Académique Documents
Professionnel Documents
Culture Documents
0 on (%0..%:)
and r(x) > 0 on (%9/2-2)/2) this group of individuals will get offspring with sizes
in the whole of (t0/2,x1/2), both in the proliferating and quiescent compartment.
It follows that for r sufficiently large, J(0)[4, v](x) > 0 a.e, in the permitted size
intervals
It is interesting to compare the behavior of solutions of the linear quiescence
‘model with the behavior of solutions of the fission model without quiescence [(26)~Se Gyllenberg and Webb
(28)]. A most remarkable difference occurs in the case of exponential individual
growth, In this case there is without quiescence no convergence toward a stable siz
distribution, whereas with quiescence the population has asynchronous exponential
growth, One can therefore say that quiescence has a stabilizing effect on the size
Uistribution, To determine whether the population grows or decays. it suffices to
consider the ordinary fission model without quiescence. Analyzing the character-
istic equations (30) and (40). one easily finds that
If A, > 0, then 0 Ay < Ab,
Tf ay <0, then Ay < Ay
This resuit means that in an exponentially growing tumor. introduction of quicse-
tence will slow down the growth rate of the tumor. but quiescence alone is not
capable of driving the cell population to extinction, In a decaying population
{quiescence cannot sive the population, only make it survive a little bit longer.
Vil. NONLINEAR QUIESCENCE MODEL
In this section we treat the full nonlinear model (16)~(20). Our approach is based
on the experimentally supported assumption that in large tumors the probability
of going into quiescence is higher than in small ones and that the probability of
returning to proliferation decreases as the tumor grows. Mathematically. this is
expressed by the requirement that r(x. N’) is increasing in N and pv. N))is deer
ing in N. Since the biological interpretation of r and p requires that r(x.N') is
bounded above by 2 and ptr.) is nonnegative (see Section IIL) the following
limits exist
roy fim rN) ext si
(9): Tim - (50)
p.(xy:= lim pix.N) Meret 51)
(x)= lim ph 5
‘The functions r, and p, describe the interaction between quiescence and protifer-
ation in very large tumors. We also introduce the functions
misers.0). eye
2
midis pv), exc tt 53)
n= plex 0). t
which can be interpreted as the transition functions for very small tumors
If one replaces the transition functions r(x,.N') and pox, N)) in (16), (17) by
vither r,(x) and p.(x). respectively, oF r(x) and p(x). respectively. one obtains
the linear quiescence model treated in Section IV. In the investigation of the
nonlinear model the idea is to use already proved properties of the linear semi-
group corresponding to that problem. We therefore define 3, and Ay as in (41)
but with 7(x0N). pix. N) replaced by r(x). pCa and rx). pul}. respectively
7. (1. 10 denotes the linear semigroup corresponding to the quiescence model
with transition Funetions r(x) and (0)Quiescence in Structured Population Dynamics 59
Theorem 8. Assume the following
(i) min p>0, min r>0.
(ii) r@,N) is nondecreasing in N, px
XE (X9/2, x2)
(iii) MAX 2meey2 {7 (UN) [re(x) ~ ror, NAN < 2,
maxy,2~x= «v2 St (UN) [0(,.N) ~ pa(x)] dN<=,
Then the following are true
(a) 1A.) >, then (pesgu]:= lime €*Tp(t-)eg(ts J] exists in Z,
where A. is the unique real root of \.(A.) = 1. Further.
Y) is nonincreasing in N for all
a> 0
(b) If 4.(0) <1, then (p(t, ).9(¢. )] is bounded in Z.
(6) If A-(0) <1 and (0) > 1. there is a nontrivial equilibrium solution
(d) If49(0) <1, the trivial equilibrium [0,0] is locally exponentially asymptoti-
cally stable,
The various types of asymptotic behavior in Theorem 8 are determined by the
values of (0) and Ao(0). These quantities correspond to the net reproductive
numbers for very large and very small tumors, respectively. If the net reproductive
number of @ large tumor is greater than I. then ultimately the solutions will
behave like those corresponding to the linear semigroup 7.(0). which according to
Theorem 7 has asynchronous exponential growth with Malthusian parameter A.
If, on the other hand, the net reproductive number of a large tumor is less than.
1. the tumor will remain bounded. Whether there in that case exists a nontrivial
equilibrium or not is determined by the net reproductive number for small tumors.
as is the stability of the trivial equilibrium.
The proof of Theorem 8 is again based on perturbation techniques and a
variation-of-constants formula. Let J(t), 120 be the nonlinear semigroup on Z
associated with the solutions of (16)-(20), that is. Fld. 0] = [ptt ). lt}
Then one easily finds the representation
Ft) = Flt~ ty Hey) + f Tle ~ 5) BN(3)) Ts) ds (ss)
where
AN Yb. AC)
= [2(r(4) ~ r(x. N)PB(2)(2K) + (—ON) = pe CO.
= 22) ~ ree. N))B(20 B28) ~ (BEN) = pM] (55)
Our last result is a structured analog of the nonlinear necrosis model of
‘Theorem 1.60. Gyllenberg and Webb
Theorem 9 Let p=0. wy =0 and min r>0. Assume that the function
(+ [p(t.). tt.) is continuously differentiable from [0. ) to Z. If r-x) > 1
¥€ [rd2. 4/2}. then O(a) is nondecreasing, lim,... Q(t) <. lim,... P(t) = 0, and
consequently, fim, .. G(t) = 0.
Vill, DISCUSSION
The size-structured quiescence model (16)~(20) can be regarded as a generalization
of both the unstructured quiescence model (3)-(6) and the size-structured cell
fission mode! without quiescence (26)-(28). This generalization has features that
fare missing in the simpler models. which come closer to reality and which aa
explain some phenomena observed.
‘The most important difference between the linear structured quiescence model
and the structured model without quiescence occurs when individual cells are
assumed to grow exponentially. In this case the quiescence model exhibits asym
chronous exponential growth, whereas in the model without quiescence the popu:
lation will grow exponentially but no convergence toward a stable size distribution
will take place. The reason for the lack of convergence in the model without
quiescence is that when individual cells grow exponentially. two sister cells will
together grow at the same rate as the mother cell would have done if she had not
divided, Thus it does not matter at what size a mother cell divides: the time at
which her daughters reach a given size is predestinated. In other words. there is
no dispersion in size. no information is lost and there is no ayynchronization,
Mathematically. this amounts to a lack of compactness of the semigroup associated
with the solutions. Quiescence provides a dispersal mechanism sufficient to make
the semigroup ultimately compact, thus implying asynchronous exponential
growth, At cell division some cells become quiescent and some do not. Quiescent
cells remain in this state for a time period that is a random variable. Information
about the initial population is indeed lost, and asynchronization takes place,
In many experiments individual exponential growth and asynchronous expo-
rnential growth of the populition have been observed simultaneously. The dis-
cussion above shows that quiescence is a mechanism allowing compatibility of
these two features. Of course, we do not claim that itis the only possible explana
tion, Heijmans (1984) showed that asynchronous exponential individual growth
are compatible in a model where fission does not result into to equal daughter
cells but where the daughter-mother size ratio is a smooth probability density
symmetrie around 1/2.
Next we compare the nonlinear structured quiescence model (16)~(20) with
its unstructured counterpart (3)-(6). In the case of necrosis we notice that structure
does not bring anything new into the model. The structured model ay described
by Theorem 9 predicts a behavior quite similar to that predicted by the unstruc-
tured model of Theorem 1. Moreover. the assumptions /, > of Theorem 1 and
(x) > L of Theorem 9 have the same biological interpretation
Looking at the unstructured model with true quiescence [7,(N)) =, = 0] we
obverve the decisive role played by uy. If there is no mortality in the quiescent
class (uy = 0). the tumor will grow without bound (Theorem 2), but even the
smallest positive mortality 4g will imply the existence of a globally asymptotically
Mable equilibrium (Theorem 3). Growth is a characteristic feature of an untreatedQuiescence in Structured Population Dynamics a
tumor and it seems very hard to find a biological justification for the stabilizing
effect of mortality in the quiescent class. Again the structured model seems to
come much closer to reality. Unlimited growth, boundedness, and existence of
nontrivial equilibria are not determined by wg alone but by the quantities 3.(0)
and (0). which involve all the vital rates (Theorem 8). If. for instance. 3..(0) > 1
and the population grows ultimately exponentially, a small perturbation of uo (or
of any other of the vital rates) will not change this mode of growth, Similar
statements are valid in situations (b), (c). and (d) of Theorem 8. The structured
model therefore has a kind of robustness that the unstructured model lacks.
ACKNOWLEDGMENT
The research by Glenn F. Webb was supported by the National Science Foundation
under Grant DMS-8922947,
REFERENCES
Bell, G. I. and Anderson. E. C. (1967). Cell Growth and Division. I. A Mathematical
‘Model with Applications to Cell Volume Distributions in Mammalian Suspension
Cultures. Biophys. J. 7: 329-351
Bertuzzi. A., Gandolfi, A., and Giovenco. M. A. (1981). Mathematical Models of the Cell
Cycle with a View t0 Tumor Studies. Math. Biosci. $3: 159-188,
Burton. A. C. (1966). Rate of Growth of Solid Tumors as a Problem of Diffusion, Grow:h
30: 157-176,
Clément, Ph.. Heijmans, H. J. A. M., Angenent, S.. van Duijn, C.1.. and de Pagter. B.
(1987). One-Parameter Semigroups, CWI Monograph 5. North-Holland, Amsterdam,
Diekmann O.. Heijmans, H. J. A. M.. and Thieme, H. R. (1984). On the Stability of the
Cell Size Disteibution, J. Math. Biol, 19: 227-248,
Dickmann, O.. Heijmans, H. J. A. M., and Thieme. H. R. (1986), On the Stability of the
Cell Size Distribution. I. Time-Periodic Development Rates. in Hyperbolic Partial
Differential Equations 111 (M, Witten, ed.), International Series in Modern Applied
Mathematics and Computer Science. Vol. 12. Pergamon Press, Elmsford, N-Y.. pp.
491-812
Feller. W. (1939). On the Logistic Law of Growth and its Empirical Verifications in Biology
‘Acta Biotheor. 5: S166
Frindel, E., and Tubiana, M. (1971). Radiobiology and the Cell Cycle. in The Cell Cycle
and Cancer (R. Baserga. ed.). Marcel Dekker, New York. pp. 391-447,
Gavosto, P.. and Pileri, A. (1971). Cell Cycle of Cancer in Man, in The Cell Cycle and
Cancer (R. Baserga, ed.). Marcel Dekker, New York, pp. 99-128.
Greenberg, W.. Protopescu, V., and Van det Mee, C. V.M. (1987). Boundary Value
Problems in Abstract Kinetic Theory: Operator Theory: Advances and Applications in
Mathematics 23. Birkhauser Verlag, Basel.
Greiner, G., and Nagel, R. (1988). Growth of Cell Populations via One-Parameter Semi
groups of Positive Operators. in Marhemarics Applied to Science. Academic Press. New
York. pp. 79-108.
Gyllenberg. M., and Webb. G. F. (1987). Age-Size Structure in Populations with Quiese
fence. Math. Biosci, 8: 67-95,
Gyllenberg. M.. and Webb. G. F_ (1989). Quiescence as an Explanation of Gopertzian
‘Tumor Growth. Growth Dev. Aging 53
Gyllenberg, M.. and Webb. G. F. (to appear). A Nonlinear Structured Population Model
(of Tumor Growth with Quiescence. J. Math. Biol.02 Gyllenbery and Webb
Hejjmans, H.J. A. M. (984), On the Stable Size Distribution of Populations Reproducing
by Fission into Two Unequal Parts, Math, Bioses, 72: 19-S0,
Heijmans. H. J. A. M. (1986). Structured Populations, Linear Semigroups and Positivity
Marl. Z. 191; 599-617.
Laird, A. K, (1964), Dynamics of Tumor Grovth, Br. J, Cancer 18° 490-802
Laird, A. K. (1965). Dynamies of Tumor Growth: Comparison of Growth Rates and
Extrapolation of Growth Curve to One Cell. Br. J. Canter 19: 278-291.
Martinez. A. O. and Griego. R. J. (1980). Growth Dynamies of Multicell Spherioids From
‘Thrve Murine Tumors, Growel 44: [12-122
MeCredic. J. A.. Inch, W. R.. Keuuy. J. aad Watson, T. A. (1965), The Rate of Tumor
Growth in Animals, Growl 29: 331-347,
Mew. J. A. J.. and Diekmann, 0. (1986). The Dynamics of Physiologically Structured
Populations. Springer Lecture Notes in Biomathematis 68. Heidelbery.
Nagel, R. ed.) (1986). One-Parumeter Semigroups of Positive Operators, Springer Lecture
Notes in Mathematics 1184, Springer-Verlag. Heidelberg,
Norton, L.. Simon, R.. Brereton. H. D.. and Bogden, A. E, (1976), Predicting the Course
‘of Gompertzian Growth, Nature 264: 542-835,
Pacey. A, (1983). Semigroups of Linear Operaiors and Applications tr Partial Differential
Eiguations, Springer-Verlag. Heidelberg.
Rotenberg. M. (1978). Correlations. Asymptotic Stability and the Gi, Theory of the Cell
Cycle. in Biomurhematies and Cell Kinetics, Developments in Cell Biologs, Nol. 2 (4.
4. Valleson and P. D. M, Macdonald. eds). Elsevier! North-Holland Biomedical Press.
Amsterdam. pp. §9-67,
Rotenberg. M. (1982). Theory of Distributed Quiewent State in the Cell Cyele. J. Theor
Biol. 96: 495-509.
Rotenbeeg, M. (983). Transport Theory for Growing Cell Populations, J. Theor. Biot
103: INL-199,
Sinko. J. W..and Stevifer, W. (1967), A New Model for Age-Size Structure of a Population
Feology 48: 910-918.
Sinko. J. W.. and Sueifer, W. (1971), A Modet for Populations Reproducing by Fission
Ecology 52: 380-38,
Skipper. H. E, (1971). The Cell Cycle and Chemotherapy of Cancer. in The Cell Cle
‘and Cancer (R. Baverga. ed.). Marcel Dekker. New York. pp. 388-387,
tannock. LF. (1968), The Relation Between Cell Proliteration and the Vascular System
in a Transplanted Mouse Mammary Tumour. Br. J. Cunicer 22: 288-273,
‘Touchiya, H. M., Fredrickson, A. G.. and Aris, P. (1966), Dynamies of Microbial Coll
Populations. Adv. Chem. Eng. 6! 125-198,
Tubiana, M. (1971). The Kinetics of Tumour Cell Proliferation and Radiotherapy. Br: J
Radiol, 44; 325-M7.
Tubiana, M. (982). L. H, Gray Medal Lecture: Cell Kinetics
Int. J Radiat. Oncol. Biol. Phys. 8: \471=1480,
Webb. G. F, (1988) Theory of Nonlinear Age-Dependemt Population Dynamics. Marcel
Dekker, New York
Radi
jon Oncology5
Altruistic Population Model
with Sex Differences
YING-HEN HSIEH
National Chung-Hsing University, Taichung. Taiwan, Republic of China
1. INTRODUCTION
‘The evolution of altruism in ecosystems is an intriguing phenomenon in biology
We define “evolution” as having a fixed ratio of the population showing one
particular characteristic. such as performing altruistic acts. The problem is interest-
ing and complicated because (1) the evolution of a self-sacrificing subpopulation
seems to be a direct contradiction to the theory of evolution—the altruistic acts
usually decreases an individual's fitness to survive, and (2) the altruistic act per-
formed can be. at times, age-dependent, sex-dependent. space-dependent. or
group-dependent. The literature on altruism includes Hamilton (1964). Maynard
Smith (1965. 1980). Trivers (1971). Cavalli-Sforza and Feldman (1978). Feldman
and Cavalli-Sforza (1981), and Akin (1984),
Population geneticists in general believe that altruism was able to evolve in a
community due to one of the three factors: reciprocal altruism, kin selection. oF
group selection (see Akin. 1984). but the relative importance of these three factors,
is still very much in controversy. The purpose of this work is not to ask why
altruism evolves. but how. We will use simple mathematical models to find the
circumstances under which altruism is able to evolve in certain communities but
rot in others. It is easy to foresee that the conditions will be both physiological
and environmental
In previous works by the author (Hsieh, 1988, 1989a), a model of coupled
Von Foerster equations of the form
Dea.t) + ela peta.t) + mara. stat) + ¥(a.0]
yilayx"(a.t)
a
© xa.t) + vat)
Dy(a.t) + aslay(a.t) + maysta.d[xta.t) + sa.)
xla)v(a. x(a.)
x(a.t) + (act)
150, a€ [0.1] 2)6 Hsielt
was used to study the possibility of evolution of altruism in a community with the
numbers of altruistic and selfish individuals represented by separate differential
equations. [For a discussion on models of this type. see Coleman (1978).| The
conditions for the persistence of the altruistic group in the population was obtained,
under the assumption of synchronized reproduction in a semelparous population
‘where reproduction occurs once in each individuals lifetime, In @ subsequent work.
(Hyich, 19895), it was also shown that under the assumption of indiscriminate
altruism for all individuals in the community. the evolution of altruism is necessiry
but not sufficient for the persistence of the community itself
‘The Von Foerster equation is generally associated with one-sex models where
only the female members of the population are considered. However. in some
altruistic communities with intricate structure such as social insects. there is usually
«difference in the performance of altruistic act between male and female. &s well
as different reproductive rates for each sex. One example is the social bees
(honeybees and bumblebees). where different forms of altruistic acts are per
formed by the (adult) male and female of the species in a complex social structure
Headed by the queen bee, whose sole function is to lay eggs (up to 20K) a dy)
the community is comprised of female workers. which build cells for the eggs and
collect enough pollen to feed the larva until it becomes a flying adult hee: male
bees, which fertilize the young queens: and the parasitic (nonaltruistic) bees. whi
do not build hives and develop in the cells of the host working bees.
In Hsieh (1988), we get around this problem by prescribing a fecundity fune-
tion that is general enough to account for the discrepaney in the sex ratio of
succeeding generations. But to deal with the different levels of altruism. we
introduce separate equations for the male and female members of the altruistic,
group so that distinct altruistic functions and fecundity functions can be assigned.
accordingly to different sex groups. However. we will not distinguish sex among.
the yelfish members of the community’ since both sexes benefit equally from
altruistic acts. Therefore. not only is the difference in ability to perform altruistic
act considered. but we can also take into account the possible change in sex ratio,
from generation to generation. The model in question is as follows:
Daxta.t) + alana.) + maysla.).N (a0)
_ narnia. nate. + a0} _
»
Nat) ‘
Dela.) + asaya. + masta. N(a.0)
= =0 ow
Dela.t) + aladeta.t) + masta. N(a.0)
Asta. + sa] 6. ae [0a] 6)
Na
voy. ¢ denotes the number of male altruists, female altruists, and nonaltruists.
respectively, and N= x + y'+ : is the total population.
In Section II we reduce the equations (3)-(S) into simpler system for the
altruistic ratio so that in Section III we can yield analytical results on sufficientAlvruistie Population Model with Sex Differences 65
conditions for a fixed sex ratio in the altruistic group. Finally. we will run computer
simulations for different cases described previously in the text and discuss the
biological and ecological implications of our results.
Although the main theme is sex differences in altruistic acts. the model can
be used to describe any situations where any three subgroups of a population
performing distinct social behavior (not necessary altruism). It also allows different
levels of mortality and reproduction for each subgroups. Hence the analysis could
have wider application in studying the complicated communities of semelparous
insets.
Ml, MODEL FOR ALTRUISTIC RATIO
Let X,(a). ¥a(a). Z,(a) be the numbers of altruistic males, altruistic females. and
selfish individuals, respectively. of age a at the nth generation. By the assumption
‘of semelparous population with synchronized reproduction, equations (3)
be simplified (see Coleman. 1978. or Coleman and Hsieh. 197) as follows
Xia) + a(a)X,(a) + m(a)X,(a)N,Aa)
Hala) + Yala) y
) ean
= nla (a) =0 (a
70) Kula) (6a)
Ya) + ax(a)¥,a) + w3(a) ¥.ladNala)
Kuta) + Yala)
= (a) MO Val y (gy = 0 (6%
(Naa) ' i
Zila) + asla)Z,fa) + msla)Za(0)N,(2)
~ ya) ** yn ela) =0, Osast (6)
Here N,(a) = X,(a) + ¥,(a) + Z,(a) is the total population number of the mth
generation at age a. the a,’s are mortality functions. the 7's denote the effect of
dispersal and migration. the ,’s are “altruistic functions” from [0. 1] + R which
describe the effects of altruism on each group (hence a group that incurs a loss in
fitness due to altruism performed may have a negative altruistic function). and the
‘generation length is normalized to unity
We assume further that the rate of dispersal and migration is the same for all
groups (i.e. m= =m). Ih we let
X,fa) + Yala)
fala) o
Nata)
be the altruistic ratio in the mth generation of age @ and
Yuta
sola) = ele) — 6)
Xa) + ¥.(@)
be the ratio of females among the altruists of the ath generation and age a. we
can combine (6a)~(6¢) to get the corresponding equations satisfied by f(a) and
gu(a) for 0=a= 1 and n €Z.66 Hsieh
Pad ~ fla = fl@Malad ~ e(aigylad
+ Fulapgaalfnla) — Fala) fala] = 0 (a)
gala) + gala ~ gaCa)lldstar ~ Fola)faCa] = 0 (9b)
with
Kila) = a(a) ~ ax(a) au
dia) = asta) ~ ala) ap
Fula) = yala) ~ yal) «2p
Fa) = yela) ~ yaad a3)
‘To equations (9). we add the fecundity functions Fy and Fy for the altruistic
ratio and the ratio of females within the altruistic group. respectively.
Fe (0) = Fulfal)) (1s)
00> Fag) (140)
The role of female reproduction as an altruistic act does not appear. althou
it certainly qualifies as altruism. However, a consideration of female reproduction
can be incorporated into the altruistic function for female altruists: hence the
females claysified in the selfish group are those females that do not perform
altruistic tasks and reproduce no offsprings. We can then look tor sutficient con-
ditions for the evolution of altruism as well as a fixed ratio of trvists in
the population using the fixed-point theory developed by Coffman and Coleman
(1978)
male
Ill ANALYSIS OF THE MODEL
Equations (9) can be rewritten as
Pale) + pil Fale). ale)-a) = 00 «say
eile) + pal Fala). alata) = 0. neZ (usb)
where
pult.y.a) = —x(1 = xyfeia) ~ ds(ayy + ¥4(ayay — ¥(arx] (ea)
pals. vod) = C1 ~ v)feista) — Fa(aday (166)
dare the loss functions for f,, and g,.
Since the conditions for evolution of altruism were given in Hyich (1988), the
first question we pose is the following: Suppose that the altruistic ratio for adults
at cach generation f',(1) becomes fixed and positive at euch generation as 1 —
What is the ratio of female altruists in the altruistic group”
It we defined the generation growth function G: [0.1] [0.1] for the adult
Female ratio among altruists g,(1) as follows:
Kall) = Glee MDD EZ an
where the initial value g,(Q0) of the IVP (15b) is given by (4b), the conditions forAltruistic Population Model with Sex Differences or
the existence of a fixed point of (17) are given as follows:
Theorem 1. Suppose that 0