Monte Carlo simulation of body height in a spatial network.

BACKGROUND/OBJECTIVES: Recent evidence suggests clustering of human body height. We want to assess the consequences of connectedness in a spatial network on height clustering in an artificial society.

SUBJECTS/METHODS: We used an agent-based computer modelling technique (Monte Carlo simulation) and compared simulated height in a spatial network with characteristics of the observed geographic height distribution of three historic cohorts of Swiss military conscripts (conscripted in 1884-1891; 1908-1910; and 2004-2009).

RESULTS

Conscript height shows several characteristic features: (1) height distributions are overdispersed. (2) Conscripts from districts with direct inter-district road connections tend to be similar in height. (3) Clusters of tall and clusters of short stature districts vary over time. Autocorrelations in height between late 19th and early 21st century districts are low. (4) Mean district height depends on the number of connecting roads and on the number of conscripts per district. Using Monte Carlo simulation, we were able to generate these natural characteristics in an artificial society. Already 5% height information from directly connected districts is sufficient to simulate the characteristics of natural height distribution. Very similar observations in regular rectangular networks indicate that the characteristics of Swiss conscript height distributions do not so much result from the particular Swiss geography but rather appear to be general features of spatial networks.

CONCLUSIONS

Spatial connectedness can affect height clustering in an artificial society, similar to that seen in natural cohorts of military conscripts, and strengthen the concept of connectedness being involved in the regulation of human height.