Wind selectivity and partial compensation for wind drift among nocturnally migrating passerines.

"Potential sources of uncertainty in analyzing radar measurements also merit consideration. In our analysis, the estimated headings were conceivably affected by the temporal coarseness (6h) in the NCEP wind data, and by the altitudinal uncertainty in the radar data (±500m). However, given the large number of tracks, such inaccuracies would have presumably led to an unbiased but less concentrated distribution of headings, which would not have strongly affected the estimated proportional drift or endogenous heading. The regression analysis illustrates how pseudo-drift can influence analysis of broad front migration tracks, since analyzing all tracks would have implied FD rather than partial compensation. Our analysis is only valid if SW and SE tracks represented distinct groups of migrants or behavioral reactions to wind, which is clearly supported by their almost nonoverlapping distributions."