Professional Documents
Culture Documents
A General Model For The Origin of Allometric Scaling Laws in Biology
A General Model For The Origin of Allometric Scaling Laws in Biology
encoding a full-length profilin cDNA (16); p989, A General Model for the Origin of Allometric
which encodes a mutant form of profilin, Pfy1p-3,
lacking the last three amino acids (18); p890, which Scaling Laws in Biology
contains the Bgl II to Stu I fragment from p182 (26),
encoding Bni1p(1227–1397); p813, which con-
tains the Bgl II to Not I fragment from p182, encod-
Geoffrey B. West, James H. Brown,* Brian J. Enquist
ing Bni1p(1414 –1953); and p951, which contains Allometric scaling relations, including the 3/4 power law for metabolic rates, are char-
the Hpa I to Not I fragment from p182, encoding
Bni1p(1647–1953). The pJG4-5 – derived plasmids
acteristic of all organisms and are here derived from a general model that describes how
were p561, which contains the Bam HI to Not I essential materials are transported through space-filling fractal networks of branching
fragment from p532 (26), encoding Bni1p(1–1953); tubes. The model assumes that the energy dissipated is minimized and that the terminal
p717, which contains the Bam HI to Eco47 III frag- tubes do not vary with body size. It provides a complete analysis of scaling relations for
ment from p532, encoding Bni1p(1–1214); p558,
which contains the Eco 47III to Not I fragment from mammalian circulatory systems that are in agreement with data. More generally, the
p182, encoding Bni1p(1215 –1953); p913, which model predicts structural and functional properties of vertebrate cardiovascular and
contains the Bgl II to Stu I fragment from p182, respiratory systems, plant vascular systems, insect tracheal tubes, and other distribution
encoding Bni1p(1227–1397); p929, which con-
tains the Bgl II to Not I fragment from p182, encod- networks.
ing Bni1p(1414 –1953); p952, which contains the
Hpa I to Not I fragment from p182, encoding
Bni1p(1647–1953); and p887, which contains the
Bam HI to Not I fragment encoding a full-length
profilin cDNA (16). The pACT-derived plasmid was
p1124, encoding full-length Act1p as isolated in a Biological diversity is largely a matter of underlies these laws: Living things are sus-
catch and release screen (22). The pGAD-C– de- body size, which varies over 21 orders of tained by the transport of materials
rived plasmid was p688, encoding the COOH-ter-
minal 311 amino acids (478 –788) of Bud6p, as
magnitude (1). Size affects rates of all bio- through linear networks that branch to
isolated in a catch and release screen (22). logical structures and processes from cellu- supply all parts of the organism. We de-
29. For localization of Bni1p, SY2625 (11) cells carrying a lar metabolism to population dynamics (2, velop a quantitative model that explains
multicopy plasmid encoding either HA-tagged Bni1p 3). The dependence of a biological variable the origin and ubiquity of quarter-power
[pY39tet1 (9)] or nontagged Bni1p were induced to
form mating projections (12). HA-Bni1p was localized
Y on body mass M is typically characterized scaling; it predicts the essential features of
by immunofluorescence with monoclonal antibody by an allometric scaling law of the form transport systems, such as mammalian
HA.11(Berkeley Antibody Company) as described [J. blood vessels and bronchial trees, plant
R. Pringle, A. E. M. Adams, D. G. Drubin, B. K. Haarer, Y 5 Y 0M b (1) vascular systems, and insect tracheal
Methods Enzymol. 194, 565 (1991)]. For localization where b is the scaling exponent and Y0 a tubes. It is based on three unifying princi-
of Bud6p, SY2625 cells expressing GFP-Bud6p (23)
or containing the control plasmid pRS316 (26) were constant that is characteristic of the kind ples or assumptions: First, in order for the
induced to form mating projections (12), then ob- of organism. If, as originally thought, these network to supply the entire volume of
served by fluorescence microscopy with the use of a relations reflect geometric constraints, the organism, a space-filling fractal-like
fluorescein isothiocyanate filter set.
30. Yeast cells of strain B5459 (MATa pep4::HIS3
then b should be a simple multiple of branching pattern (9) is required. Second,
prb1D1-6R ura3 trp1 lys2 leu2 his3D200 can1) car- one-third. However, most biological phe- the final branch of the network (such as
rying p1025 (26) were grown to mid-log phase in nomena scale as quarter rather than third the capillary in the circulatory system) is a
raffinose medium, and galactose was added to
induce the production of HA-tagged Bni1p(1215 –
powers of body mass (2–4): For example, size-invariant unit (2). And third, the en-
1953). After 1 hour, extracts were prepared by metabolic rates B of entire organisms scale ergy required to distribute resources is
grinding cells with glass beads in lysis buffer [0.6 M as M3/4; rates of cellular metabolism, minimized (10); this final restriction is
sorbitol, bovine serum albumin (1%), 140 mM heartbeat, and maximal population basically equivalent to minimizing the to-
NaCl, 5 mM EDTA, 50 mM tris-HCl (pH 7.6), 0.06%
Triton X-100, 2 mM phenylmethylsulfonyl fluoride, growth scale as M21/4; and times of blood tal hydrodynamic resistance of the system.
aprotinin (10 mg/ml)] as described (2). Escherichia circulation, embryonic growth and devel- Scaling laws arise from the interplay be-
coli strain BL 21 (Novagen) was transformed with opment, and life-span scale as M1/4. Sizes tween physical and geometric constraints
pGEX-3X (Pharmacia) or p907 (26) and induced for
expression of GST or GST-profilin, respectively. of biological structures scale similarly: For implicit in these three principles. The
GST proteins were purified on glutathione-Sepha- example, the cross-sectional areas of mam- model presented here should be viewed as
rose (Pharmacia) and washed twice with phos- malian aortas and of tree trunks scale as an idealized representation in that we ig-
phate-buffered saline (PBS) [140 mM NaCl, 2.7
mM KCl, 10 mM Na2HPO4, 1.8 mM KH2PO4 (pH M3/4. No general theory explains the ori- nore complications such as tapering of
7.3)]. Glutathione-Sepharose beads with GST or gin of these laws. Current hypotheses, vessels, turbulence, and nonlinear effects.
GST-profilin bound were then added to the yeast such as resistance to elastic buckling in These play only a minor role in determin-
extract containing HA-Bni1p(1215 –1953) and in-
cubated on ice. After 45 min, the beads were col-
terrestrial organisms (5) or diffusion of ing the dynamics of the entire network
lected and washed twice with PBS. The GST pro- materials across hydrodynamic boundary and could be incorporated in more de-
teins and associated proteins were eluted with glu- layers in aquatic organisms (6), cannot tailed analyses of specific systems.
tathione [10 mM glutathione, 50 mM tris-HCl (pH
8.0)] and subjected to immunoblot analysis with
explain why so many biological processes Most distribution systems can be de-
antibodies to GST (Pharmacia) or the HA epitope in nearly all kinds of animals (2, 3), plants scribed by a branching network in which
(29) as described (27). (7), and microbes (8) exhibit quarter-pow- the sizes of tubes regularly decrease (Fig.
31. We thank D. Amberg, B. Andrews, R. Brent, R. er scaling. 1). One version is exhibited by vertebrate
Dorer, S. J. Elledge, S. Givan, B. K. Haarer, J.
Horecka, P. James, I. Sadowski, M. Tyers, and J. We propose that a common mechanism circulatory and respiratory systems, anoth-
Zahner for plasmids and yeast strains; B. Brand- er by the “vessel-bundle” structure of mul-
horst, J. Brown, N. Davis, S. Kim, B. Nelson, and I. G. B. West, Theoretical Division, T-8, Mail Stop B285, tiple parallel tubes, characteristic of plant
Pot for comments on the manuscript; and G. Poje Los Alamos National Laboratory, Los Alamos, NM
and I. Pot for assistance with experiments. Support- 87545, and The Santa Fe Institute, 1399 Hyde Park vascular systems (11). Biological networks
ed by grants to C.B. from the Natural Sciences and Road, Santa Fe, NM 87501, USA. vary in the properties of the tube (elastic
Engineering Research Council of Canada and the J. H. Brown and B. J. Enquist, Department of Biology, to rigid), the fluid transported (liquid to
National Cancer Institute of Canada; by a grant from University of New Mexico, Albuquerque, NM 87131, and
the Swiss National Science Foundation to M.P.; and The Santa Fe Institute, 1399 Hyde Park Road, Santa Fe,
gas), and the nature of the pump (a pul-
by NIH grant GM31006 to J.R.P. NM 87501, USA. satile compression pump in the cardiovas-
* To whom correspondence should be addressed. E-mail: cular system, a pulsatile bellows pump in
2 December 1996; accepted 10 February 1997 jhbrown@unm.edu the respiratory system, diffusion in insect
k50
k k
N
k50
2
k k
k case of the classic rigid-pipe model, where
the branching is assumed to be area-pre-
eled by a single or a few representative
vessels. (iii) The scaling with M does not
must have area-preserving branching, so F(rk, lk, n) 5 W(rk, lk, nk, M) Here a [ (vr/m)1/2r is the dimensionless
that b 5 n21/2, leading to quarter-power Womersley number (13), and c0 [ (Eh/
scaling. The smaller vessels, on the other
hand, have the classic “cubic-law” branch- 1 lVb(rk, lk, nk, M) 1 O lNl 1l M
N
k50
k
3
kk M
2rr)1/2 is the Korteweg-Moens velocity. In
general, both c and Z are complex functions
ing (10), where b 5 n21/3, and play a (7) of v, so the wave is attenuated and disperses
relatively minor role in allometric scaling. as it propagates. Consider the consequences
First consider the simpler problem of Because B } Q0 and W 5 Q̇02Z, this prob- of these formulas as the blood flows through
nonpulsatile flow. For steady laminar flow lem is tantamount to minimizing the im- progressively smaller tubes: For large tubes,
of a Newtonian fluid, the viscous resistance pedance Z, which can therefore be used in a is large (in a typical human artery, a '
of a single tube is given by the well-known Eq. 7 in place of W. First, consider the case 5), and viscosity plays almost no role. Equa-
Poiseuille formula Rk 5 8mlk/pr4k , where m is where nk 5 n, so that we can use Eqs. 4 and tion 8 then gives c 5 c0 and Z 5 rc0/pr2;
the viscosity of the fluid. Ignoring small 6 for Vb and Z, respectively. For a fixed mass because both of these are real quantities,
effects such as turbulence and nonlineari- M, the auxiliary Lagrange function F, the wave is neither attenuated nor dis-
ties at junctions, the resistance of the entire which incorporates the constraints, must be persed. The r dependence of Z has changed
network is given by (14) minimized with respect to all variables for from the nonpulsatile r24 behavior to r22.
O NR 5 O pr8mln
N N the entire system (rk, lk, and n). This re- Minimizing energy loss now gives hk/rk (and,
k k
Z5 4 k
quires ]F/]lk 5 ]F/]rk 5 ]F/]n 5 0, which therefore, ck) independent of k and, most
k50 k k50 k straightforwardly leads to bk 5 n21/3. More importantly, an area-preserving law at the
@1 2 ~nb4/g!N 1 1#Rc generally, by considering variations with junctions, so bk 5 n21/2. This relation en-
5 respect to nk, one can show that nk 5 n, sures that energy-carrying waves are not
~1 2 nb4/g!nN (6) independent of k. The result, bk 5 n21/3, is reflected back up the tubes at branch points
Now, nb /g , 1 and N .. 1, so a good
4
a generalization of Murray’s finding (17), and is the exact analog of impedance
approximation is Z 5 Rc/(1 2 nb4/g)Nc. derived for a single branching, to the com- matching at the junctions of electrical
Because Rc is invariant, Z } Nc21 } M2a, plete network. Now varying M and mini- transmission lines (18). As k increases, the
which leads to two important scaling laws: mizing F in Eq. 7 (]F/]M 5 0) leads to Vb sizes of tubes decrease, so a 3 0 (in human
blood pressure Dp 5 Q̇0Z must be indepen- } M, which is just the relation needed to arterioles, for example, a ' 0.05), and the
dent of body size and the power dissipated derive Eq. 5. Although the result bk 5 n21/3 role of viscosity increases, eventually dom-
in the system (cardiac output) W 5 Q̇0Dp } is independent of k, it is not area-preserving inating the flow. Equation 8 then gives c '
Ma, so that the power expended by the and therefore does not give a 5 3/4 when i1/2ac0/4 3 0, in agreement with observa-
heart in overcoming viscous forces is a size- used in Eq. 5; instead, it gives a 5 1. It does, tion (18). Because c and, consequently, l
independent fraction of the metabolic rate. however, solve the problem of slowing now have imaginary parts, the traveling
Neither of these results depends on detailed blood in the capillaries: Eq. 2 gives uc/ u0 5 wave is heavily damped, leaving an almost
knowledge of n, b, or g, in contrast to (nb2)2N 5 Nc21/3. For humans, Nc ' 1010, steady oscillatory flow whose impedance is,
results based on Vb } M, such as Eq. 5, a 5 so uc/ u0 ' 1023, in reasonable agreement from Eq. 8, given by the Poiseuille formula;
3/4, and r0 } M3/8. From Eq. 2, Q̇0 5 pr20u0, with data (18). On the other hand, it leads that is, the r24 behavior is restored. Thus,
which correctly predicts that the velocity of to an incorrect scaling law for this ratio: for large k, corresponding to small vessels,
blood in the aorta u0 } M0 (2). However, an uc/ u0 } M21/4. Incorporating pulsatile flow bk 5 n21/3. We conclude that for pulsatile
area-preserving scaling relation b 5 n21/2 not only solves these problems, giving the flow, bk is not independent of k but rather
also implies by means of Eq. 2 that uk 5 u0 correct scaling relations (a 5 3/4 and uc/ u0 has a steplike behavior (Fig. 2). This picture
for all k. This relation is valid for fluid flow } M0), but also gives the correct value for
in plant vessels (because of the vascular bun- uc / u0.
dle structure) (11, 15) and insect tracheae A complete treatment of pulsatile flow
(because gas is driven by diffusion) (16); is complicated; here, we present a simpli-
both therefore exhibit area-preserving fied version that contains the essential
branching, which leads to 3/4 power scaling features needed for the scaling problem.
of metabolic rate. Branching cannot be en- When an oscillatory pressure p of angular
tirely area-preserving in mammalian circula- frequency v is applied to an elastic (char-
tory systems because blood must slow down acterized by modulus E) vessel with wall
to allow materials to diffuse across capillary thickness h, a damped traveling wave is
walls. However, the pulsatile nature of the created: p 5 p0ei(vt 2 2pz/l). Here, t is time,
mammalian cardiovascular system solves the z is the distance along the tube, l is the
problem. wavelength, and p0 is the amplitude aver-
Energy minimization constrains the net- aged over the radius; the wave velocity
Fig. 2. Schematic variation of the Womersley num-
work for the simpler nonpulsatile systems. c 5 2pvl. Both the impedance Z and the ber ak and the scaling parameters bk and gk with
Consider cardiac output as a function of all dispersion relation that determines c are level number (k) for pulsatile systems. Note the
relevant variables: W(rk, lk, nk, M). To sus- derived by solving the Navier-Stokes steplike change in bk at k 5 k from area-preserving
tain a given metabolic rate in an organism equation for the fluid coupled to the pulse-wave flow in major vessels to area-increasing
of fixed mass M with a given volume of Navier equations for the vessel wall (19). Poiseuille-type flow in small vessels.
HS D
and Poiseuille impedances are comparable in 2¯k ¯
implies Nc } M3/4 rather than the naı̈ve
size. The approximate value of k where this Vc b. 1 2 ~nb2,g!k expectation Nc } M, so the volume serviced
occurs (say, k) is given by r 2k /l k ' 8m/rc0, Vb 5 2 N by each capillary must scale as M1/4, and
leading to N 2 k [ N ' ln(8mlc/rc0rc2)/ln n, ~b.g! b, 1 2 ~nb2,g! capillary density per cross-sectional area of
tissue, as M21/12.
F GJ
independent of M. Thus, the number of
¯
generations where Poiseuille flow dominates 1 2 ~nb2.g!N 1 2 ~nb2.g!k A minor variant of the model describes
should be independent of body size. On the 1 2 (9) the mammalian respiratory system. Al-
other hand, the crossover point itself 1 2 ~nb2.g! 1 2 ~nb2.g! though pulse waves are irrelevant because
grows logarithmically: k } N } ln M. For the tubes are not elastic, the formula for Z is
humans, with n 5 3 (21), N ' 15 and This formula is a generalization of Eq. 4 quite similar to Eq. 8. The fractal bronchial
N ' 22 (assuming Nc ' 2 3 1010), where- and is dominated by the first term, which tree terminates in NA } M3/4 alveoli. The
as with n 5 2, N ' 24 and N ' 34. These represents the contribution of the large network is space-filling, and the alveoli play
values mean that in humans Poiseuille tubes (aorta and arteries). Thus, Vb } the role of the service volume accounting
flow begins to compete with the pulse nN11/3k } n4/3N, which, because it must for most of the total volume of the lung,
wave after just a few branchings, dominat- scale as M, leads, as before, to a 5 3/4. As which scales as M. Thus, the volume of an
ing after about seven. In a 3-g shrew, size decreases, the second term, represent- alveolus VA } M1/4, its radius rA } M1/12,
Poiseuille flow begins to dominate shortly ing the cubic branching of small vessels, and its surface area AA } rA2 } M1/6, so the
beyond the aorta. becomes increasingly important. This be- total surface area of the lung AL 5 NAAA }
The derivation of scaling laws based on bk havior predicts small deviations from M11/12. This explains the paradox (22) that
derived from Eqs. 7 and 8 (Fig. 2) leads to the quarter-power scaling (a * 3/4), observed AA scales with an exponent closer to 1 than
same results as before. For simplicity, assume in the smallest mammals (2). An expres- the 3/4 seemingly needed to supply oxygen.
that the crossover is sharp; using a gradual sion analogous to Eq. 9 can be derived for The rate of oxygen diffusion across an alve-
transition does not change the resulting scal- the total impedance of the system Z. It is olus, which must be independent of M, is
ing laws. So, for k . k, define bk [ b. 5 dominated by the small vessels (arterioles proportional to DpO2AA/rA. Thus, DpO2 }
n21/3 and, for k , k, bk [ b, 5 n21/2. This and capillaries) and, as before, gives Dp M21/12, which must be compensated for by a
predicts that area preservation only persists and u0 } M0. similar scaling of the oxygen affinity of he-
in the pulsatile region from the aorta In order to understand allometric scal- moglobin. Available data support these pre-
through the large arteries, at most until k ' ing, it is necessary to formulate an integrat- dictions (Table 1).
k. First consider the radius of the aorta r0: its ed model for the entire system. The present Our model provides a theoretical, mech-
scaling behavior is now given by r0 5 model should be viewed as an idealized anistic basis for understanding the central
rcbk.2Nb,k 5 rcn1/3N11/6k 5 rcn1/2N21/6N, zeroth-order approximation: it accounts for role of body size in all aspects of biology.
which gives r0 } M3/8 and, for humans, r0/rc many of the features of distribution net- Considering the many functionally inter-
' 104, in agreement with data (2) . Using works and can be used as a point of depar- connected parts of the organism that must
Eq. 3 we obtain, for the ratio of fluid ture for more detailed analyses and models. obey the constraints, it is not surprising that
velocity in the aorta to that in the capil- In addition, because it is quantitative, the the diversity of living and fossil organisms is
lary, u0 / uc 5 Nc(rc/r0)2 5 nN/3u0 / uc ' coefficients, Y0 of Eq. 1, can also, in prin- based on the elaboration of a few successful
250, independent of M, again in agree- ciple, be derived. It accurately predicts the designs. Given the need to redesign the
ment with data. Because g reflects the known scaling relations of the mammalian entire system whenever body size changes,
Table 1. Values of allometric exponents for variables of the mammalian with empirical observations. Observed values of exponents are taken from (2,
cardiovascular and respiratory systems predicted by the model compared 3); ND denotes that no data are available.
Cardiovascular Respiratory
Exponent Exponent
Variable Variable
Predicted Observed Predicted Observed
Aorta radius r0 3/8 5 0.375 0.36 Tracheal radius 3/8 5 0.375 0.39
Aorta pressure Dp0 0 5 0.00 0.032 Interpleural pressure 0 5 0.00 0.004
Aorta blood velocity u0 0 5 0.00 0.07 Air velocity in trachea 0 5 0.00 0.02
Blood volume Vb 1 5 1.00 1.00 Lung volume 1 5 1.00 1.05
Circulation time 1/4 5 0.25 0.25 Volume flow to lung 3/4 5 0.75 0.80
Circulation distance l 1/4 5 0.25 ND Volume of alveolus VA 1/4 5 0.25 ND
Cardiac stroke volume 1 5 1.00 1.03 Tidal volume 1 5 1.00 1.041
Cardiac frequency v 21/4 5 20.25 20.25 Respiratory frequency 21/4 5 20.25 20.26
Cardiac output Ė 3/4 5 0.75 0.74 Power dissipated 3/4 5 0.75 0.78
Number of capillaries Nc 3/4 5 0.75 ND Number of alveoli NA 3/4 5 0.75 ND
Service volume radius 1/12 5 0.083 ND Radius of alveolus rA 1/12 5 0.083 0.13
Womersley number a 1/4 5 0.25 0.25 Area of alveolus AA 1/6 5 0.083 ND
Density of capillaries 21/12 5 20.083 20.095 Area of lung AL 11/12 5 0.92 0.95
O2 affinity of blood P50 21/12 5 20.083 20.089 O2 diffusing capacity 1 5 1.00 0.99
Total resistance Z 23/4 5 20.75 20.76 Total resistance 23/4 5 20.75 20.70
Metabolic rate B 3/4 5 0.75 0.75 O2 consumption rate 3/4 5 0.75 0.76