E VOtLUT1ION1 PUBLISHED BY THE SOCIETY FOR THE STUDY OF EVOLUTION

Transcription

1 E VOtLUT1ION1 INTERNATIONAL JOURNAL OF ORGANIC EVOLUTION PUBLISHED BY THE SOCIETY FOR THE STUDY OF EVOLUTION Vol. 49 April, 1995 No. 2 Evolution, 49(2), 1995, pp THE FOUNDING OF A NEW POPULATION OF DARWIN'S FINCHES PETER R. GRANT AND B. ROSEMARY GRANT Department of Ecology and Evolutionary Biology, Princeton University, Princeton, New Jersey PRGrant@PUCC.Princeton.edu Abstract.-We report the natural colonization of the small Galapagos island Daphne Major by the large ground finch (Geospiza magnirostris). Immigrants of this species were present in every year of a 22-yr study, Typically they arrived after a breeding season and left at the beginning of the next one. Geospiza magnirostris bred on the island for the first time in the exceptionally wet El Niino year of , and bred in all subsequent years except drought years. In agreement with theoretical expectations the frequency of inbreeding was unusually high. Pronounced fluctuating asymmetry in tarsus length, together with slightly reduced breeding success of inbreeding pairs, suggests a low level of inbreeding depression. Despite this, the population increased from 5 breeding individuals in 1983 to 20 breeding individuals in 1992, and probably more than twice that number in 1993, largely through recruitment of locally born birds. The study illustrates the joint role of chance and determinism in colonization. The original colonizers were apparently a nonrandom sample with respect to morphological traits, and they and their offspring differed significantly in bill size from the immigrants that did not stay to breed. Because the traits appear to be heritable, the colonists were a genetically nonrandom sample. Genetic drift may have occurred, as only 6 of 13 founders produced recruits and small nonrandom tendencies in the colonists were amplified in the next two generations. An analogous process affected a culturally inherited trait, song type. This changed radically in frequency, apparently for reasons unconnected with properties of the song type itself; males singing one song type had better fledging and recruitment success than those singing another. The addition of a fourth species to a community of three is not expected on simple biogeographical grounds. It owes more to repeated immigration and the unusual but unidentified conditions favoring colonization than to a change in food supply on Daphne. Colonization and subsequent immigration may be the model pattern of founder events, applicable to continental and many insular situations. Key words.-biogeography, colonization, cultural drift, El Niflo, exponential increase, finches, founder effects, inbreeding, selection. Received December 10, Accepted May 23, The founding of a new population by a small number of individuals sets the stage for evolution by drift and selection. The founders contain reduced allelic diversity and unrepresentative frequencies of alleles. Further loss will occur in the first few generations as a result of inbreeding and the exposure of deleterious recessive alleles to selective elimination. On rare occasions, this may lead to substantial changes in the system of interacting genes, altered genetic correlations, and the production of phenotypes differing markedly from those of the parental population (Mayr 1954, 1992; Carson and Templeton 1984; Gray 1986). Although there is general agreement on the core of these ideas (e.g., see Provine 1989, Mayr 1992), opinions differ on whether founder events ever lead to much loss of genetic variation (Lande 1980), or to major genetic reorganizations (Barton and Charlesworth 1984; Charlesworth and Rouhani 1988) and whether populations persist long enough at low density for strong inbreeding effects to occur (Parsons 1989). Additive genetic variance may actually increase rather than decrease as a result of conversion of a portion of the nonadditive variance into additive variance (Bryant et al. 1986; Goodnight 1987, 1988; Bryant and Meffert 1993). However, effects of this conversion on the evolutionary potential of the population may be nullified by inbreeding depression (Willis and Orr 1993). Disagreements are largely a matter of genetic detail, but partly a matter of what happens in nature. With regard to the latter, ignorance prevails. For example, Charlesworth and Charlesworth (1987), reviewing empirical evidence from Drosophila and plants on the effects of inbreeding, comment that evidence from natural populations of birds and mammals is scarce and much needed. In fact, this applies to all animals. Similarly, there is little empirical data on inbreeding depression in natural populations as it might influence the prob- 229 C) 1995 The Society for the Study of Evolution. All rights reserved.

2 230 P. R. GRANT AND B. R. GRANT ability of extinction (Boyce 1992). The few data on inbreeding (e.g., Ralls et al. 1986) come from established, not recently founded, populations. The causes and consequences of colonization as a natural process are best known from studies of areas that have been partly destroyed (Diamond 1974, Thornton et al. 1990, Bush and Whittaker 1991) or created (Fridriksson 1975) by volcanic activity and other physical agents. Much less is known from observations of undisturbed or seminatural areas where communities are at equilibrial or quasiequilibrial states, because investigators are seldom in the fortunate position of being able to witness the generally rare and unpredictable process from the outset. Outstanding exceptions are demographic studies of cattle egrets (Bubulcus ibis) after they colonized Florida in 1948 (Sprunt 1953; Bock and Lepthien 1976), and barnacle geese (Branta leucopsis) after they colonized small islands in the Baltic in 1971 (Larsson et al. 1988). Both populations were shown to increase exponentially. Even so, factors affecting these colonizations in the earliest stages were not well known. Here we report the natural colonization of the small Galapagos island Daphne Major by the large ground finch (Geospiza magnirostris), an event that occurred in the middle of a 22-yr study of finches on this island. Our goal is to describe and interpret the phenomenon, with a focus on inbreeding and its possible consequences. By marking individuals uniquely and observing their mating patterns and reproductive success, we are able to show, in agreement with theoretical expectation, that inbreeding occurred. Despite signs of inbreeding depression, however, the breeding population increased in size exponentially. We conclude by discussing the roles of ecological and genetic factors in the establishment of new populations after immigration has taken place. We draw attention to the difference between the situation we studied and the situation usually modeled to explore the genetic consequences of founder events. The island is not remote, and there is no shortage of immigrants. The colonization and subsequent immigration that we document may be the model pattern of founder events, applicable to numerous continental and some island circumstances. An interplay of depleting and enhancing factors will determine the genetic consequences of such founder events. In contrast depleting factors will predominate in the unique colonization episodes typical of remote oceanic islands, for example, perhaps in the original colonization of the Galapagos archipelago by the ancestors of modern Darwin's finches. To place the colonization in perspective, we briefly summarize pertinent knowledge of Darwin's finch species elsewhere in the Galapagos archipelago. The distribution of species became known almost 100 yr ago through the collection of specimens for museums (Rothschild and Hartert 1899, 1902; Snodgrass and Heller 1904; Swarth 1931), and the observations of breeding activity made by the collectors (e.g., Gifford 1919). Modern investigations have confirmed the breeding status of almost all populations previously identified as breeding and demonstrated that nonbreeding individuals are capable of frequent movement between islands (Lack 1945; Bowman 1961; Harris 1973; Grant 1986; Grant and Grant 1989). A few populations have become extinct, all apparently in association with human activity (Grant 1986). Although very small populations may have become extinct naturally and then become reestablished through subsequent invasion and colonization (Grant and Grant 1992), no new breeding population has been reported from anywhere in the archipelago. These facts support the idea that communities of finches are stable under natural conditions, despite the potential for new colonizations to occur. This seems to be typical of tropical birds (Mayer and Chipley 1992). Thus, our documentation of a new colonization has special significance as a rarely observed event that can throw ecological and evolutionary light on the early stages of population establishment. LOCATION, MATERIALS, AND METHODS Daphne Major is a small island (0.34 km2, elev. 120 m) in the center of the Galapagos archipelago. Unlike most of the other islands in the archipelago, it has never been disturbed by human settlement or exploitation; the habitat is intact, and no organisms have been introduced. The closest islands are Santa Cruz (904 km2), Baltra (25 km2), Seymour (1.84 km2), and Daphne Minor (0.08 km2), all about 8 km away (Wiggins and Porter 1971). The next closest large island, Santiago (San Salvador: 572 km2), is 22 km from Daphne. Prevailing winds tend to be from the south and east (Boag and Grant 1984a). Two species of finches breed at high density; Geospiza fortis (medium ground finch: 17 g) and G. scandens (cactus finch: -21 g). Two others are less common; G. fuliginosa (small ground finch: 13 g) and G. magnirostris (large ground finch: -30 g). Doves (Nesopelia galapagoensis) and martins (Progne galapagoensis) breed regularly on the island, though in small numbers, and yellow warblers (Dendroica petechia) do so occasionally (Boag and Grant 1984b). Finches have been studied every year from 1973 to Methods have been described in detail elsewhere (Boag and Grant 1984a,b; Gibbs and Grant 1987) and will be only summarized here. Breeding was studied throughout the breeding season in the years and in the first half (January- March) of the 1992 and 1993 seasons. Additional breeding information was obtained in 1974 (February-March) and 1994 (January-February). In , all nests were found on two-thirds of the island, and observations were made on breeding activity in the remaining area. Almost all nests were found throughout the island in subsequent years (Grant and Grant 1992), up to March Visits to a nest were patterned to determine completed clutch size, hatching success, and fledging success (Boag and Grant 1984b). Chicks were banded usually on day 8 (range 6-10), counting hatching day as day 0. A uniquely numbered metal leg band was used in conjunction with three plastic color bands coded to correspond to the number. Adults and immatures were captured in mist nets and banded if not banded already. Experience on Genovesa had shown that G. magnirostris- individuals, if banded after fledging, were capable of squeezing the metal band onto the leg; thus, for this species, all birds captured in nets were banded solely with plastic bands. Even some of those banded as nestlings attempted to remove their bands; we therefore discontinued using the metal band on them in 1992.

3 FOUNDER POPULATION OF DARWIN'S FINCHES 231 G.magnirostrls U)G.fullglnosa LU LU Z 2 7 Z w ~~~~~~~~~~~~~~~~~~~~~LU 3- a a~~~~~~~~~~~~~ z Z 1 0- I- x~~~~~~~~~~~ LU LU z 10 YEARS YEARS w w~~~~~~~~~~~~~~ N 10- N~~~~~~~~~~~~~ ar 23G.manrotrs 37.fuignsa 13 GJafrtis, Gnds34a.ncanens YEARS YEARS FIG. 1. Annual captures of the four Geospiza species shown as numbers standardized by number of days of netting. Two-four nets were used each day for a total of approximately 8 net-hours per day (Smith et al. 1978). Annual netting effort varied from a minimum of 10 d in 1990 to a maximum of 40 d in 1989 (X? = 24.6? SD). Captures of previously banded birds are' not included. Numbers are 243 G. magnirostris, 317 G. fuliginosa, 1037 G. fortis, and 334 G. scandens. Captured birds were weighed to the nearest 0.1g with a Pesola? balance, and five dimensions were measured in millimeters: length of wing and tarsus, and length, depth and width of the bill. One of us (P.R.G.) measured 99 birds, and correction factors, calculated from birds measured by him and other measurers, were applied to an additional 166. Mean differences between measurers were small, and repeatability of measurements by different measurers was high (for comparable examples, see Price and Grant 1984). Songs were recorded with a Sony tape recorder (models TCM-5000 and TC D5PRO II) and a Sennheiser microphone (model K3U), and analyzed by means of a Kay Sonagraph (model 6061A) or Kay Elemetrics model 5500 and a Krone-Hite model 3550 filter with high (7.0 khz) and low (1.0 khz) settings. Sonagrams were printed with a Kay Gray Scale Printer (model 5509). Rainfall was measured daily in a rain gauge throughout each wet season. In March 1992, a permanent rain gauge was installed to measure the accumulation in our absence. RESULTS Colonization and Population Dynamics Immigration.-Before our study, Geospiza magnirostris had been reported on Daphne only once, in Beebe (1924) observed a pair that he thought was breeding and collected the male. They were not reported by Lack (1945, 1947), even though his colleague L. S. V. Venables had visited the island in January 1939, nor by Harris (1973) who visited the island in the 1960s and early 1970s and summarized the observations of other visitors as well as his own. Geospiza magnirostris individuals were present on the island in every year from 1973 to Their numbers varied from fewer than 10 to more than 50. A total of 243 immigrants were captured and banded in the period Others were seen but not captured in each of the years Given its proximity, the source island is probably Santa Cruz. Annual variation in the abundance of immigrants is reflected in the pattern of captures in mist nets (fig. 1). The figure shows the captures of each of the four Geospiza species standardized by a measure of trapping effort. Note that immigrant G. magnirostris became the dominant component of the captures when the captures of new individuals of the two resident species G. fortis and G. scandens declined to zero as a result of an intensive program of banding nestlings. Most immigrants were probably young birds born (hatched) in the same or previous year, because more than 90% of them were in immature plumage. The majority disappeared within a year, often at the start of the wet season when breeding began. A few stayed for 2 yr or more when conditions were too dry for breeding. Annual variation in captures of immigrants reflects variation in the production of young, which in turn is governed by the amount of rainfall: in general, the larger the amount of rain in a year the greater is the number of fledglings produced (Gibbs and Grant 1987; Grant and Grant 1989). This applies to all islands simultaneously: rainfall totals are correlated strongly among islands (Grant and Boag 1980). Number of captures of G. magnirostris immigrants (fig. 1) was not correlated with rainfall (fig. 2) in the year of capture on Daphne (r = 0.026, P > 0.1, N = 17 yr), but was correlated with the rainfall in the preceding year (r = 0.518, P < 0.05, N = 16) when, presumably, they were born. Geospiza fuli-

4 232 P. R. GRANT AND B. R. GRANT S L TABLE 1. The breeding of Geospiza magnirostris on Island Daphne Major. Males Females Pairs Nests Eggs Nestlings Fledglings ( ????) Total YEAR FIG. 2. Annual rainfall totals recorded on Isla Daphne Major. Rain recorded in November and December 1982 has been added to the 1983 totals. No rain was recorded in those months in other years. four were not known to be recruits (initially they were not ginosa, another immigrant species, shows a similar association with rainfall, except that immigration has been extreme- banded) and therefore were probably immigrants because all nests were found. These probable immigrants bred in the ly rare since 1986 (fig. 1). years Thus, by the end of the continuous study Colonization.-Colonization occurs when immigrants stay in March 1992, a majority of those that bred had been born to breed. None of the G. magnirostris immigrants bred in the on the island; after this time, we could not distinguish immigrants from locally born birds among those without bands. first part of the study, in the years , although a male had been seen nest building in 1974 (Grant et al. 1975), An important factor in the rapid increase was the occurrence of successive wet years (fig. 2). A large production of and others had been heard singing in (T. D. Price pers. comm. 1981). Breeding first occurred in fledglings in 1991 (table 1) was translated into high recruitment in 1992, when six males and six females hatched in the (Gibbs and Grant 1987), when three males and two females formed three pairs that produced a total of 17 fledglings from previous year bred. Four other members of that cohort also several broods (table 1). were present as nonbreeders. They bred in In contrast, Rain normally falls in the first few months of the year, and moderately large production of fledglings in 1983 and 1987 breeding occurs at that time (Grant and Boag 1980), but an was not translated into high recruitment because succeeding exceptional El Nifno event brought rains to the islands in years were dry or droughts (fig. 2). November 1982, and the rains persisted until July 1983 Most G. magnirostris individuals born on the island did (Gibbs and Grant 1987). Breeding began with little delay. not survive their first year. The largest sample was the 1991 The first G. magnirostris egg was laid on December 15, 1982, cohort (N = 34). Only 16 (44%) survived to 1992; nevertheless 15 of these survived to the following year, and 13 after approximately half (43%) of the G. fortis had begun breeding. Geospiza magnirostris bred in every subsequent year when the resident species bred; that is, in every year (38%) were alive in The 1991 cohort of G. fortis (N except the drought years of 1985, 1988, and 1989 (see fig. = 583) displayed an identical survival (38%) over the 3-yr 2). period. All of the original colonists had disappeared after 4 Demography.-The size of the breeding population remained approximately the same for several years at 1 to 6 yr, but two of the birds born in the first year of breeding were still alive, and breeding, in 1993 at the age of 10 yr, and one pairs, and then increased substantially in 1992 to 10 pairs was alive in Small samples preclude a quantitative (table 1), and then again in 1993 to 23 pairs (minimum). At comparison with the other finch species, which are capable the beginning of the breeding season in 1994, 41 territorial of living up to 15 yr (Grant and Grant 1992). singing males were observed, and many of them were paired. In reproduction, G. magnirostris were not as successful as For the 6 yr in which rain was sufficient to induce a full G. fortis and G. scandens. Overall success, measured as the breeding response (1983, 1984, 1987, ), growth proportion of eggs that yielded fledglings, was approximately in the population of breeding birds is described by the equation Nt = 0.75e0 46t. The exponent 0.46? 0.08 (from linear are in the range 60%-70% (Boag and Grant 1984b; Gibbs 50% (see table 1). Typical values for the other two species regression, P = ) is twice as large as the one (0.21) and Grant 1987; Grant and Grant 1992). We do not attempt calculated by Bock and Lepthien (1976) for the rate of increase in the cattle egret population in the United States. It would entail partitioning an already small sample (G. mag- a more quantitative analysis of the three species because this is almost as large as the exponent (0.49? 0.03) we calculate nirostris) to correct for annual variation (see Gibbs and Grant for the barnacle goose population on Swedish islands in the 1987). One male attempted to breed with a G. fortis female Baltic from data published by Larsson et al. (1988). Augmentation of the number of breeders can be brought about by repeated immigration or by recruitment of locally born birds. Both probably occurred, although recruitment was the more important factor in this colonization. After the initial colonization an additional 13 males and 13 females bred (eggs laid) before For each sex, nine were recruits and in 1983 but the nest was abandoned before the eggs were

5 FOUNDER POPULATION OF DARWIN'S FINCHES 233 TABLE 2. Mean measurements of immigrant Geospiza magnirostris that did not breed (noncolonists), those that did (colonists) and their offspring. Weight is in grams and the other traits are in millimeters. Principal-component (PC) 1 and PC 2 are from a PC analysis of all morphological measurements, based upon the correlation matrix of the combined sample. PC 1 is a general size factor (58.9% of variance explained), with all traits loading strongly and approximately equally. PC 2 is a shape factor (additional 13.9% of variance explained), with strongest loadings from wing (0.627) and tarsus length (0.516); all three bill dimensions load negatively ( to ). Noncolonists Colonists Offspring Jx SD SD SD Weight Wing length Tarsus length Bill length Bill depth Bill width PC PC N due to hatch (Grant 1993). Another was incorrectly reported to have hybridized with G. fuliginosa (Grant 1986, 1993). Characteristics of the Colonists and Their Offspring Mean Morphology.-Two of the five original colonists were captured and measured. Four more birds captured at the same time did not stay to breed. The two groups may be compared to see if colonization was random with respect to measured morphological traits. The samples are very small, however, and none of the trait means differed (t-tests with ln-transformed data, P > 0.1). As an alternative, the test was repeated by comparing all 238 adult immigrants captured before 1993, which did not stay to breed with the five birds known or believed to be colonists; that is, with the two initial colonists and the three measured birds among the eight not known to be recruits that bred before Table 2 gives mean measurements of the two groups and of the offspring of the colonists. Colonization appears to have been random with respect to morphology by this test. Colonists were larger than noncolonists in all means, although a MANOVA performed on the two groups was not TABLE 3. Repeatability of measurements and two estimates of heritabilities. Repeatabilities and sib-sib correlations were calculated as intraclass correlation coefficients (t) from one-way ANO- VA. Parent/offspring heritabilities were calculated as slopes (b) of regressions of midoffspring on midparent values; all offspring were at least 8 mo old when measured. Sample sizes refer to the number of birds followed by number of measurements for the repeatabilities and number of families followed by number of offspring for the heritabilities. Heritability Repeatability Parent/offspring Sib-sib Weight * Wing length Tarsus length 0.745** 0.808* Bill length 0.842*** * Bill depth 0.565* * Bill width 0.958*** ** x Sample sizes 13, 27 5, 12 7, 21 * P < 0.05; ** P < 0.005; *** P < significant (Wilks' X = 0.970; F = 1.186; df = 6, 234; P = ), nor were any of the individual trait comparisons. In a separate analysis, overall size, as indexed by principalcomponent 1 scores (table 2), was larger, on average, among the colonists than among the noncolonists, but not to a significant extent (t239 = 1.526, P = ). Sexes of the noncolonists were generally not known. Despite these negative findings, the bird with the largest bill in all dimensions among the total sample of 243 was a presumed colonist, and the individual with the next longest and widest bill was one of the two original measured colonists. The two extremely large birds (both males) in the small sagiple of colonists were responsible for large variances. The five colonists were more variable than the 238 noncolonists in weight and the three bill dimensions (F tests, P < 0.01 in each case); for the other two traits P > 0.1. They were also more variable than the 22 offspring in exactly the same four traits (again P < 0.01 in each case). With one exception, these results were confirmed by Levene's tests of deviations from the medians. The exception was the lack of difference in weight variation between colonists and noncolonists (P > 0.1). Heritable Variation in Morphology.-Offspring of the colonists were, on average, intermediate between the size of the colonists and the size of the noncolonists in three of six measured traits (table 2). They resembled the colonists (the parental group) most closely in bill dimensions. Resemblance between offspring and parents reflects some combination of shared genes and shared environments experienced during growth to maturity (Boag and van Noordwijk 1987). In the present case, locations during growth of offspring and parents are known to be different in most families, but there was no systematic tendency for offspring to be larger or smaller than their parents and therefore no indication that Daphne is a particularly favorable or unfavorable environment for the growth of G. magnirostris. The heritability of G. magnirostris morphology was estimated from an analysis of a small number of families in two ways (table 3). The slope coefficients of the regression of mid-offspring values on mid-parent values average 0.40 for the six traits. Sib-sib correlations vary less but average almost

6 234 P. R. GRANT AND B. R. GRANT TABLE 4. The frequency of inbreeding. The average coefficient of coancestry of breeding pairs is compared with that expected from random pairing of potential breeders. For statistical purposes, the data for each year were placed in a 2 X 2 contingency table, and related and unrelated pairs compared between observed and notobserved pairs with Fisher's exact tests (1987, 1991, and 1992adj) and a x2 test (1992). Related pairs comprise both classes + = 0.25 and + = One potential but unrealized pair in 1992 with + = was treated as an unrelated pair. The 1992 data were adjusted (1992adj) by eliminating three long-lived pairs to show their effect upon the mean coefficient of co-ancestry. Pairs Mean coefficients P Related Unrelated Observed Expected adj FIG. 3. The complete genealogy for Geospiza magnirostris in the years The initial breeders are shown at the top, together with all the fledglings they produced; three of the fledglings became recruits. Below that, only the fledglings that became recruits are shown. Squares, males; circles, females; diamonds, birds of unknown sex. Solid symbols represent banded birds; open symbols, birds without bands. Double horizontal lines, inbreeding pairs. Broken lines, new pairings in 1993 with unknown reproductive outcome. the same. Given the low repeatability of weights and wing lengths, heritability estimates for tarsus and bill measurements should be considered the more reliable. Averages of these estimates alone from parent regressions (0.59) and sib correlations (0.49) are similar to estimates from regression analyses of much larger samples of G. fortis (0.58) and G. scandens (0.54) obtained over approximately the same period of time (Grant and Grant 1994; see also Boag 1983). Misidentified parentage is unlikely to be a complicating factor in these comparisons as extra-pair copulations have been observed rarely in ground finch species (Grant and Grant 1995), and never in G. magnirostris. As a consequence of the similarity in size of offspring and parents, the resident population, comprising colonists and their offspring (N = 27), were significantly larger than the noncolonists in bill length (t263 = 2.437, P = ), bill depth (t263 = 2.284, P = ), bill width (t261 = 2.887, P = ), and in PC 2 (a measure of bill shape in relation to body size; t261 = 4.055, P = ), but not in weight, wing length, tarsus length, or PC 1 (all P values > 0.05). The offspring alone differed from the noncolonists in the same ways except for bill length (t258 = 1.609, P = ) and tarsus length (t258 = 2.365, P = ). Genetic Structure and Inbreeding Genetic Structure.-The full genealogy of the population is displayed in figure 3. It shows that offspring from only one of the three original pairs survived to breed, that these bred among themselves (and not with additional immigrants), and that both inbreeding and outbreeding took place in the following generation. Altogether, four different full-sib pairs were formed, as well as two half-sib pairs and one type of between-generation related pair. An additional five pairs of birds not known to be related bred but did not produce any recruits. Only six colonists produced recruits. Inbreeding.-A high frequency of inbreeding is expected in the establishment of a new population by a small number of individuals (Crow and Kimura 1970). Nevertheless, it may be less frequent than expected from random pairing if there is a tendency for birds to avoid mating with their relatives. We use the coefficient of co-ancestry (Falconer 1989) as a measure of the relatedness of breeding pairs in an analysis of the frequency of inbreeding. The mean coefficient for observed pairs is compared with the mean coefficient expected on the basis of an equal probability of all potential pairs among birds that bred. Thus, unpaired birds are assumed to be not available as mates (physiologically not ready to breed, incompetent, and so on). The results given in table 4 show that inbreeding is not avoided. In fact in three of the four years, the observed mean coefficient is higher than expected, not lower, and in one of them, the difference is statistically significant. The difference arose because full-sib pairs have tended to remain as pairs, whereas unrelated pairs have more often changed mates; in addition, the longest-lived pair breeding in all 4 yr was a full-sib pair. When the analysis is repeated with just the newly formed pairs (N = 7), the difference in 1992 is no longer significant (table 4). In this analysis, the three pairs that had bred in the previous year are assumed to be not available as mates for other birds in The mean coefficient of co-ancestry for all 20 pairs is (s = 0.103). This value is extremely high compared with the value of calculated from 583 pairs of G. fortis (Gibbs and Grant 1989). The percentage of full-sib or parent-offspring pairs (N = 3) in the G. fortis sample was 0.5%. In contrast, the percentage of such pairs in the G. magnirostris sample is either 30.0% or 33.3%, depending on whether the initial three founding pairs in 1983 are excluded or included and assumed to be unrelated. The high value of close inbreeding reflects the high probability of mating with a relative through random mate choice early in the establishment of a breeding population. In addition, the coefficient is high be-

7 FOUNDER POPULATION OF DARWIN'S FINCHES 235 cause closely related birds that formed pairs early in the establishment of a population survived well and remained as pairs, even when unrelated potential mates became more plentiful. Fitness Consequences of Inbreeding.-Clutch sizes varied from two to five eggs, with a mode of three or four depending on the year. These are typical values for the other two species on the island (Boag and Grant 1984b; Gibbs and Grant 1987) and did not differ appreciably from them in any one year. To assess the possibility that inbreeding carried a fitness cost we focus on hatching success, because this is the stage at which inbreeding depression has been found most strongly in studies of domestic birds (Falconer 1989). Data were grouped according to the coefficient of co-ancestry of the breeding pairs (0, 0.125, and 0.25), hatching success was calculated as an average value for each pair, the means for each pair were then arcsine-transformed, and the three groups subjected to ANOVA. Means differed in the direction expected from an hypothesis of inbreeding depression. Inbreeding pairs had lower hatching success (0.73 and 0.66 for the + = 0.25 and + = groups, respectively) than unrelated pairs (0.84). However, there was no statistical difference among the three groups, either in the total data (F = 1.565; df = 2, 17; P = 0.238) or the subset lacking those clutches that completely failed to hatch (F = 0.745; df = 2, 17; P = 0.490). Moreover, no one group differed from any other (Student-Newman-Keuls post-hoc tests, P > 0.1). The combined group of inbreeding birds did not differ from the noninbreeding ones (F = 1.543; df = 1, 18; P = 0.230). Further tests were carried out to see if inbreeding parents were less competent than noninbreeding ones in converting eggs to fledglings. The same procedure was used as in the analysis of hatching success with the mean number of fledglings/egg/pair. Differences among groups were not statistically significant for the total data (F = 0.830; df = 2, 17; P = 0.920), or for the subset lacking total hatching failures (F = 0.133; df = 2, 17; P = 0.876). Inbreeding effects may not be expressed, or may be weak, in the first generation of inbreeding, but could become more pronounced in subsequent generations. There are not enough data (only one family) to assess inbreeding effects in subsequent generations. Fluctuating Asymmetry.-Inbreeding can lead to a reduction in developmental homeostasis. One manifestation of this is a high level of nondirectional deviations from bilateral symmetry (Parsons 1992). The relatively inbred G. magnirostris should show deviations of this sort more frequently than the less inbred G. fortis. In 1993, we measured the lengths of the left and right tarsi of 52 G. fortis and 41 G. magnirostris to obtain estimates of the degree of fluctuating asymmetry (FA) in this trait. One G. magnirostris individual was the offspring of a brother-sister mating; three others were known from pedigrees to be more distantly related (+ < 0.125). The remainder were newly captured, probably born on the island in 1992 after we had left, and of unknown relationship. None of the G. fortis sample were known to be inbred (+ < 0.125). We used the absolute difference between the left and right measurements as an FA index for each individual and found it to be uncorrelated with the average size of the two tarsi 6 j4 A B S c SECONDS DAPHNE SANTA CRUZ 1.0 FIG. 4. Sonagrams of the three types of song sung by Geospiza magnirostris males on Daphne Major, with songs from Santa Cruz for comparison. in the samples of the two species treated separately or together (P > 0.1 in each case). Before comparing the two species, we tested for platykurtosis as an expression of antisymmetry (Palmer and Strobeck 1986). The sample of G. magnirostris displayed significant platykurtosis (k =2.33, P <0.01), whereas the sample of G. fortis did not (k , P > 0. 1). Neither sample was significantly skewed. The kurtosis of the G. magnirostris sample evidently was caused by three outliers with absolute asymmetry values greater than 1Imm. One was the single known inbred individual, and the other two were from the sample of previously unknown birds. When these three individuals were deleted from the sample, kurtosis disappeared (k = 0.283, P> 0.1). The two species were compared by Levene's test (Van Valen 1978) of absolute deviations from the median FA value rescaled by dividing by the average of left and right tarsus lengths of each individual. Deviations were significantly greater in the G. magnirostris sample than in the G. fortis sample (t91 = , P = ). Therefore, G. magnirostris displays higher levels of fluctuating asymmetry than G. fortis. Nongenetic Structure Song.-Like other members of the genus, G. magnirostris males typically sing a single and structurally simple song. Females do not sing. Three song types have been heard on Daphne (fig. 4). Two of them have been recorded on the neighboring island of Santa Cruz, giving support to the idea that Santa Cruz is the source of immigrants to Daphne. However, one (type B) has also been recorded on Santiago (Bowman 1983, fig. 52) and other islands (Bowman 1983, Grant and Grant 1989); therefore, multiple sources of colonists are possible. Darwin's finch songs are transmitted culturally and not genetically from one generation to the next and remain un-

8 236 P. R. GRANT AND B. R. GRANT altered for life (Grant and Grant 1989). In the G. fortis population, most sons faithfully copy their father's single song (Millington and Price 1985). Nevertheless, copying of other males in the population can be as frequent as 20% or more (Gibbs 1990; Green 1992). Geospiza magnirostris demonstrates a similar pattern: 10 of 11 recorded sons sang the same song type as their father. Song-type frequencies changed from 1983 to 1994, as a result of differential recruitment and of the invasion of a new type. In 1983, there were two song-a males and one song- B male on the island. One song-a male, born in 1983, was responsible for the persistence of this song type in the population through the production of sons and grandsons. In contrast, song-b males did not produce recruits, and no male sang this song in the period Their limited occurrence in the population in 1993 (two individuals) must have been due to further immigration. The first song C was recorded in By 1994, 40 of 41 males sang this song; four were sons of two males that sang this type, one was a son of a song-a singer, and the remaining 35 were either recruits born in the later part of the 1992 and 1993 breeding seasons when we were not present on the island, were immigrants, or a mixture of both. Thus the frequency of A songs changed from 0.67 in 1983 to 0.02 in 1994, and the frequency of C songs changed from 0.00 to If mate choice by a female is based on male song type in relation to the type of song sung by her father, a nonrandom breeding structure may arise (Grant and Grant 1989). Nine daughters of known paternal song type paired with males of the same song type, whereas only two paired with males singing a different song type. Owing to the predominance of C songs in the population when most of the matings were recorded, homotypic matings are to be expected by chance; in fact, the 9:2 ratio is consistent with a random-mating hypothesis (P > 0.1 for x2 test) as well as one based on assortative mating. DISCUSSION Colonization and Evolution A founder event and subsequent inbreeding can lead to the random loss of genetic variation and altered gene complexes. Berry (1992) has written that a genetic bottleneck such as is involved in most colonizations is the quickest means available of changing gene frequencies. If the new environment differs from the previous one, the population also may be subjected to directional selection from ecological pressures. For these two sets of reasons, genetic and ecological, the evolutionary trajectory of a species may be affected profoundly by events occurring in the initial stages of a colonization event. Our study shows that the founder population was a small and possibly nonrandom sample of available colonists, that recruits were produced from fewer than half of them, and that inbreeding occurred in the following two generations. A reduction in genetic variation may occur in association with each of these three conditions (e.g., see Haig et al. 1993). Nevertheless, breeding with new immigrants, as also occurred, will limit or eliminate the loss of variation, and if the source of colonists is heterogeneous variation could even be enhanced. Theoretical models usually assume for simplicity that colonization is random with respect to genotype (e.g., Charlesworth and Rouhani 1988). Colonization of Daphne by Geospiza magnirostris appears to have been nonrandom with respect to bill morphology. There is uncertainty because only two of the initial colonists were measured. Nevertheless, one of the later colonists was the largest of all 243 measured immigrants to the island, and one of the founding colonists was the next to largest in two bill dimensions. The colonists may have differed genetically as well as phenotypically from the immigrants that did not stay to breed. There is evidence of heritable variation in bill dimensions from parent-offspring regressions and sib-sib correlations, as is the case for G. fortis and G. scandens (Boag 1983; Grant and Grant 1994) and, on Isla Genovesa, G. conirostris (Grant and Grant 1989). The colonists were exceptionally variable phenotypically. One possible reason is that the largest and smallest immigrants were favored by ecological circumstances, principally food supply. Social factors such as dominance and pair compatibility may have been important as well. Variation could also be high because the colonists originated from more than one island. If this happened the islands would probably be the two large neighbors, Santa Cruz 8 km to the south and Santiago 22 km to the north. They support populations that differ morphologically from each other, more in bill dimensions than in tarsus or wing length (Grant et al. 1985). Mixing them would produce greater variation in bill traits than in the other traits. This is exactly what was observed in the sample of colonists. Song variation among the colonists is also consistent with a hypothesis of multiple invasions from these two source islands. The various uncertainties attending these interpretations underline the difficulties of demonstrating small but real and important founder effects in nature with the almost inevitably small numbers involved in colonization events, not to mention the small samples of those numbers. As in our study, samples are likely to be too small for a direct assessment of whether genetic variances and covariances change, and whether directional selection occurs in the first couple of generations. When the source of the colonists is known, and the population has built up to substantial numbers, the statistical chances of detecting small changes are much better, though identifying the causes of those changes is no easier. For example, Conant (1988a,b) demonstrated differences in bill size and shape between source and colonizing populations of Laysan finches (Telyspiza cantans) in the Hawaiian archipelago yr after the initial colonization took place. The first colonization, of Southeast island in the Pearl and Hermes group, was artificial, but nearby North island was colonized apparently naturally sometime in the next 6 yr. Conant (1988a,b) was unable to distinguish between natural selection and founder effects as causes of the differences, however, because she lacked measurements of birds in the early stages of the colonizations. There is no uncertainty about the occurrence of inbreeding in our study. Inbreeding occurred at a much higher frequency in the newly established G. magnirostris population than in the well-established and much larger population of G. fortis. The G. magnirostris inbreeding values are exceptionally high. Ralls et al. (1986) summarized data on close inbreeding (+

9 FOUNDER POPULATION OF DARWIN'S FINCHES 237 = 0.25) in natural populations of birds and mammals. Frequencies for 13 populations of 11 species of birds varied from 0.0 to 19.4%. Figures for mammals were comparable, although the maximum was lower. The frequency for G. magnirostris, 33.3%, is higher than all of them. The estimate of 19.4% for the Australian fairy wren (Rowley et al. 1986) has since been drastically reduced following the discovery of a high incidence of extra-pair fertilizations in this species (Brooker et al. 1990; Rowley and Russell 1990). This serves as a warning that the figure for G. magnirostris could similarly be inflated, although we doubt this because the species has a different (nonsocial) breeding system, and we have no observational evidence of extra-pair copulations. The frequency of inbreeding is expected to decrease as the population grows in size. There were signs that inbreeding carried a small fitness cost in terms of reproductive success. Moreover, G. magnirostris individuals displayed a higher level of fluctuating asymmetry than G. fortis in one morphological trait. This is an indication of developmental instability that can arise for genetic (and environmental) reasons, such as reduced buffering of environmental influences during growth under the influence of genes, largely in homozygous condition in inbred organisms. It is well illustrated by the work of Vrijenhoek and Lerman (1982) with Poeciliopsis fish in Mexico. They were able to link the occurrence of reduced developmental stability in no less than eight morphological traits with a reduction in heterozygosity apparently caused by a series of founder events. Numbers declined, possibly as a result of inbreeding depression. Inbreeding depression has also been implicated in the local extinction of two bird species: the heath hen (Tympanuchus cupido) on Martha's Vineyard in 1932 (Bent 1932), and the middle spotted woodpecker (Dendrocopos medius) in Sweden in 1983 (Pettersson 1985). The most striking findings of this study are first a high level of inbreeding, and second, despite indications of inbreeding depression, a strong exponential increase in population size, largely through local recruitment. The outcome of the colonization-establishment of a moderately large breeding population in the space of a few years-was apparently determined more by ecological factors than by the particular genetic constitution of individuals born on the island. A similar conclusion was reached by Berry et al. (1978) from a study of electrophoretically detectable genetic variation in colonizing populations of house mice (Mus) on various islands. This is not to say that genetic changes were unimportant in our study. First, colonization may have been nonrandom, as discussed above. Second, drift may have occurred, as suggested by the fact that small nonrandom tendencies in the composition of the colonists were amplified in the next two generations. Third, after the initial colonization. nonselective as well as selective changes may have taken place in the G. magnirostris population below the level of detection, affecting alleles with minor expression in morphological, physiological or biochemical traits. Finally, genetic variation may have been augmented by the breeding of singing one song type had better fledging and recruitment success than those singing another. This appears to be a cultural analogue to genetic drift promoted by very small numbers. Cultural drift has been repeatedly invoked to explain changes in song-type frequencies in much larger populations over longer periods of time (e.g. Payne 1988, Lynch and Baker 1993). It could be the main factor responsible for differences in song characteristics among populations of the same species, both on the same and on different islands (Ratcliffe 1981; Grant and Grant 1989). Causes of the Colonization The number of species of Darwin's finches on each island in the Galapagos archipelago varies as a function of island area, isolation, and measures of plant richness (Hamilton and Rubinoff 1967; Harris 1973; Abbott et al. 1977). On the basis of a simple regression of number of Geospiza species on ln area (r = 0.712, P = , N = 22 islands) Daphne should support populations of two finch species. The actual number until recently was three, and is now four. Daphne is the smallest island supporting breeding populations of four Geospiza species. Raibida, the next largest with four species (Abbott et al. 1977), is approximately 15 times larger than Daphne. How did G. magnirostris colonize an isl'and that was apparently full? Repeated immigration, facilitated by the proximity of Daphne to the large island of Santa Cruz 8 km away, is an obviously relevant factor. Studies of birds in Europe and elsewhere have shown that colonizations of new areas often follow population increase or range expansion in the source area; repeated immigrations of the same species, initially without attempts to breed, have preceded the colonization (O'Connor 1986; Garnett et al. 1992), as in our study. This implies that colonization of islands by mobile organisms like birds is determined by the suitability of the habitat as much as by ability to reach it (Lack 1969, 1976; Williamson new immigrants from different islands. Geospiza magnirostris does not have unique requirements Song type, which is a nongenetic, culturally inherited, trait, for breeding (Grant and Grant 1980). Thus, it appears that changed radically in frequency, apparently for reasons unconnected directly with the song itself, but because males instead they emigrated at the usual time of breeding. they could have bred on the island before , but Because 1981). A model developed by Schluter and Grant (1984) accounts for the number'of granivorous species on an island and their morphological differences in terms of the seed supply profile in the dry season. Applied to Daphne, it shows a relatively low (but not zero) population density expected for a species with the beak size of G. magnirostris. A few other islands with a low expected population density for G. magnirostris do not support a population of this species. Daphne is different in being close to a large island (Santa Cruz) that does support a population. This suggests that food supply of the island and immigration frequency are joint determinants of colonization. They are not sufficient to account for the timing of the colonization event on Daphne, however. Large and hard seeds, which constitute the diet of G. magnirostris in the dry season, were relatively and absolutely more common in the 7 yr preceding 1983, when G. magnirostris did not breed, than in the following 8 yr when they did (Grant and Grant 1993).

10 238 P. R. GRANT AND B. R. GRANT almost all the immigrants were in brown or partially black plumage they were likely to have been one- or 2-yr olds that had never bred. It is also likely that they returned to their island of origin to breed. The question then is why a few individuals broke this pattern and stayed on Daphne to breed early in the El Nifno year and before the effects of the extraordinary rainfall had developed. There was nothing unusual in the environment when G. magnirostris began to breed, such as an outbreak of the caterpillars that are fed to nestlings. Finches lay eggs in the first month of heavy rain in the wet season, and caterpillars become numerous on the flower buds of Portulaca howellii at this time, as well as on other plants later on (Millington and Grant 1983; Boag and Grant 1984b). Counts of Portulaca caterpillars in these initial months were 65 in April 1979, 15 in March 1980, 125 in April 1981, and 61 in December 1982 when G. magnirostris first bred. None were found on the leaves of Bursera graveolens and several species of annual plants in December The only unusual feature was the early arrival of heavy rain, in November The rapid breeding response of other finches on the island may have induced singing and nestbuilding activity in the G. magnirostris males before they were in a physiological state of readiness to emigrate. Breeding activity of the males would have stimulated the females. There have been no studies of the hormonal control of breeding activity, territoriality, and "migration" of Darwin's finches. The closest model is probably the Zebra finch (Taeniopygia guttata), which experiences erratic rainfall in arid Australian environments. Local and heavy rainfall induces a rapid breeding response in these nomadic finches at the site of the rainfall (Serventy 1971; Davies 1976). In conclusion, the large number of finch species on Daphne and the recent colonization event can be explained by its proximity to a large and species-rich island, repeated immigration, and an unusual climatic event. Islands may be undersaturated with species not only because they are isolated but as a result of difficulties of establishment of breeding populations of immigrants (Grant and Grant 1982). Little is known about those difficulties, which is one reason why colonization is ecologically unpredictable (Lodge 1993). Presumably, they vary from species to species. Results of the present study suggest the need for an investigation of the behavioral and physiological factors that govern the decision to stay and breed or leave. ACKNOWLEDGMENTS We thank the more than 20 people who have assisted us in the field, and L. Freed, H. L. Gibbs, T. D. Price, D. Schluter, and an anonymous reviewer for comments on the manuscript. We thank S. Nowicki and S. Peters for the use of equipment for song analysis, and R. I. Bowman for supplying some sonagrams. The study was conducted with permission and support from the Direccion General de Desarrollo Forestal, Quito, the Galapagos National Park Service, and the Charles Darwin Research Station. It has been funded by grants from Natural Sciences and Engineering Research Council (Canada) and the National Science Foundation. LITERATURE CITED Abbott, I., L. K. Abbott, and P. R. Grant Comparative ecology of Galapagos ground finches (Geospiza Gould): evaluation of the importance of floristic diversity and interspecific competition. Ecological Monographs 47: Barton, N., and B. Charlesworth Genetic revolutions, founder effects, and speciation. Annual Review of Ecology and Systematics 15: Beebe, W Galapagos: world's end. Putnams, New York. Bent, A. C Life histories of North American gallinaceous birds. Smithsonian Institution United States National Museum Bulletin 162: Berry, R. J The significance of island biotas. Biological Journal of the Linnean Society 46:3-12. Berry, R. J., J. Peters, and R. J. van Aarde Sub-antarctic house mice: colonization, survival and selection. Journal of Zoology, London 184: Boag, P. T The heritability of external morphology in Darwin's ground finches (Geospiza) on Isla Daphne Major, Galapagos. Evolution 37: Boag, P. T., and P. R. Grant. 1984a. The classical case of character release: Darwin's finches (Geospiza) on Isla Daphne Major, Galapagos. Biological Journal of the Linnean Society 22: b. Darwin's finches (Geospiza) on Isla Daphne Major, Galapagos: breeding and feeding ecology in a climatically variable environment. Ecological Monographs 54: Boag, P. T., and A. J. van Noordwijk Quantitative genetics. Pp in F Cooke and P. A. Buckley, eds. Avian genetics. A population and ecological approach. Academic Press, New York. Bock, C. E., and L. W. Lepthien Population growth in the cattle egret. Auk 93: Bowman, R. I Morphological differentiation and adaptation in the Galapagos finches. University of California Publications in Zoology 58: The evolution of song in Darwin's finches. Pp in R. I. Bowman, M. Berson, and A. E. Leviton, eds. Patterns of evolution in Galapagos organisms. American Association for the Advancement of Science, Pacific Division, San Francisco. Boyce, M. S Population viability analysis. Annual Review of Ecology and Systematics 23: Brooker, M. G., I. Rowley, M., Adams, and P. R. Baverstock Promiscuity: an inbreeding avoidance mechanism in a socially monogamous species? Behavioural Ecology and Sociobiology 26: Bryant, E. H., and L. M. Meffert The effect of serial founderflush cycles on quantitative genetic variation in the housefly. Heredity 70: Bryant, E. H., S. A. McCommas, and L. M. Combs The effect of an experimental bottleneck upon quantitative genetic variation in the housefly. Genetics 114: Bush, M. B., and R. J. Whittaker Krakatau: colonization patterns and hierarchies. Journal of Biogeography 18: Carson, H. L., and A. R. Templeton Genetic revolutions in relation to speciation phenomena: the founding of new populations. Annual Review of Ecology and Systematics 15: Charlesworth, B., and S. Rouhani The probability of peak shifts in a founder population. II. An additive polygenic trait. Evolution 42: Charlesworth, D., and B. Charlesworth Inbreeding depression and its evolutionary consequences. Annual Review of Ecology and Systematics 18: Conant, S. 1988a. Geographic variation in the Laysan finch (Telyspiza cantans). Evolutionary Ecology 2: b. Saving endangered species by translocation. BioScience 38: Crow, J. L., and M. Kimura An introduction to population genetics theory. Harper and Row, New York. Davies, S. J. J. E Environmental variables and the biology

11 FOUNDER POPULATION OF DARWIN'S FINCHES 239 of Australian arid zone birds. Pp in H. J. Frith and J. H. Calaby, eds. Proceedings of the 16th International Ornithological Congress, Canberra, Australia (1974). Diamond, J. M Colonization of exploded volcanic islands by birds: the super-tramp strategy. Science 184: Falconer, D Introduction to quantitative genetics, 3d ed. Longman, Essex, U.K. Fridriksson, S Surtsey. Wiley, New York. Garnett, S. G., A. C. Williams, and R. W. H. Hindmarsh Island colonisation after possible reconnaisance by the yellowbellied sunbird Nectarinia jugularis. Emu 91: Gibbs, H. L Cultural evolution of male song types in Darwin's medium ground finches, Geospizafortis. Animal Behaviour 39: Gibbs, H. L., and P R. Grant Ecological consequences of an exceptionally strong El Nifio event on Darwin's finches. Ecology 68: Inbreeding in Darwin's medium ground finches (Geospiza fortis). Evolution 43: Gifford, E. W Field notes on the land birds of the Galapagos Islands and of Cocos Island, Costa Rica. Proceedings of the California Academy of Sciences, ser.4. 2: Goodnight, C. J On the effect of founder events on epistatic genetic variance. Evolution 41: Epistasis and the effect of founder events on the additive genetic variance. Evolution 42: Grant, B. R., and P. R. Grant Niche shifts and competition in Darwin's finches: Geospiza conirostris and congeners. Evolution 36: Evolutionary dynamics of a natural population. The large cactus finch of the Galapagos. University of Chicago Press, Chicago Evolution of Darwin's finches caused by a rare climatic event. Proceedings of the Royal Society of London B 251: Grant, P. R Ecology and evolution of Darwin's finches. Princeton University Press, Princeton, N.J Hybridization of Darwin's finches on Isla Daphne Major, Galapagos. Philosophical Transactions of the Royal Society of London B 340: Grant, P. R., and P. T. Boag Rainfall on the Galapagos and the demography of Da'rwin's finches. Auk 97: Grant, P R., and B. R. Grant The breeding and feeding characteristics of Darwin's finches on Isla Genovesa, Galapagos. Ecological Monographs 50: Demography and the genetically effective sizes of two populations of Darwin's finches. Ecology 73: Phenotypic and genetic effects of hybridization in Darwin's finches. Evolution 48: Predicting microevolutionary responses to directional selection on heritable variation. Evolution 49: Grant, P. R., I. Abbott, D. Schluter, R. L. Curry, and L. K. Abbott Variation in the size and shape of Darwin's finches. Biological Journal of the Linnean Society 25:1-39. Gray, A. J Do invading species have definable genetic characteristics? Philosophical Transactions of the Royal Society of London B 314: Green, J. A The transmission of male song in Darwin's medium ground finch Geospiza fortis. B. A. thesis. Princeton University, Princeton, N.J. Haig, S. M., J. R. Belthof, and D. H. Allen Population viability analysis for a small population of red-cockaded woodpeckers and an evaluation of enhancement strategies. Conservation Biology 7: Hamilton, T H., and I. Rubinoff On predicting insular variation in endemism and sympatry for the Darwin finches in the Galapagos archipelago. American Naturalist 101: Harris, M. P The Galapagos avifauna. Condor 75: Lack, D The Galapagos finches (Geospizinae): a study in variation. Occasional Papers of the California Academy of Sciences 21: Darwin's finches. Cambridge University Press, Cambridge The number of bird species on islands. Bird Study 16: Island biology, illustrated by the land birds of Jamaica. University of California Press, Los Angeles. Lande, R Genetic variation and phenotypic evolution during allopatric speciation. American Naturalist 116: Larsson, K., P. Forslund, L. Gustafsson, and B. S. Ebbinge From the high arctic to the Baltic: the successful establishment of a barnacle goose Branta leucopsis population on Gotland, Sweden. Ornis Scandinavica 19: Lodge, D. M Biological invasions: lessons for ecology. Trends in Ecology and Evolution 8: Lynch, A., and A. J. Baker A population memetics approach to cultural evolution in chaffinch song: meme diversity within populations. American Naturalist 141: Mayer, G. C., and R. M. Chipley Turnover in the avifauna of Guana Island, British Virgin Islands. Journal of Animal Ecology 61: Mayr, E Change of genetic environment and evolution. Pp in J. S. Huxley, A. Hardy, and E. B. Ford, eds. Evolution as a process. Allen and Unwin, London Controversies in retrospect. Pp in D. J. Futuyma and J. Antonovics, eds. Oxford Surveys in Evolutionary Biology, Vol. 8. Oxford University Press, Oxford. Millington, S. J., and P. R. Grant Feeding ecology and territoriality of the Cactus finch Geospiza scandens on Isla Daphne Major, Galapagos. Oecologia 58: Millington, S. J., and T. D. Price Song inheritance and mating patters in Darwin's finches. Auk 102: O'Connor, R. J Biological characteristics of invaders. Philosophical Transactions of the Royal Society of London B 314: Palmer, A. R., and C. Strobeck Fluctuating asymmetry: measurement, analysis, patterns. Annual Review of Ecology and Systematics 17: Fluctuating asymmetry as a measure of developmental stability: implications of nonnormal distributions and power of statistical tests. Acta Zoologica Fennica 191: Parsons, P. A Environmental stresses and environmental conservation of natural populations. Annual Review of Ecology and Systematics 20: Fluctuating asymmetry: a biological monitor of environmental and genomic stress. Heredity 68: Payne, R. B Biological and cultural success of song memes in indigo buntings. Ecology 69: Pettersson, B Extinction of an isolated population of the middle spotted woodpecker Dendrocopos medius (L.) in Sweden and its relation to general theories on extinction. Biological Conservation 32: Provine, W. B Founder effects and genetic revolutions in microevolution and speciation: an historical perspective. Pp in L. V. Giddings, K. Y. Kaneshiro, and W. W. Anderson, eds. Genetics, speciation and the founder principle. Oxford University Press, Oxford. Ralls, K., P. H. Harvey, and A. M. Lyles Inbreeding in natural populations of birds and mammals. Pp in M. Soule, ed. Conservation biology: the science of scarcity and diversity. Sinauer, Sunderland, Mass. Ratcliffe, L. M Species recognition in Darwin's ground finches (Geospiza Gould). Ph.D diss. McGill University, Montreal. Rothschild, W., and E. Hartert A review of the ornithology of the Galapagos Islands. With notes on the Webster-Harris expedition. Novitates Zoologicae 6: Further notes on the fauna of the Galapagos Islands. Novitates Zoologicae 9: Rowley, I., and E. Russell "Philandering"-a mixed mating strategy in the splendid fairy-wren Malurus splendens. Behavioural Ecology and Sociobiology 27: Rowley, I., E. M. Russell, and M. G. Brooker Inbreeding benefits may outweigh costs. Animal Behaviour 34: Serventy, D. L Biology of desert birds. Pp in D.

12 240 P. R. GRANT AND B. R. GRANT S. Farner and J. R. King, eds. Avian biology, Vol. 1. Academic Press, New York. Schluter, D., and P. R. Grant Determinants of morphological patterns in communities of Darwin's finches. American Naturalist 123: Smith, J. N. M., P. R. Grant, B. R. Grant, I. Abbott, and L. K. Abbott Seasonal variation in feeding habitats of Darwin's ground finches. Ecology 59: Snodgrass, R. E., and E. Heller Papers from the Hopkins- Stanford Galapagos expedition, XVI. Birds. Proceedings of the Washington Academy of Science 5: Sprunt, A Newcomer from the old world. Audubon Magazine 55: Swarth, H. S The avifauna of the Galapagos Islands. Occasional Papers of the California Academy of Sciences 18: Thornton, I. W. B., T. R. New, R. A. Zann, and P. A. Rawlinson Colonization of the Krakatau Islands by animals: a perspective from the 1980s. Philosophical Transactions of the Royal Society of London B 328: Van Valen, L The statistics of variation. Evolutionary Theory 4: Vrijenhoek, R. C., and S. Lerman Heterozygosity and developmental stability under sexual and asexual breeding systems. Evolution 36: Wiggins, I. L., and D. M. Porter Flora of the Galapagos Islands. Stanford University Press, Stanford, Calif. Williamson, M Island populations. Oxford University Press, Oxford. Willis, J. H., and H. A. Orr Increased heritable variation following population bottlenecks: the role of dominance. Evolution 47: Corresponding Editor: D. Schluter