You do not mention possible confounding factors in your discussion (Nature 468, 1011; 2010) of the reported positive effect of first and last author geographical proximity on paper citations (K. Lee et al. PLoS ONE 5, e14279; 2010).

One is that these are biomedical research papers. This field has many different author-sequence conventions and citation cultures.

In basic research, the first author on a publication is typically a PhD student and the last author is his or her supervisor. The papers come from closely knit research groups, especially in molecular biology, and tend to have zero distance between the first and last authors, and to be cited more frequently than clinical research papers.

By contrast, clinical research projects typically have no clear hierarchical structure among collaborators, and often apply alphabetical ordering of co-authors. Hence, the type of research could also explain the positive correlation you discuss.

The challenge in training researchers to collaborate on publications is to find a balance between face-to-face discussion and the use of new communication technologies.