15). We next examined whether the large variability of E-vector tunings could be related to the different neuron types. Of the 25 neurons recorded from the left LAL, 22 responded to both polarized

and unpolarized light, of which SKI-606 manufacturer 13 could be classified as TuLAL1 neurons, five could be classified as TL-type neurons, and four remained unassigned to a particular neuronal type. Comparison of E-vector tuning and azimuth tuning showed no significant difference between these three groups of neuron ( Figure S3; p > 0.3). Accordingly, distributions of ΔΦmax values for none of the groups deviated from a uniform distribution ( Figure S3; p > 0.05). Thus neuronal cell type could not explain the variability of the E-vector tuning. The variability could be explained, as detailed below, by taking into account the daily changes in solar elevation and the region of the sky observed by the monarch DRA. For any sky point outside the solar meridian, the relation between the E-vector angle and the

solar azimuth is complex and depends on the location of the observed point in the sky and the solar elevation ( Figure 1B). As solar elevation changes over the day, the E-vector tuning of neurons not looking SP600125 directly at the zenith needs continuous adjustment to provide consistent azimuthal information ( Pfeiffer and Homberg, 2007). The expected ΔΦmax value of 90° for polarized light stimulation from the zenith did not match the high variability and calculated average ΔΦmax value of 35° of recorded neurons Adenosine in our studies. However, the variable ΔΦmax values we found in monarchs were similar to those from the AOTu neurons of the locust (Pfeiffer and Homberg, 2007). The locust data were explained by modeling the E-vector

angle in the lateral center of the assumed receptive fields of the locust DRA over the course of the day ( Pfeiffer and Homberg, 2007). Because of the different anatomical layout of the monarch DRA, however, the locust model cannot explain the observed E-vector tuning in monarchs. The locust DRA has a receptive field laterally centered at 60° elevation, while the monarch DRA receptive field is laterally centered at 80° elevation ( Homberg and Paech, 2002 and Stalleicken et al., 2006). As E-vector angles near the zenith only change marginally over the course of the day, the monarch ΔΦmax values predicted by the locust model are large (79° for the average recording time) ( Figure 8A). Across the entire monarch DRA, ommatidia are directed toward a narrow band of sky along the longitudinal axis of the butterfly, reaching from the apex (90°) down to elevations of 20°, restricting their view to the celestial hemisphere in front of the animal (Stalleicken et al., 2006 and Labhart et al., 2009) (Figure 8B).