Phylogeny and Evolution of the Sulidae (Aves: Pelecaniformes): A Test of Alternative Modes of Speciation

MOLECULAR PHYLOGENETICS AND EVOLUTION

Vol. 7, No. 2, April, pp. 252–260, 1997 ARTICLE NO. FY960397

Phylogeny and Evolution of the Sulidae (Aves: Pelecaniformes): A Test of Alternative Modes of Speciation V. L. Friesen* and D. J. Anderson† *Department of Biology, Queen’s University, Kingston, Ontario K7L 3N6, Canada; and †Department of Biology, Wake Forest University, Winston-Salem, North Carolina 27109 Received June 26, 1996; revised November 20, 1996

Although the allopatric model of speciation is widely accepted, it does not provide a satisfactory explanation for many evolutionary phenomena. Several alternative models exist, but they remain largely untested for vertebrate animals. In the present paper, a molecular phylogeny was used to test competing models of speciation in a seabird family, the Sulidae. A segment including 807 base pairs of the mitochondrial cytochrome b gene was sequenced from all extant sulid species, and phylogenetic methods were used to test model-specific predictions regarding tree topologies, distributions of sister taxa, timing of vicariant events, and comparative biology. Both the neighbor-joining and parsimony analyses placed sequences of gannets (Morus spp.) and boobies of the genus Sula in separate, monophyletic lineages. Sequences of Cape (M. capensis) and Australasian (M. serrator) gannets clustered together, and the sequence of Abbott’s booby (Papasula abbotti) was basal to those of the gannets. Sequences of blue-footed (S. nebouxii) and Peruvian (S. variegata) boobies were sisters and formed a monophyletic group with the masked booby (S. dactylatra). The red-footed booby (S. sula) sequence was the most divergent of the Sula boobies. All relationships received strong support from standard-error tests and bootstrap analysis. Substitution rates were similar to those suggested for mammals and suggested that most lineages arose within the last 3 million years. Lineage divergence events for which the mode of speciation could be deduced did not fit the predictions of either allopatric or sympatric models, but apparently involved either peripatric or parapatric processes. r 1997 Academic Press

The relative roles of different modes of speciation in vertebrate animals is the subject of extensive debate. The allopatric or vicariant model involves gradual divergence of two populations of moderate to large size over extended periods of separation by extrinsic barriers to gene flow (Mayr, 1970). Although widely accepted, this model is not satisfactory for several evolutionary phenomena. For example, birds encounter few 1055-7903/97 $25.00 Copyright r 1997 by Academic Press All rights of reproduction in any form reserved.

obstacles to dispersal and so violate one requirement for allopatric speciation, yet their rates of speciation are at least as high as for other groups of vertebrates (Wyles et al., 1983). Several alternative mechanisms of speciation therefore have been proposed. These mechanisms fall into three main models: parapatric speciation involves divergence of individuals from adjacent parts of a range due to local adaptation (Endler, 1977); according to peripatric models, speciation occurs in small, peripherally isolated populations (e.g., Slatkin, 1996); under sympatric models, speciation involves the establishment of intrinsic barriers to reproduction in populations having no extrinsic barriers to gene flow (e.g., phytophagous insects specializing on different hosts; Diehl and Bush, 1989; Bush, 1994). The feasibility of these alternatives, as well as their prevalence in the natural world, has been hotly debated (e.g., Gibbons, 1996). For example, Lynch (1989) used a phylogenetic approach to estimate that 20% of speciation events in birds may involve sympatric speciation, whereas Butlin (1989) contended that the conditions required for establishment of premating isolation in the absence of extrinsic barriers to gene flow are sufficiently restrictive as to be rare. Phylogenetic analyses provide new avenues for evaluating alternative mechanisms of speciation in that they enable different hypotheses associated with alternative models to be tested. Specifically, they allow tree topologies to be compared with those associated with different models (Brooks and McLennan, 1991), and they permit biogeographical correlates of alternative models to be assessed (Lynch, 1989). Furthermore, molecular phylogenies enable dates of speciation events to be estimated and compared with possible vicariant events and allow morphological, ecological, and ethological changes associated with speciation to be inferred. Such analyses suggest that many speciation events at high latitudes were driven by the climatic oscillations of the Pleistocene. For example, the Alcidae is a family of charadriiform seabirds that breed in temperate to arctic regions throughout the Northern Hemisphere; molecular phylogenetic analyses suggest that most speciation events

252

PHYLOGENY AND EVOLUTION OF THE SULIDAE

during the Pliocene and Pleistocene in this group were associated with major vicariant events such as glaciations (Friesen et al., submitted). Tropical seabirds would not have been affected by glaciers to the same extent as were their temperate and arctic counterparts, yet they are more species-rich. Thus, fundamental differences in modes of speciation may exist between these geographical groupings. In the present study, modes of speciation were investigated in the Sulidae, a family of pelecaniform seabirds that breed predominantly in tropical oceans. Although a number of phylogenetic hypotheses have been proposed for the family on the basis of general biology (Nelson, 1978), osteological characters (Warheit, 1990), and DNA sequences (Sibley and Ahlquist, 1990), a complete phylogeny based on molecular characters has not yet been published. The relationship between Abbott’s booby (see Table 1 for scientific names) and the other species, as well as the positions of masked and brown boobies relative to other Sula boobies, are especially unclear. In the present study, a segment of the mitochondrial cytochrome b gene was sequenced for all nine extant species, and phylogenetic methods were used to test predictions of alternative modes of speciation. METHODS Blood samples were obtained from two representatives of each species of sulid (Table 1), except for Abbott’s booby, for which a mature feather was obtained from each of two birds. Tissue also was obtained from a pelagic cormorant (Phalacrocorax pelagicus), a member of the putative sister-group to the sulids (Hedges and Sibley, 1994) for outgroup rooting. Samples were stored either in 70% ethanol or in a lysis buffer

TABLE 1 Scientific Names and Sampling Locations for Gannets and Boobies Common name

Scientific name

Northern gannet

Morus bassanus

Cape gannet

M. capensis

Australasian gannet

M. serrator

Abbott’s booby

Papasula abbotti a

Masked booby Brown booby

Sula dactylatra S. leucogaster

Blue-footed booby Red-footed booby Peruvian booby

S. nebouxii S. sula S. variegata

a

Olson and Warheit, 1988.

Sampling location Funk Island, Newfoundland Malgas Island, South Africa Cape Kidnappers, New Zealand Christmas Island, Indian Ocean Isla Espan˜ola, Galapagos San Pedro Martir Island, Mexico Isla Espan˜ola, Galapagos Isla Gardner, Galapagos Isla Chincha, Peru

253

(Seutin et al., 1991), or were frozen at 270°C. Total (nuclear and mitochondrial) DNA was purified by digestion of 10 µl of blood with proteinase K and removal of peptides and lipids by phenol extraction (Kocher et al., 1989). Some DNA samples were purified further using GeneClean II kits (Bio-101; Vista, CA) according to the manufacturer’s directions. Mitochondrial DNA (mtDNA) also was purified from liver tissue from one northern gannet and one Australasian gannet by ultracentrifugation on cesium chloride gradients (Carr and Griffith, 1987). Preliminary studies indicated that sulids possess a nuclear copy of the 58 end of the cytochrome b gene (V.L.F., unpublished data); due to difficulties in amplifying the mitochondrial copy, no data from this region were included in the present study. Two fragments encompassing 807 base pairs (bp) of the middle and 38 end of cytochrome b were amplified using primers and protocols described in Kocher et al. (1989) and Birt and Baker (in preparation); DNA could not be amplified from one of the feathers from Abbott’s booby. Amplified DNA was subjected to electrophoresis in 2% agarose gels, excised, and purified using GeneClean II kits. Cycle-sequencing was conducted using Amplicycle kits (Perkin–Elmer; Branchburge, NJ) with direct incorporation of [a-33P]dATP according to the manufacturer’s directions. One additional primer (CytbL 15371:58GCCATCCCATACATTGGCCAAAC-38) was designed as an internal sequencing primer. Sequences were aligned by eye and were confirmed by sequencing complementary strands (approximately 40% of base pairs). A neighbor-joining tree (Saitou and Nei, 1987) was constructed from pairwise distance estimates with MEGA (version 1.0; Kumar et al., 1993) using Kimura’s (1980) two-parameter correction for multiple hits, which accommodates unequal representation of bases and a transitional bias (see Results). The cormorant sequence was designated as an outgroup taxon. Reliabilities of phylogenetic relationships were assessed using standard error tests (Rzhetsky and Nei, 1992, 1993). Maximum parsimony analysis was conducted with PAUP (version 3.1.1; Swofford and Begle, 1993) using the branch-and-bound search algorithm with the cormorant sequence designated as the outgroup. Starting trees were generated using the ‘‘closest’’ addition option, and ‘‘tree bisection-reconstruction’’ was used for branch-swapping. Reliabilities of phylogenetic relationships were evaluated by bootstrapping (Felsenstein, 1985). To maximize the number of relationships receiving strong (.80%) bootstrap support, data were reanalyzed with transversions weighted 4 or 20 times transitions (Friesen et al., 1996) or with transitions excluded. Brown et al. (1982) argued that substitutions within mtDNA accumulate in a clock-like fashion and may be used to date lineage divergences if calibrated properly. To estimate substitution rates for cytochrome b for the sulids, dates of divergence of sulids and phalacrocoracids, and of Sula and Morus, were estimated from

254

FRIESEN AND ANDERSON

available fossil data (Olson, 1985; Warheit, 1992; Benton, 1993) and were compared with numbers of substitutions between lineages within cytochrome b. Dates of divergence of other lineages within the Sulidae were estimated using these rates. Molecular and comparative data were used to conduct four tests that differentiate among alternative models of speciation. None of these tests provide definitive evidence for or against any one model, but together they are suggestive of different alternatives. Tree topology. Different modes of speciation tend to produce different tree topologies (Brooks and McLennan, 1991): specifically, mechanisms that result in simultaneous peripatric and sympatric speciation are more likely to generate polytomies than are allopatric speciation, peripatric speciation via sequential dispersal, or parapatric speciation. However, because a polytomy represents the null hypothesis for any phylogenetic investigation, the degree of support for either simultaneous peripatric or sympatric speciation depends on the power of the analysis, which is determined largely by the number and type of characters (nucleotides; Walsh et al., submitted). Thus, lack of phylogenetic resolution is suggestive of simultaneous peripatric or sympatric speciation, but only within the limits of power of the dataset. A fully resolved, strongly supported tree does not provide any inferences about mode of speciation. Biogeography. Assuming that changes in breeding distributions following speciation are minor, modes of speciation can be inferred from the distributions of sister-taxa (Lynch, 1989; Chesser and Zink, 1994). Thus, for example, sympatric distributions of sistertaxa are suggestive of sympatric speciation. For the present study, breeding distributions of sulids were obtained from Nelson (1978). To test the assumption that breeding ranges do not change with time, distributions of species (linear distances between the two most distant colonies) were tested for correlation with lineage age as estimated from sequence data; Abbott’s booby was excluded from this analysis since its range appears to have declined due to habitat destruction over the past century (Nelson, 1978). Vicariance events. Allopatric speciation and microvicariant versions of peripatric speciation require an extrinsic barrier to dispersal; thus, absence of a vicariant event within the approximate time-frame of speciation enables rejection of allopatric or microvicariant speciation; however, presence of a potential vicariant event does not necessarily mean that speciation involved allopatric processes. For the present analysis, dates of divergence of lineages within the Sulidae were compared with geological data (Warheit, 1992) to infer possible vicariant events. Comparative predictions. Gradual divergence of lineages through genetic drift and selection will tend to

result in a correlation between numbers of changes in morphological, ecological and ethological characters, and lineage age. However, parapatric speciation probably requires strong selection (Endler, 1977) and therefore will tend to result in especially large numbers of character changes. Elevated numbers of character change due to selection (or drift in small populations) may also be associated with sympatric and peripatric speciation, but are not indicated in allopatric speciation. Rates of character change therefore can be used to investigate modes of speciation. For the present study, comparative data on skeletal morphology (Warheit, 1990), ecology, and ethology (Nelson, 1970; Kennedy et al., 1996; Appendix 1) were obtained from the literature. Numbers of unambiguous changes associated with lineages were determined using McClade (Maddison and Maddison, 1992) and were regressed against lineage age. Numbers of character changes that were significantly higher than predicted by the regression equation were considered to be inconsistent with allopatric speciation and suggestive of parapatric speciation (or less probably, sympatric or peripatric speciation). Although this is a post hoc analysis, it provides a conservative test. RESULTS AND DISCUSSION Sequence Variation Several lines of evidence indicate that sequences obtained for the Sulidae in the present study represent functional, mitochondrial copies of cytochrome b. (Sequences have been deposited in GenBank under Accession Nos. xxx). First, sequences obtained from purified mtDNA and total DNA differed by at most one silent substitution for both the northern and Australasian gannets. Second, whereas substitutions within nuclear homologs should occur randomly, with approximately 2 ⁄3 involving amino acid replacements, most substitutions within the sulid sequences involved third position transitions, and only 25 of 195 variable sites involved replacement substitutions. Finally, variable regions corresponded roughly to transmembrane domains of cytochrome b, and conserved regions matched blocks where redox centers are thought to occur (Howell, 1989). No insertions or deletions were found among the 10 species. Base composition was strongly biased: C occurred at a mean (6se) of 35.1 6 0.38% of sites, whereas G occurred at only 11.9 6 0.10% of sites; A and T were represented roughly equally, occurring at 27.4 6 0.14 and 25.6 6 0.30% of sites, respectively. These biases are similar to those found in other species of birds (e.g., Kocher et al., 1989; Edwards et al., 1991; Kornegay et al., 1993; Friesen et al., 1996; Birt and Baker, in preparation). Of 807 nucleotide sites, 250 were variable and 145 were phylogenetically informative among the 10 species. Forty-four substitutions resulted in amino acid replacements. In pairwise com-

PHYLOGENY AND EVOLUTION OF THE SULIDAE

255

parisons of sequences, transitions outnumbered transversions by a mean ratio of 3.6:1 (se 5 0.69, range 5 1.1– 29; Fig. 1): C & T transitions were the most common substitutions, and G & T transversions were the least common. None of the comparisons were approaching a transition:transversion ratio of 0.5 indicative of saturation. Genetic distances among cytochrome b sequences, using Kimura’s (1980) two-parameter correction, averaged 0.136 (se 5 0.009), and ranged from 0.0075 to 0.234. Phylogenetic Results Both the neighbor-joining and parsimony analyses placed sequences of gannets and Sula boobies in separate, monophyletic lineages (Fig. 2). Sequences of Cape and Australasian gannets clustered together, and the sequence of Abbott’s booby was basal to those of the gannets. Sequences of blue-footed and Peruvian boobies were sisters and formed a monophyletic group with the masked booby. The red-footed booby sequence was the most divergent of the Sula boobies. All relationships received strong support (.70%) from standarderror tests and bootstrap analysis. Parsimony analysis in which transversions were weighted four times transitions improved support for most relationships (Fig. 2). The phylogenetic hypothesis presented in Fig. 2 agrees largely with those of Nelson (1978; 812) and Warheit (1990), with three exceptions: (1) Nelson considered the brown booby to be the sister-species to bluefooted and Peruvian boobies; (2) Warheit did not resolve the relationships among the gannets, but Nelson considered Cape and northern gannets to be sister-

FIG. 2. Most parsimonious tree for cytochrome b sequences of sulids, with transversions weighted four times transitions. The number above each branch is support from bootstrap analysis (100 replications); the number below each branch is support from standard error tests on the neighbor-joining tree. Branch lengths are proportional to numbers of substitutions.

species; and (3) Nelson was uncertain of the phylogenetic position of Abbott’s booby but did not view a sister-group relationship between Abbott’s booby and the gannets to be a possibility, and Warheit considered Abbott’s booby to be basal to the Sula boobies. The present finding for Abbott’s booby contrasts with several ecological and morphological affinities between it and the Sula boobies (such as tropical distributions and similarities in several osteological characters; Warheit, 1990), but agrees with several ecological, ethological, and morphological similarities between it and the gannets (such as absence of a ‘‘wing flicking’’ display (Nelson, 1970), and a relatively long humerus (Warheit, 1990)). The phylogenetic position of Abbott’s booby needs to be confirmed through analyses of other genes. Otherwise, the general agreement between results of the present study and those of previous investigations suggest that the phylogeny presented in Fig. 2 represents a species tree for the Sulidae and not just a gene tree for mtDNA. Estimation of Divergence Dates

FIG. 1. Numbers of transversions (above diagonal) and total numbers of substitutions (below diagonal) within an 807-bp fragment of cytochrome b among the Sulidae. NG, northern gannet; CG, Cape gannet; AG, Australasian gannet; AB, Abbott’s booby; MB, masked booby; BB, brown booby; BFB, blue-footed booby; RFB, red-footed booby; PB, Peruvian booby; PC, pelagic cormorant.

A plot of number of transitions against number of transversions for pairwise comparisons of cytochrome b sequences for the sulids (figure not shown) indicated that parallel substitutions and reversals are beginning to accumulate both between gannets (including Abbott’s booby) and boobies and between sulids and phalacrocoracids; estimates of divergence dates based on total numbers of substitutions for these comparisons therefore may be low. Irwin et al. (1991) calculated that transversions in cytochrome b in mammals accumulate at a rate of 0.2/100 bp/million years (my). The mean

256

FRIESEN AND ANDERSON

number of transversions between sequences of gannets and Sula boobies (4.6 6 0.6/100 bp; Fig. 1) suggests a divergence date of 23 mya (95% CI 5 17–29 mya; Fig. 3); this corresponds to the Oligocene/Miocene boundary, from when fossils first could be referred clearly to either Morus or Sula (Olson, 1985; Warheit, 1990). The mean number of transversions between the cormorant and sulids (8.6 6 0.11/100 bp) suggests a divergence date of 42 mya (95% CI 5 41–43 mya), which is only slightly older than the ages of the earliest sulid and cormorant fossils (approximately 35 and 39 my, respectively; Olson, 1985; Benton, 1993). Thus, the transversion rate for mammals is consistent with available fossil data for sulids. Using this rate, the number of transversions between sequences of Abbott’s booby and the gannets suggests that these two lineages diverged ,9 my after they diverged from the other boobies, approximately 14 mya. Unfortunately, too few transversions have accumulated either among the Sula boobies or among the gannets for reliable estimates of divergence dates based on transversions only, and no fossil dates are available for calibrating total substitution rates for recent divergences within the Sulidae. However, log–log regression of the total number of substitutions against the number of transversions for pairwise comparisons of sequences suggests that transversions occur at approximately 1/13th the rate of transitions (Y 5 1.18 1 0.55 X; r 5 0.92; P , 0.001), yielding an estimated rate of 2.8%/my for all substitutions. This is slightly higher than the rate of 2%/my suggested by Brown et al. (1982) and Shields and Wilson (1987) as an average substitu-

tion rate for mtDNA, but fits the pattern described by Birt et al. (1995) for differences in substitution rates detected by sequence analysis of cytochrome b versus restriction endonuclease analysis of whole mitochondrial genomes. Application of this value to sequence divergence data suggests that all extant lineages of boobies arose within the last 3 my, since the beginning of the Pliocene (Fig. 3), and that lineages leading to blue-footed and Peruvian boobies may have diverged only since the last interglacial period (although this estimate may be inflated since it does not account for intraspecific variation, which forms a significant component of the variation between species that diverged so recently). The three gannets also appear to represent recent divergences. Warheit (1992) argued that the period from 18 to 7 mya represented a period of high diversity within the Sulidae and that diversity declined from 7 to 4 mya; the present data suggest that only three lineages of sulids survived the decline in the late Miocene and are now diversifying again. Tests of Alternative Modes of Speciation Tree topology. Phylogenetic analysis of cytochrome b sequences produced a completely resolved tree for the Sulidae (Fig. 2). Thus, no inferences about modes of speciation could be made from the topology of the phylogenetic tree. Biogeography. Breeding distributions of the three species of gannets are entirely disjunct (Fig. 4; although wintering ranges overlap to varying extents), and therefore are inconsistent with parapatric or sym-

FIG. 3. Approximate time-frame for speciation events within the Sulidae (see text for explanation). Numbers on branches are numbers of unambiguous changes in skeletal/ecological/ethological characters within lineages (from MacClade).

PHYLOGENY AND EVOLUTION OF THE SULIDAE

FIG. 4.

Approximate breeding distributions of boobies and gannets (from Nelson, 1978).

patric modes of speciation unless extensive changes in breeding distributions occurred following speciation (Table 2). Blue-footed and Peruvian boobies are parapatric in distribution (Fig. 4); thus, parapatric speciation is the most parsimonious model for these lineages. TABLE 2 Evidence Inconsistent with Alternative Modes of Speciation within the Sulidae Mode of speciation Evidence Cape gannet/Australasian gannet Breeding distributions Vicariant events Northern gannet/ (Cape, Australasian gannets) Breeding distributions Vicariant events Abbott’s booby/gannets Vicariant events Comparative biology Peruvian booby/bluefooted booby Breeding distributions Vicariant events Gannets, Abbott’s booby/Sula boobies Vicariant events

257

Allopatric Peripatric Parapatric Sympatric

x

x

x

x

With the exception of red-footed and Peruvian boobies, breeding distributions of the remaining species of Sula boobies overlap to varying extents. However, breeding ranges of sulids are positively correlated with lineage age (Pearson correlation, r 5 0.72; P , 0.05; Fig. 5), suggesting that changes in breeding ranges since speciation cannot be ignored for masked, brown, red-footed and Abbott’s boobies; inferences about modes of speciation based on breeding distributions for these lineages therefore would be tenuous. Vicariance events. Sequence data suggest that most speciation events leading to the extant sulids occurred during or after the Pliocene (Fig. 3). No major barriers to dispersal occurred either within the contemporary breeding ranges of blue-footed and Peruvian boobies (which often mix in the Gulf of Panama) or between the ranges of the gannets during this period. Thus, allopat-

x

x

x x

x x

(x)

(x)

x

x

x

Note. Parentheses indicate that results are tentative (see text).

FIG. 5. Plot of approximate breeding range of sulids (distance between most distant colonies) versus lineage age (estimated from sequence data). Species abbreviations are as described in the legend of Fig. 1.

258

FRIESEN AND ANDERSON

ric speciation may be rejected for these lineages. Changes in sea level associated with climatic oscillations of this period would have resulted in periodic emergence of the Sunda and Arafura Shelves between Thailand, New Guinea, and Australia, effectively isolating populations of tropical seabirds in the Indian and Pacific Oceans. Emergence of the Panamanian Landbridge 3 mya would have isolated Atlantic and Pacific populations. Whether these barriers provide satisfactory vicariant events for the speciation events leading to red-footed, masked and brown boobies is difficult to assess given their current, overlapping pan-tropical distributions (Fig. 4; Table 2). Fossil evidence and/or sequence data suggest that gannets (including Abbott’s booby) and Sula boobies diverged ,23 mya, and that Abbott’s booby diverged from the gannets ,14 mya; these divergence dates do not correspond with any major vicariant events that would have subdivided their ancestral species. Thus, vicariance probably does not provide a satisfactory mechanism for most speciation events within the Sulidae. Comparative predictions. Numbers of unambiguous changes in skeletal, ethological, and combined characters correlated positively with lineage age (r 5 0.85, P , 0.001; r 5 0.52, P , 0.05; and r 5 0.81, P , 0.001, respectively; Figs. 3 and 6); the relationship for ecological characters did not attain significance, probably because of the small number of characters (N 5 5). Numbers of changes in skeletal, ethological, and combined characters in the lineage leading to gannets were significantly greater than predicted by the regression equation (standard residuals 5 2.85, 2.56, and 2.82, respectively, P , 0.01; Figs. 3 and 6); this is inconsistent with allopatric speciation, and suggestive of parapatric, or possibly sympatric or peripatric processes, in this lineage (Table 2).

FIG. 6. Numbers of unambiguous changes in skeletal, ecological, and ethological characters combined versus lineage age (estimated from sequence data). The arrow indicates the data point for the lineage leading to gannets. The diagonal line represents the regression line.

Inferred Modes of Speciation Breeding distributions of the three species of gannets are inconsistent with parapatric and sympatric models of speciation, and the absence of major vicariant events within the breeding distribution and time-frame of speciation in this group is incompatable with allopatric speciation (Table 2). Peripatry therefore is the most parsimonious model of speciation for the gannets. Thus, about 2.0 mya, during the climatic oscillations of the Pliocene, the ancestral gannet appears to have split into northern and southern lineages. This was accompanied or followed by adaptation of the lineages to different oceanographic regimes (Nelson, 1978). Approximately 0.6 mya, possibly during the Kansan glacial period, the southern lineage diverged into African and Australasian populations. This scenario contrasts with that of Nelson (1978), who proposed that the Cape gannet arose from the Australasian gannet, and that the northern gannet was derived from the Cape gannet. The timing of speciation events and results of the comparative analyses are inconsistent with any mode of speciation except for parapatry for the divergence of Abbott’s booby and the gannets, although peripatric or sympatric processes may have been involved if they were associated with strong selection (or drift in small populations; Table 2). The fact that boobies are all tropical in distribution suggests that the temperate ranges of the gannets represent a derived condition. Accordingly, the divergence date suggested by sequence data for these lineages corresponds with the period (,15 mya) when the latitudinal thermal gradient in the oceans became marked (Warheit, 1992). Divergence of these lineages therefore may have involved evolution of a northern population of boobies into a form adapted to temperate regions through either parapatric or peripatric processes. Breeding distributions are inconsistent with allopatric, peripatric, or sympatric speciation for the lineages leading to Peruvian and blue-footed boobies unless ranges have changed within the ,0.2 my since they diverged (Table 2). The absence of major vicariant events within the distribution and time-frame of divergence of these lineages also is inconsistent with allopatric speciation. Parapatric speciation therefore is the most parsimonious model for this event, although peripatric processes also are possible if speciation was followed by even slight increases in breeding ranges. Thus, within the last 0.2 my, possibly during the Sangamonian Interglacial, an ancestral booby appears to have diverged parapatrically (or peripatrically) into a northern, blue-footed form and a southern, Peruvian form. The fact that the two species meet in northern Peru but do not hybridize (Nelson, 1978) indicates that premating barriers are well established, despite the short time since divergence. Given the absence of a suitable vicariant event within the time-frame of divergence of the Sula boobies

259

PHYLOGENY AND EVOLUTION OF THE SULIDAE

from Abbott’s booby and the gannets, allopatric processes probably were not involved. However, no other conclusions can be drawn regarding mechanisms of speciation within the family due to the likelihood of changes in breeding distributions of the species (Fig. 5).

(8) Sexual advertising display: 0, none; 1, ‘‘headshaking’’; 2, ‘‘sky-pointing’’; 3, ‘‘flutter wing-wave.’’

Conclusions

(11) Ritualized locomotion: 0, absent; 1, ‘‘foot-drooping’’; 2, ‘‘parading.’’

Although the mode of speciation for several divergence events within the Sulidae could not be inferred, some general conclusions are apparent. Most notably, allopatric processes could be dismissed for five of the eight speciation events involved in the evolution of the extant species. This finding contrasts with the generally accepted view that speciation in birds involves primarily allopatry (e.g., Chesser and Zink, 1994). Sympatry also did not appear to explain any of these speciation events; most or all apparently involved parapatric or peripatric processes. Population level genetic surveys, as well as the development of new tests, should provide further insight into mechanisms of speciation in the Sulidae. ACKNOWLEDGMENTS We thank A. Berrutti, N. Brothers, J. Jahncke, and B. Tershie for donation of blood samples, L. Lougheed and Y. Sherman for assisting in sample collections, and J. B. Nelson for donation of feathers from Abbott’s boobies. H. Jones was instrumental with laboratory analyses. T. Birt, B. Congdon, and K. Warheit provided insightful discussions. This work was supported by research grants from NSERC (WFA0156580 to V.L.F.), N.S.F. (DEB93-04579 to D.J.A.), and the Wake Forest University R&P Fund. We thank the Charles Darwin Research Station for logistical support, and the Galapagos National Park Service for permits during the Galapagos field work.

APPENDIX 1 Skeletal characters that were used in comparative analyses were obtained from Warheit (1990); 16 characters (HUM14, HUM15, ULN1, ULN5, COR7, COR11, STN6, CMC1, TTR8, TTR10, TMT7, SKL1, SKL13, RAD1, RAD2, and RAD3) were not included since character states could not be assigned unambiguously to taxa. Ecological and ethological characters were derived from Nelson (1970, 1978) and personal observations: (1) Nesting habitat: 0, tree; 1, ground. (2) Nesting density: 0, low; 1, high. (3) Overt fighting: 0, absent; 1, present. (4) ‘‘Wing-flailing’’ display: 0, absent; 1, present. (5) Ritualized threat display: 0, absent; 1, present. (6) Site ownership display: 0, absent; 1, ‘‘bowing’’; 2, ‘‘yes head-shaking’’; 3, ‘‘forward head-waving,’’ head high; 4, ‘‘forward head-waving,’’ head low; 5, ‘‘yes-no’’ head-shaking; 6, ‘‘head-jerking.’’ (7) Meeting ceremony: 0, absent; 1, ‘‘mutual fencing’’; 2, ‘‘mutual jabbing’’; 3, ‘‘mutual wing-waving.’’

(9) Premovement posture: 0, absent; 1, ‘‘sky-pointing’’; 2, ‘‘bill-up, face-away.’’ (10) Parallel standing: 0, absent; 1, present.

(12) Wing-shaking: 0, absent; 1, ‘‘wing-rattle’’; 2, ‘‘wing-flick.’’ (13) Ritualized flying-in to nest: 0, absent; 1, ‘‘saluting’’; 2, ‘‘ ‘V’ flight.’’ (14) Copulation: 0, without calling or biting; 1, with biting; 2, with calling. (15) Display of nest material: 0, absent; 1, ritualized display. (16) Nest building: 0, ordinary; 1, ritualized display. (17) Appeasement behavior: 0, ritualized sleeping; 1, ‘‘facing away.’’ (18) Preening: 0, absent; 1, ritualized autopreening. (19) Allopreening: 0, absent; 1, reciprocal; 2, unilateral. (20) Infantile bill-hiding: 0, absent; 1, present; 2, reduced. (21) Bill tucking display: 0, absent; 1, present. (22) Foraging site: 0, offshore; 1, inshore. (23) Clutch size: 1, one; 2, two; 3, three; 4, four or more. (24) Brood reduction: 0, absent; 1, obligate; 2, facultative. (25) Breeding distribution: 0, temperate-subarctic; 1, tropical-subtropical. Characters 4, 6–9, 11, 16, 18, 23, 25, 27, 28, and 36 from Kennedy et al. (1996) also were included; other characters from their analysis were not included either because of monomorphism within the Sulidae or because of overlap with characters from Nelson (1970).

Character Species

1111111111222222 1234567890123456789012345

Northern gannet Cape gannet Australasian gannet Abbott’s booby Masked booby Brown booby Blue-footed booby Red-footed booby Peruvian booby Pelagic cormorant

1110111110110100101100100 1110111110110100101100100 1110111110110100101100100 0000063030020010010200101 1011052221222101101100211 1011032221220110102101211 1011022221221111102101221 0011042220043310101000101 1111122220221101102101321 1000000300003000000010400

260

FRIESEN AND ANDERSON

REFERENCES Benton, M. J. (Ed.). (1993). ‘‘The Fossil Record 2,’’ Chapman and Hall, New York. Birt, T. P., Friesen, V. L., Birt, R. D., Green, J. M., and Davidson, W. S. (1995). Mitochondrial DNA variation in Newfoundland and Norwegian populations of Atlantic capelin, Mallotus villosus, detected using two techniques. Mol. Ecol. 4: 771–776. Brooks, D. R., and McLennan, D. A. (1991). ‘‘Phylogeny, Ecology and Behaviour: A Research Program in Comparative Biology,’’ Univ. Chicago Press, Chicago. Brown, W. M., Prager, E. M., Wand, A., and Wilson, A. C. (1982). Mitochondrial DNA sequences of primates: Tempo and mode of evolution. J. Mol. Evol. 18: 225–239. Butlin, R. (1989). Reinforcement of premating isolation. In ‘‘Speciation and Its Consequences’’ (D. Otte, and J. A. Endler, Eds.), pp. 158–179, Sinauer, Sunderland, MA. Bush, G. L. (1994). Sympatric speciation in animals: New wine in old bottles. TREE 9: 285–288. Carr, S. M., and Griffith, O. M. (1987). Rapid isolation of animal mitochondrial DNA in a small fixed-angle rotor at ultrahigh speed. Biochem. Genet. 25: 385–390. Chesser, R. T., and Zink, R. M. (1994). Modes of speciation in birds: A test of Lynch’s method. Evolution 48: 490–497. Diehl, S. R., and Bush, G. L. (1989). The role of habitat preference in adaptation and speciation. In ‘‘Speciation and Its Consequences’’ (D. Otte and J. A. Endler, Eds.), pp. 345–365, Sinauer, Sunderland, MA. Edwards, S. V., Arctander, P., and Wilson, A. C. (1991). Mitochondrial resolution of a deep branch in the geneological tree for perching birds. Proc. R. Soc. Lond. B 243: 99–107. Endler, J. A. (1977). ‘‘Geographic Variation, Speciation, and Clines,’’ Princeton Univ. Press, Princeton. Felsenstein, J. (1985). Confidence limits on phylogenies: an approach using the bootstrap. Evolution 39: 783–791. Friesen, V. L., Baker, A. J., and Piatt, J. F. (1996). Phylogenetic relationships within the Alcidae (Charadriiformes: Aves) inferred from total molecular evidence. Mol. Biol. Evol. 13: 359–367. Gibbons, A. (1996). On the many origins of species. Nature 273: 1496–1499. Hedges, S. B., and Sibley, C. G. (1994). Molecules vs. morphology in avian evolution: The case of the ‘‘pelecaniform’’ birds. Proc. Natl. Acad. Sci. USA 91: 9861–9865. Howell, N. (1989). Evolutionary conservation of protein regions in the proton-motive cytochrome b and their possible roles in redox catalysis. J. Mol. Evol. 29: 157–169. Irwin, D. M., Kocher, T. D., and Wilson, A. C. (1991). Evolution of the cytochrome b gene of mammals. J. Mol. Evol. 32: 128–144. Kennedy, M., Spencer, H. G., and Gray, R. D. (1996). Hop, step and gape: Do social displays of the Pelecaniformes reflect phylogeny? Anim. Behav. 51: 273–291. Kimura, M. (1980). A simple method for estimating evolutionary rate of base substitutions through comparative studies of nucleotide sequences. J. Mol. Evol. 16: 111–120. Kocher, T. D., Thomas, W. K., Meyer, A., Edwards, S. V., Pa¨a¨bo, S., Villablanca, F. X., and Wilson, A. C. (1989). Dynamics of mitochon-

drial DNA evolution in animals: Amplification and sequencing with conserved primers. Proc. Natl. Acad. Sci. USA 86: 6196–6200. Kornegay, J. R., Kocher, T. D., Williams, L. A., and Wilson, A. C. (1993). Pathways of lysozyme evolution inferred from the sequences of cytochrome b in birds. J. Mol. Evol. 37: 367–379. Kumar, S., Tamura, K., and Nei, M. (1993). ‘‘MEGA: Molecular Evolutionary Genetics Analysis,’’ version 1.0, Penn. State Univ., University Park. Lynch, J. D. (1989). The gauge of speciation: on the frequencies of modes of speciation. In ‘‘Speciation and its Consequences’’ (D. Otte and J. A. Endler, Eds.), pp. 527–553, Sinauer, Sunderland, MA. Maddison, W. P., and Maddison, D. R. (1992). ‘‘MacClade,’’ version 3.0, Sinauer, Sunderland. Mayr, E. (1970). ‘‘Populations, Species and Evolution,’’ Harvard Univ. Press, Cambridge. Nelson, J. B. (1970). The relationship between behaviour and ecology in the Sulidae with reference to other sea birds. Oceanogr. Mar. Biol. Ann. Rev. 8: 501–574. Nelson, J. B. (1978). ‘‘The Sulidae: Gannets and Boobies,’’ Oxford Univ. Press, Oxford. Olson, S. L. (1985). A selective synopsis of the fossil record of birds. In ‘‘Avian Biology’’ (D. S. Farner, J. R. King, and K. C. Parkes, Eds.), Vol. 8, pp. 79–238. Academic Press, New York. Olson, S. L., and Warheit, K. I. (1988). A new genus for Sula abbotti. Bull. Br. Ornithol. Club 108: 9–12. Rzhetsky, A., and Nei, M. (1992). A simple method for estimating and testing minimum-evolution trees. Mol. Biol. Evol. 9: 945–967. Rzhetsky, A., and Nei, M. (1993). Theoretical foundation of the minimum-evolution method of phylogenetic inference. Mol. Biol. Evol. 10: 1073–1095. Saitou, N., and Nei, M. (1987). The neighbor-joining method: A new method for reconstructing phylogenetic trees. Mol. Biol. Evol. 4: 406–425. Seutin, G., White, B. N., and Boag, P. T. (1991). Preservation of avian blood and tissue samples for DNA analyses. Can. J. Zool. 69: 82–90. Shields, G. F., and Wilson, A. C. (1987). Calibration of mitochondrial DNA evolution in geese. J. Mol. Evol. 24: 212–217. Sibley, C. G., and Ahlquist, J. E. (1990). ‘‘Phylogeny and Classification of Birds,’’ Yale Univ. Press, New Haven. Slatkin, M. (1996). In defense of founder-flush theories of speciation. Am. Nat. 147: 493–505. Swofford, D. L., and Begle, D. P. (1993). ‘‘PAUP: Phylogenetic Analysis Using Parsimony,’’ Version 3.1, Illinois Natural History Survey, Champaign. Warheit, K. I. (1990). ‘‘The Phylogeny of the Sulidae (Aves: Pelecaniformes) and the Morphometry of Flight-Related Structures in Seabirds: A Study of Adaptation,’’ Ph.D. thesis, Univ. California, Berkeley. Warheit, K. I. (1992). A review of the fossil seabirds from the Tertiary of the North Pacific: Plate tectonics, paleoceanography, and faunal change. Paleobiology 18: 401–424. Wyles, J. S., Kunkel, J. G., and Wilson, A. C. (1983). Birds, behavior, and anatomical evolution. Proc. Natl. Acad. Sci. USA 80: 4394– 4397.