Proc. Nati. Acad. Sci. USA Vol. 85, pp. 8722-8726, November 1988 Neurobiology Birth of projection neurons in the higher vocal center of the canary forebrain before, during, and after song learning (neurogenesis/area X/robustus archistriatalis/interneurons) ARTURO ALVAREZ-BUYLLA, MARGA THEELEN, AND FERNANDO NOTTEBOHM The Rockefeller University, Field Research Center, Tyrrel Road, Millbrook, NY 12545 Contributed by Fernando Nottebohm, August 1, 1988 ABSTRACT The higher vocal center (HVc) of the canary brain projects to two forebrain nuclei: robustus archistriatalis (RA) and area X of lobus parolfactorius. The time of birth of HVc neurons projecting to these two regions was determined by combining [3H]thymidine autoradiography and retrograde fluorogold uptake. Birds were sacrificed at 13 months of age, 4 days after fluorogold injections into area X or RA. A single injection of [3lH]thymidine in ovo (embryonic day 9) labeled 76% of area X-projecting cells and.8% of cells projecting to RA. The great majority of RA-projecting cells were produced during posthatching development (posthatching day 1-24; P1-P24), with a peak at P6 and a hiatus at P12. HVc reaches full adult size by P24, yet at that age the production of new RA-projecting cells continued at a rate comparable to that recorded during posthatching development. Late production of neurons interconnecting two distant regions of the brain may regulate source to target cell population size. Male canaries start to sing at P4. During subsequent months, they imitate external models and their song becomes more structured and stereotyped. At sexual maturity (P24), song is stable. Three interpretations are offered: (i) neurogenesis of RA-projecting cells is related to learning, and learning continues even after achievement of pattern stability; (ii) neurogenesis of RA-projecting cells is not related to learning; (iii) the production of RA-projecting cells serves different purposes during development and after sexual maturity. Canaries develop their song by imitating the song of other canaries they can hear (1, 2). The higher vocal center (HVc)* of the adult canary forebrain plays an important role in the control of learned song (3, 4). n canaries, this nucleus is first seen as a discrete anatomical entity on posthatching day 3 (P3) (5). At that time, it is only 1/8th of the adult volume. t continues to grow thereafter until it reaches full adult size at sexual maturity, between P21 and P24 (5). Song development, which starts at about P4, yields stable adult song by P24 (5). The time of birth of neurons in the canary HVc has not been described before, although it is known that new neurons continue to be added in adulthood (6). HVc neurons send ipsilateral projections to two forebrain nuclei, area X of lobus parolfactorius and robustus archistriatalis (RA) (Fig. 1). Previous work (7) indicated that most of the new neurons added in adulthood are local interneurons. This suggested that the great majority of projection neurons in HVc were produced earlier during development. The present report documents the birth dates of HVc neurons projecting to area X and RA. Two patterns of neuronal recruitment emerge: a brief one before hatching, typical of area X-projecting neurons; and another one lasting for many months during posthatching development, typical of RA-projecting neurons. The publication costs of this article were defrayed in part by page charge payment. This article must therefore be hereby marked "advertisement" in accordance with 18 U.S.C. 1734 solely to indicate this fact. 8722 MATERALS AND METHODS Thirty-seven male and female canaries from our close-bred colony of the Waterslager strain were used (Table 1). This colony is kept in a photoperiod matching that of New York state. Results presented here are from birds hatched during the spring of 1985. Eggs were marked on the day of laying, which was taken as embryonic day (E). [3H]Thymidine Administration. For the embryonic injections of thymidine (Table 1), a small window (diameter, 1-2 mm) was opened on the blunt end of the egg (above the air space). The shell membrane was punctured and 4,ul of [methyl-3h]thymidine (6.7 Ci/mmol; 1 Ci = 37 GBq; New England Nuclear) was injected. The window in the shell was sealed with Op-Site postoperative dressing (Smith and Nephew, Columbia, SC), and the egg was returned to the nest. Chicks hatched 13-14 days from E. Each nestling within a clutch was age-marked by clipping one of its toes at hatching. n birds receiving [3H]thymidine after hatching (Table 1), the nucleotide was administered in two i.m. injections of 2.5,uCi per g of body weight at 12-hr intervals (9-11 a.m. and 9-11 p.m.; 1-day treatment). Birds in the P24 age group received a total of four injections at the same interval (2-day treatment). All birds were sacrificed between 12 and 14 months of age. Labeling index for the birds injected after hatching is expressed as percentage of cells labeled per day of [3H]thymidine treatment. Backfills from Area X and RA. To identify cells within HVc that project to either area X or RA, fluorogold (2%) (8) injections were made into these nuclei 4 days before sacrifice. This procedure backfills many more (possibly all) projecting cells in HVc than the peroxidase protocol previously used (7). The fluorescent signal of fluorogold does not interfere with visualization of the autoradiographic grains under bright field, making identification of double-labeled cells easier. Anesthesia was induced with Chloropent (Fort Dodge Labs., Fort Dodge, A) and maintained with Metofane (Pitman-Moore, Washington Crossing, NJ). Each bird was placed in the stereotaxic apparatus and the skin over the skull was opened. RA was filled with four injections (1-2 nl each) at the following stereotaxic coordinates (9): posterior, 1.5 and 1.9; lateral, 2.5; depth, 2.5 and 2.3. To keep the injection tract into RA away from HVc, the pipette was inserted caudal to HVc at a 9 angle in the anteroposterior plane. Area X was filled with six injections (1-2 nl each) at the following coordinates: anterior, 4.5, 4, and 3.5; lateral, 1.5; depth, 4 and 3.7. Four days later, these birds were killed under deep anesthesia. Fixation was by intracardiac perfusion with 1-2 ml of saline followed by 7 ml of 2% paraformaldehyde/.25% glutaraldehyde in.1 M phosphate Abbreviations: HVc, higher vocal center; RA, robustus archistriatalis; LMAN, lateral magnocellular nucleus of the anterior neostriatum; EO, embryonic day, etc.; P3, posthatching day 3, etc. *Previously known by the misnomer hyperstriatum ventralis, pars caudalis (3).
Neurobiology: Alvarez-Buylla et al. FG. 1. Sagittal section of canary brain illustrating the HVc efferent pathways to RA and area X. RA in turn connects to the hypoglossal nucleus (nx) that innervates the syrinx. Cb, cerebellum; LH, lamina hyperstriatica; LPO, lobus parolfactorius. (Bar = 1 mm.) buffer. Brains were removed and incubated overnight in the same fixative at 4 C. Table 1. njection schedule of [3H]thymidine Age when Fluorogold Age when Bird Sex injected injection site killed 1 M E5 X/RA 41 2 M E9 X/RA 389 3 F E1 X/RA 3% 4 F E1 X/RA 395 5 F E1 X/RA 379 6 M E1 X 389 7 M E1 X/RA 375 8 M E1 X 389 9 M E12 X/RA 388 1 F P1 RA 365 11 M P1 X/RA 374 12 M P34 X/RA 366 13 M P35 RA 365 14 M P35 X/RA 366 15 M P35 X/RA 366 16 F P36 X/RA 365 17 M P36 X/RA 365 18 M P57 X/RA 393 19 M P58 X 39 2 M P58 RA 39 21 M P6 X/RA 392 22 M P61 RA 393 23 M P64 X/RA 3% 24 M P93 X/RA 42 25 M P93 X/RA 42 26 M P93 X/RA 43 27 F P94 X 43 28 M P12 X/RA 395 29 F P12 X/RA 395 3 M P121 X 395 31 M P121 X 395 32 M P239 X/RA 394 33 M P239 X/RA 394 34 F P24 X/RA 395 35 M P24 X/RA 393 36 M P242 X/RA 395 37 M P242 X/RA 395 All ages are in days; X/RA, animals in which area X- and RA-projecting neurons were successfully backfilled after injections into the left and right hemispheres, respectively; RA alone, injection in area X failed; X alone, injection in RA failed. Proc. Natl. Acad. Sci. USA 85 (1988) 8723 Histology. The brain was embedded in polyethylene glycol (1). Serial 6-,m sagittal sections were mounted on gelatinized slides. The sections were dried, delipidized in xylene, and coated with undiluted nuclear track emulsion NTB-2 (Kodak). After incubation for 1 month, slides were developed for 3 min at 17 C in D19 developer (Kodak), washed in distilled water, and fixed for 12 min. All fluorogold and double-labeled (fluorogold plus [3H]- thymidine) cells in five evenly spaced noncontiguous HVc sections were counted for each brain hemisphere. These same sections were then stained with.5% cresyl violet and the number of labeled and total neurons was determined. This sequential procedure is required because the cresyl violet staining interferes with the fluorescence of the fluorogoldfilled cells. Computer-assisted microscopy ensured that the initial HVc boundaries used to count the fluorogold-filled cells were also the boundaries for counting labeled and total neurons; this procedure has been described in detail (11). Labeling Criterion and Cell Sample Size. The criterion for considering a cell [3H]thymidine labeled was 2x background (5-7 grains per neuronal nucleus). The HVc sections examined in the 37 birds of this study yielded 3526 [3H]- thymidine-labeled HVc neurons, 1,453 RA-backfilled neurons, 7158 area X-backfilled neurons, 78 [3H]thymidinelabeled area X-projecting cells, and 76 [3H]thymidine-labeled RA-projecting cells. RESULTS Neurogenesis in HVc. The proportion of total neurons in HVc labeled with [3H]thymidine at different ages is shown in Fig. 2. Before hatching, 1-3% of all neurons in HVc are labeled by a single injection of [3H]thymidine. The highest labeling index (3%) occurred in three birds that received the [3H]thymidine injection on E5, E9, or E1. We do not know how long [3H]thymidine remains available for labeling after a single injection in the canary embryo, but in experiments with chicken embryos the earliest injections labeled cells generated up to the time of hatching (12, 13). Thus, at least 3% of the HVc neurons present in 13-month-old canaries were born before hatching. This figure could be higher if there was only partial overlap between the populations of neurons labeled by injections on E5, E9, or E1. However, the overlap is considerable (see below) and thus this figure may not be much higher. zn A cc cm # z 1 o a CO) J m > o- A o A L2-4 - *!O 6 1 14 18 22 26 AG E (DAYS) FG. 2. Proportion of all HVc neurons at 13 months ofage labeled with [3H]thymidine injected at different ages. L corresponds to time when eggs were laid (EO) and arrow indicates hatching time. Each data point is from one bird; *, males; A, females. The best fit for the posthatching injections is indicated.
8724 Neurobiology: Alvarez-Buylla et al. B.c..*.4 D44 4 4 Photomicrographs of [3Hlthymidin'e-labeled fluorogold- FG. 3. backfilled neurons in HVc. (B, D, and F) Bright-field illumination showing clusters of exposed silvergrains from [3Hlthymidine-labeled cells; (A, C, and E) the same microscopic fields but under fluorescence, revealing the fluorogold-filled cells., (A and B) Cells backfilled from area X and labeled with [3H]thymidine injected at E9 (arrows). (Cbackfiied and D) Cells from RA in bird injected at E9. Large labeled nuclei (arrows) do not c[rrehpond to the fluorescent RAprojecting neurons. (E and F) Cell backfllled from RA (arrow) and also labeled in a bird that received [3H]thymidine at P24. (Bar = 2 /AM.) Consistent with this interpretation is the observation of a slow but constant recruitment of new neurons during post- 4 Proc. Natl. Acad. Sci. USA 85 (1988) hatching development (Fig. 2). Between.1% and 2% (average,.8%) of the neurons present at 13 months of age were labeled per day of [3H]thymidine treatment between P1 and P24. A daily labeling index as low as.3% over a 24-day period could account for 7% of the HVc neurons present at 13 months. The observed rate of.8% could result from heavily labeled neuronal stem cells dividing repeatedly on successive days, each time generating a labeled neuron. Thus, a short but intense period of prehatching neurogenesis and a long but slow posthatching recruitment of neurons can account for all the neurons in HVc. The question that then arises is whether these two patterns of neurogenesis reflect the birth of different functional classes of neurons in the embryo vs. posthatching development. Birth of HVc Projection Neurons. Fluorogold injected into RA filled approximately twice as many HVc cells as when injected into area X. RA-projecting neurons were smaller [mean nuclear diameter, 6.7 ±.47,Am (SD); 3 birds, 324 neurons sampled] and showed a more rounded cytoplasmic contour than the larger (mean, 9. ±.22,um) polyhedric area X-projecting cells (Fig. 3). Thus, these two projections seem to correspond to two different populations of neurons (7). A large proportion of area X-projecting cells were labeled by a single [3H]thymidine injection into the egg (Figs. 3 A and B and 4). The highest labeling index of these cells (76%) was seen at E5 and E9. As suggested earlier, E5 and E9 injections are not likely to label separate subsets of neurons, since otherwise the total percentage of area X-projecting cells labeled during those. 2 days would come to 152%. Thus, in canaries as in chickens, [3H]thymidine injected early in embryogeny remains available for incorporation into DNA for a number of days. n the birds that received [3H]thymidine after hatching, the labeling ofarea X-projecting cells was extremely rare (Fig. 4); only 12 of 552 fluorogold-filled area X-projecting cells were also labeled with [3H]thymidine. Thus most of HVc area X-projecting neurons were produced during prehatching development. We found a dramatically different result for RA-projecting neurons. At most,.8% of all RA-projecting cells were Percent labeled neurons that project to X Percent X-projection neurons labeled X - - _ PrO P35 P6 P9 P12 P24 P1O P35 P6 P9 P12 P24 Percent labeled neurons that project to RA Percent RA-projection neurons labeled 3 X2 1 X i '"1" '- ES E EHO E2 PO P35 P6 P9 P2 P24 ES E9 EO E2 P P35 P6 P9 P12 P24 FG. 4. Proportion of [3H]thymidine-labeled HVc neurons that project to area X (Upper Left) or RA (Lower Left); [3H]thymidine was given at the ages shown. (Right) Proportion of all area X-projecting (Upper) and RA-projecting (Lower) cells labeled with [3H]thymidine injected at different ages.
Neurobiology: Alvarez-Buylla et A labeled with a [3H]thymidine injection in the egg. For example, a bird that received [3H]thymidine on E9 had 76% of the left HVc area X-projecting cells labeled (Figs. 3 A and B and 4), but only 2 of 392 (.5%) fluorogold-filled RAprojecting cells in the right HVc (five sections) were labeled with [3H]thymidine (Fig. 3 C and D). This pattern was consistently found in all of the birds treated with [3H]thymidine before hatching. [3H]Thymidine labeled.3-1.1% of RA-projecting neurons per day of treatment at all posthatching ages studied except for P12 when double-labeled cells were found in neither of the two birds injected. Even in birds injected at P24, an average of.7% of the HVc cells projecting to RA was found labeled per day of treatment (Figs. 3 E and F and 4). The highest labeling index, 1.1%, was found in the P6 group (Fig. 4). At all treatment times, the number of [3H]thymidinebackfilled area X- or RA-projecting neurons was lower than the total number of [3H]thymidine-labeled HVc neurons. So, for example, only 49.4% of the HVc neurons that were labeled with [3H]thymidine on E9 were backfilled from area X or RA at 13 months. This fraction gets lower after hatching. We do not know whether fluorogold injections were 1% effective in backfilling all HVc projection neurons. f HVc cells that did not backfill from area X or RA were interneurons, then interneurons were produced during both embryogenesis and posthatching development. Maps of labeled HVc neurons or labeled HVc projection neurons were constructed for each of the birds in this study (data not shown). The new cells were not restricted to specific areas within HVc. Cells born at any age took up positions throughout HVc. This suggests that late-born cells (such as RA-projecting neurons) can integrate with cells born early during development. We frequently observed RAprojecting neurons not labeled with [3H]thymidine in close apposition to neurons labeled with [3H]thymidine in ovo. Our observations suggest that neurons born late in development find exact positions where they are needed, as opposed to just moving to the periphery of HVc. This pattern of distribution must require sensitive pathfinding (e.g., see ref. 14). The positional integration of the new neurons must be important in the development of basic functional units within HVc. This mode of neuronal recruitment contrasts with the organized laminar histogenesis observed in mammalian cortical structures (15). DSCUSSON Recruitment of HVc Projection Neurons and Song Learning. Whereas the great majority of area X-projecting cells of the HVc are generated in ovo, most RA-projecting cells are born after hatching. Only one of these two target nuclei, RA, is part of the motor pathway for song production (4). However, area X projects to a nucleus of dorsal thalamus (16) that in turn projects to the lateral magnocellular nucleus of the anterior neostriatum (LMAN) (17). Most neurons that project from HVc to area X, from dorsal thalamus to LMAN, and from LMAN to RA are produced before hatching (this study and unpublished observations). t seems likely that the circuit linking HVc, area X, dorsal thalamus, LMAN, and RA is operational early in development and thus could influence the earliest stages in song acquisition in canaries. This circuit could induce specific patterns of differentiation and connectivity in RA. This hypothesis is supported by the observation that LMAN is necessary for song learning in zebra finches but is not necessary for the maintenance and production of learned song (18). The HVc to RA projection that develops later may play a role in later stages of song acquisition, affecting RA circuitry in a different way. Proc. Natl. Acad. Sci. USA 85 (1988) 8725 Song learning in young canaries occurs between days 4 and 24 (5). The bulk of the HVc to RA song control circuit is generated at the very time when song behavior is acquired. Song learning depends on auditory experience and motor practice; these in turn might influence the production, migration, differentiation, and survival of new cells and of the connections they form. Changes in synaptic efficacy (e.g., see refs. 19 and 2) and number (21-23) have been shown to accompany learning. These changes could take place on existing neurons. Does motor skill learning require that new neurons be inserted into a particular motor circuit at the very time when the skill is acquired? The plasticity required during song learning might go beyond the capacity for change present in the HVc neurons. A new neuron could bring uncommitted function and structure. The majority of song syllables that male canaries use during their first breeding season is established between P9 and P12 (5). At the time when the young canary is making the transition between subsong and plastic song (P6) (5), many RA-projecting neurons are recruited. The addition of these cells to the song circuitry is probably important for acquisition of motor programs underlying each new syllable type. Between P12 and P24, these syllables become more and more stereotyped and few, if any, new song syllables are added. This is a period of skill consolidation and during part of it at least (P12) recruitment of RA-projecting cells diminishes. The association between neurogenesis and song learning is not completely clear. Male canaries reach sexual maturity at about P24 when their song repertoire is stereotyped (5). RA-projecting neurons continue to be added to HVc at P24 (Fig. 4). These cells could not play a role in the acquisition of the first year's song repertoire but could be available for song modification later in the summer when song again becomes unstable and new syllables are learned (5). Nordeen and Nordeen (24) have recently shown that 22% of RA-projecting neurons in zebra finches are formed between P2 and P4 and very few are formed in adulthood. Zebra finches acquire their song between P3 and P8 (25). The ongoing recruitment of RA-projecting neurons in P24 canaries and its absence in adult zebra finches may relate to the continual modification of canary, but not zebra finch, song in adulthood (26). However, while we allowed 15 days to elapse between [3Hlthymidine treatment and the time when the birds were killed, the Nordeens allowed 37 days. The low labeling index of RA-projecting cells reported for adult zebra finches may reflect the short survival time. Projection axons may grow relatively slowly in adult brain. A previous study used horseradish peroxidase to backfill RAprojecting neurons in 1- to 2-year-old canaries treated with [3H]thymidine; these birds were killed 3-6 days later and in them very few HVc cells born in adulthood projected to RA (7). We now need to determine whether the difference between this latter study and our observations of labeled RA-projecting cells in the P24 birds resulted from differences in age, survival time, or retrograde tracer used. The song control pathway may be important for song perception (17, 27). f so, the recruitment of RA-projecting HVc neurons could be related to acquisition of new perceptual memories, which may continue well after song learning has come to an end. Late Production of Projection Neurons as a Developmental Mechanism. The late appearance of RA-projecting neurons in songbirds cannot be explained as part of the delayed development of a whole brain nucleus because, as shown, area X-projecting cells of the HVc are born in ovo. The dynamics underlying the observed late recruitment of RA-projecting neurons remain unknown. We do not know whether the RA-projecting neurons formed late in development replaced older neurons. Conceivably, RA-projecting
8726 Neurobiology: Alvarez-Buylla et al. neurons were formed in the embryo but did not survive to 13 months of age. The canary HVc grows 8 times in volume from 1 to 8 months of age (5). n zebra finches, the volume of HVc also increases during development and this growth is partly due to the addition of new neurons (28). Thus, net addition of RA-projecting neurons is probably taking place in canary HVc. However, by P24 HVc has reached its adult size, yet RA-projecting neurons continue to be generated. At P24, the recruitment of new RA-projecting cells may no longer be a developmental phenomenon of addition, but rather a case of adult neuronal replacement (29). The final size of a neuronal population projecting to a distant target is thought to be regulated by neuronal cell death (3, 31). The present results suggest that neuronal cell birth may also play a role in the final adjustment of population sizes of specific projection neurons. During the assembly of brain circuits controlling song, both neuronal cell death and neuronal cell birth are used. Type V cells in RA receive synaptic input from HVc as well as from the LMAN (23). n the young zebra finch, cell death in LMAN has been related to the development of song (32). Perhaps during development the new RA-projecting cells from HVc colonize synaptic territory previously occupied by neurons from LMAN. n both males and females, most area X-projecting cells were produced before hatching. However, 3.7% of the male HVc neurons produced after hatching were RA-projecting cells (vs. 7.1% in the females). These results are in agreement with recent findings in the zebra finch (24). Thus, recruitment of RA-projecting cells may directly contribute to the sexual dimorphism in song and HVc volume. Less than 5% of the HVc neurons labeled in the embryo are projection neurons (Fig. 4). f we assume that fluorogold backflilled all of the HVc projection neurons, then there is a class of HVc neurons produced before hatching that is not a projection cell but an interneuron. As determined by immunocytochemistry, there are two kinds of interneurons in adult canary HVc. One kind, which continues to be produced in adulthood, is y-aminobutyric acid (GABA) negative. The other kind, which is not produced in adulthood, is GABA positive (33). Perhaps this latter type is generated in the embryo. Earlier thinking emphasized that projection neurons are produced early and interneurons are produced late during central nervous system development (34, 35). We have shown that some kinds of projection neurons can continue to be produced over a long period of time, perhaps into adulthood. Why do RA-projecting neurons have to be added so late? s this a property shared with other systems that learn complex motor skills during development? Or is it a trait of some small brains, in which a very complex function is invested on circuits built in a particularly sparing fashion? Birds may have solved some problems of brain development and learning in a manner very different from mammals. We thank John Kim, Marta Nottebohm, Ellen Prediger, David Vicario, and Heather Williams for editorial help. Our work was supported by a grant from the Whitehall Foundation, by the Simpson Charitable Trust, and by National nstitutes of Health Grant MH18343. t was also supported in part by Biomedical Research Support Grant S7RR765 awarded by the Biomedical Research Proc. Natl. Acad. Sci. USA 85 (1988) Support Grant Program, Division of Research Resources, National nstitutes of Health. 1. Waser, M. S. & Marler, P. (1977) J. Comp. Physiol. Psychol. 91, 1-7. 2. Marler, P. & Waser, M. S. (1977) J. Comp. Physiol. Psychol. 91, 8-16. 3. Nottebohm, F. (1987) in Encyclopedia of Neuroscience, ed. Adelman, G. (Birkhaeuser, Boston), Vol. 1, pp. 133-136. 4. Nottebohm, F., Stokes, T. & Leonard, C. M. (1976) J. Comp. Neurol. 165, 457-486. 5. Nottebohm, F., Nottebohm, M. E. & Crane, L. (1986) Behav. Neural Biol. 46, 445-471. 6. Goldman, S. A. & Nottebohm, F. (1983) Proc. Natl. Acad. Sci. USA 8, 239-2394. 7. Paton, J. A., o'loughlin, B. E. & Nottebohm, F. (1985) J. Neurosci. 11, 388-393. 8. Schmued, L. C. & Fallon, J. H. (1986) Brain Res. 377, 147-154. 9. Stokes, T. M., Leonard, C. M. & Nottebohm, F. (1974) J. Comp. Neurol. 156, 317-374. 1. Alvarez-Buylla, A., Buskirk, D. R. & Nottebohm, F. (1987) J. Comp. Neurol. 264, 159-17. 11. Alvarez-Buylla, A. & Vicario, D. S. (1988) J. Neurosci. Methods 25, 165-173. 12. LaVail, J. H. & Cowan, W. M. (1971) Brain Res. 28, 421-441. 13. Tsai, H. M., Garber, B. B. & Larramendi, L. M. H. (1981) J. Comp. Neurol. 198, 275-292. 14. Alvarez-Buylla, A. & Nottebohm, F. (1988) Nature (London) 335, 353-355. 15. Rakic, P. (1988) Science 241, 17-176. 16. Okuhata, S. & Saito, N. (1987) Brain Res. Bull. 18, 35-44. 17. Williams, H. & Nottebohm, F. (1985) Science 229, 279-282. 18. Bottjer, S. W., Miesner, E. & Arnold, A. P. (1984) Science 224, 91-93. 19. Kandel, E. R. & Schwartz, J. H. (1982) Science 218, 433-443. 2. Fifkova, E. & Van Harreveld, A. (1977) J. Neurocytol. 6, 211-23. 21. DeVoogd, T. J., Nixdorf, B. & Nottebohm, F. (1985) Brain Res. 329, 34-38. 22. Bailey, C. H. & Chen, M. (1988) Proc. Natl. Acad. Sci. USA 85, 2373-2377. 23. Canady, R. A., Burd, G. D., DeVoogd, T. D. & Nottebohm, F. (1988) J. Neurosci., in press. 24. Nordeen, K. W. & Nordeen, E. J. (1988) Nature (London) 334, 149-151. 25. mmelmann, K. (1969) in Bird Vocalizations, ed. Hinde, R. A. (Cambridge Univ. Press, Cambridge, U.K.), pp. 61-77. 26. Nottebohm, F. & Nottebohm, M. (1978) Z. Tierpsychol. 46, 298-35. 27. Margoliash, D. (1986) J. Neurosci. 6, 1643-1661. 28. Bottjer, S. W., Miesner, E. A. & Arnold, A. P. (1986) Neurosci. Lett. 67, 263-268. 29. Nottebohm, F., ed. (1985) in Hope for New Neurology (N.Y. Acad. Sci., New York), Vol. 457, pp. 143-161. 3. Oppenheim, R. W. (1981) in Studies in Developmental Neurobiology: Essays in Honor of Viktor Hamburger, ed. Cowan, W. M. (Oxford Univ. Press, New York), pp. 74-133. 31. Cowan, W. M., Fawcett, J. W., O'Leary, D. D. M. & Stanfield, B. B. (1984) Science 225, 1258-1265. 32. Bottjer, S. W., Glaessner, S. L. & Arnold, A. P. (1985) J. Neurosci. 5, 1556-1562. 33. Paton, J. A., Burd, G. D. & Nottebohm, F. (1986) in The Biology of Change in Otolaryngology, eds. Ruben, R. W., Rubel, E. & van De Water, T. (Elsevier Biomed., Amsterdam), pp. 21-21. 34. Altman, J. (197) in Developmental Neurobiology, ed. Himwich, W. A. (Thomas, Springfield, L), pp. 192-237. 35. Jacobson, M. (1978) Developmental Neurobiology (Plenum, New York), 2nd Ed., pp. 58-6.