Proc. Natl. Acad. Sci. USAVol. 91, pp. 7849-7853, August 1994Neurobiology

The life span of new neurons in a song control nucleus of the adultcanary brain depends on time of year when these cells are born

(adult neurogeness/neuronal replacement/song system/learning/cell death)

F. NOTTEBOHM*, B. O'LoUGHLIN*, K. GOULD*, K. YOHAYt, AND A. ALVAREZ-BUYLLAt*Laboratories of Animal Behavior and tDevelopmental Neurobiology, Rockefeller University Field Research Center, Tyrrel Road, Millbrook, NY 12545

Contributed by F. Nottebohm, April 25, 1994

ABSTRACT The number of high vocal center (HVC)neurons labeled in adult male caries by systemic Injections of[3H]thymidine depended on season and survival time. This wastrue for HVC neurons projecting to the robust nucleus of thearchistriatum and for other HVC neurons that could not beretrogradely filled from the robust nucleus of the archistria-tum. Birds Injected in October and killed 40 days later hadtwice as many labeled HVC neurons as birds injected in Mayand killed 40 days later. However, this difference became muchlarger (5 times) when the birds were allowed to survive for 4months. Whereas more than half of the spring-born neuronsdisappeared between 40 days and 4 months, there was noreduction in the number of fall-born neurons present at the4-month survival point. We infer that seasonal variables affectthe life span of HVC neurons born in adulthood.

It has been known for many years that the developingnervous system overproduces neurons. Many parts of thedeveloping central nervous system retain only a fraction ofthe neurons originally produced (1-5). The conditions thatdetermine this selective death are under intensive study (e.g.,refs. 6-9). Neuronal death is also seen in some diseases oftheadult human brain-e.g., Alzheimer and Parkinson diseases.Several studies (e.g., refs. 10-12) have looked at the factorsor conditions that may prevent or delay neuronal deathduring development or in the adult diseased brain. There ishope that factors that prevent selective death during devel-opment may also rescue neurons that die during disease. Butsuch studies need not be restricted to the developing brain.Neurogenesis and neuronal replacement occur spontane-ously in a part of the song system of adult songbirds. Thisoffers the opportunity to study neuronal death and replace-ment in a healthy adult vertebrate brain and in a system thatcontrols a well-characterized behavior.Oscine songbirds acquire their song by reference to audi-

tory information (13-15). The circuits involved in the acqui-sition and production of learned song have been described(16-19). One ofthe telencephalic nuclei, the high vocal center(HVC), necessary for the production of learned song (16, 20,21), is of particular interest because it incorporates andreplaces neurons in adulthood (22-26). More than half of theadult-formed neurons added to the HVC (26-28) grow anaxon that reaches nucleus robustus archistriatalis (RA), onthe efferent pathway for song production (16). The number ofHVC neurons, including those that project to RA, ceases toincrease in adulthood and neurogenesis in this system is partof a process of neuronal replacement (25, 28, 29).

Earlier work (27) suggested that the incorporation of newneurons into the HVC is higher in early fall, when adultcanaries modify their song, than in the spring, when theyproduce a stable song. The earlier study (27) used a single

survival time, and so it was not possible to know whether thelife span of new neurons varied as a function of birth date.The present report shows that the life span ofHVC neuronsborn in adulthood is significantly longer for cells born in thefall than for cells born in the spring. Survivorship is likely tobe regulated by factors that change in a seasonal manner. Ourobservations identify an important component of a mecha-nism for regulating the turnover of neurons in an adultvertebrate brain.

METHODSAnimals. Twenty-five adult waterslager male canaries from

our breeding colony were maintained indoors at 20'C undera yearly photoperiod regime that matched that ofNew YorkState. Under these conditions, birds in our colony becomesexually mature at 8-10 months of age. All of the birds usedin the present study were 12 months old in May 1991, whenthe experiments started. They all received 10 i.m. injectionsof [3H]thymidine (pectoral muscle) over a 10-day period at24-h intervals (50 t&Ci per injection, 6.7 Ci/mmol; 1 Ci = 37GBq; New England Nuclear). These birds received thePHIthymidine treatment at two times of the year, May(spring) or September (fall), and were divided into fourgroups as described in Table 1.

All birds were weighed, anesthetized with 50 A1 of adilution of Nembutal (Abbott; 10 mg/ml), and given Fluoro-Gold [2% (wt/vol) solution in saline] microinjections into theright and left nucleus RA; this procedure used stereotaxiccoordinates and a protocol described elsewhere (26-28). TheFluoro-Gold microinjections backfilled the soma and pro-cesses ofHVC neurons that projected to RA, yielding a clearand unambiguous visualization of HVC and its boundariesand ofHVC neurons that projected to RA. Four days later thebirds were killed under deep anesthesia induced by 50 Ad ofNembutal at 50 mg/ml. The birds were quickly fixed byintracardial perfusion of4% (wt/vol) paraformaldehyde in 0.1M sodium phosphate (pH 7.4). Brains were embedded inpolyethylene glycol (30) and 6 jim frontal sections were cutusing a rotary microtome. Every fifth section was mountedon gelatin-coated slides, delipidized in xylene, and coveredwith NTB2 nuclear track emulsion (Kodak). After 30 days ofincubation, slides were developed in D19 (Kodak) for 3 minat 17'C. Sections were stained with fluorescent cresyl violet,which allows morphological identification of neurons whilepreserving the Fluoro-Gold fluorescence (31).Miroscopic Analysis. HVC perimeters revealed by the

Fluoro-Gold backfills were traced in all sections mounted andthe area was calculated. Area information was used toestimate the HVC volumes of all birds in the four groups bymultiplying the sum of the areas by the distance betweenadjacent sections. The number of neurons labeled with[3H]thymidine and the number of neurons double labeled

Abbreviations: HVC, high vocal center; RA, robustus archistriatalis.

7849

The publication costs of this article were defrayed in part by page chargepayment. This article must therefore be hereby marked "advertisement"in accordance with 18 U.S.C. §1734 solely to indicate this fact.

Dow

nloa

ded

by g

uest

on

Sep

tem

ber

14, 2

020

7850 Neurobiology: Nottebohm et al.

Table 1. Experimental groupsDate(s)

Group n [3H]Thymidine Sacrifice Survival, daysSpring I 6 May 21-30 Jul 09 40Spring II 6 May 21-30 Oct 05 128Fall I 7 Sep 20-29 Nov 09 41Fall II 6 Sep 20-29 Jan 28 120

Survival was measured from the last [3H]thymidine injection to thedate when birds in a group were killed.

with [3H]thymidine plus Fluoro-Gold were counted in theHVC of six evenly spaced sections. We also counted allFluoro-Gold-labeled neurons and all neurons unlabeled byFluoro-Gold in two sections obtained from the center of eachbird's HVC. The neuronal identity of Nissl-stained cells thatdid not project to the RA was inferred from their clear nucleiwith single or double dark-staining nucleoli. An earlier studythat used electron microscopy showed that HVC cells thatstain in this manner are neurons (24). From the number ofcells counted in each HVC section and from the area ofHVC,we calculated the number of cells per mm2. Cell counting andarea measurements were done at x827 and x82.7 magnifi-cations, respectively, by using the computer-microscopesystems described (32). The measurements and counts de-scribed above were done for both the right and left HVCs.There were no systematic differences in neuron numbersbetween the two sides so we present here an average of thedata for both sides.Grain Counts and Labeling Criteria. We determined the

background labeling by counting the number of exposedsilver grains over a unit of HVC area that included neuropilbut did not include cell nuclei. The background for each birdwas obtained by averaging 50 100-Am2 fields per bird. A cellwas considered labeled if there were seven or more exposedsilver grains over its nucleus; this corresponded to a criterionof >20 times background for all brains.

Cell Size Measurements. We measured the nuclear diame-ters of 50 HVC 3H-labeled neurons that were not backfilledwith Fluoro-Gold and 50 HVC 3H-plus Fluoro-Gold-labeledneurons per bird. Diameters were measured using the dis-placement of a computer cursor, viewed through a cameralucida (32).

Survival Estimates. We compared the proportion of 3H-labeled HVC neurons 40 days and 4 months after [3H]thy-midine injections in late spring and early fall (Table 1). Thiscomparison was made both for 3H-labeled neurons that wereor were not backfilled with Fluoro-Gold. If the number of3H-labeled HVC neurons was smaller at 4 months than at 40days, the difference was attributed to cell death.

Statistical Analysis. Comparisons between experimentalgroups were done using the Student t test (two tailed) withsignificance set at P < 0.05.

RESULTSHVC volumes [(left + right)/2] ranged from 0.191 to 0.384mm3. However, this range of sizes was not homogeneouslydistributed throughout the four experimental groups (Fig. 1).The birds of the spring-injected group that survived 4 months(spring II) and were sacrificed on October 5 had the smallestaverage HVC volumes. These volumes were significantlysmaller than those of birds sacrificed on July 24 (spring I) andNovember 9 (fall I) by 25% (t = 3.19; P < 0.024) and 21% (t= -2.73; P < 0.041), respectively. The differences in meanHVC volume between other groups were not significant (Fig.1).The number of 3H-labeled neurons per mm2 of HVC was

higher in birds injected in the fall than in the spring, and this

I

0.35-

0.3-m

E 0.25-ET-O0.2-

1o 0.15-ai)E' 0.1-

0.05-

o0-

II

Spring 40Jul 9

Spring 120Oct 5

Fall 40 Fall 120Nov 9 Jan 28

FIG. 1. HVC volume of birds in our four groups, spring I and IIand fall I and II. Spring and fall refer to the time when the birdsreceived [3H]thymidine. I and II refer to survival times of 40 days or4 months, respectively. The dates under each histogram bar indicatewhen the birds were killed. The only group whose HVC volumediffered significantly from any of the others was the spring II group,sacrificed in early October. Birds in this group had a significantlysmaller HVC. Data are the mean ± SEM.

difference was greater in the birds that survived 4 monthsthan in those that survived only 40 days (Fig. 2). Whereas thenumber of HVC neurons labeled 40 days after the last[3H]thymidine injection was approximately twice as high inthe fall- as in the spring-injected birds (t = -3.63; P < 0.004),this difference increased to 5.3-fold in the birds killed 4months after the last [3H]thymidine injection (t = -9.33; P <0.001). The increase resulted from two changes. (i) At least56% of the 3H-labeled HVC neurons present 40 days after thelast injection of [3H] thymidine in the spring were no longerpresent by 4 months (t = -5.01; P < 0.002). (ii) The numberof 3H-labeled HVC neurons seen in the fall-injected birds

100

90-.90E 80--

0 70-C1u) 60C:

50 -t

Z 40--o(D 30-

m 20-

10--0

11Spring

11Fal

FIG. 2. Histograms show the number of labeled neurons per mm2of HVC. Horizontal brackets with an asterisk show which pairwisecomparisons yielded significant differences (P < 0.05) betweengroups. Notice that the number of HVC neurons labeled per mm2decreased in the spring between survival periods of 40 days and 4months (spring I and II). In contrast, the number of labeled HVCneurons per mm2 increased considerably between similar survivalperiods in the fall (fall I and II). Data are the mean ± SEM.

Proc. Natl. Acad. Sci. USA 91 (1994)

Dow

nloa

ded

by g

uest

on

Sep

tem

ber

14, 2

020

Proc. Natl. Acad. Sci. USA 91 (1994) 7851

EE 70--a)- 60 --0

50-0cn~

,, 40-

O 30-0

20-

-5 10-a)a)-Q 0---j

I1

SpringSpring Fall

FiG. 3. Number ofnew RA-projecting neurons per mm2 ofHVC.Identification ofthe various groups is the same as in Table 1 and Fig.2. Horizontal brackets with an asterisk show which pairwise com-parisons yielded significant differences (P < 0.05) between groups:spring I vs. spring II, t = 2.55 and P = 0.04; spring I vs. fall I, t =3.52 and P = 0.005; spring lI vs. fall II, t = 8.62 and P < 0.0001; fallI vs. fall II, t = 2.81 and P = 0.02. Data are the mean ± SEM.

increased by 30% between 40 days and 4 months after the last[3H]thymidine injection (t = 2.38; P < 0.034) (Fig. 2).The number of RA-projecting 3H-labeled neurons per mm2

of HVC was also higher in the fall than in the spring (Fig. 3),in both the 40-day and 4-month survival groups. The spring-fall difference was similar to that described above for theoverall population of labeled HVC neurons; the correspond-ing P values are given in Fig. 3. Interestingly, the percentageof new neurons in HVC that projected to RA increasedbetween survival periods of 40 days and 4 months in the fall( = 2.48; P < 0.03) but did not increase in the spring (Fig. 4).The proportion ofnew HVC neurons that projected to RA at4 months was significantly higher in the fall than in the spring(67% vs. 48%; t = -3.04; P < 0.02); the percentage oflabeledHVC cells that projected to RA 40 days after the last[3H]thymidine injection was also slightly higher in the fall

100-

o0 90C-)

a) 80--0.5-70a_

co 60-

50 --C

0° 40-

z 30-

_0CD 20 -

-0 10-

v --

0os

1 1

:_..:I

l

Spring Fall

FIG. 4. Percent of labeled HVC neurons that projected to RA, ineach of the four groups, identified as in Fig. 2 and Table 1. Thepercent did not differ between the spring I and spring II groups.However, it differed between the spring II and fall II groups and alsodiffered between the fall I and fall II groups. The maximum percent-age (67%6) was seen in the fall II group. Data are the mean ± SEM.

Table 2. Nuclear diameters of labeled neuronsNuclear diameter, ian

Group RA projecting Others

Fall I 8.9 ± 0.11 8.0 ± 0.10Fallhl 8.3 ±0.08 7.3 ±0.09Spring I 9.4 ± 0.13 8.6 ± 0.10Spring II 7.5 ± 0.14 7.0 ± 0.09

Data are the mean ± SEM.

than in the spring, but this difference was not significant (Fig.4).The mean nuclear size ofthe labeled HVC neurons became

smaller with longer survival times and this effect was seen inthe RA-projecting and the non-RA-projecting neurons (Table2). This decrease in the nuclear size of the [3H]thymidine-labeled neurons was significant in both the spring-injectedbirds (RA projecting, t = 4.64, and P < 0.006; non-RAprojecting, t = 10.38 and P < 0.0001) and fall-injected birds(RA projecting, t = 3.07 and P < 0.03; non-RA projecting, t= 3.18 andP< 0.019).

DISCUSSIONOur results show that neurons born in adulthood survive forvarious durations oftime depending on the time ofyear whenthey were born. This conclusion comes from comparing twotimes-40 days and 4 months. A 40-day survival period maybe close to the time when one is likely to get a maximumcount of the new cells (33). The longer survival time of 4months was chosen so that we could look at the fate ofneurons born in the spring, when song is stable, and see howmany of these cells survived past the late summer and earlyfall period of plastic song. We used similar survivals for thefall birds. The fall birds were in plastic song when injectedin September and their song became more stereotypedthereafter (34).There were twice as many 3H-labeled HVC neurons in the

fall as in the spring 40-day survival birds. By 4 months thiswas a 5-fold difference. In the spring, there was a sharpreduction in the number of 3H-labeled HVC neurons between40 days and 4 months. In contrast, there was no evidence thatany of the 3H-labeled HVC cells that reached full neuronalstatus in the fall (40-day survival period) had disappeared by4 months. Quite the contrary, new HVC neurons continuedto be added, presumably reflecting some delayed neurogen-esis, delayed migration, or perhaps delayed phenotypic mat-uration-our data cannot distinguish between these threeinterpretations. Birds in the spring II group survived 8 dayslonger than birds in the fall II group, and it could be arguedthat this accounted for the decrease in the number of 3H-labeled cells observed in the spring II-but not in the fallII-group. However, such an explanation is unsatisfactorybecause we know from an earlier study that HVC neuronsborn in the fall persist without attrition for up to 8 monthsafter [3H]thymidine treatment (28). Thus, the robust survivalofthe HVC neurons born in the fall stands in marked contrastwith the disappearance of half of the HVC neurons labeled inthe spring. Clearly, the life span of HVC neurons born inadulthood is not rigidly set but depends on the time of yearwhen these cells are born.The magnitude ofthe differential survivorship discussed in

the previous paragraph may be even larger ifwe consider thatthe volume ofHVC became 25% smaller between 40 days and4 months in the spring-injected birds. When this drop involume is taken into account, only 33% of the neurons bornat the time of the spring [3H]thymidine injections may havesurvived by early fall (44% x 75% = 33%). However, wewould like to be cautious with regard to the apparent drop in

Neurobiology: Nottebohm et al.

Dow

nloa

ded

by g

uest

on

Sep

tem

ber

14, 2

020

7852 Neurobiology: Nottebohm et al.

the fall HVC volume. An earlier study that also used Fluoro-Gold injections into RA to backfill HVC and thus define itsboundaries did not see changes in HVC volume between lateOctober and the following 8-month period (28). If there is anend-of-summer reduction in the volume of HVC, the periodwhen it occurs may be short and restricted tojust afew weeksin September and early October. Naturally occurring sea-sonal changes in the HVC volume of adult male canariesremain controversial (35-37) and must await a more detailedstudy using larger samples.We do not know how seasonal differences in the survival

of the new neurons come about. The hormonal profiles ofadult male canaries differ markedly in the spring, summer,and fall (38). Blood testosterone levels are high in the spring,drop during the summer, and increase in the fall. Manyneurons in HVC, including those that project to RA, accu-mulate testosterone and its metabolites and have androgenreceptors (39-41). Testosterone influences the physiologyand structure of some neurons (e.g., refs. 42 and 43), andthese effects have been shown in the song system (44-47). Iftestosterone plays a role in the survival of neurons born inadult males, then it seems to affect similarly those HVC cellsthat project to RA and others, perhaps interneurons, that donot project to RA. An influence of testosterone on neuronalsurvival has been described in the nucleus RA of the devel-oping song system (48). A companion paper (49) shows thattestosterone or its metabolites increases the recruitmentand/or survival of new HVC neurons in adult female canar-ies.The long survival of the HVC neurons born in the fall

makes sense if these cells participate in encoding the newsong repertoire that the birds acquire in late summer andearly fall and retain until the next spring (28, 29). What, then,is the role (if any) of the new neurons acquired in the spring,when adult male canaries are in stable song? More than halfof these neurons will be discarded 3 months later. Are theydiscarded because they have no role or is their role ashort-lived one? Thus, whereas seasonal differences in thesurvival ofnewHVC neurons are now firmly established, thefunctional significance of these differences remains specula-tive.The nuclear size ofthe new HVC neurons that projected to

RA and that of other new HVC neurons that did not projectto RA diminished between 40 days and 4 months of survival.This drop in nuclear size is similar to that described previ-ously (28) and may reflect a pruning of dendritic and axonalbranches. Interestingly, though, the drop in the diameters ofthe 3H-labeled HVC neurons was greater in the spring-injected than in the fall-injected birds and, thus, may havebeen related to the differences in the survival of the spring-born and fall-born neurons.The changes in nuclear size that occurred between 40 days

and 4 months of survival are not likely to have affectedsignificantly the counts of labeled cells at these two times.Counts of 3H-labeled cells with nuclear diameters within therange we encountered, using tissue sections 6 ,um thick, arevery unlikely to result in over or under estimates of the truenumber of these cells. In fact, all of our 3H-labeled neurons,whether their mean diameter was close to 7 or 9 um, wereprobably equally likely to be counted (50). In accordancewith the above, there was no simple relation between thenuclear size and the number of 3H-labeled neurons in thedifferent experimental groups.Our results demonstrate that the life span ofHVC neurons

born in the spring is considerably shorter than that of HVCneurons born in the fall. We have no comparable evidence forseasonal differences in the birth rate of new HVC neurons.Our data show that twice as many new neurons are present40 days after [3H]thymidine injection in the fall as in thespring, but even by this time there could have been differ-

ential survival. Another study suggested that as many astwo-thirds of neurons born in adulthood disappear-andpresumably die-before they acquire a mature neuronalphenotype (33). It is clear from the longer survival periods inthe present study that the half-life of the new neurons willdetermine the extent to which neurons born at a specific timeof year contribute to the functions of the HVC. We know nowthat there is not a set number of months that a new neuronmust live, followed by unavoidable old age and death.Instead, seasonal factors present at the time of birth or duringthe life span of that neuron determine the time of its demise.This process might be stochastic or reflect in a very directmanner a neuron's position in a circuit and its patterns of use.In addition, neurotrophic factors that affect neuronal survivalmay be more available at some times of the year than atothers. The HVC of adult songbirds may be a good place toidentify factors that affect neuronal survival and study theirmanner of action.

We thank John Kim, Carlos Lois, and M. Notteborhn for helpfuleditorial comments. This work was supported by Public HealthService Grants MH18343 to F.N. and NS28478 to A.A.-B. andpartially by Biomedical Research Support Grant S07RR07065.

1. Hamburger, V., Brunso-Bechtold, J. K. & Yip, J. W. (1981) J.Neurosci. 1, 60-71.

2. Jacobson, M. (1991) Developmental Neurobiology (Plenum,Press, New York).

3. Cowan, W. M., Fawcett, J. W., O'Leary, D. D. M. & Stan-field, B. B. (1984) Science 225, 1258-1265.

4. Oppenheim, R. W. (1981) in Studies in Developmental Neuro-biology: Essays in Honor of V. Hamburger, ed. Cowan, W. M.(Oxford Univ. Press, Oxford), pp. 74-133.

5. Oppenheim, R. W., Schwartz, L. M. & Shatz, C. J. (1992) J.Neurobiol. 23, 1111-1115.

6. Hamburger, V. & Yip, J. W. (1984) J. Neurosci. 4, 767-774.7. Barde, Y. A. (1989) Neuron 2, 1525-1534.8. Oppenheim, R. W. (1989) Trends NeuroSci. 12, 252-255.9. Thoenen, H. (1991) Trends NeuroSci. 14, 165-170.

10. Hyman, C., Hofer, M., Barde, Y., Juhasz, M., Yancopoulos,G. D., Squinto, S. P. & Lindsay, R. M. (1991) Nature (Lon-don) 350, 230-232.

11. Oppenheim, R. W., Qin-Wei, Y., Prevette, D. & Yan, Q. (1992)Nature (London) 360, 755-757.

12. Sendtner, M., Schmalbruch, H., St6ckli, K. A., Carroll, P.,Kreutzberg, G. W. & Thoenen, H. (1992) Nature (London) 35,502-504.

13. Thorpe, W. H. (1958) Ibis 100, 535-570.14. Konishi, M. (1965) Z. Tierpsychol. 22, 770-783.15. Marler, P. (1970) J. Comp. Physiol. Psychol. 71, 1-25.16. Nottebohm, F., Stokes, T. M. & Leonard, C. M. (1976) J.

Comp. Neurol. 165, 457-486.17. Nottebohm, F., Kelley, D. B. & Paton, J. A. (1982) J. Comp.

Neurol. 207, 344-357.18. Okuhata, S. & Saito, N. (1987) Brain Res. Bull. 18, 35-44.19. Bottjer, S. W., Halsema, K. A., Brown, S. A. & Miesner,

E. A. (1989) J. Comp. Neurol. 279, 312-326.20. McCasland, J. S. (1987) J. Neurosci. 7(1), 23-39.21. Simpson, H. B. & Vicario, D. S. (1990) J. Neurosci. 10,

1541-1556.22. Goldman, S. A. & Nottebohm, F. (1983) Proc. Natl. Acad. Sci.

USA 80, 2390-2394.23. Paton, J. A. & Nottebohm, F. (1984) Science 225, 1046-1048.24. Burd, G. D. & Nottebohm, F. (1985) J. Comp. Neurol. 240,

143-152.25. Nottebohm, F. (1985) in Hope for a New Neurology, ed.

Nottebohm, F. (New York Acad. Sci., New York), pp. 143-161.

26. Alvarez-Buylla, A., Theelen, M. & Nottebohm, F. (1988) Proc.Natl. Acad. Sci. USA 85, 8722-8726.

27. Alvarez-Buylla, A., Kim, J. R. & Nottebohm, F. (1990) Sci-ence 249, 1444-1446.

28. Kim, J. R., Alvarez-Buylla, A. & Nottebohm, F. (1991) J.Neurosci. 11, 1756-1762.

29. Kim, J. R. & Nottebohm, F. (1993) J. Neurosci. 13,1654-1663.

Proc. Natl. Acad. Sci. USA 91 (1994)

Dow

nloa

ded

by g

uest

on

Sep

tem

ber

14, 2

020

Neurobiology: Nottebohm et al.

30. Alvarez-Buylla, A. & Clayton, D. F. (1993) in PolyethyleneGlycol as an Embedment for Microscopy and Histochemistry,ed. Gao, K. (CRC, Boca Raton, FL), pp. 71-80.

31. Alvarez-Buylla, A., Ling, C.-Y. & Kirn, J. R. (1990) J. Neu-rosci. Methods 33, 129-133.

32. Alvarez-Buylla, A. & Vicario, D. S. (1988) J. Neurosci. Meth-ods 25, 165-173.

33. Alvarez-Buylla, A. & Nottebohm, F. (1988) Nature (London)335, 353-354.

34. Nottebohm, F., Nottebohm, M. E. & Crane, L. (1986) Behav.Neural Biol. 46, 445-471.

35. Nottebohm, F. (1981) Science 214, 1368-1370.36. Gahr, M. (1990) J. Comp. Neurol. 294, 30-36.37. Johnson, F. & Bottjer, S. W. (1993) J. Neurobiol. 24, 400-418.38. Nottebohm, F., Nottebohm, M. E., Crane, L. A. & Wingfield,

J. C. (1987) Behav. Neural Biol. 47, 197-211.39. Arnold, A. P., Nottebohm, F. & Pfaff, D. W. (1976) J. Comp.

Neurol. 165, 487-511.

Proc. Nati. Acad. Sci. USA 91 (1994) 7853

40. Brenowitz, E. A. & Arnold, A. P. (1992) J. Neurobiol. 23,871-880.

41. Sohrabji, F., Nordeen, K. W. & Nordeen, E. J. (1989) BrainRes. 488, 253-259.

42. Breedlove, S. M. (1992) J. Neurosci. 12, 4133-4142.43. Jones, K. J. (1993) Metab. Brain Dis. 3, 1-18.44. Nottebohm, F. (1980) Brain Res. 189, 429-436.45. Luine, V., Nottebohm, F., Harding, C. & McEwen, B. S.

(1980) Brain Res. 192, 89-107.46. Bleisch, W., Luine, V. N. & Nottebohm, F. (1984)J. Neurosci.

4, 786-792.47. DeVoogd, T. J. & Nottebohm, F. (1981) Science 214, 202-204.48. Gurney, M. E. (1981) J. Neurosci. 6, 658-673.49. Rasika, S., Nottebohm, F. & Alvarez-Buylla, A. (1994) Proc.

Nati. Acad. Sci. USA 91, 7854-7858.50. Clark, S. J., Cynx, J., Alvarez-Buylla, A., O'Loughlin, B. &

Nottebohm, F. (1990) J. Comp. Neurol. 301, 114-122.

Dow

nloa

ded

by g

uest

on

Sep

tem

ber

14, 2

020