The Condor 92:93S-946 Q The cooper ornithological society 1990

SURVIVAL ESTIMATES OF INDIGO BUNTINGS: COMPARISON OF BANDING RECOVERIES AND LOCAL OBSERVATIONS

ROBERT B. PAYNE AND LAURA L. PAYNE Museum of Zoology and Department of Biology, The Universityof Michigan, Ann Arbor, MI 48109

Abstract. We comparedestimates of adultsurvival of IndigoBuntings (Passerina cyanea) from Bird BandingLaboratory banding recoveriesand from field observationsof color- bandedbirds in breeding areas in southern Michigan. Color-banded males in two breeding populations in Michigan had similar survival (S, = 2.41 years, P, = 0.586 + 0.018 SE; S, = 2.08 years, P, = 0.519 -t 0.021 SE). Survival estimatesin females were lower (P. = 0.465 and 0.335). Banding recovery data indicated a mean adult survival S, in males of 2.34 years and a probability of survival to the next year P, of 0.57 k 0.037 SE. Survival estimates from local observationsand banding recoveriesdid not differ significantlyin either sex. The agreementof survival estimatesin banding recoverieswith an area-freesampling distribution and in local observations indicates that local observations are not biased againstdispersal when the study areas and populations are sufficiently large. Key words: Indigo Bunting; Passerinacyanea; banding recoveries; demography; dispersal; jinite area sampling;population biology;adult survival.

INTRODUCTION are open to some uncertainty, however, as the Local observations of individual birds and anal- disappearance of a bird may be due to either ysis of banding recoveries provide more infor- dispersal or mortality. Nevertheless, where mation on demography and population structure banding recoverieshave been compared with ob- than does either technique alone. Recent discus- servations of marked local populations, the es- sions of estimating survival from regional band- timates of survival have been similar (Dobson ing programs emphasize statistical methods for 1990). describing variance and the implications of age- It is desirable to check the validity of local dependent survival (Anderson et al. 1985, survival estimates by comparing them with es- Brownie et al. 1985, Morgan and North 1985, timates from banding recoveries. If local esti- North 1987). Regional banding recovery data are mates of survival are lower than estimates from area-free, that is, the chancesof recovery are in- banding recoveries, we may suspecta problem dependent of the bounded area that comprises a in observing surviving birds due to their disper- local study population. However, the North sal. Here we compare estimates of survival of American Bird Banding Laboratory (BBL) re- color-marked individual Indigo Buntings, Pus- covery retrieval file currently does not include serina cyanea, in two breeding populations with returns when a bander finds a bird in the original estimates of survival from BBL banding recov- banding site, so the recoveriesare limited to birds eries. found in a second area or reported dead by the METHODS bander. In addition to regional banding and recoveries, LOCAL OBSERVATIONS local observationsand recapturesof marked birds The study areas were on and around the E. S. are useful in determining survival and are es- George Reserve in Livingston County, an area pecially useful when variation in survival can be of 10 km*, and 3 km NE of Niles, in CassCounty, related to the behavior and observed histories of an area of 4 kn?, both in southern Michigan. the individuals (Clutton-Brock 1988, Newton Nearly all territorial males and about half of the 1989). Survival estimates in local observations breeding female Indigo Buntings were color- banded in the two areasfrom 1978 through 1985. ’ ’ Received 12 February 1990. Final acceptance 13 Each bunting was marked with an USFWS metal June 1990. band and three color bands. The color bands

19381 INDIGO BUNTING RECOVERIES AND SURVIVAL 939 were plastic coils wrapped twice around the tar- ed and found later, and from 1965 the tile ex- sus and sealed with acetone. cludes birds that returned to their banding site. Males were aged at capture by the greater pri- In the present study of survival, a recovery is mary coverts, which are brown in yearlings and restrictedto an encounter in a calendar year after blue in older adults. The greater primary coverts a bird was banded and at a site other than the have long been used to age male Indigo Buntings banding site. A return is an encounter by a band- based on a study of museum specimens(Dwight er with a bird at its banding site or in the same 1900). The coverts were reliable indicators of age 10’ block. in males that we banded as nestlings in the pre- The records were examined for time elapsed vious year and recaptured on their natal area, from banding to recovery, and the records in- and in banded males whose minimal age was cluding the local returns prior to 1965 involved known when they were recaptured from year to 295 recoveries. An HY is a bird banded in the year (Payne et al. 1988, Payne and Payne 1989). calendar year of hatching (not including birds They also were reliable indicators of age in birds banded as nestlings), an SY is a bird banded in reared and observed through molts in 1 and 2 the calendar year after hatching, an ASY is a bird years in captivity (Payne 1982). Age was un- banded in the secondcalendar year or later after known in females unless they had been banded hatching, and an AHY is a bird banded at least in an earlier year. The greater primary coverts one calendar year after hatching. In male Indigo of females of known age showed no consistent Buntings, both SYs and ASYs are sexually ma- tendency for older birds to have blue coverts, or ture and breed, though SY successis lower (Payne for females to be more bluish in later years. Fe- 1982, 1989; Payne et al. 1988). An adult male males banded as nestlings in the previous year was banded as an SY or an ASY in the same year and returning to breed in their natal area (Payne or in an earlier year. An older adult is a bird 2 et al. 1987, Payne and Payne 1989) usually had years or older. brown greater primary coverts, but so did older Cases were selected from the BBL recovery females that were examined from year to year. retrieval file for estimating adult survival if they First-year females did not have darker outer pri- met the following criteria: (1) the bird was re- maries and lighter inner primaries in spring, as covered at least one calendar year after banding, Rohwer (1986) suggestedfrom a study of spec- (2) the bird was banded in Canada or the United imens collected in autumn. States, (3) when the bird returned to the same Field observations were made from early May 10’ block and was recordedby a bander, survival through August. Beginning in 1978, in each study estimates were restricted to birds recovered in area we observed all 100-l 50 resident breeding 1965 or later, and (4) the bird was banded no pairs of buntings for survival and breeding later than 1983, so survival estimates would not through 1985 (Payne 1982, 1989; Payne et al. be biased towards short-lived birds. Recoveries 1987, 1988; Payne and Westneat 1988; Payne with missing data on date and locality were ex- and Payne 1989), and we determined survival cluded. The data include birds found dead, birds through 1990. Observations on survival of res- recaptured and released, and birds of unknown ident and breeding birds on the study area are condition at recovery. thought to be complete. ESTIMATION OF SURVIVAL RECOVERIES FROM THE Survival was estimated from the same expres- BIRD BANDING LABORATORY sion for the banding recoveries and for local ob- The banding retrieval file of Indigo Buntings was servations. The selection of casesfor the sample read from 96,474 birds banded from 1955 estimates differed between the two sets of data. through 1987, including 66,360 from 1965 For estimating survival from the banding recov- through 1984. Nearly all buntings (more than eries, we included all birds recovered in a year 95%) were banded as independent birds, either after they were banded. For the local observa- adults in spring or autumn or birds of the year tions we included all resident birds in their year in autumn, and few were banded as nestlings. of banding, and observations in later years were Most were banded in migration during May, Sep- used to determine the terminal year, or the year tember, and October. of banding for birds that were not seen in a later In BBL terminology, a recovery is a bird band- year. Lifetime survival in each set of data was 940 ROBERT B. PAYNE ANDLAURA L. PAYNE calculated combining all years into a composite (1) SY males often move from one territory to cohort (Seber 1973). another during their first breeding season,(2) 28 We estimated survival from the proportion of days is the minimal time to nest and rear the birds surviving to the next year from the number young to fledging (Payne 1989), and (3) birds alive in the beginning of the year (Lack 195 1). present for a shorter period may have dispersed Beginning in year 1, s1= the proportion of birds and settled elsewhere.The secondcriterion pro- that survive only the year of banding, and n, = vides an allowance for birds being overlooked the number alive at the beginning of the year. for a short time when they first arrived and for Survival S, was estimated as the sum of bird- local dispersal between the study area and the years through the oldest survivor, divided by the nearby area within the male’s first year. Survival number alive in the beginning of the year, or was observed by repeated field census through 1990. Within each study area no demographic s, = s,n, + sznz + ssn3. . . + s,ni differences were found among the six cohorts, n, + n2 + n, . . . + ni . using an analysis of variance (mean survival for females at Niles varied significantly among years, The value S, indicates the mean expected num- but the variances also differed significantly, so ber of years of survival and was used to estimate an inferenceof significantmean differencesamong the probability of surviving from one year to the year cohorts is unwarranted). The six cohorts next, P,, where were combined within each area and sex to be P, = [l - (l/S,)]. analyzed as a composite cohort. Survival esti- mateswere similar to thoseassessed through 1987 The standard error (SE) of the survival estimate (Payne 1989). The data in Table 1 include cor- (P,) was estimated (Haldane 1955, Msller 1983, rections, an additional cohort, three additional Dobson 1990), where N = total bird-years, as years of observation of survival, and estimates SE=(l -PJ of confidence limits. Over all ages, male P, was x [P,/(N - n, - n, . . . - nJlexpn. not significantly different between the George Reserve and Niles (t = 1.33, P > 0.05). The estimate of variance allows inference of At the George Reserve, males were not sig- whether the estimatesof P, are statistically equiv- nificantly less likely to return after their SY year alent (Conroy and Williams 1984, Pyke and than after years 2 to 6 (SYs, 125 returned, 135 Thompson 1987, Dobson 1990). Two-tailed Stu- did not; older males, 153 returned, 120 did not, dent’s t-tests and chi-square tests were carried counting each male-year as a separatecase, x2 = out to determine whether the P, values differed 3.46, P > 0.05). Similarly at Niles, SY males using a significance level of 0.05. Estimates of were as likely to return as older males (SYs, 112 survival based on banding recoveries were com- returned, 8 1 did not; older males, 137 returned, pared with estimatesbased on direct observation 89 did not, x2 = 1.85, P > 0.1). No decreasein of color-banded birds returning to the breeding survival was observed in the later years of life. populations. Females were netted at the nest site and most were of unknown age when banded. Table 2 in- RESULTS cludes the minimum survival for females on ter- SURVIVAL OBSERVATIONS OF ritories in the part of the study area where nearly LOCAL RETURNS all breeding females had been banded in the pre- In the breeding populations in Michigan, we ob- vious year. Although some may have dispersed served the color-banded Indigo Buntings from from a breeding area in an earlier season, most one season to the next. A total of 260 banded females were probably in their first year when males at the George Reserve and 215 banded they were first banded. About 10% had been males at Niles were used for estimation of sur- banded on the area as nestlings. Minimum sur- vival. This sample includes all SY males that vival to the next seasondid not differ in females were banded during the breeding seasonin 1979 known to be in their first year and in females through 1984 and that remained on the study banded when their age was unknown (Payne area 28 days or longer, and all SY males that 1989). Female survival, P,, was significantly low- remained a shorter time but returned the next er at the George Reserve than at Niles (t = 2.09, year. The first time criterion is necessarybecause P < 0.05). INDIGO BUNTING RECOVERIESAND SURVIVAL 941

TABLE 1. Survival of male Indigo Buntingsbanded as SYs from 1979 through1984 in southernMichigan and observedthrough 1990. ’

Years observedon studyarea SuNival Area R 1 2 3 4 5 6 7 8 9 $1 P, SE GeorgeReserve 260 135 52 34 16 10 3 0 2 2.08 0.519 0.021 Niles 215 81 63 27 18 8 1; 3 1 2 2.41 0.586 0.018 ’ ’ In 1990, one male(9 yearsof age)was stillalive at the reserve,and four males(two of 7 years,two of 9 years)were stiIl alive at Niles;these birds are included. 2 s, = meanexpected number of yearsof suNival. 3 P. = probabilityof survivingfrom one year to the next.

BANDING RECOVERIES Michigan (BBL and George Reserve, t = 0.66, P Of the 295 recoveries of Indigo Buntings in the > 0.5; BBL and Niles, t = 0.21, P > 0.5). The BBL recovery retrieval file, 105 met the criteria estimatesof female survival P, also did not differ for estimation of survival. An additional 103 (BBL and George Reserve, t = 0.34, P > 0.5; birds returned to their banding area before 1965 BBL and Niles, t = 0.06, P > 0.5), although the and were analyzed separately (Table 3). For the sample size of BBL recoveries is low and statis- 105 recent recoveries, S, = 2.19 years, and P, = tical inference is problematic. The estimate from 0.54, combining males and females, and com- the banding recoverieswas intermediate between bining adults and birds captured as HY where the estimates from observations in the two sex was not determined. For males, S, = 2.34 breeding populations both for male and for fe- years, and P, = 0.57. For females, survival is male buntings. lower but the sample size is small and the vari- ance is large. The survival estimates for local DISCUSSION returns before 1965 were lower than for the more LOCAL OBSERVATIONS, SURVIVAL, recent and nonlocal recoveries,even though only AND THE PROBLEM OF DISPERSAL the latest return was used for returning birds that Local population studies allow direct observa- were reported more than once. In the recent re- tion of minimum lifetime survival, rather than coveries, 21 buntings were reported to be alive an indirect reconstruction from an inferred age when recovered, with S, = 1.95, P, = 0.488, and structure in a regional banding recoveries pro- SE = 0.113, suggestinga lower survival than in gram. The major limitation of local observations the total sample of 105 birds. is that birds may leave the area and settle else- COMPARISON OF SURVIVAL IN where within a seasonor between seasons.The LOCAL POPULATIONS AND question of the degree of philopatry is perhaps BANDING RECOVERIES the greatest uncertainty in extracting demo- The survival estimates from the marked breed- graphic statistics from observation of a marked ing populations at the George Reserve and at population of mobile animals such as birds (Se- Niles were similar to survival estimates from the ber 1973, Mewaldt and King 1985). Uncertainty BBL banding recoveries. The estimates of P, for is greatestif the population area is small, and if males did not significantly differ between the 60 point sampling (for example, fixed net sites)rath- recent BBL recoveries and survival observed in er than observational censusingof an area is used

TABLE 2. Survival of female Indigo Buntingsbanded from 1979 through1984 in southernMichigan and observedthrough 1990.

Years observed on studyarea SUWiVal Area n 1 2 3 4 5 6 7 So’ P.’ SE GeorgeReserve 226 159 42 13 6 2 4 0 1.50 0.335 0.036 Niles 243 127 63 30 11 7 3 2 1.87 0.465 0.023 ’ ’ s, = meanexpected number of yearsof survival. 2 P. = probabihtyof survivingfrom one year to the next. 942 ROBERT B. PAYNE ANDLAURA L. PAYNE

TABLE 3. Number of yearselapsed between banding and recoveryof Indigo Buntingsin the BBL recovery retrievalfile from 1926 throughJanuary 1988, for birds bandedby 1983 and recoveredin a later year. Row 1 includesall recoveriesof birdsbanded in 1965 or later, and birds in earlieryears except for live returns.Row 4 includesthe local returnsto the bandingsite before1965.

Years elapsed frombanding to recapture survival RLXOVCXi~ n I 2 3 4 5 6 I a So’ P,’ SE Recent,all 105 48 24 15 12 4 1 1 8 2.19 0.54 0.030 Recentmales 4: 22 11 12 9 4 8 2.34 0.57 0.037 Recentfemales 10 9 0 2 0 0 1.71 0.42 0.288 Returns 103 65 15 15 6 1 0 0 1 1.72 0.42 0.044

’ ’ s, = mean expected number of years of survival. 2 P, = probabiity of survivingfrom one year to the next. to determine the presenceor absenceof individ- season (Payne et al. 1987) suggestsome long- ual birds. distancenatal dispersal,though theseare the only Several criteria can be used to test the as- recoveries in the BBL recovery retrieval file for sumptions and accuracy of observations of sur- buntings banded as nestlings. The natal returns vival and the degreeto which survival estimates and recoveries are too few to estimate survival may by biased downward by dispersal outside in the first 12 months of life. the area of observation. Criteria to determine the (2) In local observations, individuals may dis- effectiveness of observational population sam- persewithin or out of the study area. When mor- pling include the degree of (1) natal philopatry tality and emigration cannot be distinguished, to an area, (2) breeding philopatry to an area, (3) dispersal out of a study area would lead to un- completeness of sampling, with all individuals derestimation of survival. We have observations that are still alive and present being observed in of hundreds of Indigo Buntings that indicate a each season,(4) agreement of survival estimates high degreeof local site fidelity and area fidelity. from breeding populations in different areas,and In Michigan most birds returned to the same (5) agreement of survival estimates in breeding territory in successiveyears and most relocations and wintering areas.In evaluating criteria (4) and were within 200 m of the previous site, even (5), discrepancies might be interpreted as de- though the population size and the study area mographic or dispersal differences among pop- were large enough to detect movement over ulations, and additional sampling would be re- greaterdistances. Several moved within a season quired. All five criteria involve philopatry relative and nested in territories as far as 1 km apart, and to a finite area. Since the larger the area, the some males and females were found nesting in greater the philopatry, on a scale of a single ter- their second year as far as 2 km from their year- ritory to an entire continent, it is important to ling territory. In male Indigo Buntings nearly all sample an area that is large enough to include casesof breeding dispersal involved SYs. Simi- most surviving birds that may disperseon a local larly, nearly all dispersal observations between scale. years were of males banded in the previous year (1) Few Indigo Buntings returned to their natal as SYs. Nevertheless, in parts of the study area areas. Within the natal area as defined by the 4 where buntings had been banded for the previous and 10 km2 of our study areas in Michigan, 147 4 years, 3 l-34% of the breeding adults older than of 2,544 nestlings banded were seen to return a year had immigrated into the study area in the through 1988, and most surviving young were first instance as adults rather than as SYs (Payne not recovered in a later year (Payne et al. 1987, 1989). All buntings that had been banded as nest- Payne 1989, Payne and Payne 1989). Nolan et lings and first returned to the study area as older al. (1975) found three of 185 banded nestlings adults were observed on the area in their second returning to an area of 75 ha in Indiana in the year after hatching (Payne et al. 1988). The ob- 197Os, and none of 21 young banded by W. L. servations suggestthat after the first year, Indigo Thompson in 600 ha at the George Reserve in Buntings generally return to the area of the pre- the 1960s were seen again (Emlen et al. 1975). vious year. Two recoveriesat 52 and 3 50 km from the George Ecological changes may occur that affect the Reserve in Michigan in the following breeding age structure of a local population from year to INDIGO BUNTING RECOVERIES AND SURVIVAL 943 year through survival and dispersal. The lower in Michigan. Nolan et al. (1975) found 60% sur- survival of females at George Reserve than at vival in breeding males and 50% survival in fe- Niles might be due to breeding dispersal though males in Indiana. No significant change in sur- higher predation at the nest may have occurred vival with agewas observedin breeding buntings there as well. The natural successionof vegeta- in Indiana. Using Haldane’s (1955) technique for tion and the invasion of fields by introduced estimating survival where birds were missed in shrubs,especially autumn-olive (Elueugnus um- a year of observation, Blake (1969) calculated a bellata),which overgrow the native vegetation survival of 58.5% in North Carolina; age, sex, and destroy the habitat of buntings, may have and breeding status were not noted. forced dispersal or emigration of buntings from In addition to survival estimates, the maxi- George Reserve. During our 12 years of obser- mum survival observed in banded buntings is vation on the reserve, buntings decreasedto less similar in other populations. The oldest known than 50% of their numbers in 1978, and the de- buntings, neither of which were in the BBL re- creasewas associatedwith the changein habitat. covery retrieval tile, were two females banded at The BBL banding recoveriesfile was examined Powdermill Nature Reserve in Pennsylvania, for casesof buntings that were recovered in dif- where numbers comparable to the Michigan ferent areas either within or between breeding population have been banded (Leberman and seasons.In 10 cases,the dates of banding and Wood 1983). One was banded in 1974 as an recovery suggestpossible breeding dispersal.The AHY and was recaptured for a minimum age of dates of all but one involved May, and since 10 years (Mulvihill and Leberman 1984, Leber- many buntings are in migration in May, these man et al. 1985). The other was banded as an may not have been residents. The dates for a AHY in 1972 and recaptured in 1980 with a male AHY 660-33373, banded by K. Petts near brood patch, indicating breeding at 9 years (Le- Alpena, Michigan, on 17 June 1962 and recov- berman 1982). ered 65 km west on 19 June 1964, suggestbreed- (5) Banding returns show philopatry of Indigo ing dispersal. Both dates are too late for spring Buntings to their wintering areas in the tropics migration. The bird may have moved within a (Van Tyne 1932; Nickel1 1968; Downer 1972; seasonor between seasons. Fisk 1978, 1979; Rappole and Warner 1980). (3) Observations appeared to be reasonably Fisk (1978) observed 80 winter returns of bunt- complete, as nearly all birds that were seen in ings banded from 1968 to 1976 in southern Flor- three or more seasonswere observed in all suc- ida. Two returned 8 years after banding. Sex and cessive seasons.One male was not observed in age at banding were not indicated; most of her his third year but reappeared in the following buntings were banded as HYs (Fisk 1979). From year. Another was rediscovered 2 km from his the data we estimate S, = 2.16 years, P, = 0.54, banding site 6 yearsafter banding; he was present and SE = 0.035. The survival estimate from only during his SY year at the banding site. The southern Florida is close to that observed in the proportion of caseswhere a male may have been breeding population, and suggestsa similar de- overlooked or spent a season off the area, and gree of philopatry in wintering and breeding was not seen in every year until the final year of buntings. The only recovery in different areas in observation, was less than 0.2%. Several color- two winter seasonswas a male ASY banded on banded females were seen after a year when they the Gulf coast of Florida on 9 January 1984 and were not observed, so some females may have repeating on five occasions through 14 March. avoided observation in their last year as well. It was captured again in the Bahama Islands on These unaccounted absencesinvolved about 6% 29 December 1984 and did not repeat during of the cases,where a case is a female in a season that seasonat its banding site on the Gulf coast, (Payne 1989). although it was still alive and was recovered in (4) Indigo Buntings return to their breeding Nova Scotia on 26 May 1986 (Stedman and Sted- areas with a high area fidelity from year to year man 1988; A. B. Stedman, pers. comm.). in all breeding populations where they have been observed (Emlen et al. 1975, Nolan et al. 1975, SURVIVAL ESTIMATES FROM Carey and Nolan 1979, Carey 1982, Payne et al. BANDING RECOVERIES 1988, Payne 1989). Field studies in other areas Methods of estimating survival from banding re- suggestsimilar philopatry and survival to that covery data and its statistical interpretation have 944 ROBERT B. PAYNE AND LAURA L. PAYNE receivedconsiderable attention, especiallyon age- Fewer females were recovered, probably as a re- specific differences in survival, which are most sult of their later migration and arrival on the difficult to determine from birds banded as young breeding grounds (Johnston and Downer 1968, of the year (Iakhani and Newton 1983, Ander- Taber and Johnston 1968, Blake 1969, Carey son et al. 1985, North 1987). Adult survival can 1982, Quay 1987) after most bandershave ceased be estimated with greater confidence (Dobson netting, as well as their inconspicuous appear- 1990). ance and behavior. This sampling bias in sea- Recovery data of birds banded as adults are sonal activity would not lead to a false lower subject to several sampling biases and uncer- survival estimate for females, because the bias tainties. Some assumptions in estimating a single would be the same acrossyears and the chance value of adult survival from capture-recapture of recovery of females late in the seasonshould data are that (1) probability of death is indepen- not change with the number of years since band- dent of age after a certain date, (2) survival is ing. The lower survival in female buntings sug- similar in different subsetsof data (sample ho- gested in the few recoveries available is sup- mogeneity), (3) loss of bands through age is neg- ported by the local observations, so is probably ligible, and (4) the sample is representative of not an artifact of seasonalbanding and recovery the population (Seber 1973, Lakhani and New- differences between the sexes. ton 1983, Anderson et al. 1985, Brownie et al. (3) Recovery data may underestimate surviv- 1985, Morgan and North 1985, North 1987, al, as most recoveries of Indigo Buntings were Dobson 1990). of birds found dead, and the survivors may have (1) Survival can be estimated directly when it lived longer. Few banded buntings (0.1%) were doesnot vary with age,and age-specificestimates recovered, implying that the adult lifetimes of are possible only in birds of known age at band- birds recovered may not be representative. Nev- ing. In the BBL recovery retrieval file, most bunt- ertheless,the survival estimates were consistent ings were of unknown age.A few males were aged with survival observed in local breeding popu- as SYs when banded, and a few buntings were lations, so they appear to be adequate as a first banded as HYs but most HYs were not sexed; approximation of survival in this species. so the data are of limited value for comparison (4) Loss of bands and subsequent failure to of age-specific survival of adult males and fe- recover data for the birds alive many years after males. By using the criterion that birds were re- banding was not a problem in Indigo Buntings. covered in the year after banding or in a later In the local observations, only one color-banded year, the survival estimates in the present study bunting was seen without a metal band in a later are from birds that survived through the period year. He lost that foot between his second and of juvenile dependence and the first long-dis- third year; he was recognizable by song,location, tance migration, i.e., times of highest mortality and the color bands on the other foot. Recaptures in birds (Lack 1954). Indigo Buntings do not of birds in their sixth and seventh year showed show a senescentincrease in mortality, at least the USFWS bands to be intact and legible. An- through 95% of their lifetimes. nual loss of color bands was less than 1% and in (2) Reasonable sample homogeneity is sug- each case the bird was recognizable by the re- gestedby the lack of significant demographic dif- maining bands, by location within the study area, ferencesamong year cohorts in the local obser- and in the male by song, which usually was un- vations. The survival or return of SY males in changed from year to year (Payne et al. 1988). the following year was lower than for older males in the George Reserve population, though not at COMPARATIVE SURVIVAL ESTIMATES Niles. Age was indicated for too few birds in the Survival estimates of Indigo Buntings were not BBL data for comparison of survival of age statistically distinguishable in observations of groups.Combining years and cohortsfor analysis color-banded birds and in the BBL recovery data. in the BBL data is justified by the comparable The BBL estimatesof survival were intermediate netting effort over the years included, judging between the estimates in the two color-banded from the similar number of birds banded over a populations. Few studieshave compared surviv- 40-year period. al estimatesfrom local observations and regional Although male buntings may slightly outnum- banding recoveries.Perrins (197 1) suggestedthat ber females, nearly all of known sex were males. local studies might lead to higher estimates of INDIGO BUNTING RECOVERIES AND SURVIVAL 945 survival than regional recovery data due to sam- A. B. Stedman,W. L. Thompson,and L. H. Wallcin- pling differences of the populations. In a com- shawgave information about their bandingrecords. G. E. Hill, N. K. Klein, R. C. Leberman,M. P. Lombardo, parative study of survival estimates based on V. Nolan. Jr., and W. B. Ouav commented on the local recaptures of breeding birds and on data manuscript. Field studies were- carried out under a from regional banding programs, Baillie and USFWS banding permit and supportedby the National Green (1987) found estimates of adult survival ScienceFoundation. in the Common Swift (Apus upus) to be nearly LITERATURE CITED identical. The results were similar even though the sampling of breeding birds and the age-spe- ANDERSON, D. R., K. P. Bw, ANDG. C. WHITE. 1985. Problems in estimating age-specificsurviv- cific changes in breeding biology of Common al from recovery data of birds ringed as young. J. Swifts differed from those in Indigo Buntings Anim. Ecol. 54:89-98. (Perrins 197 1, Payne 1989). In four additional BAILLIE, S. R., AND R. E. GREEN. 1987. The impor- species where standard errors could be deter- tance of variation in recovery rates when esti- mined, survival estimates in long-term popula- mating survival ratesfrom ringing recoveries.Acta Omithol. 23:41-60. tion studies and in regional banding recoveries BARROWCLOUGH,G. F. 1978. Sampling bias in dis- in Britain also were statistically similar (Dobson persal studies based on finite area. Bird-Banding 1990). The similar estimates of survival in these 49:333-341. species suggestthat the results of comparative BLAKE,C. H. 1969. Notes on the Indigo Bunting. Bird-Banding 40: 13l-l 39. studiesin the Indigo Buntings may apply to other BROWNIE,C., D. R. ANDERSON, K. P. BURNHAM,AND speciesas well, and we encourage comparative D. S. ROBSON. 1985. Statistical inference from studies of survival in other species to test the band recovery data-a handbook. 2nd ed. U.S. generality of the results. Fish and Wildlife Service ResourcePubl. 156. Regional banding recoveriesmay be more rep- CAREY,M. 1982. An analysis of factors governing pair-bonding period and the onset of laying in In- resentative of survival over most or all of a spe- digo Buntings.J. Field Omithol. 53:24&248. cies range. Also, the regional nature of the re- CAREY,M., AND V. NOLAN, JR. 1979. Population covery data is independent of local dispersalthat dynamics of Indigo Buntingsand the evolution of may lead to an underestimate of survival due to avian polygyny. Evolution 33:1180-l 192. CLW~TON-BROCK,T. H. [ED.]. 1988. Reproductive a finite area bias (Barrowclough 1978, Zeng and success.Univ. of Chicago Press, Chicago. Brown 1987, Payne 1990). The size of a study CONROY,M. J., ANDB. K. WILLIAMS. 1984. A general area as well as effort of the observersdetermine methodology for maximum likelihood inference whether a locally dispersing bird is likely to be from band-recovervdata. Biometrics 40:739-748. seen again. When local observations involve Doaso~, A. P. 1990. ~Survival rates and their rela- tionship to life-history traits in some common complete samples of populations on large areas British birds. D. 115-146. In D. M. Power led.1, and comprise a large number of adjacent birds, Current ornithology. Vol. 7. Plenum Press,‘New most individuals are not near the margin of the York. study area, and local dispersaldoes not seriously DOWNER,A. C. 1972. Longevity records of Indigo Buntingswintering in Jamaica. Bird-Banding 43: bias the estimates of survival. 287. The agreement between survival estimates in I)WIGHT,J., JR. 1900. The sequenceof plumagesand the regional BBL banding recoveries and in ob- moults of the passerinebirds of New York. Ann. servations of color-banded Indigo Buntings on N.Y. Acad. Sci. 13:73-360. their breeding grounds is encouraging for further Em, S. T., J. D. RISING, AND W. L. THOMPSON. 1975. A behavioral and morphological study of use of recovery data and of long-term field ob- sympatry in the Indigo and Lazuli Buntingsof the servations of marked populations. An implica- Great Plains. Wilson Bull. 87:145-177. tion is that local estimates of survival may not FISK, E. J. 1978. Site tenacity in wintering migrants be greatly distorted by dispersal, when birds are in Florida. N. Am. Bird Bander3: 145. FISK,E. J. 1979. Fall and winter birds near Home- observed for many years in a large and contin- stead,Florida. Bird-Banding 50:224-243,297-303. uous area. mm, J.B.S. 1955. The calculation of mortality rates from ringing data. Proc. XI Int. Omithol. ACKNOWLEDGMENTS Congr.(1954):454-458. JOHNSTON,D. W., ANDA. C. DOWNER. 1968. Migra- Amtrak permitted accessto land at Niles. Facilities at tory featuresof the Indigo Bunting in Jamaicaand the E. S. George Reserve were provided by the Uni- Florida. Bird-Banding 39~277-293. versity of Michigan. D. Bystrak provided data from LACK,D. 1951. Population ecologyin birds. Proc. X the BBL banding and recovery retrieval files. K. Petts, Int. Omithol. Congr. (1950):409448. 946 ROBERT B. PAYNE AND LAURA L. PAYNE

LACK, D. 1954. The natural regulation of animal PAYNE, R. B. 1990. Natal dispersal,area effects, and numbers. Clarendon, Oxford. effective population size. J. Field Omithol. 6 1. LAKHANI,K. H., AND I. NEWTON. 1983. Estimating PAYNE,R. B., AND L. L. PAYNE. 1989. Heritability age-specificbird survival rates from ring recov- estimates and behaviour observations:extra-pair eries-can it be done? J. Anim. Ecol. 52:83-91. matingsin Indigo Buntings.Anim. Behav. 38:457- LEBERMAN,R. C. 1982. Bird-banding at Powdermill, 467. 1980. Powdermill Nature Reserve Research Re- PAYNE,R. B., L. L. PAYNE,ANLI S. M. DoEHLERT.1987. port No. 4 1. CarnegieMuseum ofNatural History, Song, mate choice and the question of kin recog- Pittsburgh, PA. nition in a migratory songbird.Anim. Behav. 35: LEBERMAN, R. C., R. S. MULVIHILL,ANO D. S. WOOD. 35-47. 1985. Bird-banding at Powdermill, 1983. Pow- PAYNE,R. B., L. L. PAYNE,AND S. M. DoEHLaaT. 1988. dermill Nature Reserve ResearchReport No. 44. Biological and cultural successof song memes in CarnegieMuseum of Natural History, Pittsburgh, Indigo Buntings. Ecology 69: 104-l 17. PA. PAYNE,R. B., ANDD. F. W~SI-NEAT. 1988. A genetic LEBERMAN,R. C., ANDD. S. WOOD. 1983. Bird-band- and behavioral analysis of mate choice and song ing at Powdermilk twenty years reviewed. Pow- neighborhoodsin Indigo Buntings. Evolution 42: dermill Nature Reserve ResearchReoort No. 42. 935-947. CarnegieMuseum of Natural History, Pittsburgh, Pnaar~s, C. 1971. Age of first breeding and adult PA. survival rates in the Swift. Bird Study 18:6l-70. MEWALDT,L. R., AND J. R. KING. 1985. Breeding Pvxn, D. A., ANDJ. N. THOMPSON.1987. Statistical site faithfulness, reproductive biology, and adult analysisof survival and removal rate experiments. survivorship in an isolated population of C&sin’s Ecology 67~240-245. Finches. Condor 87:494-5 10. QUAY, W. B. 1987. Physical characteristicsand ar- MQLLER,A. P. 1983. Time of breeding, causesof rival times of Indigo Buntingsin easternMissouri. recoveryand survival of EuropeanSandwich Terns N. Am. Bird Bander 12:2-7. C3erna sandvicensid.Voaelwarte 32: 123-14 1. RAPPoLE,J. H., ANDD. W. WARNER. 1980. Ecological MoR&, B.J.T.,ANDP. M. NORTH[EDS.]. 1985. Sta- aspects of migrant bird behavior in Veracrux, tistics in ornithology. Notes in Statistics 29. Mexico, p, 353-393. In A. Keast and E. S. Morton Springer-Verlag, Berlin. [eds.], Migrant birds in the Neotropics: ecology, MULVIHILL,R. S., ANDR. C. LEEIERMAN.1984. Bird- behavior, distribution. and conservation. Smith- banding at Powdermill, 1982. Powdermill Nature sonian Institution Press, Washington, DC. Reserve Research Report No. 43. Carnegie Mu- ROHWER,S. 1986. A previously unknown plumage seum of Natural History, Pittsburgh, PA- of first-year Indigo Buntings and theories of de- NEWTON. I. bm.1. 1989. Lifetime reoroductionin birds. layed plumage m&ration.-Auk 103:28l-292. Academic Press, London. - SEBER,G.A.F. 1973. The estimation of animal abun- NICKELL,W. P. 1968. Return of northern migrants dance and related parameters. Griffin, London. to tropical winter quarters and banded birds re- STEDMAN,S., ANDA. STEDMAN. 1988. A remarkable covered in the United States. Bird-Banding 39: traveler. N. Am. Bird Bander 13:49. 107-l 16. TABER,W., AND D. W. JOHNSTON.1968. Pmserina NOLAN,V., JR.,M. CAREY,ANIJ C. F. THOMPSON.1975. cyaneu (Linnaeus), Indigo Bunting, p. 80-l 11. In Fidelity of Indigo Buntings Pusserinacyunea to 0. L. Austin. Jr. led.1. Life histories of North sites occupiedin previous breeding seasons.Emu American cardinals:grosbeaks, buntings, towhees, 74:289. finches, sparrows, and allies. Part 1. U.S. Natl. NORTH,P. M. [ED.]. 1987. Ringing recovery analyt- Mus. Bull. 237. ical methods. kcta Omitholr 23: l-l 75. _ _ VAN TYNE, J. 1932. Winter returns of Indigo Bunt- PAYNE,R. B. 1982. Ecologicalconsequences of song ings in Guatemala. Bird-Banding 3: 110. - matching: breeding successand intraspecific song ZENG, Z., AND J. R. BROWN. 1987. A method for mimicry in Indigo Buntings.Ecology 63:401-411. distinguishing dispersal from death in mark-re- PAYNE.R. B. 1989. India0 Buntina. D. 153-172. In capture studies.J. Mammal. 68:656-665. I. ‘Newton [ed.], Lifecme reproduction in birds. Academic Press, London.