THE SIGNIFICANCE OF HETEROCHRONY TO THE EVOLUTION OF HISPANIOLAN
PALM-TANAGERS, GENUS PHAENICOPHIL US: BEHAVIORAL, MORPHOLOGICAL
AND GENETIC CORRELATES
MARA ALESSANDRA MCDONALD
A DISSERTATION PRESENTED TO THE GRADUATE SCHOOL
OF THE UNIVERSITY OF FLORIDA IN PARTIAL FULFILLMENT
OF THE REQUIREMENTS FOR THE DEGREE OF
DOCTOR OF PHILOSOPHY
UNIVERSITY OF FLORIDA
Mara Alessandra McDonald
To Margaret K. Langworthy, who has maintained her personal and professional integrity
through all these trying times, and to all women and men who remain ethical and devoted to
excellence in their scientific and personal endeavors.
Members of my committee who have kindly supported me during my graduate tenure
include Lincoln B. Brower, John Wm. Hardy, Larry D. Harris, Carmine A. Lanciani, and
Michael H. Smith. Bill Hardy edited the several versions of the dissertation and graciously
helped get all the paperwork together in my absence. I am particularly grateful to Mike
Smith for his input into the final analyses, for his moral and financial support, and for his
many hours of advice, help, and guidance.
Robert M. Zink, Curator of Frozen Tissues, Louisiana State University Museum of
Zoology, patiently provided me with material I requested for the systematic analysis; helped
me with some cold concepts in phylogenetic analysis; and warmly encouraged my work. Bob
Crawford, formerly of Tall Timbers Research Station in Tallahassee, Florida, provided many
of the warbler specimens. James Bond at the Philadelphia Academy of Sciences, Storrs Olson
at the U.S.
National Museum in Washington, D.C.,
Kenneth Parkes at the Carnegie Museum
of Natural History in Pittsburgh, and Charles Sibley now at San Francisco State University,
made significant contributions through their writings, correspondence, and discussions.
Mercedes Foster of the U.S.
National Museum and Marcy and Robert Lawton of the
University of Alabama in Huntsville, pioneers in the embryonic field of avian heterochrony,
have provided me with the theoretical and empirical foundation for pursuing my research.
I can never repay James M. Novak at the Savannah River Ecology Laboratory (SREL)
for all the time he has taken to discuss minor points or major issues, both conceptual and
statistical. His patience in teaching me computer and other essential skills helped me
Philip M. Dixon at SREL for their moral and intellectual support.
They accepted and
encouraged me under all circumstances and provided a challenging, but supportive
atmosphere in which to work. I owe much to my friends and teachers at the University of
Florida, including Mike Binford, Mark Brenner, Lincoln Brower, Kris Brugger, Peter
Feinsinger, Rob Ferl, Larry Harris, Susan Jacobson, Rich Kiltie, Carmine Lanciani,
Margaret Langworthy, Brian McNab, Frank Nordlie, Theresa Pope, Jon Reiskind, John
Robinson, and to Bill Kilpatrick at the University of Vermont.
I am indebted to the Organization of American States for financial support while in
Haiti and to Ragnar Arneson, Roland Roi, and the staff in Port-au-Prince who helped make
my stay in Haiti a success. I am also indebted to the Chapman Memorial Fund of the
American Museum of Natural History, Sigma Xi Grants-in-Aid, the Oak Ridge Associated
Universities Program, the Department of Zoology at the University of Florida, and to the
University of Georgia Savannah River Ecology Laboratory for financial assistance. I thank
Morris and Cecilia Maizels for their friendship and financial support.
I thank the many Government of Haiti officials including Edmond Magny, Florence
Sergile, Joseph Felix, Jean-Edner Francois and Paul Paryski, who spent much time and effort
in making my field work possible. I owe much to Jean-Phillipe Audinet, Robert Cassagnol,
Stephan Dix, Santa and Scott Faiia, Joyce Flores, Tom Greathouse, Jim Keith, Susana
Molnar, Pierre-Yves Roumain, Jim Talbot, and John Thorbjarnarson, for all their help,
support, guidance and encouragement while working in Haiti.
Without the cordial help of the
various office personnel, including Janet Ziegler, Carole Binello, Grace Kiltie, Rhoda Bryant
at the University of Florida and JoAn Lowery, Miriam Stapleton and Jan Hinton at SREL,
many of the tedious little tasks could not have been finished. I thank Jean B. Coleman at
SREL for advice on the illustrations.
Vertebrate Zoology, was instrumental in providing me with the conceptual tools to undertake
my dissertation research. His lecture on heterochrony in 1978 set the stage for my
subsequent interest in the subject and his excellent teaching provided me with the conceptual
framework necessary to pursue questions in evolutionary biology. I am also deeply grateful to
Michael H. Smith, Director of the University of Georgia Savannah River Ecology Laboratory .
Although his very broad interests did not include birds, heterochrony, Hispaniola, feminism,
or systematics, he committed himself to a critical evaluation of my work and provided
significant input into the generation of hypotheses and the formulation of ideas. His
influence significantly improved the anal
yses and presentation of my data; his tenacity pulled
me through some bleak moments. I have the highest esteem for the dedication, critical
thinking, broad perspectives, and integrity of these two gentlemen.
This research was supported under contract DE-AC09-76SROO-819 between the U.S.
Department of Energy and the University of Georgia's Savannah River Ecology Laboratory.
TABLE OF CONTENTS
ACKNOW LEDGMENTS . . . .. . ........................
LIST O F TA BLES . . .. . . . . . . .. .. .. .
LIST O F FIG U RES ...............................................................
A B ST R A C T .....................................................................
Purpose and Definitions . . . . . . .
Hispaniolan Palm-Tanagers . . . . . . . . . . .
The Significance of Heterochrony and Paedomorphosis
HISPANIOLAN PALM-TANAGERS: BEHAVIORAL AND
MORPHOLOGICAL CONSEQUENCES OF HETEROCHRONY
Methods and Materials
R results . . . . . . . . . . . .
D discussion ................................................................
HISPANIOLAN PALM-TANAGERS: THE GENETIC CONSEQUENCES OF
H ETEROCH RON Y . . . . .....................................
MI ethod s ........ . . . . ......... . . .............. ..
R es lts . . . . . . . ................................ ...
D iscussion . . . . . . . . ................... ..
x-i& -o x i .*** .*^ .**************.
BIOCHEMICAL SYSTEMATICS OF HISPANIOLAN PALM-TANAGERS
Introduction . . . . .......................
CONCLUSIONS AND FUTURE DIRECTIONS
DESCRIPTION OF MORPHOLOGICAL MEASUREMENTS
ILLUSTRATION OF PLUMAGE CHARACTERS MEASURED ON
LIST OF SUBSTRATES USED BY HISPANIOLAN PALM-TANAGERS
HORIZONTAL SUBSTRATE USE OF HISPANIOLAN PALM-
LITERATU RE CITED ..........................................................
BIOGRAPHICAL SKETCH .....................................................
LIST OF TABLES
Principal Components for seventeen morphological variables measured in
Hispaniolan palm-tanagers . . . . . . . . . . .
Mean, sample size, and significance of differences of 8 foraging
behavior variables . . . . . . . . . .
List of gray-crowned Black-crowned Palm-Tanagers collected in late winter or
the earlypart of the breeding season (late March to early April
Estimates of genetic variability for Hispaniolan palm-tanagers and their hybrids
for 39 enzym e loci .......................................................
Mean % heterozygosity for adult and juvenile Black-crowned palm-tanagers
after Jackknife simulations (Lanyon, 1987) for 13 variable loci ...........
List of species, number of specimens, and genetic variability for 25 species of
Emberizidae across 20 assayed enzyme loci .............................
Allelic designations and frequencies for 20 loci used in the systematic analysis
25 species of Emberizidae
1978) unbiased genetic distance (above diagonal) and modified Rogers'
distance (Wright, 1978) (below diagonal) for 25 taxa used in the systematic
analysis . . . . . . . . . . . . . . .
Results of Jackknife procedure on the stability of phylogenetic affinities of tanagers
and warblers based on 20 electrophoretic loci ................................
Manhattan distance matrix of morphological features used in comparison of 15
tanagers and warblers . . . . . . . . . . . .
. . . . .. 33
LIST OF FIGURES
Map of the current distribution of Hispaniolan palm-tanagers
Three-dimensional graph of morphological variables that separate Hispaniolan
palm -tanagers . . . . . . . . . . . . . .
Unweighted Pair-group Method of Averaging (UPGMA) foraging behavior
phenogram for six species of birds found in Haiti .............................
Frequency of foraging behavior on each substrate types
pine, braodleaf tree,
and shrub below 3 m in height) observed in mixed pine habitat for adult
(BPA) and juvenile (BPI) Black-crowned Palm-Tanagers and Gray-crowned
Palm-Tanagers (GPT) .. . ................ . .
Expected heterozygosity (He) of colonists based on founding population size
(No) and observed heterozygosity (Ho) in the antecedent population .......
Unweighted Pair-group Method of Averaging (UPGMA) phenogram (A) and
Distance Wagner tree (B) based on modified Rogers' genetic distance
(Wright, 1978) across 20 loci for 25 Emberizidae species ..............
Unweighted Pair-group Method of Averaging (UPGMA)phenogram based on
21 plumage and skeletal features measured on 15 species of Emberizidae .
Abstract of Dissertation Presented to the Graduate School
of the University of Florida in Partial Fulfillment of the
Requirements for the Degree of Doctor of Philosophy
THE SIGNIFICANCE OF HETEROCHRONY TO THE EVOLUTION OF HISPANIOLAN
PALM-TANAGERS, GENUS PHAENICOPHIL US: BEHAVIORAL, MORPHOLOGICAL
AND GENETIC CORRELATES
MARA ALESSANDRA MCDONALD
Chairperson: John William Hardy
Cochairperson: Michael Howard Smith
Major Department: Zoology
Heterochrony, or changes in developmental timing, may explain the rapid speciation
and diversification of avian taxa. The purpose of this study is to investigate the significance of
heterochrony to the evolution of Hispaniolan palm-tanagers, genus Phaenicophilus, using
behavioral, morphological, and genetic characteristics at the population and systematic
levels. Phaenicophilus palmarum is characterized by a significant age-dimorphism in
foraging behavior, morphology, and genetic variability, as detected by starch gel
electrophoresis. Significantly higher levels of genetic variability were observed for juveniles
(H= 0.121) than for adults (H
= 0.074) across 13 of 17 variable loci in the sample of 39 assayed
1 f .
is its sibling species. Levels of genetic variability observed in P. poliocephalus
(H= 0.104) are not significantly different from either age-class of P. palmarum. A colonization
model, using observed heterozygosity of juvenile P. palmarum and varying founding
population sizes and compositions, with subsequent rapid reproduction after founding
supports the hypothesis that a small number of founders, composed mostly of juvenile P.
palmarum, could explain the high levels of genetic heterozygosity observed in current
populations of P. poliocephalu
s. The smaller body size, different social structure, retention of
juvenile behavior and morphology, and absence of a long-term age-dimorphism in relation to
P. palmarum suggest that P. poliocephalus is paedomorphic to P. palmarum.
Biochemical systematic analyses were used to determine the polarity of evolution of
the two species of Phaenicophilus. Phaenicophilus was aligned most frequently with Piranga
on Distance Wagner trees. Furthermore, average genetic distance was lower and the number
of shared alleles was higher between Phaenicophilus and Piranga than for other comparisons.
This result is consistent with the hypothesis that Phaenicophilus poliocephalus is most likely
derived from P. palmarum and paedomorphic to it.
Purpose and Definitions
The purpose of this research was to investigate the importance of heterochrony to the
evolution of two species of Hispaniolan Palm-Tanagers, genus Phaenicophilus, using
behavioral, morphological and genetic characteristics.
Population and systematic analyses
were employed to determine how the palm-tanagers were related to one another and to other
species. Heterochrony is the selective modulation of patterns of development" (Larson,
1980:1) and can be divided into two processes: recapitulation, the retention of ancestral adult
features in the early ontogeny of the descendant species, and paedomorphosis, defined on a
as the retention of ancestral juvenile features in the descendant species
(Gould, 1977; Alberch etal., 1979).
Use of the term "juvenile" in this discussion refers to
retained morphological features and not the state of sexual maturation. Paedomorphosis has
been invoked to explain the adaptive radiation of plethodontid salamanders (Wake, 1966;
Larson, 1980) and to characterize salamander populations that delay metamorphosis (Gould,
1977). Its significance as a microevolutionary process leading to macroevolutionary change
has only recently been appreciated for other taxa, such
Lawton, 1986; Foster
birds (Gould, 1977; Lawton and
't n "i n H fin n C./1 I 4- nrjr r^ a n< ,^ ,. ^ ,. ,f .n 4 LI; C. nfl 1 n 'T t n I rt aK n. iL al i n r n a a. up A
palmarum. Phaenicophilus poliocephalus resembles the juvenile of P. palmarum in crown
This resemblance suggested that heterochrony might be a process important to the
evolution of these species.
To determine the significance of heterochrony to the evolution of
Phaenicophilus, it is first necessary to reexamine the species designations of the two forms
and establish that they are not geographic variants of a single species. Both morphological
and genetic criteria are used in the analysis. Second, alternative hypotheses to explain the
similarity in morphology between the two species need to be compared to the hypothesis that
invokes heterochrony. These alternative hypotheses include those that consider character
divergence and convergence. Third, if heterochrony is involved, it is essential to know which
Phaenicophilus species was derived and which was antecedent.
The answer to this question
depends upon a knowledge of phylogenetic relationships among tanagers. Finally, changes in
behavior and body size, which correspond to the retention of juvenile morphology, can be used
as further support for the importance of heterochrony. If predictions from a model of
heterochrony by paedomorphosis are consistent with the observed patterns of foraging
behavior, social structure, body size and other morphological characters, and genetics,
this would be strong support for heterochrony
as the most parsimonious model.
The interpretation of the speciation of the two forms of Phaenicophilus
is dependent on
the distributions in Hispaniola. Early ornithologists described the eastern limit of P.
poliocephalus as extending no farther than the western edge of the Trouin Valley
,1931; Bond, 1980; Fig. 1-1). However, sightings of P. poliocephalus in the
Dominican Republic (Dod, 1981) suggest that earlier reports were either incorrect or that
changes in the distributions have occurred. Bond (pers. comm.) originally determined the
distribution of P. poliocephalus by spending a few hours in the xeric northern part of the
in this habitat (pers. obs.). Wetmore and Swales (1931) based the description of the
distribution ofPhaenicophilus at least partially on Bond's observations made in 1928. Fifty
years later, I observed P. poliocephalus in mesic habitat around the village of Trouin but not
in the xeric habitat of the Trouin Valley.
There are no published ornithological records describing the area south of the Massif de
La Selle or Cap Rouge.
The major collecting efforts occurred before the end of American
occupation of Haiti in 1930; since then there have been few ornithological studies in the
country. In the last half-century, significant environmental changes have occurred
throughout Haiti including extensive clearing of lands.
The resulting habitat disturbance
probably affected the distribution of the palm-tanagers. Habitat disruption and subsequent
contact between the species could have resulted in hybridization and called into question
their species status (Bond, 1986).
Finally, the sightings of purported P. poliocephalus within
the range of P. palmarum in the Dominican Republic suggested that P. poliocephalus might
simply be a geographic variant of P. palmarum. Consequently, the relationship between the
two forms required further investigation because of the new information on the distribution of
the black- and gray-crowned phenotypes.
Phylogeny of Tanagers.
The relationship between two forms can often be better understood when viewed from a
broader phylogenetic perspective. The affinities of Phaenicophilus to other tanagers are not
clear. Sibley (pers. comm.), using DNA-DNA hybridization techniques, regarded
Phaenicophilus as a sister group to the genus Piranga, which includes the North American
scarlet tanager, P. olivacea, summer tanager, P. rubra, and the western tanager, P.
ludoviciana. Sibley did not include species that might be considered closely allied to the
palm-tanagers based on external morphology.
The common bush-tanager, Chlorospingus
Microligea palustris and Xenoligea montan are so similar in appearance to Phaenicophilus
that their local name,
"Petite Kat-je" translates "Little Four-eyes," the diminutive form of the
Creole name given to the palm-tanagers. Hispaniolan warblers are about the
bush-tanagers, and both groups have darkish gray crowns. In addition, the warblers have
white eye-spots similar to those of the palm-tanagers.
If a clear sister relationship could be
shown between Chlorospingus-Microlige and Phaenicophilu, then the polarity of the
Phaenicophilus derivation might be more easily determined using morphological features.
The establishment of this relationship would allow a test of the hypothesis that the
Chlorospingus-Microligea-Phaenicophilu group forms a paedomorphic assemblage.
The Significance of Heterochrony and Paedomorphosis
Heterochrony depends on the evolution of regulatory genes. Small changes in
regulation can alter the growth rate of the organism, simultaneously changing suites of
characters not directly selected upon
Neo-Darwinian theory assumes that
macroevolution occurs by the accumulation of large numbers of changes across the genome
over long time periods. Heterochrony
is an alternative hypothesis to gradualism; rapid
evolutionary change can result with fewer genetic changes
Larson, 1980). There are several
that can result in heterochronic change in lineages (Gould, 1977
terminal additions, deletions, or substitutions, and nonterminal additions, deletions or
substitutions of morphological characters in the adult
vs. juvenile life history stage.
Terminal additions result in the evolution of novel features and are equivalent to
Terminal deletions are equivalent to paedomorphosis and result in the
retention of juvenile features of the ancestor in the adult of the descendant species (Alberch et
al., 1979; Kluge and Strauss,
1985). Juvenile features can be retained either by delay of
Before predictions from a model of heterochrony can be made, the polarity of
evolutionary relations (i.e., which species is derived from which) must be known. The problem
is to detect whether a terminal deletion or terminal addition has occurred; it may be resolved
by the use of outgroup analysis (Kluge and Strauss, 1985). Outgroup analysis involves the
choice of a taxon (i.e., sister group) that is closely related to the taxa being compared to
determine whether a character is common or unique to one or more of the taxa examined. If a
character is shared by one but not the other taxon with the sister group, the first taxon is most
likely antecedent to the second.
Heterochrony provides an alternative mechanism for macroevolution to occur without
major genetic changes. Paedomorphosis may result in rapid evolutionary diversification,
parallelisms, and convergences simply by changing ontogenetic sequences for one or more
The degree of concordance between genetic and morphological characters in
species assemblages is expected to be low because convergence may obscure the relationships.
Paedomorphic changes may occur in only one or a few characters producing mosaic evolution.
Congruent changes over several character states of the organism may be observed in closely
related species that have not diverged significantly from each other. Hispaniolan palm-
tanagers are closely related and may be ideal for the study of the importance of heterochrony
to their evolution.
Paedomorphosis was not expected to be a significant process in the evolution of birds
(Gould, 1977). Only recently was neoteny equated with delayed maturation in birds (Lawton
and Lawton, 1986; Foster, 1987), although delayed maturation had already been documented
for numerous bird species (Selander, 1965; Rohwer et al., 1980; Flood, 1984; Hamerstrom,
Common to delayed maturation hypotheses are several assumptions that provide the
critical link between delayed maturation and Gould's (1977) ecological constraints model to
intense intraspecific competition;
(4) some proportion of juveniles are able to breed, under
certain conditions, with the consequence that sexual maturation is decoupled from
morphological maturation; (5
the costs of sexual and somatic maturation are greater than
the increased fitness accrued to inexperienced or subordinate individuals that may
successfully breed; and (6) juveniles may benefit from cryptic or deceptive morphology
reducing predation and/or intraspecific competition (Selander, 1965; Rohwer, 1978;
1984; Lawton and Lawton, 1986; Foster,
1987). Consequently, delayed maturation, or
neoteny, results from a balance between selection to breed early and
longer to ensure successful breeding.
selection to survive
Under conditions of limiting resources, the relative
increase in fitness accrued by individuals breeding at an earlier age would be offset by the
decrease in fitness due to the inexperience of juveniles competing for scarce resources. For
example, younger-aged flocks
of Brown Jays, Cyanocorax morio, are less successful breeders
than their older-aged counterparts (Lawton and Lawton, 1985). In years of resource
abundance, the fitness differential would favor juvenile breeding.
In contrast, when species colonize new habitat, resource leve
may be relatively more
abundant. Under these circumstances, early sexual maturation is no longer constrained.
Brown Jay flocks moving into recently cleared habitat have a lower average age than flocks
observed in the older habitat (Lawton and Lawton, 1985).
With attainment of sexual
maturity, somatic development slows down or stops (Gould, 1977), resulting in individuals
with smaller body size, juvenile morphology, juvenile behaviors associated with the
morphology, and a concomitant increase in group behaviors (Geist, 1971
Lawton and Lawton, 1986). If the colonizing population becomes isolated from the main
stock, divergence can occur. It is unnecessary to invoke selection against adult phenotypes, if
rates of somatic development are tied to rates of sexual maturation, though the endpoints of
selection, can change in concordance with a selected character. For example, derived species
of mountain sheep, that have retained the juvenile morphologies of their antecedents, are
relatively more social than their antecedents (Geist, 1971).
This sociality is not a result of
selection for juvenile behaviors, but a consequence of retaining juvenile morphologies,
resulting in reduced intraspecific aggression (Lawton and Lawton, 1986).
Additional predictions can be made once the particular type of paedomorphosis can be
inferred, i.e., neoteny or progenesis. Both neoteny and progenesis achieve the same endpoint,
terminal deletion, but by different routes.
The mechanism for paedomorphic changes may be
controlled by changes in regulatory gene activity governing maturation rates. Sexual
maturation is accelerated in progenetic species, purportedly in response to abundant
resources. Progenetic species should breed earlier and be smaller than their antecedent
species; in addition, they should exist under density-independent conditions. Gould (1977)
invoked limiting resources as the necessary ecological constraint that leads to the evolution of
neoteny. Neoteny may be a function of temporal variability in resources; it should be favored
when resources are generally limiting but periods of limitations are sometimes relaxed.
Under crowded conditions, individuals delay sexual maturation with a concomitant delay in
The delay in maturation results in evolutionary change if sexual and
somatic maturation are decoupled (Lawton and Lawton, 1986).
The conditions for this
decoupling are associated with a high variance around first age to reproduction, resulting in
juveniles that breed in response to relaxation of resource limitations.
The ability to respond to fluctuating resources has genetic consequences as well. Gould
(1977) did not link changes in life history traits with genetic variation. Predictions for the
study of palm-tanagers concerning changes in genetic variability of paedomorphic systems
could not be made a priori, because patterns of genetic variability are not apparent for
heterozygosity was greater in habitats consisting of later successional stages and in higher
densities. Redfield (1973) concludes that colonizers (i.e., individuals in the earlier
successional habitat) are more homozygous than their counterparts in older habitat.
assumes that dispersal occurs into earlier successional habitat, and not throughout all
successional stages. It is unclear from Redfield's study whether heterozygous yearlings are
more successful than their homozygous counterparts in dispersing into older aged habitat,
due to behavioral dominance (Baker and Fox, 1978), or whether they result from selection for
heterozygotes in the age-class under stress. Heterozygotes may be favored under conditions of
limiting resources (Samollow and Soule, 1983: Smith, Teska and Smith,
1984). either due to
increased metabolic efficiency, superior competitive behavior, or a combination of both.
Genetic variability may be favored as a response to unpredictable resources.
variability may be higher in neotenic species, if neotenv is evolved in response to
To understand how heterochrony influences the evolution of groups,
general approaches. First, species assemblages, characterized by
due to developmental changes, can be studied.
one can take two
similarities in morphology
This approach was successfully applied to
plethodontid salamanders (Wake,
1966; Larson, 1980; Alberch, 1981).
concordance between morphological and genetic evolution is likely to be obscured by
convergences, parallelisms, and mosaic evolution in these assemblages because of significant
amounts of differentiation.
The particular ecological constraints that may eventually lead to
evolutionary change are lost in broad comparisons such as these.
The second approach focuses on population level differences of a few closely related
species. Recent ancestry will avoid obscuring the relationships of several character sets
between these species. Ideally, such species should exist in similar habitats so that selective
geographic areas imposed by island boundaries may provide less opportunity for confounding
geographic variation. Increased isolation from the the mainland results in fewer competitive
interactions that might confound patterns based on heterochrony.
Thus, Hispaniolan palm-
tanagers would seem to be suitable for the study of heterochrony because they are closely
related island endemics, and exhibit morphological patterns of resemblance that suggest
heterochrony is involved.
Figure 1-1. Map of the current distribution of Hispaniolan palm-tanagers. Diagonal lines
represent the known distribution of Gray-crowned Palm-Tanagers; Black-
crowned Palm-Tanagers occur throughout the rest of Hispaniola. Area southwest
of Port-au-Prince, bounded by Marbial and Decoze and Furcy is undescribed.
Stippled areas define the Trouin Valley and the Cul-de-Sac Plain. Enlargement
features the area of contact and hybridization bounded by Fond Jean Noel (Fd Jn
Noel) in the east, Marigot on the Caribbean Sea to the south and includes a region
southwest of Seguin. The western extent of the hybrid zone is still undefined but
may extend into the valley between the Massif de la Selle and Cap Rouge.
BEHAVIORAL AND MORPHOLOGICAL CORRELATES OF HETEROCHRONY
Few attempts have been made to integrate alterations in developmental sequence with
evolutionary processes in birds. An understanding of heterochrony (i.e., shifts in
developmental sequences) should prove useful to such an integration. Heterochrony by
paedomorphosis, resulting in the retention of juvenile features in reproductively mature
individuals, can be achieved in two ways: progenesis and neoteny. Progenesis
characterized by early sexual maturation which often results in smaller body size, while
neoteny, or delayed somatic maturation results in larger body
Gould, 1977; Lawton and
Lawton, 1986), and increased variation to age of first reproduction. Selection for change in
one character often results in simultaneous change in nonselected but associated characters
(Larson, 1980). Juvenile behaviors
associated with juvenile morphologies can be retained in
breeding individuals due to paedomorphosis (Geist, 1971; Coppinger and Coppinger, 1982;
Lawton and Lawton, 1986; Coppinger et al., in press). Suites of characters, such as behavior
and morphology, can evolve without selection acting on every character.
between closely related forms may be facilitated by paedomorphosis.
Phaenicophilus palmarum occurs throughout the island of Hispaniola except on Isle La
Gonave and on the southern peninsula of Haiti, where P. poliocephalus occurs (Fig. 2-1).
Prior to this study. the distributions of the two nalm-tananers were thought to meet but not
poliocephalus had been reported in the Dominican Republic (Dod, 1981) and eastern Haiti.
These sightings were later found to be mistaken identifications of juvenile P. palmarum
pers. comm.). In July, 1983, I observed P. poliocephalus approximately 30 km east of the
Trouin Valley. In May, 1985, I documented, with specimen collections, a narrow hybrid zone
in the area north of Marigot, extending to about 8 km south of Seguin and bordered on the
east by Fond Jean Noel (Fig. 2-1).
The western extent of this hybrid zone is not known,
although hybrids were not found along the Riviere Gosseline approximately 10-15 km east of
There were no P. poliocephalus or hybrids found in the Massif de La Selle. Both
species occur in all major habitat types of Haiti, including cloud forest, mixed pine, mesic
broadleaf woodland, xeric thornscrub, desert, mangrove swamp, and disturbed rural and
urban areas, while hybrids are found in mesic woodland, thornscrub, and disturbed rural
Phaenicophilus palmarum has a yellow-green back, a gray nape, three white eyespots
on a black face mask, a black crown and a diffused white chin and throat. It is also
characterized by an age-dimorphism in that juveniles have gray crowns that range from the
same shade of gray on the nape to darker gray. Phaenicophilus poliocephalus resembles
juvenile P. palmarum by having gray crowns, but it is distinguished from this species by a
distinct white chin against a gray throat. Juvenile P. palmarum have frequently been
confused with P. poliocephalus in the field. The resemblance of adult P. poliocephalus to
juvenile P. palmarum, the existence of an undescribed hybrid zone, and their estimated recent
divergence (Chapter 3) suggested that these taxa would be suitable for the study of
heterochronic processes on the evolution and diversification of avian species.
The effects of
heterochrony on the relationship of the two species is not likely to be confounded by
significant differentiation due to long periods of isolation.
(3) describe the relationship between behavior and morphology in these species; and
(4) determine whether P. poliocephalus is paedomorphic to P. palmarum.
Seventeen morphological variables were measured on fresh
specimens of Phaenicophilus palmarum (NI1
), P. poliocephalus (N2 = 20), and P.
X P. poliocephalus hybrids
14), before skin or skeletal preparations were
made (Baldwin et al., 1931; Table 2-1: App. 1 and 2). Hybrid specimens and representatives of
both parental species are deposited in the American Museum of Natural History in
The remaining individuals are deposited in the Florida State Museum in Gainesville.
Additional museum specimens (NI = 104 and N2 = 46) were included for the morphological
analysis to increase the sample size to conduct multivariate morphometric analysis. Hybrids
were identified in the field by the intermediate extent of black on the crown and white on the
chin. Analyses of
and species differences were subjected to Mann-Whitney U-tests
for the morphological variables.
The value for the crown character in juvenile P. palmarum
zero, because no black occurs on the crown before the final molt.
The average value for a
species class was substituted for missing values and the de
of freedom were adjusted
accordingly. Homogeneity of variance for morphological traits were tested using the Fmax
test (Sokal and Rohlf, 1969).
Juvenile P. palmarum are distinguishable in the field from adul
for P. poliocephalus. Juveniles were initially identified as fledglings when adults were seen
Once I established that juveniles had gray crowns but did not have a distinct
chin pattern, I could differentiate juvenile P. palmarum and P. poliocephalus without
lr r r *1 1 i i *, n
museum collections favors adults because juveniles are often in duller plumage and are not
collected as frequently (R. Zink, pers. comm.).
The proportion of juveniles, as detected by a
yellow wash in head plumage, in the collections I examined varied dramatically between the
two Phaenicophilus species. Juvenile P. palmarum represented 55% of collected P.
palmarum; juvenile P. poliocephalus represented only 9% of the collected specimens for this
species. Because juveniles are easily identified in P. palmarum but not in P. poliocephalus,
any collecting bias that favors adults should be greater in the former species, if juveniles take
equal time to mature in both species.
Foraging behavior. A modified version of the Cody-stopwatch method (Cody, 1968) was
used to collect foraging data for five species in Haiti, including the following: the two palm-
tanagers; the Stripe-headed tanager, Spindalis zena; the Green-tailed Ground warbler,
Microligea palustris, which is similar in plumage to Phaenicophilus; and the Black-and-white
warbler, Mniotilta varia, which was used as an outgroup in the systematic analysis of the
warbler-tanager groups (Chapter 4).
In addition to the foraging variables detailed by Cody
(1968), I included activity level (perch changes per unit time), food-catching attempts (FCA)
per perch change, FCA per minute, food-catching successes (FCS) per minute, FCS/FCA
(Table 2-2), as well as average distance per flight (Kepler, 1977; Rabenold, 1980), and
substrate zone used (see below). Multiple observations on some individuals were no doubt
made because the birds were not individually marked. Manhattan distances between the
Operational Taxonomic Units (OTU's) were calculated from the data for eight of the foraging
variables (Table 2-2) and these distances were used to cluster the OTU's
with the Unweighted
Pair-group Method of Averaging (Sokal and Rohlf, 1969; Cherry et al., 1982; Norusis, 1985).
Substrates were identified by their common Creole names (Pierre-Noel, 1971) or
classified more broadly (e.g., shrubs, broadleaf; App. 3) for subsequent diversity calculations.
of time spent in each zone for each species and for age-classes with P. palmarum were
computed. Significant differences in diversity of substrates used by the species and age-
classes were computed using
the Shannon Information Index (Peet, 1974; Pielou, 1977), and
were evaluated with a t-test (Zar, 1984).
Frequency data for flight distances were grouped into 3 m intervals, combining the
observations that occurred in the intervals
greater than 27 m.
The frequency of foragin
observations at different heights was calculated for each 1.8 m interval, up to
27 m, and
combined for the intervals beyond that. Standardization for differences in maximum
vegetation height between habitats were made by dividing the foraging height by the height
of the maximum substrate height.
Statistical analyses were conducted using parametric and
nonparametric tests from SAS or SPSS (SA
1985; Norusis, 1985). Data were standardized
to a mean of 0 and a standard deviation of 1 to perform Discriminant Function Analysis.
were not standardized prior to Principal Component Analysis (PCA) because standardization
would weight variables equally, thus eliminating the utility of the method. Covariance
instead of correlation matrices were used in the PCA.
Principal components were accepted if
they accounted for 5% or more of the variation. Null hypotheses were rejected at a probability
level of P
- 0.05; highly significant differences occurred at P
Type 1 error
reduced when multiple comparisons of data were made using the same hypothesis with the
1 (0.95)"", where n is the number of comparisons (Harris, 1975).
adjusted the experimentwide error rate to P1
< 0.05. Statistical differences between
frequency distributions were evaluated using the Kolmogorov-Smirnov test (Sokal and Rohlf,
1969). Comparisons of the tendency for the two species to form groups in the nonbreeding
season were evaluated by a modification of the Kolmogorov-Smirnov test, using a chi-square
The first two Principal Components (PC) extracted from the data for 17
morphological variables explained 92.4% of the total variance (Table 2-1). PC 1 had high
positive loadings for the width of white on the chin and PC 2 had high loadings for the extent
of black on the crown. Neither PC 1 nor PC 2 accounted for significant variation of body size
traits. Differences between the two species were highly significant. Discriminant Function
Analysis (DFA) reclassified two of the P. palmarum individuals collected on the edge of the
hybrid-contact zone near Fond Jean Noel as hybrids based on phenotype (P= 0.984 and
P= 0.998). One hybrid clustered with the parental species on the crown character and may
either represent a backcross or an outlier for one or more of the morphological characters,
since DFA weights characters equally (Fig.
as P. poliocephalus or vice versa.
No juvenile P. palmarum was misclassified
Two adult P. palmarum were misclassified as juveniles, but
the probability of correct reclassifications
was marginal (P
< 0.57), so they were still
considered adults in subsequent analyses.
Highly significant differences between age-classes within P. palmarum were detected
for the extent of black on the crown, wingchord, and proximal depth of bill (Pi
Table 2-1). There were only 6 juvenile P. poliocephalus in the samples, thus making it
statistically impossible to test for age differences in morphology.
The three variables listed
above along with width of white on the chin, length of malar stripe, bill length, bill mid-depth
and proximal width, length of dorsal anterior, posterior and ventral eyespots, and tail length
were significantly different between P. palmarum and P. poliocephalus. Significant
differences between the sexes were detected within P. palmarum for proximal depth, proximal
width, and mid-width of the bill, length of the dorsal posterior eyespot, and tail length.
within P. poliocephalus differed only in tail length.
distinct from adults for the extent of black on the crown and from P. poliocephalus for the
width of white on the chin. Adult P. palmarum and P. poliocephalus were distinct from one
another for both characters. Hybrids graded from one parental species to the other, though
they resembled P. poliocephalus more closely. Juveniles were morphologically distinct from
the hybrids and were not intergrades between the two parental species.
Homogeneity of morphological character variance between hybrids and each parental
species was evaluated under the hypothesis that hybrids were no more variable than the
parentals and that they represented first generation crosses and not backcrosses (Lerner,
Variances for all characters between P. palmarum and the hybrids were not
significantly different. Hybrid variances were highly significantly different from those of P.
poliocephalus for extent of black on crown (F
.87) and width of white on chin
The question of convergence or divergence was evaluated for bill mid-width and 12
morphological characters that differed significantly between the two taxa (Table 2-1).
Sympatric populations were designated as those that occurred in and around the contact zone
extending from Foret des Pins to LaVallee (Fig. 2-1). Differences between sympatric and
allopatric populations within P. poliocephalus and within age-classes of P. palmarum were
evaluated for each of the species-distinct characters. Significant regional differences in P.
poliocephalus were found for bill taper (mid-width to proximal width of bill) and tail length
The length of the ventral eyespot was significantly different
between sympatric and allopatric populations of adult P. palmarum. No differences were
found between sympatric and allopatric populations of juvenile P. palmarum.
Once regional differences were established between sympatric and allopatric
populations within a species, then the issue of convergence vs. divergence could be resolved.
(Grant, 1972; Cody, 1973). Differences in character means between sympatric populations of
P. palmarum and P. poliocephalus were evaluated against those in allopatry. Differences
between adults of the two species were more than two times greater for tail length and 18
times greater for bill taper in sympatry than in allopatry. For the length of the ventral
eyespot, differences were 1.5 times greater in allopatry than in sympatry.
Foraging behavior. Juvenile P. palmarum clustered more closely with adult P.
poliocephalus using data for foraging behavior than with adults of their own species (Fig. 2-2).
No significant differences between juvenile P. palmarum and P. poliocephalus were found for
the eight variables analyzed (Table
Adult and juvenile P. palmarum differed
significantly for five variables, whereas adults of both species differed significantly for four
variables (Table 2-2).
Homogeneity of variances for each of the eight foraging variables
poliocephalus and P. palmarum was evaluated under the hypothesis tha
t P. poliocephalus
observations should be more variable than juvenile P. palmarum observations if juvenile P.
poliocephalus were included in the collection of the data but not correctly identified. Only one
variable, average speed of foraging, showed significant heteroscedasticity between the two
groups. Significant differences in the use of substrate type between adult and juvenile P.
palmarum (Dmax = 0.310) and adults and P. poliocephalus (Dmax= 0.267) were found
(Fig. 2-3). Phaenicophilus poliocephalus did not differ from juveniles in substrate use
(Dmax = 0.064). Adult P. palmarum (BPA) foraged primarily in pine (58%) whereas juveniles
(BPI) and P. poliocephalus (GPT) chose broadleaf trees and shrubs (67-69%). Because foliage-
gleaning was the predominant foraging mode for all groups (BPA
73%), foraging differences occurred as a consequence of substrate use and not the mode
The diversities of the substrates used by P. poliocephalus and juvenile P.
Adult P. palmarum were highly significantly different from juveniles (Dmax = 0.158)
and from P. poliocephalus (Dmax= 0.190) and juveniles were highly significantly different
from P. poliocephalus (Dmax = 0.201) in foraging heights. Juvenile P. palmarum occurred at
equal or higher heights when foraging than did adults in all habitats except the mixed pine.
The lower foraging heights of adult P. palmarum may result from differences in the
vegetation structure of their habitat and/or resource distribution.
Highly significant differences in the frequency distribution of horizontal substrate use
were found between adult and juvenile P. palmarum (Dmax= 0.053), adults and P.
(Dmax = 0.092), and juveniles and P. poliocephalus (Dmax = 0.046) when data
were pooled across habitats
App. 4). Because sample sizes differed substantially for upper
and mid-elevation sites, horizontal substrate use was examined only for lowland habitat;
there were still highly significant differences in horizontal substrate use. In open savannah
habitat, at mid-elevation,where trees stand alone or in small clusters, foraging was more
frequent on the middle and inner zones for both species (BPA
GPT = 62%;
no data for
BPI). In closed canopy forests, foraging was concentrated on the outer zone of the substrate
(BPA = 49%, BPI = 42%, GPT= 50%).
Thus, horizontal substrate use varied with habitat
structure and/or elevational gradients.
The frequency distribution of flight distances for adult P. palmarum differed
significantly from the distributions for juveniles (Dmax = 0.209) and for P. poliocephalus
(Dmax = 0.193). Juveniles did not differ significantly from P. poliocephalus (Dmax = 0.147) in
the distribution of flight distances.
This frequency pattern is true for all habitats sampled,
supporting the hypothesis that juvenile P. palmarum and P. poliocephalus perceive the
habitats differently from adult P. palmarum.
These differences may result from differential
dispersion of resources within the habitats. For example, in mixed pine, shrubs are not as
nonbreeding season (August-April) was evaluated against the same tendency during the
breeding season. A significantly greater tendency to form groups in the breeding rather than
the nonbreeding season occurred for P. palmarum (x2(1
between seasons for P. poliocephalus.
= 3.9). Differences were not observed
Therefore, the distributions of the number of
individuals observed together on transects in the nonbreeding season were compared between
P. palmarum and P. poliocephalus. Phaenicophilus poliocephalus had a significantly greater
tendency to be observed in groups of more than two individuals than did P. palmarum in the
nonbreeding season (x2
Several hypotheses can explain the close similarity between two taxa.
lack of divergence due to insufficient time, character convergence, selection due to similar
environments, and/or heterochrony. Because the two taxa were originally chosen for study on
the basis of their close relationship, many of the similarities are due to lack of divergence.
Ninety-two per cent of the variance explained by Principal Component Analysis was
attributed to the chin and crown characters which are the primary distinguishing features of
these two taxa. However, the crown character is polymorphic within P. palmarum.
variation within this character and the frequency of its occurrence might be explained by
geographic variation within one species.
One of the predictions from the one species model of geographic variation is that
different morphotypes should correspond to particular habitat types. Observations from the
same habitats (e.g., cloud forest, mangrove swamp) within different parts of the range of each
taxa failed to document such a correspondence. For example, in mangrove swamp and cloud
forest in the southern peninsula, individuals are all gray-crowned (i.e., P. poliocephalus)
morphology, occur in areas where the two taxa are sympatric. Thus, hybrids do not occur in a
unique habitat distinct from that of the two species of Phaenicophilus. The most
parsimonious hypothesis is that this zone ofintergradation resulted from a secondary contact
between the forms which were geographically isolated in the past.
The patterns of similarity
in crown features between juvenile P. palmarum and adult P. poliocephalus, the lack of
concordance of crown color and habitat type, and the distribution of hybrids are not totally
explained by the lack of divergence model or the geographic variation model.
Additional consideration of the hybrids indicates that selection is probably involved in
reinforcing the distinctness of the two species. If selection against hybrids occurs, differences
in morphology between the parental species in sympatry should be greater than in allopatry.
Differences between the adults of the species were greater for bill taper and tail length in
sympatry than in allopatry, although comparable differences were not found for crown color.
The similarity of adult P. poliocephalus to juvenile P. palmarum in the crown character is not
easily explained by a model of character divergence between the adults of the two species,
unless the species overlapped more extensively in the past.
There is no evidence for such an
overlap and no reason to believe it occurred.
Variances of the crown and chin characters were significantly higher for the hybrids
than for P. poliocephalus, but hybrids were no more variable than P. palmarum for any
The general lack of significant differences in the degree of
morphological variation between hybrids and either of the parental species suggests that
hybrids are the result of parental crosses and are not backcrosses (Lerner, 1954).
apparent absence of introgression and the narrowness of the hybrid zone suggests that
selection against the hybrids is occurring. Similarly, convergence due to similar selective
regimes does not explain the age-dimorphism in one species and its reduction in the second for
1966; Geist, 1971) and delayed maturation within a species (Lawton and Lawton, 1986;
Foster,1987). Gould (1977) did not relate delayed maturation in birds to neoteny because of
his emphasis on phylogenetic endpoints rather than on population processes that produce
There is an absence of evidence that bird species are derived through heterochronic
changes, although variation in degrees of delayed maturation occur among closely related
species of manakins (Chiroxiphia:Foster, 1987), suggesting derivation is through
heterochronic processes. Age-dimorphisms in characters usually accompany delayed
maturation. Phaenicophilus palmarum shows a striking age-dimorphism and is likely an
example of a species characterized by delayed maturation.
The systematic consequences of
heterochrony could be a species retaining the juvenile characters of an antecedent species, as
may be the case for Phaenicophilus poliocephalus. For this to occur, there must be significant
character differences between the adults and juveniles of the antecedent species. Such a
dimorphism is often associated with differences in resource exploitation by the adult and
juvenile age-classes when resources are periodically limiting (e.g., Northern Harrier;
Juvenile P. palmarum were significantly different from adults in the extent of black on
the crown, wingchord, and proximal depth of the bill. Gray crowns in juveniles darken with
age. Although gray-crowned P. palmarum were not observed breeding, dark gray-crowned
individuals have been collected at the beginning of the breeding season in late March or early
April (Table 2-3); some individuals probably delay maturation through their first breeding
Wetmore collected two of these gray-crowned individuals which he designated as
adult females, presumably basing his judgment on ovarian maturation. Pairs consisting of
adults and juveniles were also observed in the post-breeding season, after the time of fledging,
thus increasing the likelihood of pair-bond formation. In addition, the tanager, genus
likely a neotenic species, and one reflection of this is the observed age-dimorphism. Neoteny
may be a common condition in closely related tanagers.
The principal difficulty in
establishing whether P. poliocephalus is paedomorphic to P. palmarum rests with
determining the polarity of their evolution, i.e., which species is derived from which. Derived
characters are presumed to be those unique to derived taxa (Wiley,
poliocephalus has a distinct white chin and gray throat not shared with other tanagers.
Furthermore, neoteny is not a derived condition in palm-tanagers. Because P. poliocephalus
lacks a distinct age-dimorphism, in addition to the above facts, P. palmarum are presumed to
be antecedent to them, making P. poliocephalus the paedomorph.
Several predictions follow from a model of paedomorphosis. Species
derived from other
species in a manner that involves paedomorphosis should be more social relative to their
antecedents (Lawton and Lawton, 1985). Data on mountain sheep support this prediction
(Geist, 1971). Derived species, living in environments where resources are plentiful reach
reproductive maturity earlier.
These derived species retain the juvenile morphologies and
associated behaviors of their antecedents. Moreover, they are relatively more socia
There is evidence in red-winged blackbirds, Agelaius phoeniceus, that subadult
plumage reduces intraspecific aggression (Rohwer,1978).
Therefore, delayed plumage
maturation, or the retention of juvenile morphology, may reduce intraspecific aggression,
resulting in increased sociality. Phaenicophilus poliocephalus was observed in groups of 4-6
individuals during the nonbreeding season, whereas P. palmarum was observed singly or in
This tendency to form groups in P. poliocephalus is consistent with the hypothesis that
gray crowns reduce intraspecific aggression. Additional predictions involving body size
follow from this general paedomorphic model.
sexual maturation) results in smaller body size, due to slower somatic growth after sexual
maturation (Gould, 1977).
Development arrested at an early stage often results in smaller
individuals (Larson, 1980; Alberch,1981). Thri
size between the two species were evaluated as
ee hypotheses concerning the relation of body
follows: (1) Adult P. palmarum are larger
than post-fledging juveniles; (2) Adult P. palmarum are larger than adult P. poliocephalus;
(3) Juvenile P. palmarum are no larger than adult P. poliocephalus.
needs to be tested because post-fledging juveniles can often be as lai
The first hypothesis
rge or larger than adults in
The second hypothesis tests the prediction that arrested development occurs early
enough in the ontogeny of the antecedent species that the adult of the derived species will be
The third hypothesis tests the prediction that adults of the derived species
juveniles of the antecedent species should be comparable in size.
Juvenile P. palmarum are significantly smaller (one-tailed test) than adults in
wingchord, braincase length, and proximal depth of the bill (P1
poliocephalus is significantly smaller than juvenile P. palmarum in bill length, medial depth
of bill, and length of second toe (one-tailed), and is smaller than adults in the same three
as wingchord, braincase length, and proximal bill width and depth (one-
Data from juveniles of differing ages were pooled because the criterion for
designating juvenile P. palmarum was the presence of a gray crown; elevated variances for
various characters might be expected to confound the tests for size differences. However,
variances were homogeneous between the juvenile and adult classes. Juveniles were
significantly smaller than adults in three of the eleven size-related characters.
poliocephalus was significantly smaller than adult P. palmarum in seven of the eleven
characters as predicted, supports the hypothesis of evolution by progenesis for this species.
The average first age of reproduction and/or its variance should also be lower in P.
poliocephalus is not a likely result of selection, but the consequence of its association with
juvenile morphology (Coppinger et al., in press). Juvenile P. palmarum differ significantly in
their foraging behavior from and are less efficient than adults.
This is a common observation
in birds and has been attributed to a lack of experience of juveniles (Orians, 1969; Recher and
Recher,1969; Sutherland etal., 1983; Stevens,1986). Juvenile P. palmarum are significantly
slower, spend more time stopped, and try more food-catching attempts per perch change than
adults, but the success-to-attempts ratio is lower (Table 2-2). Significant differences between
juveniles and adults were not observed in the type of foraging behavior; they are both
primarily insectivorous. Significant differences were observed in choice of substrates. In
mixed pine habitat, juveniles used broadleaf rather than pine substrates. In mesic woodland,
juveniles had a significantly lower substrate diversity than did adults. Foraging height and
flight distances were also significantly different between juveniles and adults; differences in
foraging behavior are probably not a simple function of learning and effectively result in
partitioning the foraging space between birds of different age-classes.
There were also
differences in horizontal substrate use. Juveniles and adults use the same foragin
but in different parts of the habitat.
Arrested morphology in the paedomorph should correspond to retention of juvenile
behaviors (Geist, 1971; Lawton and Lawton, 1986; Coppinger et al., in press). Adult
Phaenicophilus poliocephalus resemble juvenile P. palmarum in foraging behavior.
Homogeneity of variances were equivalent for both groups except for average speed of
foraging thus suggesting that errors in aging P. poliocephalus in the field did not contribute
significantly to the results. A lack of experience in foraging is not a likely explanation for
behavior in adult P. poliocephalus. Phaenicophilus poliocephalus uses broadleaf substrates in
mixed pine habitat as do juvenile P. palmarum and has a significantly lower substrate
selective regimes because their habitats vary.
The behavioral similarities are a result of
retaining juvenile behaviors which are associated with arrested morphological development
in P. poliocephalus.
In conclusion, the polymorphism in crown color is not due to geographic variation
within a species, nor is character displacement a viable hypothesis. Low morphological
variances in hybrids and a narrow hybrid zone point to a restriction in hybridization most
likely due to selection against the intermediates.
These data, in conjunction with a
substantial lack of gene flow (Chapter 3) support the species status of the two forms of
The close resemblance between the species is likely due to their recent
divergence, although the pattern of resemblance requires further explanation.
resemblance of adult P. poliocephalus to juvenile P. palmarum is best explained by a model
involving heterochrony by paedomorphosis.
This study demonstrated that selection among
individuals from different age-classes has significant phylogenetic consequences under a
model of heterochrony by paedomorphosis.
Phaenicophilus poliocephalus is most likely derived from and progenetic to P.
The smaller size, retention of juvenile foraging characteristics, and the
propensity to form groups in P. poliocephalus supports this conclusion.
juvenile patterns cannot be simply explained by
The retention of these
selection for every character, especially
because juvenile P. palmarum are less efficient foragers than adult P. palmarum.
retention of juvenile foraging behavior and smaller size of P. poliocephalus is probably a
consequence of arrested development in morphology. Selection for early maturation in P.
poliocephalus was probably concomitant with the retention of gray crowns, which serves to
reduce intraspecific aggression. Reduction of intraspecific aggression resulted in changes in
the degree of sociality between species of Phaenicophilus, which can have significant
Principal Components for 17 morphological variables measured in Hispaniolan
Table 2-1. (Cont'd)
Table 2-1. (Cont'd)
Table 2-1. (Cont'd)
Sample size (N), mean in mm, standard error
S.E.), and loadings on the first two
Principal Components for 17 morphological variables of adult (BPA) and juvenile (BPI)
Black-crowned Palm Tanager (BPT), Gray-crowned Palm Tanager (GPT), and their
Percent variation accounted for by each Component given in parentheses.
Significant differences in morphological character between BPA and BPI (P1
Significant differences between BPT and GPT.
Significant differences between sexes within BPT.
Significant differences between sexes within GPT.
Mean, sample size, and significance of differences of eight foraging behavior
VARIABLE BPA BPI GPT
- - - - -------------------------- -
VARIABLE BPA BPI GPT
FCSperb 2.05 1.80 1.53
minute (126) (30) (54)
Successes/ 0.59 0.46 0.39
I ------------------ ----------------- --- I
FCA pera 0.26 0.37 0.30
perch change (233) (84) (136)
Black-crowned Palm-Tanagers are divided into adults (BPA) and juveniles (BPI).
Gray-crowned Palm-Tanagers (GPT) are not.
Significance differences between
designated by a line with asterisks,
P < 0.001 (***).
P s 0.01
= food-catching attempts.
= food-catching successes.
List of gray-crowned Black-crowned Palm-Tanagers collected in the early part of
(late March to early April) with associated comments.
6 Feb 1916
Yellow wash on crown, nape, chin and
throat, and Dorsal Anterior Eyespot
1 Feb 1928
Dark gray crown;
Dorsal Anterior Eyespot
9 Feb 1932
Some gray on crown
26 Mar 1931
Gray in crown.
16 Apr 1919
Crown more dark gray than black;
difficult to distinguish crown from
11 Apr 1927
Back of crown dark gray;
listed this as adult female
9 Mar 1917
Dark gray crown;
26 Mar 1929
Crown mostly dark gray
Abbreviations for museums are:
= Carnegie Museum of Natural History in
= National Academy of Sciences in Philadelphia; USNM = U.S.
National Museum at the Smithsonian Institute in Washington, D.C.
* Mfl -
*' 4 C
- c 1
bifl & r
a- *'-4 ^
0 to 0
w '- 1
CI o c
'^ a ^2
5 5 "
* C 5
* U u-
a a *ii Oh S
S t" II
1. : -; v-v -;..:-:1
h #1 -
THE GENETIC CONSEQUENCES OF HETEROCHRONY
Heterochrony, or variation in developmental patterns, provides a framework from
which to integrate populational processes into a general evolutionary model (Gould, 1977)
and to explain how these processes lead to different levels of genetic variability within and
between species. Paedomorphosis, a type of heterochrony, results in the retention of juvenile
characters in reproductively capable individuals (Gould, 1977; Lawton and Lawton, 1986) and
can be achieved by small changes in regulatory genes early in development
Alterations in body size and shape, fecundity, age-structure, and significant changes in social
structure within populations can result by increasing growth rates and decreasing age of
sexual maturity (Geist, 1971; Gould, 1977; Lawton and Lawton, 1986). Life history
characteristics, such as fecundity and growth rates, are correlated to levels of genetic
variability (Smith et al.,
1975; Cothran etal., 1983; Mitton and Grant, 1984), although there
is no unifying theory to predict the genetic consequences of paedomorphosis or its effects on
the rate of speciation.
Avian groups that are paedomorphic may represent ideal systems from which to
develop a coordinated theory relating ecological constraints, age-structure, life history traits,
and genetic variability to rapid speciation. Many avian groups, with the possible exception of
ratites, have only recently been recognized as paedomorphic (Lawton and Lawton, 1986;
recognized as a paedomorphic process. The alternative process to delayed somatic maturation
is early onset of sexual maturation, or progenesis. This latter process has not been
documented for avian groups. Both neoteny and progenesis result in individuals breeding in
juvenile morphology (i.e., paedomorphosis). Hispaniolan palm-tanagers represent a species
pair in which one is paedomorphic to the other. Gray-crowned Palm-Tanagers,
Phaenicophilus poliocephalus, resemble the juveniles of Black-crowned Palm-Tanagers, P.
palmarum, in foraging behavior and morphology (Chapter 2).
My purpose is to describe the patterns of genetic variation within and between the two
palm-tanager species. My specific objectives are to: (1
quantify the degree of genetic
differentiation between these species and use it to calculate estimated divergence time; (2)
compare genetic variability, as measured by multilocus heterozygosity, between the two
species; (3) describe age-specific genetic variation within these species and relate it to
differences in behavior and morphology; and, (4) discuss speciation in these tanagers
relates to isolation, founder effect, and genetic variability.
Methods and Materials
Collections were made of two species of Phaenicophilus and their hybrids in Haiti from
May through September, 1985. Field identifications were based on crown and chin characters
Juveniles of both species could be identified in hand by the presence of a yellow
wash in the plumage. Because no dry ice or liquid nitrogen is available in Haiti, tissues were
stored in 1.5% buffered 2-phenoxyethanol solution 0.5 to 4 hours after collection (Nakanishi
., 1969; Barrowclough, pers. comm.), and vials were stored in a conventional freezer at -
4oC until September 1985.
Tissues were stored at -600C thereafter.
Liver and muscle extracts were ground in the phenoxyethanol solution. Samples
Thirty-nine presumptive loci were assayed. Locus designations, abbreviations, and
buffer conditions are given in Table 3-1 except where listed below. Loci were numbered
according to the mobility of the products with the most anodal as 1 when two or more isozymes
appeared on the same gel. Monomorphic loci included AAT-2 (Acid Citrate 6.2
Malate 7.4 buffers), CK-1, FH-2, aGPD-1, lactate dehydrogenase (LDH)- 1 and
dehydrogenase (MDH)- 1 and 2, malic enzyme (ME), and SOD-2 (all on Acid Citrate 6.2
buffer), and general proteins (GP)-1 and
8.2 buffer). Alleles were
designated alphabetically, with A corresponding to the allele with the fastest migrating
product. No locus had more than three alleles. A third locus for AAT appeared occasionally in
the most anodal position, although attempts to visualize this third and highly variable locus
on a regular basis were futile. Because only two AAT loci have been reported for birds, the
appearance of this third AAT locus may be a staining artifact.
General statistical tests were conducted with the Statistical Analysis System (SAS,
1985) or Statistical Package for the Social Sciences (SPSS, Norusis, 1985). Significance
set at P<0.05; highly significant rejection of null hypotheses occurred when P <0.01.
Acceptance levels for multiple comparisons involving the same data set were adjusted to an
experiment-wide error of P < 0.05 (Harris, 1975). Degrees of freedom are given as subscripts
to the reported statistics.
Tests are reported as two-tailed except where noted. Substrate
diversities were calculated for the two species ofPhaenicophilus and age-classes within P.
palmarum, using the Shannon Information Index (Peet, 1974; Pielou, 1977). Differences
between diversities were evaluated using t-tests (Zar, 1984).
Allele frequencies and genetic variability, as measured by the proportion of
heterozygous loci determined by direct count per individual averaged across 39 loci (H), the
average number of alleles per locus (A), and the per cent loci polymorphic (P) with the
Statistical analyses of individual heterozygosities that were arcsine square-root
transformed (Archie, 1985) were performed using the t-test. Because samples were small in
some categories, data for rare or unique alleles were excluded on subsequent analyses if the
expected frequency of observation
was less than one individual
as calculated from the gene
frequencies in the alternate sample.
Genetic distance (Nei, 1978) between P. palmarum and P. poliocephalus
F-statistics (Fst) (Wright, 1965, 1978) were employed to evaluate the amount of genetic
differentiation between P. palmarum and P. poliocephalus (Fst =
0.122). Fst were used in this
to make comparisons to within-species values possible. No significant differences in
genetic heterozygosity existed between the two parental species (t(40)= -
palmarum and hybrids (t34
=-0.17), or between P. poliocephalus and hybrids (t(32)= 0.54)
No fixed allelic differences were found between the species although there were
shifts in allele frequencies and the distribution of rare alleles.
Unique alleles, defined as
those observed in only 1 group, were observed at
21 of the
polymorphic loci. Phaenicophilus poliocephalus had 15 and P. palmarum had 16 unique
out of a total number of 63 and 64 alleles, respectively.
size for each species
was sufficient to detect 56
.2% and 74.1%
, respectively, of the unique alleles in at least one
individual at the observed frequencies in the other species.
Nine rare alleles across seven loci
were shared between the two species. Rare alleles are those found in both samples with a
frequency of 0
.25 or less.
There were 9 unique alleles in juvenile P. palmarum and 6 unique
alleles in adults across 11 loci; they shared 10 rare alleles. Differences in heterozygosity were
highly significant between juvenile (H
= 0.121) and adult P. palmarum (H = 0.074)(t(19)=
Juvenile or adult P. palmarum did not differ significantly from either P. poliocephalus or the
hybrids in genetic heterozygosity (P1 0.05).
To assess the size of the founding populations required for colonization and
maintenance of current heterozygosity levels in P. poliocephalus, expected heterozygosity
(He) was computed for the colonists as follows (Crow and Kimura, 1970; Baker and Moeed,
where No is the size of the founding population and Ho is heterozygosity of the founders.
Founding populations derived from ancestral P. palmarum were assumed for the calculation
to consist of one of three groups: all adults, all juveniles, or a mixture of adults and juveniles
collected at random.
Three curves were generated and compared with the current levels of
heterozygosity in P. poliocephalus (H =
0.104) (Fig. 3-1
The curve generated assuming a
founding population of all juveniles asymptoled at He =
0.106 at No = 4, where the curve
generated assuming a mixed group reached He = 0.090 at No = 40.
This latter value is more
than one standard error (S.E.) below the current level of heterozygosity of P. poliocephalus.
The curve generated assuming a founding population of all adults reached an aymptote at
more than two S.E. below the current heterozygosity level of P. poliocephalus.
Of the 17 variable loci in P. palmarum (aGPD-2
was omitted due to low sample sizes),
only four did not have higher levels of heterozygosity in juveniles than adults.
across loci was significantly different from random (x2(1)= 4.8). Data from the 4 loci (glucose
dehydrogenase, phosphoglucosemutase-1, peptidase-1, and xanthine dehydrogenase) that did
not follow this trend were dropped.
The jackknife procedure (Lanyon,
1987) was performed on
the remaining data to evaluate single-locus effects on heterozygosity differences between age-
loci. Heterozygosity difference between age-classes was tested under the hypothesis that the
differences were not due to significant single-locus effects. Levels
significantly different between age-classes (t(24)=-24.33).
of heterozygosity were still
Therefore, differences in
heterozygosity between age-classes were not due to the effects of one or two loci. Moreover,
the direction of the difference between age-classes remained unchanged; juvenile
palmarum were always more heterozygous than were adults, regardless of which data were
removed (Table 3-2).
Substrate and behavioral diversities were used
as estimates of niche breadth. Adult
and juvenile P. palmarum differed significantly from P. poliocephalus in foraging behavior
diversity (t(964)= 9.95 and t(819)= 6.49).
Age-classes within P. palmarum were not
statistically different from one another (t(724)= 2.18).
Adult P. palmarum were highly
significantly different from juveniles of the same species and from P. poliocephalus
5.00 and t(235)=
, respectively) in the diversity of substrates used.
F-statistics were used as an indirect measure of gene flow in palm-tanagers.
interpret the statistic, both species were assumed to represent populations of a single form
(Wright, 1965, 1978). The differentiation observed between the two forms ofPhaenicophilus
(Fst= 0.122) is twice as high as that observed among populations of most other avian species
(Barrowclough, 1980b, 1983), suggesting that gene flow is restricted within Phaenicophilus.
Comparable Fst's are observed for species with low dispersal rates that are caused by
geographic isolation between populations (Corbin et al.,
1974; Yang and Patton, 1981
and Moeed, 1987).
Because species of Phaenicophilus are not now isolated by any geographic
barrier, the high Fst is indicative of reduced gene flow on the same order as most avian
and the apparent selection against hybrids in the contact zone (Chapter 2) supports the
hypothesis of reduced gene flow.
The genetic distance found between the two Phaenicophilus taxa was low
even for avian species that have an overall average D= 0.044 (Barrowclough, 1980a; 19
However, Johnson and Zink (1983) observed D= 0.004 in two closely related species of
sapsuckers (Sphyrapicus) that were undergoing character displacement due to assortative
mating in sympatry. Low genetic distance should not be used a priori as a criterion for species
recognition in birds.
Genetic distance is not just a function of current gene flow, but is a
measure of the time since gene flow has been restricted (Yang and Patton, 1981). Moreover, D
is a function of the relative magnitude and divergence of allelic frequencies and consequent
shifts in heterozygosities and associated variances (Novak, pers. comm.).
The low D for
Phaenicophilus is likely an indication of two recently diverged but distinct species.
The estimated time of divergence
Nei, 1975; Johnson and Zink, 1985) is between 5.0
104 and 2.6
x 105 years ago in the middle of the Pleistocene.
This range in the estimate
corresponds remarkably well to the time of the most recent interglacial period when sea levels
rose 8-10 m some 6.5 x 104 years ago and multiple times throughout the Pliocene and
Pleistocene (Pregill and Olson, 1981).
The rise in sea level would have inundated the Cul-de-
Sac Plain, which runs from west to east across Hispaniola, thus cleaving it into north and
This plain is presently below sea level; during interglacial periods when
glaciers melted and sea levels rose it would have formed an open water barrier to gene flow
(Pregill and Olson, 1981).
Phaenicophilus is likely to have fragmented in allopatry on the two islands. Current
distributions of the two species suggest that P. poliocephalus arose on the south island. If the
divergence is a result of a vicariance event, then P. poliocephalus would be expected to be
in P. poliocephalus is consistent with the hypothesis that it is derived from P. palmarum as a
result of colonization to the south island. Low competition and relatively abundant resources
on the colonized south island could have favored the evolution of progenesis, which is
characterized by earlier sexual maturation often at smaller body size than normally expected
If the colonizers were few in number, then genetic variability would have
declined due to founder effect (Crow and Kimura,1970; Kilpatrick, 1981).
Both species have
similarly high levels of heterozygosity (H = 9-10%). Relatively high H is not unusual for
island birds (Yang and Patton, 1981) or for other vertebrates (Nevo et al., 1984), though an
"island" effect has been observed in small mammals (Kilpatrick, 1981; Aquadro and
Kilpatrick, 1981). A decline in heterozygosity could be avoided by rapid population growth
after colonization, a large number of founders, or multiple invasions. If heterozygosity were
reduced by founder effect, it could have been restored over long periods through mutation.
The simplest explanation for the current situation is that heterozygosity was not
reduced on the south island in ancestral P. poliocephalus.
There seems to be insufficient time
for the re-establishment of high heterozygosity since the estimated divergence. Large
numbers of colonizers could explain the high heterozygosity, but is inconsistent with the idea
of geographic isolation and rapid subsequent speciation (Templeton, 1980). A small number
of highly heterozygous colonizers and rapid population growth would also explain
maintenance of high heterozygosity in the founder population on the south island (Nei, 1975;
The major difficulty with this explanation is how to get a group of highly
heterozygous colonizers. A founding group of adults would not have had sufficient levels of
genetic variability to explain levels of heterozygosity currently observed in P. poliocephalus
(Fig. 3-1). A mixed group of adult and juveniles colonizers would have had to consist of four
times the number of founders than a group composed only of juveniles to maintain levels of
other avian species (Greenwood and Harvey, 1982). Flocks of neotenic Brown jays,
Cyanocorax morio, colonizing recently cleared habitat, have a lower mean age than flocks in
the main population (Lawton and Lawton, 1985). Geographic isolation, small numbers of
founders, and few founding events set the stage for rapid speciation (Templeton,
progenesis provides a basis for explaining the resemblance of P. poliocephalus to juvenile P.
palmarum in behavior and morphology. An alternative explanation, using a vicariance
model, could be used but it is not as parsimonious as the colonization model in explaining both
levels of genetic variability and the resemblance of adult P. poliocephalus to juvenile P.
of heterozygosity are similar between adult P. poliocephalus and
juvenile P. palmarum. Although no significant difference between age-classes exists within
P. poliocephalus (P= 0.10), adults are relatively more heterozygous than are juveniles
(H = 0.115 and 0.069, respectively), and have comparable levels of variability with juvenile P.
palmarum (H = 0.121). Juvenile P. palmarum are significantly more heterozygous than
adults (H = 0.074) of their own species. Age-related changes in multi-locus heterozygosity
have not been documented previously for birds, but are known for other vertebrates (Cothran
et al., 1983; Samollow and Soule, 1983).
The similarity of genetic variability between adult P.
poliocephalus and juvenile P. palmarum may be serendipitous, but the pattern across several
character sets (i.e., behavior, morphology; Chapter 2) suggests there may be a single
underlying biological proce
ss. Rapid divergence of P. poliocephalus on the south island may
have been facilitated by selection in a new habitat with abundant resources.
Colonists on the
south island may have experienced abundant resources relative to their north island
counterparts, resulting in increased fecundity, faster growth rates, and earlier sexual
These characteristics are the essential features of progenesis.
founded by a few individuals and isolated from the ancestral form satisfy the requirements of
1980) model for rapid speciation.
The size of founder populations should be just
small enough to cause a rapid accumulation of inbreeding without a severe reduction in
These conditions enhance the probability of a reorganization of the
genome, with especially important consequences for regulatory genes (Templeton, 1980).
Even small changes in regulatory loci can effect significant phenotypic changes (Larson,
1980) and provide the basis for the establishment of isolating mechanisms. Few differences in
structural loci nor are radical shifts in ecological niches expected in Templeton's model or in
(1977) model of paedomorphosis.
Phaenicophilus poliocephalus appears to be
exploiting the foraging niche of juvenile P. palmarum (Chapter 2) and there are few
differences between the two species in structural gene loci (Table 3-1
The conditions that
favor rapid speciation after a founding event are the same ones as prescribed by Gould's
model for the evolution of progenesis in colonizing populations.
Rapid speciation in paedomorphic assemblages would be promoted by higher genetic
variability in juvenile age-classes. Higher levels of genetic variability in the founding
populations would result in greater evolutionary changes than populations with lower levels
of genetic variability (Futuyma, 1979:p. 442).
Faster evolution may increase the probability
of speciation and not require as long a period of geographic isolation.
species in which juveniles are more heterozygous and are also the primary dispersers should
be more prone to speciation.
Age-related differences in heterozygosity observed in P. palmarum might be spurious
and not justify the predictions for rapid speciation in Phaenicophilus.
levels of heterozygosity are observed for juvenile P. palmarum in 13 out of 17 variable loci.
Removal of the data for any single locus does not alter this pattern (Table 3-2).
Age-related differences in heterozygosity observed for Phaenicophilus might also be
expected to occur in amphibians that undergo paedomorphosis. No changes in the levels of
genetic variability have been observed in populations of salamanders that experience varying
degrees of paedomorphosis (Pierce and Mitton,1980; Shaffer, 1984). However, age-class
differences in single- and multi-locus heterozygosity have been observed in other vertebrates
(Redfield, 1973; Tinkle and Selander, 1973; Ramsey et al., 1979; Chesser et al., 1982; Cothran
et al., 1983; Samollow and Soule, 1983), although higher genetic variability is not always
found in juveniles. If the heterozygosity levels in adults are relatively constant over time and
juveniles are more variable than adults, one mechanism is required to explain higher
heterozygosity in juveniles and another for its reduction in adults.
The differences in genetic variability across life history stages
e explained in
several ways that are not mutually exclusive.
These explanations include developmental
changes in isozyme patterns, negative assortative mating, or dispersal of individuals with
different genotypes across spatially heterogeneous areas for allele frequencies. None of these
processes account for the loss of heterozygosity in adults observed for 13 loci.
the number of rare alleles in juveniles compared to adults
operating (Samollow and Soule,
The decrease in
indicates selection is probably
1983). Increased genetic heterozygosity in juveniles may be
due to one process while selection eliminates certain adults, thus decreasing heterozygosity
for that age-class. Selection may also act on both life history stages, first increasing, then
decreasing genetic variability. Juvenile Blue grouse (Dendrogapus obscurus), another
neotenic species (Lewis and Jamieson, 1987), are more heterozygous for the Ng locus than are
adults in the same late successional habitat.
These data suggest that genetic variability may
also be positively correlated with resource availability (Redfield, 1973).
Genetic variability is often correlated with various characteristics closely associated
individuals might be at a selective disadvantage among adult P. palmarum. If exploratory
behavior is positively correlated with heterozygosity (Garten, 1977), then heterozygous
juveniles might be more active and exposed to higher levels of accidental mortality.
Alternatively, since adult birds tend to return to the same breeding territory (Greenwood and
Harvey, 1982), selection may result in microgeographic adaptation to particular habitats.
The type of genetic data collected do not allow an evaluation of the relative importance of any
of these mechanisms or the basis of selection for or against relatively heterozygous
Low heterozygosity in adults might be a response to decreased niche breadth
(Levins, 1968). Contrary to this prediction, substrate diversity for adults is higher than that
of juveniles. Because niche breadth is an N-dimensional concept (Hutchinson, 1958), other
variables need to be examined to test this hypothesis more fully.
In conclusion, neoteny in P. palmarum seems to be expressed both behaviorally and
morphologically (Chapter 2).
The age-dimorphism in behavior and morphology is congruent
with observed heterozygosity differences between age-classes. Selection is probably operating
to produce the observed life history shifts in genetic variability between juveniles and adults.
Higher juvenile genetic variability, combined with a greater amount of dispersal by juveniles
and the high genetic variability observed in P. poliocephalus, are consistent with the
derivation of P. poliocephalus from small founding populations of juvenile P. palmarum on the
south island of Hispaniola during the Pleistocene. Speciation will be more rapid when life
history traits of the ancestral populations are variable
Derivation of new species by
progenesis is more likely if the ancestral species is neotenic. Speciation will occur for neotenic
species expanding into new environments, if (1) the colonizers possess high genetic
variability, (2) the colonizers have high reproductive and low mortality rates at low densities,
(3) founder populations are small, and (4) reproductive isolation follows the reorganization of
Estimates of genetic variability for Hispaniolan palm-tanagers and their
hybrids for 39 enxyme loci.
Table 3-1. (Cont'd)
Table 3-1. (Cont'd)
Table 3-1. (Cont'd)
Note: Allele frequencies, direct count heterozygosity (H
, percent polymorphic loci
, mean number of alleles (A), and
sample sizes (N) for
Hispaniolan palm-tanagers and their hybrids across 39 enzyme loci.
enzyme names and international nomenclatural numbers, according to Harris and
Hopkinson (1976), and pH/buffer systems is included for each isozyme
= Adult Black-crowned palm-tanager; BPI=juvenile
tanager; GPT = Gray-crowned palm-tanager; HYB = hybrids.
Enzyme names and numbers recommended by the Commission on Biological
Nomenclature ("Enzyme Nomenclature", Elsevier, Amsterdam, 1973).
Abbreviations for buffers are: AC
=Acid Citrate (Clayton and Tretiak, 1972);
=Tris Citrate 7.0 (Ayala et al., 1972); TM =
Tris Malate; TC
=Tris Citrate 8.0
EDTA = Ethylenediamine Tetraacetic Acid; LIOH = Lithium Hydroxide (Selander et
al., 1971; Harris and Hopkinson, 1976).
When more than one buffer condition is
JI 1 f r '1 I, *4 ..A- '
Mean % heterozygosity for adult and juvenile Black-crowned palm-tanagers
after Jackknife simulation
Lanyon, 1987) for 13 variable
This procedure removes
data for one
ocus at a time and computes
heterozygosity after removal. A t-test was performed to test the null hypothesis that
the average difference across all 13 loci between age-classes is not significant.
Differences were highly significant (t(24)= 24.3); the null hypothesis was rejected at
P< 0.001. Abbreviations for loci are explained in Table 3-1.
Standard errors ranged from 0.010-0.013 for adults and 0.007-0.013 for immatures.
Probabilities, after removal, ranged from 0.004-0.037 for single locus t-tests.
C dQ) -
c wo ^ '"
C o .
4 a? t
~o o ^
m ^ a
r-l C *-
-. aj c,
3b I l
ed g g
;I S 0 l
*I I 0r
I I .-
IIC < ;(DQ
QI I -
II r O
S 0 I-
II I r
S I a
OF HISPANIOLAN PALM-TANAGERS
In recent years, increasing confidence has been placed in the genetic resolution of
phylogenetic relationships and the molecular tools used to assay the genetic differences
underlying these relationships.
What is often not appreciated is that the various molecular
techniques provide indirect or incomplete assessments of genetic differentiation. Moreover,
correlations between character sets, such
as morphology and electrophoretically detectable
proteins, are often weak (Zink et al.,
Incongruencies in character sets often raise more
questions than they answer, thus generating more phylogenetic hypotheses. Character
reversals and convergences also create problems in resolving phylogenetic relationships as
has been amply demonstrated in birds (Sibley and Ahlquist, 1986; Neff, 1987).
The resolution of the derivation of Hispaniolan palm-tanagers from one another cannot
be based on morphological criteria alone. Crown color is an important character in
distinguishing between these species and may represent an example of character reversal in
relation to other tanagers. Additional characters must be used to resolve this relationship.
Resulting phylogenies must withstand rigorous scrutiny and be understood from a broader
taxonomic perspective. Although avian taxa are well studied, many phylogenetic
relationships remain to be resolved, including those of neotropical tanagers.
The subfamily Thraupinae tanagerss) in the Emberizidae represents a diverse South
American radiation (
> 200 species) of small to medium-sized birds that are largely
(Storer, 1969) and how they are related to warblers, vireos, and finches remain unclear. Bond
(1978) proposed that Greater Antillean tanagers were originally derived from Central
American species that are now extirpated on the mainland. Alternatively, some Greater
Antillean tanagers may have been derived from migrants either island hopping across the
Caribbean or flying over during annual migrations from and to North America. In general
plumage characters, Hispaniolan palm-tanagers resemble the Central American bush-
tanagers, genus Chlorospingus.
Therefore, the derivation of Phaenicophilus from a
Chlorospingus-like ancestor is plausible. Sibley (pers. comm.) placed species belonging to the
genus Piranga in close association to Phaenicophilus, though Chlorospingus was not included
in the DNA hybridization analysis. Because Hispaniola lies on one of the major migratory
corridors between North and South America, and Piranga olivacea is a migrant along this
corridor, derivation of Phaenicophilus from a Piranga-like ancestor is also possible. Species of
the two genera resemble one another in overall body-size, but otherwise are not similar in
A third phylogenetic hypothesis depends on ontogenetic similarities in plumage
being potentially important for the analysis.
Whereas bill morphology
changes a great deal
across species, plumage patterns tend to be conservative (Storer, 1969).
phylogenies built on plumage characters may be robust. Although adults of Spinda
are morphologically and behaviorally distinct from adult Phaenicophilus, female Spindalis
from Jamaica are quite similar in appearance to Phaenicophilus fledglings.
may be phylogenetically significant.
Spindalis zena is common throughout the West Indies.
Colonization during the Pleistocene with subsequent isolation on Hispaniola could have
resulted in differentiation of populations and subsequent divergence of Phaenicophilus
plumage patterns may be important to determine phylogenetic relationships among tana
Consequently, there are five taxa that might be the closest relative to Phaenicophilus.
taxa include the following: the Central American common bush-tanager, Chlorospingus
ophthalmicus, based on plumage patterns and geography; the scarlet tanager, Piranga
olivacea, based on DNA hybridization and biogeographical patterns; the stripe-headed
tanager, Spindalis zena, based on ontogenetic plumage patterns; the palm tanager endemic to
the Lesser Antilles, Thraupis palmarum, based on geography, behavior and ecology; and the
South American hooded tanager, Nemosia pileata based on plumage patterns.
Which of these species is the closest relative to Phaenicophilus is critical in the
understanding of the evolution of the two species in this genus.
The determination of the
polarity of the speciation event (i.e., which species is derived from which) depends on
phylogenetic reconstruction to demonstrate that one species is derived from another either by
terminal deletion (i.e., paedomorphosis) or by terminal addition (i.e., peramorphosis) of
characters (Kluge and Strauss, 1985).
The assumption made throughout earlier discussions
of Phaenicophilus (Chapter 2) was that Piranga is the closest sister group to Phaenicophilus.
Because neoteny occurs in Piranga, it is a shared character with Phaenicophilus palmarum.
The chin and throat pattern observed in P. poliocephalus is not found in other tanagers
included in the analysis.
The existence of a shared character state between Piranga and
Phaenicophilus palmarum and the presence of a unique character in P. poliocephalus support
the hypothesis that the latter species is derived from P. palmarum. If true, the absence of a
black crown in adult P. poliocephalus constitutes a terminal deletion and P. poliocephalus
would be paedomorphic to P. palmarum. In contrast, if Chlorospingus is the closest relative of
Phaenicophilus, then the polarity of the derivation of Phaenicophilus species is not clear,
because there is not much information on the breeding biology of Chlorospingus. An
warblers (Parulinae) represent a group of small, colorful insectivorous birds with North
American affinities (Bond, 1978). Beecher (1953) considered warblers and tanagers to be
sister groups derived from the same vireo-like ancestor. Subsequent work with DNA
hybridization removed vireos to the corvine assemblage (Sibley and Ahlquist, 1982), thus
suggesting similarity in plumage and size among these three groups constitutes a case of
convergence rather than a close relationship. Additional analyses (Sibley, pers. comm.) point
to a more distant relationship between tanagers and warblers than previously thought.
Plumage similarities between the Hispaniolan palm-tanagers, Phaenicophilus,
green-tailed ground warbler, Microligea palustris, and the mainland common bush-tanager,
Chlorospingus ophthalmicus raise questions about the true affinities of the palm-tanagers to
other tanagers and to warblers.
If the taxonomic designations of the Hispaniolan forms are
correct, then plumage
similarities could be due to the retention of primitive characters in both warblers and
tanagers. Alternatively, the similarity in plumage may be the consequence of convergences
or is due to closer taxonomic affinities of these species than DNA hybridization reveals. If the
latter is true, then the Hispaniolan complex might provide clues to the origin of warblers and
tanagers, possibly putting Hispaniola as the site of origin of these two groups.
If Microligea and Xenoligea are more closely allied to Chlorospingus and
Phaenicophilus than current taxonomic designations suggest, then the relationship might be
crucial to the final determination of the polarity of speciation in Phaenicophilus.
Consequently, several morphological features could be used in the determination of the
polarity if Hispaniolan warblers link Chlorospingus to Phaenicophilus to form a species
. For example, both Microligea and Chlorospingus have gray crowns. If
Phaenicophilus were derived from a Chlorospingus-like ancestor, then gray crowns are a
addition. Consequently, the relationship of Microligea and Chlorospingus to Phaenicophilus
is crucial to the determination of the polarity of evolution in Phaenicophilus and whether or
not paedomorphosis has occurred.
My purpose was to describe the relationships of Hispaniolan palm-tanagers to other
tanagers and to Hispaniolan warblers.
The specific objectives were as follows: (1
determine from which group Hispaniolan palm-tanagers are most likely derived;
2) to use
close relatives as outgroups in the determination of the polarity of speciation in palm-
tanagers; and (3) to determine the relationship of Microligea palustris to Hispaniolan palm-
A total of 163 specimens from
25 species of tanagers
and warblers was used for the
electrophoretic analysis of 20 presumptive loci.
(Operational Taxonomic Units
An additional 422 specimens from 15 OTU's
representing each of the genera used in the genetic analysis,
were chosen for morphological analysis (Table 4-1
and Chlorospingus, Microligea palustris was included to test the relationship of warblers to
tanagers. Xenoligea montana
too rare to collect in Haiti.
With the inclusion of
Microligea, seven species of wood warblers were added as outgroups for the tanager-warbler
comparisons. These species included Mniotilta varia, Geothlypis trichas, Dendroica
pensylvanica, D. palmarum, D. pinus, D. coronata, and D. dominica. The addition of the
warblers also allowed comparisons between this study and earlier ones. The White-eyed
vireo, Vireo griseus, was chosen as an outgroup for the tanager-warbler comparisons
Species were selected from several additional genera including Euphonia,
Eucometis, and Coereba based on the phylogeny of Sibley (pers. comm.).
In addition to Phaenicophilus, Piranga,
Floridian material are deposited in the Louisiana State University Museum of
(LSUMZ) frozen tissue collection.
Whole tissue homogenates of liver and pectoral muscle tissue were prepared for
horizontal starch gel electrophoresis (Chapter 3). Stain recipes were from Selander et al.
and Harris and Hopkinson (1976); isozyme nomenclature followed from Harris and
Collection of the Haitian material is described in Chapter
collected in Florida and South Carolina was treated similarly, except that tissues were stored
within 8 hours at
-600C in an ultracold freezer. Specimens from Tall Timbers Research
Station near Tallahassee, Florida, may have been dead for up to 8 hours before being frozen.
Neotropical tanagers and some North American wood warbler tissues were obtained from the
LSUMZ tissue collection.
These tissues were collected and stored in liquid nitrogen (Johnson
Subsamples of the material were stored in 1
.5% 2-phenoxyethanol solution for
Three to five gels per stain were scored for different sequences of species
to determine the relative mobility of allelic products.
Two or more buffer systems were used
on all enzymes except fructose diphosphate aldolase (FDA-1), fumarate hydratase (FH),
malate dehydrogenase (MDH) and malic enzyme
ME) (Table 4-2).
The computer program BIOSYS-1 (Swofford and Selander, 1981) w
allelic frequencies, expected heterozygosity (He
used to calculate
observed heterozygosity by direct count
averaged over loci (Ho
1978) and modified Rogers' distance
UPGMA (Unweighted Pair Group Method of Averaging) phenograms and
Distance-Wagner trees (Fig. 4-1). Rogers' D was used in the generation of UPGMA and
Distance Wagner trees and in comparisons with morphological distances, since it satisfies the
triangle inequality (Swofford and Selander, 1981
Distance Wagner trees were evaluated
before and after optimization and after using the multiple additions criterion for 20 loci
The Jackknife procedure (Lanyon, 1987), deleting data from loci or Operational
Taxonomic Units (OTU's) one at a time, with replacement, was used to determine the
stability of phylogenetic clusters. Data for fifteen variable loci were subject to jackknifing;
the data for the remaining five loci were not sufficiently variable to provide useful
information for this analysis. Data for OTU's of single species or
several species from a single
genus were systematically jackknifed. Stability of phylogenetic affinities was assessed by per
cent of association in clusters for each phylogram when data for loci or OTU's
after corrections were made for changes in sample
Clusters were defined by
the presence of "core" species,
such as Piranga olivacea, Chlorospingus ophthalmicus, or
Eucometis penicillata. For example, Piranga olivacea, P. rubra, and P. ludoviciana were
associated 100% of the time on both UPGMA and Distance Wagner algorithms, whether data
from loci or OTU's
Morphological data were collected for 21 skin and skeletal variables.
at least ten males and ten females of each species were measured to the nearest 0.01 mm
using Helios dial calipers (Table 4-1
Manhattan distances, considered to be the most robust
metric for cross taxonomic comparisons (Cherry et al.,
1982), were calculated for all species
used in the genetic analysis, except Piranga ludoviciana, Hemispingus atropileus and H.
superciliaris, Geothlypis trichas, Dendroica pensylvanica, D. palmarum, D. pinus, D. coronata,
Manhattan distances are calculated as the absolute difference between two
When more than one variable is used in comparing groups, the sums of distances
between pairs of variables are used in calculating the final distance. Morphological distance
matrices were constructed for all 21 morphological features, for body-size related features
only (Table 4-5) and for plumage features only. Body-size related features included
wingchord, length of braincase, body length from the gonys to insertion of rectrices, length of
from exposed culmen, extent of color on the chin, lateral extent of chin color from the gonys,
and length of the dorsal anterior, dorsal posterior, and ventral eyespots. Bill size and shape,
measured by proximal and medial width and depth of the bill, were included in the total body
The congruency between genetic and morphological matrices was tested using
Mantel analysis (Mantel,1967). Significance levels for Mantel analyses for rejecting the null
hypothesis are P
s 0.05 and P
Levels of genetic variability within species are given in Table 4-1; allelic frequencies
are presented in Table 4-2. No locus was monomorphic across all species. Average observed
heterozygosity was significantly higher in tanagers, excluding Microligea palustris and
Coereba flaveola, (Ho
= 0.108) than in warblers
= 0.055) (t(19)
= 2.60; P
< 0.02) and was
higher than the average heterozygosity reported for other birds (Ho
= 0.063: N
= 85; Evans,
Observed heterozygosity for each species was within one standard error of expected
heterozygosity except for Dendroica pinus and D. coronata.
Values of observed heterozygosity
for each warbler species fell within one standard error of values reported elsewhere
(Barrowclough and Corbin, 1978; Avise et al., 1980), with the exception of Mniotilta varia.
The mean value found for this species was about four times higher than previously reported
(Barrowclough and Corbin, 1978; Avise etal., 1980).
Heterozygosity (S.E.) for Microligea palustris, an wood warbler endemic to Hispaniola,
was 0.087 (0.042).
was higher than that reported earlier (Ho
= 0.017; McDonald,
1987) based on 14 loci. Because estimates of heterozygosity are dependent on sample size and
the number and kinds of enzymes assayed, variation in levels of genetic variability across
studies is not unusual (Johnson, 1974; Archie, 1985; Simon and Archie, 1985). For example,
39 loci Ho
= 0.091 whereas for P. poliocephalus the values were Ho
= 0.099 and Ho
The number of alleles
, A (S.E
, averaged over loci for all species
was 4.6 (0.41).
ranged within species from 1.0 for Chlorospingus canigularis to 1.8 for Phaenicophilus
poliocephalus while per cent polymorphic loci (frequency of common allele s 0.99) ranged from
5% to 60%, respectively.
There were 15 unique alleles (found only in a single species):
creatine kinase-2, amino aspartate transaminase (AAT
-2, lactate dehydrogenase (LDH)-2,
malic enzyme (ME), purine nucleoside
phosphorylase (NP), 6-phosphoglucose dehydrogenase (6PGD)-2, and phosphoglucose
mutase (PGM)-3; two for LDH-1 and peptidase (PEP-L); and three for 6PGD-1.
Four alleles in
fumarate hydratase (FH)-1, isocitrate dehydrogenase (ICD)-1, NP, and PGM-1 were shared
between Phaenicophilus and Chlorospingus but not with Piranga. Eight alleles in GPI-
beta-glucorunidase (bGUS), ICD-1, PEP-L and PGD-1 were shared between Phaenicophilus
and Piranga but not with Chlorospingus (Table 4-2).
Matrices of Rogers' and Nei's
genetic distances (D) are provided in Table 4-3.
Nei's (1978) interspecific genetic distance in tanagers was D
= 0.128 and in warbler
= 0.234 (D
= 0.265 calculated using Nei, 1972).
The warbler value was an order of
magnitude higher than those reported by Avise et al. (1980)
= 0.043 (Nei, 1972).
loci were common to the two studies. Small sample
different kinds of enzymes assayed
and the distribution of variation across the enzymes will affect estimates of genetic distance
(J. Novak, pers. comm.).
Nei's (1978) D for intergeneric comparisons of tanagers (excluding
Microligea palustris and Coereba flaveola) was D
= 0.600 (N
= 110), for all warblers was
= 11), and between tanagers and warblers was D
= 0.650 (N
distance was highest between species of Chlorospingus and between Piranga ludoviciana and
(1978) genetic distance between species of Phaenicophilus and
Chlorospingus was D
= 0.542, while it
= 0.404 between Phaenicophilus and Piranga.
Average Manhattan distances, based on 21 morphological characters, were 23.6 and 15.4,
Both UPGMA and Distance-Wagner trees are presented since the affinities of the
palm-tanagers differed with the two methods (Fig. 4-1).
UPGMA phenograms are calculated
assuming equal importance for all sources of genetic variation, while Distance Wagner trees
weight certain types of variation. In addition, Distance Wagner trees make no assumptions
about homogeneous rates of divergence for different characters and thus, the branch lengths
for the OTU's may vary.
The cophenetic correlation for the UPGMA phenogram was 0.932,
with a % standard deviation (Fitch and Margoliasch,
Wagner trees, the goodness-of-fit statistics were 0.95
1967) equal to 7.12. For Distance
and 5.72, respectively, before
optimization. Neither the multiple additions criterion nor optimization added to the
interpretation of the patterns observed in the Distance Wagner trees.
The resolution of the relationships of the tanagers and the warblers based on the
dendrograms was complicated by the variation in the intra- and intergeneric branch lengths
(Fig. 4-1). Several species retained their same relative affinities when the data were
jackknifed: four main groups could be distinguished.
The groups included Piranga olivacea,
Chlorospingus ophthalmicus, Thraupis palmarum, and Eucometis penicillata as core species.
Euphonia music and Vireo griseus were more distantly related to the tanagers and warblers
than tanagers and warblers were to one another.
Wood warblers, with the exception of
Microligea palustris, branch off directly from the Eucometis penicillata-Coereba flaveola line.
Estimated divergence time for the warbler-tanager split was 1.7 to 10.5
X 106 years ago.
Microligea palustris, a purported wood warbler, was never aligned with the Eucometis-
were removed under these conditions; Spindalis zena was aligned with the Eucometis-C
flaveola cluster with Microligea palustris more distantly related to this group. Microligea and
Spindalis were most frequently aligned with the Piranga complex in the UPGMA phenogram
(100%) and in the Distance Wagner tree (67-95
%), except when the data for GPI-2 and PGM-3
were removed. The Piranga assemblage formed a sister group to the Chlorospingus
The three species of Piranga were associated 100%
of the time on both methods.
Eucometis was directly associated with Coereba 93-100% of the time (Table 4-4; Fig.
The Chlorospingus-Hemispingus association was less stable under all conditions with a
per cent association ranging from 60-90%.
Phaenicophilus was chiefly aligned with Piranga
in the Distance Wagner tree (87-95%) but
ciated with Chlorospingus in the UPGMA
phenogram (80-81%). Thraupis palmarum formed a consistent association with Nemosia
pileata and often was associated with the species in the Chlorospingus-Piranga branch
Thraupis palmarum was not closely associated with Spindalis or Phaenicophilus.
Wood warblers formed two main branches with one containing Mniotilta varia and the other
Dendroica coronata was closely associated with Mniotilta varia, and D.
palmarum and D. pinus were associated with Geothlypis.
The morphological phenogram (Fig. 4-3) showed patterns different from either the
genetic phenogram or tree.
Microligea palustris was paired with Chlorospingus
ophthalmicus, and this species pair formed a sister group to the Euphonia-Nemosia-Dendroica
dominica association. Phaenicophilus formed a sister group to the Piranga-Eucometis-
Mantel analysis of the relationship of genetic and combined
morphological (21 characters) distances gave a significant r
= 0.332 (P
Thus, only a
small amount of the pattern of variation in morphology can be accounted for by the
correlation with the pattern of variation in genetics.
The pattern of genetic differences among
Two clustering techniques, UPGMA and Distance Wagner trees, were used to
summarize phylogenetic relationships among tanagers with a focus on the relationship of
Phaenicophilus to other tanagers. Relationships hypothesized from UPGMA phenograms do
not necessarily reflect historical relationships among species because the phenograms are
based on statistical similarity between groups (Neff, 1987). Convergences can obscure the
true phylogenetic relationships, particularly if character differences are few and the direction
of character change is critical to defining phylogenetic branching sequences.
are sensitive to changes in character polarity and represent phylogenetic relationships more
accurately (Neff, 1987).
Three of the phylogenetic hypotheses proposing a close relationship between Spindalis,
Thraupis or Microligea to Phaenicophilus can be rejected based on the branching patterns
observed for the UPGMA phenogram and the Distance Wagner tree (Fig. 4-1).
The affinity of
Phaenicophilus to either Piranga or Chlorospingus remains ambiguous when the results from
the two methods are considered.
When data from loci or OTU's used to construct UPGMA
phenograms and Distance Wagner trees were jackknifed, the results were also equivocal.
degree of association (80-95%) of Phaenicophilus to Chlorospingus or to Piranga varied with
the clustering technique used.
The instability in these associations may result from to
problems of convergences in both morphology and genetics, and these convergences may be a
common feature of heterochronic systems. If the Distance Wagner tree more accurately
reflects true phylogenetic relationships, then Piranga is more closely related to
Phaenicophilus. Piranga olivacea is neotenic (Lawton and Lawton, 1986); therefore, neoteny
would not a derived condition in Phaenicophilus palmarum. Because this relationship is
critical in determining the polarity of evolution within PhaenicoDhilus. criteria other than
Data based on the number of shared rare alleles and genetic and morphological
distances can be used to test the two phylogenetic hypotheses regarding the closest relative of
(1978) distance between species of Phaenicophilus and
Chlorospingus ophthalmicus (D
= 0.542) is greater than that between species of
Phaenicophilus and Piranga olivacea (D
When all morphological features are
combined to make the same comparisons, the average Manhattan distances were Dm
and 15.38, respectively. In addition, Phaenicophilus shares twice
with Piranga than it does with Chlorospingus (8
Therefore, Phaenicophilus seems to be
more closely related to Piranga than with Chlorospingus, and thus, Phaenicophilus
poliocephalus is derived from P. palmarum.
The alignment of Phaenicophilus to Chloropsingus on the UPGMA phenogram seems
counterintuitive to the above conclusions. Chlorospingus is less similar to Phaenicophilus
based on genetic distance and shared rare alleles. Average genetic distance between two
points in the UPGMA phenogram is influenced by the average distance between clusters and
would probably differ from the absolute genetic distance between two taxa. Average genetic
distance between the species clusters of Phaenicophilus and Chlorospingus
the effect of averaging in the technique.
is lower because of
The polarity of the relationship of the two species of
Phaenicophilus should depend more on its relationship to its closest relative (i.e.,
olivacea) rather than its average relationship to tanagers.
The relationship of Microligea to tanagers and not warblers is clear from the UPGMA
phenogram and the Distance Wagner tree (Fig. 4-1
The relationship remains stable even
when data are jackknifed. If the genetic relationship of Microligea palustris to tanagers is
accepted, then similarities in body size and bill morphology must be due to morphological
convergence and not to their close phylogenetic affinities.
general foraging ecology, although Microligea palustris is considered more closely related to
the warbler genus Dendroica (A.O.U., 1983).
The morphological and ecological similarities
may result in competitive replacement of one species for the other that essentially result in
allopatric distributions in Haiti. Microligea palustris occurs in the Massif de La Selle and the
northwest peninsula of Haiti (McDonald, 1987) where Geothlypis is rare or absent. In
contrast, Geothlypis is common in the Massif de La Hotte where Microligea is absent.
similarity in general foraging ecology may be indicative of selection for convergence in bill
and body size and shape. However, a detailed analysis of foraging behavior reveals that
Microligea is more similar to tanagers than to warblers (Chapter 2). Interspecific similarities
between species may result from evolutionary convergences more often than
do similarities in behavior or genetics. For example, New World vultures resemble Old World
vultures in morphology and foraging ecology. Analyses based on DNA hybridization suggest
that New World vultures are related to storks (Sibley and Ahlquist, 1986). Furthermore, both
New World vultures and storks share a common behavior of defecating on their legs for
thermoregulation. In this example, consideration of the superficial morphological data leads
to the wrong conclusion, while the conclusion based on the concordance of behavior and
genetics seems correct.
The genetic and behavioral relationship of Microligea to tanagers and its similarity in
plumage to Phaenicophilus and Chlorospingus initially suggested the possibility that it
formed a link between Central American and Hispaniolan species. Microligea's alignment
with Piranga may seem perplexing. However, if heterochrony is an important process to the
adaptive radiation of tanagers, interspecific similarities due to convergences would not be
surprising. Microligea did cluster with Chlorospingus based on morphological similarities,
although these species were not closely aligned with Phaenicophilus because of body size
species of Phaenicophilus may be due either to retention of a primitive pattern or
heterochrony that leads to convergence. My analysis of plumage patterns cannot distinguish
between these two alternative hypotheses.
Convergence in morphology, relying on repetition of ontogenetic themes,
is more likely
if heterochrony commonly occurs in the adaptive radiation of species (Larson, 1980).
similarities in basic plumage patterns of the four Hispaniolan species, two Phaenicophilus,
Microligea, and Xenoligea, can be explained by assuming convergence due to heterochrony.
Using this explanation, the Hispaniolan species and Chlorospingus probably represent a
paedomorphic assemblage. Characters may be common within an assemblage because of
repetition of ontogenetic themes. Heterochrony may be more prevalent in tanager evolution
than considered previously. For example, the gray crown in Phaenicophilus poliocephalus
may be a terminal deletion with respect to the crown color in its closest antecedent, P.
palmarum, but when compared to the character state in Chlorospingus may represent a
character reversal. If true, then the lack of concordance between the UPGMA phenogram and
the Distance Wagner tree with respect to the affinities of Phaenicophilus could be attributed
to character reversal in P. poliocephalus.
The close affinities of warblers to Eucometis-Coereba require further evaluation by
other molecular techniques.
There may be a closer phylogenetic relationship than previously
thought based on the electrophoretic results.
Warblers do not form a distant sister group to
tanagers as suggested from the DNA hybridization data (Sibley, pers. comm.) but branch off
directly from Eucometis. Jackknifing data from loci or OTU's failed to disturb the stability of
The relationship of warblers to tanagers is closer than that of Euphonia to
other tanagers. Coereba flaueola forms the link between Eucometis and wood warblers and is
not clearly a tanager or a warbler based on the genetic data.
Phaenicophilus rather than Spindalis seems justified. Genetic similarities between Thraupis
palmarum and Nemosia pileata were unexpected because of differences in body
Although Thraupis and Nemosia resemble Phaenicophilus in basic plumage patterns, they
are not the closest relatives based on the genetic data (Fig. 4-1).
Intergeneric genetic distances are higher in tanagers than in warblers, though this
may be an artifact of the diversity of tanager genera sampled (9
between congeneric species of warblers (D
congeneric species of tanagers
vs 3). Genetic distances
= 0.238) are greater than those between
= 0.128) and exceed values reported for warblers elsewhere
(Barrowclough and Corbin, 1978; Avise et al., 1980).
The reason for these differences is not
obvious, but could be due to differences in heterozygosity reported in various studies, with
high heterozygosities possibly resulting in higher genetic distances. Average heterozygosity
for Dendroica species is essentially the same as values reported in other studies. Higher
average D values may result from fewer shared alleles between warbler species, although
there are fewer fixed allelic differences between warblers than in tanagers (Table 4-2).
Distances for tanagers fall within the range of those in other avian groups (Johnson and Zink,
1983; Christidis, 1987).
Phaenicophilus. This ih
The smallest genetic distance occurs between species of
s probably due to their recent divergence during the Pleistocene that
may have been facilitated by heterochronic mechanisms that rely on small changes in
The largest genetic distance in tanagers is between Piranga ludoviciana
and its North American congeners. Although Piranga ludoviciana superficially resembles
subadult P. olivacea, it more closely resembles P. erythrocephala, a resident species of Mexico
and Central America (R. Greenberg, pers. comm.) and subadult P. rubriceps, a South
Therefore, P. ludoviciana may not be directly derived from its North
American congeners, but rather from a series of heterochronic speciation events within
Tanagers have higher heterozygosities than do other birds and have significantly
higher levels than do warblers.
The higher tanager heterozygosity is not an artifact of the loci
examined, because values for both groups depend upon the same enzymes, and warbler values
are generally the same as those reported elsewhere. Higher genetic variability may be
correlated with differences in life history.
Although life history studies have not been
conducted for all of the tanagers, there may be a number of species with predefinitive
plumages (Isler and Isler, 1987) indicating a more widespread occurrence than is currently
recognized. For example, both Phaenicophilus and Piranga are neotenic tanagers that also
have higher than average levels of heterozygosity.
Thus it is tempting to speculate that high
levels of heterozygosity and the occurrence of heterochrony are correlated. Plasticity in life
history characteristics associated with high genetic variability may allow selection for
heterochrony to occur.
Convergence in a species assemblage might be expected to be more frequent if
heterochrony is a common phenomenon.
The low congruence observed between
morphological and genetic characters is not unique (Zink et al.,
1985) nor is it unexpected for
groups that diverged long ago. Rates of evolution for a variety of characters differ (Schnell
and Selander, 1981). Few studies have demonstrated a strong correlation between the degree
of morphological and genetic divergence in homeotherms as measured by structural gene loci.
Part of the reason for this may be that there is not a strong correlation because quantitative
assessments of structural gene changes may not provide the appropriate measure.
measure of assessing regulatory gene change is necessary. Standard methods of quantifying
structural gene changes may not be appropriate to assess changes in regulatory genes. A
small change in regulatory genes could effect a large change in morphology whereas a change
of similar magnitude in structural gene loci might have little or no affect on external
significant (Mantel test).
The relationship abstracted from the morphological phenograms
are most likely the result of similarities in body size (Fig. 4-2).
The six species in the upper
part of the phenogram have the largest body size; the remaining nine are generally small,
with the exception of Spindalis zena that is intermediate in body size.
When only body-size
related features are analyzed, the relationships among the largest species change little.
When only plumage features are analyzed, striking changes occur in the phenogram. Coereba
flaveola and Dendroica cluster separately; Piranga, Chlorospingus, Phaenicophilus,
Microligea-Spindalis cluster together, because they share similar plumage characters.
similarity in basic plumage pattern may be the result of small changes in regulatory genes
is consistent with the frequent convergences in morphology leading to a lack of
correlation between genetic and morphological features as observed.
Phylogenies should be evaluated by independent sets of characters. Some of these may
be more useful than others in calculating correct phylogenies. In the case of tanagers,
plumage similarities are more likely the result of convergences. Microligea is more closely
allied with tanagers based on the genetic and behavioral evidence (Chapter 2) but based on
body size and bill shape is convergent to warblers. If a choice of competing phylogenetic
hypotheses must be made, patterns based on two sets of characters, such as genetics and
special behaviors, may be more correct than those based on a limited set of morphological
The assessment of the utility of character
sets in defining phylogenies must be based on
an understanding of the limitations of the methodologies used in measuring characters.
Although the biochemical evidence supported the close relationship between Phaenicophilus
and Piranga originally determined by DNA hybridization data (Sibley, pers. comm.), the
conclusions from this evidence conflicted with the higher taxonomic relationships of warblers
study, there were sufficient numbers of shared alleles among species to compare the
phylogenies generated from electrophoresis and DNA hybridization. Based on the UPGMA
phenogram and Distance Wagner tree generated from the electrophoretic data, warblers are
more similar to Eucometis penicillata than Eucometis is to Euphonia music.
problems in assessing higher ordered relationships using DNA hybridization data
(Houde,1987; but see Ahlquist etal., 1987). Consequently, higher order relationships remain
ambiguous. Apparent similarities in structural genes used to support higher levels of
taxonomic relationships may be influenced by convergences for electrophoretic mobility more
than true genetic similarities. Future analyses must continue to test the robustness of the
various methodologies and to compare results from molecular techniques to those obtained
from other character sets.
List of species, number of specimens, and genetic variability for
25 species of
20 assayed enzyme loci.
P. ludoviciana "
Spindalis zena d
Microligea palustris d
C. canigularis e
Nemosia pileata t
Eucometis penicillata c
Coereba flaveola c
Thraupis palmarum t
Hemispingus atropileus C
P. poliocephalus d
Geothlypis trichas e
Dendrlim a npnsvhnunir* e
Table 4-1. (Cont'd)
D. palmarum "
D. coronata '
D. dominica c
Mniotilta varia t
Vireo griseus '
Note: Observed heterozygosity (Hobs), expected (Hexp), per cent polymorphic loci (P), and
average number of alleles per locus (A) are computed by BIOSYS-1 program (Swofford
and Selander, 1981). First sample size given is for the genetic analysis; sample sizes
listed for male (M) and female (F) are for morphological analysis.
Number and sex of specimens used in analysis. F
= female; M = male; U = unknown.
Specimens used in the morphological analysis were from Harvard University Museum of
Comparative Zoology, Smithsonian Institute U.S. National Museum in Washington, D.C.,
Carnegie Museum of Natural History in Pittsburgh, Philadelphia Academy of Sciences,
and collections made in Haiti in 1985.
Tissue samples provided by Louisiana State Museum of Zoology (LSUMZ) Frozen Tissue
Collection, Robert M. Zink, Curator.
Tissue samples and skins collected in Haiti from May-September, 1985.
deposited in LSUMZ; specimens are deposited in the American Museum of Natural
History in New York and the Florida State Museum in Gainesville.
9 4 4 0
'3 0 SOt
^- *^r- -^
SO cO to
0 CO O
-N- --^ *-t
t0 03 t
*-' .-^ ^
I* A *m
*"- t rl
^ 0 .
? S "
* C .-