At every scale of life, from microscopic bacteria to the towering giants of Amazonian trees we see a similar pattern of diversity where not every individual gets to be equally abundant. Rather, there is a pattern in which common species are rare and rare species are common. But how does this seemingly universal pattern of diversity arise and how is it is maintained? What mechanisms account for the rarity of some and the commonness of others? Hypotheses can be categorized in two different perspectives: classical niche ecology, arguing that natural selection has shaped and altered species to outcompete others or neutral ecology, stating that a never ending game of chance alters community structure according to fixed laws of probability. One might say that each organism is either struggling for life or rolling the dice in life’s casino. The main theme of the thesis by Edwin Pos is to quantify the relative importance of these different mechanisms shaping community composition. In this thesis, theoretical models were developed to test the relative importance of selection, migration and stochastic events on structuring community composition. But as any theoretical model needs empirical input, the first chapters of his thesis are devoted to validating and testing important assumptions and caveats of using and analyzing ecological data. First it is shown that having a large number of unidentified species in a dataset, which often is the case for ecological field data, may not affect our results as much as is often thought. Second, that estimators of species diversity based on a logseries distribution are much more effective in estimating species richness for highly diverse ecosystems than other, far more commonly, used nonparametric estimators. Third, a comparison between many different mathematical methods for the estimation of migration shows that in many cases this is more an approximation of the homogenization among local communities over time than a direct measurement of migration. Combining the above and developing a new simulation model, a level of biological reality was added to predictions from neutral theory, simultaneously focusing at both local and regional scales of life. The results showed that no matter how well parameterized or how well the output of simulations fitted the regional patterns, an accurate simultaneous prediction on both regional and local diversity patterns was impossible to attain. It was concluded that other non-neutral processes must be at work, at least at the local level. Finally the use of mathematical principles from information theory, the Maximum Entropy Formalism showed an overall low, but strong, environmentally dependent effect of specific functional traits on genus level composition. Very strong effects of dispersal accompanied this pattern of selection, with a strong spatial gradient in dispersal that depends on geographical distance. The combined results of this thesis show that both niche and neutral driven processes play important roles in structuring community composition and provides clues as to where we should focus our attention towards understanding the game of life.