1: \documentclass[aps,pre,superscriptaddress]{revtex4}
2: \usepackage{amssymb,amsmath,epsfig}
3: \begin{document}
4: \title {Extinction dynamics of Lotka-Volterra ecosystems on evolving networks}
5: \author{F. Coppex}
6: \affiliation{Department of Physics, University of Geneva, CH 1211 Geneva 4, Switzerland}
7: \author{M. Droz}
8: \affiliation{Department of Physics, University of Geneva, CH 1211 Geneva 4, Switzerland}
9: \author{A. Lipowski}
10: \affiliation{Department of Physics, University of Geneva, CH 1211 Geneva 4, Switzerland}
11: \affiliation{Faculty of Physics, A.~Mickiewicz University, 61-614 Pozna\'{n}, Poland}
12:
13: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
14: \pacs{}
15: \begin {abstract}
16: We study a model of a multi-species ecosystem described by
17: Lotka-Volterra-like equations.
18: Interactions among species form a network whose evolution is determined
19: by the dynamics of the model.
20: Numerical simulations show power-law distribution of intervals between
21: extinctions, but only for ecosystems with sufficient variability of
22: species and with networks of connectivity above certain threshold that is
23: very close to the percolation threshold of the network.
24: Effect of slow environmental changes on extinction dynamics, degree
25: distribution of the network of interspecies interactions, and some
26: emergent properties of our model are also examined.
27: \end{abstract}
28: \maketitle
29: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
30: \section{introduction}
31: Lotka-Volterra models of interacting species have a well established
32: role in population ecology~\cite{LV}.
33: Being inspired by an oscillatory behavior in some prey-predator
34: systems,
35: they are typically used to model populations on time scale shorter than
36: lifetime of describing species.
37: It means that long-term properties of ecosystems (macro-evolution) are
38: usually not captured within such an approach.
39: On the other hand, models used to describe macro-evolution very often
40: use the dynamics that operates at the level of species rather than
41: individuals.
42: Such coarse-grained models usually refers to the notion of fitness of a
43: species that is not commonly accepted~\cite{BAKSNEPP}.
44:
45: Recently, there has been some attempts to study macro-evolution using
46: models equipped with dynamics that operates at the level of
47: individuals~\cite{STAUFFER,ZIA,HALL}.
48: Taking into account that Lotka-Volterra models are relatively
49: successful in describing many aspects of population dynamics it would be
50: desirable to apply such an approach also to macroevolution.
51: Some time ago Abramson introduced a discrete version of Lotka-Volterrra
52: ecosystem~\cite{ABRAMSON} and studied certain characteristics of
53: extinctions.
54: His model is an example of a one-dimensional food chain with $M(\sim
55: 100)$ trophic levels and a single species occupying a given trophic
56: level.
57: Since in realistic food webs $M\sim 4-6$ with typically many species
58: belonging to a given trophic level~\cite{MCKANE,QUINCE}, these are highly
59: nonrealistic assumptions.
60: Nevertheless, extinction dynamics in Abramson's model shows some
61: features that are characteristic to Earth biosystem.
62:
63: In the present paper we introduce a Lotka-Volterra model that describes
64: a simplified ecosystem of $N$ species of predators and one species of
65: preys.
66: Our model can be thus considered as a simple food web model with only
67: two trophic levels.
68: Competition between predator species is described by a certain
69: network~\cite{ALBERT} of interactions whose evolution is coupled with
70: dynamics of the model.
71: Namely, when a certain species becomes extinct (i.e., its density falls
72: below a certain threshold) it is replaced by new species with a newly
73: created set of interactions with some of existing species.
74: Despite obvious simplifications the model exhibits some properties that
75: are typical to more complicated ecosystems, as for example power-law
76: distributions of intervals between extinctions.
77: Within our model we can also examine how robust this power-law
78: distribution is.
79: We find that under certain conditions, as for example very sparse
80: interactions between species, or too strong dominance of a small group of
81: species, these power-law characteristics disappear and the model is
82: driven into a regime where extinctions have exponential distributions or
83: where there are no extinctions and the ecosystem enters a steady state.
84: In our opinion, such regimes might be relevant when a restricted
85: (either in space or time) evolution of an ecosystem or its part is studied.
86: Interestingly, a threshold value of connectivity that separates
87: power-law extinctions and steady state is very close to the percolation
88: threshold of the random network of inter-species interactions.
89:
90: According to a large class of statistical physics
91: models of biological evolution, avalanches of extinctions do not
92: require external factors to trigger them, but might be a natural consequence
93: of the dynamics of an ecosystem.
94: As a result, these external factors, as e.g., climate changes, solar
95: activity or impact of a big meteorite, are very often neglected in such
96: studies~\cite{ROBERTS}.
97: But such factors certainly affect the ecosystem and there is a good
98: evidence of it~\cite{NEWMAN}.
99: Let us emphasize that even the basic mechanism that triggers avalanches
100: of extinctions is not known and is a subject of an intensive
101: multidisciplinary
102: debate~\cite{newman_palmer_book}.
103:
104: One possibility to take external factor(s) into account in our model is
105: to modify a growth rate of prey.
106: Since dynamics of the model is nonlinear, such a change might have more
107: dramatic consequences than merely a change of densities of species.
108: And indeed we noticed that dynamics of extinctions is strongly
109: dependent on the growth rate.
110: It turns out, that in our model abundance of preys leads to a larger
111: frequency of extinctions, and in periods of hunger there are less
112: extinctions.
113: This is clearly due to nonlinearity of the dynamics.
114: Larger growth rate increases the density of preys that in turn
115: increases densities of predators.
116: With increased densities, dynamics becomes more competitive and
117: extinctions become more frequent.
118: Such a periodically modulated growth rate leaves some traces also in
119: the probability distribution of extinctions.
120: It might be interesting to notice that paleontological data also show
121: some traces of periodic events, but their proper understanding is still missing~\cite{NEWMAN,RAUP}
122:
123: During evolution some species are favored and selected at the expense
124: of less fortunate ones.
125: Evolution constantly searches for best solutions that resembles an
126: optimization process.
127: For example a large size of organisms of a given species might be of
128: advantage in some situations, but might cause some problems in the
129: other.
130: What might be Nature's solution to this problem?
131: Will it be middle-size species or rather two groups of species sitting
132: at the extremes of conflicting requirements?
133: In our opinion, this aspect of evolution is also often omitted in
134: models of macroevolution.
135: Within our model we looked at such emergent properties of species
136: selected by evolution.
137: It turns out that depending on some dynamical details, our model can
138: reproduce both types of solutions of such an optimization problem.
139:
140: In Section II we introduce our model and briefly describe the
141: numerical method we used.
142: Obtained results are presented in Section III.
143: In Section IV we summarize our results and suggest some further
144: extensions of our work.
145: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
146: \section{Model and numerical calculations}
147: We study a Lotka-Volterra ecosystem that consists of $N$ species of
148: predators with densities $\rho_i\ (i=1,2,\ldots ,N)$ who are all feeding
149: on one species of preys with density $\rho_0$.
150: We assume that each predator species $i$ is characterized by a
151: parameter $k_i$ ($0<k_i<1$) that enters evolution equations of the
152: model through death and growth terms
153: \begin{subequations}
154: \label{lveq}
155: \begin{eqnarray}
156: \dot{\rho_0} & = & g(t)\rho_0(1-\rho_0)-\frac{\rho_0}{N}\sum_{i=1}^N
157: f(k_i)\rho_i\label{lv1}\\
158: \dot{\rho_i} & = &
159: -d(k_i)\rho_i(1-\rho_0)+f(k_i)\rho_i\rho_0\left(1-\frac{k_i\rho_i+\sum_{j}' k_j\rho_j}{k_i+\sum_{j}' k_j}\right),
160: \label{lv2}
161: \end{eqnarray}
162: \end{subequations}
163: where $i=1,2,\ldots,N$.
164: In our model we assume that species interact mainly through
165: environmental capacity terms (the last term in Eq.~(\ref{lv2})).
166: Namely, the growth rate of a given species $i$ is reduced not only due
167: to its density but also due to weighted (with the factor $k$)
168: densities of a group of randomly selected neighboring species.
169: In Eq.~(\ref{lv2}) summation over these neighboring species is denoted
170: by ($\sum'$).
171: Approximately, we might interpret the coefficient $k_i$ as the size of
172: organisms of $i$-th species -- the bigger they are the bigger their
173: role in the environmental capacity term.
174: We also assume that the growth rate of preys is corrected by the
175: environmental capacity term and due to external factors might be a slowly
176: varying function of time ($g(t)$).
177: In principle, external factors might affect also other terms of model
178: (\ref{lveq}), but for simplicity we restrict its influence only to the
179: growth rate of preys.
180: Functions $d(k)$ and $f(k)$ reflect the $k$-dependence of death and
181: growth of our species.
182: Explicit form of functions $g(t),\ f(k)$ and $d(k)$ will be given
183: later.
184:
185: Differential equations (\ref{lveq}) are solved using Runge-Kutta
186: fourth-order method.
187: Multi-species Lotka-Volterra ecosystems were subject to intensive
188: studies since the pioneering work of May~\cite{MAY}.
189: It is known that such systems might evolve toward the steady-state with
190: positive densities.
191: However, in some cases, in the steady state density of some species
192: might be zero.
193: Each time a density of a certain species in model (\ref{lveq}) drops
194: below a threshold
195: value which we fix as $\varepsilon=10^{-7}$ we consider such a species
196: as extinct~\cite{EPSILON}.
197: Such a species is then replaced by a new species with a randomly
198: assigned density (from the interval (0,1)), the coefficient $k$ ($0<k<1$)
199: that is randomly drawn from the distribution
200: $p(k)$, and a new set of neighbors (all links of the 'old' species are
201: removed).
202: With such rules the model rather describes $N$ niches, and we assume
203: that a time to create a species that will occupy a niche is relatively
204: short comparing to the typical lifetime of species.
205:
206: We assume that a newly created species makes $z$ links with randomly
207: selected neighbors.
208: Links are not directional so a newly created species will also enter
209: the evolution equation of species it is neighboring.
210: If the extinct species would be chosen randomly the network of
211: interactions would have been a random graph.
212: However, it is the dynamics (\ref{lveq}) that determines which species
213: are extinct.
214: Thus, extinct species are not selected randomly and the resulting
215: network is in general not a random graph.
216: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
217: \section{Results}
218: In the following we describe numerical results obtained for some
219: particular cases of model (\ref{lveq}).
220: \subsection{Intervals between extinctions}
221: Various paleontological data suggest that dynamics of extinctions has
222: some power-law distributions of sizes or durations~\cite{NEWMAN}.
223: In our model we measured time intervals $t$ between successive
224: extinctions.
225: In this calculations we used a constant growth term of preys
226: $g(t)\equiv 1$.
227: We examined two cases: (i) model I: $f(k_i)\equiv 1,\ d(k_i)\equiv 1$
228: and (ii) model II: $f(k_i)=k_i ,\ d(k_i)\equiv 1$.
229: Unless specified otherwise we select $k_i$ randomly with a homogeneous
230: distribution on the interval (0,1) ($p(k)=1$).
231: Our results are shown in Fig.~\ref{intervals}.
232: In the simplest case, model I with $z=4$ and $k_i\equiv 1$ (i.e., all
233: species during the evolution have identical $k_i(=1)$) we obtain
234: exponentially decaying distribution of intervals between extinctions $P(t)$.
235: Such a decay is also seen for model I (z=4) with linear distribution of $k_i$ namely $p(k)=2k$.
236: We expect that such a behavior appears when a distribution of $k_i$ in
237: the ecosystem is relatively narrow and shifted toward unity.
238: Such an effect might be due to the small width of distribution $p(k)$
239: (i.e., a distribution from which we draw $k_i$) or might be dynamically
240: generated
241: as in model II.
242: In this case even though $k_i$ are chosen from a homogeneous
243: distribution, the dynamics favors large $k_i$ species (due to their larger growth
244: rate) and they dominate the ecosystem.
245: When the distribution of $k_i$ in the ecosystem is more uniform (model
246: I with $p(k)=1$) our simulations suggest that $P(t)$ decays as a power
247: law.
248: Let us notice, however, that a power-law behavior is seen only on
249: approximately one decade and we cannot exclude that on a larger time scale a
250: different (perhaps exponential) behavior appears as was already
251: observed in some other macroevolutionary models~\cite{STAUFFER}.
252: Let us also notice that for model I with $p(k)=k^{-1/2}/2$ the
253: power-law distribution $P(t)$ seems to decay as $t^{-2}$, i.e., with the
254: exponent consistent with some paleontological data~\cite{NEWMAN} as well as
255: with predictions of some other models~\cite{ZIA}. However, one has to
256: recognize that the error bars on experimental data are rather large and
257: that a non-power law behavior cannot be excluded.
258: %%%%%%%%%%%%%%%%%%%%
259: \begin{figure}
260: \centerline{\epsfxsize=0.7\columnwidth
261: \epsfbox{fig2.eps}
262: }
263: \caption{
264: Probability distribution of intervals between successive extinctions
265: $P(t)$ calculated for some particular cases of model (\ref{lveq}) for
266: $N=100$.
267: Inset shows the same data but plotted on a lin-log scale.
268: }
269: \label{intervals}
270: \end{figure}
271: %%%%%%%%%%%%%%%%%%%%%%%
272:
273: Note that a power-law decay of $P(t)$ is seen only for sufficiently
274: large $z$. When $z$ is too small, we observed that the ecosystem enters the
275: steady state where all $\rho_i$ are positive and there are no
276: extinctions.
277: This is probably due to the fact that the competition among predators
278: is too weak (or rather too sparse).
279: To examine the transition between this two regimes in more detail we
280: measured the averaged time between extinctions $\tau$ and the results are
281: seen in Fig.~\ref{tau}.
282: One can see that $\tau$ diverges around $z\sim 1.8$~\cite{COMMENT}.
283: Such a value of the threshold parameter suggests that this transition
284: might be related with the percolation transition in our network of
285: interspecies interactions.
286: To examine such a possibility we measured the average size of the
287: largest cluster of connected links in the network $R$ (normalized by the
288: number of species $N$) and the results are shown in Fig.~\ref{tau}.
289: Vanishing of this quantity locates the percolation
290: transition~\cite{STAUFFERAHARONY}.
291: One can see that the percolation transition takes place at a larger
292: value namely around $z\sim 2.0$.
293: Our results suggest that these two transitions take place at different
294: values of $z$.
295: However the analysis of finite size effects especially in the
296: estimation of $\tau$ is rather difficult and we cannot exclude that these two
297: transitions actually overlap, as might be suggested by their proximity.
298: Such a result would show that the dynamical regime of an ecosystem
299: (i.e., steady state or active with power-law distribution of extinctions)
300: is determined by the geometrical structure of its interactions.
301: %%%%%%%%%%%%%%%%%%%%
302: \begin{figure}
303: \centerline{\epsfxsize=0.7\columnwidth
304: \epsfbox{tau.eps}
305: }
306: \caption{
307: The inverse average time between extinctions $\tau^{-1}$ and the
308: percolation probability $R$ as a function of $z$.
309: Plotted results are based on calculations for $N=100$, 200, 300 and 400
310: and extrapolation $N\rightarrow\infty$.
311: }
312: \label{tau}
313: \end{figure}
314: %%%%%%%%%%%%%%%%%%%%%%%
315: \subsection{Effect of a modulated growth rate}
316: Now we examine the role of a modulated in time growth rate of preys.
317: Such a modulation is supposed to mimic the influence of an external
318: factor like a change of a climate.
319: One of the questions that one can ask in this context is how such a
320: change affects the extinction dynamics.
321: We studied model I with $p(k)=1$ and $d(k_i)\equiv 1$.
322: The growth rate of preys we chose as $g(t)=1+A{\rm sin}(2\pi t/T)$,
323: where
324: $A$ and $T$ are parameters.
325: A typical behavior in case of model I with such a growth rate is shown
326: in Fig.~\ref{modul0}.
327: %%%%%%%%%%%%%%%%%%%%
328: \begin{figure}
329: \centerline{\epsfxsize=0.7\columnwidth
330: \epsfbox{fig3.eps}
331: }
332: \caption{
333: A time evolution of the density of preys $\rho_0$, average density of predators $\rho_a=\frac{1}{N}\sum_{i=1}^{N}$, and the number of extinctions $M$ (divided by 20) in the time interval $\Delta t=10^3$ for the model I with $N=100$ and $z=4$. A rescaled modulated growth rate $(g(t)-1)/10=0.09 \sin(\frac{2\pi t}{T})$ ($T=10^5$) is also shown.
334: }
335: \label{modul0}
336: \end{figure}
337: %%%%%%%%%%%%%%%%%%%%
338: One can see that increased growth rate increases the density of preys
339: $\rho_0$ that increases the density of predators.
340: However, it increases also the frequency of extinctions.
341: Such a behavior, namely increased extinction rate during abundance of
342: food, might at first sight look as counterintuitive.
343: This effect is related with the form of environmental capacity terms
344: in
345: in the growth rate in Eq.~(\ref{lv2}), namely
346: $1-(k_i\rho_i+\sum_{j}' k_j\rho_j)/(k_i+\sum_{j}' k_j)$.
347: Such term certainly has a larger variability for increased density of
348: predators $\rho_i$, and for some species (depending on the distribution
349: of links, coefficients $k_i$ and densities) it causes faster
350: extinction.
351: Let us also notice that since period of modulation $T$ is quite large,
352: there is no retardation effect between density of preys and predators.
353: We observed such retardation for smaller values of $T(\sim 1000)$.
354:
355: Modulated growth rate of prays affects also the probability
356: distribution of intervals between extinctions $P(t)$ as shown in
357: Fig.~\ref{modul}.
358: One can see that period of modulation $T$ is imprinted in $P(t)$.
359: Let us notice that certain paleontological data do show some signs of
360: periodicity but its origin still remains unclear~\cite{RAUP,NEWMAN}.
361:
362: It is known that slowly changing ecosystems sometimes undergo
363: catastrophic shifts~\cite{SCHEFFER}.
364: As a result, the ecosystem switches to a contrasting alternative stable
365: state.
366: It would be interesting to examine whether multi-species ecosystems, as
367: described by our model (\ref{lveq}), might also exist in such
368: alternative states.
369: If so, one can ask whether for example structure of the network of
370: interspecies interactions or extinction dynamics are the same in such
371: states.
372: %%%%%%%%%%%%%%%%%%%%
373: \begin{figure}
374: \centerline{\epsfxsize=0.7\columnwidth
375: \epsfbox{fig1.eps}
376: }
377: \caption{
378: Probability distribution of intervals between successive extinctions
379: $P(t)$ calculated for model I with modulated growth rate ($N=100$).
380: }
381: \label{modul}
382: \end{figure}
383: %%%%%%%%%%%%%%%%%%%%%%%
384: \subsection{Emergent properties of species}
385: It might be interesting to ask what are the characteristics of species
386: that are preferred by the evolution in our ecosystem.
387: Since species are characterized only by the number $k_i$ it is
388: equivalent to calculating the distribution of $k_i$ in the steady state of
389: model (\ref{lveq}).
390: Of course, due to selection this distribution in general will be
391: different than the distribution $p(k)$, i.e., the distribution from which we
392: draw
393: $k_i$ of a newly created species.
394: Some of our results are shown in Fig.~\ref{emer} (all results for
395: $g(t)\equiv 1$,\ z=4,\ N=100).
396:
397: In the case of model I ($f(k_i)\equiv 1,\ d(k_i)\equiv 1$) with
398: homogeneous initial distribution of $k_i$ ($p(k)=1$) one can see that the
399: steady state distribution is also approximately homogeneous
400: (with a slight bias favoring small-$k$ species).
401: We checked that model I shows this behavior also for other
402: distributions $p(k)$ (what you put is what you get).
403: Different behavior appears for model II ($f(k_i)=k_i,\ d(k_i)\equiv
404: 1$).
405: In this case the growth rate factor $f(k_i)$ of $i$-th species is
406: proportional to $k_i$ that certainly prefers species with large $k_i$.
407: Numerical results for homogeneous distribution $p(k)=1$ confirm such a
408: behavior (Fig.~\ref{emer}).
409: We observed similar strong preference of large $k_i$ species also for
410: Model II with other distributions $p(k)$.
411:
412: We also examined the selection pattern in presence of some competing
413: effects.
414: To compensate a strong preference toward large-$k$ species we made
415: simulations for our model with $f(k_i)=k_i,\ d(k_i)=\sqrt{k_i}$ and
416: $p(k)=1$.
417: Such term reduces the death rate of small-$k$ species.
418: Our results show (Fig.~\ref{emer}) that in this case distribution of
419: $k_i$ has two maxima at the extremities of the interval (0,1).
420: On the other hand with the same model but for
421: $d(k_i)=(1-\rho_0)^{-k_i}$ (that also reduces the death rate of small-$k$ species) we obtain a
422: distribution with a single maximum around $k=0.45$.
423: It would be desirable to understand the origin of the qualitative
424: difference between these two cases.
425: %%%%%%%%%%%%%%%%%%%%
426: \begin{figure}
427: \centerline{\epsfxsize=0.7\columnwidth
428: \epsfbox{fig4.eps}
429: }
430: \caption{
431: Distribution of $k_i$ in the steady state of some particular cases of
432: model
433: (\ref{lveq}) (see text).
434: }
435: \label{emer}
436: \end{figure}
437: %%%%%%%%%%%%%%%%%%%%%%%
438:
439: Actually, there is yet another property of our species that is
440: subjected to evolutionary selection, namely the number of links $l_i$ (degree)
441: of a given species.
442: Although at the beginning each species acquires $z$ links this number
443: changes during the evolution because some links of a given species
444: might be created or removed due to creation or extinction of another
445: species.
446: And since it is the dynamics of our model and not the random process
447: that determines which species are removed, one can expect that the degree
448: distribution might be different from the Poissonian distribution that
449: is characteristic for random graphs (see~\cite{BOLLO} for a precise
450: definition of random graphs).
451:
452: To check statistical properties of the network of interactions in our
453: model we calculated the degree distribution.
454: Our results for model I with $z=4$ and $N=100$ are shown in
455: Fig.~\ref{link}.
456: Let us notice that although each species has $z$ links at the
457: beginning
458: it does not mean that the average number of links connected to a given
459: site $\langle l_i \rangle$ equals $z$ since dynamics of the model might
460: preferentially remove sites of certain connectivity.
461: And indeed, numerical calculations show that in this case $\langle
462: l_i\rangle =2.98<z=4$, i.e., dynamics preferentially removes sites of
463: larger connectivity.
464: For comparison with the random graph we plot also the Poissonian
465: distribution
466: $r(l)={\rm e}^{-\langle l_i \rangle}\langle l_i \rangle^l/l!$, where
467: $\langle l_i \rangle=2.98$. It should be emphasized that the distribution
468: might be approximately fitted using a Poisson distribution, for
469: example with $\langle l_i \rangle=2.65$. However, it is then not a physically
470: relevant distribution since the average connectivity $\langle l_i
471: \rangle=2.65$ differs from the value $\langle l_i \rangle=2.98$ obtained
472: from the simulations. In this sense the distribution is not Poissonian.
473: One can see that for large connectivity the degree distribution decays
474: faster than the Poissonian distribution.
475: This result confirms that dynamics of the model preferentially removes
476: highly connected species.
477: Such sites are probably more susceptible to fluctuations in the system
478: due to extinctions and creation of new species.
479: On the other hand, poorly connected species are more likely to arrive
480: at a relatively stable state.
481: Similar results concerning the degree distribution were obtained in
482: some other cases of our model.
483: %%%%%%%%%%%%%%%%%%%%
484: \begin{figure}
485: \centerline{\epsfxsize=0.7\columnwidth
486: \epsfbox{fig5.eps}
487: }
488: \caption{
489: Probability distribution $r(l)$ of sites with a given connectivity $l$
490: for model I with $z=4$ and $N=100$ compared with the corresponding
491: Poissonian distribution ($\langle l_i \rangle = 2.98$).
492: }
493: \label{link}
494: \end{figure}
495: %%%%%%%%%%%%%%%%%%%%%%%
496: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
497: \section{Conclusions}
498: In the present paper we studied extinction dynamics of a Lotka-Volterra
499: model of a two-level food web.
500: In our model $N$ species of predators feed on a single species of
501: preys.
502: Competition between predators, that is specified by a certain network
503: of interactions, leads to their extinction and replacement by new
504: species.
505: Distribution of intervals between successive extinctions in some cases
506: has power-law tails and thus resembles extinction pattern of the real
507: ecosystem.
508: However, when the network of interactions between predators is too
509: sparse the ecosystem enters the steady state.
510: We have shown that such a change of behavior might be related with a
511: percolation transition of the network.
512: We also examined an influence of external factors on the evolution of
513: the ecosystem.
514: More specifically, we studied the evolution of our model in case when
515: the growth rate of preys is changing periodically in time.
516: It turns our that such a modulation substantially changes the frequency
517: of extinctions.
518: Counterintuitively, periods with abundance of preys have higher
519: frequency of extinctions than periods with lesser amount of preys.
520: Moreover, we examined some properties of species that are
521: preferentially selected by the dynamics of our model.
522: Under some conditions preferred species are a compromise to the
523: conflicting dynamics.
524: Under some other conditions, preferred species form two antagonistic
525: (with respect to the conflicting rules) groups.
526: We also examined the degree distribution of the network of interactions
527: between species.
528: It turns out that dynamics of the model has a slight preference to
529: remove species of higher connectivity.
530: As a result degree distribution shows some deviation from Poissonian
531: distribution that is characteristic to random graphs.
532:
533: It would be desirable to examine some extensions of our model.
534: For example one can introduce additional trophic levels or other forms
535: of interspecies interactions.
536: One can also examine a variable number of species that would allow to
537: create new species using certain mutation mechanism rather than assuming
538: that they appear as soon as a niche becomes empty.
539: Another possibility that is outside the scope of majority of
540: macroevolutionary models would be to make further study of emergent properties of
541: species.
542: For example, one can imagine that a group of species in the ecosystem
543: is well adapted and essentially not subjected to evolutionary changes.
544: On the other hand there is a group of 'newcomers' where evolutionary
545: changes are much more frequent.
546: How evolution and properties of 'newcomers' are influenced by the
547: properties of well-adapted species?
548: Such problems might be easily approached within our model.
549: Selection of a certain group of species (with a given value of $k$ for
550: example) can be considered as a selection of a certain strategy.
551: One can examine models of this kind where species are have
552: multi-component parameters [$k=(k^a,k^b,\ldots)$].
553: Consequently, one can study evolutionary selection of more complicated
554: traits, strategies, or behaviors.
555: Such an approach would provide an interesting link with certain
556: evolutionary aspects of game theory~\cite{GAME}.
557:
558: This work was partially supported by the Swiss National Science
559: Foundation
560: and the project OFES 00-0578 "COSYC OF SENS".
561: Some of our calculations were done on 'openMosix Cluster' built and
562: administrated by Lech D\c{e}bski at the Institute of Physics at the
563: Adam Mickiewicz University (Poland).
564: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
565: \begin{thebibliography}{}
566: \bibitem{LV} J.~D.~Murray, {\it Mathematical Biology}, (Springer,
567: 1989).
568: J.~Hofbauer and K.~Sigmund {\it The Theory of Evolution and Dynamical
569: Systems}, (Cambridge University Press, 1988).
570: \bibitem{BAKSNEPP} P.~Bak and K.~Sneppen, Phys.~Rev.~Lett.~{\bf 71},
571: 4083 (1993).
572: B.~Drossel, Adv.~Phys.~{\bf 50}, 209 (2001).
573: \bibitem{STAUFFER} D.~Chowdhury, D.~Stauffer, and A.~Kunwar,
574: Phys.~Rev.~Lett.~{\bf 90}, 068101 (2003).
575: \bibitem{ZIA} P.~A.~Rikvold and R.~K.~P.~Zia, Phys.~Rev.~E {\bf 68},
576: 031913 (2003).
577: \bibitem{HALL} M.~Hall, K.~Christensen, S.~A.~di Collobiano, and
578: H.~J.~Jensen, Phys.~Rev.~E {\bf 66}, 011904 (2002).
579: \bibitem{ABRAMSON} G.~Abramson, Phys.~Rev.~E {\bf 55}, 785 (1997).
580: \bibitem{MCKANE} B.~Drossel and A.~J.~McKane, {\it Handbook of Graphs
581: and Networks: From the Genome to the Internet}, S.~Bornholdt and
582: H.G.~Schuster (Eds) (Wiley-VCM, Berlin, 2002), e-print: nlin.AO/0202034.
583: D.~Chowdhury and D.~Stauffer, e-print: q-bio.PE/0311002.
584: \bibitem{QUINCE} C.~Quince, P.~G.~Higgs, and A.~J.~McKane, in {\it
585: Biological Evolution and Statistical Physics}, eds. M.~L\"{a}ssig and
586: A.~Vallerian (Springer Verlag, Berlin-Heidelberg 2002).
587: \bibitem{ALBERT} R.~Albert and A.~-L.~Barab\'asi, Rev.~Mod.~Phys.~{\bf
588: 74}, 47 (2002).
589: \bibitem{ROBERTS} One of the few macroevolutionary models that takes
590: into account external factors was studied by Roberts and Newman
591: (B.~W.~Roberts and M.~E.~J.~Newman, J.~Theor.~Biol.~{\bf 180}, 39 (1996)).
592: This model, however, is a variant of Bak-Sneppen model with dynamics
593: operating at the level of species.
594: \bibitem{NEWMAN} M.~E.~J.~Newman and R.~G.~O.~Palmer, e-print:
595: adap-org/9908002.
596: \bibitem{newman_palmer_book} M.E.J.~Newmann and R.G.~Palmer, \emph{Modelling Extinction}, Oxford University Press (2003).
597: \bibitem{RAUP}D.~M.~Raup and J.~J.~Sepkoski,
598: Proc.~Natl.~Acad.~Sci.~{\bf 81}, 801 (1984).
599: \bibitem{MAY} R.~M.~May, Nature {\bf 238}, 413 (1972).
600: \bibitem{EPSILON} Statistical properties of extinctions that we study
601: in this paper are not affected by the precise value of the $\varepsilon$
602: as soon as it has a small value ($\varepsilon\ll 1$).
603: \bibitem{COMMENT} For noninteger $z$ (i.e., the number of links of
604: newly created species) we used the following procedure: with probability
605: $z-[z]$ we created $[z]+1$ links and with probability $1+[z]-z$ we
606: created
607: $[z]$ links.
608: On average such a recipe produces $z$ links.
609: \bibitem{STAUFFERAHARONY} D.~Stauffer and A.~Aharony, {\it Introduction
610: to Percolation Theory} (Taylor \& Francis, London 1982).
611: \bibitem{SCHEFFER} M.~Scheffer, S.~Carpenter, J.~A.~Foley, C.~Folks,
612: and B.~Walker, Nature {\bf 413}, 591 (2001).
613: \bibitem{BOLLO} B. Bollob\'as, {\it Random graphs}, 2nd ed. (Cambridge
614: University Press, 2001).
615: \bibitem{GAME} R.~Axelrod, {\it The evolution of cooperation} (Basic
616: Books, New York, 1984).
617: J.~Maynard Smith {\it Evolution and the theory of games} (Cambridge
618: University Press, 1982).
619: \end{thebibliography}
620: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
621: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
622: \end {document}
623: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
624: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%