floater 1996
TRANSCRIPT
-
8/10/2019 Floater 1996
1/8
Aust ra l ian Jou rnal
of
Entomology , 1996, 35: 271 -278
27
1
The Brooks-Dyar Rule and Morphometrics of the Processionary
Caterpillar Ochrogaster Zunifer Herrich-Schaffer Lepidoptera:
Thaumetopoeidae)
GRAHAM
J.
FLOATER
Departm ent of En tomology , University of Queensland, Q ld 4072.
ABSTRACT
Vario us insect species display a unif orm geo me tric increase in size during th e larval
stage (that is they follow the B rooks-Dyar rule). Here , results of larval development are presented
for the bunny-tailed moth , Ochrogaster lunifer. The processionary larvae of this species live in
a com munal coh ort, an d mou lt en masse in the silken nest sp un at the base of their host tree (usually
a phyllodinous acacia). Th e exuviae, which remain buried in the accumulated silk and frass of the
nest, provide a life history record of the larval co ho rt. Larval exuviae were collected fro m 773 cohorts
a t 37 localities in southeastern Q ueensland between Novem ber 1993 and May 1994. Th e 6,948 exuviae
examined were from cohorts feeding on Acacia
concurrens
Pedley. Head-capsules showed a strongly
uniform geometric increase in size through eight larval instars, supporting the Brooks-Dyar rule.
The number
of
instars did not vary between trees
or
localities.
A
bimodal distribution
of
final instar
head-capsule widths was shown to b e a sexual dimorph ism, an d s imilar bimodal dis tr ibutions were
foun d fo r instars V-VII. P upal size was also sexually dimorp hic. Th e geometric size increase from
one larval instar to the next holds for both males and the larger females. Th e geometric rule was
tested using larval cohorts reared
on
A . concurrens in the greenhouse through instars
I-IV;
development was rem arkably s imilar to that in the f ield. Larval growth pa t terns of 0 lunifer a r e
very different from the structurally similar bag-shelter moth. The ability to distinguish different
instars
of
0
unifer
with a high degree of precision from field-collected exuviae will allow accura te
comparisons of developme nt, survival and dispersal of larvae in different gr oup sizes, on different
trees and in different localities.
Introduction
Ground-nesting populations of the bunny-tailed
moth, Ochrogaster lunifer Herrich-Schaffer
(Lepidoptera: Thaumetopoeidae), are common
and widespread along the eastern seaboard of
Australia (Froggatt 1896; Floater 1996a). The
processionary larvae live in a communal cohort
at the base of the host tree (usually a species of
phyllodinous acacia), moving up the trunk in
single file to feed in the canopy at night. The larvae
moult
en
masse in the silken nest spun at the base
of the tree, where the exuviae remain buried in
accumulated silk and frass. These exuviae
therefore represent a life history record of the
larval cohort.
I
reasoned that
if
different larval
stages could be distinguished within the same nest,
the moth would be an ideal subject for ecological
studies on the mortality and development of
individuals within and between populations.
The number of larval instars in insect species
is often determined from
a
frequency distribution
of larval size. A representative sample of larvae
of all ages is collected from the field, and various
measurements (often the width of the head)
recorded. Alternatively, the moulted skins of
larvae may be collected, as long as the head
capsules, or other well-sclerotised structures,
remain intact.
I f
the growth rates of larvae are
relatively uniform, the resulting frequency
distribution should consist of
a
series of peaks,
with each peak representing one instar (see Daly
1985 for review). In practice, however, the
frequency distribution is often ragged with
overlapping peaks caused by sampling error,
environmental variation, sexual dimorphism,
parasitism and genetic differences between
individual animals.
A second method of analysis involves the
Brooks-Dyar rule (Brooks 1886; Dyar 1890;
Hutchinson and Tongring 1984). The rule states
that larval head widths in successive stages
describe a regular geometric progression (Dyar
1890) with the following equation:
where X is the instar number (1, 2, 3, etc.); Y is
head-capsule width; and a and b are
constants. The equation serves both as a growth
curve and as
a
method of checking for an
overlooked instar in
a
frequency distribution (Daly
1985). To check for a hidden instar, equation
1)
is made linear by taking the natural logs of both
sides, giving:
1nY = c
+
bX. (2)
where c = In(a). The relationship between 1nY and
X should be a straight line with slope b, and
therefore a significant deviation from a straight
line indicates
a
missing instar.
While the Brooks-Dyar rule holds for many
species (e.g. Dyar 1890; Taylor 1931; Gaines and
Campbell 1935; Mackay 1978), many exceptions
exist (e.g. Kishi 1971; Craig 1975; Schmidt
et
al.
1977; Allsopp and Adams 1979; Jobin
et
al. 1992).
Here the different larval stages of
0
lunifer are
distinguished using the Brooks-Dyar rule and the
subsequent model tested using cohorts reared in
the greenhouse.
Y = aebX (1)
-
8/10/2019 Floater 1996
2/8
212 0
.
FLOATER
Methods
0
unifer
is univoltine.
A
survey of larval exuviae
f rom one genera t ion was conducted f rom
November 1993 (when eggs hatched) to Ma y 1994
(when final instar larvae dispersed to enter
prepupa l diapause). In November 1993, over 4,000
acacia trees of various species were examined f or
egg batches at 37 localities in southeastern
Quee nsland. D etails of localities can be f ou nd in
Floater (1996a). A tota l of 773 egg batches w ere
recorded on six
Acacia
species: A .
concurrens
Pedley,
A . leiocafyx
Domin (Pedley),
A .Jimbriata
Cunn. ex
G .
Don ,
A . aulacocarpa
Cunn. ex
Bentham,
A . implexa
Bentham and
A . maidenii
F.
v. Mueller. Over 95 of trees with eggs were
A . concurrens
and the rest of the study focused
on larvae feeding on this species.
Localities w ere revisited
2
m onth s later (in late
Janu ary 1994), when larvae were in the fifth instar.
Co hor t extinction (100 mortality) is comm on
in the early instars, an d the rem ains of all extinct
cohorts were collected and examined for larval
skins. Head-capsule widths of the 10largest skins
in each extinct coho rt were then measured t o give
an estimate of larval size at the time o f,
or
shortly
before, extinction. Depending on the time of
ex t inc t ion , the head-capsu le wid ths could
therefore represent instar I, 11, I11 or VI.
Measurements were made
to
the nearest 0.1 m m
using a graticule set in a stereo microscope.
In Ju ne 1994, after the final instar larvae had
migrated from trees to pupate underground, all
remaining nests were collected from 29 of the 37
localities. Th e nest m aterial was examined in the
4001
labo rato ry. Exuviae of the penultimate (seventh)
instar (i.e. the last larval moult before the final
instar larvae leave the nest) were estimated by eye,
and th e head-capsule of each removed. Instar VII
cast skins appeared to be distinctly larger than
other skins in the nest, and were often matted
together. If doubt existed over assigning a skin,
the head-capsule was removed an d added to the
rest. For each cohort, head-capsule widths were
measured t o the nearest 0.1 m m, a nd a frequency
distribution of sizes plotted. If a discontinuity
existed in the lower end of the distribution (i.e.
there appeared to be skins in distinct size classes
below instar VII; see Fig. l) , only da ta above the
discontinuity were used in subsequent analyses.
In nests where larvae had defoliated their host-
plant an d emigrated before the penultimate mo ult,
all skins present were removed for analysis. Once
again, for each cohort, a frequency distribution
was plotted, and only data above the highest
discontinuity (i.e. skins tha t appeared to be in the
highest distinct size class) were subsequently used.
This distribution represented either instar
V
or
instar VI, depending on the stage at which
emigration to ok place. In all, 6,948 head-capsules
were measured fro m exuviae collected in Jan ua ry
and June.
T o measure the head capsule widths of known
larval instars, cohorts were reared
to
the fourth
instar stage in :he greenhou se. I n December 1994,
five egg batches were collected from Kuraby in
eastern Brisbane (this locality was not par t of the
regional survey from the year before). Each batch
was then ke pt in a small plastic tu b until the first
larval moult. Th e first instar larvae d o not feed,
1
I Ins tar
VII
Instar VI
I
1
I
I
I \
.___
1.8 2 2.2 2.4
2.6
2.8 3 3.2 3.4
3.6
3.8 4 4.2 4.4 9 6
head-capsule width
(mm)
Fig. 1.
Frequency distribution
of
head-capsule widths in a larval cohort. The distribution describes instars
V-VII,
with each
instar displaying a bimodality of sexes.
-
8/10/2019 Floater 1996
3/8
MORPHOMETRICS
OF
0 LUNIFER 213
an d remain in the m at of scales deposited by the
female moth over her egg batch. When all the
larvae in a cohort had m oulted, 20 head-capsules
were removed at random from the tub for
subsequent examination.
The newly moulted second instar larvae were
transferred from the tubs to greenhouse plants.
Plants were well watered and fertilised. Each
cohort was placed, along with the scale mat, at
the base of a potted plant of
A . concurrens
whereupon the larvae ascended the plant to feed.
30 1
I n=25)
5
0
The co horts were kept on the plants for
6
weeks,
and during this time the larvae moulted four times.
After each moulting event, all skins from a cohort
were removed fro m the base of the plant. Of these,
20
skins were chosen at random and the head
capsules removed. C onseque ntly, at the end of the
experim ent, a collection of head capsules had been
made, with
20
capsules of each instar fr om each
cohort. The width of each head capsule was
measured to the nearest
0.04
mm using a stereo
microscope.
0 5 0.9 1 . 3 1 .7 2 . 1 2 . 5 2 . 9 3 . 3 3 . 7 4.1 4 . 5 4.9
head-capsule width
mm)
30 -
I
25
I1
20 w
G
c
2
1 5
ct:
\
I
10
I11
VI
0.5
0.9
1 .3 1 .7 2 .1
2.5 2.9
3 . 3 3 . 7
4.1
4 . 5
4.9
head-capsule width mm)
Fig.
2.
Frequency distribution of larval size modal head-capsule width) from field-collected cohorts, with seven peaks representing
seven instars:
upper)
Distribution o f fema le sizes; the upper female) mode head-capsule width was calculated for the largest
exuviae in each coh ort and plotted a s a single data entry;
lower)
Distribution
of
male sizes; lower mo de head-capsule widths.
-
8/10/2019 Floater 1996
4/8
214
G.
J.
FLOATER
To measure final instar moults and to
investigate sex differences, eight nests were
collected from trees of
A . concurrens
in May 1995.
The nests contained final instar caterpillars that
were close to prepupal diapause. The resulting
pupae were removed from their silken cocoons in
September 1994, sexed and measured (length and
width). The head-capsule of the final moult
contained in the cocoon was measured also.
Consequently, the sex and size of final instar
larvae could be ascertained.
2 1
1.5
.I
Results
Sexual
dimorphism.
Female pupae, with length
24.4
k
1.3 mm (mean
SD,
n = 29) and width
9.8
0.5
mm, were significantly larger than male
pupae, with length 21.1
k
1.2 mm (n
=
25) and
width 8.2 0.6 mm. Furthermore, the bimodal
distribution of head capsule widths of final instar
larvae was shown to be the result of sexual
dimorphism. Female head capsule widths were
significantly larger 5.51 0.39 mm, n = 29) than
- . 5
1
INSTAR
2
c
w
.5
.
m
1
0 1 3 4 5 6 7 8
INSTAR
Fig. 3. A
regression showing'the geometric increase
of
larval size (represented as ln[head-capsule width]) with
instar number: upper) Female larvae; lower) Male larvae.
corresponding
-
8/10/2019 Floater 1996
5/8
MORPHOMETRICS
OF
0 LUNIFER 275
those of male larvae (4.65 f 0.19 mm, n =
25);
ANOVA F
= 17.13;
P =
0.002.
This difference
between the sexes can be seen in earlier instars.
Fig. 1 shows the distribution of head capsule sizes
from a single nest. The distribution represents
three instars (instars
5
6,
7)
and clearly
demonstrates the bimodality of each instar, even
in instar 5 The differences in size between males
and females must be taken into account when
3
2 5
3 2
2
.
1.5
2
-
l
.
. 5
a 0
- - . 5
1
1.5
W
--
c
W
c:
N
d
n
z
d
assessing the validity of the Brooks-Dyar rule.
Consequently, in analyses of larval growth, the
upper an d lower modes in the distribution of head-
capsule widths for each instar from instar
5
to
8
are considered separately.
Larval development in field-collected cohorts. By
plotting the modal head-capsule width of cohorts
from the Janu ary an d J un e collections in 1994,
a
continuous frequency distribution of head-
1
2
3
4
5
6
7
8
INSTAR
Fig.
4. A regression showing the geometric increase in the variation of larval size with increasing instar number.
I
120
100
80
60-
L 2
G
40
IV
20
1 1
0.64 0.8 0.96 1.12 1.28
1.44 1.6
1.76 1.92
head-capsule width
(mm)
Fig. 5 .
Frequency distribution of head-capsule widths of larvae from five cohorts reared in the greenhouse. The four peaks
represent instars I-11.
-
8/10/2019 Floater 1996
6/8
216 0
.
FLOATER
capsule widths was created for instars
I-VII
(Fig.
2). Becauseof the differences in head size between
males and females, two distributions were created.
The first distribution (Fig. 2a) represents upper
(i.e. female) modes, while the second (Fig. 2b)
represents lower (male) modes.
In
both graphs, a
single mode was calculated for cohorts collected
in January (instars
I-IV),
and hence both graphs
have an identical distribution of head sizes from
0 5 to 1.7 mm.
Examining the frequency distribution of female
modes, there appear to be seven major peaks with
modes at approximately 0.6 mm (peak l), 0.85
mm(peak2), 1.1 mm(peak 3), 1.4 mm(peak4),
2.15 mm (peak
5),
2.95 mm (peak
6 ,
4.4 mm
(peak 7). Peaks 5-7 (from the June collection) are
quite distinct, and peak 7 appears to be bimodal.
Cohorts collected at some localities had distinctly
larger larvae than cohorts at other localities,
creating a dichotomy
of
sizes rather than a
continuous range of sizes. The reason for this
geographical distinction is unclear. Peaks 1-4
(from the January collection) are also quite
distinct, though a possible gap exists between
peaks 4 and
5 . The first instar moult (peak 1) is
particularly pronounced.
Assuming the seven peaks in Fig. 2a to represent
seven instars, the following categories were
created: instar
1:
0.6; instar 2: 0.8-1.0; instar
3:
1.1-1.3; instar 4: 1.4-1.7; instar 5 : 1.9-2.4; instar
6:
2.6-3.4; instar 7: 3.7-4.9. A regression was then
plotted of larval size against corresponding instar
number for each cohort (Fig. 3a). The result is a
near perfect straight line with equation InY
=
0.326X 0.845; R* = 0.994, and consequently,
the seven peaks do appear to correspond to seven
consecutive instars. Furthermore, the line predicts
the mode head-capsule width of final instar
females (instar 8) accurately (Fig. 3a). A similar
analysis
of
lower size modes (corresponding to
male head-capsule widths) gives a straight line with
equation
InY =
0.305X 0.803;
R2
0.994 (Fig.
3b). Because the increase in size from one instar
to the next is geometric, the size ratio of one instar
to the next is a constant (eb). This ratio is 1.39 for
females and 1.36 for males.
The mean larval size of males and females is not
the only variable to increase exponentially with life
stage. The range
of
larval head-capsule widths
around the mode also gets progressively larger as
the larvae moult from one instar to the next. While
the relationship is not as tight as for the modes,
Fig. 4 clearly shows the geometric increase in size
range with instar number. The straight-line
equation describing the relationship is 1nY
=
0 517X 1.22; RZ 0.961 (where
Y
is the range
of head-capsule widths in each instar).
Larval development
in
greenhouse-reared
cohorts.
From the field results, larval development of 0
lunifer
does appear to follow the Brooks-Dyar
rule, with a geometric growth rate through eight
larval instars. To test the validity of the findings
from the field-collected data, the size distributions
of greenhouse-reared larvae were compared to the
regression models calculated from the field data.
When pooled together, the size distribution of
I
i
-.6
.4
- . 2
0
. 2
.4
.6
predicted
larval
size
In
hcw)
(mm)
Fig. 6 . A
test of the Brooks-Dyar rule in
0 lunifer
comparing observed head-capsule widths
of
greenhouse-reared larvae
with those predicted for respective instars (I-IV). The solid line shows the 1:l ratio of observed to predicted sizes, on which
the points were predicted to lie.
-
8/10/2019 Floater 1996
7/8
MORPHOMETRICS OF
0
LUNZFER 2
greenhouse-reared larvae showed distinct classes
for instars 1-4, with prominent modes at 0.60,
0.84, 1.12 and 1.60 mm (Fig. 5 . The results from
the greenhouse experiment were compared with
the geometric model calculated from the field
data. A predicted head-capsule size was first
calculated as the mean of male and female head-
capsule widths for each instar from the field data.
Fig. 6 shows the strong relationship between
predicted head-capsule sizes and those observed
in the experiment for instars 1-4. The points lie
very close to a 1 relationship, and the regression
is highly significant (Y
= 0.006 +
1.022X; R 2 =
0.996), supporting the Brooks-Dyar rule. The
straight-line equation describing the relationship
between reared-larval size and instar number is InY
= 0.322X 0.835 RZ = 0.996), which lies
between the lines predicted for male and female
head-capsule widths.
Discussion
Assigning larval instars of 0 lunijer.
When
comparing larval development and survival
between cohorts on different trees and in different
localities, it is essential to be able to distinguish
between larval instars within the same cohort.
Different insect species exhibit different growth
patterns, which may aid or hinder the researchers
ability to assign exuviae to particular instars. In
the Lepidoptera, growth patterns may display
consecutive size ratios that remain constant
throughout development (Dyar 1890; Gaines and
Campbell 1935); that decrease with each moult
(Jobin et al. 1992); or that are highly variable from
one moult to the next (Schmidt et al. 1977). In
some insect species, larval development is so
irregular that assigning individuals to age classes
is simply impossible. For example, AIlsopp and
Adams (1979), in a study of the tenebrionid beetle,
Pterohelaeus darlingensis Carter, were only able
to distinguish one larval instar (the first) and that
due to its colour rather than size. The problem
with assigning independent individuals to stadia
is that even when development is uniform for each
individual, the different developmental rates
between individuals will usually produce overlaps
in instar size; particularly in later instars.
Larvae of 0 lunijer possess two features that
allow accurate assignment of larval exuviae to
respective instars. First, rate of development
appears to be relatively uniform for all larvae
within a cohort. And second, because the rate of
development follows a geometric progression,
instars remain relatively distinct. Even cohorts
reared on young, fertilised plants in the
greenhouse had a remarkably similar rate of
growth to those on older trees of varying condition
in the field. Variation in size exists within each
instar, but rarely overlaps between instars within
the same cohort. As Fig. 4 shows, the range of
larval sizes increases exponentially with instar
number. It is for this reason that in insect species
that do not follow the Brooks-Dyar rule, later
instars with large size ranges overlap with one
another. Only if the mean instar size increases
exponentially along with the range of sizes around
the mean, will later instars remain distinct.
When comparing rates of larval development
between cohorts, it is enough to distinguish the
respective peaks of male and female size for a
particular instar. However, when comparing larval
survival, the researcher must know the number of
individuals in a particular instar. Consequently,
it is more important to distinguish between the
upper and lower size bounds of an instar rather
than the position of the mode. As Fig. 1 shows,
these bounds are quite distinct in a cohort. Clearly,
some subjective judgment is inevitabIe when
assigning the one or two individuals at the
boundary between adjacent instars, but these will
be all but irrelevant in statistical comparisons of
cohort numbers. Comparing larval numbers at
instar VII has the added advantage that instar VIII
exuviae are not present in the nest. Furthermore,
in over 90 of the nests sorted, a distinct gap
existed between instar VI and instar VII, and in
cases where an overlap existed, the number of
individuals at the boundary was never more than
two.
Morphometrics and the 0 lunifer species
complex. While Common (1990) has suggested
that the name
0
lunifer
be restricted to ground-
nesting populations in eastern Australia, the name
has been used by other authors to describe the
canopy-nesting bag-shelter moth (van Schagen et
al. 1992). Although adults of the bag-shelter moth
are structurally similar to those of the bunny-tailed
moth (Common 1990), various behavioural and
morphological differences exist between the
immature stages (Floater 1996a).
The morphometrics o f both moths also suggest
that they are different species. Van Schagen et al.
(1992) described the larval morphometrics of the
bag-shelter moth, and found the following: six
larval instars;
a
decrease in the size ratio from
instar to instar, contradicting the Brooks-Dyar
rule; no relationship between size range and mean
size for each instar; and no sexual dimorphism in
larval size. In contrast, the results of my study on
the bunny-tailed moth show eight larval instars;
a
constant size ratio;
a
geometric increase in size
range from instar to instar; and marked sexual
dimorphism in larval size. It is not clear why van
Schagen et al. (1992) found a greater range (and
variance) of head capsule widths in instar I than
in instar VI, while the mean size increased almost
by an order of magnitude (instar
1:
0.57 0.15
mm S.D.; instar
VI ;
4.80 0.10 mm). The scope
of the investigation was limited (a total of 240
larvae were examined) and a more comprehensive
study of the early stages would be required to
verify whether the range of first instar sizes in the
bag-shelter moth was indeed large, or whether
a
-
8/10/2019 Floater 1996
8/8