An avian basal ganglia-forebrain circuit contributes differentially to syllable versus sequence variability of adult Bengalese finch song

Abstract

Behavioral variability is important for motor skill learning but continues to be present and actively regulated even in well-learned behaviors. In adult songbirds, two types of song variability can persist and are modulated by social context: variability in syllable structure and variability in syllable sequencing. The degree to which the control of both types of adult variability is shared or distinct remains unknown. The output of a basal ganglia-forebrain circuit, LMAN (the lateral magnocellular nucleus of the anterior nidopallium), has been implicated in song variability. For example, in adult zebra finches, neurons in LMAN actively control the variability of syllable structure. It is unclear, however, whether LMAN contributes to variability in adult syllable sequencing because sequence variability in adult zebra finch song is minimal. In contrast, Bengalese finches retain variability in both syllable structure and syllable sequencing into adulthood. We analyzed the effects of LMAN lesions on the variability of syllable structure and sequencing and on the social modulation of these forms of variability in adult Bengalese finches. We found that lesions of LMAN significantly reduced the variability of syllable structure but not of syllable sequencing. We also found that LMAN lesions eliminated the social modulation of the variability of syllable structure but did not detect significant effects on the modulation of sequence variability. These results show that LMAN contributes differentially to syllable versus sequence variability of adult song and suggest that these forms of variability are regulated by distinct neural pathways.

FIG. 1.

Bengalese finch song before and after lateral magnocellular nucleus of the anterior nidopallium (LMAN) lesion. A: spectrograms of undirected songs from a Bengalese finch before (top) and after (bottom) LMAN lesion. Above each spectrogram are labels for each distinct syllable. Bars above labeled syllables highlight examples of sequence variability for syllable “b”, a branch point that could be followed by the syllables “c” (white bar), “b” (dark gray bar), or “e” (light gray bar). B: representative images of calcitonin gene-related peptide (CGRP) staining in LMAN and robust nucleus of the arcopallium (RA) in a control bird (top) and LMAN lesioned bird (bottom). CGRP is expressed in LMAN neurons and their terminals in RA. Lesions of CGRP-expressing cells in LMAN lead to a loss of CGRP-immunoreactivity in RA (Bottjer et al. 1997). C: a diagram of 2 major pathways in the song system, the motor pathway [HVC (proper name), RA, nXIIts (tracheosyringeal portion of the hypoglossal nucleus), and syrinx], and the anterior forebrain pathway [(AFP); (AreaX), DLM (medial nucleus of the dorsal lateral thalamus), and LMAN]. LMAN is the output nucleus of the AFP and is positioned to modulate activity in the motor pathway.

FIG. 2.

Effect of LMAN lesion on syllable variability for undirected song. A: spectrogram of the type of syllable, with flat harmonic structure, used in the analysis of fundamental frequency (FF). To the right is the distribution of FF values for undirected (UD) song before (Pre; top) and after (Post Lesion; bottom) LMAN lesion. In this example, LMAN lesions led to a decrease in the variability of FF (2-sample F-test for equal variances, P < 0.001) but no change in mean FF (t-test, P = 0.6083). B: CVs of 31 syllables from 11 birds before (x-axis) and after (y-axis) LMAN lesion (filled symbols) and of 14 syllables from 8 control birds at 2 time points (open symbols). Overall, CV did not change over time for control birds (P = 0.3784) but was significantly reduced after LMAN lesion (paired t-test: P < 0.0001). C: plotted on the y-axis are values for percent change in mean FF of 31 syllables from 11 lesion birds (filled symbols) and of 14 syllables from 8 control birds (open symbols) vs. prevalues on the x-axis. Mean FF did not change significantly over time for either control or lesion birds.

FIG. 3.

Effect of LMAN lesion on sequence variability for undirected song. A: transition diagrams before (Pre-lesion; left) and after (Post-lesion; right) LMAN lesion for the same song shown in Fig. 1. Arrow thicknesses are proportional to the probability of transitions. In this example, all transitions persisted after LMAN lesion, with no gross changes to probabilities. B: spectrograms showing maintained sequence variation at a branch point for the same bird: syllable “b” could be followed by “c”, “e”, or “b”. In this case, the probabilities for each transition and the branch point entropy (right) were only modestly affected by LMAN lesion. C: branch point entropy values for 28 branch points in 11 lesion birds before and after LMAN lesion (filled symbols) and for 23 branch points in 8 control birds at 2 time points (open symbols). Branch point entropy did not change in a consistent manner in lesion or control birds. D: repeat entropy values for 12 repeated syllables (repeats) in 8 lesion birds before and after LMAN lesion (filled symbols) and for 12 repeats in 6 control birds at 2 time points (open symbols). Repeat entropy did not change significantly over time for either lesion or control birds. E: CVs of repeat numbers for 12 repeats in 8 lesion birds before and after LMAN lesion (filled symbols) and for 12 repeats in 6 control birds at 2 time points (opens symbols). The CV of repeat number did not change significantly over time for either lesion or control birds. Power analyses corroborated these findings (see Supplementary Table S3).

FIG. 4.

Effect of LMAN lesion on tempo for undirected song. A: example of the largest change in tempo (6% faster) following LMAN lesion. At top is a spectrogram of the sequence used for measurement of song tempo (sequence duration). The sequence duration was calculated as the time from the onset of “a” to the onset of “e”, denoted by the dotted lines. Below are histograms of the duration of “abcde” before and after LMAN lesion. In this case, LMAN lesion led to a significant speeding up of song (t-test, P < 0.01). B: plotted on the y-axis are values for percent change in sequence duration for lesion (filled symbols) and control birds (open symbols) versus prevalues on the x-axis. Despite individual cases of tempo changes, there was no significant change in song tempo over time for lesion or control birds.

FIG. 5.

Effect of LMAN lesion on social modulation of song. A–E: social modulation is plotted for each song feature from pre- and post-time points for control and LMAN lesion birds. Gray boxes indicate means ± SE. For all 5 features, there was significant social modulation during the pre period, consistent with a prior report (Sakata et al. 2008). Lesions eliminated social modulation of the CV of FF but did not significantly affect social modulation of other features. A: social modulation of the CV of FF for 14 syllables in 8 control birds and 25 syllables in 8 lesion birds. The magnitude of the social modulation of the CV of FF did not change over time for control birds (P = 0.5462) but decreased significantly from the 1st to the 2nd timepoint for lesion birds (P < 0.0001; MANOVA: group × time, P = 0.0035). B: social modulation of mean FF for 14 syllables in 8 control and 25 syllables in 8 lesion birds. The social modulation of mean FF did not change significantly across time for control or lesion birds. C: social modulation of branch point entropy for 23 sequences in 8 control birds and 19 sequences in 8 lesion birds. There was no significant effect of lesion on social modulation of branch point entropy (MANOVA: group × time; P = 0.2835), although an attenuation of social modulation of branch point entropy across testing sessions (MANOVA: effect of time; P = 0.0084) was present for both control (P = 0.0451) and lesion birds (P = 0.0675). D: social modulation of repeat entropy for 12 sequences in 6 control birds and 9 sequences in 6 lesion birds. There was no significant effect of LMAN lesion on social modulation of repeat entropy. E: social modulation of song tempo for 8 control and 8 lesion birds. There was no significant effect of LMAN lesion on the social modulation of song tempo.

FIG. 6.

Effect of LMAN lesion on the social modulation of sequence variability in birds with only 1 social context test. Five birds were administered LMAN lesions and tested for social context effects on branch point entropy 2 wk after LMAN lesion (LMAN lesion). A control group of birds (n = 13) with intact LMAN (Intact) were similarly administered only 1 social context test. For both groups, branch point entropy was reduced in the female-directed (FD) vs. undirected (UD) condition, but this modulation achieved significance only for the control group (controls: −19.4% change in branch point entropy from UD to FD song for 37 sequences in 13 birds; t-test; H₀ = mean is equal to zero; P < 0.01; lesions: −13.0% change for 12 sequences in 5 birds; P = 0.3203). However, although the magnitude of social modulation was attenuated in lesion birds relative to controls, this difference was not significant (P = 0.5520).