You can use higher gain antenna's I guess... I remember that we did a range test here, similar to the hidden station problem with three and four devices, with 3dBi antennas on a soccer field... with lowest power level of zero and sometimes we had a maximum transmission range of 90 meters, sometimes around 135 meters. It depended on the weather plus the exact orientation of the antenna, since every antenna behaves differently when it comes to transmission characteristics.
Ah and we used normal XBee 802.15.4 modules, didn't test the pro modules...
In total I guess that the maximum range values from the Libelium test in the Monegros desert seem to be either really optimistic since nobody could confirm their results up to now. But I guess that the specifics of the test environment "improved" the range... it seems for example strange that RSSI values from different devices and frequencies for 356m and 639m LOS are the same. (http://www.libelium.com/libelium-downloads/EWSN2010PosterFinal.pdf
). I guess the specific Fresnel zone of the testing area helped here... but I cannot prove it due to the lack of proper testing ground.