Introduction

"You now see how, from the things demonstrated thus far, there clearly
follows the impossibility (not only for art, but for nature herself) of increasing
machines to immense size. Thus it is impossible to build enormous ships,
palaces, or temples, for which oars, masts, beamwork, iron chains, and in sum
all parts shall hold together; nor could nature make trees of immeasurable
size, because their branches would eventually fail of their own weight; and
likewise it would be impossible to fashion skeletons for men, horses, or other
animals which could exist and carry out their functions proportionally when
such animals were increased to immense height — unless the bones were
made of much harder and more resistant material than the usual, or where
deformed by disproportionate thickening, so that the shape and appearance of
the animal would become monstrously gross."

Galileo Galilei, 1638

Discourses and Mathematical Demonstrations
Relating to Two New Sciences

Size, shape, and the relationships of each to function are important topics in organismal biology,
and in many ways are the fundamental themes of this entire module. Whether you are interested
in understanding the significance of a particular morphology or interpreting the factors that
produce a particular evolutionary trend, an understanding of the physical consequences and
interdependence of form and size is mandatory. These exercises will give you a basic
understanding of the issues involved: the literature in this area is huge and spans topics from
physiology, to biomechanics, to comparative anatomy, to ecology (and even aeronautics,
engineering, and robotics!). Yet even such a basic treatment will give insights into the functional
and structural consequences of size per se.

In this practical you will take a sample of related biological specimens of different sizes, devise
hypotheses about whether the shapes of these specimens vary with size and how that might relate
to some biologically relevant function and test your hypotheses by taking a series of
measurements. First, we will review the major concepts and equations.
 Review of concepts

Geometric Similarity: Two objects are geometrically similar if they have the same shape. For
geometrically similar objects, one linear measure is proportional to another linear measure: if
height equals one half the diameter for a given bowl, then any geometrically similar bowl will
show this same relationship, no matter its size. Other relationships include:

Area ∝ Length2 ↔ Length ∝ Area 1/2

Volume ∝ length3 ↔ Length ∝ Volume1/3
Volume ∝ Area 3/2 ↔ Area ∝ Volume2/3

In each of the above, the term to the left of the proportion symbol would be considered the “y-
variable”, while the termon the right the “x-variable”.

Power Law Function: The relationships among various dimensions of objects of different sizes
often can be represented by the equation:
𝑦 = 𝑎𝑥 𝑏
↔
log 𝑦 = log 𝑎 + b ∗ log 𝑥,

where x and y are two variables (e.g., a diameter and a surface area), b is the scaling exponent,
and a is a constant that depends on shape. We have seen this in lecture.

The scaling exponent, b, is a useful way to describe the relationship between the measurements x
and y. If a set of objects are geometrically similar, b will take on a value that is easily predictable
knowing whether x and y are lengths, areas, or volumes (from the relationships above).

Isometric Scaling: This describes the condition where objects of different sizes share the same
shape. Objects or organisms that scale isometrically are geometrically similar with one another
(see above). The relationship between any two measures will fit a power law function with a
scaling exponent, b, equal to the expected relationship (or at least statistically close to it).

Allometric Scaling: This describes a set of objects or organisms that differ in shape as well as
size, and therefore are non-isometric. The relationship between any two size measures that are
allometrically related will have a scaling exponent, b, different from that predicted by isometry.
Scaling exponents larger than the predicted isometric value are said to be “positively allometric”,
while lower than predicted are “negatively allometric”.

NOTE HERE: As seen above, the Power Law relationship becomes a

simple line function when you log-transform both variables. When you
do this, you can simply calculate the slope of the regression through
the data, and that value is b. (Technically it doesn't much matter
whether you use base-10 logs or natural logs, but to keep things
simple, just use the natural log of the measurements.)
Determining if your slope (“b”) is significantly different from a value: Even in the case
where you have an isometric relationship, the slope of the regression you calculate is NEVER
going to be exactly equal to the predicted value. This is due to myriad places where tiny amounts
of error can be introduced to the process. This is to be expected.

So, for example, you could have a predicted value of 2 between x-variable leg length, and y-
variable femur cross sectional area (the x-variable is one dimensional and the y-variable is 2
dimensional). But in your experiment, you may measure a slope of 2.078. Is that positively
allometric? Is that close enough to 2 to be called isomeric? This is why we must always analyze
our observed values statistically.

What we can say is the following: if the observed value of b is not significantly different from
the predicted value of isometry, then the scaling is isometric. If, and only if, we can demonstrate
the observed value of b is significantly different from the predicted value, will we consider the
scaling allometric.

The tricky part is to determine what “statistically significant” means. To determine if an

observed slope is significantly different from a predicted slope, you must first figure out what the
predicted slope should be. This is done using the relationships above.

Next, calculate a value called the standard error of the estimate (SEb). Consider this a standard
deviation of the error derived from the estimated log(y) values – that is, the difference between
your actual measurements and the values that would be predicted by the regression. To get this,
you need to use the STEYX function in Excel.

From here the formula for a 95% confidence interval (CI) around your slope values is rather
easy:

𝐶𝐼 = 𝑏 ± 𝑡(𝛼, 𝑑𝑓) ∗ 𝑆𝐸𝑏

In the above equation, t is calculated using the TINV function in Excel. The syntax is
TINV(α,df), where α is 0.05 (that is a 5% significance threshold), and df (degrees of freedom) is
the number of data points minus 2.

If you want to test the hypothesis that b is different from an expected value:
1. Calculate b from your data. This is your observed value of b.
2. Calculate the confidence interval value by multiplying t by SEb using the above
formula.
3. Add and subtract the CI value in step 2 from the b in step 1.

Adding and subtracting the CI from b produces a range called the Confidence Interval around
the observed value of b. You can be certain, with 95% confidence that the real slope falls within
this range. If the predicted value for isometry falls between these upper and lower CI limits, then
the slope you calculated for your data IS NOT STATISTICALLY DIFFERENT from the
expected value. If the CI that you calculated does not contain the predicted value, then you have
a significant difference, and hence allometry.
 Allometry of brain and body masses in mammals

Brain size is something that evolutionary biologists have been fascinated with since Darwin (he
discussed it at length in his The Descent of Man). Here, we are going to look at phylogenetic
allometries for two ecologically and economically important groups of mammals: Carnivora and
Artiodactyla. Carnivorans include the cats, dogs, bears and other taxa spanning a wide range of
ecological and environmental niches. Artiodactyls are the even-toed ungulates, the group that
includes cows, sheep, pigs, antelopes, and many, many more species. Attached to this
assignment are two data sheets in comma-delimited format (.csv) with body masses (in grams)
and brain volumes (in milliliters) taken from Smith et al.’s (2003) database and from several
studies on brain evolution in mammals (Finarelli 2006, 2011, Finarelli and Flynn 2009).

For the provided data sets: calculate scaling equations for log-body mass (x-axis) vs. log- brain
volumes (y-axis) for the species data for Carnivora and Artiodactyla. You will need to think
about what the expectation is for brain volume and body mass in terms of expected slopes.

Produce graphs for Artiodactyla and Carnivora.

For each group, plot the log (Brian vol.) on the y-axis against log (Body Mass) on the x-axis. Do
this in a single graph. Compare the distribution of points in each clade

Questions
Do the brains of larger animals differ in their relative sizes from those of smaller ones?
If so, how?
Does a power function describe this pattern?
Is the scaling of the brain isometric or allometric with respect to body size?
Is there a difference in the scaling of brain volume to body mass between Carnivora and
Artiodactyla?
Does one group have fundamentally larger brains than the other?

Attach your results and graphs to your report, making sure to label columns and axes and include
units. Include all statistics you calculated and give the power function equation for each graph
(the easier slope-intercept form is fine). Determine whether the observed exponent matches your
prediction from isometry. If it is allometric, determine if it follows negative or positive
allometry.

Final Exercise!
Can you predict a brain volume?
The aurochs (Bos primigenius) is an extinct species of large, wild cattle that inhabited Asia,
Europe, and North Africa. It is the ancestor of modern domestic cattle. The last known aurochs
died in Poland in 1627. We do not have brain volume data for this species, but the species body
mass averaged about 364 kilograms. Can you calculate an estimated volume of the brain in ml?
References
Finarelli, J. A. 2006. Estimation of endocranial volume through the use of external skull
measures in the Carnivora (Mammalia). Journal of Mammalogy 87:1027–1036
Finarelli, J. A. 2011. Estimating endocranial volume from the outside of the skull in
Artiodactyla. Journal of Mammalogy 92:200-212.
Finarelli, J. A., and J. J. Flynn. 2009. Brain size evolution and sociality in Carnivora.
Proceedings of the National Academy of Sciences of the United States of America
106:9345-9349.
Smith, F. A., S. K. Lyons, S. K. M. Ernest, K. E. Jones, D. M. Kaufman, T. Dayan, P. A.
Marquet, J. H. Brown, and J. P. Haskell. 2003. Body mass of late Quaternary
mammals. Ecology 84:3403.