Differences between Genotype and Phenotype in Multiobjective Genetic Algorithm?

I would like to optimise a machine with the 4 conflicting objective functions, mass, cost, losses and diameter. Could anyone can advice me which type of distance crowding I should use? Genotype or Phenotype? What is the differences between them?

I did try to use both of them to generate the Pareto fronts but I am not sure which one is the right one. If I use Genotype, I got the Pareto fronts (Red dots) spreading all around the constraints area.

http://i124.photobucket.com/albums/p40/vunteng/ScreenShot2013-02-22at025444_zps65eaf291.png

While if I use Phenotype, it is more concentrate to one area which causing the results with high cost.

http://i124.photobucket.com/albums/p40/vunteng/ScreenShot2013-02-22at024906_zps367ab26d.png