Many ecologically relevant life-history traits of organisms, (such as generation time and ingestion rate) are significantly correlated to body size. Since these individual and species characteristics can affect the interactions between the species in a community, it is possible that the distribution of body sizes (in communities) can affect different properties of foodwebs as well. To explore this possibility, this thesis analyzes the effect of real body size distributions on various aspects of food web organization. To begin with, data on the body sizes of the organisms in a collection of documented food webs was estimated (resulting in data on almost 700 animal consumers in 52 food webs).

Part of this data is used to analyze the "cascade model" assumption of a constant predation probability for every potential predator-prey pair in a number of these web. The original assumption is compared to four heterogeneous alternatives termed predator-dominance (i),prey-dominance (ii), distance-dominance (iii) and size-ratio dominance (iv). In these alternatives the predation probabilities are non-constant and determined either by the identity of the predator (i) or the prey (ii), or by the difference in rank (iii) or size (iv) between thepredator and its prey. In the restricted set of webs for which adequate body size data was available (16 webs) the null-hypothesis (equal predation probabilities) was rejected in favor of any of these alternatives in 7 out of 16 cases (and specifically, in favor of the size-ratio alternative in 4 out of 16 cases) at a significance level of P=0.06.

Furthermore, the relationship between prey size, predator size and the trophic position of consumers have been analyzed and compared with the predictions of the cascade model. It is shown that the relative size difference between a consumer and its average prey tends to decrease with the trophic position of the consumer in these webs. Furthermore, the trophic links do not seem to be distributed randomly between consumers and their potential prey, as assumed by the cascade model. Instead large predators tend to eat prey of larger size than the cascade model predicts.

The implications of these findings, for the stability of Lotka-Volterra food chains, is analyzed. It is suggested that the trophic interaction coefficients of the Lotka-Volterra model should be affected by the relative size difference between a predator and its prey. It is shown that assuming that the prey interaction strength is positively correlated to the predator-prey body mass ratio, (which can motivated by energetical arguments), in combination with decreasing size ratios between consumers and their prey, with trophic position of the consumer, affects the stability properties of a food chain. The probabilities of local stability and resilience are both increased, when compared to chains where the size ratio does not change with trophic position.

The effect of the generation time of the primary producer, relative to that of the primary consumer, on food chain length is analyzed. In phytoplankton based systems the turnover rates of the primary producers exceed that of the primary consumers while the opposite is true in many terrestrial (tree based) systems. Furthermore, pelagic food webs appear to have longer food chains on average than terrestrial webs. In particular, phytoplankton chains seem to be longer than for example tree based food chains. We show that one reason for this could be that in model Lotka-Volterra food chains (with linear functional responses), resilience increases with the growth rate of the primary producer.

Finally, the capacity of Lotka-Volterra food chain models to produce feasible food chains where the abundance pattern is pyramidal ("the pyramid of numbers") is analyzed. It is shown that density-dependent consumer mortality as well as the distribution of body sizes may affect this capability. Without density-dependent consumer mortality it is difficult (and under some assumptions impossible) for a model Lotka-Volterra food chain to be pyramidal for more than one chain length. With density-dependent consumer mortality this is not a problem. Furthermore, if some of the parameters (i.e. the interaction strengths and death rates) are scaled to body size in a realistic way, it may be possible for model food chains of adjacent lengths to be pyramidal to some extent, even without density-dependent consumer mortality.

The main theme of this thesis is that considering realistic distributions of body sizes in communities can give a better description of how trophic links are distributed and an understanding of how dynamical instability or unrealistic patterns of abundance can be avoided in real communities.