Content uploaded by Jesus Rivas
Author content
All content in this area was uploaded by Jesus Rivas on Nov 08, 2020
Content may be subject to copyright.
Contemporary Herpetology 2008(2): 1-3 1
1 Department of Ecology and Evolutionary Biology, University of Tennes-
see, Knoxville TN 37996-0900, e-mail: anaconda@prodigy.net
2 Departamento de Biologia de Organismos, Universidad Simón Bolívar,
Valle de Sartenejas, Baruta, Venezuela
3 Current Address: Department of Mathematics and Natural Sciences,
Somerset Community College, 808 Monticello Street, Somerset, KY
42501, e-mail: jesus@anacondas.org
WHAT IS THE LENGTH OF A SNAKE?
INTRODUCTION
The way that herpetologists have traditionally mea-
sured live snakes is by stretching them on a ruler and
recording the total length (TL). However, due to the thin
constitution of the snake, the large number of interver-
tebral joints, and slim muscular mass of most snakes,
it is easier to stretch a snake than it is to stretch any
other vertebrate. The result of this is that the length of
a snake recorded is infl uenced by how much the animal
is stretched. Stretching it as much as possible is per-
haps a precise way to measure the length of the speci-
men but it might not correspond to the actual length of
a live animal. Furthermore, it may seriously injure a live
snake. Another method involves placing the snake in a
clear plexiglass box and pressing it with a soft material
such as rubber foam against a clear surface. Measur-
ing the length of the snake may be done by outlining its
body with a string (Fitch 1987; Frye 1991). However, this
method is restricted to small animals that can be placed
in a box, and in addition, no indications of accuracy of the
technique are given. Measuring the snakes with a fl ex-
ible tape has also been reported (Blouin-Demers 2003)
but when dealing with a large animals the way the tape
is positioned can produce great variance on the fi nal out-
come. In this contribution we revise alternative ways to
measuring a snake and propose a method that offers re-
peatable results. We further analyze the precision of this
method by using a sample of measurements taken from
wild populations of green anacondas (Eunectes murinus)
with a large range of sizes.
METHODS
To record the natural measure of the length of the ani-
mal we muzzled the snake with a sock and tape (Rivas
et al. 1995). Next, we followed an imaginary midline of
the body from head to tail (not necessarily the spine,
depending on the position of the snake) with a string and
then measured the length of the string by laying it loose-
ly on a ruler (Figure 1). This allowed us to record the
actual length of the animal regardless of its position and
without having to stretch it. A total of 82 newborn and 42
stillborn snakes from 14 wild captured pregnant females
JESUS A. RIVAS1,3, RAFAEL E. ASCANIO2,
AND MARIA D C. MUÑOZ2
were measured for this study. Three measurements of
each animal were taken; and these were slightly different
due to errors caused by the snake struggling and mov-
ing from under the string, as well as from inaccuracies
in the placement of the string. The average of the three
measures was then calculated. We also recorded the TL
of each neonate in the sample using the conventional
method of stretching it on a ruler and used a sign test to
compare both measurements of each animal. We divided
the measurements obtained with the stretching method
by the measurements obtained with the string method in
order to calculate a relationship between the two. In or-
der to analyze the changes of this relationship in respect
to size, we used the mass as an independent measure of
the size of the animal. We performed a Spearman cor-
relation test between the variables. The use of stillborn
snakes in this study was to remove the error introduced
by the struggle of live animals, allowing a determination
of the “actual” TL of the animals as close as possible.
A different sample of 68 animals from a wild population
(ranging from 84.7 cm to 494.7 cm TL) was measured in-
dependently by three people. Each snake was measured
by one of two researchers who had three years of experi-
ence performing the procedure, and by two people well
instructed in the technique but without much previous
experience. Thirteen of the animals were measured by
the two researchers (MDCM and JAR) that had previous
experience.
We calculated the coeffi cient of variation (CV) on the
three measurements collected on animals from the wild
to study the changes in the precision of the measure-
ment of snakes of different sizes. The CV was calculated
by dividing the mean by the standard deviation (see
formula in Sokal and Braumanm 1980) and provided a
measurement of the variance in units of the mean, so it
was not dependent on the absolute value of the variable
measured. This is especially important when dealing with
variables that vary in a wide range of values. All statisti-
cal analysis was made using the program SSPS 8.0.
RESULTS AND DISCUSSION
The string technique described here is comparable to
Contemporary Herpetology
ISSN 1094-2246
Volume 2008, Number 2 17 April 2008 contemporaryherpetology.org
© Contemporary Herpetology 2
the squeeze box except that it can be used on larger ani-
mals that cannot fi t in a box; or on animals that cannot
be pinned and restrained allowing broader applicability.
Measurements taken with the string were consistently
shorter than measurements with the ruler (Z= 6.82; p <
0.000; Table 1). The quotient among the measurements
is smaller in larger animals (r = –0.362; p< 0.001; Fig-
ure 2) which suggests that smaller animals are being sig-
nifi cantly stretched when measured on a ruler.
All the measurements estimate a unique parameter:
“the size of the neonate”. However, measurements from
the two methods using stillborn snakes were more dis-
parate than measurements on live individuals (Table 1).
Measurements of stillborn snakes with the ruler were
the largest of all and the measurements of stillborn
snakes with the string were the shortest of all (Table 1).
An ANOVA test shows a signifi cant difference between
the measurements of all the groups (F = 70.47; df =
3; p<0.0001). We used only stillborn animals that were
completely formed and whose cause of death was most
likely due to dystocia of the female or other problems
at the end of the gestation (Ross and Marzec 1990). We
believe that the size of the stillborns was not signifi cantly
different than the size of live neonates, which is support-
ed by the fact that there was no signifi cant difference in
mass (t= 1.252; df = 120; p = 0.21; Table 1). Thus the
difference in the measurements of the live snakes and
the stillborns are most likely due to inaccuracies resulting
from the struggling of live animals.
If we assume that the “real” length of the animal is the
length when it is relaxed, and not struggling or being
over-stretched (as is usually the case when most other
vertebrates are measured), then the length of the still-
born measured with the string should be a more realistic
estimate of true length. However, the data suggest that
even this method is not error-free either.
Repeated measurements collected with the string on the
same wild-caught animals showed a relatively high vari-
ance. The average variance in animals around 80 cm was
0.514 cm and the maximum was up to 2.35 cm. It was
clear that while processing calmer animals the repeated
measurements on them were more similar than the mea-
surements were for more active animals. In animals that
struggled a lot, the fi rst measurement tended to be the
most different. After the snake had been measured once,
it tended to calm down.
The struggle of the animal during the measuring can po-
tentially infl uence the repeatability of the measure. The
data collected by the experienced versus the inexperi-
enced researchers were signifi cantly different (Z= -3.13;
p< 0.002); where inexperienced researchers consistently
obtained shorter measurements than experienced ones.
The data collected by the two experienced researchers
were consistent with each other and were not signifi cant-
ly different in a Wilcoxon sign test (z= -0.27; p< 0.79).
The variance of the measurements changed with the
size of the animal being measured (Figure 2). Between
snake size of 2 to 3 meters the variance was particularly
high, mostly due to a few animals that had a very high
CV. This might be a consequence of the higher level of
struggling found in some smaller animals. The smallest
animals can be easily subdued during the process and
the measurements are more consistent with each other
(but see below). Beyond a certain size the snakes are
stronger and some of them are able to put up more of
a struggle, which decreases the precision of the mea-
surements. Larger animals are calmer and although they
could make the measuring much harder they tended to
be easier to measure consistently (Figure 4). However,
the CV was high in all the smaller sizes and decreased
after three meters. Thus, the lower variance found in Fig-
ure 2 for smaller sizes is probably an artifact of smaller
values. The fi rst measurement of each animal tended to
be more different than the following two; this was es-
pecially true in medium-sized animals. Larger animals
are only females and the medium-sized ones are mostly
males so some differences in the behavior of each sex
could be involved in this trend. However, the effects of
size versus sex could not be tested because adult males
are always smaller and females are typically larger (Ri-
vas 2000; Rivas and Burghardt 2001).
Stretching a snake apparently has a considerable ef-
fect on the measurements collected for the length of the
snake. Smaller animals seem to provide less resistance
to being stretched than larger ones, thus studies involv-
ing measuring animals among several size classes must
Table 1. Total length of neonate green anacondas measured by stretching
them on a ruler and by following their midbody line with a string. Lengths
are the mean of three independent measurements of each snake given in
centimeters. Mass is recorded in grams.
Length Ruler Length String Mass N
Live 79.72 77.57 228.11 82
Stillborn 85.12 76.0 225.54 42
Figure 1. Measuring technique of stretching the string over the back of the
anaconda to assess its length.
Ratio of measures
Mass (g)
140 160 180 200 220 240 260 280
1.14
1.08
0.96
1.06
0.98
1.12
1.10
1.04
1.02
1.00
Figure 2. Scatter plot of ontogenetic change of the quotient between the
measurements of neonate anacondas obtained by stretching them on a ruler
and follwing the midline of the body with a string. Notice how the relation-
ship between the two measurements changes with the size (r = –0.362 p<
0.001; n= 124).
Contemporary Herpetology 2008(2): 1-3 3
consider this issue. This method is not different in theory
from the method of the squeeze box (Fitch 1987; Frye
1991) but it has a much broader application to snakes of
larger sizes. The squeeze box method is most often used
to measure animals that are diffi cult to handle such as
venomous snakes or very small animals. Here we sug-
gest that measuring the snake with a string is the best
way to obtain accurate measurements of the length of a
snake and should be used in a more generalized manner.
We also recommend that animals be measured several
times to account for errors in the measurements that
are always present. We do not believe that using a tape
(as done by Blouin-Demers 2003) is a good idea since
the tape is not as malleable as the string and it may
make it more diffi cult to track the middle of the snake.
Furthermore, when the measure is done with a tape the
different measures are not truly independent since the
fi rst measure may infl uence how measurer places the
tape in the later ones.
The size of newborn anacondas is within the size range
of what is considered a relatively small snake. Herpe-
tologists have traditionally known that measuring large
snakes is problematic, but measuring animals smaller
than, say, 1.3 meters has not been perceived as a prob-
lem. We have shown that this is not the case. Stretch-
ing the animal on a ruler is less time consuming and in
some situations it might seem appropriate. However, the
degree that the animal is stretched can be infl uenced by
the size and behavior of the animal, or even by the mood
of the researcher!
There are many records and claims of large sizes in
snakes and what the record is for the longest snake
seems to still be a question that many herpetologists de-
bate. Here we introduce more fuel to that discussion,
and perhaps rendering it pointless, since the measured
length of the animal can vary wildly depending on the
temperament of the snake and skills of the researcher.
Our data show signifi cant difference between data col-
lected by skilled and unskilled collectors even when they
use the same technique. An error of two or three feet
does not seem to be out of the question when measuring
a snake that measures, say, more than 20 feet in length.
Anecdotally, we may point out that in 2003 an anaconda
was measured by a well-known herpetologist to be 5.5
meters long in the llanos of Venezuela. However, the
following week we had the opportunity to measure the
same snake, and it proved to be only 4.3 meters long!!
Measuring the animals with a clean, unmarked, non-
elastic string is a more reliable method especially if it is
done by people properly trained in the technique. We dis-
courage the use of a measuring tape since its consistency
and shape compromises the ability to track accurately the
middle line of the snake. In addition, it has the potential
to bias consecutive measurements that the researcher
takes. Research involving mark and recapture, or growth
studies must give special attention to these issues.
Acknowledgements: We thank The Wildlife Conserva-
tion Society and The National Geographic Society for the
fi nancial support of this research. We also thank COVEG-
AN and Estacion Biologica Hato El Frio for allowing us to
work in their facilities, and Anaconda Investments llc. for
logistic support. J. Thorbjanarson, C. Chávez, R. Kays,
D. Holtzman, B. Holsmtrom, C. Foster, M. Barcasky, and
N. Ford helped in the fi eld work. We are in debt to G.
Burghardt, P. Andreadis, M. Waters, M. Krause, and L
DiGangi for editorial comments on the ms.
LITERATURE CITED
Blouin-Demers, G. 2003. Precision and accuracy of body-
size measurements in a constricting, large-bodied snake
(Elaphe obsoleta). Herpetological Review 34: 320–323.
Fitch, H. S. 1987. Collecting and life-history techniques. In
R. A. Seigel, J. T. Collins, and S. S. Novak (eds.), Snakes:
Ecology and Evolutionary Biology, pp. 143–164. MacMillan
Publishing, New York.
Frye, F. L. 1991. Biomedical and Surgical Aspects of Cap-
tive Reptile Husbandry. 2 vols Krieger Publ. Co., Malabar,
Florida.
Rivas, J. A., Munoz M. C., Thorbjarnarson, J. B., Holmstron,
W., AND P. Calle. 1995. A safe method for handling large
snakes in the fi eld. Herpetological Review 26: 138-139.
Rivas, J. A. 2000. The life history of the green anaconda
(Eunectes murinus), with emphasis on its reproductive
biology. Unpublished dissertation at the University of Ten-
nessee at Knoxville. 287p.
Rivas, J. A. and Burghardt G. M. 2001 Sexual size dimor-
phism in snakes: wearing the snake’s shoes. Animal Be-
haviour. 62(3): F1-F6.
Ross, R. A. and G. M. Marzec. 1990. The reproductive hus-
bandry of pythons and boas. Institute of Herpetological
Research, Stanford.
Sokal, R. R and C. A. Braumanm. 1980. Signifi cance test
for coeffi cients of variation and variability profi les. Sys-
tematic Zoology 29: 50-66.
Figure 3. Size related change of variance of three measurements of SVL
obtained from each wild-caught anaconda using a string to follow the mid-
line of the body.
Figure 4 Relationship of the coeffi cient of variation from three measure-
ments on the same individual wild-caught anaconda measured with a string.
Note the decrease in the larger sizes.
Variance
0 100 200 300 400 500
600
500
400
300
200
100
0
Mean SVL (cm)
CV
0 100 200 300 400 500
6
5
4
3
2
1
0
Mean SVL (cm)
