an evolutionary understanding
of insects increases, important clarifications and extensions of the basic
theory probably will occur.
Entomologists that employ evolutionary theory should consider the basic
nature of the hypothetico-deductive model of science and the four methods
for applying it: lab experiments, field experiments, observational analysis
and the comparative method. Here I outline the general model of science
and the pros and cons of each of the four methods for applying it. I argue
that the Popperian' view of science is useful because it causes scientists
to consider predictions that may lead to elimination of hypotheses. An ex-
treme interpretation of this approach, however, is inappropriate because it
ignores the value of positive evidence for understanding nature. All four
methods of applying the scientific model can lead to precise understanding of
cause and effect in biology, especially when alternative hypotheses with
mutually exclusive predictions are carefully considered. Lab and field experi-
ments and observational analysis can provide answers to questions about
the nature of selection presently acting on traits of organisms. But only
the comparative method can yield answers to questions about evolved function
(selective history) of traits.
There are evolutionists who argue that the comparative method is at
best a method of obtaining correlative patterns that provide tentative hypo-
theses that must be subjected to experimental or observational tests in
order to determine their value. But regardless of the method, the goal is to
discover significant differences between sets of data that allow construction
and refinement of correlations. I argue that scientific knowledge actually ac-
cumulates and science progresses toward understanding simply as a result
of improving correlations between presumed cause and effect.
INTRODUCTION
"For answering questions on function in biology, comparative evidence
is more reliable than mathematical reasoning." (Williams 1975, p. 7)
"Only experiments can truly test theory." (Stearns 1976, p. 42)
"The only valid method by which the adaptive significance of a feature
can be determined is by direct analysis which includes observing the
animal in its natural environment and direct determination of the bio-
logical roles and selection forces." (Bock 1977, p. 79)
"There is no fundamental difference between the comparative method
and the experimental method in biology." (Alexander 1978, p. 95)
"... observation and comparison are methods in biological research that
are fully as scientific and heuristic as the experiment." (Mayr 1982, p.
76)
*Randy Thornhill is an Associate Professor in the Department of Biology at the
University of New Mexico. His research focuses on insect (especially Mecoptera) and
human biology. His book with John Alcock, The Evolution of Insect Mating Systems
(Harvard University Press, 1983) provides a modern analysis of the diversity of mating be-
havior and reproductive tactics in insects. Current Address: Department of Biology, Uni-
versity of New Mexico, Albuquerque, New Mexico 87131.
Insect Behavioral Ecology-'83 Thornhill
What is considered appropriate scientific methodology varies among
areas of science, between disciplines of biology, and even among practition-
ers within a biological discipline. The above quotes by evolutionary biologists
reveal that within this field of biology large differences in opinion exist
about appropriate methodology. Some critics of the comparative method
(e.g., Stearns 1976, Bock 1977, Reznick 1982) argue that experiments repre-
sent the only methodology that can yield an understanding of cause and
effect. This view also is held by some biologists who are unfamiliar with
evolution (e.g., many "laboratory" biologists) and by many physical
scientists. On the other hand, some evolutionists state that the compara-
tive and experimental methods are not fundamentally different in quality
(e.g., Alexander 1978, 1979, Mayr, 1982) or imply this in their work (e.g.,
Darwin 1859, 1874, Williams 1975, Maynard Smith 1978, Alcock 1979,
Clutton-Brock and Harvey 1979, Mayr 1982, 1983).
The view that the comparative method is inferior stems in large part from
lack of recognition that each method has strengths and weaknesses and is
consistent with criteria of the general model of science. I will argue in this
paper that the comparative method is the only method we have for deter-
mination of the selection that has shaped characteristics of organisms (as
opposed to the selection presently acting on traits) and it is as precise
as any other method in science.
The study of the ultimate meaning of living things was brought into
focus by Charles Darwin. Not only did he provide the general theory of life
but he also gave us powerful methodology-comparative analysis-for
examining and clarifying the theory. The profundity of Darwin's contribu-
tions was incompletely understood until recently. Hamilton's classic papers
(1964 I and II) and Williams' book Adaptation and Natural Selection: A
Critique of Current Thinking in Evolution (1966) had a revolutionary im-
pact on biological investigation. They caused investigators in many bio-
logical disciplines to focus their thinking on reproductive competition be-
tween individuals and thus return to a Darwinian framework.
Like molecular biology (Lewin 1981) and certain other fields devoted
to the study of life, entomology has generally lacked a strong evolutionary
foundation. This is changing in molecular biology (e.g., Orgel and Crick
1980, Lewin 1981), and there is indication of such a shift within entomology
as a result of efforts by several individuals, most notably Lloyd (see symposia
volumes 1979-1982), but there is need for a stronger evolutionary basis in
most insect studies. The lack of rigorous hypothesis testing in many areas
of entomological research is a reflection of the absence of the powerful
theoretical basis that evolution provides. Though I direct this paper
primarily toward an entomological audience, the message I am trying to
convey is general and applies to all areas of biology that do not presently
rely on modern evolutionary analysis for research direction.
Undoubtedly, some of my entomological colleagues will reject my view of
the current state of entomology. Such colleagues might point to some ex-
perimentally elegant piece of entomological research done in total absence
of any understanding of evolution. However, the question is not how elegant
research is in terms of control of confounding variables, or how clever the
idea behind the work is, but, does the research yield a better general under-
standing of life? We see over and over again in the history of biology that
specific results do not make any general sense until placed in the framework
Florida Entomologist 67(1)
of evolutionary theory (Dobzhansky 1973, Mayr 1982). An excellent example
is the present attempt to couple molecular studies with evolutionary analysis
(Lewin 1981). As Dobzhansky (1973) put it, "Nothing in biology makes
sense except in the light of evolution."
CAUSATION AND THE GENERAL THEORY OF LIFE
The effects of interest to biologists are the traits or features of organ-
isms. Biologists consider causation of any trait from two perspectives:
proximate and evolutionary. Proximate explanations for the existence of
biological traits deal with genetic, biochemical, physiological, develop-
mental, social, or other immediate causes leading to the expression of the
characteristics. Evolutionary explanations address causes that operated
during evolutionary history to lead to present biological phenomena. The
evolutionary approach focuses on the relationship between biological
characteristics and the selective forces that produced them-that is, the
contribution of traits to differential reproduction of individuals in the en-
vironments of evolutionary history.
Consider the warning coloration of the monarch butterfly. The colors of
adults and juveniles are caused by genetic, biochemical, physiological, and
developmental proximate factors. In terms of ultimate causation, the colors
probably stem from a history of nonrandom differential reproduction of
individuals in the context of avoiding visual predators. The two levels of
explanation are complementary, but as discussed below, they are not
alternatives.
It has been claimed by some that viewing traits in terms of evolutionary
causation solely as the product of selection ignores the roles of agents other
than selection in shaping them. For example, drift and mutation cause
changes in gene frequencies from generation to generation (evolution). But,
relative to selection, mutation and drift are impotent as evolutionary
forces, because they act randomly with regard to fitness and thus are un-
likely to bring about significant cumulative change (see Alexander 1979 for
detailed discussion). Viewing features of life as shaped by selection provides
the best general working hypothesis. Any other view is completely im-
pervious to test, as critics of the selectionist approach admit (Lewontin
1978).
The selectionist approach is referred to as the ultimate approach be-
cause it is selection that accounts for the existence of proximate mechan-
isms. Thus, the two forms of causation are not alternatives in any sense.
This means that there is really only one approach in biology and that it is
the ultimate one. The theory of evolution by selection is not a theory of life,
it is the theory of life. Charles Darwin did not invent the idea of selection;
he discovered the process, as did Alfred R. Wallace independently at the
same time. Selection has acted continuously on all living things throughout
the history of life and continues to do so today. Thus selection is omnipo-
tent. The features of life are what they are because of selection in the past,
and thus all features of all living things are expected ultimately to promote
reproduction or genetic propagation of individuals in evolutionarily relevant
environments. This provides the foundation for scientific study of all life
and tells the biologist how to proceed in order to gain further understanding
of life through experiment, observation, and comparative analysis, regardless
March, 1984
Insect Behavioral Ecology-'83 Thornhill 77
of whether he is interested in molecules, behavior, physiology, morphology,
proximate or evolutionary causation, or beetles or human beings.
This is not to say that study of proximate causation is unimportant.
A complete understanding of any feature of life includes elucidation of both
proximate and ultimate causation. I am saying that the theory of evolution
by selection provides the best direction for investigating proximate causa-
tion. This approach has recently been used in the study of insect phero-
mones. Until recently, pheromone structure was approached only from a
descriptive perspective, but now ideas are being developed regarding chemi-
cal composition of insect pheromones expected on the basis of sexual selection
theory (Marshall 1982, Arnold and Houck 1982, Thornhill and Alcock
1983). Selection theory is being used successfully to predict the actual in-
gredients of insect pheromones (Marshall 1982, in preparation). In the
sense I have outlined, the "evolutionary" in evolutionary biology is re-
dundant. Biology is the scientific study of the evolution of life. No other
definition of biology makes any sense.
THE GENERAL METHOD OF SCIENCE
Science is a very complicated endeavor, and definitions of science are
probably as numerous as scientists. Although it is perhaps a waste of time
to attempt a perfect definition of the enterprise, I like Peter Medawar's
(1967) definition: the art of the soluble. Most scientists that I am ac-
quainted with feel that things are ultimately knowable, and when we do
not know, or only know incompletely, it is because we have not asked the
right questions. The hypothetico-deductive method provides the way for
scientists to ask questions and seek to answer them. The hypothetico-deduc-
tive model consists of the following stages: observation, hypothesis forma-
tion, identification of predictions, and testing predictions. The stages are
interactive and all stages are creative. One can do science by generation
and testing of hypotheses and/or by testing assumptions of hypotheses.
These endeavors include locating errors in the observations, hypotheses, and
tests of others, and for this reason science is often defined in terms of its
repeatability and self-correcting nature (e.g., Simpson 1964).
An observation in the right hands leads to scientific inquiry. Hypotheses
begin with speculation, a hunch about the cause of some effect of interest.
Scientists should speculate. Without imaginative speculation there would
be no hypotheses and thus no direction for seeking understanding (see
Lloyd's discussion of speculation in this volume).
Gould (1978) and Lewontin (1978) have argued that evolutionary hypo-
theses are often ad-hoc (only specific to the trait in question and without
sufficient generality to allow testing). They consider evolutionary hypothesiz-
ing the art of creating just-so stories the same way Rudyard Kipling did in
explaining the leopard's spots, the camel's hump, etc. Ad-hoc arguments do
not represent valid speculation. The initial speculations of the working
biologist are post-hoc, not ad-hoc, and there is nothing wrong with using
post-hoc leads for generating true hypotheses that are general enough for
testing. As Clutton-Brock and Harvey (1979) have pointed out, post-hoc
explanations represent an inevitable first step in any observational science.
The suggestion that monarch butterfly coloration is an anti-predator trait
can be used to generate a real hypothesis, which is testable by observational
Florida Entomologist 67(1)
March, 1984
analysis or experiments dealing with effects of the colors on predator
preferences, or by comparative analysis across species of animals with
and without warning colors.
For a hypothesis (or theory) to be scientific it must be testable. This
means it must be both predictive and empirically falsifiable. The require-
ments for predictions are that they be logically derived from the hypothesis
and be statements about the unknown. It has been argued that evolution-
ary theory is not truly predictive because it focuses on historical causes and
not future events (Peters 1976). But, as numerous people have pointed
out, prediction of future events is not a requirement of a scientific theory.
Prediction of the unknown, whether past, present, or future, is the important
issue.
A hypothesis that predicts everything imaginable is not within the
realm of science because it cannot be falsified. That is, a hypothesis that
explains everything explains nothing. The theory of evolution by selection
erroneously has been called a nonfalsifiable theory because of this (Popper
1934). It is puzzling how Popper and others could read Darwin's work and
conclude that the theory of evolution is nonfalsifiable. Darwin's writing
reveals his care in identifying observations that would falsify his theory
(for discussion see Ghiselin 1969, Alexander 1977). Popper (1934) is
usually given credit for recognizing that in order for an idea to be con-
sidered scientific it must be falsifiable in principle (the criterion of demarca-
tion), but this procedure is apparent throughout Darwin's work and in
the research of other early scientists using hypothetico-deductive analysis
(e.g., Mendel, Newton, Pascal, Pasteur).
One way to look at the value of the criterion of demarcation is in terms
of what logicians call the fallacy of affirming the consequent. Consider a
hypothesis and its derived predictions. Assume that the predictions are
found to be true. Now consider the following argument: if the hypothesis
is true, then the predictions must also be true; the predictions are true,
therefore, the hypothesis is true. According to logicians the conclusion is
not valid even if the hypothesis is correct. If the predictions of a hypothesis
are confirmed, it is logically invalid to conclude that the hypothesis is
correct, because some other hypothesis (es) might yield the same pre-
dictions. Thus, logicians argue that attempts to falsify hypotheses avoid
this fallacy. It is logical to conclude that a hypothesis is false when its pre-
dictions are false.
But the major value of the criterion of demarcation is simply that it
causes scientists to look for predictions that potentially can eliminate a
given hypothesis rather than only evaluating predictions that, regardless
of whether they are true or false, will support the hypothesis or multiple
hypotheses. The importance of falsification of hypotheses to the advance of
a science was eloquently stated by Darwin: ". . false views, if supported
by some evidence, do little harm, for everyone takes a salutary pleasure in
proving their falseness; and when this is done, one path towards error is
closed and the road to truth is often at the same time opened" (C. Darwin
1874, p. 606). Envisioning empirical observations that would disprove
hypotheses is vital for scientific advance. However, extreme applications
of Popperian philosophy and the fallacy of affirming the consequent in
scientific endeavors are inappropriate because they deny the significance
Insect Behavioral Ecology-'83 Thornhill 79
of positive evidence in the achievement of true understanding and promote
the erroneous view that knowledge is an illusion (see below).
The strongest test of any hypothesis involves identification of competing
hypotheses that predict mutually exclusive, empirically falsifiable outcomes.
Such a test in its strongest form supports one hypothesis and falsifies the
alternative hypotheses (see Platt 1964). Another aspect of strong scientific
testing is that predictions be precise, which increases the likelihood of
falsifying the hypothesis generating the predictions. A hypothesis that has
passed many crucial tests involving precise mutually exclusive predictions
from alternative hypotheses can be said to be corroborated (Popper 1934).
The number of tests is not the important factor for evaluating the reliability
of a hypothesis. Insead, it is the number of severe or crucial tests that de-
termines our confidence in a hypothesis. The degree to which hypotheses are
corroborated varies, and science can lead to a degree of corroboration ap-
propriately labelled certainty.
The use of alternative causal hypotheses also avoids the natural tendency
of investigators to become attached to a pet hypothesis. Of course, one can
favor a pet hypothesis by setting up weak alternatives that have no chance
of matching observation. But this will fail ultimately given the self-correct-
ing nature of science.
By alternative hypotheses I mean alternative routes of causation, and
not simply a test of a null hypothesis and its alternative, because only one
causal hypothesis is identified in such a test.
Before outlining an example of the use of alternative hypotheses, I will
address the view that knowledge is an illusion. That is, the notion that,
given the necessity of employing the criterion of demarcation, we can never
know anything with certainty (e.g., Harris 1982), which stems from ex-
treme interpretations of the Popperian philosophy of science. An extreme
Popperian accepts only negative results. With this view only falsified
hypotheses are valuable; positive results do not count and certainly do not
imply understanding, because to accept positive evidence commits the
fallacy of affirming the consequent. This is where the actual practice of
good scientists deviates from a Popperian perspective. Although one cannot
actually prove a hypothesis to be true in the sense of logical proof, science
can lead to proof of a hypothesis in the sense of meaningful understanding
of natural phenomena. For example, the planets actually orbit the sun
and not the earth, and the earth is roughly spherical and not flat. The
hypotheses that the earth is the center of the universe and that the earth
is flat turned out to be totally wrong. The notion of pangenetic inheritance
is completely wrong. We now know that, barring cultural inheritance,
genes (in interaction with environment) are responsible for parent-offspring
correlations in similarity. The list of what we know for sure is long and
growing. Any view of science that pretends we don't know for sure that
insulin is produced in the isles of Langerhans, that bacteria and viruses
can cause disease, that cells are the building blocks of higher organisms,
that natural selection acts incessantly, that chromosomes house genes, etc.
is nonsense.
TESTING ALTERNATIVE HYPOTHESES
In this section I provide an example from my own research of how use-
ful alternative causal hypotheses with mutually exclusive predictions can
Florida Entomologist 67(1)
be. I use my own work because of my intimate familiarity with it. It deals
with the nature of selection currently acting on a morphological feature of
scorpionflies rather than the selection that has produced the structure;
only the latter selection directly addresses the evolved function of a charac-
ter. Later I discuss the distinction between selective maintainance and se-
lective history and the methodologies for studying each of these categories
of selection.
DORSAL CLAMP OF Panorpa: SPERM COMPETITION OR FORCED COPULATION
Sperm competition is competition between ejaculates of two or more
males for the fertilization of eggs of a single female. That sperm competi-
tion can be a potent selective force leading to male morphologies and be-
havior was first discussed by Parker (1970). This now classic paper has led
to many studies of sperm competition as the selective force that has molded
male reproductive characteristics. However, sperm competition is the only
context considered when many investigators study insect characteristics
such as copulation duration, copulatory frequencies, interactions of males
and females during copulation, and post and precopulatory interactions of
females and males. There are alternative hypotheses to explain these
characteristics, and they require examination.
I began work on scorpionflies (Mecoptera) in 1971, the year following
the publication of Parker's seminal paper. The males have behavioral and
morphological features that I initially interpreted as evolved in the context
of reducing sperm competition. As my studies developed it became more
and more difficult to accept this interpretation in all cases. This led to ex-
periments beginning in 1977 designed to analyze alternative explanations of
the traits.
There is a clamp-like structure on the dorsum of the male's abdomen in
scorpionflies of the genus Panorpa. The dorsal clamp is formed from parts
of the dorsum of the male's third and fourth abdominal segments. The
clamp holds the female's wings during mating. Solitary males often attempt
to disrupt copulating pairs and such males are occasionally successful. This
led me initially to the interpretation that the clamp is important in prevent-
ing the female from being usurped and inseminated by an intruder, reducing
the probability of the ejaculate of the usurped male being the fertilizing
ejaculate. This interpretation was in keeping with Parker's (1970) view of
male-grasping morphologies-he saw them as evolved to prevent "take-overs".
(As it turns out, the dorsal clamp appears to be used for forced copulation;
see below).
Male Panorpa exhibit three alternative forms of mating behavior that
are present within the behavioral repertoire of each individual. Two alterna-
tives employed by males to obtain copulations involve nuptial feeding-the
male presents a food item to the female during courtship and the female
feeds on it throughout copulation. (1) A male may secrete a salivary
mass. After saliva secretion, males stand near their salivary mass and
disperse distance sex pheromone. A female attracted by the pheromone
feeds on the saliva. (2) A male may feed a female a dead arthropod. In this
case a male locates a dead arthropod, feeds on it briefly, and then disperses
sex attractant while standing next to it. (3) A male may employ forced
copulation in which a male without a nuptial offering (dead insect or
salivary mass) rushes toward a passing female and lashes out his mobile
March, 1984
Insect Behavioral Ecology-'83 Thornhill
abdomen at her. (Males engaging in forced copulation do not release
pheromone.) If such a male successfully grasps a leg or wing of the female
with his genital claspers, he then attempts to position her to secure the
anterior edge of her right forewing in his dorsal clamp. Then the male
attempts to grasp her genitalia with his genital claspers. The male retains
hold of the female's wing with the dorsal clamp throughout copulation.
Forced copulation in Panorpa is not an abnormal or "aberrant" behavior, but
an aspect of the evolved behavioral repertoire of individual males that is
widespread among species of the genus Panorpa (Thornhill 1980a, 1981,
1984a).
The behavior of females toward males with and without a nuptial offer-
ing is distinctly different. Females flee from males that approach without a
nuptial offering but approach males that have nuptial offerings and behave
"coyly" toward them. Females struggle to escape from the grasp of force-
ful males, but do not resist copulation with resource-providing males.
I have shown in laboratory and field experiments and observations in-
volving several species of Panorpa, that the extent of use of each of the
three behavioral alternatives by males is related to the availability of dead
arthropods, which is determined by absolute abundance of arthropods and
by male-male competition for the arthropods. Individual males prefer to
adopt the three alternatives in the following sequence: dead arthropod >
salivary mass > forced copulation. That is, when males are excluded from
dead arthropods via male-male competition, they secrete saliva if they can
(a male's ability to secrete saliva is determined by his recent history of ob-
taining food), and males only adopt forced copulation when the other two
alternatives cannot be adopted. A male's body size influences his ability in
male-male competition, and large males tend to adopt the use of dead arthro-
pods as nuptial gifts, medium-sized males most frequently use saliva, and
forced copulation is adopted most frequently by small males.
The behavioral alternatives contribute differently to male fitness. The
preference of alternatives employed by males is consistent with female
choice, and thus with male mating success. Females prefer males with
arthropods over males with salivary secretions and actively attempt to
avoid force copulating males. Also, the alternatives appear to be associated
with different male mortality probabilities. Relative to large and medium-
sized males, small males tend to lose in the competition for food, and thus
are forced to feed on dead arthropods in the webs of spiders, which results
in high mortality. Finally, force copulators have relatively low reproductive
success compared to resource-providing males because females lay few eggs
following forced copulation (Thornhill 1980a, 1981, 1984a).
Lab experiments have revealed that the dorsal clamp is essential for
forced but not unforced copulation (Thornhill 1980a, 1984a). In one experi-
ment beeswax was used to cover the dorsal clamps of males that had been
starved, which prevents them from secreting saliva. The dorsal clamp of
starved control males was left functional. Treated and control males at-
tempted copulation with equal frequency, but only control males succeeded.
Treated males tried to reposition females so as to secure their forewing in
the clamp, but the females escaped by struggling. In other experiments males
with saliva or dead crickets were treated with beeswax. In these tests treated
males readily copulated with females. Furthermore, insemination rates for
treated males in unforced copulations was 100%.
Florida Entomologist 67(1)
March, 1984
The experimental results confirm predictions of the hypothesis that the
dorsal clamp is important in the context of increasing the success of forced
copulation attempts when sexual competition forces individuals into this
alternative behavior. Yet, despite the apparent uselessness of the clamp in
unforced copulation, the female's wing is placed in the clamp during both
forced and unforced copulation. Could the dorsal clamp be important solely or
in part in some other context? The experiments only superficially address
this question. They were designed to test predictions of a forced copulation
hypothesis. The predictions that were tested could be consistent with those
stemming from an alternative hypothesis(es) for the role of the dorsal
clamp.
A reasonable alternative was identified earlier. It views the dorsal
clamp as important in the context of sperm competition-as a structure
that prevents disruption of copulating pairs and the insemination of the
female by an intruding male. Aspects of the reproductive behavior of male
Panorpa are consistent with this hypothesis-pair disruptions by intruding
males are not infrequent, disruptions sometimes result in the intruder
copulating with the female, and the clamp is used in both forced and un-
forced copulations. Mutually exclusive predictions from the two alternative
hypotheses are easily identified. If the clamp serves a male's reproduction
by reducing the probability of the takeover of a copulating male's mate, one
would expect treated males (dorsal clamp occluded with beeswax) to ex-
perience higher takeover rates than untreated males. But if the clamp is
used solely for something else (i.e. is important in the context of forced
copulation, or some other context consistent with the predictions examined
earlier from the forced copulation hypothesis), one would expect the treated
and untreated males to experience similar rates of takeover. If the former
prediction (takeover highest when clamp functionless) is supported, the
forced copulation hypothesis would be eliminated as the sole explanation of
the selection presently acting on the dorsal clamp. If the latter prediction
(takeover rate not influenced by clamp) is supported, the takeover hypothesis
would be falsified. The predictions are strong in the sense that they offer
potential for falsification.
The predictions from the two hypotheses were tested with a lab experi-
ment. The results reveal that copulating males whose clamps were covered
with beeswax had the same takeover rate as untreated copulating males
(Thornhill 1984a). Thus, the clamp apparently does not presently aid a
male's reproduction in the context of take-over attempts-i.e. there is no
selection on the clamp in the context of take-over. If the experiment had
shown that treated males experienced significantly higher take-over rates,
the forced copulation hypothesis would remain potentially very important
because of the findings from experiments described earlier.
DORSAL CLAMP OF Panorpa: OTHER HYPOTHESES
There are other alternative hypotheses that can be considered in an
attempt to understand the selective maintenance of the dorsal clamp. Felt
(1895) observed female Panorpa debilis palpating the dorsal clamp of males
during courtship. I have observed palpation of the dorsal clamp during
courtship by females of several Panorpa species. Felt concluded that the
structure probably secretes a volatile oil that attracts the female to the
male and primes her for mating. Felt's hypothesis is incorrect. I have
Insect Behavioral Ecology-'83 Thornhill
examined the histology of the dorsal clamp of several species of Panorpa,
including debilis, and found no associated glandular tissue. The hypothesis
is also inconsistent with experimental results. When the dorsal clamp of re-
source-providing males is covered with beeswax, presumably preventing
any odors from being released, females mate readily with the males.
The dorsal clamp could be of value to males in species recognition. The
dorsal clamp varies in morphology across Panorpa species. The examination
(olfactory and apparently visual) by the female prior to copulation could
reduce the probability of an interspecific mating error. This hypothesis must
be examined because it has been, and still is, a widely used evolutionary
explanation for species differences in courtship and mating behavior and
associated morphological features. Darwin argued that sexual selection
was the most important context for the evolution of sexual differences in
sexual behavior and morphology, but Wallace identified species and sex
recognition as the more likely contexts (see Thornhill 1980b). After Darwin
and until recently, premating and mating behavior and associated morpho-
logical features have been generally viewed as functioning as reproductive
isolating mechanisms-that is, as adaptations that prevent wasted repro-
ductive effort by individuals in heterospecific interactions (Thornhill and
Alcock 1983, West-Eberhard 1983). The theory of sexual selection appears
to have far greater predictive power for understanding the diversity of
these traits (Thornhill and Alcock 1983, West-Eberhard 1983). But the
species' and sex identification hypotheses can serve as alternative hypotheses
in studies of traits presumed to be important in the context of sexual se-
lection or reproductive competition in general.
Although interspecific mating errors could potentially occur because of
the co-occurrence of sexually active adults of several species of Panorpa in
time, evidence indicates that females do not depend on cues from the dorsal
clamp for species (or sex) discrimination. Male Panorpa produce species-
specific sex pheromones that attract females from a distance and may also
serve in close-range interaction (Thornhill 1979). Males exhibit species-
specific courtship actions (wing and body movements). Also, when females
approach a heterospecific male they do so to obtain a meal-i.e., they attempt
to feed on the nuptial offering. Females never behave coyly toward hetero-
specific males as they always do toward conspecifics. Females of large species
attempt to usurp, via aggression, the nuptial offerings of males of small
species, and are sometimes successful. Resource-holding males behave
aggressively toward heterospecific females that approach them. These con-
siderations indicate that species and sex identity are discerned prior to
close-range courtship by both sexes. Finally, the experiments discussed above
are strongly inconsistent with the sex-and species-discrimination
hypotheses. In these experiments the dorsal clamp was covered with opaque
beeswax and thus it would not apparently emit normal visual (or olfactory)
cues. Yet, females readily mated with treated males who offered nuptial
gifts.
At this point in my research on the dorsal clamp of Panorpa, results
suggest that this structure is maintained solely by selection in the context
of forced copulation. The alternative hypotheses I have considered were
either falsified (the sperm competition hypothesis and the pheromone-
emission hypothesis) or inconsistent with existing evidence (the sex and
species identification hypotheses). Only further tests would falsify the sex
Florida Entomologist 67(1)
and species identification hypotheses. The general point I want to emphasize
is that alternative hypotheses should be impartially examined in relation
to all available evidence. In a subsequent section I further consider the
dorsal clamp in relation to comparative evidence that could provide the
best understanding of its evolved function (as opposed to its selective
maintenance).
METHODS OF APPLYING THE HYPOTHETICO-DEDUCTIVE MODEL
The hypothetico-deductive system for ascertaining cause and effect can
be applied in four ways: lab experiments, field experiments, observational
analysis, and comparative analysis. All four methods are equally valid
scientific procedures: all four have strengths and weaknesses and they are
based on different premises. The first three methods primarily yield in-
formation about present selection maintaining a trait of interest (i.e. about
a trait's present contribution to reproduction), but only the comparative
method can provide information about the selective history or evolved
function of a trait.
EXPERIMENTATION
The essence of experimentation is manipulation. The experimental method
involves some systematic variation of a variable of interest. But lab and
field experiments represent very different scientific procedures. Lab experi-
ments typically take the form of attempts to control all variables but one.
Field experiments often do not involve control of variables that may confuse
results; instead, all parameters except the manipulated one are allowed to
vary naturally. Randomization of treatments in field experiments can pro-
duce limited control of confounding variables (e.g., site effects) but always
many variables remain uncontrolled. Typically, one is less certain about the
influence of other variables on the result with field experiments compared to
lab experiments. Even, with lab experiments, however, all potentially con-
fusing variables cannot always be controlled. The number of confounding
variables that could cause a given result is potentially infinite. Also, lab and
and field experiments are very different in that the lab is at best seminatural,
and thus one can never be sure that lab results address nature.
A major problem with the experimental method is that manipulation
effects often confuse results. It is always difficult to determine if one's
result is due to the presumed cause, the manipulated variable, or is an ex-
perimental artifact unrelated to presumed causation. For example, it would
seem from many lab and field experiments that nitrogen content of soil is a
cause of plant growth and health (the effect). However, the correlation
between nitrogen content and plant growth may be spurious (Grover 1982),
because when nitrogen is added other changes occur (e.g., redox potential).
These changes may be the actual cause of improved plant growth (effect).
Other experiments will be necessary to control parameters that confuse the
result.
Another problem with experimentation is the difficulty of determining
the appropriate parameters to manipulate in order to yield biologically rele-
vant information. It may take years of natural history observation on a
biological system before an appropriate experimental procedure can be
identified. For example, the availability of the limited resource (dead
March, 1984
Insect Behavioral Ecology-'83 Thornhill
arthropods) is a major determinant of the behavioral variation among
male Panorpa (see above). Which alternative behavior is adopted by a male
depends on his ability to obtain resources. Resource abundance for a male
is a function of absolute abundance of dead arthropods in the habitat, sex
ratio (number of competing males), and size of conspecific and hetero-
specific male competitors (Thornhill 1981). It took several field seasons to
obtain sufficient natural history information to identify the field and lab
experiments that would provide results relevant to understanding Panorpa
in nature, and it is my experience that biologists are often willing to initiate
experimental work without a proper understanding of the natural history of
the system they are interested in. Until recently experimental psychology
was the epitome of such an approach. Elegant, elaborately controlled lab
experiments were conducted with only repeatability in mind and without
concern for the biological meaning of the results.
OBSERVATIONAL ANALYSIS
By observational analysis I mean the testing of predictions with ob-
servational data in the absence of manipulation or comparative analysis.
The absence of manipulation is one of its strengths, but its weakness is a
lack of rigorous controls. Confounding variables plague observational work.
Some can be eliminated by refining observations via further observational
analysis pertaining to some question about a trait (e.g., see Clutton-Brock
and Harvey 1979 and Skinner and Charnov in this symposium volume
for examples), but many problem variables will remain. Despite this
problem, crucial tests can be conducted via observational analysis (i.e., tests
involving mutually exclusive quantitative predictions from alternative hypo-
theses).
Parker's (1978) work on the dung fly Scatophaga stercororia provides an
example of observational analysis. Initial experimental research on sperm
competition revealed that the longer a male mates the more eggs of his
mate he fertilizes, up to '100 minutes of mating which results in all eggs
fertilized. But the egg gain for the copulating male rapidly diminishes at
about 40 minutes, a duration which yields 80% fertilization. With this
information and an understanding of the average time necessary for a
male to guard his mate from rivals until she lays her eggs, and then begin
searching for another mate, Parker predicted that the optimal copulation
time should be 41 minutes. This prediction is precise and thus easily dis-
proved, and is logically derived from the hypothesis that males behave so as
to maximize the number of eggs fertilized per minute. The actual average
time in the field turned out to be 36 minutes. The prediction about copula-
tion duration might be improved by considering the cost of extended copula-
tion in relation to male size, rather than as an average for all males, because
small vs. large male dung flies have different opportunities for access to
multiple mates (see Borgia 1980).
Although explicit alternative hypotheses were not considered in the
dung fly work they could be in future studies. It is not correct to argue
that the use of alternatives is unnecessary because the prediction is precise
enough to exclude a great many alternative hypotheses. Both the Bohr
theory of the atom and the Schrbdinger theory predict exactly the same
Rydberg constant!
Observational analysis can be conducted in the same fashion as evolution-
Florida Entomologist 67(1)
March, 1984
ary comparative analysis (below). By using observational comparisons
which by their nature and number randomize and thus control confounding
variables, one can arrive at robust conclusions. A good example from every-
day life is the work on the role of seat belts in preventing injury during
automobile accidents (see Alexander 1979, p. 12). Another example is the
research on the role of genetically inherited tendencies in criminal activity.
With regard to the latter, sociologists have realized that with appropriate
comparisons the quality of the result obtained is as good as in a rigorously
controlled lab experiment (see Ellis 1982).
COMPARATIVE ANALYSIS
The comparative method involves species or population comparisons
conducted so as to randomize the influence of confusing variables on the
effect of interest (Alexander 1978, 1979). This method is based on con-
vergent and divergent evolution. The former involves distantly related
forms converging on an adaptation because of similar selection pressures.
The latter pertains to closely related forms diverging in adaptation to some
condition because of different selection pressures.
One real challenge in biology is understanding events of the distant past
without which our understanding of evolutionary history would remain in-
complete. The comparative method is analogous to a time machine, and with
it we can ask what selective force operated to lead to the present expression
of a characteristic of interest?
The comparative method randomizes the influence of confounding
variables on a result in such a way that single presumed causal forces can
be examined, though complete randomization is difficult, perhaps impossible.
The difference between the comparative and experimental methods is not in
controls, both methods include controls, but in the presence or absence of
manipulation. Because manipulation effects can confuse results, the absence
of manipulation in comparative analysis is one of its strengths.
With the appropriate comparisons, the comparative method can lead to
results as precise as those obtained through other methods. Because life is
incredibly diverse it provides vast numbers of appropriate comparisons for
almost any question about the selective history of any biological character-
istic. The ingenuity in use of the comparative method involves recognition
of appropriate comparisons, i.e., those comparisons that because of their
number and diversity are likely to randomize and thus control the influence
of other variables on the result.
None of the methods I've discussed is perfect; all have inherent
problems. Experiments designed to ascertain cause and effect are improved
through time by investigators interested in a given cause-effect relationship.
Experimenters strive for refinement; better controls and manipulations are
created. Likewise, investigators using comparative analysis to elucidate a
given cause-effect relationship make new and better comparisons. It is
incorrect to suggest that a finding derived from the comparative method is
less accurate than a finding derived from the experimental method.
One's confidence in a given finding obtained by either method should depend
only on the quality of controls employed.
The erroneous opinion that only experiment provides reliable results has
led to inappropriate conclusions and research directions in biology. After
Insect Behavioral Ecology-'83 Thornhill
discussion of a few well documented cases of this in the history of biology,
Mayr (1982, p. 856) said, "It would be interesting to go through the history
of science and see how often a misplaced insistence on experiment has
caused research to move into unsuitable directions."
It has been said that the comparative method can provide insights about
general patterns in nature but cannot elucidate cause and effect (see Reznick
1982). That is, the comparative method can generate correlations but is
not a method for examining causation; its results must be tested by other
methods. This view is puzzling. All scientific findings are correlations. I
don't mean that all scientific findings are represented by Spearman's or
Pearson's correlation coefficients. I mean that all scientific knowledge is
based on relationships between variables-presumed cause and effect re-
lationships between variables. Regardless of whether we use experiments,
observational analysis, or the comparative method to determine significance
between sets of observations, and whether we use regression, t-test, etc., we
are examining and attempting to construct and refine correlations.
Examples of Comparative Method
Two studies that should drive home the essence of comparative work
are one dealing with the function of sexual reproduction and another with
sexual dimorphism in vertebrates. Williams' (1975) analysis of the selective
background of sexual reproduction is classic in this regard. Williams
hypothesized that sex is a parental adaptation to the likelihood that offspring
will face changed or unpredictable conditions. He made several predictions
from this hypothesis about the occurrence and nature of sexual and asexual
reproduction in organisms and tested these via the comparative method. One
of the most important predictions he made was: in organisms that employ
both sexual and asexual reproduction, sex should occur in the life cycle
prior to changed or unpredictable conditions. This prediction was met. In
effect he was attempting to randomize the influence of confounding variables
on the timing of sex in life cycles of organisms with both modes of repro-
duction. In this analysis he examined the evolved function or selective history
of sex. His analysis allows us to go back in time and begin to understand
factors influencing differential reproduction of individuals as a result of
variations in sexuality. His analysis is not the final answer, but it provides
a beginning answer to an important question. Present attempts to elucidate
the function of sex appropriately focus on tests of alternative hypotheses
via comparative analysis (Bell 1982).
It has been assumed since Darwin that a positive relationship exists
between adult sexual size dimorphism (with males larger than females)
and degree of polygyny in vertebrate mating systems. Darwin reasoned that
as the degree of polygyny increased, more and more males would be ex-
cluded from reproduction, which in turn would cause selection for male
combative traits, including large body size. Comparative evidence suggested
that the predicted relationship holds for birds and mammals, but no rigorous
comparative analysis had been conducted until recently. The most rigorous
tests have been done by Alexander and colleagues (1979), Clutton-Brock and
Harvey (1977, 1979), and Payne (1983).
The approach is to make comparisons of vertebrates that differ in sexual
dimorphism and breeding systems. If appropriate comparisons can be made
the influence of confounding variables on the result can be controlled.
Florida Entomologist 67(1)
Alexander and colleagues found significant positive relationships between
degree of sexual size dimorphism and degree to which breeding systems
deviate from monogamy toward extreme polygyny in ungulates, primates,
and pinnipeds. Closely related species diverged, and distantly related
species, even across orders, converged in sexual dimorphism in relation to
the extent of sexual competition in evolutionary history.
Clutton-Brock and Harvey analyzed 42 species of primates and found
the same relationship that Alexander and colleagues found for this taxon.
Clutton-Brock and Harvey's work is stronger in the sense that they used
more species, carefully considered which taxonomic level should be used for
analysis (e.g., species, genus, or family), paid careful attention to alterna-
tive hypotheses, and attempted to eliminate allometric effects, but their
analysis is weaker because it did not cut across distantly related taxa,
as did the analysis of Alexander and his coworkers. Payne's work on a
diversity of bird taxa provides a comparative test of alternative hypotheses
for the relationship between mating systems and sexual dimorphism; his
results also support Darwin's hypothesis for sexual dimorphism.
The comparative method has made some sense out of seemingly chaotic
natural variation but there is still unexplained variance in the correlations
that have been discovered. The presence of unexplained variance in a sig-
nificant relationship between two variables does not in itself mean that
the relationship or the prediction behind the relationship are questionable. I
know of no presumed cause and effect relationship in any area of science
in which all the variance is understood. This is why statistical analysis is
used to detect significant differences in results. Regardless of one's pro-
cedure-experimental, observational, or comparative-there are always ex-
ceptions to expected patterns.
Other Applications of the Comparative Method
I have focused my discussion of the comparative method on its value in
studies of evolved function. This is where this method has been most
successful. But Darwin, the inventor of the comparative method, used it to
examine phylogeny, speciation, biogeography, soil formation by earthworms,
and coral reef formation. Darwin even took a stab at community structure
using the comparative method. Ghiselin (1969) and Gould (1982) address
in detail the advances in the scientific study of historical phenomena that
Darwin provided us with. The comparative method should be equally po-
tent for studying all long-term events.
All methods need constant refinement, and it is likely that some con-
siderations useful in analyzing evolved function will not work for bio-
geography or community structure and vice versa. Some of the problems
that ecologists have encountered recently in attempts to apply the compara-
tive method to community structure (Case and Sidell 1983 and references
therein) may stem from not separating community organization into
evolutionary and ecological components. The latter may change rapidly and
lead to lack of fit between comparative predictions and pattern. The
evolutionary component of community organization, perhaps synonymous
with biogeography, is historical and thus subject only to comparative
analysis. Clutton-Brock and Harvey (1977, 1979), Clutton-Brock (1982),
Harvey and Mace (1982), and Jarman (1982) have provided some refine-
March, 1984
Insect Behavioral Ecology-'83 Thornhill
ments for use of the comparative method in study of evolved function
which may be of value in analysis of other historical problems.
SELECTIVE HISTORY VS. SELECTIVE MAINTENANCE
Perhaps the most rigorous approach for examining a hypothesis is the
use of critical tests involving all methods, because each method is based
on different assumptions. When results from all methods point to the
same conclusion, one usually achieves considerable confidence about cause
and effect. But the combination of methods to examine a question must be
done in a fashion that considers the relative potency of methods for as-
certaining information about the role of selection in the history vs. the
current maintenance of a trait.
Both experimental and observational analysis may cause one to reject
a correct hypothesis about evolved function when predictions are not met
because the tests are done in an evolutionarily novel environment. One
may show via experiment or observation that a prediction from a hypothesis
about the evolutionary history of some presumed adaptation is not correct,
but one does not know if the validity of the hypothesis has really been
examined.
Suppose a hypothesis predicts that the reproductive success of males of
a species should be positively correlated with body size (e.g., the sexual
dimorphism hypothesis discussed earlier). When one tests this prediction,
no correlation is found. Thus, it seems that the hypothesis has been falsified.
But is the hypothesis wrong or is the species living in an evolutionarily novel
environment in which body size is not related to sexual access, despite the
correlation between fitness and male size in the evolutionary history of the
species. The novel environment might involve artificially high population
numbers or very low food supply brought about by human activity. Under the
former condition even the strongest and largest males may be unable to
control the activities of 'other males that persist in copulation attempts.
Under evolutionarily abnormally low food levels, large males may be too
weak to fight for females or to intimidate males through display (because
of the greater energetic cost associated with large body size). Sometimes
the evolutionary novelty in the environment is apparent and research can be
modified to take this problem into account (Jarvis 1974). But in most cases
novel circumstances are unknown or incompletely understood.
Although novel environments may allow detection of counter selection
on a trait-e.g., natural selection may reduce body size and sexual selection
increase it-such circumstances present problems for testing functional
hypotheses via experimental and observational methods. Furthermore, there
is the difficulty of measuring fitness in a way that will provide meaningful
answers via short-term and site specific experimental or observational
analysis. On the other hand, observational and experimental results pro-
vide the only means of obtaining information about which selective
forces) is presently favoring a trait or if the trait is being selected
against. This information is of great interest.
But only the comparative method can yield answers to questions about
selection pressures that have led to present features of living things (also
see Curio 1973 and Mayr 1982, pp. 855-6). If a functional hypothesis pre-
dicts a correlation between sexual size dimorphism and breeding system,
Florida Entomologist 67(1)
the more species that fit the assumptions of the hypothesis and also fit the
prediction, the more one has confidence in the hypothesis. The more evi-
dence for predicted convergence and divergence, the more assured the in-
vestigator is that the hypothesis is correct.
Thus the comparative method can corroborate results pertaining to
possible evolved function obtained by other methods, but other methods
cannot question results from comparative analysis. Observational and ex-
perimental studies can be coupled with comparative analysis in order to
determine if a force presently favoring a trait is the same as the selection
that produced it.
For these reasons, when possible in my own work on scorpionflies, I em-
ploy all methods in an attempt to examine major questions about specific
traits (see Thornhill 1981). The experiments and observations I de-
scribed earlier provide understanding of selection maintaining the dorsal
clamp of male Panorpa. I consider the following prediction an important test
for elucidating the evolved function of the dorsal clamp: the frequency
with which males of the various species of Panorpa exhibit forced copulation
should be positively related to the size of the clamp (or other morphological
correlates of the effectiveness of the structure as a clamp). The alternative
hypotheses pertaining to the dorsal clamp would not generate this prediction,
but they would yield other mutually exclusive ones that could be tested
against comparative data. Panorpa spp. and species in related genera exhibit
considerable diversity in shape and size of the dorsal clamp (Byers and
Thornhill 1983), but too little information is presently available for ap-
propriate comparative work.
It is argued that the nature of selection that has shaped a trait can
never be ascertained for sure because multiple selection pressures probably
operated on the trait and in the same direction. This view is implicit in any
attempt to rank selection pressures in importance in studies of the evolved
function of biological characteristics. This argument is invalid. If two se-
lection pressures are acting in the same direction, and one is stronger than
the other, the degree of evolutionary modification of a trait will reflect
the stronger of the two, because the organism cannot "feel" the weaker
pressure. This holds regardless of number of selective pressures. It is highly
unlikely that two selection pressures acting on a trait will be equally potent,
but even if this does occur the degree of adaptive modification will reflect
the strength of any one of them alone. See Curio (1973, p. 1046) for a full
discussion of this. Any trait has many effects, some beneficial and others
harmful to fitness, but only one fitness effect is its function-its reason for
being (Williams 1966). This is why it is appropriate for biologists to
attempt to locate a single evolved function for each trait they are inter-
ested in. Ecological phenomena, including selective maintenance of a trait,
may have multiple causes, but a question about evolved function has only
one causal answer.
SCIENTIFIC PROGRESS: IMPROVING CORRELATIONS
I think it is safe to say that there is no satisfactory account of how
science progresses or accumulates knowledge. Some argue that science does
not progress; that any impression of progress in understanding is an il-
lusion. Others argue not only that science doesn't progress but that it
changes in an arbitrary way. This stems from the view that, at any time,
March, 1984
Insect Behavioral Ecology-'83 Thornhill
the theories being examined in a science change like automobile styles.
There is no right or wrong theory, only popular and unpopular theories. And
some feel that the major determinate of popularity of an idea is cultural
attitude. For example, when football and other sports are popular and
during World Wars ecologists are expected to believe that competition is
important as an organizer of community structure! Undoubtedly, our be-
liefs, which are culturally inherited from parents, friends, etc., influence
how we view the world scientifically. This is especially apparent in the
writing of scientists who adhere to Marxist philosophies, but is also present
in more subtle form in the work of anyone who adheres to any strong
ideology. But as Ruse (1982) has pointed out, ideas in science that stem
from ideology, cryptic or blatant, are not necessarily bad, because they are
subject to the same criterion as other ideas. When ideological perspectives
are paraded as science they will eventually meet the criterion of demarca-
tion and be eliminated if they are wrong.
I suggest that scientific progress ultimately has nothing to do with
arbitrary popularity of ideas, but instead, advances happen via improved
understanding of the relationship between variables-that is, by increased
understanding of correlations between presumed cause and effect. Every
scientific hypothesis (or theory) describes an expected correlation between
a causes) and an effect. To elucidate the variance in such relationships via
the hypothetico-deductive model is the occupation of scientists. In no case is
all the variance understood, even for simple relationships. I'll take a
relatively simple example from my own work to illustrate my view of
scientific progress and the achievement of greater understanding in general.
I began work on female choice in the scorpionfly Hylobittacus apicalis in
1971. H. apicalis exhibits nuptial feeding: the male feeds a female a prey
arthropod during courtship and throughout copulation. The sizes of prey
carried by males vary. Initial observations led to the hypothesis, based on
evolutionary theory, that females will value material and genetic benefits
of males in mate choice. I.assumed that the material benefits (nuptial prey
size) and genetic benefits (offspring quality) a male could deliver to a
female would be positively correlated. I examined two qualitative predictions
from this hypothesis: 1) females will sometimes refuse to mate with males
with small prey, and 2) the duration of mating will be positively related to
nuptial prey size. The general correlation I sought to define and explain
was choice behavior of females (effect) in relation to the prey size males
possess (presumed cause).
Initial research revealed that the predictions were upheld (Thornhill,
1976, 1977). Many females refused to copulate with males with small prey
(i.e., less than 16 mm2 in surface area) and there was a positive relation-
ship between mating duration and nuptial prey size. I also discovered that
copulations involving large prey (- 16 mm2) are terminated by males,
whereas copulations involving small prey (when a female allows copulation)
are terminated by the female. Furthermore, by interrupting lab copulations
involving virgin females I found that in the first 5 minutes of copulation
few or no sperm are transferred to the female. From 5 to 20 minutes
there is a direct positive relationship between number of sperm transferred
and mating duration, and beyond 20 minutes of mating no further sperm
are transferred. Finally, studies revealed that females lay eggs and do
not mate again following matings with males with large prey, but females
Florida Entomologist 67(1)
remain sexually receptive and do not lay eggs when they mate with a
male with small prey.
The relationship between copulation duration and sperm transfer yielded
the prediction that when females mate with males with small prey they will
terminate copulation after 5 minutes; that is, at about the time when sperm
begins to flow. This way a female obtains a brief meal but receives few or
no sperm from inferior hunters. Field observations revealed that mean
time of mating involving small prey was 5.8 minutes (Thornhill 1980c).
Although this is close to the predicted 5 minutes there was still much vari-
ance in mating times when small prey were involved. But I was making pro-
gress in terms of understanding the initial relationship of interest to me. I
now knew that 50% of the time females reject males with small prey and
when females allow such males to mate they terminate mating on average
at about the time sperm begins to flow from male to female. The correlation
between female mate choice and male nuptial prey size was being improved.
At this point in the study I modified the hypothesis in an attempt to
understand more variance in the relationship, but the modification did not
change the general correlation I was studying. The modified hypothesis is:
The relative value of material and genetic benefits for females in choice
decisions depends on female conditions of body size, feeding history and
mate availability. Basically, I was proposing that every female will strive
to maximize material and genetic benefits received from males. But I need
to clarify how I derived the modified hypothesis. Energetic cost of body
maintenance is positively related to body size; the larger the animal the
more nutrients required. Also, feeding history of females was expected to
vary and thus some females should be more willing to mate with males
with small prey. Mate availability should also influence female choice; the
more males available the more choosy females might be.
The major predictions the modified hypothesis yielded were: 1) large fe-
males will be more likely to mate with males with small prey than will
small females, 2) there will be a positive relationship between female body
size and duration of matings involving small prey, 3) large females will
behave like small females when fed prior to placing them with males
possessing small prey, and 4) independent of female body size, females will
become more choosy as potential mate availability increases.
I have begun testing these predictions. All four are supported. The first
three are well substantiated, but the fourth will need more testing in order
to clarify completely the role of male density in female mate choice (Thorn-
hill 1984b). Thus, I have considerable understanding of the variance in
the correlation between female mate choice and mate nuptial prey size. I
do not have all the variation explained; there are still exceptions. But if I
know a female's body size, her recent feeding history and the mate
availability I can predict with considerable accuracy not only whether the
female will mate with a male with small prey but also the duration of the
mating if it occurs.
I emphasize that the the sequence my work followed is distinctly different
from endeavors involving the addition of parameters that explain more
and more variance in a data set. A common procedure in areas of social
science and biology is to employ multiple regression analysis to determine
the ability of presumably important parameters to explain variation. In
some cases, parameters are added until most or all of the variance is ac-
March, 1984
Insect Behavioral Ecology-'83 Thornhill
counted for. This procedure involves ad-hoc explanation. It is not valid
science to modify a hypothesis to account for unpredicted observations and
then claim that the hypothesis is confirmed. Likewise, it is invalid to
construct a model from data and claim that the model is confirmed. It is
scientifically accurate, however, to use unpredicted observations to modify
or eliminate a hypothesis and then test the predictions from the modified
or alternative hypothesis with new observations (or simply suggest the
direction that testing should involve).
I feel that the sequence of events the Hylobittacus work went through is
the appropriate and typical sequence for scientific hypotheses and even
general theories when they are successful. (See Skinner and Charnov's
paper in this symposium volume for an additional example of this sequence.)
Hypotheses other than the one outlined that I considered in my research
sequence with Hylobittacus are discussed in Thornhill (1980c, 1984b). A
particular sequence may involve only one or all methods of applying the
hypothetico-deductive model. Also, a particular sequence may involve one
or multiple investigators examining the same presumed cause-effect re-
lationship. At any stage in a sequence a hypothesis (or theory) may be
disproved and replaced by another hypothesis (or the original hypothesis
modified) that attempts to account for the relationship.
Even great theories like those of Darwin, Einstein, and Newton portray a
relationship or correlation between variables. Darwin's theory is by far the
most comprehensive theory in science in the sense that it is directed at ex-
plaining life, the most complex and diverse phenomenon known to human-
kind. But still even this theory rests on the relationship between the di-
versity of life (effect) and a history of differential reproduction of in-
dividuals (cause). Since Darwin, biologists have been attempting to under-
stand the variance in this relationship. Biologists ask questions about the
relation between imagined selection pressures and diversity in sexual di-
morphism, life history, chromosome structure and number, mating be-
havior, etc. The imagined selection pressures of biologists serve as alterna-
tive hypotheses which succeed or fail to explain subrelationships of the
general correlation Darwin's theory generated.
FOOTNOTE
'Karl Popper (1934), a philosopher of science, has argued that only falsifiable ideas are
within the realm of science, and that tests of scientific hypotheses (or theories) should
focus on attempts to falsify them. His ideas have been very influential in many areas of
science, including biology.
ACKNOWLEDGEMENTS
I thank John Alcock, Bill Kuipers, Jim Lloyd, Larry Marshall, John
Sivinski, Nancy Thornhill, John Wiens, and Bruce Woodward for construc-
tive criticisms on the manuscript. The Population Biology and Psychobiology
Programs of the National Science Foundation have generously supported
the research from which this paper is derived (grants BNS-7912208, DEB-
7910293, BSR-8219810).
LITERATURE CITED
ALCOCK, J. 1979. Animal behavior: an evolutionary approach, 2nd ed.
Sinauer and Assoc., Sunderland, Massachusetts.
Florida Entomologist 67(1)
March, 1984
ALEXANDER, R. D. 1977. The changing scenes in the natural sciences, 1776-
1976. Philadelphia Acad. Nat. Sci. Special Pub. 12, pp. 283-337.
1978. Evolution, creation, and biology teaching. American Biol.
Teacher 40: 91-107.
1979. Darwinism and human affairs. Univ. of Washington Press,
Seattle, Wash.
J. L. HOOGLAND, R. D. HOWARD, K. M. NOONAN, AND P. W. SHERMAN.
1979. Sexual dimorphism and breeding systems in pinnipeds, ungu-
lates, primates, and humans. Pages 402-35 In N. A. Chagnon and
W. Irons, Eds. Evolutionary biology and human social behavior: an
anthropological perspective. Duxbury Press, North Scituate, Mass.
ARNOLD, S. J., AND L. HOUCK. 1982. Courtship pheromones: evolution by
natural and sexual selection. Pages 173-211 In M. Nitecki, Ed. Bio-
chemical aspects of evolutionary biology. Univ. Chicago Press,
Chicago.
BELL, G. 1982. The masterpiece of nature: the evolution and genetics of
sexuality. Univ. Calif. Press, Berkeley.
BOCK, W. J. 1977. Adaptation and the comparative method. Pages 57-82
In M. K. Hecht, P. C. Goody, and B. M. Hecht, Eds. Major patterns
in vertebrate evolution. Plenum Press, New York.
BORGIA, G. 1980. Sexual competition in Scatophaga stercoraria: size- and
density-related changes in male ability to capture females. Behaviour
75: 185-206.
BYERS, G. W., AND R. THORNHILL. 1983. Biology of the Mecoptera. Annu.
Rev. Ent. 28: 203-28.
CASE, T. J., AND R. SIDELL. Pattern and chance in the structure of model
and natural communities. Evolution 37: 832-849.
CLUTTON-BROCK, T. H. 1982. The problems of comparisons. Pages 319-22.
In King's College Sociobiology Group, Eds. Current problems in
sociobiology. Cambridge University Press, Cambridge.
AND P. H. HARVEY. 1977. Primate ecology and social organization.
J. Zool. 183: 1-39.
AND P. H. HARVEY. 1979. Comparison and adaptation. Proc. R. Soc.
Lond. B 205: 547-67.
CURIO, E. 1973. Towards a methodology of teleonomy. Experientia 29:
1045-58.
DARWIN, C. 1859. On the origin of species. Facsimile of the first edition,
1964. Harvard Univ. Press, Cambridge, Mass.
1874. The descent of man and selection in relation to sex, 2nd ed.
A. L. Bunt Co., N. Y.
DOBZHANSKY, TH. 1973. Nothing in biology makes sense except in the
light of evolution. American Biol. Teacher 35: 125-9.
ELLIS, L. 1982. Genetics and criminal behavior. Criminology 20: 43-66.
FELT, E. P. 1895. The scorpionflies. In Tenth Report of the State Entom-
ologist on the Injurious and Other Insects of the State of New York.
GHISELIN, M. T. 1969. The Triumph of the Darwinism Method. Univ. of
California Press, Berkeley.
GOULD, S. J. v978. Sociobiology: the art of story telling. New Scientist
80: 530-3.
1982. The importance of trifles. Natural History: 1 (April) 16-23.
GROVER, H. 1982. Analyses of the resistant properties of forest ecosystems
in response to nitrogen enrichment. Univ. New Mexico Dissertation.
HAMILTON, W. D. 1964. The genetical evolution of social behavior, Parts I
and II. J. Theor. Biol. 7: 1-52.
HARRIS, C. L. 1982. Evolution: genesis and revelations. State Univ. of
New York Press, Syracuse, New York.
|
RSS
