Comparative demography of new world populations of thrushes (Turdu spp.).
Geographical variation in the life tables of birds has been key to the development of ideas about the evolutionary adjustment of life history variables, especially life-span, population turnover rate, and the capacity of populations to respond to perturbation (Caughley 1977, Ricklefs 1977, Murray 1985, Roff 1992, Stearns 1992). Although latitudinal gradients in offspring number of birds and other groups have been documented (e.g., Moreau 1944, Lack 1947, Cody 1966, 1971, Ricklefs 1980), geographical patterns in adult survival are poorly known (Ricklefs 1973, Curio 1989, Karr et al. 1990). Resolving latitudinal patterns of demography may help us to understand environmental determination of survival rates, evolution of reproductive effort, and life-span (Ricklefs 1977, 1993, Rose 1991, Brawn et al. 1995), density-dependent feedbacks among demographic variables (Ricklefs 1980, 1983), and such behavioral problems as delayed plumage maturation (Rohwer and Butcher 1988) and helping at the nest in tropical populations (Brown 1987), which may reflect restricted opportunities for recruitment (Brown 1974, Ricklefs 1975, Emlen 1982, Brown 1989).
In this paper, I estimate annual survival rates of adults (S) for thrushes of the genus Turdus from proportions of adults and immatures ([is less than] 1 yr old) in museum collections. Because members of the genus Turdus have similar morphology, behavior, and ecology (Snow and Snow 1963, Simms 1978, Skutch 1981), demographic differences between populations may show how ecological variables directly affect survival and fecundity. In this sense, Turdus thrushes may be thought of as a bioassay for the response of population turnover rate and other demographic variables to environmental conditions. Briefly, the results of this analysis reveal that annual survival in the tropics (60-90%), particularly at high altitude (77-90%), exceeds that at temperate latitudes (46-74%). Within the Western Hemisphere as a whole, S varies inversely with seasonal range of temperature. Additional analyses suggest that annual fecundity varies in direct proportion to annual adult mortality, so that survival of fledglings to the following breeding season ([S.sub.0]), typically [approximately equals] 20%, is independent of latitude and annual adult survival.
Proportions of adults in samples of birds have been used to estimate annual adult survival only infrequently (e.g., Miller 1931, Snow 1956, Hanson 1963, Greenberg 1980, Mewaldt and King 1985), and sampling errors and biases inherent to this method have not been explored in detail. Because the method potentially has broad application to populations for which banding studies have not been conducted, I set out the assumptions of this approach, discuss potential errors and biases, and illustrate its application by several detailed examples from populations of Turdus.
COMPARATIVE AVIAN DEMOGRAPHY
Within most groups of terrestrial birds, both clutch size and reproductive rate (defined here as number of offspring fledged per adult each year) increase with latitude from equatorial to polar regions (Moreau 1944, Lack 1947, 1954, Cody 1966, 1971, Ricklefs and Bloom 1977, Ricklefs 1980, Skutch 1985). In a nongrowing population, annual recruitment (R) balances annual adult mortality (M) (Ricklefs 1977, Sutherland et al. 1986). Recruitment is the product of prereproductive survival ([S.sub.a]) and fecundity (F). Mortality and survival (S) are related by the expression S = 1 - M. Annual survival is generally independent of age in adult birds (Lack 1954, Ricklefs 1973, Clutton-Brock 1988; but see Newton 1989, Holmes and Austad 1995a, b), and so the variables M (or S), [S.sub.a], and F characterize fundamental attributes of avian demography and life history.
Because tropical birds produce relatively few offspring each year (low F), many ecologists believe that tropical birds live longer (high S) than their temperate and boreal counterparts (Moreau 1944, Ricklefs 1973, Skutch 1985, Curio 1989). Studies on small- to medium-sized passerine birds ([is less than] 100 g body mass) have revealed annual adult survival rates (S) of 40-60% in northern temperate zones, primarily in North America and Europe (Lack 1954, Dobson 1990). Tropical passerines have been reported to exhibit annual survival rates of 60-90% (Snow 1962, Fogden 1972, Willis 1974, 1983, Greenberg and Gradwohl 1986, additional references in Karr et al. 1990). The temperate data have come primarily from recoveries of dead individuals in large regional banding programs (Farner 1955, Brownie et al. 1985, Clobert and Lebreton 1991) or from local studies of individually marked, territorial birds followed over several years (e.g., Savidge and Davis 1974, Loery et al. 1987, Saether 1989). The tropical data derive from a small number of local population studies.
Recently, Karr et al. (1990) compared survival of temperate and tropical passerine birds, which they calculated from recaptures of banded, unaged individuals (i.e., 1st-yr and adult individuals were not distinguished) obtained by mistnetting in Maryland and central Panama. Estimated survival of 25 tropical species of passerines (S = 0.56 [+ or -] 0.11, mean [+ or -] 1 SID; range 0.33-0.73) generally overlapped the range of a sample of eight temperate species of passerines studied in Maryland (S = 0.52 [+ or -] 0.09; range 0.35-0.61) and a sample of seven temperate species from other localities in eastern North America (S = 0.55 [+ or -] 0. 10; range 0.42-0.67) (Nichols et al. 1981; see Karr et al. 1990). Even though nine of the tropical species had values [is greater than] 0.64, that is, beyond the range of the temperate species in the Maryland study, the two samples nonetheless did not differ significantly (Mann-Whitney U = 124, [n.sub.1] = 8, [n.sub.2] = 25, P = 0.16). Furthermore, survival rate in the combined sample of 15 temperate passerines did not differ from that of the 25 tropical species ([F.sub.1,38] = 0.6, P =1 0.45; Wilcoxon Z = 0.6, P = 0.56). Thus, the result of Karr et al. challenges the long-held notion of higher survival in tropical species.
For six species of antbird (Formicariidae) included in the sample of Karr et al. (S = 0.63 [+ or -] 0.08), independent estimates of survival obtained from population studies of banded, adult birds in the same region (Barro Colorado Island, Panama) have yielded values that are statistically indistinguishable from those of Karr et al. (S = 0.67 [+ or -] 0.10; Willis 1974, 1983, Greenberg and Gradwohl 1986). The major discrepancy between tropical data from Karr et al. and other local population studies within the tropics pertains to manakins (Pipridae), for which Karr et al. estimated S to be 0.47, 0.5 1, and 0.72 for three species, based on recoveries of both males and females and including prereproductive age classes. Disappearances of adult males from communal display grounds (leks) over long periods suggest that S = 0.82-0.89 for three different species in Costa Rica and Trinidad (Snow 1962, Snow and Lill 1974, McDonald 1989). Although these estimates differ from those of Karr et al. (1990), the samples are not comparable with respect to sex and age class.
If S did not differ between tropical and temperate species, population biologists either would have to revise upward the accepted estimates of annual reproductive success (e.g., Oniki 1979) or they would have to consider the possibility that lower reproductive rates are offset by greater prereproductive survival of young birds (S.) in tropical areas than in temperate areas (Karr et al. 1990). This would be difficult to reconcile with the case in which adults had similar survival in the two regions. Furthermore, observations on temperate species indicate that 1st-yr survival is generally correlated with adult survival (Ricklefs 1973, Saether 1989).
High survival in tropical passerines has been claimed primarily for suboscine taxa, particularly the manakins (Pipridae) and antbirds (Formicariidae). Because these taxa do not inhabit temperate regions, taxonomically and ecologically controlled comparisons cannot be made. Ideally, to study the influence of environment on demography, including tropical-temperate comparisons, one should use unbiased analytical techniques to estimate survival within a single, widely distributed taxon. In an attempt to approach this goal, this study assesses geographical (including latitudinal) variation in annual adult survival (proportion of adult individuals alive after a period of 1 yr) of large thrushes of the genus Turdus in the Western Hemisphere. I estimated S from proportions of 1st-yr and adult birds in samples of populations in museum collections (Miller 1931, Snow 1956, Hanson 1963, Greenberg 1980, Hilden 1982). In some species of birds, subtle plumage characters, which cannot be observed at a distance in the field, distinguish birds in their 1st yr from older individuals. As shown in this study, the age distribution of specimens in museum collections can provide robust estimates of annual adult survival that may be used reliably to characterize patterns of variation in demographic parameters among populations.
The present study analyzes 8653 specimens from 30 populations of 19 species of Western Hemisphere Turdus (Muscicapidae: Turdinae) from Alaska (T. migratorius) to southern Patagonia (T. falcklandii). Specimens were examined during 1971 and 1972 in the following museums: Academy of Natural Sciences, Philadelphia, Pennsylvania; American Museum of Natural History, New York, New York; Smithsonian Institution, Washington, D.C.; Carnegie Museum, Pittsburgh, Pennsylvania; Peabody Museum, New Haven, Connecticut; Museum of Comparative Zoology, Cambridge, Massachusetts; University of Michigan Museum of Zoology, Ann Arbor, Michigan; Los Angeles County Museum, Los Angeles, California; Dickey Collection, University of California, Los Angeles, California; San Diego Museum of Natural History, San Diego, California.
In the genus Turdus, the secondary coverts on the wings of 1st-yr birds are edged or tipped with buffy coloring (Snow 1969, Simms 1978). These feathers appear at the postjuvenal molt and are replaced at the first postnuptual molt, at [approximately equals] 1 yr old (Fig. 1). Feather wear may make it difficult to distinguish age in museum specimens collected at certain times of the year. Even among freshly molted birds, however, the plumage difference between immature and adult birds usually is unnoticeable in the field and, thus, should not produce collecting bias.
[Figure 1 ILLUSTRATION OMITTED]
Immature and adult specimens were tabulated separately by sex and month, which were taken from specimen labels (tags). In addition, notes were tabulated by month on breeding (egg sets, brood patches, size of reproductive organs, large follicles, or oviducal eggs noted on tags), numbers of individuals in juvenal plumage, individuals molting from juvenal to immature plumage, individuals with worn plumage, particularly worn secondary coverts, immatures and adults in molt, and adults in fresh plumage. These data were used in combination with some published accounts to establish the approximate periods of breeding and molting, to which monthly proportions of adults in collections could be compared.
ESTIMATION OF ANNUAL ADULT SURVIVAL
When populations are constant from year to year and 1st yr and older individuals are collected in proportion to their abundance in the population, annual adult survival may be estimated by E(S) = A/(A + I), where 1 is the number of 1st-yr birds in the sample and A is the number of older individuals. This estimate of S is unbiased when, at the time of sampling, collection probabilities and subsequent survival rates of adult and immature birds are identical; adult survival rates are achieved at an age of [approximately equals] 6 mo in many temperate-zone songbirds (Lack 1946), but no comparable information exists for tropical populations. Estimates of S may be biased when the population is growing at a constant rate, population size fluctuates over time, plumage maturation is delayed, breeding is aseasonal, or collecting is biased with respect to either age or sex. These sources of bias are evaluated and shown to be relatively small compared to differences between some populations.
A model with discrete age classes
In a discrete-reproduction model of an avian population, the breeding period is represented as a point in the annual calendar at which time all offspring are produced. Individuals mature and reproduce on their first birthday, and on each succeeding birthday thoughout their lives. When the probability of survival between breeding seasons (annualized value, S) is independent of age from birth onwards and population size is stable (geometric growth rate [Lambda] = 1), the number of offspring surviving to maturity per parent (R) equals the annual probability of death (1 - S); furthermore, the proportion of adults (age [is greater than or equal to] 1 yr) in the population not only remains constant throughout the year but also equals the annual adult survival probability. In such a population, the proportion of adults in samples of individuals obtained at any time during the year estimates the annual probability of survival without bias.
It is well known that young birds survive less well than do older individuals (Ricklefs 1973, Saether 1989). As a result, young age classes are overrepresented in a population, compared to the relative size of the same cohort later on. Accordingly, the proportion of adults in samples of individuals obtained when young survive less well, on average, than older individuals underestimates annual adult survival. This can be illustrated by the example of a simple life table, shown in Table 1, which represents a balanced population with an annual adult survival probability of 0.55. Three age classes of individuals are distinguishable by plumage: juveniles (0-1 mo old), immatures (1-13 mo), and adults ([is greater than] 13 mo). Molt (plumage change) takes place at the end of month 1. Breeding occurs in a discrete episode at time 0 during the annual cycle, at which point all individuals produce 2.5 offspring. The annualized exponential mortality rates (m, where the annual survival probability S = [e.sup.-m]) are 0.6 for adults, 7.0 during the 1st mo of life juveniles), 2.0 for 1st-yr birds during months 2-6 of life, and 0.6 (the adult rate) for 1st-yr birds during months 6-13. Juveniles molt to immature plumage at age 1 mo; immatures molt to adult plumage at age 13 mo.
TABLE 1. Life table for a hypothetical population with a discrete annual breeding event and increasing annualized probability of survival during the 1st yr of life. The annualized mortality rate is in parentheses.
Month Population status Adults 1st yr 0 Initial population 55 000 45 000 - deaths - 2673(0.60) - 2187(0.60) 1 Premolt population 52 327 42 813 molting out of stage 0 -42 813 molting into stage 42 813 139,619 Postmolt population 94 140 139,619 - deaths - 20 978(0.60) -78 941(2.00) 6 Midwinter population 74 162 60 678 - deaths - 19 162(0.60) -15 678(0.60) 12 Breeding population 55 000 45 000 Month Total Offspring 0 100,000 250,000 4860 -110,381(6.99) 1 95 140 139,619 -139,619 0 233,759 0 6 134,840 12 100,000
In this example, the proportion of adult birds in the postmolt population at the end of the 1st mo of the annual cycle is 0.405, which underestimates S by 0. 145. However, at the end of the 6th mo of the annual cycle, when immature birds are assumed to attain adult survival rates (Lack 1946), the proportion of adult birds has increased to 0.55. It remains at that level for the rest of the annual cycle, that is, until the molt at the end of month 1. In this example, any sample of adults and immatures from months 6-13 of the annual cycle would provide an unbiased estimate of annual adult survival. Samples including individuals tabulated during months 2-6 underestimate S. If the population were sampled monthly (postmolt at month 1) in proportion to the total number of individuals in the population at any one time, the overall proportion of adults in the sample would be 0.515.
The proportion of adults in a population provides an estimate of annual survival probability, but this proportion is based on sampling and has an associated error. Sampling consists of drawing individuals from a population at random, with the probability of drawing an adult being equal to the proportion of adults in the population (p). The number of adults in a sample of size n has a binomial distribution with mean np and variance np(1 - p). Setting the number of adults in the sample (np) equal to A and the number of immatures [n(1 - p)] equal to I, the proportion of adults in the sample is estimated without bias by A/(A + 1), which has a standard error of [[AI/[(A + I).sup.3]].sup.1/2]. Thus, for a total sample (A + I) of 100 individuals and proportions of adults of 0.5, 0.65, and 0.80, the standard errors of the estimates of p are 0.050, 0.048, and 0.020. For A + I 400, the standard errors of p are half of those values.
To verify the expression given for the standard error of S, sampling distributions of p were obtained by simulation by drawing different numbers of individuals from infinitely large populations containing proportion p adults. Samples were 50, 100, 200, and 400; values of p were 0.50, 0.65, 0.80, and 0.90. These span the ranges of both sample sizes and estimated annual survival probabilities in this study. The resulting estimates of p were within 0.003 of the population parameter and standard errors were within 0.002 of the predicted values.
Life table formalization
Any simple computational model of a population may be represented by a life table, which includes the survival probabilities ([S.sub.x]) and fecundity rates ([b.sub.x]) of individuals at each age x (Table 2). Survivorship ([l.sub.x]), which is the probability of survival of an individual to age x, is the product of [S.sub.i] from [S.sub.0] to [S.sub.x-1]. In this discussion, following common practice for models of bird populations, age intervals are 1 yr and sexes are not distinguished. Life table analysis is greatly simplified when one assumes constant fecundity (B) and survival probabilities (S) for adults ([is greater than or equal to] 1 yr old). In this case, the characteristic equation relating population growth rate ([Lambda]) to the life table
[TABULAR DATA 2 NOT REPRODUCIBLE IN ASCII]
[MATHEMATICAL EXPRESSION NOT REPRODUCIBLE IN ASCII].
can be expressed as
[MATHEMATICAL EXPRESSION NOT REPRODUCIBLE IN ASCII].
This equation can be solved for [lambda] = S + [S.sub.0]B. That is, the factor by which the population grows each year is equal to the survival of adults plus recruitment of young to the breeding population, B. The relative sizes of the age classes ([w.sub.x]) in such a population ([w.sub.0] = 1) depend only on the survival probabilities in the life table, and may be calculated by [w.sub.x] = [l.sub.x] [[Lambda].sup.-x]. The proportion of adults in this population just prior to the breeding episode is the ratio [w.sub.2+]/([w.sub.1] + [w.sub.2+]), where [w.sub.2+] is the relative number of individuals aged [is greater than or equal to] 2 yr. When survival is constant after age 1 yr, [w.sub.2+] is equal to
[MATHEMATICAL EXPRESSION NOT REPRODUCIBLE IN ASCII].
[w.sub.2+] = [S.sub.0] S/[Lambda]([Lambda] - S).
If we now estimate annual survival probability by E(S) = [w.sub.2]+/([w.sub.1] + [w.sub.2+]), the result is E(S) = S/[Lambda]. Thus, when [Lambda] = 1, E(S) is an unbiased estimator. When the population is either growing or declining, E(S) is biased with respect to S by B(S) = E(S) - S = S(1 - [Lambda]/[Lambda]. Therefore, when the population increases, E(S) underestimates S, because young age classes are augmented relative to older age classes by excess of births over deaths in the population. When a population is decreasing, E(S) overestimates S for the converse reason. This bias may be quite large for rapidly increasing or decreasing populations. In most populations, however, periods of increase must be roughly balanced by periods of decrease.
Suppose that a population alternates between periods of increase and decrease, and that samples of immature and adult birds are obtained equally from both phases of population change. Further suppose that the population grows by a factor of [Lambda] = [Phi] annually when it increases, and by a factor of 1/[Phi] when it decreases. If periods of increase and decrease are of equal length, the mean bias in E(S) can be estimated as the average of the biases during periods of population increase and decrease, hence
B(S) = S [([Phi] - 1).sup.2]/2 [Phi].
For values of [Phi] = 1.2, 1.5, and 2.0, this bias is +0.017S, +0.083S, and +0.250S. This equation applies to situations in which deviation of [Lambda] from 1 is caused by variation in the annual recruitment R. When variation is caused by variation in S, the bias in E(S) is
B(S) = -(R + 1) [([Phi] - 1).sup.2]/2 [Phi].
In this case, when [Phi] is 1.2, 1.5, and 2.0, E(S) underestimates S by -0.017(R + 1), -0.083(R + 1), and -0.250(R + 1), respectively. The magnitude of the bias when R varies is greater than that when S varies, because S is always less than R + 1.
Long-term studies of local populations have revealed, in many cases, decreases in annual adult survival rates with increasing age. It is reasonable to ask whether or not age-related variation in the survival of adults biases estimates of survival rate based on proportions of adults in museum collections. The mean survival rate of adults in a population is the mean of the age-specific survival rate weighted by the frequency of adults in each age class. Thus,
[MATHEMATICAL EXPRESSION NOT REPRODUCIBLE IN ASCII].
The proportion of adults in the total population, our estimate of S, is
[MATHEMATICAL EXPRESSION NOT REPRODUCIBLE IN ASCII].
Because [l.sub.x+1] = [l.sub.x] [S.sub.x], [[Sigma].sub.x=2] [l.sub.x] = [[Sigma].sub.x=1] [l.sub.x] [S.sub.x] and S = S. Thus, regardless of changes in age-specific rate of survival, the proportion of adults in a population always estimates the mean survival rate of adults in the population.
Delayed reproduction and delayed maturation of plumage
Because the relative sizes of age groups within a population ([w.sub.i]) depend only on the age-specific survival probabilities, delayed maturation (age at first breeding) has no effect on the proportions of individuals in different age classes in the population and, thus, does not influence the estimated survival rate. Of course, delayed first reproduction might be associated with high adult survival rates through density-dependent feedbacks on maturation, but S is nonetheless estimated without bias from the proportions of adult individuals in such populations.
When the acquisition of adult plumage is delayed from the 1st to the 2nd yr, which has been shown to occur in several temperate-zone passerine birds (Rohwer et al. 1980, Procter-Gray and Holmes 1981, Lyon and Montgomerie 1986), annual survival probability is underestimated by the proportion of adults in the population, because birds in their 2nd yr are not classed as adults. Consider a population with survival values [S.sub.0], [S.sub.1] and S (for ages [is greater than or equal to] 2 yr). The relative sizes of the age classes are [w.sub.1] = [S.sub.0] [[Lambda].sup.-1] [w.sub.2] = [S.sub.0] [S.sub.1] [[Lambda].sup.-2], and [w.sub.3+] (adult plumage) = [S.sub.0] [S.sub.1] S/[[Lambda].sup.2]([Lambda] - S). In such a population, when one estimates S by the proportion of adults in the population, that is, by A/(A + I), then E(S) = [w.sub.3]/([w.sub.1] + [w.sub.2] + [w.sub.3+]), or
E(S) = [S.sub.2] S/[Lambda]/[Lambda] + [S.sub.1] - S).
Note that when [Lambda] = 1, and [S.sub.1] = S, E(S) = [S.sup.2]. The bias in the estimate of S, that is, B(S), = E(S) - S, can be expressed as
B(S) = S([S.sub.1][[Lambda].sup.-1] - ([Lambda] + [S.sub.1] - S)/[Lambda] + [S.sub.1] - S).
When [Lambda] = 1 and [S.sub.1] = S, then B(S) = S(S - 1). Thus, when plumage maturation is delayed to the 2nd yr, E(S) may underestimate S substantially, but the bias decreases with increasing S, specifically the demographic condition likely to favor delayed plumage maturation.
When the postjuvenal molt is delayed until the first breeding season, and immature molt is delayed until after the second breeding season following birth, then E(S) = A/(A + 1) = [n.sub.3+]/([n.sub.3+] + [n.sub.2]). This is equivalent to E(S) = S/[Lambda], which provides an unbiased estimate of S in a population of constant size when individuals are sampled outside the molting and breeding season, and assuming that the survival rate of individuals with immature plumage is the same as that of adults.
The proportion of adults in a population provides an unbiased estimate of S only when immatures and adults have the same probability of survival. Thus, samples obtained shortly after the end of a breeding episode, before immatures of the year have acquired adult survival probabilities, underestimate S, as shown by example (Table 1). Samples obtained shortly prior to breeding, by which time immature birds have acquired experience and their survival rates have reached adult levels, can provide unbiased estimates of S. However, this reasoning applies only to populations in which breeding is restricted to a short period during the year, and immature birds mature as a cohort of similarly aged individuals.
When breeding continues throughout the year, immatures are produced during every month and samples obtained at any time contain young birds with low survival rates. As a result, the proportion of adults in such a population will always underestimate adult survival probability. The magnitude of this effect is difficult to obtain analytically, but can be shown by a numerical example (Table 3). This example, which is based on that in Table 1, assumes that reproduction in the population occurs in discrete episodes once each month and is spread evenly throughout the year; annualized mortality rates are 7.0 for juveniles (0-1 mo), 2.0 for immatures to 6 mo, 0.6 for immatures older than 6 mo, and 0.6 for adults. As in Table 1, the immature-adult molt occurs at 13 mo. Each month, losses from each segment of the population through death or molt must be balanced by gains through molt or reproduction. The balance sheet is shown in Table 3. The steady-state numbers in the population are 55 000 individuals in adult plumage and 55 299 in immature plumage. Thus, the proportion of individuals in adult plumage is 0.499, which underestimates S by [approximately equals] 0.05. As in the case of delayed plumage maturation, aseasonal breeding is likely to be more prominent in tropical latitudes; therefore, it would tend to obscure higher survival rates in tropical populations.
TABLE 3. Balance sheet for a population in which reproduction is spread evenly throughout the year at monthly intervals, with molt at ages 1 and 13 mo. The annualized mortality rate is in parentheses.
Month Cause of change Adults 6-12 mo 0 Initial population 55 000 23 045 - death 2 682 (0.60) 1 124 (0.60) 1 Premolt population 52 318 21 921 + molting out of stage 0 2 682([dagger]) + molting into stage 2 682 3 806 + birth 0 0 1 Postmolt population 55 000 23 045 Month 1-5 mo Offspring 0 32 254 15 681 4 952 (2.00) 6 923 (6.99) 1 27 302 8 758 3 806([double dagger]) 8 758 8 758 0 0 15 681 1 32 254 15 681
([dagger]) Individuals ready to molt (13 mo old) are proportion 0.122 of the subpopulation of individuals 7-13 mo old.
([double dagger]) Individuals ready to molt (7 mo old) are proportion 0.139 of the subpopulation of individuals 2-6 mo old.
Although the plumages of immature and adult Turdus differ by a small number of subtle characters difficult to detect in the field, differences in behavior or habitat distribution may result in collecting bias. Suppose that immatures are collected with efficiency [Psi] relative to adults. Accordingly, in a population with A adults and I immatures, the estimate of adult survival probability would be
E(S) = A/A + [Psi]I'
which has a bias, E(S) - S, of
B(S) = S(1 - [Psi]) I/A + [Psi]I
B(S) = S(1 - E[S]) 1 - [Psi]/[Psi].
Thus, the bias in E(S) is positive for [Psi] [is less than] 1 and negative for [Psi] [is greater than] 1. The magnitude of the bias decreases as E(S) increases. These relationships are shown in Fig. 2 for values of S equal to 0.50 and 0.75.
[Figure 2 ILLUSTRATION OMITTED]
Age-specific sex bias
In most species of Turdus, males sing conspicuously from exposed perches (Simms 1978). Males appear more frequently than females in collections made during these periods. If this bias were to affect adults more (or less) than it affected immatures, then the total sample of the less biased age would be underrepresented for those months. Specifically, if adults showed male sex bias and immatures did not (suggesting that 1st-yr birds do not breed), then adults as a whole would be overrepresented in the sample, and E(S) would show a positive bias.
If we assume that the true sex ratio in a population is 1:1 for both adults and immatures, but adults of one sex are collected with bias [Beta] ([Beta] = 1 indicates absence of bias), then the numbers of adults in the collection (A) will be biased relative to the number adults in the population ([n.sub.2+]) by the factor 0.5(1 + [Beta]). Accordingly,
E(S) = [n.sub.2+] (1 + [Beta]) / [n.sub.2+] (1 + [Beta]) + [2n.sub.1].
Assuming that the true value of S = [n.sub.2+]/([n.sub.2+] + [n.sub.1]), bias of E(S) is
B(S) = 0.5S I([Beta] - 1) / A + I
B(S) = 0.5S(1 - E[S])([Beta] - 1).
This bias could be corrected by basing E(S) on the number of adult females in the collection multiplied by 2.
Accuracy and bias: conclusions
S can be estimated from the proportion of adults in collections with 1 SE [is less than or equal to] 0.05 in samples of [is greater than or equal to] 100 individuals. Thus, this technique is capable of detecting differences in S between populations on the order of 0.10 or less. Significance tests can be based either on analysis of 2 X 2 contingency tables (population X age) or on the differences between estimates of S divided by the standard error of the estimates. In general, the difference between two means may be accepted as being significant when it exceeds the sum of the standard errors of each mean (that is, standard errors do not overlap).
Several sources of bias potentially are important. Factors that produce bias alter the actual or apparent proportion of adults in the population and, consequently, estimates of annual adult survival (Table 4). Depending on the conventional wisdom about differences between tropical and temperate populations in sources of bias, these factors may either reinforce or weaken a pattern of increasing adult survival rates towards the equator. For example, temperate populations are thought to crash and recover more frequently and more severely than tropical populations, and this would make adult survival appear relatively lower in temperate populations. Conversely, breeding is less seasonal in tropical than in temperate populations (Ricklefs 1966, Ricklefs and Bloom 1977), which should reduce estimated survival rates in tropical populations and obscure any trend of increasing survival towards the tropics. On balance, factors that tend to reinforce or weaken such a trend are about equal in number, suggesting that estimates of survival from age ratios in museum collections may be relatively unbiased (Table 5).
[TABULAR DATA 4 NOT REPRODUCIBLE IN ASCII]
TABLE 5. Likely influence of sources of bias in estimating S on the relationship between estimated annual adult survival rate and latitude.
Effect on trend of Zone with increasing greater survival source of towards Source of bias bias the tropics Constant population growth temperate reinforces Population fluctuation due to variable recruitment temperate reduces Population fluctuation due to variable adult survival temperate reinforces Delayed molt from immature to adult plumage tropical reduces Aseasonal breeding tropical reduces Collecting bias favoring immatures over adults temperate reduces Collecting bias favoring adults of one sex tropical reinforces
Application: Samples of the collection data
Both the potential and the problems of using data from museum collections to estimate annual adult survival are readily seen by inspecting the data themselves. These are presented for four representative species of Turdus in Tables 6 through 9.
TABLE 6. Monthly distribution of numbers of individuals of different age classes and indications of breeding and molt in collections of Turdus amaurochalinus.
Status Jul Aug Sep Oct Nov Dec Breeding 3 5 4 Juveniles 1 2 Molting juveniles 1 1 Worn adult plumage 1 3 Immatures 46 25 16 17 3 5 Adults 56 46 36 32 18 10 Proportion adult 0.549 0.648 0.692 0.653 0.778 1 SE 0.049 0.057 0.064 0.068 0.069 Status Jan Feb Mar Apr May Jun Year Breeding 1 Juveniles 4 8 4 2 21 Molting juveniles 1 1 3 Worn adult plumage 6 1 Immatures 2 1 9 17 26 31 198 Adults 9 8 9 23 35 31 313 Proportion adult 0.850 0.551 0.574 0.500 0.613 1 SE 0.080 0.065 0.063 0.064 0.022
TABLE 7. Monthly distribution of numbers of individuals of different age classes and indications of breeding and molt in collections of Turdus m. migratorius from east of the Appalachian Mountains in the United States.
Status Jan Feb Mar Apr May Jun Breeding 3 5 4 Juveniles 1 14 Molting juveniles Worn adult plumage Immatures 35 34 61 47 13 21 Adults 27 33 65 90 22 43 Proportion adult 0.436 0.493 0.515 0.680 0.629 0.672 1 SE 0.063 0.061 0.045 0.039 0.082 0.059 Status Jul Aug Sep Oct Nov Dec Year Breeding 1 Juveniles 33 30 12 90 Molting juveniles 1 4 1 Worn adult plumage Immatures 14 1 13 64 42 11 356 Adults 20 6 15 37 14 8 380 Proportion adult 0.588 0.600 0.366 0.293 0.516 1 SE 0.084 0.083 0.048 0.053 0.018
TABLE 8. Monthly distribution of numbers of individuals of different age classes and indications of breeding and molt in collections of Turdus leucomelas.
Status Jan Feb Mar Apr May Jun Jul Aug Breeding 3 2 3 1 1 1 Juveniles 3 3 5 1 1 2 5 Molting juveniles 2 1 2 3 1 Worn adult plumage 2 2 6 7 2 2 1 Immatures 9 18 11 14 17 17 12 5 Adults 16 34 26 38 33 19 30 22 Proportion adult 0.649 0.719 0.605 0.754 1 SE 0.054 0.048 0.053 0.052 Status Sep Oct Nov Dec Year Breeding 1 1 Juveniles 1 3 3 5 32 Molting juveniles Worn adult plumage Immatures 20 10 12 12 157 Adults 16 25 20 15 294 Proportion adult 0.578 0.593 0.652 1 SE 0.059 0.064 0.022
TABLE 9. Monthly distribution of numbers of individuals of different age classes and indications of breeding and molt in collections of Turdus ignobilis.
Status Jan Feb Mar Apr May Jun Jul Aug Breeding 3 1 1 2 1 Juveniles 2 1 2 1 Molting juveniles 1 1 1 1 Worn adult plumage 1 3 9 1 3 1 Immatures 4 5 5 3 2 1 12 7 Adults 34 34 28 32 24 25 17 10 Proportion adult 0.883 0.882 0.942 0.587 1 SE 0.037 0.039 0.032 0.073 Status Sep Oct Nov Dec Year Breeding 1 1 3 2 Juveniles 1 3 5 15 Molting juveniles Worn adult plumage 1 Immatures 2 6 10 2 59 Adults 18 33 40 19 314 Proportion adult 0.864 0.831 0.842 1 SE 0.045 0.045 0.019
Turdus amaurochalinus is a nonmigratory, temperate South American species. Museum data indicate highly seasonal breeding during October through January (the Southern Hemisphere summer) (Table 6). Juveniles appear in collections a month after the appearance of egg sets and other signs of breeding (enlarged ovaries, eggs in oviduct, brood patches) that are often noted on specimen labels. Juveniles molting into immature plumage do not appear in collections until 4 mo after the first appearance of juveniles, suggesting that juveniles retain their plumage for [approximately equals] 4 mo. Adult and immature birds with heavily worn plumage occur during the breeding season, but not after, suggesting that molt occurs shortly after the end of the breeding season, perhaps in January and February. Collecting intensity, success, or both apparently vary over the course of the year, because relatively few specimens have been taken during the period December through March. Collectors tend to avoid collecting during the late breeding season and just after the breeding season to avoid specimens with badly worn plumage. Estimates of S based on monthly or bimonthly tabulations vary on a seasonal basis in a predictable fashion. During the 3 mo prior to and during the early part of the breeding season (August through October), E(S) varies between 0.65 and 0.69. During November through February, E(S) increases to [approximately equals] 0.80, as the plumage of some immatures becomes so worn that they cannot be distinguished from adults while other immatures molt into adult plumage. From March through July, E(S) drops considerably, to 0.50-0.57, as large numbers of juveniles molt into immature plumage. The estimate of S least likely to suffer bias is that from the period just before and at the beginning of the breeding season, after immatures have achieved adult survival rates and before plumage wear and molt become complications. For the period August-October, E(S) = 0.663 [+ or -] 0.036 (mean [+ or -] 1 sE). Because biases during other seasons of the year nearly cancel each other, E(S) for the entire sample (0.613 [+ or -] 0.022) does not differ substantially.
Turdus migratorius is also a temperate species (North America) with a breeding season restricted to May through July in the eastern United States and Canada (Howell 1942, Young 1955, Knupp et al. 1977). In collections from this region, juveniles occur in June through September and apparently molt at an age of [approximately equals] 2 mo (Table 7). Worn plumage appears in June and July, suggesting that molt occurs in August after breeding is finished. As in T. amaurochalinus, collecting is seasonal, reaching an ebb during the breeding season and late summer, when plumage is worn or in molt, and during the winter, when most collectors would rather be indoors. E(S) exhibits a similar seasonal trend to that in T. amaurochalinus, in that E(S) increases during the breeding season and then decreases to a low value following molt of the juvenal plumage. Feather wear in T. migratorius is such that immature plumages are difficult to distinguish by April. Thus, the months of February and March probably provide the least biased estimate of S (0.508 [+ or -] 0.036). As in T. amaurochalinus, the estimate of S obtained from the entire sample (0.516 [+ or -] 0.018) does not differ significantly from the more restricted estimate obtained prior to the breeding season.
Turdus leucomelas is a lowland, tropical Central American species with somewhat seasonal breeding, apparently concentrated during the dry season, January through April, but with sporadic breeding recorded throughout the year (Skutch 1981; Table 8). Estimation of its demography from museum collections may be complicated because juvenile molt appears to be delayed until the following breeding season. This suggests, additionally, that immatures do not molt into adult plumage until they are up to 2 yr old. Based on the seasonal occurrence of worn plumage, immatures and adults molt primarily in June-August, after the peak of breeding. Presumably, immatures have achieved adult survival rates by this time, so estimates of S obtained after the molt and prior to the breeding season (that is, September through December) should be unbiased. For this period, E(S) = 0.585 [+ or -] 0.044; for the entire year, E(S) = 0.652 [+ or -] 0.022.
The last example, that of Turdus ignobilis, was chosen to illustrate a tropical species with little seasonality (Miller 1963). Indications of breeding are recorded during nine months (Table 9). Juveniles also occur in collections throughout the year and, although the sample is small, juvenile molting also is sporadically distributed. Worn plumage occurs in many months, but appears to be concentrated during March through June. This suggests that molt may occur during a restricted period in the middle of the calendar year, perhaps coinciding with a lull in breeding suggested by the absence of breeding indicators in June and July. Birds in immature plumage are scarce in collections. The proportion of adults in bimonthly samples varies between 0.83 and 0.94, except during July/August (0.59), that is, just after the presumed molting period. Based on samples collected during the other 10 months, E(S) = 0.878 [+ or -] 0.018. If immatures have lower survival rates, on average, than adults, this value would underestimate S.
The sex ratio of adults of Turdus migratorius from eastern North America during March through June was 144 males to 77 females, or 1.87 males per female ([chi square] = 20.3, P [is less than] 0.001). During the remainder of the year, numbers were 87 males and 82 females, or a ratio of 1.06 ([chi square] = 0.15, P [is greater than] 0.2). Immature birds exhibited the same sex bias in collections: during March through June, 4 males to 44 females, or a ratio of 2.23 ([chi square] = 20.5, P [is less than] 0.00 1); during the remainder of year, 110 males to 100 females, or a ratio of 1.10 ([chi square] = 0.48, P [is greater than] 0.3). Sex ratios of immatures and adults did not differ either during the period of sex-biased collecting ([chi square] = 0.12, P [is greater than] 0.5) or during the remainder of the year. Thus, although there is a strong bias towards collecting males during certain months of the year, this bias affects adults and immatures equally and, therefore, should not cause a bias in E(S).
In Turdus amaurochalinus, adult males were collected with considerable bias during the months of August through November, including 2 mo prior to, and 2 mo after, the beginning of breeding in October. During this period, there were 89 adult males and 43 females, for a ratio of 2.07 males per female ([chi square] = 16, P [is less than] 0.001). During the remainder of the year, the ratio was 99 males to 82 females, or 1.21 males per female ([chi square] = 1.60, P = 0.2). During the period August through November, in contrast, immatures in the sample numbered 32 males and 29 females, for a ratio of 1. 10 ([chi square] = 0.14, P = 0.3). The distribution of adults and immatures between the sexes differed significantly ([chi square] = 3.99, P [is less than] 0.05). During the remainder of the year, immatures numbered 76 males and 61 females, with a ratio of 1.25 ([chi square] = 1.64, P = 0.2), which did not differ from that of adults during the same period.
In both species of Turdus, there appears to be a slight collecting bias in favor of males throughout the year, but this becomes highly significant during the prebreeding and breeding periods. Because the bias is similar in adults and immatures of T. migratorius, it should not cause a bias in E(S). In T. amaurochalinus, however, only adult males are collected with bias, which should increase the proportion of adults in the collection and give E(S) a positive bias. Earlier, S was estimated from individuals collected during August-October. These numbered 77 adult males (AM), 37 adult females (AF), 29 immature males (IM), and 29 immature females (IF). The proportion of adults was 114 out of 172, or 0.663. Because immatures appeared to be collected without sex bias, adult females may provide a better index to the relative size of the adult population than males and females together. Assuming a 1:1 sex ratio, and setting A = 2AF, one obtains 74 adults and 58 immatures, and E(S) = 0.561, suggesting that the sex X age interaction in collecting bias created a bias in E(S) of [approximately equals] 0.10 in this species. In this particular case, the estimate of S based on the entire collection (0.61) was better than that obtained from the prebreeding sample without correction for sex bias.
COMPARATIVE DEMOGRAPHY OF TURDUS
Only samples with [is greater than or equal to] 100 individuals of adults and immatures combined were included in the comparative analyses that follow. An additional 11 samples failed to meet this criterion. These are listed with details, including estimates of S, in Table 10. Of the 30 samples with [is greater than or equal to] 100 individuals (see Table 11) several represent regional populations or subspecies of the same species.
TABLE 10. Demographic characteristics of Turdus populations represented by <100 museum specimens. The total sample of adults and immatures (A + I) is n; E(S) is the estimate of annual adult survival, with 1 SE; J(total) is the total no. juveniles in the museum samples divided by A; and E([S.sub.a]) is the relative rate of recruitment per juvenile, i.e., (1 - E(S))/J(total).
Turdus species Region([dagger]) n E(S) 1 SE ardiosacea (plumbeus) trop. 57 0.667 0.062 aurantius trop. 28 1.000 0.000 flavipes trop. 12 0.583 0.142 fulviventris Andean 53 0.811 0.054 hauxwelli trop. 87 0.816 0.042 jamaicensis trop. 23 0.652 0.099 lawrencii trop. 11 0.818 0.116 migratorius permixtus temp. 50 0.440 0.070 maranonicus trop. 33 0.788 0.071 nigriceps Andean 83 0.723 0.049 reevei trop. 45 0.933 0.037 Turdus species J(total) E([S.sub.a])([double dagger]) ardiosacea (plumbeus) 0.105 3.2 aurantius 0.000 flavipes 0.000 fulviventris 0.186 1.0 hauxwelli 0.042 4.4 jamaicensis 0.000 lawrencii 0.000 migratorius permixtus 0.091 6.2 maranonicus 0.000 nigriceps 0.017 16.6 reevei 0.167 0.4
([dagger]) Regions: temp., temperate North America; mont. CA, mid- and high-elevation montane Central America; trop., lowland tropical Central and South America; Andean, mid- and high-elevation montane South America; subtrop., subtropical zones of eastern South America.
([double dagger]) Blank cells indicate that the estimate of [S.sub.a] is infinite because no juveniles ([J.sub.total]) were recorded from collections.
TABLE 11. Demographic characteristics of Turdus populations represented by [is greater than or equal to] 100 museum specimens. Parameters are as in Table 10.
Turdus species Region([dagger]) n E(S) 1 SE albicollis assimilis mont. CA 131 0.786 0.036 albicollis leucauchen trop. 315 0,765 0.024 albicollis daguae trop. 372 0.715 0.023 albicollis albicollis subtrop. 250 0.736 0.028 amaurochalinus subtrop. 511 0.613 0.022 chiguanco Andean 213 0.784 0.028 falcklandii Andean 208 0.827 0.026 Jumigatus trop. 215 0.595 0.033 fuscater gigas Andean 116 0.905 0.027 grayi casius trop. 287 0.777 0.025 grayi grayi trop. 440 0.764 0.020 ignobilis trop. 373 0.842 0.019 leucomelas trop. 451 0.652 0.022 migratorius achrusterus temp. 294 0.493 0.029 migratorius caurinus temp. 321 0.676 0.026 migratorius confinus temp. 249 0.743 0.028 m. migratorius central temp. 300 0.460 0.029 m. migratorius east temp. 736 0.516 0.018 m. migratorius north temp. 136 0.507 0.043 migratorius nigrideus temp. 198 0.510 0.036 migratorius propinquus temp. 768 0.538 0.018 nigrescens mont. CA 115 0.774 0.039 nudigenis trop. 213 0.836 0.025 olivator Andean 139 0.856 0.030 plebejus mont. CA 174 0.776 0.032 plumbeus trop. 218 0.899 0.020 rufitorques mont. CA 150 0.840 0.030 rufiventris subtrop. 422 0.680 0.023 rufopalliatus mont. CA 121 0.802 0.036 serranus Andean 222 0.856 0.024 Turdus species J(total) E([S.sub.a]) albicollis assimilis 0.039 5.5 albicollis leucauchen 0.095 2.5 albicollis daguae 0.109 2.6 albicollis albicollis 0.043 6.1 amaurochalinus 0.067 5.8 chiguanco 0.180 1.2 falcklandii 0.134 1.3 Jumigatus 0.109 3.7 fuscater gigas 0.114 0.8 grayi casius 0.063 3.6 grayi grayi 0.068 3.5 ignobilis 0.048 3.3 leucomelas 0.109 3.2 migratorius achrusterus 0.000 migratorius caurinus 0.092 3.5 migratorius confinus 0.000 m. migratorius central 0.123 4.4 m. migratorius east 0.237 2.0 m. migratorius north 0.000 migratorius nigrideus 0.089 5.5 migratorius propinquus 0.153 3.0 nigrescens 0.045 5.0 nudigenis 0.028 5.8 olivator 0.084 1.7 plebejus 0.022 10.1 plumbeus 0.036 2.8 rufitorques 0.071 2.2 rufiventris 0.049 6.6 rufopalliatus 0.062 3.2 serranus 0.132 1.1
([dagger]) Regions: see Table 10 footnote.
Bias in E(S) is minimized when S is estimated from samples obtained immediately prior to the beginning of the breeding season, by which time all juveniles will have molted into immature plumage, and the survival rate of immatures approaches that of adults. Because breeding seasons could not be determined, or because samples immediately prior to breeding were small, S could not be estimated from prebreeding samples for several of the populations considered in Table 11. Of the 30 populations surveyed in this study, only 14 had sufficient specimens collected during the 1-2 mo prior to the breeding season to calculate E(S). However, estimates of S from the total samples of these populations collected throughout the year were virtually identical to those obtained from the prebreeding samples ([E[S].sub.pre] = -0.021 [1 SE = 0.070, P [is greater than] 0.05] + 1.046 [1 SE = 0. 1141 [E[S].sub.Total]; [F.sub.1,12] = 84, P [is less than] 0.0001, [R.sup.2] = 0.88; H,: slope b = 1, P [is greater than] 0.05). Therefore, [E(S).sub.Total] is used to estimate S in this study for all 30 populations.
An estimate of annual adult survival for Turdus m. migratorius in eastern North America, based on numbers of 1st-yr and older birds in collections (E[S] = 0.516 [+ or -] 0.015, mean [+ or -] 1 SE; A + I = 1157) compared favorably with survival established from recoveries of banded birds (E[S] = 0.494 [+ or -] 0.005; Henny 1972), using composite-dynamic life table method (see Anderson et al. 1981), and did not differ significantly from a higher estimate of S based on the survival of adults in a local Pennsylvania population (E[S] = 0.67 [+ or -] 0.086, n = 30; Savidge and Davis 1974). On an isolated mountain in Oregon, annual survival in a local population of T. migratorius was determined to be 0.70 (1 SE 0.03, n = 80) for males and 0.83 (1 SE = 0.02, n 72) for females (King and Mewaldt 1987). It is, however, difficult to compare estimates of S from local studies with those from the broader geographical representation of museum collections. The annual mortality (1 - S) of the European Blackbird (Turdus merula), determined from banding studies in Britain (0.37 [+ or -] 0.04, mean [+ or -] 1 SD over 7 years), was closely matched by estimates of the proportions of 1st-yr birds in March populations (0.37 [+ or -] 0.06, mean [+ or -] 1 SE) during those same years (Snow 1966).
E(S) differed significantly among populations grouped according to geographical locality, being highest in tropical montane populations and lowest in north-temperate populations (Fig. 3 and Table 12). Although E(S) varied considerably among eight north-temperate populations and among nine lowland tropical populations, estimates of S differed substantially between them (E[S] = 0.56 and 0.76, respectively; Mann-Whitney U = 5, P = 0.001). Ten montane populations in Central America and the Andes of South America exhibited minimum values of E(S) of 0.77, and mean values of 0.80 and 0.85, respectively. South-temperate populations also had high values of E(S), exceeding all the north-temperate samples except for two that occur along the moderate Pacific coast (T. m. caurinus, northwestern USA, southwestern Canada; T. m. confinus, southern, montane Baja California). Thus, survival rate of Turdus shows a strong geographical pattern of variation. In particular, tropical populations appear to have higher survival rates than those of north-temperate regions. Possible biases resulting from delayed plumage maturation and a seasonal breeding in tropical species would tend to reduce, rather than augment, survival rates of tropical species. Thus, the latitudinal pattern would appear to be robust. Additional potential sources of bias will be considered.
TABLE 12. Estimated annual survival rates for populations of Turdus grouped according to geographical locality.
Locality n([dagger]) Mean Temperate North America 8 0.555 Subtropical South America 3 0.676 Lowland tropical 9 0.761 Montane Central America 5 0.796 Montane Andean 5 0.846 SNKT([double Range dagger]) Temperate North America a 0.46-0.74([sections]) Subtropical South America b 0.61-0.74 Lowland tropical bc 0.60-0.90 Montane Central America bc 0.77-0.84 Montane Andean c 0.78-0.90
([dagger]) No. species in sample. Temperate North America: migratorius (6 subspecies, 8 populations); subtropical South America: albicollis, amaurochalinus, rufiventris; lowland tropical: albicollis (2), fumigatus, grayi (2), ignobilis, leucomelas, nudigenis, plumbeus; montane Central America: albicollis, nigriceps, plebejus, rufitorques, rufopalliatus; Andean: chiguanco, falcklandii, fuscator, olivater, serranus.
([double dagger]) Student-Newman-Keuls multiple range test; samples with the same letter do not differ significantly (P < 0.05). ANOVA: [F.sub.4,15] = 13, P < 0.0001, [R.sup.2] = 0.68; Kruskal-Wallis [chi square] = 19.8, P = 0.0005.
([sections]) The highest value for temperate North America represents T. migratorius confinus from the mountains of Baja California, near the Tropic of Cancer. The next highest value is 0.68 for T. m. caurinus from the Pacific coast of the northwestern USA, British Columbia, and Alaska.
[Figure 3 ILLUSTRATION OMITTED]
Survival and climate
The geographical pattern in survival rate suggests that survival might be inversely related to the seasonality of the environment. To determine whether variation in survival is associated with variation in conditions of the environment, values of E(S) were correlated with several climate measurements. Minimum and maximum monthly mean temperatures, mean annual temperature, minimum and maximum monthly precipitation, and annual precipitation (Wernstedt 1972) were tabulated for one locality close to the center of the geographical and elevational range of the breeding distribution of each population (Table 13). For 28 populations for which such data were available, E(S) was significantly correlated with minimum mean monthly temperature (r = 0.69, P [is less than] 0.0001) and the difference between maximum and minimum mean monthly temperatures (r = -0.84, P [is less than] 0.0001) (Fig. 4), indicating that seasonal range of temperature is the most important correlate of annual adult survival among the climate variables considered. Simple correlations of survival with other variables, including latitude, altitude, maximum mean monthly and annual temperature, maximum mean monthly, minimum mean monthly, and annual precipitation, were between r =-0.4 and r = 0.4. A stepwise regression of survival rate on the climate variables revealed that only the annual mean monthly temperature range made a strong, unique contribution to survival rate ([F.sub.1,25] = 70.7, P = 0.0001, b = -0.0148 [1 SE = 0.00181), with a weak contribution from total annual precipitation ([F.sub.1,25] = 4.7, P = 0.04, b = -0.00058 [1 SE = 0.00026]; [R.sup.2] = 0.75); no other variables were significant.
[TABULAR DATA 13 NOT REPRODUCIBLE IN ASCII]
[Figure 4 ILLUSTRATION OMITTED]
Of the north-temperate populations for which climate data were available, the highest survival was recorded for the subspecies T. m. caurinus (E[S] = 0.68) from coastal northwestern United States, British Columbia, and southern Alaska, having a maritime climate. Although survival rates for three European species of Turdus were not included in the regressions, their positions in Fig. 4 indicate higher survival in England, having a maritime (less seasonal) climate, than in the more continental climate of eastern Europe. Relatively high survival rates of south-temperate species also reflect a less seasonal climate. Indeed, survival of populations located outside the tropics (at latitudes higher than 23 [degrees] N and 23 [degrees] S) is related uniquely only to annual mean monthly temperature range ([F.sub.1,12] = 49, P [is less than] 0.0001, [R.sup.2] = 0.80, b = -0.0156 [1 SE = 0.0022]).
Within the tropics, seasonal temperature range is uniformly small and explains little of the variation in annual survival. The lowest annual survival rates for tropical species in this study were for T. fumigatus (Amazon Basin, E[S] = 0.595) and T. leucomelas (tropical and subtropical South America, E[S] = 0.652).
Sex ratio and age at first breeding
As we have seen in the detailed examples of Turdus populations presented, males are more frequent in collections during certain times of the year than are females, presumably due to increased conspicuousness of males when they are singing. If 1st-yr birds show a similar male bias at the beginning of the breeding season, we might infer that the behavior of 1st-yr males is similar to that of older adults, and that males enter the breeding population at the end of their 1st yr. This appears to be the case for the North American T. migratorius. If, however, 1st-yr males are not collected with bias, it seems likely that reproduction may be delayed until the 2nd yr.
Many of the samples included in this study were too small to be analyzed, with statistical confidence, for breeding-season sex ratios. For populations with larger samples, sex ratios are analyzed for the entire sample collected throughout the year (Table 14), because in many tropical species, the breeding season, including the period of male singing, continues through much of the year. The annualized sex ratio for adults and immatures should bear a close relationship to the sex ratio during the period of territorial singing, and this appears to be the case. For example, in T. m. migratorius from the eastern United States, the annualized proportion of males among adults was 0.59, and among immatures it was 0.66. Both values differed significantly from an even sex ratio (0.50), but did not differ from each other, consistent with the more detailed results described. In the case of T. amaurochalinus, collections of adults had a significant excess of males (0.60) whereas those of immatures (0.55) did not. Proportions of males did not, however, differ significantly between adults and immatures, even though they did in samples collected during the breeding season (0.67 vs. 0.52). Thus, it appears that the annualized sex ratios show the same qualitative patterns as the breeding-season sex ratios, but may have less statistical power to resolve differences between samples.
TABLE 14. Proportions of males among immature and adult specimens of Turdus species. The [chi square] values for immatures and adults test the numbers of each sex against an equal distribution (sex ratio 1:1). The pooled [chi square] values test the homogeneity of numbers of males and females among immatures and adults.
Turdus species Immatures [chi square] Adults albicollis assimilis 0.79 9.1 0.61 albicollis leucauchen 0.68 9.1 0.62 albicollis daguae 0.53 0.3 0.58 albicollis albicollis 0.61 3.0 0.68 amaurochalinus 0.55 1.6 0.60 chiguanco 0.54 0.3 0.63 falcklandii 0.69 5.4(*) 0.54 fumigatus 0.59 2.6 0.50 fuscater gigas 0.45 0.1 0.57 grayi casius 0.66 6.3 0.56 grayi grayi 0.67 12.5(***) 0.64 ignobilis 0.49 0.0 0.54 leucomelas 0.64 11.8(***) 0.54 migratorius achrusterus 0.40 5.6(*) 0.48 inigratorius caurinus 0.64 8.7(**) 0.72 migratorius confinus 0.56 1.0 0.59 m. migratorius central 0.60 6.3 0.67 m. migratorius east 0.66 32.1 0.59 m. migratorius north 0.73 14.3(***) 0.67 migratorius nigrideus 0.68 12.6(***) 0.69 migratorius propinquus 0.67 44.0(***) 0.50 nigrescens 0.61 1.4 0.66 nudigenis 0.34 3.5 0.67 olivator 0.75 5.0(*) 0.66 plebejus 0.72 7.4(**) 0.56 plumbetis 0.41 0.7 0.66 rufitorques 0.50 0.0 0.71 rufiventris 0.69 19.3(***) 0.58 rufopalliatus 0.63 1.5 0.59 serranus 0.59 1.1 0.56 Pooled Turdus species [chi square] [chi square] albicollis assimilis 5. 1(*) 2.9 albicollis leucauchen 14.4(***) 0.7 albicollis daguae 7.3(**) 0.9 albicollis albicollis 23.7(***) 1.2 amaurochalinus 12.7(***) 1.5 chiguanco 11.1 1.1 falcklandii 1.1 2.9 fumigatus 0.0 1.5 fuscater gigas 2.1 0.6 grayi casius 2.9 1.9 grayi grayi 27.4(***) 0.3 ignobilis 2.3 0.5 leucomelas 2.3 3.6 migratorius achrusterus 0.3 1.6 inigratorius caurinus 41.6(***) 1.8 migratorius confinus 6.6(*) 0.2 m. migratorius central 16.7(***) 1.8 m. migratorius east 13.3(***) 3.0 m. migratorius north 7.7(**) 0.7 migratorius nigrideus 15.1(***) 0.0 migratorius propinquus 0.0 25.3(***) nigrescens 9.4(***) 0.2 nudigenis 20.2(***) 13.1 olivator 11.5(***) 0.7 plebejus 2.1 3.0 plumbetis 19.6(***) 5.3 rufitorques 23.1(***) 4.3 rufiventris 8.4(***) 4.2(*) rufopalliatus 3.0 0.1 serranus 3.0 0.1
(*) P < 0.05; (**) P < 0.01; (***) P < 0.001.
Table 14 reveals that both immatures and adults of most populations show male-biased sex ratios in museum collections. In only one case, that of T. migratorius achrusterus immatures, was the sex ratio significantly biased in favor of females. In addition, immatures exhibited significant male-biased sex ratios as frequently as did adults. Two aspects of Table 14 deserve further note. First, nine species of Turdus do not exhibit significantly male-biased sex ratios among adults, suggesting that males do not sing conspicuously, that both sexes sing equally conspicuously, or that the population was not adequately sampled during the breeding season.
Second, in a small number of species, adult and immature sex ratios differ from each other. Two of these cases seem unusual in that immature sex ratios are significantly male-biased, whereas adult sex ratios are significantly less so. The two samples belong to T. migratorius propinquus, which is widely distributed in western North America, and T. rufiventris, a subtropical South American species.
Three species exhibit male-biased sex ratios as adults but not as immatures, suggesting that breeding might be delayed until the 2nd yr. These are T. nudigenis (tropical), T. plumbeus (tropical), and T. rufitorques (montane Central America). All three have very high estimated adult survival rates (0.835, 0.899, and 0.840, respectively). Collecting bias favoring adults (males in this case) also biases the estimate of annual adult survival rate. The magnitude of this bias is B(S) = 0.5S(I - E(S))([Beta] - 1), where [Beta] is the factor by which adult males are collected over adult females. This equation can be rearranged to express B(S) in terms of the measured variables E(S) and [Beta], giving
B(S) = E(S)(1 - E[S1)([Beta] - 1)/ 2 + (I - E[S])([Beta] - 1).
For the three species showing age-specific sex bias, values of [Beta] were 2.02, 1.93, and 2.50, yielding biases of +0.065, +0.040, and +0.090, respectively. Accordingly, the corrected estimates of S would be 0.770, 0.859, and 0.750.
Survival and fecundity
The plumage of juvenile Turdus differs conspicuously from that of either immatures or adults (Fig. 1), and juveniles undoubtedly are collected with a negative bias. Yet, juveniles are commonly represented in museum collections. If the collecting bias were unrelated to other attributes of life history and demography, then the relative proportions of juveniles in collections would provide a useful index to the production of offspring in populations. Little is known about the behavior of juveniles in different regions, or the duration of the juvenile plumage, so indices of production based on representation in collections of individuals in juvenal plumage must be used with caution. If the juvenal plumage were retained longer in the tropics than in temperate regions, indices of breeding productivity would be biased upward in the tropics. However, we observe the opposite trend.
Two indices of the relative abundance of juveniles in populations are used here. The first is restricted to the period of the year during which juveniles appear in populations, generally the latter part of the breeding season and prior to the postjuvenal molt. This period was arbitarily determined to be those consecutive months with the highest numbers of juveniles in museum collections. The number of juveniles was divided by the total number of adults (not including immatures, that is, I st-yr birds) collected during the same months to calculate the index J(breed). It was not possible to calculate a meaningful value for J(breed) for species lacking well-defined breeding seasons or with small numbers of juveniles in collections. For all species, the number of all juveniles collected throughout the year was divided by the total number of adults in collections to obtain the index J(total).
In a stable population, adult mortality must be balanced by recruitment of young into the breeding population. The annual mortality rate of adults (1 - E[S]) divided by the total number of juveniles per adult (J[total]) is related to the fraction of juveniles entering the breeding population, and thus provides a relative index to the survival of juveniles to maturity (E[[S.sub.a]).
J(breed) and (total) are correlated (r = 0.74, P [is less than] 0.0001, n = 21). J(breed) is inversely related to annual adult survival ([F.sub.1,20] = 6.2, P = 0.02, [R.sup.2] = 0.24; J[breed] = 1.054 [1 SE = 0.290] - 0.988 [1 SE = 0.397] E[S]). As one would expect, J(total) also is inversely related to adult survival ([F.sub.1,25] = 5.0, P = 0.035, [R.sup.2] = 0.17; J[total] = 0.208 [1 SE = 0.054] 0.163 (1 SE = 0.073] E[S]). A graph of J(total) as a function of E(S) (Fig. 5) reveals that four Andean populations of Turdus are conspicuous outliers, exhibiting much higher values of J(total) than lowland tropical and Central American montane species with similar adult survival rates. When the Andean populations are deleted from the analysis, the relationship between J(breed) and E(S) improves slightly ([F.sub.1,16] = 6.5, P = 0.02, [R.sup.2] = 0.29; J[breed] = 1.185 [1 SE = 0.335] - 1.210 [1 SE = 0.476] E[S]), and that between J(total) and E(S) improves substantially ([F.sub.1,20] = 20.7, P = 0.0002, [R.sup.2] = 0.51; J[total] = 0.279 [1 SE = 0.044] - 0.281 [1 SE = 0.062] E[S]).
[Figure 5 ILLUSTRATION OMITTED]
The relative survival of juveniles to maturity (E[[S.sub.a]]) is independent of annual adult survival E(S) (Fig. 6; [F.sub.1,25] = 1.6, P = 0.21).
[Figure 6 ILLUSTRATION OMITTED]
A second approach to estimating rates of fecundity and recruitment uses field data on breeding productivity to calculate the number of fledglings produced per individual per year. In a constant population, adult mortality must be balanced by recruitment, which is the product of fecundity and prereproductive survival. Fecundity (F, fledglings per individual per year) is a function of clutch size, nest survival rate, season length, interclutch interval (Ricklefs and Bloom 1977), and proportion of the population breeding. Rate of nest initiation during the egg-laying season is estimated from nest failure rate and the lengths of successful and unsuccessful nesting cycles. Fecundity is then the product of initiation rate, season length, clutch size, and nest survival rate (see Ricklefs and Bloom 1977). Survival from fledging to the beginning of the following breeding season (E[[S.sub.0]]) is equal to the ratio of annual adult mortality (E[Al] = 1 - E[S]) to annual fecundity (fledglings per individual parent), assuming that all adult individuals breed.
Fecundity in temperate populations of Turdus (F 2.6) exceeds that of most lowland tropical populations (F = 1.2-2.4; Table 15), paralleling the latitudinal trend observed in passerine birds more generally (Ricklefs and Bloom 1977). In this limited sample, F and J(total) appear to be generally correlated (Fig. 7). Furthermore, differences in fecundity nearly balance differences in adult mortality between tropical and temperate populations. Consequently, survival from fledging to the beginning of the next breeding season appears to be similar in temperate (E[[S.sub.0]] = 0.19-0.21) and tropical (0.12-0.21) species of Turdus.
[TABULAR DATA 15 NOT REPRODUCIBLE IN ASCII]
[Figure 7 ILLUSTRATION OMITTED]
The lowest value of E[[S.sub.0]] is that for T. nudigenis, a lowland tropical species of northern South America. T. nudigenis was one of the few species for which the absence of a male-biased sex ratio among immature individuals suggested that reproduction is delayed until the end of the 2nd yr. Assuming that 2nd-yr and adult survival both equal S, a balanced population requires that (1 - S) = F[S.sub.0] S, which can be rearranged to [S.sub.0] = (1 - S)/FS. Accordingly, for T. nudigenis, for which E(S) = 0.836 and F = 1.36, [S.sub.0] may be estimated as 0.144, and survival to first reproduction, [S.sub.a], as 0.120. However, the age-dependent sex ratios also cause a bias in the estimation of S. When corrected for this bias, the value of E(S) for T. nudigenis becomes 0.770, which results in estimated values of [S.sub.0] = 0.220 and S. = 0.169, assuming that breeding begins at the end of the 2nd yr. These corrected values of E([S.sub.0]) and E(S.) are similar to those of the other species in Table 15, thus supporting the previous finding that rate of recruitment is independent of annual adult survival.
This analysis suggests a number of general conclusions concerning methods of demographic analysis and geographical patterns in the demography of songbird populations. (1) Museum collections can provide useful information concerning the demography of populations. Errors in estimation, and biases resulting from relaxing assumptions about estimating demography from age ratios, are small compared to the range of values for annual adult survival among populations. (2) Populations of Turdus reveal a striking increase in survival rate from high latitudes towards the equator, especially in the Northern Hemisphere. Survival is particularly high among montane populations in the tropics. (3) Survival rate is strongly correlated with climate variables, particularly the difference between maximum and minimum mean monthly temperatures, i.e., temperature seasonality. (4) Fecundity, assessed by the relative frequency of juveniles in collections, is inversely related to adult survival rate. This finding is consistent with broader demographic comparisons among avian taxa (e.g., Ricklefs 1977, Johnson and Geupel 1996), which have shown that the ratio of fecundity to mortality is a life history invariant (Charnov 1993). (5) A consequence of the constant fecundity/adult mortality ratio is that the estimated prereproductive (generally 1st-yr) survival of juveniles is independent of adult survival rate; this rate assumes values close to 0.2 in Turdus, which appears to be typical of other avian taxa (Ricklefs 1977). Indeed, the low apparent fecundity of tropical populations, the inverse relationship between fecundity and adult survival, and the independence of prereproductive and adult survival observed in this study are generally consistent among comparative studies of avian life histories (Ricklefs 1973, 1977, Zammuto 1996, Bennett and Harvey 1988, Sather 1988).
Data obtained for Turdus clearly indicate low fecundity and high survival of adults in the tropics compared to temperate latitudes. Thus, the results of this study support the widely held idea that survival rate decreases with increasing latitude. The results also contradict the conclusions of Karr et al. (1990) that survival rates of temperate and tropical birds do not differ. However, the data presented for Turdus in this study do not directly address the results of Karr et al. because their study did not include Turdus or other members of the subfamily Turdinae. If, however, adult survival in tropical populations of Turdus were as low as that of temperate populations, 1st-yr survival of fledglings would have to be on the order of 0.40, that is, nearly as high as that of adults in the same population, and twice the value required to balance the demographic equation in temperate populations.
The basic demographic patterns observed in this analysis, including the relationship of adult survival rate to latitude and altitude, the inverse relationship between fecundity and adult survival, and the apparent invariance of prereproductive survival, raise basic issues of life history evolution in birds. Because annual adult survival is strongly correlated with seasonality of temperature, it is tempting to link variation in survival to factors that cause mortality during the cold season. Presumably, this is a time of stress for many birds, particularly when ice storms and snow cover may make foraging difficult, or when low temperatures increase food requirements and reduce food availability (Martin 1987). Latitudinal variation in predation may also influence survival patterns, although it is difficult to determine the relative contributions of predation and food stress to mortality.
The inverse correlation between fecundity and adult survival may arise through (1) adjustment of reproductive effort with respect to expectation of further life (Cody 1971), (2) density-dependent interactions between population density and the food supply available for reproduction (Ricklefs 1977, 1980, 1992), or (3) correlation of environmental conditions that separately influence adult survival and fecundity (Ricklefs 1983). Based on a variety of considerations, Ricklefs (1977) concluded that variation in reproductive effort was unlikely to cause much of the difference in fecundity or survival between tropical and temperate birds. Indeed, reproductive effort is adjusted in direct relation to adult mortality rate, and, therefore, variation in risk incurred by reproduction should merely enhance underlying differences in adult survival rates between populations. In addition, Cooch and Ricklefs (1994) concluded that annual variation in survival or reproductive success would not influence optimal reproductive effort enough to account for observed latitudinal patterns. Furthermore, important components of fecundity seemingly are relatively insensitive to variation in parental investment: length of the breeding season is controlled primarily by the seasonal course of food availability; nest predation probably is reduced little by defense behaviors of adults. Thus, although parental reproductive effort is predicted to decrease in response to increased adult survival rate and reduced variation in survival rate, this adjustment probably has little consequence for realized fecundity or adult survival.
Ashmole (1963) and Ricklefs (1980) argued that clutch size is controlled by density-dependent feedback of adult population density on food resources available for reproduction. It was hypothesized that, in highly seasonal environments, adult populations are kept low by unfavorable conditions during the nonbreeding season, so that each surviving adult has a relatively large food resource available for breeding during the favorable period of the year. In less seasonal environments, high adult survival during the nonbreeding season (if, indeed, there is any pause in breeding) results in high population densities, reduced absolute levels of food, and greatly reduced per capita food supplies. Few data are available, but population densities of songbirds do appear to be higher in tropical than in temperate regions, even when the higher biological productivity of tropical environments is taken into account (Ricklefs 1980). Furthermore, annual adult survival rates and breeding population densities of Neotropical migrants appear to be higher than those of resident taxa breeding in the same locations in North America (e.g., Greenberg 1980), and breeding productivity of migrants is markedly lower. These observations are consistent with a more benign tropical environment resulting in higher adult over-winter survival, and with density-dependent feedback of adult population density on breeding success.
Possibly, attributes of the environment that independently determine clutch size and adult survival are themselves correlated, although such factors have not been analyzed. For example, length of the breeding season is a large component of fecundity, and it is possible that less seasonal environments that permit prolonged reproduction also result in higher adult survival. Comparative analysis of Turdus populations show, however, that low seasonality is associated with high adult survival and low, rather than high, fecundity, so this pathway of environmental correlation cannot contribute to the demographic patterns observed here. Indeed, without density-dependent feedbacks or life history trade-offs, it is difficult to envision how attributes of the environment that promote adult survival do not also promote high fecundity: high food availability, low predation, and short periods of environmental stress.
The absence of marked differences in prereproductive survival between populations with divergent rates of fecundity and adult survival is consistent with population studies of birds more generally (Ricklefs 1977, Charnov 1993). Adult mortality ([time.sup.-1]) and fecundity ([time.sup.-1]) vary in direct proportion to each other, and because prereproductive survival (dimensionless) is the ratio of the first to the second, it becomes a life history invariant in the terminology of Charnov (1993). In a broad comparison of avian taxa, prereproductive survival of fledglings, estimated as the ratio of adult mortality to fecundity, was found to range narrowly between [approximately equals] 0.15 and 0.20 in 14 species (Ricklefs 1977). This invariance suggests that prereproductive survival may be under strong social control in a density-dependent system. Otherwise, one would expect ecological factors that determine adult survival to similarly affect prereproductive survival. For example, if long life-span derives from living in an environment essentially free of density-independent mortality factors, such as snowstorms making food inaccessible, then survival should be high in all age classes. In broad comparisons, 1st-yr survival and adult survival generally are positively correlated (Ricklefs 1973, Saether 1989).
The invariance in prereproductive survival over a wide variety of birds having different adult survival rates is the result of delayed maturity. Because individuals in most populations of Turdus reach maturity at age 1 yr, invariance in prereproductive survival must be produced by the intrinsic symmetry of adult survival and fecundity, followed by density-dependent feedbacks on either 1st-yr survival rate or the fraction of 1-yr-old birds that fail to become breeders. These two possibilities cannot be distinguished in the type of analysis presented here. Density dependence may be achieved by social feedbacks that act by restricting access of young birds to food or by forcing young to occupy suboptimal habitats (e.g., Krebs 1971). Alternatively, the acquisition of feeding skills by young tropical birds may be so difficult that fledglings initially have poor survival and require prolonged behavioral development to attain adult levels of survival. Although demographic patterns are clarified by analyses of the kind presented in this paper, understanding the origin and maintenance of these patterns will require additional observations, experiments, and analyses.
I am grateful to S. C. White-Schuler for assistance in compiling and analyzing the data, and to J. Brawn, J. Clobert, R. Colwell, C. Francis, R. S. Greenberg, S. Heard, J. R. Karr, J. D. Nichols, R. J. O'Connor, J. Rotenberry, D. W. Snow, and several anonymous reviewers for comments on an earlier version of this manuscript. The manuscript was completed while the author was a Regents' Fellow at the Smithsonian Tropical Research Institute.
Anderson, D. R., A. P. Wywialowski, and K. P. Burnham. 1981. Tests of the assumptions underlying life table methods for estimating parameters from cohort data. Ecology 62:1121-1124.
Ashmole, N. P. 1963. The regulation of numbers of tropical oceanic birds. This 1036:458-473.
Beklova, M. 1972. Age structure and mortality of the Czechoslovakian populations of Turdus merula L. 1758, Sturnus vulgaris L. 1758, and Parus major L. 1758. Zoologica Listy 21:337-346.
Bennett, P. M., and P. H. Harvey, 1988. How fecundity balances mortality in birds. Nature 333:216.
Brawn, J. D., J. R. Karr, and J. D. Nichols, 1995. Demography of birds in a neotropical forest: effects of allometry, taxonomy, and ecology. Ecology 76:41-51.
Brown, J. L. 1974. Alternate routes to sociality in jays--with a theory for the evolution of altruism and communal breeding. American Zoologist 14:63-80.
--. 1987. Helping and communal breeding in birds: ecology and evolution. Princeton University Press, Princeton, New Jersey, USA.
--. 1989. Habitat saturation and ecological constraints: origin and history of the ideas. Condor 91:1010-1013.
Brownie, C., D. R. Anderson, K. P. Burnham, and D. S. Robson, 1985. Statistical inference from band recovery data: a handbook. Second edition. U.S. Fish and Wildlife Service Resource Publication 156.
Caughley, G. 1977. Analysis of vertebrate populations. John Wiley, New York, New York, USA.
Charnov, E. L. 1993. Life history invariants. Some explorations of symmetry in evolutionary biology. Oxford University Press, New York, New York, USA.
Clobert, J., and J.-D. Lebreton, 1991. Estimation of demographic parameters in bird populations. Oxford Ornithological Series Number 1:75-104.
Clutton-Brock, T., editor, 1988. Reproductive success: studies of individual variation in contrasting breeding systems. University of Chicago Press, Chicago, Illinois, USA.
Cody, M. L. 1966. A general theory of clutch size. Evolution 20:174-184.
--. 1971. Ecological aspects of reproduction. Pages 461-512 in D. S. Farner and J. R. King, editors. Avian Biology, Volume I. Academic Press, New York, New York, USA.
Cooch, E. G., and R. E. Ricklefs. 1994. Do variable environments significantly influence optimal reproductive effort in birds? Oikos 69:447-459.
Curio, E. 1989. Some aspects of avian mortality patterns. Mitteilungen aus dem Zoologischen Museum Berlin 65, Supplement: Annalen der Ornithologie 13:47-70.
Dobson, A. 1990. Survival rates and their relationship to life history traits in some common British birds. Current Ornithology 7:115-146.
Emlen, S. T. 1982. The evolution of helping. I. An ecological constraints model. American Naturalist 119:29-39.
Farrier, D. S. 1955. Bird-banding in the study of avian population dynamics. Pages 397-491 in A. Wolfson , editor' Recent studies in avian biology. University of Illinois Press, Urbana, Illinois, USA.
Fogden, M. P. L. 1972. The seasonality and population dynamics of equatorial birds in Sarawak. This 114:307-343.
Greenberg, R. 1980. Demographic aspects of long-distance migration. Pages 493-504 in A. Keast and E. S. Morton, editors. Migrant birds in the neotropics. Smithsonian Institution Press, Washington, D.C., USA.
Greenburg, R., and J. Gradwohl. 1986. Constant density and stable territoriality in some tropical insectivorous birds. Oecologia 69:618-625.
Hanson, W. R. 1963. Calculation of productivity, survival, and abundance of selected vertebrates from sex and age ratios. Wildlife Monographs 9:1-60.
Henny, C. J. 1972. An analysis of the population dynamics of selected avian species. Wildlife Research Reports 1:1-99.
Hilden, O. 1982. Proportion of juveniles as a measure of adult mortality in a breeding population of Great Tits Parus major and Blue Tits P. caeruleus. Ornis Fennica 59:149-151.
Holmes, D. J., and S. N. Austad. 1995a. The evolution of avian senescence patterns: implications for understanding primary aging processes. American Zoologist 35:307-317.
Holmes, D. J., and S. N. Austad. 1995b. Birds as animal models for the comparative biology of aging: a prospectus. Journal of Gerontology 50A:B59-B66.
Howell, J. C. 1942. Notes on the nesting habits of the American Robin (Turdus migratorius L.). American Midland Naturalist 28:529-603.
Johnson, M. D., and G. R. Geupel. 1996. The importance of productivity to the dynamics of a Swainson's Thrush population. Condor 98:133-141.
Karr, J. R., J. D. Nichols, M. K. Klimkiewicz, and J. D. Brawn. 1990. Survival rates of tropical and temperate forest birds. American Naturalist 136:277-291.
King, J. R., and L. R. Mewaldt. 1987. The summer biology of an unstable insular population of White-crowned Sparrows in Oregon. Condor 89:549-565.
Knupp, D. M., R. B. Owen, Jr., and J. B. Dimond. 1977. Reproductive biology of American Robins in northern Maine. Auk 94:80-85.
Krebs, J. R. 1971. Territory and breeding density in the Great Tit Parus major L. Ecology 52:2-22.
Lack, D. 1946. Do juvenile birds survive less well than adults? British Birds 39:258-264.
--. 1947. The significance of clutch-size. Parts 1 and 2. Ibis 89:302-352.
--. 1954. The natural regulation of animal numbers. Oxford University Press, London, UK.
Loery, G., K. H. Pollock, J. D. Nichols, and J. E. Hines. 1987. Age-specificity of avian survival rates: an analysis of capture-recapture data for a Black-capped Chickadee population, 1958-1983. Ecology 68:1038-1044.
Lyon, B. E., and R. D. Montgomerie. 1986. Delayed maturation in passerine birds: reliable signalling by subordinate males. Evolution 40:605-615.
Martin, T. E. 1987 . Food as a limit on breeding birds: a lifehistory perspective. Annual Review of Ecology and Systematics 18:453-487.
McDonald, D. B. 1989. Cooperation under sexual selection: age-graded changes in a lekking bird. American Naturalist 134:709-730.
Mewaldt, L. R., and J. R. King. 1985. Breeding-site faithfulness, reproductive biology, and adult survivorship in an isolated population of Cassin's Finches. Condor 87:494-510.
Miller, A. H. 1931. Systematic revision and natural history of the American shrikes (Lanius). University of California Publications in Zoology 38:11-242.
--. 1963. Seasonal activity and ecology of the avifauna of an American equatorial cloud forest. University of California Publications in Zoology 66:1-78.
Moreau, R. E. 1944. Clutch size: a comparative study, with reference to African birds. Ibis 86:286-347.
Murray, B. G., Jr. 1985. Evolution of clutch size in tropical species of birds. Ornithological Monographs 36:505-519.
Newton, I. 1989. Synthesis. Pages 153-173 in I. Newton, editor. Lifetime reproduction in birds. Academic Press, London, UK.
Nichols, J. D., B. R. Noon, S. L. Stokes, and J. E. Hines. 1981. Remarks on the use of mark-recapture methodology in estimating avian population size. Studies in Avian Biology 6:121-136.
Oniki, Y. 1979. Is nesting success of birds low in the tropics? Biotropica 11:60-69.
Proctor-Gray, E., and R. T. Holmes. 1981. Adaptive significance of delayed attainment of plumage in male American Redstarts: tests of two hypotheses. Evolution 35:742-751.
Ricklefs, R. E. 1966. The temporal component of diversity among species of birds. Evolution 20:235-242.
--. 1973. Fecundity, mortality, and avian demography. Pages 366-435 in D. S. Farrier, editor. Breeding biology of birds. National Academy of Sciences, Washington, D.C., USA.
--. 1975. The evolution of co-operative breeding in birds. Ibis 117:531-534.
--. 1977. On the evolution of reproductive strategies in birds: reproductive effort. American Naturalist 111:453-478.
--. 1980. Geographical variation in clutch size among passerine birds: Ashmole's hypothesis. Auk 97:38-49.
--. 1983. Comparative avian demography. Current Ornithology 1:1-32.
--. 1992. The megapopulation: a model of demographic coupling between migrant and resident landbird populations. Pages 537-548 in J. M. Hagan III and D. W. Johnston, editors. Ecology and conservation of neotropical migrant landbirds. Smithsonian Institution Press, Washington, D.C., USA.
--. 1993. Sibling competition, hatching asynchrony, incubation period, and life-span in altricial birds. Current Ornithology 11: 199-276.
Ricklefs, R. E., and G. Bloom. 1977. Components of avian breeding productivity. Auk 94:86-96.
Roff, D. A. 1992. The evolution of life histories. Chapman and Hall, London, UK.
Rohwer, S., and G. S. Butcher. 1988. Winter versus summer explanations of delayed plumage maturation in temperate passerine birds. American Naturalist 131:556-572.
Rohwer, S., S. D. Fretwell, and D. M. Niles. 1980. Delayed maturation in passerine plumages and the deceptive acquisition of resources. American Naturalist 99:129-141.
Rose, M. R. 1991. Evolutionary biology of ageing. Oxford University Press, Oxford, UK.
Saether, B.-E. 1988. Patterns of covariation between life history traits of European birds. Nature (London) 331:616-617.
--. 1989. Survival rates in relation to body weight in European birds. Ornis Scandinavica 20:13-21.
Savidge, I. R., and D. E. Davis. 1974. Survival of some common passerines in a Pennsylvania woodlot. Bird-Banding 45:152-155.
Simms, E. 1978. British thrushes. Collins, London, UK.
Skutch, A. F. 1981. New studies of tropical American birds. Publications of the Nuttall Ornithological Club 19:1-281.
--. 1985. Clutch size, nesting success, and predation on nests of Neotropical birds, reviewed. Ornithological Monographs 36:575-594.
Snow, D. W. 1956. The annual mortality of the Blue Tit in different parts of its range. British Birds 49:174-177.
--. 1962. A field study of the Black-and-White Manakin, Manacus manacus, in Trinidad. Zoologica 47:65-104.
--. 1966. The population dynamics of the Blackbird. Nature 211:1231-1233.
--. 1969. The moult of British thrushes and chats. Bird Study 16:115-129.
Snow, D. W., and A. Lill. 1974. Longevity records for some neotropical land birds. Condor 76:262-267.
Snow, D. W., and B. K. Snow. 1963. Breeding and the annual cycle in three Trinidad thrushes. Wilson Bulletin 75:27-41.
Stearns, S. C. 1992. The evolution of life histories. Oxford University Press, Oxford, UK.
Sutherland, W. J., A. Grafen, and P. H. Harvey. 1986. Life history correlations and demography. Nature 320:88.
Wernstedt, F. L. 1972. World climatic data. Climatic Data Press, Lemont, Pennsylvania, USA.
Willis, E. O. 1974. Populations and local extinctions of birds on Barro Colorado Island, Panama. Ecological Monographs 44:153-169.
--. 1983. Longevities of some Panamanian forest birds, with note of low survivorship in old Spotted Antbirds (Hylophylax naevioides). Journal of Field Ornithology 54:413-414.
Young, H. 1955. Breeding behavior and nesting of the Eastern Robin. American Midland Naturalist 53:329-352.
Zammuto, R. M. 1986. Life histories of birds: clutch size, longevity, and body mass among North American game birds. Canadian Journal of Zoology 64:2739-2749.
Manuscript received 28 September 1995; accepted 19 February 1996; final version received 9 May 1996.
Robert E. Ricklefs Department of Biology, University of Missouri, 8001 Natural Bridge Road, St. Louis, Missouri 63121 USA,(1) and Smithsonian Tropical Research Institute, Apartado 2072, Balboa, Republic of Panama
(1) Address to which correspondence should be sent.
|Printer friendly Cite/link Email Feedback|
|Author:||Ricklefs, Robert E.|
|Date:||Feb 1, 1997|
|Previous Article:||A model of nitrogen uptake by Eriophorum vaginatum roots in the field: ecological implications.|
|Next Article:||Estimates and tests of per capita interaction strength: diet, abundance, and impact of intertidally foraging birds.|