Figure 7. DCN lesions impair song learning and reduce the similarity to tutor song after crystallization.

(A) Diagram of the song system, as in Figure 2A, representing DCN lesion. (B) Top: Diagram of the song learning periods in songbirds: the sensory period, the sensorimotor period in which juveniles start to produce sounds, and the crystallization phase. Lesions were made at 60 dph. Bottom: Examples of three spectrograms of tutor and pupil song motifs at crystallization: top: tutor, bottom left: a pupil with DCN lesion, bottom right: control pupil. Solid lines connect two similar syllables found in the tutor and juvenile song motifs, dotted lines between two syllables reflect a partial copy of the tutor syllable (red lines for the juvenile with DCN lesion, black lines for the control juvenile). (C) Nissl staining on horizontal slices showing the deep cerebellar nuclei. Top: control bird. Bottom: bird with DCN lesion (Scale bar: 100 µm). (D) Normalized imitations score (imitation score at crystallization divided by the imitation score before surgery) plotted as a function of the total area left from the lateral DCN (%) for juveniles with DCN lesion (red and purple dots, n = 10 birds). The mean and SD of the normalized imitation score over the sham group are represented as an error bar (n = 6 sham birds). For the following analysis, we consider the birds with a significant lesion (<75% of intact lateral DCN, vertical dotted line) represented with red dots. A significant correlation was revealed between the similarity and the proportion of lateral DCN left intact (r = 0.68, p<0.05). (E) Population data showing the evolution of the imitation score between the day before the lesion (pre) and the crystallization period (90 dph) in juveniles with sham lesions (black lines for individual birds) and DCN lesions with larger lesion size (red lines) or small lesion size (purple lines). (F) The imitation score at crystallization is significantly larger in the sham group than in the DCN large lesion group (n = 7 birds with large lesion, n = 6 sham birds, Wilcoxon test, p<0.05).

Figure 7—figure supplement 1. Song rate is not affected by DCN lesion.

Representation of the song rate in the sham (black) and lesion (red) groups around the day of surgery. Each bird is represented with a single line. Day +1 corresponds to the first day with song production following surgery. Quantification of the singing rate among the two groups did not reveal any significant difference between the sham and lesion groups (n = 10 birds in sham group, n = 8 birds in lesioned group, Wilcoxon test, p=0.24). Moreover, all birds resumed singing in the 1–4 days following surgery and there was no difference in the time delay between surgery and the first day with song in the two groups (n = 10 birds in sham group, n = 8 birds in lesioned group, Wilcoxon test, p=0.5).

Figure 7—figure supplement 2. Example spectrogram of tutor and pupil (control and lesion) song motifs as a function of lesion size.

Pupil received DCN partial lesions at 60 dph. (A–D) Example spectrograms of the tutor song motif (left) and the song motif at crystallization of its pupil with a cerebellar lesion (right), in four different families. Histological control indicates that the proportion of intact lateral DCN respectively covered around 3, 40, 39.5, and 18% of the lateral DCN volume in the pupils shown in (A, B, C and D). Note that the three first examples represent large DCN lesions (>50% lateral affected), while the last example is from a bird with a small DCN lesion.

Figure 7—figure supplement 3. Effect of DCN lesions revealed by a custom-written similarity score based on the peak cross-correlation between the spectrograms of the tutor’s motifs and of the pupil’s songs.

Among all songs produced by the pupil in each considered condition: before lesion or at crystallization (all recordings from a single day of recording were considered for analysis in each condition: pre-surgery or after crystallization), 10 randomly-selected songs were compared to the tutor’s selected motifs using the following procedure. Cross-correlations of the spectrograms were computed between all possible pairs defined as follows: a pair consisted in a tutor’s motif and a pupil’s song. For each pair, a cross-correlation index was calculated as the sum of the cross-correlation function between their two spectrograms, normalized by the square root of the product of their auto-correlation function. The average cross-correlation index over all 100 pairs was called the ‘spectral similarity index’ between tutor and juvenile in that condition. (A) Population data showing the evolution of similarity between the day before the lesion (pre) and the crystallization period (90 dph) in pupils with sham lesions (black dots for individuals, solid black line for the mean) and DCN lesions (red dots for individual, solid red line for the mean). Data are normalized over the pre-lesion period. (B) Normalized similarity score plotted as a function of the total area left from the lateral DCN (%) for juveniles with DCN lesion (red dots, n = 10 birds) or sham lesion (black dots, n = 6 birds). A significant correlation was revealed between the quality of the tutor song imitation and the proportion of lateral DCN left unaffected (Pearson correlation coefficient, r = 0.57, p<0.05).