Cultural inheritance of song and its role in the evolution of Darwin's finches.
Studies of bird song over the last 40 years have provided observational and experimental evidence that species can discriminate between conspecific and heterospecific song (Marler 1960; Thielcke 1962; Emlen 1972; Gill and Murray 1972; Becker 1977; Marler and Peters 1977; Thielcke et al. 1978; Grant 1984; Ratcliffe and Grant 1985; Grant and Grant 1989; Nelson 1989; Baker and Baker 1990; Lynch and Baker 1990; Searcy 1990; Baker 1991, 1994; Nelson and Marler 1993; Searcy et al. 1995). Mating decisions by females may be based largely on that discrimination (Eriksson and Wallin 1986; Searcy and Brenowitz 1988; Searcy 1990; Ten Cate et al. 1993). If song is a factor in species discrimination, it is important to know whether or not it is learned; from whom it is learned; and whether the learning is subject to error and, if so, whether or not it leads to hybridization (Nicolai 1959; Gill and Murray 1972; Immelmann 1975; Kroodsma 1978; Guttinger 1979; Bohner 1983, 1990; Gelter 1987; Wallin 1987; Baptista and Morton 1988; Alatalo et. al. 1990; Baker 1991; Immelmann et al. 1991; Kruijt and Meeuwissen 1991, 1993; Ryan et al. 1994).
Species of birds vary considerably in their song characteristics and how they acquire them (Nottebohm 1972; Kroodsma 1978; Payne 1981; Payne et al. 1981; Konishi 1985; Marler and Peters 1987, 1988a,b). Some, like the ringed turtledove, Streptopelia risoria (Nottebohm and Nottebohm 1971), and the eastern phoebe, Sayornis phoebe (Kroodsma and Konishi 1991) have a genetically determined, structurally simple and stereotyped song, whereas in all oscine passetines song is learned (Marler and Peters 1977). Some learn their song during a short sensitive period early in life (Marler 1987), before or after dispersing from the natal territory, or both; examples are the chaffinch, Fringilla coelebs (Thorpe 1958), zebra finch, Taeniopygia guttara (Immelmann and Suomi 1981), song and swamp sparrows, Melospiza melodia and M. georgiana (Marler and Peters 1977, 1988a), marsh wren, Cistothorus palustris (Kroodsma and Pickett 1980), and marsh tit, Parus palustris (Rost 1990). Others, like the red-winged blackbird, Agelaius phoeniceus (Yasukawa et al. 1980), the canary, Serinus canaria (Nottebohm and Nottebohm 1978), and the starling, Sturnus vulgaris (Bohner et al. 1990), increase and modify their repertoire throughout life.
Studies by Konishi (1985), Nottebohm (1993), and others on the neural pathways and song nuclei in the brain have shown that the sensitive period for song learning coincides with low plasma testosterone levels and with neurogenesis and synapse formation in the brain. In species such as the canary that modify their song repertoire throughout life, episodes of neuronal replacement in the song nuclei recur each spring, allowing the learning and production of new song types (Nottebohm 1993). Termination of the sensitive period for song learning coincides with a rise in testosterone levels, and can be influenced by the environment (Kroodsma and Pickert 1980; Bohner et al. 1990; Nottebohm 1993). The ability to recognize and discriminate between songs of other individuals continues in both males and females throughout life, involving neural pathways not used in imitative learning (Nottebohm 1993).
Understanding the source of a culturally inherited character like song is important because the mode of inheritance influences its rate of change, and the degree to which individuals within a population vary (Cavalli-Sforza et al. 1982). Cavalli-Sforza et al. (1982) describe the learned transmission of culturally inherited characters as "vertical" if it proceeds from parent to offspring, "horizontal" if it derives from neighbors from the same generation, and "oblique" if it derives from members of the previous generation. They show that vertical transmission is the most stable. This pathway results in high heterogeneity within and between populations, whereas horizontal and oblique transmission result in a sharing of traits by members of a group and a rapid turnover of those traits between generations.
Song, when learned, is a culturally inherited trait having some features in common with speech (Marler and Peters 1977; Cavalli-Sforza et al. 1994). Like speech, song frequencies can change across generations faster than gene frequencies as a result of small inconsistencies in transmission from one generation to the next (Bohner 1983; Lynch and Baker 1990; Cavalli-Sforza et al. 1994). Song differences between populations are likely to arise faster in small isolated populations than in large, continuously distributed, populations. When learning of song is restricted to a sensitive period early in life, imprinting may occur. Imprinting simultaneously guides mating within groups and hinders mating between groups, thereby tending to strengthen the genetic isolation between groups (Immelmann 1975, 1980; Baker and Marler 1980; Bischoff and Clayton 1991; Immelmann et al. 1991; Kruijt and Meeuwissen 1991, 1993; Casey and Baker 1992; Baptista et al. 1993; Ten Cate et al. 1993). On the other hand, migration of individuals between groups, if followed by interbreeding, could introduce new song types into the population.
In this paper we demonstrate the method of song transmission and stability of song across generations in two sympatric congeneric species of Darwin's finches on the small island of Daphne Major in the center of the Galapagos archipelago. We examine the mating pattern of females with respect to the songs of their fathers and their mates, and discuss the importance of song in the evolutionary dynamics and speciation of Darwin's finches. Although valuable information on the inheritance of song has been obtained from a variety of bird species in the laboratory and in nature, transmission between generations has rarely been examined in a natural population over several generations with individually marked birds (but see Payne et al. 1987; Payne and Payne 1993). In this study we have followed the transmission of song for up to five generations in natural and unmanipulated populations.
The species concept we work with is fundamentally biological, and we apply it by a combination of morphological and biological methods (for details see Grant 1993).
Darwin's Finches and Their Songs
Geospiza fortis (medium ground finch) and G. scandens (cactus finch) are resident on Isla Daphne, and are socially monogamous. Males defend their territories throughout the breeding season, and in G. scandens, but not G. fortis, this defense continues throughout the year. Once successful breeding has taken place males retain their territories for life with only minor boundary changes. Males initiate nest building, females incubate the eggs, and males and females feed their nestlings and fledglings. Only males sing, but both males and females discriminate between conspecific and hetero-specific males on the basis of song (Grant 1984; Ratcliffe and Grant 1985; Grant and Grant 1989). Males sing vigorously on their territories before obtaining a mate. After egg laying has begun song rate declines, but singing continues periodically near the active nest and particularly after nestlings and fledglings have been fed (Millington and Price 1985). Singing occurs frequently during interactions with intruders and neighboring males (Ratcliffe 1981). Therefore young are exposed repeatedly to their father's song, and to a lesser extent their natal neighbors' songs, as well as songs of other species.
Darwin's finches resemble zebra finches in that they have a sensitive period for song learning extending from approximately age 10-40 d (Bowman 1983). This corresponds to a period that starts a few days before nest departure, extends through the fledgling phase while young are fed by their parents, and ceases soon after they have become independent. Typically a male sings an initially undeveloped subsong that crystallizes into adult song before he establishes a territory (Grant and Grant 1989). This song remains unaltered throughout life (Millington and Price 1985; Grant and Grant 1989; Gibbs 1990). In possessing a repertoire of only one song, these finches resemble some other emberizines in the temperate zone, including the North American Spizella passerina (Marler and Isaac 1960), S. arborea (Naugler and Ratcliffe 1994) and Zonotrichia leucophrys (Baptista 1990).
Earlier studies examining song transmission in G. fortis (Ratcliffe 1981; Millington and Price 1985; Gibbs 1990) used qualitative descriptions and some sonagrams to document the persistence of song types across generations. Millington and Price (1985), and Gibbs (1990) described discrete song types by what they heard. Although these types can be distinguished, intermediate songs are also heard, and measurements of the expanded sample of sonagrams used here reveal that song types are not discrete but show continuous and extensive variation among individuals.
Populations of finches on Isla Daphne Major have been studied by banding individuals since 1973, and intensively since 1976. From 1979 to 1992 between 95-100% of the birds were banded with a unique color band combination. The identities of parents, sibs, mates, and offspring and the location of territories and nests were determined by observation and mapping. In 1976-1978 and 1984-1995, singing males were recorded with a Uher 4000, Uher 4200, and Sony TCM 5000 recorder, and Uher M516, Sennheiser AKG D900, and ME80 microphones. Sonagrams were made from recordings with a Kay Elemetrics model 5500 and a Krohn-Hite model 3550 filter with high (8.0 kHz) and low (1.0 kHz) settings. They were printed with a Kay Gray Scale Printer model 5509 on paper 120 cm wide. The sample included a few songs of two other, relatively rare, species: G. fuliginosa, the small ground finch, and G. magnirostris, the large ground finch (Grant and Grant, in press a).
Song Measurement and Analysis
From an initial set of nine variables (Green 1992), selected to capture a large part of the structural complexity of the songs, we chose five reliably measured variables for analysis: the duration and maximum frequency of the first and second units and the interval between them, where a unit is defined as uninterrupted sound. All songs were measured on the sonagrams in millimeters with a Pacific Arc metal ruler by PRG. Frequency measurements were made from the base 0 kHz line, and duration (length) measurements were made from the start of the unit being measured. Minimum frequency measurements proved unreliable due to sound fading at low frequency and were not used in any of the analyses. Measurements were analyzed by univariate and multivariate methods as described in the results. The correlation matrix was used for principal components analysis (PCA). Regression analysis was used to predict the song characteristics of sons or the mates of daughters from song characteristics of fathers. The approach is basically the same as that used in heritability analysis of morphology or other continuously varying traits, in which inheritance of traits is assessed by estimating the degree of resemblance between family members, for example parents and offspring (e.g., Falconer 1989).
Repeatability of Measurements
Repeatabilities of measurements made on the same sonagram on different days provide an estimate of measurement error (cf. Lessells and Boag 1987). Measurement error is low for all measured traits of G. fortis, as shown by the high repeatabilities (Table 1A). Repeatabilities of G. scandens song measurements are lower (Table 1). Song recording quality varies according to atmospheric conditions and recording distance from the bird. However, despite this variation, high consistency in song is shown both within and across years. Results of repeatability analyses (Table 1B,C) demonstrate that any deviation in song within or across years can be accounted for by measurement error alone, and not by changes in the song of the bird. Thus each bird sings one song that is retained throughout its lifetime [ILLUSTRATION FOR FIGURE 1 OMITTED].
Ratcliffe (1981), Millington and Price (1985), and Gibbs (1990) classified the songs of G. fortis into four types by listening to them and confirming the classification with sonagrams. Gibbs (1990) gives the most detailed criteria for this classification by ear. A few songs were found by him to be difficult to classify or were misclassified. Figure 2 gives a representative sample of the songs recorded on the island over a much longer period. The variation is more continuous, [TABULAR DATA FOR TABLE 1 OMITTED] less discrete, then previously encountered or recognized. Therefore, to capture the quantitative variation among individual male songs, we used PCA of the song measurements (Nowicki and Nelson 1990; Green 1992). The first two principal components [ILLUSTRATION FOR FIGURE 3 OMITTED] account for 79% of the variance. Factor loadings for each component are given in Table 2.
Geospiza scandens individuals, in contrast to G. fortis, sing basically the same, unvarying, repeated-note song [ILLUSTRATION FOR FIGURE 4 OMITTED]. In its simplicity and uniformity, it resembles the song of a North American emberizine, the chipping sparrow Spizella passerina, which also has a repertoire of one (Marler and Isaac 1960). Song length varies within the same G. scandens individual, apparently due to changes in motivational state. Like G. fortis, G. scandens individuals sing only one song throughout life.
The two species differ strikingly in quantitative features of their songs (Table 3, [ILLUSTRATION FOR FIGURE 5 OMITTED]). For the multivariate comparison, the PCA was repeated with just the first five variables listed in Table 2 [ILLUSTRATION FOR FIGURE 5 OMITTED]. Means differ in PC1 and PC2 scores (t-tests, both P [less than] 0.0001, G. fortis N = 460, G. scandens N = 134). A univariate duplication of these tests showed that all of the five measured song traits (ln-transformed) differ between species (t-tests, all P [less than] 0.0001). Geospiza scandens songs have more repeated units than G. fortis songs, and each unit is shorter and of lower frequency. These differences are conspicuous to the human ear. Variation between individuals is significantly lower for G. scandens songs than for G. fortis songs in PC1 and PC2 scores (F-tests, P [less than] 0.001), and in all the individual traits (F-tests, ln-transformed data, all P [less than] 0.0001), except duration of the first unit. In this trait G. scandens songs vary significantly more than G. fortis (ln-transformed data, [F.sub.133,549] = 272.8, P [less than] 0.0001).
Transmission of Song: Vertical
Gibbs (1990), using a classification of G. fortis songs into song types, found that 78% of 76 sons sang their father's song. We repeated the analysis with a much larger sample of sons (263), ignoring songs that were difficult to classify, and obtained a similar value of 72%. Analysis of principal components scores was then undertaken to avoid the problem of ambiguous classification.
Among the 460 G. fortis males with recorded songs, there were 273 father-son pairs, involving 133 different fathers. In separate analyses, PC1 and PC2 scores of the songs of sons were regressed on the scores of their fathers ([ILLUSTRATION FOR FIGURE 6 OMITTED], upper). Both relationships were positive and significant: for PC1, b = 0.705 [+ or -] 0.047 SE, r = 0.671, P [less than] 0.0001; for PC2, b = 0.365 [+ or -] 0.059, r = 0.349, P [less than] 0.0001. All of the original variables showed significant positive relationships at P [less than] 0.0001.
Thus sons closely resemble their fathers, not only in song type but in quantitative features of their songs. Siblings also resemble each other in the songs they sing. The intraclass correlation coefficient ([r.sub.1]) calculated from ANOVA of 43 same-nest sibs was significant for PC1 scores ([r.sub.1] = 0.655, P = 0.0001). Significant correlations of PC1 scores were also found for 69 different-nest sibs ([r.sub.1] = 0.59, P = 0.0001). These results show that the resemblance between sibs cannot be attributed to the same-nest environment. Father's song is a feature common to all rearing environments.
As shown above, songs sung by G. scandens males vary little among individuals, in contrast to G. fortis. There is no statistically detectable transmission of multivariate song characteristics from father to son, as revealed by PCA that used all variables listed in Table 2: for PC1, N = 56 father-son pairs, b = 0.442 [+ or -] 0.365, r = 0.163, P = 0.23; for PC2, N = 56, b = 0.201 [+ or -] 0.287, r = 0.095, P = 0.49. However, as also noted above, G. scandens songs vary appreciably in the length of the first unit, and this is the only individual song trait showing significant transmission from father to son, as indicated by the father-son associations (b = 0.803 [+ or -] 0.12, r = 0.666, P [less than] 0.0001). The fidelity of transmission might be underestimated by our use of only frequency and duration variables that do not fully capture the structural complexity of the songs.
Genetic or Cultural Inheritance of Song
Song transmission from father to son could be either genetically determined or learned (culturally determined). Analyses of song resemblance between sons and their paternal and maternal grandfathers reveal whether or not song type is genetically inherited. If there is genetic inheritance, and it is polygenic, the slopes (b) of the regressions of songs (PC scores) of sons on songs of fathers provide estimates of heritabilities. Heritabilities estimated by regressing songs of sons on the paternal and maternal grandfather songs should be equal and approximately half of the heritability values calculated by regressing songs of sons on songs of fathers. In contrast to these expectations, if the song type is learned and not genetically determined it will be transmitted only through the paternal line. Thus there should be no resemblance between songs of sons and the songs of their maternal grandfathers, but there should be a resemblance between the songs of sons and the songs of their paternal grandfathers.
The sample of G. fortis comprises 136 son-paternal grandfather pairs involving 47 different paternal grandfathers, and 124 son-maternal grandfather pairs involving 45 different maternal grandfathers. The regression of PC1 scores of sons on paternal grandfather scores is significant (b = 0.540 [+ or -] 0.076, r = 0.523, P [less than] 0.0001), whereas the regression of son scores on maternal grandfather scores is not significantly different from zero (b = -0.036 [+ or -] 0.101, r = 0.032, P [greater than] 0.1), but clearly different from the other regression [ILLUSTRATION FOR FIGURE 6 OMITTED]. The expected value for the slope of sons on grandfather scores is the square of the son-father regression (0.725) from the same data set, 0.525. The slope of the paternal grandfather regression, b = 0.540 [+ or -] 0.076, is close to the expected value. The slope of the maternal grandfather regression, b = -0.036 [+ or -] 0.101, is far from the value (0.523) expected from slope of the son-father regression (0.723) for this data set. Neither of the PC2 score regressions are significant (P [greater than] 0.1).
The positive and highly significant song transmission from paternal grandfather to father to son, and lack of song transmission from maternal grandfather to grandson, reveals that song is culturally and not genetically transmitted across generations.
Songs Not Copied from Fathers
Although most sons sing songs similar to their father's, a few produce very different songs [ILLUSTRATION FOR FIGURE 7 OMITTED]. Either they acquired them by extemporizing or they learned them from other males. During the sensitive period the most frequently heard song, after its father's, is likely to be the song of the nearest neighboring male; or, if a mother switches mates while still feeding the son, the song of her next mate. However, in many species of song birds, young males sing one or more songs similar to their breeding neighbor (Payne 1981; Baptista and Morton 1988; Marler and Nelson 1993; Payne and Payne 1993; Nelson et al. 1995). We therefore examined the influence of father, natal and breeding neighbor, and female mate changes, on male song.
Natal Neighbors (Oblique Transmission). - Songs were recorded from 116 G. fortis sons and both their fathers and nearest natal neighbor; the nearest neighbor was identified by map distances between nests. We performed multiple regression analysis of both potential song influences to determine if the song of the nearest natal neighbor has an independent influence on the song of the sons. Results showed that sons sing a song similar to the father (as before) and not similar to their nearest natal neighbor. Much of the variation in son PC1 scores is statistically accounted for by variation in father PC1 scores (partial regression coefficients [Beta] = 0.825 [+ or -] 0.056, [r.sup.2] = 0.66, P [less than] 0.0001), whereas nearest natal neighbor PC1 scores make no additional contribution to the statistical explanation ([Beta] = 0.058 [+ or -] 0.055, [r.sup.2] = 0.00, P = 0.29). For PC2, again much of the variation in son scores is statistically accounted for by variation in father scores ([Beta] = 0.405 [+ or -] 0.090, [r.sup.2] = 0.21, P [less than] 0.0001), but in this case nearest natal neighbor scores contribute significantly to explaining the residual variation ([Beta] = -0.219 [+ or -] 0.080, [r.sup.2] = 0.06, P = 0.007). Note however that the sign of the coefficient is negative. The opposite signs of the coefficients imply that some features of the song of fathers and nearest neighbors are dissimilar, and that sons copy their fathers but not their nearest neighbors. The correlation between PC2 scores of father and nearest neighbor is negative, but not significantly different from zero (r = -0.08, P = 0.35).
Natal neighbors other than the nearest one could possibly influence the song of a young male. To assess this possibility we performed a biased analysis with the most similar neighbor. First, all neighbors were considered that had nests within twice the map distance between the son's natal nest and its nearest neighbor's nest. Next we computed euclidean distances on a PC1/PC2 plot to identify the neighbor with the song most similar to the son's. The multiple regression analysis of PC1 scores yielded a statistical association with most similar neighbor ([Beta] = 0.193 [+ or -] 0.053, [r.sup.2] = 0.08, P = 0.0004). In contrast, the [r.sup.2]-value for the independent association with father was 0.62. Therefore natal neighbors may have a small influence on the song a male sings as an adult, but it is dwarfed by the magnitude of the paternal influence.
TABLE 2. Factor loadings for the first two principal components of Geospiza fortis (N = 461) and G. scandens (N = 135) song measurements. G. fortis G. scandens PC1 PC2 PC1 PC2 Proportion of variance 0.557 0.229 0.614 0.199 Factor loadings Duration of 1st unit 0.883 0.200 0.581 -0.301 Duration of 2nd unit 0.884 0.256 0.824 -0.503 Max. freq. of 1st unit 0.697 -0.562 0.769 0.554 Max. freq. of 2nd unit 0.832 -0.223 0.859 0.480 Duration of interunit in- terval 0.212 0.820 0.689 -0.297 Duration of penultimate unit - - 0.835 -0.470 Duration of last unit - - 0.801 -0.466 Max. freq. of penultimate unit - - 0.819 0.551 Max. freq. of last unit - - 0.843 0.490 Duration of last interunit interval - - 0.778 -0.183
Breeding Neighbor (Horizontal Transmission). - Geospiza fortis males sing the same song throughout life and do not change songs when the breeding neighbors are replaced by new birds singing different songs. Nevertheless, the possibility remains that some birds could select from the many songs experienced during their sensitive period early in life and sing a song that is similar to their first breeding neighbor, copy it, and ignore the rest, as suggested by Nelson and Marler (1993) in a slightly different context. This possibility is not supported by the observation that males do not change songs even though many change territories, thereby gaining neighbors with different songs, prior to obtaining a mate.
Neither is the possibility supported by the results of a regression analysis of the songs of 125 sons and both their father and first breeding (nearest) neighbor. For PC1 scores, the partial regression coefficient for sons regressed on fathers is high and significant ([Beta] = 0.696 [+ or -] 0.064, [r.sup.2] = 0.49, P [less than] 0.0001), whereas the partial regression coefficient for nearest neighbor is low and nonsignificant ([Beta] = 0.071 [+ or -] 0.065, [r.sup.2] = 0.001, P = 0.28). For PC2 scores the results are similar: the coefficient for father is significant ([Beta] = 0.323 [+ or -] 0.084, [r.sup.2] = 0.11, P = 0.0002) [TABULAR DATA FOR TABLE 3 OMITTED] and the coefficient for nearest neighbor is not significant ([Beta] = -0.038 [+ or -] 0.094, [r.sup.2] = 0.001, P = 0.69).
We repeated the multiple regression analysis of PC1 scores, with most similar breeding neighbor substituted for nearest neighbor, and obtained an almost identical result. Even though a weak statistical association with most similar neighbor was found ([Beta] = 0.141 [+ or -] 0.067, P = 0.0372), the [r.sup.2]-value was only 0.02. By these analyses we have detected at most a minor influence of breeding neighbors on the song sung by a male when an adult.
Song Copying by Geospiza scandens. - The regression analyses were repeated with G. scandens data because, like G. fortis, some G. scandens sons sing songs that differ from their fathers' songs [ILLUSTRATION FOR FIGURE 8 OMITTED]. Sample sizes were N = 35 for the analysis of father and either nearest or most similar natal neighbor, and N = 28 for the analysis with father and either nearest or most similar breeding neighbor.
With PC1 and PC2 scores, none of the regressions was significant. However, variation in the duration of the first unit (ln-transformed) of the song of sons was significantly related to the variation in the song of fathers ([Beta] = 0.746 [+ or -] 0.160, [r.sup.2] = 0.48, P [less than] 0.0001), as found above, although not to the variation in the song of nearest natal neighbors ([Beta] = 0.176 [+ or -] 0.193, [r.sup.2] = 0.01, P = 0.3675). Substituting most similar neighbor for nearest neighbor, following the procedure adopted for the analysis of G. fortis above, we obtained a weakly significant relationship between son and most similar neighbor ([Beta] = 0.358 [+ or -] 0.163, [r.sup.2] = 0.05, P = 0.0355), and a virtually unchanged relationship between son and father ([r.sup.2] = 0.48). Similarly, in the analysis of breeding neighbors father song variation made a significant statistical contribution to son song variation ([Beta] = 0.569 [+ or -] 0.256, [r.sup.2] = 0.19, P = 0.035), but nearest breeding neighbor song variation did not ([Beta] = -0.117 [+ or -] 0.201, [r.sup.2] = 0.03, P = 0.57). In a separate analysis the most similar breeding neighbor made no significant contribution to son song variation ([Beta] = 0.263 [+ or -] 0.196, [r.sup.2] = 0.07, P = 0.1923). Thus the duration of the first unit of the song was copied by sons from fathers' songs. Natal (but not breeding) neighbors may have a small additional influence on the song a male sings as an adult.
Influence of Mate Changes
The above analyses give little evidence that birds copy the songs of their breeding or natal male neighbor. After fledging, offspring of either sex may be fed by both parents, or by either the father or mother exclusively. Fathers do most of the feeding of fledglings when food is abundant and females are renesting (Price and Gibbs 1986). Occasionally a female switches mates while feeding a fledgling. When this happens, the auditory environment previously dominated by the father's song changes to one dominated by another male's song. Learning from either the mother's second male or a male in another area of the island may occur. If it does, it could account for some of the noncopying of father's song.
We do not have a large enough sample of G. fortis fledglings known to be fed exclusively by their mother to test this hypothesis directly. However, we can indirectly assess the possibility of G. fortis sons learning from the mother's second mate by comparing those sons reared by pairs where the female changed mates with those reared by pairs where the female did not change mates. We compared the euclidean distances between father and son in the two groups on a plot of PC1 against PC2 scores; 78 where the female changed mates and 194 where the female did not change mates. The mean euclidean distance between father and son's song in the group where the female changed mates was significantly greater than in the group where the female did not change mates (ln-transformed data, [t.sub.270] = 3.253, P = 0.0013). These result suggest that sons were less likely to copy father's song when mothers switched mates than when the parents remained paired, perhaps because they copied the songs sung by the mother's new mate.
Song Changes across Generations
Inconsistencies in transmission from one generation to the next can affect the evolution of song. Gibbs (1990) reported that males with a minority song type in this G. fortis population had enhanced survival, and observed the predicted trend of minority song types rising in frequency from 1979 to 1985. He suggested that the increase in frequency was due to a correlation between song type and the genetically heritable trait body size that had been subject to selection. Using a slightly modified version of Gibbs's original classification into four song types, we found basic stability in the proportions of song types. The small tendency for the rarer song types to increase from 1979 to 1985 (Gibbs 1990) continued over the next 10 yr [ILLUSTRATION FOR FIGURE 9 OMITTED]; the rarest two types combined increased from 0.24 in 1985 to 0.32 in 1995.
However, our measurements of song and morphology of a larger sample of 210 G. fortis males do not confirm the suggestion of a selection induced change in song frequencies. There was no correlation between PC1 song scores and PC1 morphology (overall body size) scores (r = 0.016, P = 0.82). Analyses of cohorts of birds born in 1978, 1983, 1987, and 1991 showed that their survival over the first four years of life was not significantly associated with different song types in any year ([[[Chi].sup.2].sub.3] tests, all P [greater than] 0.1). Furthermore, we found no significant association between song type and order of breeding in any of these four cohorts.
Nevertheless small changes occurred across years in quantitative features of songs. PC1 scores of song of G. fortis males born (hatched) in the productive years 1978, 1983, 1987, and 1991 were significantly heterogeneous (ANOVA [F.sub.3,383] = 8.138, P = 0.0001). Student-Newman-Keuls post hoc tests showed that 1991 songs differed significantly from 1978 and 1983 songs (P = 0.01) and from 1987 songs (P = 0.05). No major new song types entered the population, rather changes were due to small new variants [ILLUSTRATION FOR FIGS. 2, 9 OMITTED] in quantitatively measured characters, probably due to the accumulation of small innovations as a result of inconsistencies in copying.
Differential Mating of Females with Respect to Song
Young females entering the breeding population for the first time have a limited choice of mates because they pair up later in the season than older and experienced females, some of which retain their mate from the previous year. Young females are therefore likely to take the first available mate. If song has any influence on choice of mates, it is unlikely to be expressed strongly in the first year, owing to restricted opportunities, but could be expressed in the second breeding year when there is a greater choice of males available at the start of the breeding season. Results of analyses are consistent with this reasoning. In the first breeding season, females mated randomly with respect to father's song, as indicated by the lack of a significant association between PC1 song scores of females' mates and females' fathers (r = 0.01, P = 0.9, N = 211 pairs). However in the second breeding year of the females, the relationship is significant and negative (r = -0.22, P = 0.014, N = 122). Thus, in their second year when they have a greater opportunity to choose among males, females pair with those that have songs significantly different from their father's song.
The clear differences between the songs of the two species, and their general long-term stability, provide an opportunity for unambiguous species recognition on Daphne. Heterospecific singing is very rare, and females almost always pair with males that sing the same species song as their fathers.
During the more than 20 yr of study on Daphne there have been 11 instances of males singing heterospecific song; three paired heterospecifically. An example of misimprinting on heterospecific song is shown in Figure 10. It probably arose through intense social interactions with an aggressively dominant neighboring bird (G. magnirostris), overriding an inherent tendency to learn conspecific song, as in known to occur, rarely, in other emberizines (Baptista and Petrinovich 1986; Baptista 1990),
We do not know how the apparent misimprinting occurred in most of the other cases (Grant and Grant, in press a), but the consequence was hybridization between the species. In a sample of 482 females in which both the father's and the mate's song were recorded, there are only two exceptions to the rule that females pair with a male singing the same species song as their father. Of 90 G. scandens females for which both the father and the mate were recorded, 86 mated with a G. scandens male singing a G. scandens song. The remaining four females were daughters of a G. scandens male that sang a G. fortis song, and all four mated with G. fortis males. Of the 392 G. fortis females, 378 mated with G. fortis males singing G. fortis song. The 14 that did not mate conspecifically included 12 females that mated with G. fuliginosa males and two that mated with G. scandens males. All the G. fuliginosa males had songs indistinguishable from G. fortis songs.
The main findings of this study are first, songs of the medium ground finch (G. fortis) resemble their father's songs in quantitative details, as well as the songs of their paternal grandfathers but not their maternal grandfathers; second, their songs differ quantitatively from songs of a sympatric species, the cactus finch (G. scandens); and third, females avoid mating with males that sing heterospecific song, with very rare exceptions.
These results are consistent with studies of song in other species of Darwin's finches. In all species studied, individuals have a short song that remains constant across years (Ratcliffe 1981; Bowman 1983; Grant 1984; Millington and Price 1985; Grant and Grant 1989; Gibbs 1990), transmission is from father to son, sympatric species usually differ in their songs, and daughters mate with males having the same species song as their father's (Grant and Grant 1989).
Thus song, an evolving and culturally inherited trait, is an important factor in species recognition and mate choice. We discuss these findings in the context of evolution by considering the role of song in mate choice, speciation and hybridization.
Song and Mating Patterns
The songs of G. fortis and G. scandens on Isla Daphne Major differ conspicuously from each other and provide an unambiguous cue for females to choose conspecific mates. The differences between species appear to be maintained by vertical (cultural) transmission of song from father to son, a short sensitive period for song learning, and an imprinting mechanism that effectively constrains daughters to mate with males having the same species song as their fathers. The transmission of song from father to son is not completely error free [ILLUSTRATION FOR FIGS. 7, 8 OMITTED], and as such it resembles the inheritance of song in zebra finches (Bohner 1983, 1990; Zann 1990; Mann and Slater 1994). We suspect the same is true for the learning of father's song by the daughter. Occasional mis-imprinting results in a low frequency of hybridization (Grant and Grant, in press a).
Song, which is paternally inherited, can be thought of as a cultural analogue to maternally inherited mitochondrial DNA. As a phenotype it has the advantage over mtDNA of being usable as a detectable expression of familial (and phylogenetic) affinity. We suggest that females use it to choose a particular conspecific mate.
For example, not only do G. fortis females generally mate with conspecific males, and hence avoid extreme outbreeding, they appear to pair preferentially with males that possess song features somewhat different from their father's, and perhaps as a result they avoid extreme inbreeding (cf. Bateson 1982; Ten Cate and Bateson 1988). The most thorough analysis to date of inbreeding in this species did not find evidence of an avoidance of inbreeding (Gibbs and Grant 1989). But in a study of G. conirostris on Isla Genovesa, no female was found paired with a male singing a similar song subtype to her father's (Grant 1984; Grant and Grant 1989), and this observation is compatible with a hypothesis of inbreeding avoidance.
A father-son transmission of a short song that remains unaltered throughout life facilitates individual recognition of songs by females, and this could serve as a cue for the avoidance of inbreeding. Females that avoid mating with males of similar song to their father's would have a higher probability of not mating with kin than those mating with males with songs similar to their father's songs. On small islands, with restricted dispersal and fluctuating population sizes, the chance of mating with relatives is high. This has been demonstrated with a small population of G. magnirostris on Daphne (Grant and Grant 1995a; for G. fortis see Gibbs and Grant 1989). If inbreeding carries a reproductive cost, individuals who avoid it would be at an advantage. A cost has not been demonstrated in Darwin's finches, but data for G. magnirostris are suggestive of such a cost (Grant and Grant 1995a). Elsewhere a reproductive cost has been demonstrated for inbreeding pairs of passerine birds in the form of depressed hatching success (van Noordwijk and Scharloo 1981; Bensch et al. 1994; Kempenaers et al. 1996). The cost of inbreeding may arise later in life through the low survival of inbred birds subjected to environmental stress (Keller et al. 1994).
The Evolution of Song Variation in Darwin's Finches
In contrast to the wide variety of songs in the G. fortis population, songs sung by G. scandens males vary to a small extent. The difference between the species may be explained by differences in population size and chance. Population size of G. fortis fluctuated, but never dropped as low as the G. scandens population, which decreased to 16 mated males and remained low between 1984 and 1992 (Grant and Grant 1992b, 1993). Episodic population bottlenecks in the past may have restricted song heterogeneity in the G. scandens population, through the chance differential survival and breeding of a few individuals, a situation observed in the population of G. magnirostris on Isla Daphne (Grant and Grant 1995a). During our study, depletion of the variation has not been compensated by immigration of birds singing new types. Geospiza scandens immigrants have been observed on Daphne, but they have never stayed to breed (and sing).
The difference between the species on Daphne mirrors the situation on many islands in the Galapagos archipelago (Ratcliffe 1981; Bowman 1983). A large array of finch songs has been recorded, and just as we found on Daphne, some species vary much more than others (Bowman 1983). Populations of finches on large islands have a higher diversity of songs than on smaller islands. On most islands the songs of species differ from each other and transmit the species identity of the singer unambiguously. Song playback experiments carried out on several islands with different species of finches have shown strong discrimination between sympatric species (Ratcliffe and Grant 1985).
However, populations of two species on different islands may sing similar or nearly identical songs. For example, G. conirostris on Espanola have songs similar to G. magnirostris on Santa Cruz and Daphne, whereas G. conirostris on Genovesa have songs similar to G. scandens on Rabida and Daphne and to G. fortis on Santiago (Ratcliffe 1981). These song similarities may have reproductive consequences if the allopatric populations ever meet. Although G. conirostris on Espanola did not respond strongly to playback of the song of allopatric G. magnirostris, despite the structural similarity to their own song, G. conirostris on Genovesa did respond strongly to playback of the song of G. scandens from Daphne, and as strongly as they did to their own song (Ratcliffe and Grant 1985).
To explain the pattern of differences between allopatric populations of the same species and occasional similarity in the songs of allopatric species, we suggest the following. Small satellite islands are colonized by a few individuals from nearby larger islands, and the founding populations on satellite islands contain by chance different subsets of the songs found on the larger island (cf. Baker and Jenkins 1987). These songs then change over time, first as the result of small copying errors in transmission from father to son, and second through random extinction of song types, in other words by a process of cultural change analogous to genetic drift (e.g., see Mundinger 1980). A low level of song misimprinting combined with changes in ecological conditions that permit the survival of intermediate morphologies, as seen on Daphne after the El Nino event of 1982-1983 (B. R. Grant and P. R. Grant 1993, 1996), could result in episodes of hybridization with introgression. During such times songs of one species could become incorporated into the repertoire of another species, producing the observed situation of songs of one species on one island resembling the songs of another species on other islands.
Alternative hypotheses for similarities in the songs of different allopatric species are selection for convergence arising from similar sound-transmitting environments (Bowman 1983), and common inheritance. Both make specific predictions about the direction of change between populations that could be tested when a well supported phylogeny becomes available (see Stern and Grant 1996).
Song and Speciation
The allopatric model of speciation can account for the adaptive radiation of Darwin's finches in the Galapagos archipelago (P. R. Grant and B. R. Grant 1996). During allopatry the strength of the barrier to gene flow would be proportional to the distance between islands. In isolation, the sister taxa would accumulate differences in both morphology and song. On secondary contact the strength of the barrier to gene flow would be a function of the fidelity of the song imprinting process (Grant and Grant, in press a).
This last point deserves elaboration as imprinting is often thought of solely as an intraspecific phenomenon. Imprinting is restricted to species with parental care where learning to recognize parents occurs early in life. Information acquired in early life is consolidated during courtship (Bischoff and Clayton 1991; Immelmann et al. 1991; Kruijt and Meeuwissen 1991, 1993; Casey and Baker 1992; see also Domjan 1992). To the extent that both sexes are involved in mate choice (Wynn and Price 1993), both are affected by their early learning experiences, though in different ways (Ten Cate et al. 1993; Vos et al. 1993; Vos 1995). We interpret the process of sexual imprinting to be one of narrowing the range of possible mates, reducing or eliminating the chance of mating heterospecifically under normal circumstances. In this respect, sexual imprinting has both intraspecific (mating) and interspecific (avoidance) consequences. They are two sides of the same coin.
The consolidation phase of sexual imprinting in initial courtship may be especially important in learning not to respond to heterospecific stimuli at secondary contact. This is indicated by the tendency for sympatric, closely related, pairs of species to show little or no response to each other's songs whereas allopatric populations of the same species respond strongly to both conspecific and heterospecific songs (Gill and Murray 1972; Becker 1977; Lynch and Baker 1990; but see also Emlen et al. 1975; Salomon 1989; Baker and Baker 1990). An evolved difference in perceptual abilities between sympatric and allopatric populations is the alternative explanation. It is hardly likely to apply to those situations where the populations are separated by a few hundred meters (see Lynch and Baker 1990).
The strength of the discrimination between songs at maturity might be a consequential effect of selection on the length of the sensitive period in early life (Immelmann 1980; Baptista et al. 1993; Nottebohm 1993; Grant and Grant, in press a). An important factor affecting the length of the sensitive period is the duration of dependence of young birds on their parents. Young Darwin's finches are dependent on their parents for food for 20-30 d after fledging, rarely for longer (Grant 1984; Grant and Grant 1989). On hearing their father's call notes and song, they fly toward him begging. The use of song later in life to recognize conspecifics might be a secondary consequence of the survival value of learning to recognize father's song early in life. In other species the learning of a single song from father takes place over a shorter period of dependence: perhaps for one week and certainly no more than 19 d after fledging in the European marsh tit, P. palustris (Rost 1990).
Song and Hybridization
Geospiza fortis and G. scandens are unlikely to be sister taxa (Stern and Grant in press). Nevertheless the situation on Daphne is similar to that of secondary contact between recently formed sister species that differ in morphology and song. Hybridization between them, sometimes a consequence of a low incidence of misimprinting on heterospecific song, is rare (1-3%). From 1976 to 1982 no hybrid survived to breed. However, under altered ecological conditions following the El Nino event of 1982-1983, hybrids of intermediate bill and body size and shape survived and backcrossed to members of their parental species (Grant and Grant 1993). No genetic penalty to interbreeding was detected, as their survival and reproduction was as good as or better than the performance of nonhybrids born in the same year (Grant and Grant 1992a). Choice of a mate by the backcrossing hybrids was strongly influenced by paternal song (Grant and Grant, in press b).
The circumstantial evidence suggests that learning of song though imprinting, or an imprinting-like process (Baptista et al. 1993; Ten Cate et al. 1993), may be one factor among several influencing the likelihood that hybridization will occur (Grant and Grant, in press a). This is paradoxical, in that imprinting constrains courtship and mating to conspecifics. However, hybridization is possible when the normal imprinting process is perturbed, for example when the male dies and the female alone feeds the fledglings.
Hybridization occurs widely but nonuniformly among avian taxa (Grant and Grant 1992a). In some groups, song (vocalizations) may play no role in the hybridization, for example in those where plumage variation is pronounced and paternal care is lacking (Sibley 1961; Christidis and Schodde 1993: see also Pierotti and Annett 1993). However hybridization is relatively frequent among passerines, specifically the oscines, where song differences between species are ubiquitous and song is learned (Marler and Peters 1977; Marler 1987; Searcy et al. 1995). In contrast, the suboscines do not culturally transmit song (Kroodsma and Konishi 1991), and perhaps as a result they do not hybridize as often (Graves 1992). Likewise, doves do not culturally transmit song (Nottebohm and Nottebohm 1971; Baptista and Trail 1992), and the incidence of natural hybridization in this group is low also (Grant and Grant 1992a). Vocal learning occurs in two of the taxa that hybridize at a moderate to high frequency (e.g., parrots and hummingbirds; Baptista and Trail 1992).
The potential to hybridize in birds remains for more than 20 million yr on average since the time of separation of taxa (Prager and Wilson 1975). It is therefore not surprising that Darwin's finch species, although differing from each other in body mass, bill size and shape, and song, show no indication of genetic incompatibility after a period of less than 3 million yr since divergence from a common ancestor (P. R. Grant and B. R. Grant 1996). Prager and Wilson's (1975) results imply that prezygotic isolation arises considerably faster than postzygotic isolation, and that species may differ phenotypically as a result of selection and drift long before they are incapable of exchanging genetic material. This may be especially true of the passerines, whose radiation in approximately 60 million yr (Boles 1995) has given rise to more than half of the bird species alive today. Song is considered to have been one of several factors contributing to the evolution of passerine diversity (Raikow 1986; Baptista and Trail 1992). Our findings suggest that the song imprinting process may have been an important contributor by providing a leaky barrier to gene exchange on secondary contact.
The consequences of gene exchange can be evolutionarily significant. We estimated that the additive genetic variance introduced by hybridization of finches is two to three orders of magnitude greater than that introduced by mutation alone (Grant and Grant 1994). Evolutionary responses to selection would be enhanced as a result of introgression subject to constraints imposed by genetic correlations between characters. In G. fortis and G. fuliginosa with similar allometries, the effect of hybridization is to strengthen genetic correlations, thereby rendering evolution in a new direction less likely. On the other hand, in G. fortis and G. scandens with different allometries, the effect of hybridization is to enhance genetic variation but weaken some of the genetic correlations, thereby relaxing the genetic constraints on evolution in a new direction (Grant and Grant 1994, 1995b). Such episodes of introgression may have arisen many times in the evolution of the finches on the Galapagos, and may or may not have lasted long enough to break down the morphological and song identities of sympatric congeneric species. Thus introgression, at least partly caused by misimprinting on a cultural trait (song), has important implications for evolution.
We thank the Galapagos National Park Service and the Charles Darwin Research Station for permission to conduct the research and for logistic support; the National Science Foundation for continued support, most recently grant number DEB 93-06753; L. M. Ratcliffe and H. L. Gibbs for some tape recordings and sonagrams; J. A. Green and the many assistants who have helped us, and especially S. Nowicki and S. S. Peters for use of their song analysis equipment; and T. D. Price, and two reviewers for valuable comments and suggestions on the manuscript.
ALATALO, R. V., D. ERIKSSON, L. GUSTAFSSON, AND A. LUNDBERG. 1990. Hybridization between pied and collared flycatchers - Sexual selection and speciation theory. J. Evol. Biol. 3:375-389.
BAKER, A. J., AND P. F. JENKINS. 1987. Founder effect and cultural evolution of songs in an isolated population of chaffinches, Fringilla coelebs, in the Chatham Islands. Anita. Behav. 36:1793-1804.
BAKER, M. C. 1991. Response of male indigo and lazuli buntings and their hybrids to song playback in allopatric and sympatric populations. Behaviour 119:225-242.
-----. 1994. Does exposure to heterospecific males affect preferences of female buntings (Passerina)? Anim. Behav. 48:1349-1355.
BAKER, M. C., AND A. E. M. BAKER. 1990. Reproductive behavior of female buntings: Isolation mechanisms in a hybridizing pair of species. Evolution 44:332-338.
BAKER, M. C., AND P. MARLER. 1980. Behavioural adaptations that constrain the gene pool in vertebrates. Pp. 59-80 in H. Markl, ed. Evolution of social behaviour: Hypotheses and empirical tests. Proceedings of the Dahlem conference. Weinheim Verlag Chemie, Berlin, Germany.
BAPTISTA, L. F. 1990. Song learning in White-crowned Sparrows (Zonotrichia leucophrys): Sensitive phases and stimulus filtering revisited. Pp. 143-152 in R. van den Elzen, K.-L. Schuchmann, and K. Schmidt-Koenig, eds. Proceedings of the international centennial meeting of the Deutsche Ornithologen-Gesellschaft, Bonn 1988. Verlag der D. O.-G., Bonn, Germany.
BAPTISTA L. F., AND M. L. MORTON. 1988. Song learning in montane white-crowned sparrows: From whom and when? Anita. Behav. 36:1753-1764.
BAPTISTA, L. F., AND L. PETRINOVICH. 1986. Song development in the white-crowned sparrow: Social factors and sex differences. Anim. Behav. 34:1359-1371.
BAPTISTA L. F., AND P. W. TRAIL. 1992. The role of song in the evolution of passerine diversity. Syst. Biol. 41:242-247.
BAPTISTA L. F., D. A. BELL, AND P. W. TRAIL. 1993. Song learning and production in the white-crowned sparrow: Parallels with sexual imprinting. Neth. J. Zool. 43:17-33.
BATESON, P. 1982. Preference for cousins in Japanese quail. Nature 295:236-237.
BECKER, P. H. 1977. Verhalten auf lautausserungen der zwillingsart, interspezifische territorialitat und habitatanspruche von Winterund Sommergoldhanchen (Regulus regulus, R. ignicapillus). J. Ornithol. 118:233-260.
BENSCH, S., D. HASSELQUIST, AND T. VON SCHANTZ. 1994. Genetic similarity between parents predicts hatching failure: Nonincestuous inbreeding in the great reed warbler? Evolution 48:317-326.
BISCHOFF, H.-J., AND N. CLAYTON. 1991. Stabilization of sexual preferences by sexual experience in male zebra finches Taeniopygia guttata castanotis. Behaviour 118:144-155.
BOHNER, J. 1983. Song learning in the zebra finch (Taeniopygia guttata): Selectivity in the choice of tutor and accuracy of song copies. Anim. Behav. 31:231-237.
-----. 1990. Early acquisition of song in the zebra finch, Taeniopygia guttata. Anim. Behav. 39:369-374.
BOHNER, J., M. L. CHAIKEN, G. F. BALL, AND P. MARLER. 1990. Song acquisition in photosensitive and photorefractory male European starlings. Horm. Behav. 24:582-594.
BOLES, W. E. 1995. The world's oldest songbird. Nature 374:21-22.
BOWMAN, R. I. 1983. The evolution of song in Darwin's finches. Pp. 237-537 in R. I. Bowman, M. Berson, and A. E. Leviton, eds. Patterns of evolution in Galapagos organisms. Am. Assoc. Adv. Sci., Pac. Div., San Francisco, CA.
BUSH, G. L. 1993. A reaffirmation of Santa Rosalia, or why are there so many kinds of small animals? Pp. 229-245 in D. R. Lees and D. Edwards, eds. Evolutionary patterns and processes. Linn. Soc. Lond., London.
CASEY, R. M., AND M. C. BAKER. 1992. Early social tutoring influences female sexual responsiveness in white-crowned sparrows. Anim. Behav. 44:983-986.
CAVALLI-SFORZA, L. L., M. W. FELDMAN, K. H. CHEN, AND S. M. DORNBUSCH. 1982. Theory and observation in cultural transmission. Science 218:19-27.
CAVALLI-SFORZA, L. L., P. MENOZZI, AND A PIAZZA. 1994. The history and geography of human genes. Princeton Univ. Press, Princeton, NJ.
CHRISTIDIS, L., AND R. SCHODDE. 1993. Sexual selection for novel partners: A mechanism for accelerated morphological evolution in the birds-of-paradise (Paradisaeidae). Bull. Brit. Ornithol. Club 113:169-172.
DOMJAN, M. 1992. Adult learning and mate choice. Am. Zool. 32: 48-61.
EMLEN, S. T. 1972. An experimental analysis of the parameters of bird song eliciting species recognition. Behaviour 41:130-171.
EMLEN, S. T., J. D. RISING, AND W. L. THOMPSON. 1975. A behavioral and morphological study of sympatry in the indigo and lazuli buntings of the Great Plains. Wilson Bull. 87:145-179.
ERIKSSON, D., AND L. WALLIN. 1986. Male bird song attracts females: A field experiment. Behav. Ecol. Sociobiol. 19:297-299.
FALCONER, D. S. 1989. Introduction to quantitative genetics. 3d ed. Longman, Harlow, Essex, UK.
GELTER, H. 1987. Song differences between the pied flycatcher Ficedula hypoleuca, the collared flycatcher F. albicollis, and their hybrids. Ornis Scand. 18:205-215.
GIBBS, H. L. 1990. Cultural evolution of male song types in Darwin's Medium ground finches, Geospiza fortis. Anim. Behav. 39: 253-263.
GIBBS, H. L., AND P. R. GRANT. 1989. Inbreeding in a population of Darwin's Medium ground finches, Geospiza fortis. Evolution 43:1273-1284.
GILL, F. B., AND B. G. MURRAY JR. 1972. Discrimination behavior and hybridization of the blue-winged and golden-winged Warblers. Evolution 26:282-293.
GRANT, B. R. 1984. The significance of song variation in a population of Darwin's finches. Behaviour 89:90-116.
GRANT, B. R., AND P. R. GRANT. 1989. Evolutionary dynamics of a natural population. The large cactus finch of the Galapagos. Univ. of Chicago Press, Chicago.
-----. 1993. Evolution of Darwin's finches caused by a rare climatic event. Proc. R. Soc. Lond. B. 251:111-117.
-----. 1996. High survival of Darwin's finch hybrids: Effect of beak morphology and diets. Ecology 77:500-509.
GRANT, P. R. 1993. Hybridization of Darwin's finches on Isla Daphne Major, Galapagos. Philos. Trans. R. Soc. Lond. B. 340:127-139.
-----. 1994. Population variation and hybridization: Comparison of finches from two archipelagos. Evol. Ecol. 8:598-617.
GRANT, P. R., AND B. R. GRANT. 1992a. Hybridization of bird species. Science 256:193-197.
-----. 1992b. Demography and the genetically effective sizes of two populations of Darwin's finches. Ecology 73:766-784
-----. 1994. Phenotypic and genetic effects of hybridization in Darwin's finches. Evolution 48:297-316.
-----. 1995a. The founding of a new population of Darwin's finches. Evolution 49:229-240.
GRANT, P. R., AND B. R. GRANT. 1995b. Predicting microevolutionary responses to directional selection on heritable variation. Evolution 49:241-251.
-----. 1996. Speciation and hybridization in island birds. Philos. Trans. R. Soc. Lond. B. 351:765-772.
-----. In press a. Hybridization, sexual imprinting and mate choice. Am. Nat.
-----. In press b. Mating patterns of Darwin's finch hybrids determined by song and morphology. Biol. J. Linn. Soc.
GRAVES, G. R. 1992. Diagnosis of a hybrid antbird (Phlegopsis nigromaculata x Phlegopsis erythroptera) and the rarity of hybridization among suboscines. Proc. Biol. Soc. Wash. 105:834-840.
GREEN, J. A. 1992. The transmission of male song in Darwin's medium ground finch Geospiza fortis. Bachelor's thesis, Princeton Univ., Princeton, NJ.
GUTTINGER, H. R. 1979. The integration of learnt and genetically programmed behaviour: A study of hierarchical organization in songs of canaries, greenfinches and their hybrids. Z. Tierpsychol. 49:285-303.
IMMELMANN, K. 1975. Ecological significance of imprinting and early learning. Annu. Rev. Ecol. Syst. 6:15-37.
-----. 1980. Genetical constraints on early learning: A perspective from sexual imprinting in birds. Pp. 121-133 in J.R. Royce, ed. Theoretical advances in behavior genetics. Van Nijhoff, Amsterdam, The Netherlands.
IMMELMANN, K., AND S. J. SUOMI. 1981. Sensitive phases in development. Pp. 395-431 in K. Immelmann, G. W. Barlow, L. Petrinovich, and M. Main, eds. Behavioral development. Cambridge Univ. Press, Cambridge.
IMMELMANN, K., R. PROVE, R. LASSEK, AND H.-J. BISCHOFF. 1991. Influence of adult courtship experience on the development of sexual preferences in zebra finch males. Anim. Behav. 42:83-89.
KELLER, L. F., P. ARCESE, J. N. M. SMITH, W. M. HOCHACHKA, AND S. C. STEARNS. 1994. Selection against inbred song sparrows during a natural population bottleneck. Nature 372:356-357.
KEMPENAERS, B., F. ADRIAENSEN, A. J. VAN NOORDWIJK, AND A. A. DHONDT. 1996. Genetic similarity, inbreeding and hatching failure in blue tits: Are unhatched eggs infertile? Proc. R. Soc. Lond. B. 263:179-185.
KONISHI, M. 1985. Birdsong: From behavior to neurone. Annu. Rev. Neurosci. 8:125-170.
KROODSMA, D. E. 1978. Aspects of learning in the ontogeny of bird song: Where, from whom, when, how many, which and how accurately? Pp. 215-230 in G. Burkhardt and M. Bekoff, eds. Ontogeny of behavior. Garland, New York.
KROODSMA, D. E., AND M. KONISHI. 1991. A suboscine bird (eastern phoebe, Sayornis phoebe) develops normal song without auditory feedback. Anita. Behav. 42:477-487.
KROODSMA, D. E., AND R. PICKERT. 1980. Environmentally dependent sensitive periods for avian vocal learning. Nature 288: 477-479.
KRUIJT J. P., AND G. B. MEEUWISSEN. 1991. Sexual preferences of male zebra finches: Effects on early and adult experience. Animal Behav. 42:91-102.
-----. 1993. Consolidation and modification of sexual preferences in adult male zebra finches. Neth. J. Zool. 43:68-79.
LESSELLS C. M., AND P. T. BOAG. 1987. Unrepeatable repeatabilities: A common mistake. Auk 104:116-121.
LYNCH, A., AND A. J. BAKER. 1990. Increased vocal discrimination by learning in sympatry in two species of chaffinches. Behaviour 116:109-126
MANN N. I., AND P. J. B. SLATER. 1994. What causes young male zebra finches Taeniopygia guttata, to choose their father as song tutor? Anim. Behav. 47:671-677.
MARLER, P. 1960. Bird songs and mate selection. Pp. 348-367 in W. E. Lanyon and W. Tavolga, eds. Animal sounds and communication. Am. Inst. Behav. Sci., Washington, DC.
-----. 1987. Sensitive periods and the role of specific and general sensory stimulation in birdsong learning. Pp. 99-135 in J. P. Rauschecker and P. Marler, eds. Imprinting and cortical plasticity: Comparative aspects of sensitive periods. Wiley, New York.
MARLER, P., AND D. ISAAC. 1960. Physical analysis of a simple bird song as exemplified by the chipping sparrow. Condor 62: 124-135.
MARLER, P., AND S. S. PETERS. 1977. Selective vocal learning in a sparrow. Science 198:519-521.
-----. 1988a. Sensitive period for song acquisition from tape recordings and live tutors in the swamp sparrow, Melospiza georgiana. Ethology 77:76-84.
-----. 1988b. The role of song phonology and syntax in vocal learning preference in the song sparrow Melospiza melodia. Ethology 77:125-149.
MILLINGTON, S., AND T. PRICE. 1985. Song inheritance and mating patterns in Darwin's finches. Auk 102:342-346.
MUNDINGER, P. C. 1980. Animal cultures and a general theory of cultural evolution. Ethol. Sociobiol. 1:183-223.
NAUGLER, C. T., AND L. RATCLIFFE. 1994. Character release in bird song: A test of the acoustic competition hypothesis using American tree sparrows Spizella arborea. J. Avian Biol. 25:142-148.
NELSON, D. A. 1989. The importance of invariant and distinctive features in species recognition of bird song. Condor 91:120-130.
NELSON, D. A., AND P. MARLER. 1993. Innate recognition of song in white-crowned sparrows: A role in selective vocal learning? Anim. Behav. 46:806-808.
NELSON, D. A., P. MARLER, AND A. PALLERONI. 1995. A comparative approach to vocal learning: Intraspecific variation in the learning process. Anim. Behav. 50:83-97.
NICOLAI, J. 1959. Famlientradition in der Gesangsentwicklung des Gimpels (Pyrrhula pyrrhula L.). J. Ornithol. 100:39-46.
NOTTEBOHM, F. 1972. The origins of vocal learning. Am. Nat. 106: 116-140
-----. 1993. The search for neural mechanisms that define the sensitive period for song learning in birds. Neth. J. Zool. 43: 193-234.
NOTTEBOHM, F., AND M. E. NOTTEBOHM. 1971. Vocalizations and breeding behaviour of surgically deafened ring doves. Anim. Behav. 19:313-327.
-----. 1978. Relationship between song repertoire and age in the canary Serinus canaria. Z. Tierpsychol. 46:298-305.
NOWICKI, S., AND D. A. NELSON. 1990. Defining natural categories in acoustic signals: Comparisons of three methods applied to chickadee call notes. Ethology 86:89-101.
PAYNE, R. B. 1981. Song learning and social interactions in indigo buntings. Anim. Behav. 29:688-697.
PAYNE, R. B., AND L. L. PAYNE. 1993. Song copying and cultural transmission in indigo buntings. Anim. Behav. 46:1045-1065.
PAYNE, R. B., W. L. THOMPSON, K. L. FIALA, AND L. L. SWEANY. 1981. Local song traditions in indigo buntings: Cultural transmission of behavior patterns across generations. Behaviour 77: 199-201.
PAYNE, R. B., L. L. PAYNE, AND S. M. DOEHLERT. 1987. Song, mate choice and the question of kin recognition in a migratory songbird. Anim. Behav. 35:35-47.
PIEROTTI, R., AND C. A. ANNETT. 1993. Hybridization and male parental care. Condor 95:670-679.
PRAGER, E. R., AND A. C. WILSON. 1975. Slow evolutionary loss of the potential for interspecific hybridization in birds: A manifestation of slow regulatory evolution. Proc. Nat. Acad. Sci. USA 72:200-204.
PRICE, T. D., AND H. L. GIBBS. 1986. Brood division in Darwin's finches. Anita. Behav. 35:299-300.
RAIKOW, R. J. 1986. The role of song in the evolution of passerine diversity. Syst. Biol. 41:242-247.
RATCLIFFE, L. M. 1981. Species recognition in Darwin's ground finches (Geospiza Gould). Ph.D. diss., McGill University, Montreal, PQ, Canada.
RATCLIFFE, L. M., AND P. R. GRANT. 1985. Species recognition in Darwin's Finches (Geospiza, Gould). III. Male response to playback of different song types, dialects and heterospecific songs. Anim. Behav. 33:290-307.
ROST, R. 1990. Song dialects of the marsh tit (Parus palustris) and their functional significance: A test of models. Pp. 111-122 in R. van den Elzen, K.-L. Schuchmann and K. Schmidt-Koenig, eds. Proceedings of the International Centennial Meeting of the Deutsche Ornithologen-Gesellschaft, Bonn 1988. Verlag der D. O.-G., Bonn, Germany.
RYAN, P. G., C. L. MOLONEY, AND J. HUDON, 1994. Color variation and hybridization among Nesospiza buntings on Inaccessible Island, Tristan da Cunha. Auk 111:314-327.
SALOMON, M. 1989. Song as a possible reproductive isolating mechanism between two parapatric forms. The case of the chiff-chaffs Phylloscopus c. collybita and P. c. brehmi in the western Pyrenees. Behaviour 111:270-290.
SEARCY, W. A. 1990. Species recognition of songs by female red-winged blackbirds. Anim. Behav. 40:1119-1127.
SEARCY, W. A., AND E. A. BRENOWITZ. 1988. Sexual differences in species recognition of avian song. Nature 332:152-154.
SEARCY, W. A., J. PODOS, S. PETERS, AND S. NOWICKI. 1995. Discrimination of song types and variants in song sparrows. Anim. Behav. 49:1219-1226.
SIBLEY, C. G. 1961. Hybridization and isolating mechanisms. Pp. 69-88 in W. F. Blair, ed. Vertebrate speciation. Univ. of Texas Press, Austin.
STERN, D., AND P. R. GRANT. 1996. A phylogenetic reanalysis of allozyme variation among populations of Galapagos finches. Zool. J. Linn. Soc. 118:119-134.
TEN CATE, C. R., AND P. P. G. BATESON. 1988. Sexual selection: The evolution of conspicuous characteristics of birds by means of imprinting. Evolution 42:1355-1358.
TEN CATE, C., D. R. Vos, AND N. MANN. 1993. Sexual imprinting and song learning: Two of one kind? Neth. J. Zool. 43:34-45.
THIELCKE, G. 1962. Versuche Tit klangatrappen zur klarung der verwandtschaft der Baumlaufer Certhia familiaris L., C. brachydactyla Brehm und C. americana Bonaparte. J. Ornithol. 103: 266-271.
THIELCKE, G., K. WUSTENBERG, UND P. H. BECKER. 1978. Reaktionen von Zilpzalp und Fitis (Phylloscopus collybita, P. trochilus) auf verschiedene gesangformen des Zilpzalps. J. Ornithol. 119:213-226.
THORPE, W. H. 1958. The learning of song patterns by birds, with special reference to the song of the chaffinch, Fringilla coelebs. Ibis 100:535-570.
VAN NOORDWIJK, A. J., AND W. SCHARLOO. 1981. Inbreeding in an island population of the great tit. Evolution 35:674-678.
VOS, D. R. 1995. Sexual imprinting in Zebra-finch females: Do females develop a preference for males that look like their father? Ethology 99:252-262.
VOS, D. R., J. PRIJS, AND C. TEN CATE. 1993. Sexual imprinting in zebra finch males: A differential effect of successive and simultaneous experience with two colour morphs. Behaviour 26: 137-154.
WALLIN, L. 1987. Integration of a call into the song of the collared flycatcher: Adaptive compensation for broadcast efficiency? Ornis Scand. 18:1205-1215.
WYNN, S., AND T. PRICE. 1993. Male and female choice in zebra finches. Auk 110:635-638.
YASUKAWA, K., J. L. BLANK, AND C.B. PATTERSON. 1980. Song repertoires and sexual selection in the red-winged blackbird. Behav. Ecol. Sociobiol. 7:233-238.
ZANN, R. 1990. Song and call learning in wild zebra finches in south-east Australia. Anim. Behav. 40:811-828.
|Printer friendly Cite/link Email Feedback|
|Author:||Grant, B. Rosemary; Grant, Peter R.|
|Date:||Dec 1, 1996|
|Previous Article:||Ecological diversification and community structure in the Old World leaf warblers (genus Phylloscopus): a phylogenetic perspective.|
|Next Article:||Competition between segregation distorters: coexistence of "superior" and "inferior" haplotypes at the t complex.|