ECHOLOCATION IN BATS: SIGNAL PROCESSING OF ECHOES FOR TARGET RANGE

• Document Type: CREST
• Collection: General CIA Records
• Document Number (FOIA) /ESDN (CREST): CIA-RDP88R00729R000200030017-6
• Release Decision: RIFPUB
• Original Classification: K
• Document Page Count: 4
• Document Creation Date: December 12, 2016
• Document Release Date: March 25, 2002
• Sequence Number: 17
• Publication Date: March 5, 1971
• Content Type: MAGAZINE

ations (spurious or
resence o; the illuminated disk. If
dventitio'us reward were effective dur-
ould continue to be effective in sus-
aining a high rate of pecking in the
As a further check on the importance
lumination with feeding, we studied
ve new birds on the nondifferential
ondition. After 14 days of nondif-
rential exposure to grain presentation,
total of ten pecks had been recorded
r all five birds together. All of these
cur-red during the intertrial interval
hen the key was not illuminated. Ap-
the first experiment was not an arti-
ct of changing the procedures, nor
fference in reinforcement density. Ac-
isition, as well as maintenance of
eking, is dependent on a differential
sociation of key and reinforcer.
the original differential procedure,
eventually began pecking the disk.
en after 35 days of exposure, bow-
er, the mean rate of response was
ly 20 per minute, and there was no
se birds and those of the first group,
ose mean terminal rate was 101 re-
cement, even after successful acqui-
rtant aspects of the autoshaping phe-
h the assumption that classical con-
ioning is a fundamental factor in the
pecific signaling relationship is not
intenance of behavior. Second, the
essity for differential pairing in
intenance, as well as acquisition, in-
ates that informational properties of
stimulus, rather than its mere asso-
the phenomenon. Third, the phe-
enon, although obviously suscepti-
to analysis by principles of classical
he pecking engendered by autoshap-
is directed to a significant part of
environment-that is, a part cor-
ted with the opportunity to eat. The
ew environ-
co lied without the involvement of ments for to any "shaping" effect by
re and punishment. The findings rewar nd punishments.
of the Brelands (5) in a number of ELKAN GAMZU
nonavian species suggest that such DAVID R. WILLIAMS
mechanisms are not peculiar to pigeons. Department of Psychology,
Although study of the way in which University of Pennsylvania,
complex activities are developed and Philadelphia 19104
learned has largely excluded effects
other than those of reward and punish-
ment, it now seems necessary to include
some other factors as well, if the prin-
ciples of adaptive learning are to pro-
vide an adequate account of the devel-
opment and maintenance of effective
but often nonarbitrary behavior. It is
apparent that animals do not select be-
haviors randomly' from their repertoire
in new situations. The manifestation of
References and Notes
1. P. L. Brown and H. M. Jenkins, J. Exp.
Anal. Behav. 11, 1 (1968).
2. D. It. Williams and H. \Villiams, ibid. 12,
511 (1969).
3. R. A. Rescorla, Psychol. Rev. 74, 71 (1967).
4. -, 1. Comp. Physiol. Psychol. 66, 1 (1968);
Psychonorn. Sc!. 4, 383 (1966).
S. K. Breland and M. Breland, Amer. Psychol.
16, 681 (1961).
6. This work was supported by grant G14055 from
the National Science Foundation. T. Allaway,
B. Schwartz, A. Silberberg, H. Williams, and
K. Zonana contributed substantially to our de-
velopment of this approach.
Echolocation in Bats: Signal Processing of Echoes
for Target Range
Bats of the suborder Microchiroptera
use a type of active sonar for orienta-
tion (1). Biologists, psychologists, and
physicists' have speculated often about
the kinds of information available to
the bat from echoes and about the
nature of the mechanism which proc-
esses the echoes from targets in the
bat's environment. The possibility of
depth perception or target ranging by
echolocation has received particular
attention (2-5). The ease with which
bats detect and avoid obstacles and
detect, track, and capture airborne
targets seems to require some means
of determining the distance to tar-
gets (6).
Three specimens of the North Amer-
ican insectivorous bat, Eptesicus fuscus,
and three specimens of the neotropical,
carnivorous and frugivorous bat, Phyl-
lostomus hastatus, learned to discrimi-
nate target range in the experiment
reported here. The bats were blinded
(enucleated) several months prior to
the experiment to eliminate the pos-
sible use of vision, since the experi-
ment could not be conducted in dark-
ness. Each bat learned to fly from a
Approved For Release 2002/06/19 : CIA-RDP88R0072
Abstract. Echolocating bats Eptesicus fuscus and Phyllostomus hastatus can
discriminate between the nearer and farther of two targets. Their errors in dis-
crimination are predicted accurately by the autocorrelation functions of their
sonar cries. These bats behave as though they have an ideal sonar system which
cross correlates the transmitted cry with the returning echo to extract target-
range information.
small, elevated platform to the closer
of two other platforms (Fig. 1). A
triangular target 10.0 cm wide and
5.0 cm high was mounted at the back
of each of the two landing platforms.
The platforms were separated by an
angle of 40? when viewed from the
bat's position on the starting platform.
The landing platforms differed from
each other in the distance from the
bat on the starting platform to the
target. The bat learned to fly to the
nearer platform in a straightforward
simultaneous discrimination procedure
with food as reward (a piece of a
mealworm offered in forceps) and
without correction of error trials. To
make training easier, each bat was
deprived of some of its normal food
intake until it reached 90 to 95 per-
cent of its weight when captured. The
closer platform alternated left to right
according to a pseudo-random sched-
ule (7).
At first the nearer target ap-
peared at a distance of 50 cm. The
farther target was 60 cm from the
bat throughout the experiment. After
the bat reached a criterion of better
 Approve
or Release 2002/06/19 : CIA-RDP88R00729R000200030017-6
=The cries were frequency-modulated
Fig. 1. The apparatus for target-range dis-
crimination by echolocating bats, showing
the starting platform and the two landing
platforms with the targets.
than 95 percent correct responding for
three consecutive days at 25 trials per
day, the nearer target was presented in a
series of new positions, a new one
every other day.' The bat responded
for 50 trials on each position of the
nearer target. Changes were preceded
by five warm-up trials at the position
of the previous day. The nearer target
appeared at 51 cm, then 52 cm, then
53 cm, and so forth until both targets
appeared at the same distance, 60 cm.
On succeeding days the difference in
target range was 10, 9, 8, 7, 6, 5, 4,
3, 2, 1.5, 1, 0.5, and finally 0 cm.
After the trials on each of the last
four differences in the series, the bat
was tested at a difference of 3 cm to
determine whether exposure to dif-
ficult discriminations had impaired the
bat's responding.
The bats readily learned the flying
response, reached the initial criterion,
and proceeded through the discrimina-
tion series without difficulty. None of
the bats suffered in performance on
the 3-cm difference after the trials on
more difficult discriminations. During
each trial, the bat scanned both targets
with its sonar and then flew directly
to the left or the right platform.
Eptesicus can discriminate a range
difference of about 13 mm 75 percent
correctly, and Phyllostotnus can dis-
criminate 12 mm (Figs. 2 and 3). At
an 'absolute distance of 60 cm, these
two species have an acuity of about
2 percent in discriminating target
range. The choice of 75 percent correct
is an arbitrary criterion for discrimina-
tion.
Condenser
microphones were
(FM) and swept from about 50 khz
to about 25 khz with very little sec-
ond harmonic energy present. Phyl-
lostomus used 0.5- to 1.0-msec cries
with an amplitude of 2 to 15 dyne/cm2.
These cries were also FM with har-
monically related sweeps covering a
range of frequencies from around
65 to about 30 khz. The echoes. re-
turned to Eptesicus by the target at
60 cm were in the vicinity of 0.3 dyne/
cm2. The echoes for Phyllostomus
were near 0.05 dyne/cm2. They were
certainly audible to the bats (8, 9).
Since the bats were without vision, the
discriminations were mediated by so-
nar.
The nearer target subtended a larger
angle in the bat's sonar field, and it
produced a slightly stronger echo than
the farther target due to the smaller
attenuation factor for a shorter air
path traveled by the echo. Although
such artifacts might be discriminable
to the bat for range differences of
several centimeters or more, it seems
unlikely that they would be useful for
a range difference of only 12 to 13
mm. The minimum discriminable size
difference for targets similar to those
used here is 17 percent of the area
of the larger triangle for Eptesicus
(10). Such a size difference would re-
quire a range difference of over 30
mm, so it does not appear that the
"apparent size" difference between the
nearer and farther targets influenced
the data on distance discrimination.
The difference in arrival time of
echoes from the nearer and farther
targets probably carried the informa-
tion about target range. The outgoing
and returning time difference is 70 to
75 ?sec for range differences of 12
to 13 mm.
All of the information potentially
available in a returning echo about
echo arrival time, and hence target
range, is contained in the cross-correla-
tion function of the transmitted and
received sonar signals (11-13). The
inputs to a sonar receiver are signals
separated by some difference in arrival
time associated with the target's range.
The input also includes noise. In the
laboratory the ambient noise in the
bat's frequency band was below the
measuring limits of the available equip-
ment, so the environmental signal-to-
noise ratios for the echoes reflected
back to Eptesicus and Phyllostotnus
exceeded + 30 to + 40 db. Under
mounted on the landing platforms and
on the starting platform to monitor
the bat's scanning during each trial and
to obtain good recordings of the cries
for analysis. Eptesicus emitted 1.0- to
2.5-msec cries with a peak sound pres-
sure of about 50 to 100 dyne/cm2.
such favorable noise conditions the sonar cry (theoretical curve).
?0
90
C~ - E. fuscus
60 cm
ernpiricat a?? --?
theoretical O--O
0 2 4 6 8
Difference in distance to targets (em)
Fig. 2. The average performance of three
Epiesicus in discriminating differences of
target range (empirical curve), and the per-
formance predicted from the autocorrela-
tion function of the Eptesicus sonar cry
(theoretical curve).
output of an ideal sonar . receiver is
essentially the cross-correlation func-
tion of the input signals (13). In prac
tical situations the envelope of the
cross-correlation function is usually
taken to represent the ambiguity en-
countered by an ideal receiver in esti-
mating target range from the signals
appearing at the input (12, 13). In
the absence of target motion and con-
sequent Doppler shift of the echo, the
autocorrelation function of the sonar
transmission is a satisfactory approxi-
mation of the desired cross-correlation-
function (12)..
Autocorrelation functions for the
cries of Eptesicus and Phyllostomus
were derived from the recordings made
during discrimination trials (14). As-
sume that the rate at which the bat
makes errors in judging target distance
or echo arrival time is more or less
directly related, for any given time
separation, to the magnitude of cor-
relation between signals as represented
by the envelope of the autocorrelation
0 2 4 6 8 10
Difference in distance to targets (cm)
Fig. 3. The average performance of three
Phyllostomus in discriminating differences
of target range (empirical curve), and the
performance predicted from the autocor-
`. @ 60 cm.
empirica! o-.>e
 roved Fotr. Release 2002/06/19 : CIA-RDP88R00729R000200030017-6
function (Fig. 4).gh correla loner
would lead to poor discrimination, and
low correlation would result in good
discrimination. At zero time separation,
when correlation is highest and there
-is in fact no difference in target range
to be discriminated, the bat would
be unable to perform beyond chance
levels (50 percent correct responses).
The "theoretical" curves in Figs. 2
and 3 show the distance discriminations
that would be predicted from the en-
velopes of the autocorrelation func-
tions of the cries of Eptesicus and
Phyllostoinus if one assumes a rough
correspondence between signal corre-
lation and number of errors made by
the bat (15). Both Eptesicus and Phyl-
lostomus perform as though they used
ideal sonar systems operating on the
echoes of their respective cries with
a cross-correlation receiver.
It has been proposed that bats de-
rive target information from the enve-
lopes of the outgoing cries and re-
turning echoes or from a perceived
pitch associated with the time interval
separating trains of cries and echoes
(5). In such cases much of the in-
formation carried in the waveforms of
the individual- signals would be dis-
carded. The results of the distance-
discrimination experiment indicate that
most of the information in the signals
is actually preserved and processed by
the bat, and that target ranging is de-
pendent upon the detailed frequency
structure of the echo rather than upon
echo envelopes or trains of cries and
echoes. Electrophysiological observa-
tions on evoked potentials and single-
unit responses in the bat's auditory
system establish that precise analysis
of individual echo signals is possible
(2, 3, 9, 16).
One form of cross-correlation proc-
essing, pulse compression, has been
suggested for target ranging by bats
(4). The basilar membrane of the
bat's cochlea does not, however, act
as a dispersive delay line in a manner
suitable for pulse compression (4, 11).
Cross correlation of transmitted and
received signals may take place at
some point in the bat's auditory brain
without requiring a delay line at the
cochlea serving as a filter matched
to the bat's cries (2, 3, 9, 11, 16).
The most probable location for the
neural cross correlator is in the inferior
colliculus, an auditory center much
enlarged in the bat's brain. The acuity
of obstacle avoidance by echolocation
in Myotis is relatively unaffected by
S MARCH 1971
C(T)
i t t
0 0.5 1.0
msec
Fig. 4. The autocorrelation function of an Eptesicus sonar cry for time-delay differences
up to 200 ssec, showing the upper envelope of the function. (Inset) The oscilloscope
trace of the cry itself. The envelope of the autocorrelation function represents the am-
biguity in echo arrival time (target range) confronting the bat when it processes the
echoes of its cries.
bilateral ablation of the auditory cortex
or by unilateral damage to the inferior
colliculus, whereas bilateral ablation
of the ventral portion of the inferior
colliculus severely impairs echolocation
(17). Single units in the cochlear nu-
cleus show little of the sophistication
of units in the inferior colliculus for
the analysis of bat-like, FM signals.
The response properties of neurons in
the inferior colliculus suggest that
these units function as a neural "tem-
plate" mechanism for the processing
of cries and echoes (3, 9, 16).
The target-ranging performance re-
ported here is direct behavioral evi-
dence that the bat processes individual
echo signals in detail for the arrival-
time information they contain. Ap-
parently the bat possesses a sonar re-
ceiver with ideal properties. The bat
effectively stores the outgoing sonar
signal in the main nucleus of the in-
ferior colliculus. The storage mecha-
nism probably involves the complex
response characteristics of neurons in
the inferior colliculus for analyzing
FM signals. The returning echo also
enters the inferior colliculus where it
undergoes cross correlation with the
stored replica of the sonar transmis-
sion. Neurons in the bat's auditory sys-
tem are selectively sensitive to echo.
like sounds (2, 3), and the cross-corre.
lation operation probably involves such
sensitivity. The existence of matched-
filter properties as suggested by elec-
trophysiological studies of the bat's
auditory brain is supported by the
range-discrimination data.
At present this cross-correlation model
is applicable to echo processing by
bats that use short-duration, FM sonar
cries. The model should prove useful
in accounting for the well-known re-
sistance of bats to confusion from
multiple-target clutter interference and
in explaining their remarkable pro-
ficiency at target identification, track-
ing, and interception (6). If the alter-
nate expression for the correlation
function, the power spectrum, is also
available in the bat's brain, then the
bat could distinguish many target
characteristics from their echo signa-
tures. Bats that use long-duration, con-
stant-frequency signals with a short,
terminal FM sweep (Rhinolophus and
Chilonycteris, for example) may use
cross-correlation' processing on the last
few milliseconds of their echoes.
JAMES A. SIMMONS
Auditory Research Laboratories,
Princeton University,
Princeton, New Jersey 08540
References and Notes
I. E. Ajrapetjantz and A. I. Konstarcinov,
Echolocation in Nature (Soviet Academy of
Sciences Press, Leningrad, 1970); D. R.
Griffin, Listening in the Dark (Yale Univ.
Press, New Haven, Conn., 1954).
2. J. H. Friend, N. Suga, R. A. Suthe^s, J.
Cell. Physiol. 67, 319 (1966); A. D. Grin-
nell, J. Physiol. 167, 67 (1963); O. W. Hen-
son, Jr., in Anintal Sonar Systems, R. G.
Busnel, Ed. (Laboratoire de Physiologic
Acoustique, Jouy-en-Josas-78, France, 1967),
vol. 2, p. 949.
3. N. Suga. 1. Physiol. 175, 50 (1964); ibid. 179,
26 (1965).
Approved For Release 2002/06/19 : CIA-RDP88R00729R000200030017-6 927
 4. W. A. van Bergcijk, J. Acoust. Soc. A q7190
36, 594 (1964); l~pproal dl ot- t sI s %-RDPHR007291&200030017-6
McCue F. A. Wel5ster, Nature 201,
(1964): J. J. G. McCue, J. Acoust. Soc. Amer. The usual hypothesis given for the compared
40, 545 (1966); G. K. Strother, ibid. 33, 696 etiology of goiter in man is that the
(1961).
S. L. Kay, Nature 190, 361 (1961); Antm. Behav enlargement of the thyroid gland is a
10, 34 (1962); M. A. Mogus, AD-650476,
Ordnance Research Lab., Penn. State Univ. consequence of dietary iodine defi-
(1967): J. Nordmark, Nature 188, 1009 (1960): ciency. Studies of iodine intake of goi-
ibid. 190, 363 (1961); J. D. Pye, J. Laryngol.
Otol. 74, 718 (1960); Endeavour 20, 101 trous and nongoitrous persons living in
(1961); Nature 190, 362 (1961); J. L. Stewart, the same environment have not shown
AMRL-TR-68-40, U.S.A.F. Systems Command
(1968). significant differences (1). The hypoth-
6. D. R. Griffin and R. Galambos, J. Exp. Zoo?. esis that goiter is caused by, or is
86, 481 (1941); D. R. Griffin, J. H. Friend,
F. A. Webster, ibid. 158, 155 (1965); D. R. associated with, infection has not been
Griffin and A. Novick, ibid. 130, 251 (1955):
D. R. Griffin, F. A. Webster, C. R. Mi rejected nor has it been adequately
D. Anim. Behav. 8, 141 (1960): A. D. tested. Endemic goiter occurs, in gen-
Grinnell and D. R, Griffin, Biol. Bull. 114,
10 (1958); R. A. Grummon and A. Novick. eral, among populations living in rural
Physiol. Zool. 36, 361 (1963); A. 1. Konstanti- areas and belonging to lower socio-
nov, B. V. Sokolov, I. M. Stosman, Dokl.
Akad. Nauk SSSR 175, 1418 (1967): A. No- economic groups. Several studies have
vick, Ergebn. Biol. 26, 21 (1963): Biol. Brill. shown that the drinking water of such
128. 297 (1965); A. Novick and J. R. Vaisnys,
ibid. 127, 478 (1964); H. U. Schnitzler, in populations is polluted with bacteria.
Animal Sonar Systems, R. G. Busnel, Ed. Since shallow wells are more likely to
(Laboratoire de Physiologic Acoustique, Jouy-
en-Josas-78, France, 1967). vol. 1, p. 69; R. be polluted than either deep wells or
A. Suthers, J. Mammal. 48, 79 (1967); F. A. public water supplies we made the
Webster and O. G. Brazier, AMRL-TR-65- p
172, U.S.A.F. Systems Command (1965); hypothesis that goiter is associated with
AMRL-TR-67-192, U.S.A.F. Systems Com-
stand (1968). drinking water obtained from shallow
7. B. J. Fellows, Psychol. Bull. 67, 87 (1967); wells. In 1965 and 1966 we tested this
L. W. Gellermann, J. Genet. Psychol. 42, 206
(1933). hypothesis among people living in
8. J. I. Dalland, J. And. Res. 5, 95 (1965); Richmond County in the tidewater
Science 150, 1185 (1965); -, J. A.
Vernon, E. A. Peterson, J. Neurophysiol. 30, area of Virginia. We found that there
697 (1967). = was an increased prevalance of goiter
9. A. D. Grinnell, in Animal Sonar Systems,
R. G. Busnel, Ed. (Laboratoire de Physi- among persons from households sup-
ologie Acoustique, Jouy-en-Josas-78, France, plied with water from shallow wells
1967), vol 1, p. 451.
10. J. A. Simmons and J. A. Vernon, J. Exp.
Zool., in press.
11. J. J. G. McCue, J. Aud. Res. 9, 100 (1969).
12. D. A. Cahlander, Tech. Rep. 271 (M.I.T.
Lincoln Laboratory, Lexington, Mass., 1964);
in Aninml Sonar Systems, R. G. Busnel, Ed.
(Laboratoire de Physiologic Acoustique, Jouy-
en-Josas-78, France, 1967), vol. 2, p. 1052.
13. P. M. Woodward, Probability and Informa-
tion Theory, with Applications to Radar
(Pergamon, New York, ed. 2, 1964).
14. The autocorrelation functions of the bat
cries were obtained by using a Princeton Ap-
plied Research Corp. model 101A correlation
function computer. For the use of this instru-
ment I thank Mr. S. Letzter, Mr. W. Atkin-
son, and the staff of P.A.R.
15. The bat made slight head movements of a
centimeter or two during the discrimination
trials. These head movements altered the dis-
tance to each target by a few millimeters from
one trial to another. The movements were
recorded, and the envelope of the autocorre-
lation function was corrected to compensate
for such small variations in target range after
the time scale of the autocorrelation function
was converted into an equivalent distance
scale based on the travel time of the echoes.
16. E. Ajrapetjantz, A. I. Konstantinov, D. P.
Matjushkin, Acta Physiol. Acad. Sci. Hung.
35, 1 (1969); D. R. Griffin, in Animal Sonar
Systems, R. G. Busnel, Ed. (Laboratoire de
Physiologic Acoustique, Jouy-en-Josas-78,
France, 1967), vol. 1, p. 273; N. Suga, J.
Physiol. 200, 555 (1969).
17. A. I, Konstantinov, Dokl. Akad. Nauk SSSR
161, 989 (1965); N. Suga, J. Physiol. 203,
707 (1969); ibid., p. 729.
18. Supported by grants from the National In-
stitute of Neurological Diseases and Stroke,
by an ONR contract, and by Higgins funds
allotted to Princeton University. I thank E. G.
Wever (Princeton University) and D. R.
Griffin (New York Zoological Society and
Rockefeller University) for their advice and
assistance. I also thank R. G. Busnel, J.
Chase. D. Kelley, A. I. Konstantinov, J.
Phenylthioacetate as a
Stain for Cholinesterase
Booth and Metcalf (1) suggest the
substitution of phenylthioacetate (PT)
for acetylthiocholine (ATCh) as a histo-
chemical stain for detection of cholines-
terase. In the adult summer form but
not the winter form of the female spi-
der mite (Tetranychus urticae), PT was
specific for the walls of the midgut and
insensitive to 1 X 10-7M paraoxon;
ATCh was specific for the synaptic area
of the brain and the surface of nerves
in formalin-fixed tissue (2). The cholin-
esterase sensitivity to paraoxon was
found to vary in different strains of
spider mites (3). Differences in histo-
chemical staining of PT and ATCh can
be expected among arthropods.
W. D. MCENROE
Waltham Field Station,
Waltham, Massachusetts 02154
1. G. M. Booth and R. L. Metcalf, Science 170,
455 (1970).
2. W. D. McEnroe, in Advances in Acarology,
with people who received
their water from the public supply (2).
Now Werner at a]. report that IgM
levels are elevated in persons with
goiter as compared with appropriate
nongoitrous controls (3). It seems to us
that, although other interpretations are
also possible, these data provide addi-
tional support for the infectious hy-
pothesis. Other tests of this hypothesis
(which does not exclude the iodine
hypothesis) are warranted.
W. THOMAS LONDON
Institute for Cancer Research,
Fox Chase,
Philadelphia, Pennsylvania 19111.
ROBERT L. VOUGHT
National Institute of Arthritis and
Metabolic Diseases, Bethesda, Maryland
References
1. W. T. London, D. A. Koutras, A. Pressman,
R. L. Vought, J. Clin. Endocrinol. 25, 1091
(1965); B. Malamos, D. A. Koutras, S. G.
Marketos, G. A. Rigolpoulos, X. A. Yataganas,
D. Binopoulos, J. Sfontouris, A. D. Pharmakio-.
tis, R. L. Vought, W. T. London, ibid. 27,
1372 (1967).
2. R. L. Vought, W. T. London, G. A. Stebbing,
ibid., p. 1381.
3. S. C. Werner, S. Bora, D. A. Koutras;
P..Wahlberg, Science 170, 1201 (1970).
6 January 1971
Pesticide Concentration in Seawater
The assumption of Blanchard and
Syzdek (1) that DDT might be concen-
trated in natural surface films, of sea-
water should not be left to speculation
for the readers of Science. Apparently
these and other authors (2) are unaware
that we have reported concentration
factors of up to 100 for chlorinated
pesticides in sea slicks (3). Their ex-
pectation that slicks would be areas of
high biologic activity was similarly con-
firmed (3). It has been our express con-
cern that this phenomenon may lead to
much more rapid concentration of
these toxicants in marine food chains
than would be anticipated if dilution
were homogeneous. .
DOUGLAS B. SERA
E. F. CORCORAN
Rosenstiel School of Marine and
Atmospheric Science, University of
Miami, Miami, Florida 33149
References
D. C. Blanchard and L. Syzdek, Science 170,
628 (1970).
B. Parker and G. Barsom, BioScience 20, 91
(1970).
D. B. Seba and E. F. Corcoran, Pestic. Moult.
R. A. s, ~wt A J 3 190 (1969).
their suggestiggestionfUd ~"6 1?'~'~6`S@ L roe pJRQ, U/LK00?,~^Q0A 1315.
19 August 1970; revised 14 December 1970 17 November 1970 I 6~ U t L FC 14 January 1971