| ||
Network DynamicsScaling and Distribution NetworksGeoffrey West, Physics, Los Alamos National Laboratory Van Savage, Physics, Washington University, NSF/SFI Physics Graduate Fellow Geoffrey West, Jim Brown and Brian Enquist have developed a general mechanistic explanation for the origin of a wide variety of allometric scaling relationships in biology Their model solves a long-standing problem: why metabolic rate scales as the 3/4 power of body mass instead of a k/3 power. Their model assumes that the distribution of resources within organisms is the rate limiting process, and they derive the observed properties of fractal-like distribution networks based on basic physical and biological principles. The general model is based on the observation that life at all scales is sustained by hierarchical fractal-like branching networks. The basic idea is that, in order to supply the enormous number of localized fundamental units of an organism (e.g., cells) a hierarchical, branching, space-filling network is required. The terminal units of the network (capillaries, mitochondria, cytochrome oxidase molecules, petioles, etc) are taken to be invariant. It is further assumed that, of the infinitude of possible networks, those that have evolved by natural selection minimize the energy required to sustain the organism. Given these general postulates it can be shown that such networks must have self-similar, fractal-like characteristics, leading to many scaling laws both between organisms (dominated by 1/4 power exponents) as well as within an individual organism. The model puts severe physical constraints on the structural design and functional characteristics of organisms resulting in a complete quantitative description of the entire system. This has been achieved for the circulatory, respiratory, and plant vascular systems resulting in remarkably good agreement with a plethora of data. This is one of the most exciting results SFI has achieved in the past year or two, in particular because these same principles have been successfully extended to plant ecosystems predicting energy and frequency distributions as functions of size. It is natural to generalise this even further and extend the paradigm to the hierarchies inherent in man-made structures such as the food-web, energy distribution systems, urban development and corporate/economic hierarchies. These have much in common with biological organisms: they have typically evolved under competition suggesting an optimisation principle; they tend to fill the space they occupy, and their terminal units (the individual worker, or the individual dwelling) are approximately invariant. As such, they should be subject to a similar analysis. As one example, consider the scaling of settlement sizes, observed by geographers (since Le Maitre in 1682) and first dealt with in the literature by Zipf (1948). The empirically observed regularity is that settlements of rank r in the descending (size) array of settlements have a size equal to 1/r of the size of the largest settlement in the system. In other words, when the population size of towns is plotted against their frequency on a logarithmic scale, we see an approximately straight line with an exponent of -1. This relationship is known as the 'rank-size rule'. We would like to know why these scaling laws exist, how they might be altered, and to what extent they are optimal. First, however, we need to understand more carefully what the scaling laws really are. The cumulative frequency distribution of the sizes of cities is also similar to a Pareto function or a lognormal distribution. The exact nature of the power law involved has been actively debated, in part because the data used generally did not include the tail end of the distribution (the geographers concentrated on the towns above 10 000 inhabitants, and did not take any smaller settlements into account). SFI visitor Sander Van der Leeuw and SFI post-doctoral researcher Eric van Nimwegen are interested in exploring the effect of including all settlements in an area, regardless of size. In this case, the tail end of the distribution is skewed. Preliminary results suggest that the exponent might be close to 0.82. The explanations for the phenomenon are many. The basic spatial model of urban settlement (Christaller, 1856 or so) is that settlements are indeed organised according to a hierarchical pattern, that larger towns are most distant from each other, and that distance decreases with size. Possibly one of the most convincing explanations (Beckmann 1958) relates the rank-size rule to this model of urban settlement by assuming spatial scale effects in the areas of population, demand, transport facility, etc. If one attempts to understand the causes of this relationship by looking at the variations on the rule observable in real-life urban systems, one has to explain the existence of more or less concave, as well as convex distributions. Johnson (1981) argues that in highly integrated systems, the size of any settlement is dependent on a conditional function of the sizes of all other settlements. To argue this, he draws upon Richardson's theory (1973) which argues that if a given random variable y is equal to the sum of a series of independent random variables, then as the number of independent variables increases, and if no single variable dominates the others, then the distribution approaches the normal distribution. But, if the variable y is equal to the product of the independent random variables, the distribution approaches log-normal. Johnson then presents a series of case studies where evolution in the rank-size relationship (both towards more concave and towards more convex) can convincingly be linked to changes in the intensity of trade patterns. He links the deviations from the norm to the degree of integration of the systems observed, basically arguing that in concave distributions, the degree of integration between the largest centre and the smaller ones is suboptimal, whereas in the case of convex distributions, the system is overintegrated (hypercoherent). This, he posits, is due to the fact that to get linearity and a slope of -1, the variance of the distribution has to approximate 1.0. If it is higher, concave distributions are generated, and if it is lower, convex ones. This constraint on the variability of the products of a multiplicative process, he tentatively proposes, may be found in the functionally hierarchical structure of integrated systems. Johnson's hypothesis suggests an explanation based on work by Brown and West. The idea is to connect the body of theory and the data on settlements to their work on allometric scaling and its central hypothesis---the need to distribute resources efficiently to each member of a population spread into settlements. It is interesting to note that in settlement systems the hierarchical networks that distribute resources are of three different kinds: those that transport matter, those that transport energy, and those that are administrative (and effectively transport information). The basic idea is that it is the differences between the spatial configurations and the hierarchical structures of these three networks that determine the precise forms of scaling observed, as well as the aberrations and the dynamics of settlement systems. Were we able to demonstrate this, we would have a better theoretical grip on such phenomena as metropolisation. And that is an issue which is threatening to change the geography of entire continents in the next century or so. People predict that most towns of 10 000 inhabitants or less will disappear, and that the inhabitants will move towards the megalopoles. Can we find another, equally efficient, way of distributing resources to everyone? References West, G.B., J.H. Brown, and B.J. Enquist. 1997. A general model for the origin of allometric scaling laws in biology. Science 276:122-126. Enquist, B.J., J.H. Brown, and G.B. West. 1998. Allometric scaling of plant energetics and population density. Nature 395: 163-165. West, G.B., Brown, J.H., and B.J. Enquist. 1999. A general model for the structure of plant vascular systems. Nature 400: 664-667. Enquist, B.J., West, G.B., Charnov, E.L., and J.H. Brown. 1999. Allometric scaling of production and life-history variation in vascular plants. Nature 401: 907-911. West, G.B., J.H. Brown, and B.J. Enquist. 1999. The fourth dimension of life: fractal geometry and allometric scaling of organisms. Science 284:167-169. | ||