Louisiana State University LSU Digital Commons
LSU Doctoral Dissertations Graduate School
2010 The oler of physical and ecological barriers in the diversification process of birds in the Guiana Shield, northern Amazonia Luciano Nicolás Naka Louisiana State University and Agricultural and Mechanical College, [email protected]
Follow this and additional works at: https://digitalcommons.lsu.edu/gradschool_dissertations
Recommended Citation Naka, Luciano Nicolás, "The or le of physical and ecological barriers in the diversification process of birds in the Guiana Shield, northern Amazonia" (2010). LSU Doctoral Dissertations. 1702. https://digitalcommons.lsu.edu/gradschool_dissertations/1702
This Dissertation is brought to you for free and open access by the Graduate School at LSU Digital Commons. It has been accepted for inclusion in LSU Doctoral Dissertations by an authorized graduate school editor of LSU Digital Commons. For more information, please [email protected].
THE ROLE OF PHYSICAL AND ECOLOGICAL BARRIERS IN THE DIVERSIFICATION PROCESS OF BIRDS IN THE GUIANA SHIELD, NORTHERN AMAZONIA
A Dissertation
Submitted to the Graduate Faculty of the Louisiana State University and Agricultural and Mechanical College in partial fulfillment of the requirements for the degree of Doctor of Philosophy
in
The Department of Biological Sciences
by Luciano Nicolás Naka B.S., Universidade Federal de Santa Catarina, 1999 M.S., Instituto Nacional de Pesquisas da Amazônia, 2001
August, 2010
To the memory of Jürgen Haffer.
Pa’ mis viejos, quienes siempre apoyaron mi raro gusto por las aves, confiando en mí y en el
(improbable) futuro de un ornitólogo en Argentina. Su amor incondicional y educación libre de creencias y prejuicios, hicieron que yo sea lo que hoy soy. Espero poder ofrecer a la pequeña
Sofia la libertad de pensamiento que ellos me brindaron, para que piense por si misma y sea todo lo que quiera ser. Pero novios, después de los 18…
Para Carolina e Sofia, os amores da minha vida, que esperaram por mim por tempo demais. Sem elas, a vida não e tão colorida, os cantos das aves não são tão bonitos, e o tempo não voa. Tou louco pra voltar pra casa, pra recuperar meu lugar de papai, amigo e confidente. Vocês nunca mais estarão sozinhas!
ii ACKNOWLEDGMENTS
Sometime ago, I heard that good PhD students do not need advisors, whereas lousy and outstanding ones need at least two. Not sure where I stand in that gradient, but I need to thank the two great advisors I had: J. Van Remsen and Robb T. Brumfield for being the best advisory team one can ever ask for. Support, confidence, guidance, and encouragement rarely mix in such good dose. The Museum of Natural Science at LSU has been an outstanding place to conduct my doctoral studies, and I honestly could not think of a better place on earth other than LSU to be a student in ornithology. I will be proud to have my picture depicted next to all the leading characters of Neotropical ornithology.
I am also very grateful to my academic committee, Mike Hellberg, Richard Stevens, Philip C
Stouffer, and Kent Mathewson, as well as the non-official members: Fred Sheldon, Jessica
Eberhard, and Mario Cohn-Haft. Fred has been a great director of the Museum, and he has always supported me and other students in making our time at LSU a fulfilling experience.
Jessica, has been like an older sister, sharing her insights and humor. Mario has been a mentor and friend, a shinning star of Amazonian ornithology worth following. He has taught me an awful lot, introducing me to the world of Amazonian biogeography, and sharing his experience and passion for the natural world.
Several LSU professors were vital in helping design my project by sharing their thoughts and experiences with me, including Chris Austin, Bryan Carstens, Prosanta Chakrabarty, David
W. Foltz, Mark Hafner, Kyle Harms, Anthony J. Lewis, and Kam-biu Liu. At the Museum and the Life Sciences Department, several people were extremely helpful: Peggy Simms & Gwen
Mahon, Allaina Ferguson, Thomas Moore, and particularly Tammie Jackson and Prissy
Milligan, who I pestered way too often with overdue deadlines, last minute requests, and obnoxious inquiries. Steven W. Cardiff and Donna L. Dittmann were outstanding in providing
iii help and advice during curatorial assistantship duties, preparation of specimens, and sharing their experience on dealing with matters such as obtaining exporting permits and genetic samples.
They were never too busy, or too tired to stop whatever they were doing to help me find invisible and rotten gonads inside oiled warblers from the Gulf. Natalie Rigby, the director of
International Services has been particularly helpful in matters related to visas and all paperwork.
Financial support within LSU was provided by the Department of Biological Sciences,
Graduate School, Biograds, and the Museum of Natural Science. My dissertation was an expensive one, and would have not been possible without the financial support generously provided by the following funding agencies: American Ornithologists’ Union (Research and
Tucker Travel Awards), American Museum of Natural History (Frank Chapman Grant Award),
Neotropical Ornithological Society (Travel Award), IdeaWild (Research Grant), Sigma-Xi
(Grants-in-Aid of Research), National Science Foundation (Doctoral Dissertation Improvement
Grant, DEB-1011512). Laboratory work was supported in part by NSF grants DBI-0400797 and
DEB-0543562 to Robb T. Brumfield. Most fieldwork in Brazil was supported by a Conselho
Brasileiro de Ciência e Tecnologia (CNPq) Research Grant to Magalli Pinto Henriques (Edital
Universal 486253/2006-6). For two years I was generously supported by funding from the
Instituto Internacional de Educação do Brasil (IIEB), and an Overseas Doctoral Fellowship from
CNPq (202131/2006-8). I hope all funding agencies and individuals that trusted me on this project are satisfied with this dissertation and the upcoming publications.
Fieldwork in Brazil was only possible because of the collaboration of several institutions, particularly the Instituto Nacional de Pesquisas da Amazônia (INPA) which provided essential logistic support for my research, and the Instituto Brasileiro do Meio Ambiente e dos Recursos
Naturais (IBAMA) for providing collecting permits. The local office of IBAMA in Roraima has provided substantial financial and logistic support to work in ecological stations and national
iv parks under their care. I am particular indebt to Antônio Lisboa and Bia Ribeiro-Lisboa,
Marlucia Lima and Iran das Chagas Almeida and their family, Bruno, Miúdo, and Branco for making my time at Parque Nacional do Viruá an invaluable experience. Antônio Galdino de
Souza (Estação Ecológica Niquiá), Bruno de Campos Souza and Andrea Lambert (Estação
Ecológica Maracá), and Inara Rocha Santos (Parque Nacional Serra da Mocidade) also were essential for the success of my research. I am particularly grateful to the following friends, who shared their time and experience with me in the field: Marina Anciães, Cathy Bechtoldt, Phred
Benham, Jean Philippe Boubli, Sergio Borges, Gustavo Bravo, Luiz Mestre, Ingrid Macedo,
Chico Pontual, and Marcela Torres. Many field assistants joined me in the expeditions. Murilo was particularly important in the success of several trips to remote places. Caçula was also a great assistant, and should now answer his never ending question: yes, we probably lost our place in heaven for collecting so many innocent birds. Sorry.
Several people were responsible for converting me from a total novice into a sequencing expert, who experienced every single issue that could go wrong in a lab. Katie Faust introduced me to the world of extractions; Matt Carling, Zac Cheviron, Susan Murray and Robb Brumfield were essential in the early days of DNA amplification and sequencing. Gustavo Bravo, Santiago
Claramunt, Andrés Cuervo, Elizabeth Derryberry, Haw-Chuan Lim, and James Maley always shared their expertise on those horrible days when good plates were as elusive as Yapacana
Antbirds. I am also greateful to the Whitehead lab, whose centrifuge saved me hours upon hours of extraction time.
My dissertation would have not been possible without tissue samples from scientific collections. I am indebt to all institutions and people who were involved in collecting and carefully preserving bird specimens and their valuable tissues. Six ornithological collections provided the bulk of the samples, and were particularly important for this project. Specifically, I
v would like to thank Robb Brumfield and Donna Dittmann from LSU; Mario Cohn-Haft and
Ingrid Macedo from INPA; Alexandre Aleixo and Fabíola Poletto from Museu Paraense Emilio
Goeldi (MPEG); Joel Cracraft, Paul Sweet, Thomas Trombone, and Peter Capainolo from the
American Museum of Natural History; Gary Graves and James Dean from the National Museum of Natural History; and John Bates and David Willard from the Field Museum of Natural
History. Without their generous collaboration, many patterns would have not been unveiled.
Those very same institutions also provided distributional data for the locality database on which my entire dissertation stands. I have no words to fully acknowledge the collaboration of those people above, but also those data provided by Gary Stiles (Instituto de Ciencias Naturales);
Mauricio Alvarez (Instituto Humboldt), David Caro (ProAves), Adriana Nieves (Colección
Biológica Rancho Grande), Mark Robbins (Kansas University); Kimball Garrett (Los Angeles
County Museum of Natural History), J.V. Remsen (LSUMZ), and F. Lima (MPEG). The following colleagues kindly provided unpublished distributional data to feed my ever hungry database: Diego Calderón, Mario Cohn-Haft, Anthony Crease, Fernando d’Horta, Renato Gaban-
Lima, Steve L. Hilty, Morton Isler, Curtis Marantz, Brian O’Shea, and Camila Ribas.
A special paragraph need to be devoted to Miguel Lentino, who opened the doors (and the entire database) of the Coleccion Ornitológica Phelps (COP) in Venezuela. He has been a friend and a central part in this project, always eager to help me understand individual distribution patterns. Jhonathan Miranda, Robin Restall, Margarita Martínez and particularly Yemayá López and Josmar had also been very helpful on my visits to Caracas (gracias Yemayita!). Clemencia
Rodner and her family were incredible to me, and treated me as one of their own.
Some parts of this dissertation benefitted from insightful comments from Catherine
Bechtoldt, Robb Brumfield, Carolina V. de Castilho, Santiago Claramunt, Mario Cohn-Haft,
Jessica Eberhard, Mike Hellberg, J.V. Remsen, Richard Stevens, and Phil Stouffer. The
vi implementation of BARRIER was greatly facilitated by the help of Michael Patten; Santiago
Claramunt and Sebastian Tello were invaluable in the execution of some statistical analyses in R;
Cathy Bechtoldt helped in the nuisances or raster transformations and is responsible for many of the spatial analyses that we will soon publish together; and Mike Hickerson and Naoki
Takebayashi were fundamental in the implementation of ABC analyses in MSBAYES. I also need to thank Wolfgang Buermann, Sassan Saatchi, Henri Thomassen, John McCormack, and Tom
Smith at the Center for Tropical Research for facilitating use of remote-sensing layers.
I am very grateful to the outstanding group of current and former LSU students who made my life in Baton Rouge such a nice part of my life. They had become friend for life and I will miss them: David, Eli and Yasmina (Wong)derson, Adri Bravo, Bravín, Curt Burney, Matt
Carling, Zacarias Cheviron, Claramuntia, Cuervito, Jessica Deichmann, Juicy Fernandes, Martín,
Erica (and now Mathilda) Fraile, Mechita Gavilanez, Ricardo Gibono, Danielito Lane, Haw-
Chuan Lim, Jamesito Maley, Jonathan Myers, Fabiana Mendoza, Brian O’Shea, César mae
Sánchez, Mary Seigot, Sebas Tello, Thomas Valqui, Andrésito Vidal, and many others.
I thank my family: pá, má, Martín y Florencia, who have always provided unlimited doses of love and support. My mom, who never fully understood how I could make a living studying pajaritos, has been my biggest fan. Ale, Javi, Kaia, Male, Mel, Joaquin, and Facu: los adoro!
And of course, I need to thank Carolina, my girlfriend, wife, and life companion. Whitout her help, I could have never finished my dissertation in a timely manner. She sacrificed a year of her life to come with me to Baton Rouge. Carolina has been the most amazing companion with whom I could share a life. I am very happy that we have endured and survived the difficult times of being an ocean apart for many months. And best of all, we have been rewarded with la pequeña Sofia! I know you cannot read these words yet Sofi, but when you do, you should know that você é o sol das nossas vidas.
vii TABLE OF CONTENTS
ACKNOWLEDGMENTS………………………………………………………………….…….iii
ABSTRACT……..………………………………………………………………………….……ix
CHAPTER 1 INTRODUCTION…………………….…………………………………………..….…..1
2 AVIAN DISTRIBUTION PATTERNS IN THE GUIANA SHIELD: IMPLICATIONS FOR THE DELIMITATION AND UNDERSTANDING OF AMAZONIAN AREAS OF ENDEMISM..……………...... ………………………..……….5
3 THE ROLE OF RIVERS IN GENERATING AND MAINTAINING AVIAN DIVERSITY IN THE RIO NEGRO BASIN: A COMPARATIVE APPROACH……26
4 THE ROLE OF GEOGRAPHIC AND ECOLOGICAL BARRIERS IN THE LOCATION OF AVIAN CONTACT ZONES AND PHYLOGEOGRAPHIC BREAKS IN THE GUIANA SHIELD……………………………………………………….….. 66
5 CONCLUSIONS………………………………………………………………….…...109
REFERENCES………………………………….……………………………………...... 113
APPENDIX A LIST OF 24 GUIANA SHIELD POSSIBLE ENDEMIC TAXA NOT INCLUDED IN THE ANALYSES ………………………………………..………………….……….133
B COMMUNITY COMPOSITION MATRIX DATA FROM 34 LOCALITIES FROM THE GUIANA SHIELD ……………………….…………………………………… 134
C BEST-FIT MODELS AND PARAMETERS FOR ALL TAXA…………...……...…141
D EXAMPLES OF PUTATIVE HYBRIDS FROM THE HEADWATERS……...... 143
E RESULTS FROM THREE INDEPENDENT JOINT POSTERIOR DENSITIES FOR THE RIO BRANCO USING ABC SIMMULATIONS……………………………...144
F RESULTS FROM THREE INDEPENDENT JOINT POSTERIOR DENSITIES FOR THE RIO NEGRO USING ABC SIMMULATIONS………………………………. 145
G LIST OF 26 PAIRS OF TAXA THAT REPLACE WITHIN THE GUIANA SHIELD NOT INCLUDED IN THE ANALYSES……………………………………..……...146
VITA…………………………………………………………………………………………....147
viii ABSTRACT
Understanding the factors that influence the formation and location of distribution boundaries is important for the study of evolutionary processes. These factors can be studied effectively at suture zones, regions containing disproportionally high numbers of contact zones (CZs) and phylogeographic breaks (PBs). Together, CZs and PBs offer complementary views of current and historical factors that separate or bring together populations of closely related taxa. For my dissertation, I studied a suture zone in northern Amazonia, where ~100 pairs of taxa replace one another geographically.
I analyzed the Guiana Shield avifauna, using bird distributions to redefine the boundaries of the Guianan area of endemism. I showed that the Rio Branco is an important biogeographical barrier and a natural western limit for this area, although smaller rivers, savannas, and mountains also play a significant role. A multivariate approach revealed that the Branco/Negro interfluvium represents a transitional zone for birds, suggesting that the longstanding view of
Amazonia as a mosaic of parapatric areas of endemism likely represents an oversimplification of current patterns.
I investigated the role of rivers in maintaining and generating biodiversity by testing predictions of the ‘riverine barrier hypothesis’. Using a comparative phylogeographic approach,
I found that phenotypically differentiated populations across rivers are reciprocally monophyletic. The lower Rio Negro represents a stronger barrier to gene flow than does the upper Rio Negro, but no genetic homogenization occurs towards the headwaters. Most ‘riverine barrier hypothesis’ predictions were not supported, suggesting that rivers are key to maintaining biodiversity, but not for its generation.
Finally, I explored the role of physical and ecological factors in the location of CZs and PBs.
PBs cluster along physical barriers, whereas CZs aggregate at the headwaters of large rivers.
ix Nearly half of the pairs that come into contact hybridize, and show significantly lower genetic distances than pairs that come into contact and do not interbreed, suggesting that time of isolation as inferred from genetic data may predict their likelihood of hybridization. Ecological niche models showed significant levels of niche divergence between pairs, suggesting that environmental variables cannot be ruled out as factors influencing the location of suture zones.
x CHAPTER 1
INTRODUCTION
“On this accurate determination of an animal's range many interesting questions depend. Are very closely allied species ever separated by a wide interval of country? What physical features determine the boundaries of species? Do the isothermal lines ever accurately bound the range of species? What are the circumstances which render certain rivers and certain mountain ranges the limits of numerous species, while others are not?” Alfred Russel Wallace (1852)
The Amazon basin has fascinated biologists ever since Wallace revealed to the Zoological
Society of London the unique patterns he had observed during his travels in the Amazon. He found that different forms of similar-looking monkeys and birds replaced one another across some large rivers, and even mentioned that natives were “perfectly acquainted with this fact, and always cross over the river when they want to procure particular animals, which are found even on the river's bank on one side, but never by any chance on the other” (Wallace, 1852).
Many regions on earth have impressive rivers, such as the Congo, the Mississippi, and the
Nile, yet nowhere else on the planet are rivers known to subdivide animal distributions to the extent observed in Amazonia. During the 160 years since Wallace’s visit to the Amazon, biologists documented the role of rivers in the delimitation of biotic distributions in groups as diverse as birds (Haffer, 1969), primates (Hershkovitz, 1977; Ayres & Clutton-Brock, 1992;
Silva & Oren, 1996), butterflies (Brown, 1982; Hall & Harvey, 2001), frogs (Ron 2000), lizards
(Vanzolini, 1973; Avila-Pires, 1995), and plants (Prance, 1982).
Despite the voluminous work of researchers and collectors, many of Wallace’s questions remain unanswered. For example, we do not know whether many closely related species have gaps between their ranges or are truly parapatric, simply because many areas in Amazonia have
1 never been sampled before (Oren & Albuquerque, 1991). We are also still unaware which physical features of the landscape determine the boundaries of species, given that few species’ entire distributions are known with certainty. Although good range maps are available from many different sources, particularly for birds (Ridgely & Tudor, 1994; Hilty, 2003; Ridgely et al., 2007), these are either geographically restricted or limited to the species level, neglecting the patterns of independent evolutionary lineages currently recognized as subspecies. Furthermore, whereas we may have a good idea of the general distribution patterns of Amazonian birds, we rarely know details at the river’s headwaters, where rivers are too narrow to act as physical barriers to dispersal. To date, the headwaters of large rivers may well represent the final frontier in Amazonian biogeography, although in some regions, such as southern Amazonia, the vegetated headwaters are disappearing at unprecedented rates. Likewise, we have no idea whether isothermal lines (or any other climatological feature) ever bound the ranges of species in
Amazonia. And at present, we can only speculate as to why some rivers limit the distributions of numerous species, whereas others do not.
These are some of the questions I address in my dissertation, using a comparative phylogeographic approach, which involves examining geographical and genetic patterns across multiple co-distributed taxa (Bermingham & Moritz, 1998; Avise, 2000; Arbogast & Kenagy,
2001; Riddle et al., 2008). Those are questions of broad interest that have remained unanswered for too long, and can now be addressed using a combination of modern methods and new analytical approaches, many of which were unavailable just ten or twenty years ago. Our ability to study population genetics has opened outstanding opportunities for studying Amazonian biogeography: old but long-standing hypotheses can now be tested directly (Moritz et al., 2000;
Aleixo, 2004), novel cryptic patterns are being described (Moritz et al., 2009), and new statistical frameworks add rigor to a field that a few years ago relied on subjective interpretations
2 (Carstens et al., 2005; Richards et al., 2007; Hickerson et al., 2009). Likewise, Geographical
Information Systems (GIS) and ecological niche models, with newly available environmental layers and ever improving algorithms, have provided the baseline for an outburst of spatial analyses (Kidd & Ritchie, 2006). Unfortunately, these types of analyses, which have been successfully applied in other regions and which can potentially resolve the complexities of animal distributions, have yet to be used extensively in the study of Amazonian biogeography
(Kozak et al., 2008).
The main challenge of my dissertation was to integrate these new techniques into the study of avian distribution patterns in the Amazon basin. I focused on one of the most extraordinary regions of the Amazon basin: the Guiana Shield. Covering much of northern South America, the
Guiana Shield is located between the Amazon and the Orinoco rivers, in a region dominated by a rare mixture of ancient and recent geological features, with some of the most pristine and least- populated areas in the world (Hammond, 2005a). In the Guiana Shield, evolutionary processes can be studied in their natural state. Biogeographical barriers are still active, without the confounding factors of deforestation, and the smoke of civilization has not yet blurred the evolutionary history of its landscapes and biota. From a biogeographical perspective, the Guiana
Shield lies at the confluence of four areas of endemism (Cracraft, 1985). The meeting of these distinct avifaunas has already been described by Haffer (1974), but not fully studied until now.
From an ecological standpoint, the Guiana Shield represents one of the most environmentally heterogeneous regions of the Amazon, where terra firme forests, flooded forests, savannas, white-sand forests, and the magnificent tabletop mountains known as Tepuis all intermingle in a fashion not seen elsewhere in Amazonia.
This study was motivated by my personal fieldwork in the Rio Negro basin since 2001, when
I observed that many Guianan endemic birds showed different distribution patterns. As I
3 surveyed the avifauna along the Rio Negro, I realized that the terra firme forest avifaunas on both banks became more similar to one another as I moved towards the headwaters of the river, an observation made by Wallace more than 150 years before!
My dissertation is divided into three main research chapters (Chapters 2-4), which aim to explore different aspects of a single theme: understanding the patterns of avian distributions and replacement, exploring the role of physical and ecological factors in the delimitation of contact zones and phylogeographic breaks, and evaluating their implications for the diversification process. In the first research chapter (Chapter 2), I focus on the endemic avifauna of the lowland terra firme forests of the Guiana Shield and use their distributional patterns to redefine the boundaries of the Guianan area of endemism. In the second research chapter (Chapter 3), I shift my interest to the patterns of replacement of pairs of closely related taxa across large rivers, testing the predictions of the ‘riverine barrier hypothesis’. My goal here is to investigate the role of rivers as evolutionary forces by analyzing their contribution to the maintenance and generation of biodiversity. In the last research chapter (Chapter 4), I explore the role of physical and ecological factors in the location of avian contact zones and phylogeographic breaks in the
Guiana Shield. I also analyze the outcomes of secondary contact (hybridization or reproductive isolation), and use the time of isolation between pairs as inferred from molecular data to correlate with their likelihood of hybridization. Finally, I use ecological niche models to evaluate the role of ecological factors in the geographic location of suture zones.
4 CHAPTER 2
AVIAN DISTRIBUTION PATTERNS IN THE GUIANA SHIELD: IMPLICATIONS FOR THE DELIMITATION AND UNDERSTANDING OF AMAZONIAN AREAS OF ENDEMISM
INTRODUCTION
The Amazon basin has long fascinated ecologists and biogeographers, not only for harboring the highest biodiversity on earth, but particularly because of the uniqueness and apparent simplicity of the distribution patterns of its biota. During the early days of Amazonian exploration, Alfred
Russel Wallace realized that rivers represented meaningful barriers to birds and monkeys
(Wallace, 1852). On opposite banks of some large rivers, closely related, yet well differentiated taxa, replaced one another. These patterns of geographic and ecological substitution across large rivers, covering an entire river basin of continental proportions, are unique. This apparent simplicity implies that by determining which rivers represent the region’s main biogeographical barriers, general distribution patterns can be identified without undue attention to the distribution of every taxon, many of which are still poorly known.
The importance of large Amazonian rivers as biogeographical barriers results in spatial congruence in the distributions of plants and animals, and was used to define what Wallace
(1852) called ‘districts’, which were later described as ‘areas of endemism’ by Cracraft (1985).
Areas of endemism are based on shared congruent distributions (Platnick, 1991) and are vital for historical biogeography, representing the smallest unit of study (Nelson & Platnick, 1981).
Despite the controversy related to whether they really are minimal study units (Hausdorf &
Hennig, 2003; Szumik & Goloboff, 2004), areas of endemism remain essential in the quest to understand the distribution of life on earth, and are the entities compared when the goal is to search for explanations of biological patterns based on earth history (Linder, 2001). Although
5 general congruence in species distributions is required, precise congruence at all scales is not
(Morrone, 1994; Hausdorf, 2002), because different ecological requirements and dispersal abilities among species will prevent exact congruence. Nonetheless, the correct delimitation of areas of endemism is crucial for understanding historical processes (Harold & Mooi, 1994), because the assumption is that the biota of an area of endemism shares a common history, and that the biogeographical relationships of its elements should be relatively similar, resulting in incongruence when this assumption is not met (Linder, 2001).
In Amazonia, areas of endemism are well established (Cracraft, 1985) and have rarely been disputed. They have been widely used to study area relationships (Cracraft & Prum, 1988; Bates et al., 1998; Marks et al., 2002; Eberhard & Bermingham, 2005), to test evolutionary and historical hypotheses (Aleixo, 2004; Ribas et al., 2005; Aleixo & Rossetti, 2007), and even to suggest conservation priorities (Conservation International, 2003; Silva et al., 2005). At present, up to nine Amazonian areas of endemism are recognized for birds (Cracraft, 1985; Silva et al.,
2002; Borges, 2007). The generality of these areas of endemism has been confirmed by studies of other groups, such as plants (Prance, 1982), butterflies (Brown, 1982), lizards (Avila-Pires,
1995), primates (Silva et al., 1996), and frogs (Ron, 2000).
In spite of this apparent generality, a recent review of avian phylogeographic studies in
Amazonia (Aleixo & Rossetti, 2007) and a community-wide Parsimony Analysis of Endemicity
(Bates et al., 1998) showed that common historical and biogeographical patterns among most taxa are rarely found, and intricate evolutionary scenarios and species-specific historical events have been suggested to account for these incongruence (Bates et al., 1998; Cohn-Haft, 2000;
Bates, 2001; Cheviron et al., 2005). Complex and species-specific evolutionary scenarios are likely (Bush, 1994), but a potential source of error in such analyses is that Amazonian areas of endemism, as currently defined, do not accurately represent avian distribution patterns.
6 Therefore, it is critical to define unambiguously the boundaries of these areas and the taxa that characterize them.
Here, I investigated in detail the distribution patterns of birds in the Guiana Shield, studying in particular, the western limit of the region known as the Guiana area of endemism, one of the most heterogeneous regions of Amazonia in terms of vegetation, geology, and topography (Naka et al., 2006). My specific goals were to: i) identify the endemic avian taxa that characterize this area of endemism, ii) define the boundaries of this area of endemism for birds, iii) evaluate whether individual boundaries coincide with landscape features, and iv) investigate degree of congruence in distributions of Guianan endemics. By answering these questions, I attempted to provide a critical assessment of the utility of Amazonian areas of endemism, as presently described, to the field of biogeography.
METHODS
The Guiana Area of Endemism
The Guiana Shield covers much of north-eastern South America, accounting for ~13% of the continent, and representing a complex land of ancient soils and recent geological features
(Hammond, 2005a). This region is typically defined by the Orinoco, Amazon and Japurá rivers, going as far west as the state of Vaupés in Colombia (Conservation International, 2003;
Hammond, 2005b) (Fig. 1). Although the exact boundaries of this geological region are still unclear (Gibbs & Barron, 1993; Hammond, 2005a), the existence of the Guiana area of endemism has not been controversial. Back in 1852, Wallace defined the ‘Guiana district’ as bounded by the Negro and Amazon rivers. Although three other districts defined by Wallace’s were subsequently subdivided by biogeographers into smaller areas, the Guiana district remained largely unchanged. Haffer (1969) used avian distributions to delineate the ‘Guianan center of dispersal’, defined by the same boundaries as Wallace. Muller (1973) reduced the ‘Guianan
7 centre’ to only include the lowland forests of Guyana, Surinam, French Guiana, and Amapá, in
Brazil, whereas Cracraft (1985) defined its limits more precisely as bounded by the Amazon
River to the south, and the highlands of Guyana and south-eastern Venezuela to the west, recognizing that several species extended further westward, “some as far as the Rio Negro and the southern border of the Venezuelan highlands”.
Figure 2.1. General location of the Guiana Shield (outlined) in northern South America, as defined by the conservation priority-setting workshop (Conservation International, 2003), available on-line at http://www.guayanashield.org. General location of landscape features mentioned in the text: 1. Lower Rio Negro; 2. Middle Rio Negro; 3. Upper Rio Negro; 4. Upper Rio Orinoco; 5; Sierra de Parima; 6. Río Caura; 7. Rio Paragua; 8. Río Caroni; 9. Sierra de Lema; 10. Gran Sabana; 11. Roraima-Rupunnuni savannas.
8 Selection of Taxa
The Guiana Shield is home to nearly 1,000 species of resident birds (Conservation
International, 2003). Of these, ~ 400 taxa (including species and subspecies) are endemic or near endemic to the region (Naka, unpublished data). To avoid confounding different factors, I restricted my analyses to species that are mainly restricted to the primary habitat type of the region: lowland terra firme forest, excluding species that occupy other habitats, such as savannas, white-sand, dry, flooded, or montane forests. I defined Guianan endemics as those taxa that do not occur south of the Amazon River and west of the Rio Negro. Technically, taxa occurring west of the Rio Negro could still be considered Guiana Shield endemics (in geological terms), yet the ‘Negro-Solimões’ interfluvium has traditionally been considered part of the Napo or Imeri areas of endemism (Cracraft, 1985), and more recently as an endemic area on it own, the
Negro area of endemism (Borges, 2007). Therefore, a taxon found throughout the Guiana Shield would occur in at least two areas of endemism and would no longer be a ‘Guianan endemic’.
I created a preliminary list of endemic taxa (species and subspecies) based on distributional data from the ornithological literature, including the works of Hellmayr (1929-1938), Peters et al., (1931-1986), Zimmer (1933-1944), Hellmayr & Conover (1942-1949), Pinto (1978), and more recent works such as del Hoyo et al., (Vols. 1-12, 1992-2008), but the final inclusion of a given taxon was based on the most updated taxonomic revision available (Remsen et al., 2010).
The use of subspecies in biogeography is controversial because some subspecies may not represent independent evolutionary units (McKitrick & Zink, 1988; Phillimore & Owens, 2005).
Most subspecies included in this study, however, represent well-defined, diagnosable units, and many of them may merit species-level rank under the biological species concept. Under other species concepts, such as the phylogenetic species concept, all diagnosable taxa merit species- status. Therefore, I decided to use described taxa (whether they represent species or subspecies)
9 as study units. Described subspecies from the Guiana area of endemism that were not unambiguously diagnosable, or had too few samples to create distributional maps, were excluded from analyses (see Appendix A).
Distribution Data
I created distribution maps for each taxon based on a large distributional database assembled from museum specimens, published sources, audio recordings, and my own fieldwork in the region. I obtained data from many institutions, including the Instituto Nacional de Pesquisas da
Amazônia (INPA) and Museu Paraense Emílio Goeldi (MPEG), for Brazil; Colección
Ornitológica Phelps (COP) and Colección Biológica Rancho Grande (CBRG), for Venezuela;
United States National Museum of Natural History (NMNH), for Guyana; Instituto Humboldt and Project BIOMAP (which includes collections from several sources, such as the Instituto de
Ciencias Naturales, ICN) for Colombia (see BIOMAP homepage for details, http://www.biomap.net); the American Museum of Natural History (AMNH), Field Museum of
Natural History (FMNH), and the Museum of Natural Science at the Louisiana State University
(LSUMZ) for several countries. I also included a database for the Brazilian State of Roraima
(Naka et al., 2006), which includes specimens and published locality data from 11 institutions and 78 published sources, respectively. Finally, I used S. Hilty’s databases for Venezuela (Hilty,
2003) and M. Isler’s unpublished database for the Thamnophilidae. Most data were already georeferenced, but when geographic coordinates were not available I used national gazetteers
(Paynter & Traylor, 1982, 1991; Paynter et al., 1981) to obtain geographic coordinates as accurately as possible. Between 2001 and 2008 I conducted extensive fieldwork in the region, conducting general avian surveys, collecting specimens, and recording vocalizations. These specimens and recordings were deposited at the INPA bird collection and the Amazonian Sound
Archive (ASA). At present, the distributional database contains ~ 28,000 entries.
10 Subspecies identification from collections require careful revision (Graham et al., 2004).
Given the large amount of data, it was not feasible to positively verify every specimen. I focused on examining specimens in Brazil (INPA and MPEG) and Venezuela (COP and CBRG).
Curators and collection managers from several institutions kindly reviewed specimens under their care. I visually checked for potential errors by studying individual maps, paying particular attention to records located at the edge of the distributions, where western allotaxa could be present and generate identification mistakes. I built a database using Microsoft Access 2007, and generated maps by importing georeferenced locality data into ArcView 9.3.1 (Environmental
Systems Research Institute, 2009).
Avian Distribution Patterns and Boundaries of the Guiana Area of Endemism
To investigate distribution patterns and to define the geographic boundaries of the Guiana area of endemism, I used avian range maps and community composition data to: i) search for common patterns among individual distributions, and ii) to build a hotspot map of endemic species richness in the region. Range maps were built by connecting point localities with the minimum convex polygon method (Swenson, 2008), using a maximum buffer of 50 km to enclose forested areas around those points, except when such buffers would traverse a potential barrier. To build the endemic taxa richness map, individual polygons were converted into raster files, and these were overlaid using the ArcGIS Spatial Analyst Extension through the raster calculator in ArcView 9.3.1 (Environmental Systems Research Institute, 2009). Geographic boundaries were proposed based on the areas that contained 25%, 50%, and >75% of endemics.
I used the 50% threshold as the value to define boundaries of the Guiana area of endemism.
Community composition data were analyzed using nonmetric Multidimensional Scaling
(MDS), an ordination method used to explore similarities within a dataset and to illustrate clustering and regional transitions (Conran, 1995; Proches & Marshall, 2001). To run the MDS,
11 I compiled a presence/absence data matrix for 34 different localities (Table 2.1) spanning ~2,500 km from Amapá to the Andean foothills (Fig. 2.2). Each locality was characterized by the presence or absence of 181 taxa (see Appendix B), including Guianan endemics and their western allotaxa, when present. Widespread taxa were excluded because they would be uninformative in these analyses. Data from these 34 sites were obtained form different sources, including my own fieldwork, published data, unpublished data provided by colleagues, web sources, and museum specimens (Table 2.1). When specimens from a given locality were not available, I used specimens from nearby localities to determine subspecies. To quantify composition similarity among localities I used a modified Jaccard distance (Jm) proposed by
Patten & Smith-Patten (2008), where Jm = 100 - [!A"B!/!A#B!] x 100 x [!B!/!A!], with ‘A’ representing a site with a set of species, ‘B’ another site with another set of species, ‘"’ the species in common between the sites, and ‘#’ the addition of all species. Jaccard's distance is well suited for presence/absence data (Birks, 1987; Patten & Smith-Patten, 2008). The correction implemented by Patten & Smith-Patten (2008) deals with false absences. This is necessary because some of the localities used in the analyses represent heavily collected and well-known areas, whereas others were only studied during short surveys. For the latter cases, some absences (0 in the matrix) may represent false negatives.
I used point localities, rather than grids, because in a region as heterogeneous as the Guiana
Shield (with mountains, savannas, flooded and white-sand forests), large cells would probably not include comparable areas of terra firme forest, whereas many randomly located small cells would have few or no samples. The use of point localities also allows the inspection of nearby sites as independent units, which are useful to evaluate the role of physical barriers (see below).
The ordination analysis was performed in R (R Core Development Team, 2008) using the isoMDS function in the ‘Mass’ library (Venables & Ripley, 2002), with 25 iterations using
12 different random initial configurations to minimize the stress due to local optima. The Jaccard function of the ‘Prabclus’ library (Hennig & Hausdorf, 2008) was adapted for computing the modified Jaccard distance.
Figure 2.2. Map of northern South America showing localities at which bird composition data were gathered (see Table 2.1 for locality names) and output results from the BARRIER analysis. Polygons around each locality represent Voronoi tessellations and lines connecting the points are the Delaunay triangulations. Darker lines represent the barriers found, with line width proportional to their relative strength, according to their bootstrap support (see Table 2 for average values).
Identification of Biogeographical Barriers
I used Monmonier’s Maximum Difference Algorithm (Monmonier, 1973) as implemented in
BARRIER v. 2.2 (Manni et al., 2004) to determine the location of major biogeographical barriers in the region. Monmonier’s algorithm identifies boundaries from a distance matrix from specific localities. For this analysis, I employed the same 34 localities and the modified Jaccard distance matrix as used for the ordination analysis. In BARRIER, localities are mapped using an x,y
13 coordinate system (i.e., decimal degrees). A series of polygons are created around each point
(Voronoi tessellations), and from them, the algorithm builds Delaunay triangulations connecting adjacent points on a map (see Fig. 2.2). The distance (dissimilarity) matrix is mapped onto the triangulation such that each pair of localities has an associated distance. Monmonier’s algorithm then builds biogeographical boundaries, beginning with the maximum pairwise distance and continuing until i) the edge of the map is hit, ii) a loop is formed, or iii) a previously computed barrier is reached (Manni et al., 2004). A great advantage of this program is that it allows the use of a bootstrap method to determine the relative strength of the barriers identified. I used a specific function in R (R Core Development Team, 2008) to resample, with replacement, species within each site and to generate 100 pseudoreplicate matrices. For a more detailed description of
Monmonier’s algorithm and its implementation in BARRIER see Manni et al. (2004).
RESULTS
I identified 90 avian taxa (25 species and 65 subspecies representing 28 families) as endemic to terra firme forests of the Guiana area of endemism (Table 2.2). Avian families with the highest number of representatives are Thamnophilidae (17 endemic taxa), Dendrocolaptidae
(11), Tyrannidae (9), Picidae (5), and Ramphastidae (5). Another 24 taxa that are possible endemics were not included in the analyses because of poor sampling (8 taxa) or uncertain taxonomic status (16 taxa) (subspecies identification was ambiguous with the material studied)
(see Appendix A).
An endemic taxa richness map of individual ranges showed that the number of endemic taxa drops consistently towards the west (Fig. 2.3). All endemic taxa occur in the forests east of the
Rio Branco, whereas less than half occurs west of this river. Therefore, the Amazon River, lower Rio Negro, and Rio Branco and its associated savannas (Roraima-Rupunnuni and the Gran
Sabana) seem to define well the boundaries of the Guiana area of endemism for birds. Further
14 north, the Río Caroni and Sierra de Lema also coincide with the boundaries of many bird species
(Fig. 2.3).
Table 2.1. Localities from which species lists of avian taxa were gathered for the analyses (see geographic location in Fig. 2.2). No. Locality Sources 1 Brazil: Amapá, Porto Grande C.A. Marantz and M. Cohn-Haft (in litt.) 2 Brazil: Amapá and Pará, Jari Barlow et al. (2007) 3 French Guiana: Nouragues field station Thiollay (1994) Brazil: Amapá, Parque Nacional Montanhas do 4 Tumucumaque Bernard (2008) 5 Suriname: Brownsberg J.H. Ribot (http://webserv.nhl.nl/~ribot/) 6 Suriname: Raleigh falls/Voltzberg J.H. Ribot (http://webserv.nhl.nl/~ribot/) 7 Guyana: Iwokrama Forest Reserve Ridgely et al. (2005) 8 Guyana: Acary Mountains Robbins et al. (2007) 9 Guyana: West Kanukus Parker et al. (1993) 10 Brazil: Amazonas, Manaus Cohn-Haft et al. (1997), Naka (2004). 11 Brazil: Roraima, Parque Nacional do Viruá Naka (unpubl.) 12 Venezuela: Bolívar, Reserva Florestal Imataca COP database 13 Brazil: Roraima, Caracaraí Naka et al. (2006), Naka (unpubl.) 14 Brazil: Roraima, Estação Ecológica Maracá Silva (1998), Naka (unpubl.) 15 Venezuela: Bolívar, El Paují Crease (2009), Naka (unpubl.) 16 Brazil: Amazonas, Rio Demeni/Aracá C. Ribas and R. Gaban (in litt.), Naka (unpubl.) 17 Venezuela: Bolívar, Cerro Guaiquinima COP database 18 Brazil: Amazonas, Campina do Rio Preto Naka (unpubl. data) Brazil and Venezuela: Amazonas, Serra do Naka and M. Cohn-Haft (unpubl.), AMNH 20 Tapirapecó database 21 Brazil: Amazonas, Rio Marauia Naka (unpubl.) 19 Venezuela: Bolívar, lower Río Caura COP database 22 Venezuela: Amazonas, Cerro de la Neblina Willard et al. (1991), COP database 23 Venezuela: Amazonas, Campamento Junglaven Zimmer & Hilty (1997) 24 Venezuela: Amazonas, San Carlos del Rio Negro COP database Brazil: Amazonas, São Gabriel da Cachoeira (east 25 bank) Naka (unpubl.) 26 Venezuela: Amazonas, Yavita-Pimichin COP database Brazil: Amazonas, São Gabriel da Cachoeira (west 27 bank) Naka (unpubl.) C. Ribas, F. Horta and R. Gaban (in litt.), and C.A. 28 Brazil: Amazonas, Manacapuru Marantz (in litt.) 29 Brazil: Amazonas, Parque Nacional do Jaú Borges et al. (2001), Naka (unpubl.) Brazil: Amazonas, Reserva de Desenvolvimento 30 Sustentavel Amanã Naka and M. Cohn-Haft (unpubl.) 31 Brazil: Amazonas, Estação Ecológica Juami-Japura M. Cohn-Haft (in litt.) D. Calderón (in litt.), BIOMAP, COP and MPEG 32 Colombia: Vaupés, Mitú databases Colombia: Caquetá and Guaviare, Parque Nacional 33 Natural Serranía de Chiribiquete Álvarez et al., (2003), BIOMAP database 34 Colombia: Meta, Parque Nacional Tinigua Cadena et al., (2000), BIOMAP database
15
Figure 2.3. Endemic taxa richness map based on 88 taxa. White dashed line represents the 50% threshold overlap point, and the delimitation of the Guiana area of endemism suggested in this study. Darker colours represent areas with higher number of endemic taxa.
A two-dimensional multidimensional scaling (MDS) ordination (stress = 5.02) illustrates that terra firme forests of the Guiana Shield (as a geomorphological unit) contain two distinct avifaunas, divided roughly by the Rio Negro (Fig. 2.4). Bird species composition east of the Rio
Branco, and west of the Rio Negro appears to be quite homogeneous. Localities within the
Branco/Negro interfluvium, however, seem to represent a large transitional zone between the two former areas (Fig. 2.4).
The Monmonier’s algorithm identified 10 putative barriers for the Guianan avifauna within the Guiana Shield (Table 2.3). These barriers coincide with known physical features of the landscape, strongly suggesting a causal effect. The barriers include five rivers, two savannas, and two mountain ranges. According to these analyses, the Rio Branco represents the second- most important barrier for birds within the Guiana Shield, second only to the lower Rio Negro.
16 Table 2.2. List of 90 taxa endemic to the Guiana area of endemism. Those taxa with trinomials indicate that the endemic taxon is currently recognized as a subspecies (65 taxa), whereas binomials indicate a species-level recognition (25 taxa). The distribution pattern refers to the general location of its western boundary (B: Rio Branco, BN; Branco/Negro interfluvium; N: Rio Negro). Physical boundaries indicate the specific barrier that seems to coincide with the western edge of their distributions. Avian Family Endemic taxon1 Dist. patt.3 Physical boundaries4 Cracidae Penelope m. marail1 BN B-Río Caura Psophidae Psophia c. crepitrans N N-Orinoco Psittacidae Pyrrhura p. picta BN N-Orinoco Brotogeris c. chrysoptera BN BN-Orinoco Pyrilia caica B B-Río Caura Trochilidae Topaza p. pella BN BN-Sierra Parima Phaethornis s. superciliosus N N-Orinoco Thalurania f. furcata1 BN BN-Orinoco Trogonidae Trogon r. rufus BN BN-Sierra Parima Momotidae Momotus m. momota N N-Orinoco Galbulidae Galbula a. albirostris BN B-RR-GS-Orinoco Galbula d. dea B B-RR-Caroni Bucconidae Notharchus m. macrorhynchos B B-RR-GS- Río Caura Monasa atra N N-Orinoco Capitonidae Capito niger B B-RR-GS-RC-Sierra de Lema Ramphastidae Ramphastos t. tucanus BN BN-Upper Orinoco Ramphastos v. vitellinus BN BN-Upper Orinoco Selenidera culik BN BN (Araca)-Caroni Pteroglossus viridis N N-Orinoco Pteroglossus aracari nigricollis B B-Caura/Orinoco Picidae Veniliornis cassini BN BN (C de la Neblina)-Alto Orinoco Piculus f. flavigula N N-Orinoco Celeus u. undatus1 B B-RR-GS-RC Celeus e. elegans1 B B-RR-GS-RC-Sierra de Lema Celeus t. torquatus BN BN-Río Caura Furnariidae Synallaxis rutilans dissors N N-Orinoco Automolus infuscatus cervicalis B B-RR-RC Automolus rubiginosus obscurus B B-RR Xenops minutus ruficaudus N N-Orinoco Dendrocolaptidae Dendrocincla f. fuliginosa B B-GS-RC-Sierra de Lema Dendrocincla m. merula B B-RR-GS Deconychura l. longicauda B B-RR-GS-Paragua Sittasomus griseicapillus axillaris BN BN-Alto Orinoco/Ventuari Glyphorynchus s. spirurus B B-RR-Paragua Hylexetastes perrotii BN BN (Maracá)-RC-Sierra de Lema Dendrocolaptes c. certhia N N-Orinoco Dendrocolaptes p. picumnus N N-Orinoco Xiphorhynchus pardalotus1 N N-Orinoco Xiphorhynchus guttatus polystictus BN BN-Orinoco/Ventuari Lepidocolaptes a. albolineatus B B-RR-GS- RC-Sierra de Lema Thamnophilidae Cymbilaimus l. lineatus B B-RR-GS- RC-Sierra de Lema Frederickena viridis BN BN (Aracá)-Caura Thamnophilus amazonicus divaricatus B B-RR-GS-RC-Sierra de Lema Epinecrophylla gutturalis B B-RR-GS-RC Myrmotherula guttata N N-Orinoco Myrmotherula a. axillaris BN BN (Aracá)-Ventuari
17 (Table 2.2 con’d.) Avian Family Endemic taxon1 Dist. patt.3 Physical boundaries4 Thamnophilidae Myrmotherula m. cinereiventris B B-RR-GS-Paragua Herpsilochmus stictocephalus Trombetas-GS-Sierra de Lema Hypocnemis cantator1 B B-RR-GS-RC Cercomacra cinerascens immaculata B B-RR-GS-RC Cercomacra tyrannina saturatior BN BN-Alto Orinoco-Cuchivero Percnostola rufifrons1 B B-RR-GS-Upper Essequibo Schistocichla l. leucostigma B B-RR-GS-RC-Sierra de Lema Myrmeciza f. ferruginea B B-RR-GS-RC-Sierra de Lema Pithys a. albifrons N N-Orinoco Gymnopithys rufigula1 N N-Orinoco Willisornis p. poecilinotus BN BN (Demeni)-upper Orinoco Grallaridae Grallaria v. varia B B-RR-GS-RC-Sierra de Lema Hylopezus m. macularius B B-RR-GS-RC-Sierra de Lema Myrmothera c. campanisona B B-RR-GS-RC-Sierra de Lema Conopophagidae Conopophaga a. aurita B B Tyrannidae Zimmerius acer5 BN BN (Aracá)-upper Orinoco Lophotriccus vitiosus guianensis B B-Essequibo Tyrannidae Hemitriccus zosterops rothschildi6 B B-RR-GS Todirostrum pictum N N-Orinoco Tolmomyias assimilis examinatus B B-RR-RC Platyrinchus coronatus gumia BN BN- Sierra Parima- Río Caura Onychorhynchus c. coronatus N N-Orinoco Terenotriccus e. erythrurus BN BN-RC-Sierra de Lema Neopipo cinnamomea helenae B B-Guyana Pipridae Tyranneutes virescens B B-RR-GS- RC Lepidothrix serena B B-RR-GS-Guyana Chiroxiphia pareola pareola BN BN-(Aracá)-Paragua/Río Caura Corapipo gutturalis N N-Orinoco Tityridae Schiffornis turdina olivacea B B-RR-GS-Guyana Iodopleura fusca BN BN (Maracá)-RC Insertae sedis Piprites chloris chlorion N N-Orinoco Vireonidae Hylophilus m. muscicapinus N N-Orinoco Hylophilus ochraceiceps luteifrons B B-RR-GS-RC Corvidae Cyanocorax cayanus B B- Río Caura Troglodytidae Microcerculus b. bambla B B-RR-GS-RC-Sierra de Lema Pheugopedius c. coraya B B-RR-GS-RC Cyphorhinus a. arada1 BN BN (Maracá)-RC Polioptilidae Polioptila g. guianensis B B-RR-GS Thraupidae Tachyphonus s. surinamus BN BN (Aracá)-upper Orinoco Tangara m. mexicana B B-RR-GS-Guyana Cyanerpes c. caeruleus BN BN (Rio Cauaburi)-Sierra Parima Hemithraupis f. flavicollis B B-RR-GS- RC -Sierra de Lema Parulidae Phaeothlypis rivularis mesoleuca N N-Orinoco Fringillidae Euphonia cayennensis B B-RR-GS- RC 1 Includes two endemic subspecies (see Appendix B). 2 Coarse western boundary of each endemic taxon: RR: Roraima-Rupunnuni savannas; GS: Gran Sabana; RC: Río Caroni; areas in parenthesis after BN represent likely barriers within the Branco/Negro interfluvium. 5 Z. acer occurs in NE Brazil, but its stronghold remains within the Guianan area of endemism 6 The form rothschildi has been considered valid by Cohn-Haft (2000), based on molecular and vocal data (2000).
18 Whereas the Rio Branco seems to represent a significant barrier to many avian taxa, the
savannas north of the Branco appear to be more permeable than the river. At least four taxa
bounded by the Rio Branco do bridge the Roraima-Rupunnuni savannas, whereas seven have
populations on both sides of the Gran Sabana. Similarly, at least seven taxa with distributions
bounded by the Rio Branco occur west of the Río Caroni, two of which appear to be bounded by
the Río Paragua and five by the Río Caura in Venezuela.
Despite representing a single area of endemism, Guianan endemics do not share a single
geographic distribution towards the west, and show at least three distinct distribution patterns: i)
taxa limited by the Rio Branco and its associated savannas (42 taxa); ii) taxa bounded by the
Negro and Orinoco Rivers, coinciding with the Venezuelan-Colombian border (21 taxa); and iii)
taxa that occur west of the Rio Branco and the savannas, but do not reach the upper Rio Negro or
the Orinoco (27 taxa). One endemic taxon (Herpsilochmus stictocephalus) does not reach the
Rio Branco from the east, where it is replaced by its allotaxon H. dorsimaculatus.
Table 2.3. Main physical barriers identified as biogeographically important for birds in the Guiana Shield. Average bootstrap support is the number of times a barrier appeared in BARRIER, using resampled data matrices. The number of endemic taxa bounded by each barrier was independently assessed with individual taxon maps (Table 2.1). Ave. bootstrap value Putative barrier (100 reps.) No. of taxa delimited Lower Rio Negro 100 86 Rio Branco 97.16 41 Roraima-Rupunnuni savannas 96.5 36 Middle Rio Negro 87.25 32 Río Caroni 66.5 30 Gran Sabana and Sierra de Lema 65 34 and 16 Upper Rio Negro 55.5 18 Sierra de Parima 45.2 12 Río Caura 38.5 8 Upper Orinoco 35.66 8
19
Figure 2.4. Multidimensional scaling (MDS) ordination of the 34 localities analyzed, according to their differences in bird species composition (modified Jaccard distance for presence/absence data). Localities are numbered as in Table 2.1. Localities 1-12 are located east of the Rio Branco, and its associated savannas (a); localities 13-26 are located within the Branco/Negro interfluvium (b); and localities 27-34 are located west of the Rio Negro (c).
DISCUSSION
The delimitation of areas of endemism is a central issue in historical and descriptive biogeography, and several methods have been applied recently to define these areas in a more objective way than Wallace could have done 150 years ago (Morrone, 1994; Harold & Mooi,
1994; Mast & Nyffeler, 2003). Amazonian areas of endemism have already been identified and described, but their coarse boundaries remain as preliminary as when they were formally proposed 25 years ago (Cracraft, 1985). The role of large rivers in defining animal distributions has apparently made detailed distributional studies seem less critical in Amazonia than in other
20 regions, because they remain as default explanations for most distribution patterns. The present study suggests otherwise. Not only is it likely that many of their limits need to be redrawn, but entire areas of endemism may not be well supported for birds (Borges et al., 2001; Borges, 2007) or other organisms (Racheli & Racheli, 2004).
The role of large rivers has been critical in the biogeographic history of Amazonia
(Capparella, 1991; Haffer, 1992). My analyses, however, reveal that minor barriers (medium- size rivers, savannas, and mountain ranges) also define boundaries of an area of endemism.
Distribution maps revealed that Guianan endemics show at least three distinct distributional patterns (Table 2.2), suggesting a shared effect of different barriers on phylogenetically independent taxa with disparate ecological requirements and unequal dispersal abilities. Some endemic taxa, however, have species-specific distribution patterns, whose boundaries do not appear to coincide with physical barriers. In these cases, ecological interactions with closely related taxa, rather than physical barriers, may play a significant role in boundary delimitations
(Naka, unpublished data)
This study represents one of the first attempts since Haffer (1969, 1974) and Cracraft (1985) to quantify the distribution patterns of an entire endemic Amazonian avifauna. Although some individual patterns may change with further sampling, I predict that general patterns described here are robust. Geographic Information Systems, easier access to distant field sites, and more complete sampling allow studying avian distributions at a level of detail not possible in the past.
I expect that similar studies of other areas of endemism may expose the importance of minor barriers for birds and unveil different patterns within apparently homogeneously distributed avifaunas. Recent fieldwork in the Rio Madeira basin showed that small rivers, such as the Rio
Aripuanã, also represent barriers to birds, a phenomenon called ‘mini-interfluves’ by Cohn-Haft et al. (2007). Also, focusing on taxa restricted to habitats other than terra firme forest, such as
21 savannas, flooded, or white-sand forests, may prove to be fruitful avenues of research in terms of revealing novel patterns of endemism (Naka et al., 2007, Cohn-Haft et al., 2008).
Guianan Endemics and the Utility of Amazonian Areas of Endemism
In this study, I identified 90 avian taxa as endemic to the lowland terra firme forest of the
Guiana area of endemism. Previous estimates for this area of endemism identified 47, 34, and 60 avian taxa by Haffer (1974), Cracraft (1985), and Nores (1999), respectively. The difference between these previous studies and the current one is even greater than these numbers suggest, because I excluded nearly half of the taxa noted by previous authors because they were more typical of habitats other than terra firme forest, such as savannas, white-sand, or dry forests.
Also, some taxa previously considered endemic (e.g. Pachyramphus surinamus, Cotinga cotinga, Haematoderus militaris, Cyanicterus cyanicterus and Periporphyrus erythromelas), are more widespread than previously thought (e.g., Stotz et al., 1997) and occur outside this area of endemism. The reason why I identified more endemic taxa than previous studies seems to simply be the result of a more exhaustive search, suggesting the need to apply the methods used here to other areas of endemism.
Identification of areas of endemism is important for historical biogeography and conservation biology (Linder, 2001). If the main goal is to identify regions with a common history, the inclusion of species from different habitats should be avoided. The endemic taxa of the Guiana area of endemism suggested by previous authors encompassed habitats with contrasting geological histories. For example, when a forest interior species undergoes a population contraction, a savanna species could be experiencing a population expansion (Haffer, 1969).
Although the hypothesis of areas of endemism as areas of common diversification in Amazonia has not been well tested, the inclusion of taxa from different habitats under a single area of endemism is likely to create composite areas that may obscure any possible historical event that
22 could be revealed by study of taxa with a common history. Similarly, it is useful to distinguish between narrow and broad endemics. Species with abnormally small distributions not shared by other taxa are also likely to have had an idiosyncratic history that may not add towards the understanding of areas of endemism, and should therefore be analyzed separately (Stott, 1981).
I suggest that Amazonian areas of endemism, as defined today, are only useful because of the apparent simplicity of having large rivers defining them, rather than because of a careful analysis of endemic taxa species composition in each area. If we are to find common histories and phylogeographic patterns, we should first define properly the boundaries, extent, and species composition of all areas of endemism. Bates et al. (1998) suggested that too much emphasis has been placed on finding evolutionary processes when general patterns remain poorly understood.
This study represents an attempt to describe patterns more accurately, and similar studies should be conducted in other areas of endemism.
The Role of Physical Barriers and Boundaries of the Guiana Area of Endemism
All evidence presented here suggests that the Rio Branco and its associated savannas act as significant barriers to avian dispersal in the Guiana Shield, representing the most natural boundary for the Guiana area of endemism. This configuration is consistent with that suggested by Cracraft (1985), who despite presenting relatively similar boundaries did not mention specifically the Rio Branco or the savannas. Recent publications continue to depict the Guiana area of endemism as bounded by the Amazon River to the south and the Rio Negro and Orinoco to the west, just as Wallace suggested in 1852 (e.g., Bates et al., 1998; Nores, 1999; Silva et al.,
2005). This is hardly surprising, given that many Guianan endemics, as acknowledged by
Cracraft (1985), do reach the Orinoco and upper Rio Negro. Despite this apparent ambiguity, the
Branco/Negro interfluvium holds both eastern and western elements, and should not be included within the Guiana area of endemism.
23 Monmonier’s algorithm provides a powerful tool for identifying biogeographical barriers
(Kidd & Ritchie, 2006; Patten & Smith-Patten, 2008). This algorithm revealed that the Rio
Branco represents a major barrier for birds in the Guiana Shield, second only to the lower Rio
Negro, the third-largest Amazonian river (Goulding et al., 2003). Individual range maps support this finding in showing that at least 42 taxa are bounded by this river, positioning the Rio Branco within the top five most important Amazonian rivers in terms of biogeographical importance to birds (together with the Amazon, the lower Negro, the Madeira, and the Tapajós).
Although the Rio Branco is relatively small by Amazonian standards (it ranks 18th among the 20 longest Amazonian rivers (Goulding et al., 2003), and at least 12th in discharge volume
(Ferreira et al., 2007)), it possesses several features that possibly enhance its effect as a biogeographical barrier. First, like other Amazonian rivers, the Rio Branco has relatively large floodplains and flooded forests along its banks, which also act as dispersal barriers to terra firme forest species (Remsen & Parker, 1983; Naka et al., 2007). Second, the Rio Branco lies in a very ecologically heterogeneous region, which could directly affect avian dispersal patterns by diminishing opportunities for dispersal (Naka et al., 2006). Finally, the Branco headwaters are located within the Roraima savannas, which also act as barriers to terra firme birds. Therefore, the Rio Branco in conjunction with the flooded forests and the savannas at its headwaters collectively represent an effective barrier for terra firme birds. West of the Rio Branco other barriers seem to play a role, including the Sierra de Parima, the Paragua, and the Caura Rivers.
Monmonier’s algorithm detected the differential effect of different sections of the Rio Negro as biogeographical barriers, showing the diminishing power of this river along a gradient from its mouth towards the headwaters. Wider portions of Amazonian rivers represent stronger barriers to animals than their narrower headwaters (Beven et al., 1984; Ayres & Clutton-Brock, 1992;
Haffer, 1992; Hayes & Sewlal, 2004). In the case of the Rio Negro, however, the weakening of
24 the barrier shown by Monmonier’s algorithm seems to be a consequence of filters for dispersal preventing Guianan endemics to reach the Rio Negro headwaters.
Ecological and Biogeographical Importance of Transition Zones
The longstanding view of the Amazon Basin as constituting a mosaic of parapatric areas of endemism delimited by major rivers appears to be an oversimplification of the biogeographic patterns. Delimiting the Guiana area of endemism by the Rio Branco and its associated savannas results in the Branco/Negro interfluvium being outside any formal area of endemism. This region can be characterized as a large transition zone between two distinct avifaunas –one that occurs east of the Rio Branco and the other, west of the Rio Negro. That several contact zones occur within the interfluvium, away from major rivers (Haffer, 1974), illustrates the importance of the region as a transitional zone between areas of endemism that flank the interfluvium.
The conservation importance of transition zones has been debated (Brooks et al., 2001; Smith et al., 2001; Araújo, 2002), but the ecological and biogeographical interest of such areas is undisputed. By definition, transition zones should represent areas of high species richness and beta and gamma diversity (Odum, 1971), but there has been debate concerning whether this is the result of the overlap of two or more biogeographic assemblages along boundaries, or whether these regions also hold high concentration of rare or endemic species (Araújo & Williams, 2001;
Kark et al., 2007). In the case of the Branco/Negro, the absence of endemics in the interfluvium supports the idea that its high species richness comes from species turnover. This does not devalue the evolutionary importance of this region, because it harbors a disproportionately high number of contact zones and phylogeographic breaks (Naka unpublished data). Transition zones represent natural laboratories for taxonomic, evolutionary, and ecological studies (Hewitt, 1988;
Schilthuizen, 2000; Spector, 2002), and the Branco/Negro interfluvium, seems to represent a hotspot of evolutionary processes for birds.
25
CHAPTER 3
THE ROLE OF RIVERS IN GENERATING AND MAINTAINING AVIAN DIVERSITY IN THE RIO NEGRO BASIN: A COMPARATIVE APPROACH
“During my residence in the Amazon district I took every opportunity of determining the limits of species, and I soon found that the Amazon, the Rio Negro and the Madeira formed the limits beyond which certain species never passed. The native hunters are perfectly acquainted with this fact, and always cross over the river when they want to procure particular animals, which are found even on the river's bank on one side, but never by any chance on the other. On approaching the sources of the rivers they cease to be a boundary, and most of the species are found on both sides of them. Thus several Guiana species come up to the Rio Negro and Amazon, but do not pass them… In going up the Rio Negro the difference in the two sides of the river is very remarkable.” Alfred Russel Wallace (1852)
“I do not mean to say that Amazonian rivers were unimportant, but they appear to have been overrated as barriers.” Jürgen Haffer (1997a)
INTRODUCTION
One of the most striking features of the Amazon basin is that some large rivers define the distributions of many animal taxa, particularly among birds and monkeys. These rivers separate the ranges of phenotypically distinct populations of closely related taxa, which are either recognized as different subspecies or species pairs. This biogeographical pattern was first described by Wallace (1852) and later acknowledged by many naturalists, biogeographers, and evolutionary biologists (Bates, 1863; Snethlage, 1913; Mayr, 1942; Sick, 1967), and continues to captivate contemporary biologists (Hershkovitz, 1977; Capparella, 1988; Ayers & Clutton-
Brock, 1992; Haffer, 1992; Colwell, 2000; Gascon, 2000; Moritz et al., 2000; Aleixo, 2002).
Not surprisingly, rivers were considered the main cause of speciation in the first diversification model proposed to explain Amazonian species richness. Although Wallace
26 himself did not explicitly put forward what is now known as the ‘riverine barrier hypothesis’, it is believed that he saw rivers as sources of speciation, as he often cited them (Wallace, 1876) when discussing barriers to dispersal in what would later become the modern model of vicariant cladogenesis (Colwell, 2000). Under the ‘riverine barrier hypothesis’, the formation of the extensive Amazonian river system induced speciation in forest-dwelling animals by fragmenting their ranges and preventing gene flow between once-continuous populations (Sick, 1967;
Hershkovitz, 1977; Capparella, 1988). Although it represented a compelling idea to explain current patterns of differentiation and distribution, it rapidly lost prominence with the proposal of
Haffer’s (1969) ‘Pleistocene refugia hypothesis’. Not only did Haffer provide an elegant mechanism that could be applied to any region in the world, but he also argued that rivers were barriers to secondary contact, rather than the original causes of speciation. An alternative proposed by Ayres (1986), and later supported by Ayres & Clutton-Brock (1992) and Capparella
(1991), was that whereas the broad lower courses of several rivers isolated animal populations
(as stated by the ‘riverine barrier hypothesis’), the forests in the headwaters (often at the periphery of the Amazonian forest) contracted during glacial periods isolating populations in
‘semi-refugia’ away from the headwaters. This hypothesis was later designated as the ‘river- refuge hypothesis’ by Haffer (1992, 1993), because it combined aspects of the ‘riverine- barrier’ and ‘refuge’ hypotheses of diversification. An expert on avian distributions, Haffer recognized that the lower portions of Amazonian rivers effectively divide populations of forest birds, with consequent genetic differences between populations separated by wide river courses, yet he was convinced that rivers alone could not have led to full speciation at such a large scale (Haffer,
1978; 1987; 1992; 1997a; 2002).
One of the major problems of the ‘riverine barrier hypothesis’ is that although the lower (and wider) portions of Amazonian rivers may represent unsurpassable barriers to forest species, their
27 narrower headwaters could, in theory, be easily crossed by dispersing individuals (Haffer, 1992;
1997a, 2001). In that case, gene flow would still occur between populations, and chances of full speciation, although still possible, would be reduced or slowed down considerably (Haffer,
1997b; Coyne & Orr, 2004; Price, 2008). Data on population interactions and gene flow in the headwaters of Amazonian rivers, however, are limited. Therefore, although population pairs could potentially come into contact there, generalized gene flow is not known to occur in these headwaters. Other aspects of ‘riverine biogeography’ that remain poorly explored include the levels of genetic divergence across rivers in many pairs of taxa, the phylogenetic relationship of species pairs, and whether replacing taxa are the result of simultaneous divergence.
Fortunately, these issues have clear predictions under the ‘riverine barrier hypothesis’ and are amenable to falsification in a phylogeographic framework: i) Replacement of taxa should cluster along rivers, not within interfluvia (Haffer, 1974;
Moritz et al., 2000). This prediction (a topic explored in chapter 4) can be tested using both morphological and genetic data, although it is important to evaluate the concordance between phenotypic and phylogeographic breaks. Few comparative studies have investigated whether populations across rivers are genetically distinct (but see Capparella, 1998; Bates et al., 2004;
Burney & Brumfield, 2009), although some individual species-level phylogenies show the importance of different rivers in delimiting the distributions of distinct clades (Aleixo, 2004;
Marks et al., 2002; Sardelli, 2005; Fernandes, 2007). ii) Pairs of taxa divided by rivers should represent each other’s closest relative (Haffer,
1997a; Moritz et al., 2000). Species-level phylogenies should provide the information to answer this question. Although it is arguable whether a subsequent divergence event in one of the populations could render a non-sister relationship between a pair truly separated by a barrier, a distant phylogenetic relationship between pairs would suggest a scenario of secondary contact
28 (Haffer, 1997a; Moritz et al., 2000; Patton et al., 2000). Unfortunately, species-level phylogenies are available only for a limited number of species. iii) Genetic differentiation between populations across rivers should decrease towards the headwaters (Hershkovitz, 1977; Haffer, 1978; Ayres & Clutton-Brock, 1992). Given that headwaters are much narrower than lower portions of rivers, cross-river gene flow in these areas are expected to increase. Although this is relatively easy to test, it requires a hierarchical sampling design that includes samples from both along and across a river. Surprisingly, these empirical data are only available from the Rio Juruá in Amazonia, where decreases in levels of genetic divergence towards the headwaters has not been found (Patton et al., 1994; Gascon et al.,
1996, 1998, 2000; Lougheed et al., 1999). Unfortunately, the Rio Juruá is possibly one of the worst rivers to test for this, because it has never been proposed as a biogeographical barrier
(Haffer, 1992). iv) Pairs of co-distributed taxa across a primary barrier should show evidence of a single divergence event (Avise, 2000). Testing for this is central for our understanding of evolutionary processes (Bermingham & Moritz, 1998; Avise, 2000; Arbogast & Kenagy, 2001; Hickerson et al., 2006a), and particularly for assessing the role of rivers in the creation of biodiversity. A major issue in addressing this question is that simple comparisons of genetic levels of divergence across co-distributed taxa (Capparella, 1988, 1991; Bates et al., 2004; Whinnett et al., 2005) are difficult to interpret, mostly due to the stochastic nature of gene coalescence (Edwards & Beerli,
2000). Under the coalescent theory, several population parameters, such as mutation rate, effective population size, gene flow, and different demographic histories such as population expansions or bottlenecks, could dramatically affect time estimates (Edwards & Beerli, 2000;
Arbogast et al., 2002), resulting in either type I errors (relatively similar temporal estimates in populations that diverged at different times) or type II errors (different estimates in species that
29 diverged simultaneously). Therefore, to make comparative phylogeographic inferences, models that account for these sources of variation must be used; these characterize multiple phylogeographic datasets in a single analysis and focus on quantifying the statistical probability of simultaneous divergence (Hickerson et al., 2006a; 2007). In recent years, analytical frameworks such as approximate Bayesian computation (ABC) have been developed exactly for this aim (Beaumont et al., 2002; Hickerson et al., 2006a; 2006b; 2007; 2009) and seem appropriate to test whether multiple populations of birds were simultaneously dissected by the formation of rivers. v) Rivers should be older than the populations they divide. If rivers were important in the genesis of populations across their margins, then their estimated date of origin necessarily must predate the inferred time of diversification of the animal populations that they divide. The palaeohydrological tools available to answer this question, however, are limited, and our current understanding of the age and origin of Amazonian rivers remains fragmentary (Latrubesse, 2003) and controversial (Hoorn & Vonhof, 2006).
In this chapter, I explore some of these predictions by investigating avian distribution patterns in the Guiana Shield, along and across two biogeographically important Amazonian rivers: the Rio Negro and the Rio Branco. The main goal of this chapter is to evaluate the role of these two Amazonian rivers as forces involved in the maintenance and generation of biodiversity, thereby testing the main predictions of the ‘riverine barrier hypothesis’. In particular, I address the following questions: i) is the location of phenotypic differentiation in populations across the Branco and Negro rivers concordant with the location of phylogeographic breaks? ii) are pairs of taxa across rivers closely or distantly related phylogenetically? iii) how effective are Amazonian rivers in preventing gene flow between populations across rivers, and what happens at the rivers’ headwaters, where replacing populations can potentially come into
30 contact? iv) are there differences in the levels of genetic divergence across the lower, middle, and upper section of a river? and v) are pairs of co-distributed taxa across the two rivers the result of simultaneous divergence?
To answer these questions, I gathered the largest molecular and distributional datasets ever assembled for any Amazonian river, using a multi-species approach. Museum specimens and genetic samples across rivers allowed evaluating the role of rivers in the location of breaks, as well as investigating the concordance of phenotypic and genotypic variation. A hierarchical sampling design across the Rio Negro was used to estimate levels of genetic divergence along the river, whereas reciprocal monophyly and museum specimens were used to infer levels of gene flow, both along the main course of rivers and at the rivers’ headwaters. Finally, I used coalescent simulations under an ABC approach to test for simultaneous divergence across both rivers. Although both rivers are important biogeographical barriers to birds and are part of the same basin, these rivers differ in their time of formation, geomorphological features, and ecology. By investigating two very different river systems, I was able to identify common patterns that might be generalizable to other river systems, and I also found important differences that illustrate the idiosyncratic nature of Amazonian evolutionary history.
METHODS
Study Area: Rio Negro and Rio Branco
The Rio Negro is the largest black-water river in the world and the second largest tributary of the Amazon (Goulding et al., 2003). Based on water volume, it is the world’s fifth-largest river, discharging 28,400 m3/s into the Amazon River (Latrubesse et al., 2005). Despite its size, the amount of sediments carried by the Negro is very low compared to other Amazonian rivers (e.g., up to 30 times lower than the Rio Madeira). The sources of the Negro come from the confluence of the Guainia, a river originating in the Colombian Llanos, and the Brazo Casiquiare, along the
31 Colombian-Venezuelan border. Unlike most other Amazonian rivers (and like the Congo river), the Negro headwaters are located in a region of low relief and run through cratonic areas on stable Precambrian crystalline basement, characterized by wide channels with islands alternating with narrow reaches with rapids, and with a very low yield of sediments (Latrubesse et al., 2005;
Franzinelli & Igreja, 2002). The course of the Rio Negro can be divided into three sections according to its geological features (Franzinelli & Igreja, 2002). First, the upper Negro runs from its sources to the locality of Santa Isabel do Rio Negro, where the river leaves the crystalline rocks of the Guiana Shield and flows across the sedimentary deposits of the Amazon
Basin. There the channel is straight with slow-flowing sections alternating with rapids and waterfalls. The maximum depth is about 12 m in the dry season. Second, the middle Negro flows on the Tertiary sedimentary deposits, in a general NW–SE direction, where the river has large channels, with numerous sandbars especially near its southern end at the confluence with the Rio Branco, where the crystalline rocks of the basement form outcrops that force the river into a narrow channel. Upstream from the Rio Branco, where the Negro attains widths greater than 20 km and depths greater than 18 m in the dry season, it resembles an active sedimentation basin, with the crystalline rocks at the mouth of the Branco forming a barrier that causes an upstream accumulation of sediments (Latrubesse & Franzinelli, 1998). Third, the lower Rio
Negro extends from 62°W to its confluence with the Amazon River (60°W), also trending NW–
SE, where it displays tectonic control. In this section the river is up to 20-km wide and is bordered by discontinuous cliffs. The lineament belongs to a larger transcurrent system of geological features (faults, folds, and rhombochasms) that occur in the whole Amazon Basin. In biogeographical terms, the Rio Negro represents one of the most important dividers of the entire
Amazon (Wallace, 1852; Haffer, 1974), second only to the Amazon River itself (Ayres &
Clutton-Brock, 1992). Nearly 100 pairs of phenotypically differentiated bird populations replace
32 each other across the lower sections of this river (see chapters 2 and 4). The upper Rio Negro, however, ceases to act as such strong barrier, and only ~15 pairs remain phenotypically differentiated on opposite banks along this section.
The Rio Branco, despite being the main tributary of the Rio Negro, is a much smaller river, discharging ~2,500 m3/s at its mouth (Ferreira et al., 2007). The Branco is one of the few rivers in the Guiana Shield that can be considered a ‘muddy’ river. Although not as turbid as the
Amazon or the Madeira rivers, its waters contrast strongly with the dark waters of the Negro. In satellite imagery, this contrast remains visible for more than 50 km downstream from their confluence. The Rio Branco originates in the Roraima-Rupunnuni savannas by the convergence of two rivers: the Tacutu and Uraricoera, and runs in a N-S direction for its entire course until it reaches the Rio Negro.
The geomorphological history of the Rio Branco is relatively well understood (Schaefer &
Dalrymple, 1995; Schaefer & Vale, 1997, Ferreira et al., 2007), and the presence of dated palaeochannels suggests that this river represents a recent Pleistocene capture of a previous protobasin that drained towards the Essequibo River less than 1Mya. By the end of the
Cretaceous, both the Tacutu and Uraricoera rivers flowed north towards the Berbice River and the Caribbean, creating a fluvial system called the Proto-Berbice, which ran in that direction until the Pleistocene (Schaefer & Dalrymple 1995). A recent change of direction is evident in the present geometry of several tributaries that flow in a NE direction, which change southwards when they meet the present course of the Rio Branco (Schaefer & Vale, 1997).
The Rio Branco can be divided into two main geographical and ecological regions, the upper and lower Rio Branco, with the city of Caracaraí (01o 50’ N) at the dividing point. About 70 km north of Caracaraí, coinciding with the confluence of the Rio Branco and the Rio Mucajaí, annual rainfall rates decrease sharply (Barbosa, 1997) with accompanying dramatic changes in
33 soil conditions (Brown & Prance, 1987). In this region savannas replace lowland terra firme
Amazonian forests, and gallery forests replace flooded forests along the rivers (Naka et al.,
2007). Unlike the Rio Negro, the headwaters of the Rio Branco are located within the Roraima-
Rupunnuni savannas (a barrier itself to forest-dwelling species). Therefore, to evaluate the area where isolated populations could potentially meet, I consider the ‘headwaters of the Rio Branco’ as the region where terra firme forests bridge Guianan populations and western Amazonian taxa.
This region is located in the Venezuelan state of Bolivar, north of the Roraima-Rupunnuni savannas and the Gran Sabana (Fig. 3.1). Biogeographically, the Rio Branco represents an important barrier to the terrestrial fauna, including birds (Naka et al., 2006) and monkeys (J. F.
Boubli, unpubl. data), although its importance has only recently been revealed (Naka et al.,
2006; 2010).
Selection of Population Pairs and Collection of Molecular Data
I created a list of all bird species or allospecies that show taxonomically designated phenotypic variation across the lower Rio Negro, and I then built a distributional dataset for those taxa based on museum specimens, audio recordings, and fieldwork (see chapter 4). These phenotypically distinct populations are presently considered either different subspecies or allospecies. Because data are often missing for whether these population pairs should be ranked as species or subspecies, I opted for treating all pairs equally, regardless of their current taxonomic rank. To explore levels of genetic divergence across rivers, I obtained tissues, both from museum collections and my own fieldwork in the region (Fig. 3.1). To assess the phylogenetic relatedness of pairs of taxa across the Branco and Negro rivers, I examined 20 published and unpublished phylogenies.
To obtain a comparable measurement of genetic divergence, I sequenced the mitochondrial protein-coding NADH dehydrogenase subunit 2 (ND2) gene for all samples. The ND2 gene has
34 been a useful marker for estimating levels of genetic divergence among populations of
Neotropical birds (Lovette, 2004; Weir & Schluter, 2004; Burney & Brumfield, 2009). Although using a single locus has inherent limitations, particularly because of the stochastic nature of gene coalescence (Edwards & Beerli, 2000; Arbogast et al., 2002; Brumfield et al., 2003), I opted for maximizing the taxonomic breath and number of samples of this study over collecting data from an increased number of loci that would provide greater precision in population genetic parameter estimation.
I extracted genomic DNA from ~25 mg of fresh tissue using Qiagen DNeasy Tissue Kits, following the manufacturer’s instructions (QIAGEN, Inc., Valencia, California). Following standard PCR protocols, I amplified the complete ND2 gene (1041 base pairs) using the following primers: L5215 for the light (Hackett, 1996) and H6313 for the heavy strand
(Sorenson et al., 1999). Internal primers (H5766 and L 5758) were used when needed to resolve ambiguous reads (Sorenson et al., 1999). PCR amplifications were obtained from 25 µL reactions, consisting of 2.5 µL of template DNA (~50 ng), 1 µL of each primer (10 mM), 1 µL of dNTPs (10 mM: 2.5 mM each dATP, dTTP, dCTP, dGTP), 2.5 µL of 10X reaction buffer with
MgCl2 (15 mM), 0.1 of Taq DNA polymerase (5 U/µL AmpliTaq, Applied Biosystems Inc.,
Foster City, California), and 17.9 µL sterile dH2O. Thermocycler protocols for PCR consisted of an initial denaturation of 2 min at 94°C followed by 35 cycles of 30 sec at 94°C, 30 sec at 50°C, and 2 min at 72°C, with a final extension of 5 min at 72°C. PCR products were purified using
20% polyethylene glycol (PEG), and cycle sequence reactions were completed using a Big Dye terminator Cycle Sequencing Kit, version 3.1 (ABI). Cycle sequence reactions were cleaned using Sephadex (G-50 fine) columns and electrophoresed on an ABI 3100 Genetic Analyzer.
Consensus sequences were compiled from both forward and reverse sequences, and complementary strands were aligned and edited using Sequencher, version 4.6 (GeneCodes
35 Corporation, Madison, Wisconsin). The entire length of each sequence was examined by eye to confirm base calls, and particular attention was paid to look for gaps, insertions or deletions in the data. To corroborate the absence of pseudogenes, I confirmed that all codons translated into amino acids and that no stop codons occurred within the open reading frame.
Role of Rivers as Current Barriers
I calculated levels of genetic divergence between and within 58 phenotypically differentiated populations of 21 bird families distributed across the Negro and Branco rivers. Most population pairs analyzed represent distinct subspecies (42), but over a fourth (16) are currently recognized as species pairs. For all 58 species pairs, the distributions of phenotypically distinct populations coincide with the lower Rio Negro. However, only 14 of these species pairs are isolated by the
Rio Negro along its entire length. Instead, the distributions of 28 of the 58 species pairs coincide with the Rio Branco, with the distributional termini of the remaining 16 pairs occurring somewhere in the Negro/Branco interfluvium (see chapters 2 and 4).
I analyzed sequence data for the complete ND2 gene from 851 individuals of 60 pairs of taxa
(Table 3.1). Mean levels of both pairwise corrected and uncorrected genetic distances were calculated using PAUP*4.0 (Swofford, 2001). Corrected pairwise distances were calculated using the optimal finite-sites substitution models of DNA evolution for each pair obtained using the Akaike Information Criteria (AIC) test implemented in MODELTEST 3.6 (Posada & Crandall,
1998) (see Appendix C). Average pairwise distances, standard deviations, means and medians were calculated directly from the distance matrices.
I used reciprocal monophyly of populations across rivers as an indication of no ongoing gene flow between populations. Reciprocally monophyletic populations do not share haplotypes, suggesting absence of gene flow, whereas lack of monophyly suggests present or past gene flow or incomplete lineage sorting, which is the persistence of ancestral alleles in current populations.
36
Fig. 3.1. Map of the study area indicating the localities from were I obtained tissue samples (white dots). The white circle indicates the what I considered the headwaters of the Rio Branco, and the black circle represents the headwaters of the Rio Negro. The yellow lines define the three main sections of the Rio negro, according to Franzinelli & Igreja (2002), and the white arrows point towards that particular section of the river.
37 Phylogenetic trees and bootstrap support for each phenotypically differentiated clade were reconstructed with Geneious (Drummond et al., 2010) using PHYML (Guindon & Gascue, 2003) on the corrected distance matrix. When the finite-sites model selected by MODELTEST was not available in PHYML, I chose the most similar one according to the MODELTEST hierarchy
(Posada & Crandall, 1998).
Population Interactions at the Headwaters
Because little genetic material was available from the Branco and Negro headwaters, I used distribution data based on museum specimens to infer whether pairs of phenotypically distinct populations occurred in allopatry, parapatry, or sympatry. I considered a species pair to be allopatric if the geographically closest specimen records were separated by a physical barrier
(such as a savanna, a river, or a mountain range, see chapters 2 and 4), or if no specimen records closer than 100 km were available. I arbitrarily considered as parapatric those phenotypically distinct populations that occurred within 100 km of each other without any obvious intervening physical barriers. Using these criteria, populations 10 km apart would be considered allopatric if separated by a river, or parapatric if no barrier was apparent. When both species occurred at the exact same locality, the pair was considered to be sympatric.
To evaluate whether introgression occurred between a pair of taxa, I examined museum specimens from the areas of contact. I also searched for published descriptions of hybrids in the literature (e.g., McCarthy, 2006; del Hoyo et al., 1991-2010). Museum specimens were intensively studied in both Venezuelan (COP and CBRG) and Brazilian (INPA, MPEG) collections (see chapter 2 for acronyms), particularly to assess possible areas of contact.
Although genetic data would provide a more direct assessment of hybridization in population pairs, a lack of genetic samples from the headwaters prevented such analyses using a comparative approach. Nonetheless, I considered non-reciprocal monophyly of the species pair
38 gene trees as a possible indication of gene flow. Because ancestral polymorphism cannot be ruled out in these cases, I only used lack of monophyly as an ancillary, supporting assessment of the plumage data. Fortunately, the many museum specimens collected along the Branco and
Negro headwaters in Venezuela were informative in documenting whether pairs come into contact and interbreed.
Levels of Genetic Differentiation Along the Rio Negro
I compared levels of genetic divergence across the lower, middle, and upper regions of the
Rio Negro in 37 pairs of species (8) or subspecies (29) that show taxonomically recognized phenotypic variation across the lower reaches of the river. For this comparison, I only selected those pairs for which I had samples from both across and along the river. To perform the test, I made pairwise comparisons of populations across the three sections of the Rio Negro (as defined by Franzinelli & Igreja, 2002, see Fig. 3.1). For each population pair, corrected pairwise genetic distances between individuals were averaged to obtain a single value across each portion of the river. Due to differential sampling across species and river sections, the number of individuals used for each comparison was different (see results). In cases where I lacked samples from one specific section of the river (i.e., the middle), I merged samples from the two other sections if I found no evidence of genetic structure between the two. I tested levels of pairwise genetic distance across the three sections of the river using a one-way ANOVA. I tested for heteroscedasticity using the Fligner-Killeen test of homogeneity of variances (Crawley, 2009) and square-root transformed the data to attain homogeneity. I repeated this analysis excluding those 19 pairs known to replace one another along the Rio Branco, and again including only the
14 pairs known to replace one another only along the Rio Negro. I used Tukey’s honest significant differences (Crawley, 2009) to make pairwise comparisons among the three river
39 sections. All statistical tests described above were performed using R (R Core Development
Team, 2008).
Testing for Simultaneous Divergence
To test for simultaneous divergence among the several pairs of taxa co-distributed across the
Branco and Negro rivers, I used an approximate Bayesian computation (ABC) approach, implemented in MTML MSBAYES (Hickerson et al. in prep), which works under a hierarchical coalescent model using multiple co-distributed population pairs. An advantage of this method is that instead of conducting independent analyses on every population pair and testing the hypothesis of temporal concordance, it tests simultaneous divergence using all the data in a single analysis. It does this by estimating a set of three hyper-parameters (!) that characterize and quantify the degree of variability in divergence times across populations, while allowing for variation in various within population-pair demographic parameters (sub-parameters) that can affect the coalescent (Hickerson et al., 2007). These hyper-parameters are: i) ", the number of possible divergence times; ii) #($), mean divergence time; and iii) %, the ratio of variance to the mean across all population pairs, which is Var ($)/#($). The subparameters of each pair (&) are allowed to vary independently across all populations, and include parameters such as divergence times, current and ancestral population sizes, post-divergence founding population sizes, recombination rates, and post-divergence migration rates. Rather than directly calculating likelihood expressions, this method uses simulation-based summary statistics that approach an approximate likelihood, which can be sampled from the posterior probability. The choice of summary statistics is vital for the success of the method, and they should show a strong correspondence with parameter values (Hickerson et al., 2006b). Extensive simulations conducted by Hickerson et al., (2006a) suggest the use of four summary statistics that can be summarized in a summary statistic vector (D), and these are the ones calculated by default by
40 MTML MSBAYES. These summary statistics are: i) !, average number of pairwise differences among all sequences within each population pair; ii) "w, number of segregating sites within each population, normalized for sample size; iii) Var (!="w) in each pair; iv) !net, Nei and Li’s net nucleotide divergence between each pair of populations. This last summary statistic is actually represented by the difference between the average pairwise difference between (!b) and within each population pair (!w). The implementation of these analyses follows three basic steps, which include: i) calculating the observed summary statistic; ii) running coalescent simulations of the DNA sequence data, using parameters drawn from the hyper-prior (#) and prior ($); and iii) sampling from the posterior distributions and obtaining posterior estimates of the three hyper- parameters (%, &('), and () across all pairs.
I tested for the simultaneous divergence of 24 population pairs across the Rio Branco and 14 populations across the Rio Negro. For the Rio Negro, I only selected those populations that are separated by the entire extension of the river. For both rivers, I selected those pairs for which I had at least three samples from each bank, although ABC simulations perform well with even smaller sample sizes (Hickerson et al., 2007). I used the following prior parameters for both simulations: i) bounds of " per site: upper"= 0.1, lower "= 4e-8; ii) upper limit of ' (time of divergence) = 1.0; iii) the number of ' classes (%, possible times of divergence) was set to 0, which means that there could be as many differentiation events as pairs of populations in the comparison; iv) migration rate=0 (which disables migration); and upper ancestral population sizes were set to 0.025, and v) subparameters unconstrained (0). These prior parameters have worked well in simulations and are expected to perform well under a wide number of cases
(Hickerson, pers. comm). I ran three independent simulations of 1,000,000 replicates for each analysis (which is the number of random draws from the hyper-prior) and set the tolerance to
0.0005, which is the proportion of accepted draws from the prior. To explore whether the
41 tolerance value would affect the results, I also ran analyses with this value set to 0.001 (which provides 1,000 posterior values). Considering 1,000,000 replicates, this tolerance value provides the 500 closest matches between the summary statistic calculated from the observed and simulated data in order to sample from the posterior distribution. Given that these simulations were very similar, I concatenated the replicates and the final results presented are the result of
3,000,000 simulations, and a tolerance of 0.00016 (which still represent ~500 posterior values).
RESULTS
Role of Rivers as Present Day Barriers
In general, phenotypic divisions observed in plumage characters across the Branco and
Negro rivers were concordant with the location of phylogeographic breaks. Pairs of taxa showed phenotypic differentiation across only one of the two rivers, and the genetic structure of the taxa matched this pattern. In 80% of the cases (48 pairs) the mean level of genetic divergence between pairs across rivers was at least one order of magnitude higher than within populations from the same bank of the river (Table 3.1), and in all cases, mean levels of genetic divergence between phenotypically differentiated populations were higher than within similar-looking phenotypes (Table 3.2). Because mean values of corrected and p-uncorrected genetic distances were highly correlated (Pearson's product-moment correlation: 0.966; t=29.5792, df = 62, p- value < 2.2e-16), I only present corrected values. Levels of genetic divergence were highly variable among phenotypically differentiated pairs (Table 3.2). In ~85% of the pairs (49 out of
57), phenotypically distinct populations represented reciprocally monophyletic clades (Table
3.2). In addition, no shared haplotypes were observed, suggesting little ongoing gene flow between populations across rivers.
42 Table. 3.1. Mean percentage of pairwise genetic divergence (corrected) across rivers in 60 pairs of closely related taxa, divided by river (Rio Branco and Rio Negro), and compared to those not divided by rivers (interfluvium). Location of phenotypic N Genetic divergence divergence range Mean (std) median between same-bank populations (east) (west) Lower Rio Negro 60 0.68 - 13.34 5.74 (±3.60) 5.55 0.34 0.48 Rio Branco 28 0.76 – 13.34 6.25 (±3.46) 6.48 0.35 0.54 Rio Negro 16 1.26 – 11.34 4.86 (±3.61) 2.66 0.28 0.43 Interfluvium 16 0.68 - 12.93 5.72 (±3.87) 6.05 0.38 0.42
Phylogenetic Relationships of Replacing Pairs of Taxa
Available phylogenies informed the evolutionary relationships of 20 species pairs. Pairs of taxa are each other’s closest relatives only in approximately half of the cases (Table 3.3).
Although many of these pairs are not sister taxa, it remains clear they replace one another geographically and ecologically.
Population Interactions at the Headwaters Most pairs of taxa are parapatric (25 pairs) or allopatric (24 pairs) at the headwaters, with at least seven pairs occurring in sympatry (Table 3.2). I found intermediate-looking individuals in at least 13 pairs of taxa (see Appendix D, for examples), suggesting that at least limited hybridization is occurring between some pairs of taxa at the headwaters (Table 3.4). Seven of the 13 pairs are not reciprocally monophyletic. That the samples with incompletely sorted genes are those closest to the headwaters provides evidence that hybridization and not incomplete lineage sorting accounts for the lack of reciprocal monophyly. Interestingly, 10 of 13 putative cases of hybridization at the headwaters are limited to three avian families: Ramphastidae
(toucans), Picidae (woodpeckers), and Thamnophilidae (antbirds).
Levels of Genetic Differentiation Along the Rio Negro An analysis of 37 pairs of taxa showed that corrected pairwise genetic distances between populations on opposite margins of the Rio Negro diminish towards the headwaters (Table 3.5).
The data required two square-root transformations to attain homogeneous variances (Fligner-
43 Killeen test after transformation: chi-squared = 2.83, df = 2, p-value = 0.24). Mean corrected genetic distances between the pairs studied across the lower, middle, and upper Rio Negro were
5.83%, 2.71%, and 1.59%, respectively (Fig. 3.2a). A multiple comparisons of means (Table
3.6) showed that mean levels of genetic divergence across rivers were significantly higher along the lower Rio Negro, but that there were no significant differences between the middle and upper
Rio Negro.
To investigate whether the significant results were because the narrow upper Rio Negro was easier to cross, or whether the difference between the lower and upper portions could be attributed to the presence of a second barrier, such as the Rio Branco, I first excluded 18 taxa that show phenotypic differences across the lower Rio Negro and do not extend west of the Rio
Branco. When these taxa were excluded, the data did not require any transformation (Fligner-
Killeen test: chi-squared = 2.35, df = 2, p-value = 0.30), and there were no significant differences, although it remained marginally non-significant (Table 3.7, Fig. 3.2b). Furthermore, when I included only pairs divided by both the lower and upper Rio Negro (no transformations needed; Fligner-Killeen test: chi-squared = 0.09, df = 2, p-value = 0.95), I found no signs of genetic homogenization towards the headwaters (i.e., no significant variation in mean levels of genetic divergence along the river (Table 3.8, Fig. 3.2c).
Testing for Simultaneous Divergence
I conducted two independent analyses to test for simultaneous divergence among pairs of co- distributed taxa across the Rio Branco and Rio Negro. Pairs of taxa showed variable mtDNA p- uncorrected genetic distances, which ranged from ~1 % to 7.8 % across the Negro (Table 3.9a), and between 1.4 % and 9.8 % across the Branco (Table 3.9b). Hyper-posterior estimates suggest that the rivers have two very different histories. Whereas estimates of ! (the number of possible divergence times) across the Rio Negro could be consistent with either one or two diversification
44 Table 3.2. List of 60 pairs of taxa with taxonomically recognized populations across the lower Rio Negro for which I collected molecular data. Distr. Sample Corrected genetic Rec. Boot. support Avian Family Pairs of taxa (E/W) Patt.1 size distance (SD)2 Corr. gen. dist.2 monoph? (100 reps) Head.4 E W between groups E Pop. W Pop. E clade W clade Tinamidae Tinamus m. major/zuliensis B 3 2 0.18 (0.07) 0.00 0.00 no ? Psittacidae Pyrilia caica/P. barrabandi B 7 5 7.83 (0.57) 1.09 0.60 yes 100 80 S Trochilidae Topaza pella/T. pyra BN 3 7 7.63 (0.16) 0.98 0.06 yes 85 99 A Phaethornis s. superciliosus/insolitus N 12 10 10.82 (0.27)Yesyes yesddddd0.17 0.02 yes 78 100 P Thalurania f. furcata/nigrofasciata BN 7 7 5.74 (0.13) 0.18 0.16 yes 99 100 S Trogonidae Trogon r. rufus/sulphureus BN 2 5 1.89 (0.10) 0.00 0.43 yes 100 100 A Momotidae Momotus m. momota/microstephanus N 9 4 2.33 (0.20) 1.01 0.20 yes 71 79 P Galbulidae Galbula a. albirostris/chalcocephala B 7 14 7.67 (0.37) 0.26 0.67 yes 100 82 P Galbula d. dea/brunneiceps B 8 4 8.17 (0.31) 0.26 0.14 yes 99 100 P Bucconidae Monasa atra/M. morphoeus N 3 4 8.45 (0.19) 0.13 0.85 yes 96 95 S Capitonidae Capito niger/C. auratus B 3 3 7.65 (0.23) 0.65 3.44 yes 94 100 A Picidae Veniliornis cassini/V. affinis BN 6 2 8.50 (0.07) 0.23 0.10 yes 98 100 P Piculus f. flavigula/magnus N 9 7 1.97 (0.09) 0.06 0.27 yes 100 55 P Celeus u. undatus/C. grammicus B 2 10 1.95 (0.11) 0.08 0.17 no S Celeus e. elegans/jumanus B 2 10 0.77 (0.16) 0.10 0.36 no P Celeus t. torquatus/occidentalis BN 7 2 1.12 (0.16) 0.59 0.10 yes 72 69 P Furnariidae Synallaxis rutilans dissors/confinis N 10 6 1.67 (0.10) 0.18 0.26 no ? Automolus infuscatus cervicalis/badius B 4 10 3.65 (0.12) 0.26 0.64 yes 97 100 P Xenops minutus ruficaudus/remoratus N 26 5 6.344 (0.11) 0.18 0.04 yes 100 100 P Dendrocolaptidae Dendrocincla f. fuliginosa/phaeochroa B 7 16 6.05 (0.17) 0.11 0.16 yes 100 97 S Dendrocincla m. merula/bartletti B 1 10 4.37 (0.07) 0.00 0.10 yes 100 100 A Sittasomus griseicapillus axillaris/amazonus BN 6 5 1.63 (0.17) 0.37 0.82 no A Dendrocolaptidae Glyphorynchus s. spirurus/rufigularis B 6 20 8.53 (0.23) 0.77 1.91 yes 100 90 P Dendrocolaptes c. certhia/radiolatus N 9 6 2.14 (0.35) 0.06 0.63 yes 99 87 A Xiphorhynchus pardalotus/X. ocellatus N 9 4 5.26 (0.18) 0.25 0.45 yes 99 100 A Dendrocolaptidae Xiphorhynchus guttatus polystictus/ guttatoides BN 12 7 6.36 (0.40) 0.63 0.42 yes 100 100 P Lepidocolaptes a. albolineatus/duidae B 5 6 6.90 (0.78) 0.16 0.84 yes 100 96 A Thamnophilidae Cymbilaimus l. lineatus/intermedius B 5 10 2.73 (0.16) 0.16 0.53 yes 100 88 A Thamnophilus m. murinus/canipennis N 11 10 1.95 (0.12) 0.12 0.09 yes 100 93 ? Thamnophilus amazonicus divaricatus/ cinereiceps B 4 15 4.291 (0.17) 0.57 0.10 yes 95 100 A Epinecrophylla gutturalis/E. haematonota B 4 6 7.36 (0.13) 0.26 0.12 yes 100 100 P Myrmotherula guttata/M. hauxwelli N 2 2 9.08 (1.04) 0.19 2.17 yes 100 100 A Hypocnemis cantator /H. flavescens B 12 20 9.04 (0.73) 0.81 0.16 yes 100 100 S Cercomacra cinerascens immaculata/ cinerascens B 7 10 1.95 (0.27) 0.06 0.14 yes 93 86 A 45 (Table 3.2 con’d.) Distr. Sample Corrected genetic Rec. Boot. support Avian Family Pairs of taxa (E/W) Patt.1 size distance (SD)2 Corr. gen. dist.2 monoph? (100 reps) Head.4 E W between groups E Pop. W Pop. E clade W clade Thamnophilidae Cercomacra tyrannina saturatior/tyrannina BN 17 5 1.88 (0.10) 0.08 0.16 yes 100 95 P Percnostola r. rufifrons/minor B 9 7 5.35 (0.15) 0.24 0.13 yes 96 100 A Schistocichla l. leucostigma/infuscata B 3 2 1.9 0.12 0.09 yes 99 100 A Pithys a. albifrons/peruvianus N 16 14 1.26 (0.17) 0.24 0.32 no P Gymnopithys rufigula/G. leucaspis N 13 9 2.29 (0.58) 0.48 0.58 no P Willisornis p. poecilinotus/duidae BN 10 23 10.82 (0.76) 0.77 0.49 no P Grallariidae Myrmothera c. campanisona/dissors B 3 4 2.84 (0.13) 0.19 0.24 yes 90 99 A Tyrannidae Zimmerius acer/Z. gracilipes BN 8 12 12.93 (0.47) 0.38 0.63 yes 63 93 P Tyrannidae Hemitriccus zosterops rothschildi/zosterops B 7 4 1.23 (0.43) 0.85 0.40 yes 68 100 A Tolmomyias assimilis examinatus/ examinatus B 9 5 5.31 (0.22) 0.29 0.08 yes 100 100 A Tolmomyias p. klagesi/ poliocephalus N 2 6 2.98 (0.13) 0.19 0.17 yes 99 99 A Platyrinchus coronatus gumia/coronatus BN 4 4 8.81 (0.27) 0.38 0.15 yes 87 100 A Onychorhynchus c. coronatus/castelnaui N 3 5 8.18 (0.17) 0.15 0.58 yes 100 100 A Terenotriccus e. erythrurus/venezuelensis BN 3 5 8.15 (0.87) 0.19 2.08 yes 91 100 P Pipridae Tyranneutes virescens/T. stolzmanni B 5 5 11.38 (0.46) 0.25 1.36 yes 100 96 P Pipridae Chiroxiphia pareola pareola/regina BN 7 3 5.04 (0.14) 0.26 0.06 yes 100 100 A Tityridae Schiffornis turdina olivacea /amazonus B 14 12 9.69 (0.18) 0.12 0.20 yes 100 100 A Iodopleura fusca/I. isabellae BN 3 4 8.753 (0.22) 0.14 0.15 yes 99 100 A Incertae sedis Piprites chloris chlorion/tschudii N 4 2 1.65 (0.20) 0.19 0.19 no S Vireonidae Hylophilus m. muscicapinus/H. hypoxanthus N 6 3 11.34 (0.71) 0.83 0.13 yes 73 100 P Hylophilus ochraceiceps luteifrons/ ferrugineifrons B 7 13 13.34 (0.82) 0.51 0.57 yes 100 98 P Troglodytidae Microcerculus b. bambla/albigularis B 3 7 10.13 (0.47) 0.06 0.51 yes 100 100 A Pheugopedius c. coraya/griseipectus B 7 13 12.89 (0.68) 1.44 0.98 yes 100 100 A Thraupidae Tachyphonus c. cristatus/cristatellus ? 2 6 1.18 (1.79) 0.29 1.85 no P Tachyphonus s. surinamus/brevipes BN 6 9 1.52 (0.13) 0.31 0.29 yes 75 64 P Tangara chilensis paradisea/chilensis? N 4 2 0.409 (0.14) 0.16 0.68 no P Cyanerpes c. caeruleus/microrhynchus BN 4 5 0.684 (0.33) 0.68 0.65 no P Hemithraupis f. flavicollis/aurigularis B 1 2 5.006 (0.07) 0.00 0.10 yes 100 98 A Fringillidae Euphonia cayennensis/E. rufiventris B 2 3 6.94 (0.01) 0.00 0.39 yes 100 100 S 1 Distribution pattern: All pairs have distinct taxa on opposite banks of the lower Rio Negro, but are distinguished by the location of their phenotypic break further upstream. B, pairs of taxa that replace each other across the Rio Branco; N, pairs of taxa that replace each other across the entire length of the Rio Negro; BN, pairs of taxa with the phenotypic break in the Rio Negro/Rio Branco interfluvium. 2 Corrected pairwise genetic distance. 3 Refers to whether mitochondrial gene trees are reciprocally monophyletic with respect to phenotype. 4 Distribution patterns of each pair of taxa at the headwaters. A, allopatric; P, parapatric; S, sympatric. 46 events (Table 3.10a; Fig. 3.3a), estimates for the Rio Branco strongly suggest a history consistent with at least five independent vicariant events (Table 3.10b; Fig. 3.3b).
The mean time of divergence (E(t)) has its mode around 0.17 for the Rio Negro, and around
0.41 for the Rio Branco dataset (Table 3.10; Fig. 3.4), suggesting that the taxa divide across the
Rio Branco are twice as old than those divided by the Rio Negro. On the other hand, exactly 100 accepted posterior values of ! for the Rio Negro were equal to zero, which suggests that the time of divergence across pairs is similar. None of the distributions of the priors accepted for the Rio Branco included values of zero for !.
I obtained similar results in at least three independent 1,000,000 draws from the priors
(Appendices E, F), and results remained largely unchanged using a higher number of posterior draws (tolerance value), although the mode of the number of possible divergence times almost doubled for the Rio Branco dataset (Table 3.10).
Table 3.3. Phylogenetic relationships of 20 taxon pairs that replace each other across rivers in the Guiana Shield. The location of phenotypic break was inferred from the distributional dataset presented here (B: Rio Branco; N: Rio Negro; Low N: only the lower portion of the Rio Negro). Sister taxa are those that share a common ancestor and no additional descendents. Location of Pair of taxa phenotypic break Sister taxa? Reference Pyrilia caica/P. barrabandi B No Ribas et al. (2005) Phaethornis s. superciliosus/insolitus N No LNN & M. Miller (unpubl.) Capito niger/C. auratus B Unresolved Armenta et al., (1995) Veniliornis cassini/V. affinis Low N No Moore et al., (2006) Xenops minutus ruficaudus/remoratus N Yes Burney (2009) Dendrocincla m. merula/bartletti B No Henriques (unpubl.) Glyphorynchus s. spirurus/rufigularis B No Marks et al., (2002) Xiphorhynchus pardalotus/X. ocellatus N Yes Aleixo (2004) Xiphorhynchus guttatus polystictus/guttatoides Low N No Aleixo (2002) Lepidocolaptes a. albolineatus/duidae B No Rodrigues (2008) Epinecrophylla gutturalis/E. haematonota B Yes Bravo et al. (unpubl.) Myrmotherula guttata/M. hauxwelli N Yes Bravo et al. (unpubl.) Hypocnemis cantator/H. flavescens B No Tobias et al., (2008) Percnostola r. rufifrons/minor B No LNN and Bravo (unpubl.) Pithys a. albifrons/ peruvianus N Yes LNN (unpubl.) Gymnopithys rufigula/G. leucaspis N Yes Brumfield et al., (2007) Willisornis p. poecilinotus/duidae Low N No Bates (unpubl.) Zimmerius acer/Z. gracilipes Low N No Rheindt et al., (2008) Hemitriccus zosterops rothschildi/zosterops B Yes Cohn-Haft (2000) Schiffornis turdina olivacea/amazonus B No Nyári (2007)
47 Table 3.4. List of 13 taxon pairs with intermediate-looking birds known from collections or observations along the Branco/Negro interfluvium (See Appendix D). All localities are from Venezuela, unless otherwise specified. Species Family Locality Reference1 Reciprocal monoph? Pteroglossus aracari x pluricinctus Ramphastidae Río Carapo, Bol. COP 76590 No mol. data Ramphastos v. vitellinus x Ramphastidae Branco/Negro interfluv. Haffer (1974) No mol. data culminatus Ramphastos t. tucanus x cuvieri Ramphastidae Branco/Negro interfluv. Haffer (1974) No mol. data Piculus f. flavigula x magnus Picidae Yavita-Pimichin, Am. COP 34338 Yes Celeus undatus x C. grammicus Picidae Manaus, AM (Brazil) Pers. obs. No Celeus e. elegans x jumanus Picidae Auyan Tepui, Bol. Short (1972) No COP 7935 C. t. torquatus x occidentalis Picidae Chimanta-Tepui, Bol. COP 35656 Yes Xiphorhynchus g. polystictus x Dendrocolaptidae Río Orinoco, Am. Marantz, Yes guttatoides (unpubl.) P. a. albifrons x peruvianus Thamnophilidae La Ceiba, Colombia ICN 32218 No Gymnopithys rufigula x G. Thamnophilidae Yavita-Pimichin, Am. COP 34563 No leucaspis Willisornis p poecilinotus x duidae Thamnophilidae Mt. Duida, Am. Zimmer (1934) No Piprites chloris chlorion x tschudii Incertae sedis Yavita-Pimichin, Am. COP 34608 No Cyanerpes c. caeruleus x Thraupidae Branco/Negro interfluv. COP No microrhynchus 1COP: Colección Ornitológica Phelps; ICN: Instituto de Ciencias Naturales, Universidad Central, Colombia
Fig. 3.2. Mean values of corrected pairwise genetic divergence along three sections of the Rio Negro (see Fig. 3.1). Panel a represents all 37 species or pairs of species for which I had genetic samples along the Rio Negro. Panel b represents those same taxa, excluding those pairs that are known to replace across the Rio Branco. Panel c represents those same taxa as panel b, excluding those pairs that are known to replace within th Branco/Negro interfluvium. Panel c, is composed only by pairs that replace across the Rio Negro, and therefore are more suitable for comparisons.
48 Table 3.5. Mean pairwise genetic distance of 37 species or species pairs across different sections of the Rio Negro. Note that all pairs have phenotypically differentiated populations across the lower Rio Negro.
Location of Sample size2 phenotypic E of Rio W of Rio Corrected levels of mean Taxa break1 Negro Negro genetic divergence (*100)1 L M U L M U L M U Phaethornis superciliosus N 4 7 4 1 3 10.80 10.32 7.92 Momotus momota N 3 2 2 2 2.41 2.56 2.58 Galbula albirostris BN 6 4 1 4 2 9.04 2.77 0.29 Galbula dea B 7 1 1 1 1 8.65 0.12 0.09 Monasa atra/M. morphoeus N 2 1 2 1 1 8.67 7.60 1.42 Piculus flavigula N 4 5 1 2 1 2 1.95 1.88 0.83 Celeus undatus/C. grammicus B 3 1 2 3 1 0.27 0.000 0.05 Celeus elegans B 2 7 2 2 1 1.37 0.17 0.19 Automolus infuscatus B 4 1 3 2 4 3.64 1.82 0.50 Xenops minutus N 22 2 2 2 1 6.36 6.42 6.42 Dendrocincla fuliginosa B 8 6 3 3 5.34 0.24 0.24 Glyphorynchus spirurus B 3 7 5 2 1 2 8.65 1.53 0.75 Dendrocolaptes certhia N 2 6 1 1 0 2.98 2.87 2.88 Xiphorhynchus pardalotus/ X. ocellatus N 3 3 2 1 1 5.37 5.18 5.08 Xiphorhynchus guttatus BN 2 6 1 1 1 2 6.47 6.37 0.54 Lepidocolaptes albolineatus B 5 1 2 3 7.47 0.90 0.90 Cymbilaimus lineatus B 5 1 1 2 1 3.11 0.50 0.50 Thamnophilus murinus N 3 7 1 5 2 3 1.94 1.99 1.91 Thamnophilus amazonicus B 4 6 2 5 2 4.19 0.02 0.20 Epinecrophylla gutturalis/ E. haematonota B 4 3 1 1 1 7.47 0.09 0.06 Myrmotherula axillaris BN 14 6 3 2 1.72 0.94 0.94 Hypocnemis cantator/ 1 H. flavescens B 12 6 2 1 9.42 0.72 0.71 Cercomacra cinerascens B 7 2 1 2 2 2.09 0.14 0.05 Cercomacra tyrannina BN 10 7 1 3 1 1.72 1.56 0.00 Pithys albifrons N 7 3 3 5 1 7 1.29 1.30 1.24 Gymnopithys rufigula/ G. leucaspis N 7 3 3 2 3 2.42 2.14 1.34 Willisornis poecilinotus BN 8 7 3 2 7 6 10.13 3.15 0.86 Zimmerius acer/ Z. gracilipes BN 6 3 2 3 4 12.81 8.88 0.56 Tolmomyias assimilis B 9 3 1 1 5.39 0.13 0.09 Tolmomyias poliocephalus N 2 4 1 2 3.07 2.97 2.87 Schiffornis turdina B 8 2 2 3 2 9.67 9.58 0.19 Hylophilus muscicapinus/ H. hypoxanthus N 5 1 1 1 2 11.33 10.96 12.82 Hylophilus ochraceiceps B 7 7 3 2 13.29 0.26 0.26 Microcerculus bambla B 3 2 1 1 2 9.68 0.58 0.90 Pheugopedius coraya B 7 7 2 3 1 13.04 0.76 0.64 Microbates collaris BN 2 3 1 2 4 3.10 1.60 1.07 Tangara chilensis N 1 1 2 2 0.43 0.33 0.43 1 Location of replacement of phenotypically distinct taxa. 2 Location of samples: L: lower, M: middle, U: upper Rio Negro (see Fig. 3.1).
49 Table 3.6. Results of one-way ANOVA (pairwise genetic distances and section of the river) and Tukey HSD test (multiple comparisons of means) of three river sections (95% family-wise confidence level) for 37 bird species or species pairs along the Rio Negro. Source Sum of squares df Mean square F ratio P value River section 5.56 2 2.84 20.29 3.3e-08 2 Error 15.12 108 S =0.14
River section Difference lower upper P adj Low-mid -0.385 -0.592 -0.178 6.7e-5 Low-upp -0.537 -0.744 -0.330 0.000 Mid-upp -0.152 -0.359 0.054 0.191
Table 3.7. Results of one-way ANOVA (pairwise genetic distances and section of the river) and Tukey HSD test (95% family-wise confidence level) for 19 species or species pairs along the Rio Negro (18 taxa divided by the Rio Branco were excluded). Source Sum of squares df Mean square F ratio P value River section 68.04 2 34.02 2.82 0.067 2 Error 686.90 57 S =12.05
River section Difference lower upper P adj Low-mid -1.113 -3.755 1.528 0.571 Low-upp -2.599 -5.241 0.042 0.055 Mid-upp -1.486 -4.128 1.155 0.372
Table 3.8. Results of one-way ANOVA (pairwise genetic distances and section of the river) for 14 species or species pairs with phenotypic differentiation across the Rio Negro. No Tukey multiple comparisons of means are presented because results of ANOVA show no significant difference. Source Sum of squares df Mean square F ratio P value River section 5.11 2 2.55 0.20 0.82 2 Error 459.04 36 S =12.75
DISCUSSION
The role of Amazonian rivers as promoters of biodiversity has been in the center of the discussion for decades (Sick, 1967; Hershkovitz, 1977; Haffer, 1969, 1974, 1992, 2002;
Capparella, 1988, 1991; Ayres & Clutton-Brock, 1992; Lougheed et al., 1999; Gascon et al.,
2000). However, surprisingly few empirical data are available to support or refute their role as mechanisms responsible for maintaining current levels of biodiversity (but see Bates et al.,
2004). This study represents the first comparative, multi-species approach that involves a significant portion of an Amazonian avian community to assess the roles of rivers in both promoting and preserving biodiversity. 50 Table 3.9. List of 38 pairs of taxa used for testing simultaneous divergence across the Negro and Branco rivers, including number of samples per population, number of base pairs of the ND2 gene, and mean pairwise p-uncorrected mtDNA distance.
(a) Rio Negro (n=14 pairs) Species pair Sample size Bp(n) P-uncorr. mean E pop. W pop. mtDNA div. (%) Phaethornis s. superciliosus/insolitus 12 11 967 7.83 Momotus m. momota/microstephanus 8 4 1033 2.08 Piculus f. flavigula/magnus 9 7 1039 1.90 Synallaxis rutilans dissors/confinis 10 6 1041 1.63 Xenops minutus ruficaudus/remoratus 26 5 1035 6.56 Xiphorhynchus pardalotus/X. ocellatus 10 4 1036 3.62 Thamnophilus m. murinus/canipennis 10 8 1021 1.78 Thamnomanes caesius E/W* 8 8 1025 2.00 Myrmotherula brachyura E/W* 8 4 1033 1.99 Pithys a. albifrons/peruvianus 16 14 1026 1.14 Gymnopithys rufigula/G. leucaspis 13 9 1016 2.01 Onychorhynchus c. coronatus/castelnaui 3 5 856 7.67 Piprites chloris chlorion/tschudii 4 4 1040 1.53 Hylophilus muscicapinus/H. hypoxanthus 6 3 1040 7.60 b) Rio Branco (n=24 pairs) Species pair Sample size Bp(n) P-uncorr. mean E pop. W pop. mtDNA div. (%) Galbula a. albirostris/chalcocephala 7 14 1037 5.76 Galbula d. dea/brunneiceps 8 3 1041 7.42 Capito niger/C. auratus 3 3 1031 7.06 Automolus infuscatus cervicalis/badius 4 10 1041 3.51 Dendrocincla f. fuliginosa/phaeochroa 7 16 1008 5.32 Glyphorynchus s. spirurus/rufigularis 8 15 1028 6.96 Lepidocolaptes a. albolineatus/duidae 5 6 1009 4.98 Cymbilaimus l. lineatus/intermedius 5 9 1040 2.13 Thamnophilus amazonicus divaricatus/cinereiceps 4 14 1028 3.58 Epinecrophylla gutturalis/E. haematonota 4 6 1033 6.82 Herpsilochmus dorsimaculatus E/ W* 6 6 1037 2.23 Hypocnemis cantator/H. flavescens 11 20 1002 5.54 Cercomacra c. immaculata/cinerascens 5 10 1039 1.91 Percnostola r. rufifrons /minor 9 7 1023 5.07 Formicarius colma E/W* 8 8 1041 2.47 Tolmomyias assimilis examinatus/neglectus 9 5 1032 5.01 Tyranneutes virescens/T. stolzmanni 5 5 1039 7.20 Schiffornis turdina olivacea/amazonus 14 12 1018 8.83 Hylophilus ochraceiceps luteifrons/ferrugineifrons 7 12 1039 9.62 Microcerculus b. bambla/albigularis 3 7 1021 8.40 Pheugopedius c. coraya/griseipectus 7 12 1025 8.57 Euphonia cayennensis/E. rufiventris 3 3 1040 6.43 * Population pairs with genetic structure across rivers, but no clear phenotypic variation.
51 Table 3.10. Hyper-posterior estimates and 95% confidence intervals across two rivers for two tolerance values (0.0005 and 0.001 that represent the acceptance of 500 and 1000 posterior values). a) Rio Negro Hyper- Estimate Quantiles (%) Estimate (tol=0.001) Quantiles (%) parameters (tol=0.0005) mode mean median 0.025 0.097 mode mea median 0.025 0.0975 5 n ! 2 3.20 3.0 1 10 1 3.08 2 1 10 " 0.006 0.15 0.13 0 0.49 0.003 0.13 0.10 0 0.46 E(t) 0.17 0.15 0.15 0.04 0.28 0.147 0.14 0.14 0.02 0.28 Var.t 8e-4 0.03 0.02 0 0.10 6e-4 0.02 0.01 0 0.10
! b) Rio Branco Estimate Quantiles (%) Estimate Quantiles (%) Hyper- (tol=0.0005) (tol=0.001) parameters mode mean median 0.025 0.0975 mode mean median 0.025 0.0975
! 5 11.46 11 2 22 9 11.63 11 2 22 " 0.17 0.16 0.16 0.05 0.26 0.16 0.17 0.17 0.05 0.27 E(t) 0.41 0.39 0.39 0.28 0.50 0.40 0.39 0.40 0.28 0.50 Var.t 0.065 0.06 0.06 0.02 0.11 0.06 0.07 0.07 0.01 0.11
! Rivers as Current Biogeographical Barriers
Both rivers under examination are significant biogeographical barriers and should be rated
among the top five most important Amazonian riverine barriers for birds, along with the Amazon
itself, the Madeira, and the Tapajós rivers (see Haffer, 1987). The Rio Negro has been known as
a barrier to animals for more than 150 years (Wallace, 1852), yet this study is one of the first to
show that the patterns observed by the British explorer, based on phenotypic characters, are
supported by genetic data. Capparella’s (1988) pioneering study along the Rio Napo linked
phenotypic and genotypic variation across rivers for the first time, but his dataset consisted of
only five species of understory birds and was too limited for generalizations. Here, I presented
similar results using genetic data on ~60 bird species or species pairs from 24 different avian
families.
Phenotypic data presented in the two previous chapters show that more than 90 pairs of taxa
replace one another along the lower Rio Negro, making it the most important biogeographical
52 barrier in the Guiana Shield. The Rio Branco, on the other hand, has only recently been proposed as an important biogeographical barrier (Naka et al., 2006, 2010), and the genetic data here presented strongly support previous observations by confirming the importance of this river as a current barrier to gene flow. Levels of genetic divergence across rivers (and between phenotypically distinct populations) are much higher than those within similar-looking populations across interfluvia. Thus, large areas lacking genetic structure are abruptly broken by the presence of a river. This has been previously observed in many species-level phylogenies
(Cohn-Haft 2000; Aleixo, 2002; Marks et al., 2002; Fernandes, 2007; Nyári 2007) and seems to be the general pattern in Amazonia. Most population pairs that show phenotypic differentiation are also reciprocally monophyletic, further supporting the role of rivers as current and historical barriers to gene flow. Mean levels of corrected genetic distances between pairs of taxa across phenotypically distinct populations) are much higher than those within similar-looking populations across interfluvia. Thus, large areas lacking genetic structure are abruptly broken by the presence of a river. This has been previously observed in many species-level phylogenies
(Cohn-Haft 2000; Aleixo, 2002; Marks et al., 2002; Fernandes, 2007; Nyári 2007) and seems to be the general pattern in Amazonia. Most population pairs that show phenotypic differentiation are also reciprocally monophyletic, further supporting the role of rivers as current and historical barriers to gene flow. Mean levels of corrected genetic distances between pairs of taxa across rivers are highly variable (ranging from virtually 0 to ~13%), and no patterns are apparent from the data. I cannot discount, however, that this might simply be the consequence of the low resolution of single mitochondrial markers in estimating divergence time (Arbogast et al., 2002).
Levels of Gene Flow at the Headwaters
The main criticism of the ‘riverine barrier hypothesis’ might be that full speciation and reproductive isolation could not be attained if populations are not completely isolated, and that
53 populations divided by the wide sections of Amazonian rivers come into contact at the headwaters (Haffer, 1978; 1987; 1993; 1997a; 2001). Gene flow, the transfer of alleles from one population to another, can effectively homogenize populations and potentially delay, or even prevent, the diversification process (Endler, 1977). However, the degree of gene flow needed for preventing speciation is still uncertain (Porter & Johnson, 2002). In recent years, several examples of speciation with gene flow have emerged (Niemiller et al., 2008), and such examples may be more common than previously thought (Nosil, 2008). Despite the possibility of speciation with gene flow, extensive gene flow clearly can affect the speciation process in
Fig. 3.3. Posterior probability densities for ! (number of possible divergence times) given Y pairs of taxa. Estimates are based on 3,000,000 simulations (draws from the hyperprior) and 500 draws from the joint posterior.
54
Fig. 3.4. Joint posterior probability densities for E(t) and !, using the approximate Bayesian computation estimator in MTML msBayes. Tolerance values were set to 0.000166 in order to obtain ~500 posterior values from the 3,000,000 draws from the priors. several ways. Therefore, it remains important to determine how widespread is the admixture of largely isolated populations that can potentially come into contact.
The headwaters of Amazonian rivers are one of the last frontiers in Neotropical biogeography, and little is known from these regions. In the next chapter, I will show that many areas of contact cluster at the headwaters of the Negro and Branco rivers. This contact, however, seems to translate rarely into admixture of populations. Three cases suggest the presence of extensive hybrid zones on the Rio Negro basin, two of which have been studied by Haffer (1974) 55 in detail and involved two pairs of Ramphastos toucan subspecies, and one which is under current study that involves two subspecies of Xiphorhynchus woodcreepers (C. A. Marantz, pers. comm.). Cases of occasional hybridization seem to be more common, in that I found a few intermediate-looking specimens in populations that otherwise looked “pure” in at least 10 other pairs of taxa, including three in the Ramphastidae, three in the Picidae, and three in the
Thamnophilidae, the latter all closely related and forming a well supported ant-following clade
(Brumfield et al., 2007). Given the wide array of taxa investigated, it seems surprising that putative hybrids are concentrated largely in three families. Although preliminary, these data suggest that some common characters might influence the likelihood of hybridization between pairs. In the cases above, weak sampling prevents knowing whether these represent isolated hybridization events or examples of well-established hybrid zones.
Levels of Genetic Differentiation Along the Rio Negro
The lower and wider sections of Amazonian rivers are thought to represent stronger biogeographical barriers than do their narrower headwaters (Beven et al., 1984; Ayres &
Clutton- Brock, 1992; Haffer, 1992; Hayes & Sewlal, 2004). Whereas the lower Amazon divides ~150 pairs of avian taxa, the upper Amazon (the Marañón river) acts as a barrier to fewer than 20 taxa (Haffer, 1978). In the case of the Rio Negro, the number of taxa divided by the lower section of the river (~90 pairs) is also much higher than the number divided by the upper portion of the river (15 pairs) (see chapter 2 and 3). This is hardly surprising, given that whereas the lower Negro can reach up to 20 km in width, the middle and upper Negro are much narrower and have two extensive fluvial archipelagos that could serve as dispersal stepping-stones for the avifauna (Cintra et al., 2007), although it remains unclear whether there is any gene flow through those islands.
56 Despite the obvious reduction of the number of taxa divided by the upper portions of rivers, it remains largely untested whether the pairwise levels of genetic divergence decrease clinally towards the headwaters, as previously suggested (Haffer, 1993, 1997a; Ayres & Clutton-Brock,
1992; Patton et al., 1994, 1996; Patton & Silva, 1998; Silva & Patton, 1998; Gascon et al.,
2000). The expected genetic homogenization of populations towards the headwaters was first shown in a population of Saddle-back Tamarins (Saguinus fuscicollis) along the Rio Juruá (Peres et al., 1996), but this was not found in a group of spiny rats (Patton et al., 1994) or in a species of poison-dart frog along the same river (Lougheed et al., 1999). Unfortunately, the Rio Juruá, where extensive sampling has occurred (Patton et al., 1994; Patton et al., 2000; Gascon et al.,
2000), is not a good choice for biogeographical studies, because this river is not known as an important biogeographical barrier (Haffer, 1997a).
Here, I showed that at a community level, taking into consideration 37 species (or species pairs) along the Rio Negro, genetic distances between the lower and the upper Rio Negro are highly significant, with a reduction in divergence towards the headwaters. This phenomenon, however, is not clinal, and appears to simply represent the function of having many taxa that do not reach the upper sections of the Rio Negro. When comparing only the 14 pairs divided along the entire river, I found no evidence of genetic homogenization towards the headwaters.
Therefore, whereas in general fewer taxa are divided on the upper portions of a river, levels of genetic divergence are not lower in these regions, and they do not decrease clinally towards the headwaters, as predicted by Haffer (1997a). Thus, pairs remain genetically distinct from the mouth of the river up to the upper section. It is possible that the upper Rio Negro does not accurately represent the headwaters, and that farther upstream populations coalesce, although this is not evident from the morphological data. It could also be argued that mitochondrial genes are not appropriate to assess homogenization of populations, because each individual will only
57 have a maternal copy of the mitochondria (and all its genes). Homogenization, however, would still be apparent by clinal phenotypes and more intermediate-looking birds, but this is not found along the Rio Negro.
Unfortunately for a comparative analysis, there are no similar studies from elsewhere in
Amazonia. Although Bates et al. (2004) found that six of 10 avian populations showed breaks across the Rio Teles Pires (which could be considered the headwaters of the Rio Tapajós), they did not have samples from the lower portion of the Tapajós for comparison. Aleixo (2002) found no signs of gene flow between sister clades of a woodcreeper species at the headwaters of the Xingú and Tapajós rivers. The current investigation represents the first case study but of just one Amazonian river. The data presented here strongly suggest that avian populations are not more similar (genetically) at the headwaters than across the mouth of an extensive river such as the Rio Negro.
Testing for Simultaneous Divergence
Distinguishing between simultaneous and non-simultaneous divergence using comparative phylogeographic datasets remains one of the most challenging goals of statistical phylogeography. Although multi-locus data can greatly improve parameter estimates, extensive simulations on single-locus datasets have shown that they perform extremely well under hierarchical ABC approaches (Hickerson et al., 2007; Hickerson & Mayer, 2008; Carnaval et al.,
2010). Instead of obtaining multiple parameters for a number of genes, HABC achieves this by pooling information across groups without assuming that they come from similar populations
(Hickerson & Mayer, 2008). This method estimates temporal congruence across groups by obtaining information from the entire sample. According to Gelman and Hill (2007), the use of hierarchical models "borrows strength" from the dataset, which means that “sub-parameter
58 estimates (!) are pooled together from hyper-parameter estimates of ", rather than estimated independently from each phylogeographic dataset” (Hickerson & Mayer, 2008).
Several lines of evidence suggest that the history of the taxa divided by the Rio Branco differs from those divided by the Rio Negro. First, the values of Psi (number of possible divergence times) are higher for the Rio Branco. Whereas a scenario of one or two vicariant events is consistent with the simulations for the Rio Negro, such a scenario is not supported by the Rio Branco dataset. Differences in values of Psi have been important in the inference of two waves of diversification in mammals and reptiles in Baja California (Leache et al., 2007) and represent one of the most important pieces of information provided by ABC approaches.
Second, the values of the mean time of divergence (E(t)) suggest that the pairs of taxa divided by the Rio Branco are more than twice as old as those divided by the Rio Negro. Although the estimates of E(t) are relative to each other, they are clear in this difference. Finally, the values of
!, the ratio between the variance and the mean, tend towards zero when divergence times are equal for all pairs of taxa. Therefore, having 20% of the values of ! as zero for the Rio Negro dataset supports a history of simultaneous divergence. On the other hand, this was not found across the Rio Branco, and no value of ! equaling zero was selected in the 500 posterior values.
Hyper-posterior estimates and 95% confidence intervals across both datasets remained similar under different tolerance values; therefore, it seems that whether 500 or 1,000 accepted posterior values were used did no change the conclusions. Similarly, I ran three independent simulations, each with 1,000,000 draws from the prior, and all resulted in very similar joint posterior densities for E(t)(mean divergence times), !, and " (number of possible divergence times) (Appendices E, F). The strength of results using different priors and parameters has been found in simulations (Hickerson et al., 2007) and suggests that the analyses are robust to changes in such parameters.
59 The Role of Rivers in the Speciation Process
Rivers potentially play an important role in the evolution of vertebrate diversification in
Amazonia. Their ubiquitous presence in the Amazon and their role in defining current distributions is unmatched by any other landscape feature in the region. The predictions of the
‘riverine barrier hypothesis’, however, obtained mixed support from the data, and these seem to differ for the two rivers investigated. This is not surprising, however, given that the Branco and
Negro rivers differ in their history, ecology, and hydrology.
The history of the Rio Branco region seems to have been much more traumatic that most of
Amazonia, with severe arid periods that affected large areas currently covered by forest. The
Pleistocene history in the Rio Branco basin seems to be in concert with the refuge-driven
Amazon envisioned by Haffer (Schaeffer & Vale, 1997; Ab’Saber, 1997). The presence of large savannas (the largest in Amazonia), extensive areas covered by white-sands, and the existence of paleo-dunes seem to support the idea that this region has been exposed to climatic fluctuations
(Latrubesse & Nelson, 2001; Carneiro Filho et al., 2002), although this remains controversial
(Colinvaux et al., 2000). Furthermore, the Pleistocene origin of the Rio Branco (Schaeffer &
Dalrymple, 1995) would strongly argue against a role of this river in the diversification of the avifauna, given that the high levels of genetic differentiation found in many pairs of avian taxa across its margins would not be consistent with such a recent origin. Furthermore, from the nine pairs of taxa that replace across the Branco for which species-level phylogenies are available
(Table 3.3), only in two cases are these populations each other’s closest relatives. Coalescent simulations also suggest several independent diversification events and argue against simultaneous diversification. Therefore, although the Rio Branco represents an important biogeographical barrier that plays a role in maintaining current levels of biodiversity, it seems clear that it was likely not involved in the diversification process of those pairs of taxa it divides.
60 These results suggest a scenario in which the Rio Branco is simply the natural meeting point for taxa that diverged elsewhere and subsequently spread to their present opposite-bank positions
(Simpson & Haffer, 1987; Haffer, 1992). Furthermore, the Rio Branco is not the only landscape feature in the region to divide avian populations of closely related but phenotypically distinct pairs of taxa. Savannas, campinas, flooded forests, and a mountain chain seem also to play significant roles in defining present-day distributions (see previous chapter).
The history of the Rio Negro is less known than that of the Rio Branco. However, the actual origin of the Rio Negro seems to be older and more related to other tectonic and geomorphological changes in Amazonia. Although palaeohydrological data are scarce and still fragmentary in Amazonia (Latrubesse, 2003), there is some consensus that the Amazon river probably obtained its present configuration on the Late Miocene-Pliocene (c. 2.6-7 mya) (Irion
& Kalliola, 2010). On the other hand, Campbell and co-workers (2006) have suggested an earlier date for the present Amazon configuration, which could be as late as the Late Pliocene -
Early Pleistocene (~2 mya). The lack of consistent dates on the origin of most Amazonian rivers
(Hoorn & Vonhof, 2006) prevents us from testing explicit historical hypotheses. Therefore, to create a plausible scenario that would account for the diversification process that created modern biodiversity is difficult.
Some of the evidence arguing against a role of the Rio Branco in the process of avian diversification, however, has not been found for the Rio Negro. For example, five of six species- level phylogenies of pairs of taxa that replace along the entire Rio Negro are each other’s closest relatives (Table 4.2), and importantly, coalescent simulations are consistent with a single vicariant event (although two vicariant events are nearly as probable). This means that we cannot reject the possibility that the Rio Negro, besides being the most important biogeographical barrier in the Guiana Shield and possibly the second most important Amazonian
61 river in terms of the number of taxa that divides, may have played a role in the original diversification of those taxa. Those simulations, however, were performed on only 14 pairs of taxa that replace one another along the entire extension of the Rio Negro. At least another 75 pairs that replace each other across the lower Rio Negro were not included in the analyses. The lack of even coarse time estimates of the origin of the Rio Negro also prevented me from explicitly testing the likelihood of this river existing prior to the avian populations that it divides.
Below, I include the predictions of the ‘riverine barrier hypothesis’ and their support from the available data.
Do replacement patterns cluster along rivers, rather that within interfluvia? Most phenotypically distinct populations studied here replace each other across the lower Rio Negro, but at least 15 pairs of the replacing pairs meet in the Branco/Negro interfluvium (see chapter 4).
The rivers’ headwaters, particularly on the Rio Branco, act as a connection between ‘Guianan endemics’ and western Amazonian taxa. In this region, however, rivers are not the only landscape features that seem to coincide with avian distribution limits; non-forested habitats
(savannas and campinas), smaller rivers, and mountains also seem to play an important role (see chapter 2). These observations are consistent with Haffer’s view that Amazonian rivers represented natural suture zones upon expanding populations that came into secondary contact within interfluvia (Haffer, 1974, 1978, 1992, 1997a, 2002; Simpson & Haffer, 1978).
Are pairs of taxa each other’s closest relatives? Although species-level phylogenies are currently available for only a limited group of taxa, this number is rising quickly, and we are now in a position to evaluate this question using nearly 20 pairs of taxa with replacements across the lower Rio Negro (see Table 4.2). Among these pairs, it is evident that at least half are not sister taxa. It is questionable, however, whether a non-sister relationship invalidates a history of primary diversification as previously suggested (Haffer 1993, Moritz et al., 2000). The
62 evolutionary process is a continuous one and as such does not stop after a given vicariant event.
Therefore, despite being separated by a common barrier in the past, it is possible that each population continued its evolutionary history and that subsequent diversification events blur the role of a primary barrier. On the other hand, distantly related pairs are likely the result of secondary contact, and that seems to be the case in many of the pairs under investigation.
Fortunately, the number of species-level phylogenies is increasing, and in a few years, we will have a better understanding of the historical processes involved in their divergence.
Does the genetic differentiation between populations across rivers decrease towards the headwaters? Although the lower portions of the Rio Negro divide more species, and therefore have larger values of genetic divergence as a whole, levels of genetic divergence are not clinally reduced towards the headwaters. In fact, clinal reduction has not been shown for any Amazonian river so far.
Are pairs of co-distributed taxa across the two rivers the result of simultaneous divergence? Coalescent simulations suggest different results for the two rivers studied, strongly arguing against simultaneous divergence for pairs across the Rio Branco, but supporting few diversification events accounting for the diversification across the Rio Negro. These results suggest that rivers are likely to have their own idiosyncratic histories, and that generalizations are not warranted. Similar studies need to be performed across more Amazonian rivers to find common historical patterns.
Is the inferred timing of divergence across pairs posterior to the formation of the rivers? Biogeographical studies on Amazonian rivers often neglect one fundamental piece of information: the age of the rivers under investigation. Fortunately, the two rivers under study in this chapter represent two dramatically different histories in terms of their geomorphological features and time of formation. Whereas the Rio Negro is a craton river, and as such is
63 characterized by channel-dominated processes in which sand deposition takes place in the form of sand bars within river channels (Hoorn et al., 2010), the Rio Branco is thought to be of a much more recent origin, possibly Pleistocene, and the result of a recent capture of an entire protobasin that drained towards the Essequibo River (Schaefer & Vale, 1997). If these estimates are accurate, then current levels of genetic divergence across pairs could hardly be consistent with such a recent origin.
Final Remarks on Diversification
The data from the Guiana Shield, where two of the most biogeographically important
Amazonian rivers are located, offer mixed answers in our quest to understand vertebrate diversification processes. Rivers as different as the Negro and Branco show comparable levels of biogeographical importance. This suggests that independently of a river’s origin and geomorphology, large Amazonian rivers are powerful barriers to many animals. This, however, is not universal, as nearly two thirds of the region’s avifauna is not bounded by any river, and their populations extend unbroken throughout the Guiana Shield.
Acting as single vicariant events upon continuous populations, however, is not the only way rivers may affect the speciation process. Large Amazonian rivers could also act as barriers after dispersing individuals (founders) establish new populations on ‘empty territory’ (Haffer, 1978,
1987, 1993, 1997a, 2001). If these founders manage to establish viable populations and are isolated long enough, then they could create a biogeographical pattern with many pairs each diverging at its own pace. Two problems with this alterative hypothesis are that: i) the inferred age of the pairs should still be younger than the formation of the rivers, and this does not seem to be the case, particularly on the Rio Branco, and ii) it would still be expected that taxa across rivers would not be distantly related, as seen in several of the pairs under investigation.
Therefore, this alternative seems to be less likely than the original ‘riverine barrier hypothesis’.
64 Another variant of the ‘riverine’ hypothesis is the ‘river-refuge hypothesis’ (Ayres, 1986;
Capparella, 1991; Ayres & Clutton-Brock, 1992), under which rainforest contractions on the
Amazon’s periphery, could isolate forest-dwelling animals in `semi-refugia' between the wide lower courses of rivers (see Fig 3.5). This model, dismissed by Haffer (1997a) as less likely than the ‘refuge hypothesis,’ has one particular advantage over the ‘riverine barrier hypothesis’, because it could explain present distributional patterns in the rivers’ headwaters, where successive forest expansions would have created areas of secondary contact at the headwaters.
Differential dispersal abilities and pair-specific histories would locate each contact zone in one particular region, although they would cluster at the headwaters. Sister relationships between every taxon pair would not be expected, because not every interfluvium would be isolated at the same time, thereby creating idiosyncratic patterns of gene flow between pairs.
Fig. 3.5. Schematic representation of the ‘river-refuge hypothesis’, where reduction of forests would only occur in Amazonian peripheral areas, and populations would remain isolated along the lower (and wider) stretches of Amazonian rivers (from Haffer, 1997a).
65 CHAPTER 4 THE ROLE OF GEOGRAPHIC AND ECOLOGICAL BARRIERS IN THE LOCATION OF AVIAN CONTACT ZONES AND PHYLOGEOGRAPHIC BREAKS IN THE GUIANA SHIELD
INTRODUCTION Allopatric speciation, or speciation by geographic isolation, is considered the main mode of diversification among vertebrates (Coyne & Orr, 2004; Price, 2008). Under this model, biological populations are physically isolated by an extrinsic barrier and subsequently evolve intrinsic (genetic) reproductive isolation mechanisms (Mayr, 1963). Therefore, understanding the factors that limit sympatry in closely related taxa provides an important framework to understand evolutionary processes.
The formation of boundaries and the role of geographic and ecological barriers to gene flow and dispersal can be studied along phylogeographic breaks where the distributions of distinct, yet closely related taxa, come close together but remain isolated along sharply defined regions, usually as a result of a physical barrier (Avise, 2000; Coyne & Orr, 2004). When barriers are absent, pairs of taxa can potentially come into actual contact, forming areas known as contact zones (Mayr & O’Hara, 1986). Populations that come into contact can either interbreed or remain reproductively isolated, and in some cases they may form stable hybrid zones, where genetically distinct populations meet, mate, and breed (Barton & Hewitt, 1985). The outcome of secondary contact between these populations depends on whether they have evolved enough changes (genetic, morphological, or behavioral) to restrict interbreeding sufficiently to remain distinct. Populations that remained isolated for longer periods of time are more likely to have attained reproductive isolation (Coyne & Orr, 1989). Therefore, the outcome of secondary contact may be directly related to the time of isolation (Mayr & O’Hara, 1986), which can be indirectly measured by levels of genetic divergence (Coyne & Orr, 1989, 2004; Foltz, 1997).
66 Together, phylogeographic breaks, contact zones, and hybrid zones offer complementary views of the past and present of a region’s biota, offering hints about factors important for the generation and maintenance of species diversity. Some geographic regions, known as suture zones, are of particular interest, because they mark areas with disproportionally high numbers of phylogeographic breaks, contact zones, and hybrid zones (Anderson, 1948; Remington, 1968;
Swenson & Howard, 2004, 2005). These regions are particularly interesting for biological studies, because they allow the use of multiple species and independent lineages in a comparative framework.
A prominent avian suture zone is found on the Guiana Shield, in northern Amazonia, where two distinct avifaunas meet within a relatively small geographic area (Haffer, 1974; Naka et al.,
2006; chapter 2). In this region, defined by two important Amazonian rivers (the Rio Branco and the Rio Negro) nearly 100 pairs of closely related, forest-dwelling avian taxa replace one another geographically (see chapters 2 and 3). Within the Branco/Negro interfluvium some pairs exhibit sharp phylogeographic breaks without evidence of direct contact, others have contact zones but no evidence of hybridization, and others from hybrid zones where they come into contact. The valleys of the Branco and Negro rivers and their interfluvium have been previously identified by
Haffer (1974; 1987) as having “many contact zones” and separating “conspicuously different eastern and western populations”, but little is known about the exact locations of these areas, and no quantitative or molecular studies are available to define the regions of contact. Furthermore, little is known about the interactions in those regions where pairs of closely related taxa come into contact. In fact, studies of the Amazonian suture zones are few at the community level for any biotic group (but see Haffer, 1997b; Whinnett et al., 2005).
The location of suture zones and their relation to the origin and rearrangement of species distributions has captured the attention of biologists in temperate regions since the mid 20th
67 century, when Anderson (1948) suggested that North American hybrid zones clustered along hybrid landscapes. Hybrid landscapes could either result from human interference or from natural transitional areas, such as “tension zones” or ecotones (Barton & Hewitt, 1985).
Remington (1968) hypothesized that suture zones should cluster along mountain foothills, and particularly within midpoints between putative refugia, following post-glacial biological expansions. Contemporary ideas on hybrid zones suggest that they cluster along mountain passes and are the result of glacier retreats, invoking similar causes for the location of suture zones in Europe and North America (Hewitt, 1996; 1999). In general, climatic changes and physical barriers seem to be responsible for current distributional patterns in temperate zones
(Graham et al, 1996; Hewitt, 2004; Peters et al., 2005; Swenson & Howard, 2005; Price, 2010).
In Amazonia, where mountain chains do not represent major biogeographical barriers, and where ice sheets apparently did not affect the landscape directly (Bush, 1994; Colinvaux, 1997;
Colinvaux et al., 2000), analogous ideas have been put forward to explain the location of suture zones. Instead of mountains, large Amazonian rivers have been invoked as barriers (Wallace,
1852; Sick, 1967; Hershkovitz, 1977; Capparella, 1988; Ayers & Clutton-Brock, 1992), and forest refugia within savannas rather than within glaciers were suggested to be important in the present distribution of contact zones (Haffer, 1969; 1974; Brown & Ab’Saber, 1977). Similarly, headwaters of large rivers (rather than temperate mountain passes or foothills) may represent natural places where pairs of allopatric taxa could come into secondary contact.
Physical barriers are not the only causes of distribution breaks. The location of distribution boundaries could be defined by sharp changes in climatic variables or ecological gradients
(Endler 1977, 1982a; Schneider et al., 1999), biotic interactions (Hutchinson, 1957; Case &
Taper, 2000; Anderson et al., 2002), or simply by historical factors (Gaston, 2003; Hawkins &
Porter, 2003). Therefore, history, geography, and ecology are powerful forces that could act in
68 concert to define the geographic location where closely related pairs of taxa separate, meet, or interbreed (Endler, 1982b; Wiens & Donahue, 2004; Case et al., 2005).
In recent years, the development of ecological niche models (ENMs), based on species occurrence data and the availability of environmental data from virtually every corner of the planet (Guisan & Zimmermann, 2000; Peterson, 2001), has allowed the use of ENMs in evolutionary biology and has catalyzed studies on the role of ecology in the speciation process
(Peterson et al. 1999; Graham et al. 2004; Kozak & Wiens, 2006; Swenson, 2008). Using ENM frameworks, levels of divergence, overlap, breath, and similarity in the ecological niches of pairs of closely related taxa can be quantified and evaluated (Warren et al., 2008). The influence of topography, temperature, rainfall, and other abiotic factors on the location of contact zones and phylogeographic breaks can also be investigated by using null models and spatially explicit approaches (Costa et al., 2008; Swenson, 2008; Warren et al., 2008; McCormack et al., 2009).
Alternatively, in the same way that species richness gradients can arise through simple geometric constraints on species range boundaries in the absence of any environmental gradients
(Colwell et al., 2004), clusters of contact zones and phylogeographic breaks could peak in the center of a given region because of the ‘mid-domain effect’, independently of any environmental, historic, or biotic factors (Colwell & Hurtt, 1994; Colwell & Lees, 2000).
Disentangling the role of these forces remains one of the most challenging issues in evolutionary biology (Gaston, 2003; Swenson, 2008) but needs to be taken into consideration for understanding current patterns of geographic distributions.
In this chapter, I make use of an extraordinary biogeographical setting in one of the most biologically diverse regions on earth to study the role of physical barriers and ecological factors in the location of avian contact zones and phylogeographic breaks using ~80 pairs of closely related avian taxa. In particular, I describe the geographical location of avian contact zones and
69 phylogeographic breaks in the terra firme forests of the Guiana Shield and address the following questions: i) are contact zones and phylogeographic breaks geographically clustered? ii) are they associated with obvious physical features? iii) what role, if any, is played by ecological factors? and iv) can they be explained by geometric constraints of the study area (‘mid-domain effect’)?
The answers to those questions lead us naturally to address further questions, which include: v) what is the outcome of secondary contact (hybridization and introgression, parapatric exclusion, or sympatry) in pairs of closely related taxa that come into contact? And vi) is this outcome associated with the time of species divergence, as inferred from molecular data?
To answer these questions, I used a combination of distributional and molecular data obtained from two large independent datasets that include ~28,000 locality records of nearly 200 taxa from museum specimens and field records and ~1,500 mtDNA sequences from the study region. These data were integrated using geographic information systems (GIS) mapping tools and a phylogeographic approach. I also used recently developed spatial analytical tools to assess data clustering, and I used ENMs to test whether environmental variables are related to the distributional patterns observed.
METHODS
Study Area
The Guiana Shield, 2.5 million km2 in size, covers much of northern South America north of the Amazon River and east of the Japurá-Caquetá rivers, ranging from the Brazilian state of
Amapá on the Atlantic Ocean to the Sierra de Chiribiquete near the Colombian Andean foothills
(Conservation International, 2003; Hammond, 2005b). Most of the Guiana Shield is covered by pristine lowland terra firme forests, but large areas are dominated by open vegetation types, such as savannas and white-sand forests, and flooded forests generally border the rivers (Naka et al.,
2006; 2007). The Guiana Shield also stands as one of the most environmental and
70 topographically complex areas of the entire Amazon, where savannas, campinas (white-sand forests), flooded forests, terra firme forests, and the spectacular table-top mountains known as tepuis intermingle in a fashion hardly seen elsewhere in Amazonia (Hammond, 2005a; Naka et al., 2006). For more information of the Guiana Shield and its avifauna see Chapter 2. A more detailed description of the two river systems involved in this chapter was presented in Chapter 3.
Selection of Taxa and Distributional Database
I created a preliminary list of pairs of taxa (species or subspecies) that replace each other geographically and ecologically in the terra firme forests of the Guiana Shield. I based this list on distributional data from my own fieldwork in the region, but mainly from the traditional ornithological literature, including the works of Hellmayr (1929-1938), Peters et al., (1931-
1986), Zimmer (1933-1944), Hellmayr & Conover (1942-1949), Pinto (1978), and more recent compilations such as del Hoyo et al. (Vols. 1-12, 1992-2008). Species order and classification follow the most updated taxonomic revision available (Remsen et al., 2010).
One major concern in spatial analyses of multiple species is to obtain accurate digital maps that can be used to identify distribution patterns. Such maps are not available at the level of detail necessary for this study, especially for those pairs of taxa presently considered different subspecies within a polytypic species. Therefore, I had to build my own maps for each pair of taxa. With that in mind, I assembled a large distributional dataset, based on museum specimens, published sources, audio recordings, and field data obtained from several museums, my own fieldwork in the region, and unpublished distributional data from colleagues (see chapter 2). At present, this database has over 28,000 records for ~200 taxa, including all pairs studied here.
Most locality points were already georeferenced, but when geographic coordinates were not available from collection databases, I used national gazetteers (e.g., Paynter & Traylor, 1982,
1991; Paynter et al., 1981) to obtain geographic coordinates as accurately as possible. Between
71 2001 and 2008 I conducted extensive fieldwork in the region, carrying out general avian surveys, collecting specimens, and recording vocalizations. Specimens and recordings were deposited at the INPA bird collection and the Amazonian Sound Archive (ASA). For a more complete description of the sources of the database, see chapter 2.
Georeferenced distributional data were archived in Microsoft Access 2007 and later imported into ARCVIEW 9.3.1 (Environmental Systems Research Institute, 2009) to build the maps. Some pairs of taxa described in the literature could not be securely diagnosed (separated) from museum specimen evidence alone. If molecular data was available for this pair, then I relied exclusively on molecular evidence to assess their phylogeographic breaks. Described pairs for which I had too few samples to create distributional maps were not included in these analyses.
Subspecies identification from collections required care. Given the large amount of data, it was not feasible to check every specimen. I focused on checking specimens from Brazil (INPA and MPEG) and Venezuela (COP and CBRG), and I personally studied many of the specimens mapped. In addition, curators and collection managers from several institutions kindly reviewed specimens under their care. Because pairs often exclude one another geographically, I visually checked for potential identification errors by displaying each pair geographically. I paid close attention to specimens that fell within the range of its sister taxon and also to specimens located near putative contact zones, where either member of a pair could be present.
Mapping the Location of Phylogeographic Breaks and Contact Zones
• Phylogeographic Breaks. I used the distributional dataset to define each pair’s morphological break. Because patterns of phenotypic variation was shown to be highly concordant with the geographic structure of the genetic marker studied (see chapter 3), I refer to these plumage breaks as phylogeographic breaks, although I am aware that this term is generally used as a population genetic phenomenon. To map phylogeographic breaks, I included each pair
72 of taxa in a single map and created a break polygon for each pair. A polygon is superior to a line in this case because it allows the inclusion of the uncertainty in the position of the break due to incomplete sampling. Within the polygon the two forms replace one another. Therefore, each pair of taxa resulted in a polygon that separated their easternmost (for western taxa) and westernmost (for eastern taxa) records of each pair (Fig. 4.1). When fieldwork or available specimens indicated that a given pair replaced each other across a river or other obvious physical barrier, I represented the break as a polygon encompassing the extent of the barrier within the study area.
Fig. 4.1. Example of a phylogeographic break polygon (in red) created from the distributional data to represent the region where two taxa divide.
• Contact Zones. I used the distributional dataset to define which pairs come into direct contact. Pairs were considered ‘in contact’ if they i) occur in sympatry (actual contact), ii) hybridize, or iii) occur in parapatry (potentially in contact). I considered ‘parapatric’ those pairs of taxa that have been recorded at nearby localities (<100 km) and which have no obvious
73 physical barrier between their nearest specimen localities. To map contact zones, I created a 50- km buffer around the point of contact (location of syntopy or putative hybrid), providing independent 50-km-radius circles around each point. For those pairs in potential contact (i.e., no actual point is known to be a meeting point), I used the midpoint in between records as the potential point of contact, creating 50-km buffer circles around this point.
Clustering and Geographic Constraints
I used two different methods to assess whether phylogeographic breaks and contact zones were geographically dispersed or clustered throughout the study area. I first used the Getis-Ord general G (Global) statistic, which can be calculated within ARCVIEW and has been successfully used to map suture zones in North America (Swenson & Howard, 2004, 2005). The general G statistic identifies whether high or low grid cell values cluster across the landscape (Getis & Ord,
1992; Ord & Getis, 1995), but does not identify the location of these clusters. Given that I found significant values of clustering, I used the local G statistic to detect the locations of these clusters. To conduct the analyses, I converted both polygons (phylogeographic breaks) and 50- km radius circles (contact zones) into raster layers at a resolution of 5 minutes (10-km cells).
Raster layers for each pair of taxa were then overlaid, resulting in a map layer that had the number of breaks or contact zones as a value for each cell in the landscape. One map layer was created and for phylogeographic breaks and another for contact zones. These rasters were then converted to features for separate analyses using Spatial Statistics Tools in ARCVIEW.
The second method included an algorithm that uses cellular automata, as suggested by
Swenson (2008). This method has been implemented in BIOGEOSIM (Gotelli et al., 2007) and has been used in the past to test for the mid-domain effect and for environmental gradients of species diversity (Rahbek et al., 2007). This method relies on the “geometric constraint model”
(Jetz & Rahbek, 2001), also known as the ‘spreading dye’ model, which assumes range
74 continuity. BIOGEOSIM simulates the stochastic origin and spread of species geographic ranges in a landscape (Rahbek et al., 2007). Instead of using species ranges, we used the polygons and circles (phylogeographic breaks and contact zones) in the analyses. We used a homogeneous landscape to generate a null distribution of two-dimensional simulated ranges that could be compared to empirical data. The null dataset created has the same size as the species’ ranges in the empirical dataset within the study area. BIOGEOSIM first places a random point in the study area (the defined domain), allowing it to grow in random directions until it reaches the same size as each of the observed data rasters (polygons for the breaks or circles for the contact zones).
Given that each phylogeographic break has a known size, BIOGEOSIM creates similar-sized polygons for each species, but randomly shaped and located. Because contact zone circles are all the same size, the null model consists of one-sized polygons. All polygons are then added, resulting in two independent maps from which is estimated the statistical expectation of range overlap of phylogeographic breaks and contact zones in each grid cell. This process is then repeated and iterated multiple times giving multiple ‘random’ phylogeographic breaks and contact zones maps. These replicates of ‘random’ maps are then used as the null distribution to which the observed map is compared. I used 300 repetitions, 1,000 iterations, and an 8-cell contiguous dispersal pattern across a 10-km2 grid (166 x 278 cells). The observed distributions were compared to the expected values with a goodness-of-fit statistic and a regression. Although possible to include a heterogeneous landscape, I considered the expansion of each cell as equiprobable, as would be in a uniform environment. I also used this method to determine whether the distribution of phylogeographic breaks and contact zones across the study area could be explained by the domain’s area geometry, referred to as the mid-domain effect (Colwell &
Hurtt, 1994; Colwell et al., 2004).
75 To create hotspot maps of phylogeographic breaks and contact zones, I transformed each taxon’s observed polygons or circles into 5-km resolution rasters and overlaid them using the raster calculator in ARCVIEW’s Spatial Analyst Extension.
Testing for Niche Divergence
One possible way of testing for the role of ecology in the location of phylogeographic breaks is by comparing the levels of niche overlap, breath, and similarity between pairs of closely related taxa (Warren et al., 2008). A significant difference in their ecological niches suggests niche divergence, and the location of the break between a pair of replacing taxa could be in response to environmental variables, whereas a non-significant difference is consistent with niche conservatism and can be interpreted as evidence that ecological variables may not be playing such an important role in the location of current breaks (Peterson et al., 1999).
I selected five pairs of taxa to conduct niche comparisons, encompassing the three replacement patterns described later in this chapter. Two pairs of taxa (Epinecrophylla gutturalis/E. haematonota and Hypocnemis cantator/H. flavescens) represent cases in which phenotypic and phylogeographic breaks coincide with the Rio Branco and the savannas in Brazil and Venezuela, but both pairs apparently come into contact at the Branco headwaters. Two other pairs (Brotogeris chrysoptera/B. cyanoptera and Willisornis p. poecilinotus/W. p. duidae) replace each other within the Branco/Negro interfluvium, where they come into contact. One pair (Pithys a. albifrons/P. a. peruvianus) replace each other across the Rio Negro and come into contact at the Negro headwaters (see chapter 3).
• Ecological Niche Models. The first step in comparing ecological niches requires the creation of taxon-specific ENMs. Among the several algorithms available for ENMs, I selected the Maximum Entropy algorithm (Maxent) (Phillips et al., 2006). Maxent is a widely used method that has outperformed other approaches in simulations (Elith et al. 2006). Maxent uses
76 maximum entropy density estimation to create a probability distribution over a study area that represents the potential distribution of a species, based on specimen locality data and environmental predictor variables (Phillips et al., 2004; 2006; Phillips & Dudik, 2008).
Distribution models in Maxent were created using 25% of the locality points as test data and 25 replications to evaluate model performance and stability of results. I examined the Area Under
Curve (AUC) scores on test and training data, consistency of results among replicates and the model’s ability to predict a taxon’s distribution to determine model performance, because interpreting model accuracy using one measure alone can be problematic (Lobo et al., 2008;
Jiménez-Valverde et al., 2008; Peterson et al., 2008). I created the final Maxent models for the comparative analyses from the average of all 25 replicates, using all available points. To run niche overlap, niche breadth, identity, background, and the linear range-breaking tests I used
ENMTools, which uses null distributions to compare the observed data (Warren et al., 2008).
• Environmental Variables. I obtained two types of data: climatic and elevational data from the Worldclim database (Hijmans et al., 2005) and vegetation data available from Saatchi’s website (Saatchi et al. 2007; Saatchi et al. 2008; Buermann et al. 2008). An additional unpublished vegetation data layer was obtained from J. McCormack. In total, I obtained an initial subset of 22 layers that were later reduced to 11 (Table 4.1). To avoid using highly correlated variables, I ran Pearson’s correlation on 5,000 points randomly located in the Guiana
Shield. First, I compared correlations between temperature, precipitation, and vegetation variables separately, noting which variables were highly correlated. Second, I performed a
Principal Component Analysis (PCA) to determine which variables loaded on the same PC axes
(data not shown). Finally, I eliminated variables that showed a Pearson’s correlation > 0.670, which loaded on the same PC axis, and were in the same category for temperature, precipitation, or vegetation variables. Three additional layers from the vegetation data were eliminated due to
77 a high proportion of missing data, which created problems when generating and comparing
ENMs. I used the program ENMTOOLS (Warren et al., 2008) to compare ecological niches among the five pairs of replacing taxa selected. All layers were at a 1-km (30’’) resolution and were rescaled to 5-km (2.5’) resolutions to account for the margin of error of my locality points, which in many cases do not have such high resolution.
Table 4.1. Eleven environmental and vegetation layers used in the ecological niche models. Original layers were available at a resolution of 30’’ but were rescaled to a lower resolution of 2.5’ to match the resolution of the specimen data. Layer Layer biological interpretation Data source Mean Diurnal Range (Mean of monthly (max temp - Bio2 min temp)) Bio4 Temperature Seasonality (standard deviation*100) Bio5 Max Temperature of Warmest Month Worldclim data Bio15 Precipitation Seasonality (Coefficient of Variation) (Hijmans et al. 2005) Bio16 Precipitation of Wettest Quarter Srtm Elevation; Shuttle Radar Topography Mission Srtm_std Change in elevation over each km Jerswettxr Canopy texture. Wet season texture radar data Saatchi et al. 2000 Jerswetbk Flooded forest. Wet season backscatter radar data Qscat_mean Aboveground biomass and canopy roughness Long et al. 2002 Qscat_stdv Degree of phenological change over a year
• Niche Overlap. This test compares the probability distribution over geographic space generated by Maxent between two taxa, using a similarity metric (Warren et al., 2008).
ENMTOOLS generates values for Schoener’s D (Schoener, 1968) and the “I” niche overlap statistic proposed by Warren et al. (2008) based on the Hellinger distance. Because I found no difference between the outcomes of the comparisons made with each metric, I only report results based on Schoener’s D, where D = 1 if niche overlap between the two taxa is identical and D= 0 when two species share no aspect of their niche.
• Niche Breadth. This parameter is measured as in Levins (1968), where Breadth (B)=
1/!pi2, and pi is the proportion of all resources used by the population that consists of items in
78 state i. Given that pi is the denominator, species that use many resources (or many environments) will have a large value of B (Warren et al. 2010).
• Identity Test. This test determines whether the observed niche overlap score for two pairs of taxa is lower than a null distribution of niche overlap scores. The null distribution is generated by randomly sampling geographic coordinates from the pooled locality data of the two taxa to create two new samples with the same number of observations as the observed data, repeating this process 100 times. If the observed niche overlap value falls outside of the lower
95% confidence interval, then the two taxa exhibit greater niche divergence than expected by chance.
• Background Test. This test compares the observed niche overlap score between two sister taxa to a null distribution that takes into account the differing background environmental heterogeneity of the two taxa’s ranges. This test creates a null distribution of overlap scores by comparing Maxent models calculated from taxon A’s locality data and random points drawn from the background of its pair (taxon B) range. An identical comparison is made using taxon
B’s locality data and random background points from taxon A’s range. To determine whether niches where different or not from a random null distribution, I treated the hypothesis of niche divergence as a one-tailed test. Overlap values smaller than the null distribution supported niche divergence, whereas values equal or larger than the null supported lack of niche differentiation.
These results were then tested using a one-tailed t-test.
• Linear Range-Breaking Test. I used this test to test the hypothesis that a boundary between a pair of taxa corresponds to a significant environmental boundary (i.e., more different than expected by chance). This is done by randomizing the location of the boundary, by pooling occurrences for the two taxa and randomly a new boundary that separates the pooled occurrences into artificial pairs with similar sample sizes as the original data set (Warren et al., 2010). The
79 following step is to construct ENMs for the newly generated pair many times, measuring their overlap and building a null distribution of expected overlaps across a randomly drawn barrier that splits populations (Warren et al., 2010). These results were tested using a one-tailed t-test.
Collection of Molecular Data and Phylogenetic Analyses
To provide an independent assessment of the location of major breaks in the study area, I gathered molecular data for each pair of taxa. I only included pairs for which I obtained samples from the three main regions of the study area (see chapter 2): i) east of the Rio Branco, ii)
Branco/Negro interfluvium, and iii) west of the Rio Negro. A total of 68 pairs fulfilled these requirements (Table 4.2).
To obtain a common measurement of genetic structure, I sequenced the complete mitochondrial NADH dehydrogenase subunit 2 (ND2) protein-coding gene. This gene has shown to be well suited to recover population monophyly and to assess population structure, both in recently diverged and older related taxa (Price, 2008; Zink & Barrowclough, 2008).
Although single locus studies have limitations, particularly to obtain population parameters due to the stochastic nature of the gene coalescence (Edwards & Beerli, 2000; Arbogast et al., 2002;
Brumfield et al., 2003; Balloux, 2010), my intention here was simply to describe the general genetic structure (and its breaks) in as many pairs as possible, a role that mtDNA genes play surprisingly well (Avise, 2000; Zink & Barrowclough, 2008). For a detailed explanation of the protocols used for DNA extraction, gene amplification, and sequencing, see chapter 3. To establish the general location of the phylogeographic breaks, I built phylogenetic trees and then determined how the split from the most recent common ancestor of each pair correlated with physical barriers in the landscape. For a complete description of tree-building methods see chapter 3.
80 Outcome of Secondary Contact
Within those pairs of taxa that come into actual or potential contact, I looked for evidence of hybridization or introgression. Using museum specimen data, I looked for intermediate-looking birds as evidence of hybridization (see chapter 3). Molecular data were also used to identify cases of hybridization, evidenced by discordance between morphological and molecular data.
On the other hand, lack of putative hybrids was accepted as evidence that pairs of taxa were reproductively isolated. Although it would be ideal to make a quantitative genetic analysis defining and measuring hybrid zones and clinal width, this was not feasible because tissue samples from the contact zones were not available.
Once the outcome of secondary contact was established, a second aim was to test whether the likelihood of introgression or hybridization between a given pair is related to their inferred time of isolation, using levels of genetic divergence as proxies. I compared average genetic distances between pairs of taxa that come into contact and apparently interbreed with those that come into contact and do not hybridize. The use of genetic variation to assess absolute divergence times has many caveats (Arbogast et al. 2002; Graur & Martin, 2004), particularly in the application of molecular clocks in single-locus studies. I avoided using tissue samples from areas where gene flow may occur, and only used samples located away from the inferred contact zones.
RESULTS
I identified a total of 103 pairs of taxa (29 species pairs and 74 subspecies pairs, representing
27 avian families) that replace one another geographically and ecologically in the terra firme forest of the Guiana Shield (Table 4.2, Appendix F). Avian families with the highest number of replacing pairs were the Thamnophilidae (18 pairs), Dendrocolaptidae (13), Tyrannidae (9),
Thraupidae (8), Picidae (6), Furnariidae (5) and Pipridae (5).
81 To assess general replacement patterns in the Guiana Shield, I integrated both the distributional and phylogeographic data for 77 pairs of taxa (Table 4.2). Whereas most pairs replace one another across the lower Rio Negro, three recurrent distributional replacement patterns are evident from the data above this region (farther west). The three patterns were: i) pairs replacing along the Rio Branco and the Roraima-Rupunnuni savannas and the Gran Sabana in Venezuela (40 pairs) (see Fig. 4.2 for an example); ii) pairs replacing along the entire Rio
Negro (19 pairs) (Fig. 4.3); and iii) pairs replacing along the Branco/Negro interfluvium (18 pairs) (Fig. 4.4). The replacement patterns of 26 pairs could not be established with confidence, due mostly to lack of geographic or molecular sampling (19 pairs), but also due to weak morphological and genetic differentiation of some of the pairs (7) (Appendix F). I obtained data on both phenotypic and molecular characters for 53 pairs of taxa (Table 4.2).
Location of Phylogeographic Breaks
Using distributional data, I built phylogeographic break polygons for 69 pairs of taxa, which is the number of pairs for which I obtained sufficient samples to draw sound conclusions of their distribution patterns. Two independent metrics showed that breaks are not randomly distributed within the landscape. The Global Getis-Ord G statistic yielded positive and significant Z scores
(Zg =36.89; P<0.001) indicating that phylogeographic breaks cluster spatially. Local Getis-Ord
G statistics provided the visual answer to where these clusters are located, by depicting all regions where Zgi>1.96, and p<0.05 (Fig. 4.5). A hotspot map of phylogeographic breaks shows that the highest proportion of breaks occurs along the lower Rio Negro, the Rio Branco, and its associated savannas, but also along the middle Rio Negro, and to a lesser extent the upper Rio
Negro (Fig. 4.6). Besides the main rivers and non-forested regions, a number of breaks also aggregate within the Branco/Negro interfluvium and in some cases seem to be independent of landscape features.
82 Table 4.2. Pairs of forest-dwelling avian taxa (N = 77) that replace one another geographically and ecologically in the Guiana Shield. Replacement patterns were separated in morphological and phylogeographic breaks to highlight coincidence and differences, and both datasets were integrated in the second column. B represent pairs of taxa that replace across the Rio Branco (pattern I distributions); N are pairs of taxa that replace across the Rio Negro (lower and upper); and BN are pairs that replace within the Branco/Negro interfluvium. When available, I included the mean corrected genetic distance between phenotypically distinct populations. Contact pattern refers to the areas where the distributions of each pair come close together (i.e., rivers’ headwaters) A: allopatric; PS: parapatric or sympatric (in contact with no known hybrids); H: hybrids or intermediate plumage birds reported or found in collections.
Mean Replacement pattern genetic Contact Pair of taxa Integrated morphology genetic distance pattern Brotogeris chrysoptera/B. cyanoptera BN BN BN 1.99 A Pyrilia caica/P. barrabandi B B Poor sampling 7.82 PS Topaza pella/T. pyra BN BN Poor sampling 7.62 A Phaethornis bourcieri whitelyi/ bourcieri B unclear B ? Phaethornis s. superciliosus/insolitus N N N 10.81 PS Thalurania f. furcata/nigrofasciata B unclear B 5.74 ? Trogon r. rufus/sulphureus B B B 1.88 A Momotus m. momota/microstephanus N N N 2.32 PS Galbula a. albirostris/chalcocephala BN or B BN B 7.67 ? Galbula d. dea/brunneiceps BN or B BN B 8.17 ? Notharchus macrorhynchos/ N. hyperrhynchus B B Poor sampling A Monasa atra/M. morphoeus BN BN BN 8.45 S Capito niger /C. auratus B B B 7.65 A Ramphastos t. tucanus/couvieri B B Poor sampling 0.32 H Ramphastos v. vitellinus/culminatus B B Poor sampling 0.73 H Selenidera culik/S. nattereri BN BN Poor sampling A Pteroglossus aracari/P. pluricinctus B B Poor sampling 2.75 H Veniliornis cassini/V. affinis BN BN BN 8.50 PS Piculus f. flavigula/magnus N N N 1.97 H Celeus undatus/C. grammicus B B B 1.95 H Celeus e. elegans/jumanus B B B 0.77 H Celeus t. torquatus/occidentalis BN BN BN 1.12 H Synallaxis rutilans dissors/confinis N N N 1.67 A Automolus infuscatus cervicalis/badius B B B 3.66 A Automolus rubiginosus obscurus/venezuelanus B B no genetic data A Xenops minutus ruficaudus/remoratus N N N 6.34 PS Dendrocincla f. fuliginosa/phaeochroa B B B 6.05 PS Dendrocincla m. merula/bartletti B B B 4.37 A Sittasomus griseicapillus axillaris/amazonus N unclear N 1.63 A Glyphorynchus s. spirurus/rufigularis B B B 8.53 PS Hylexetastes perrotii/H. stresemanni BN Poor sampling ? Dendrocolaptes c. certhia/radiolatus N N N 2.14 A Xiphorhynchus pardalotus/X. ocellatus N N N 5.26 A Xiphorhynchus guttatus polystictus/ guttatoides BN BN BN 6.36 H Lepidocolaptes a. albolineatus/duidae B B B 6.90 A Cymbilaimus l. lineatus/intermedius B B B 2.73 A
83 (Table 4.2 con’d.) Mean Replacement pattern genetic Contact Pair of taxa Integrated morphology genetic distance pattern Frederickena viridis/F. unduligera N N Poor sampling 10.06 A Thamnophilus murinus cayennensis/ murinus N unclear N 1.94 ? Thamnophilus amazonicus divaricatus/ cinereiceps B B B 4.29 A Epinecrophylla gutturalis/E. haematonota B B B 7.37 PS Myrmotherula guttata/M. hauxwelli N N Poor sampling 9.08 A Myrmotherula a. axillaris/melaena BN BN BN 1.19 H Myrmotherula menetriesii cinereiventris/ pallida B B Poor sampling PS Hypocnemis cantator/H. flavescens B B B 9.04 PS Cercomacra cinerascens immaculata/ cinerascens B B B 1.95 A Cercomacra t. saturatior/tyrannina BN BN BN 1.88 PS Percnostola r. rufifrons/minor B B B 5.36 A Schistocichla l. leucostigma/infuscata B B B 1.94 A Pithys a. albifrons/peruvianus N N N 1.26 H Gymnopithys rufigula/G. leucaspis N N N 2.29 H Willisornis p. poecilinotus/ duidae BN BN BN 10.82 H Myrmothera campanisona campanisona/ dissors B B B 2.85 A Zimmerius acer/Z. gracilipes BN BN BN 12.93 PS Hemitriccus zosterops rothschildi/zosterops B B B 1.24 A Todirostrum pictum /T. chrysocrotaphum N N N 10.18 A Tolmomyias assimilis neglectus/examinatus B B B 5.32 A Tolmomyias poliocephalus sclateri- klagesi/ poliocephalus N unclear N 2.98 A Onychorhynchus c. coronatus/castelnaui N N N 8.18 A Terenotriccus e. erythrurus/venezuelensis BN BN Poor sampling 8.15 A Phoenicircus carnifex/P. nigricollis B B Poor sampling 2.92 A Lepidothrix serena/L. coronata B B Poor sampling ? Tyranneutes virescens/T. stolzmanni B B B 11.38 PS Schiffornis t. turdinus/amazonus B B B 9.69 A Iodopleura fusca/I. isabellae BN BN Poor sampling 8.75 A Piprites chloris chlorion/tschudii N N N 1.65 H Hylophilus muscicapinus/H. hypoxanthus N N N 11.34 A Hylophilus ochraceiceps luteifrons/ ferrugineifrons B B B 13.35 PS Microcerculus b. bambla/albigularis B B B 10.14 A Pheugopedius c. coraya/griseipectus B B B 12.89 PS Microbates collaris torquatus/collaris B unclear B 2.99 A Polioptila g. guianensis/facilis B B No genetic data ? Tachyphonus s. surinamus/brevipes BN BN Poor sampling 1.52 A Tangara m. mexicana/media B B Poor sampling A no gen. Cyanerpes c. caeruleus/microrhynchus BN BN structure 0.68 A Hemithraupis f. flavicollis/aurigularis B B B 5.01 A Euphonia cayennensis/E. rufiventris B B B 6.56 PS
84
Fig. 4.2. Example of a type I replacement pattern in the Guiana Shield, where a pair of closely related taxa (Hypocnemis cantator [white dots] and H. flavescens [red dots, and red portion of tree]) replace each other at the Rio Branco and the Roraima-Rupunnuni savannas. ML phylogenetic tree shows reciprocal monophyly of the two populations. Forty other pairs share this general pattern. (Photograph by Arthur Grosset).
85
Fig. 4.3. Example of a type II replacement pattern in the Guiana Shield, where a pair of closely related taxa (Pithys a. albifrons [white dots] and Pithys a. peruvianus [red dots, and red portion of tree]) replace each other at the Rio Negro. ML phylogenetic tree shows reciprocal monophyly of the two populations. Nineteen other pairs share this general pattern. (Photographs by L.N. Naka).
86
Fig. 4.4. Example of a type III replacement pattern in the Guiana Shield, where a pair of closely related taxa (Willisornis p. poecilinotus [white dots] and Willisornis p. duidae [red dots, and red portion of tree]) replace each other in the Branco/Negro interfluvium. ML phylogenetic tree shows general reciprocal monophyly of the two populations, except one individual with eastern phenotype located to the east of the Rio Branco (Virua 6) that falls within western populations. Eighteen other pairs share this general pattern. (Photograph by L.N. Naka).
87
Fig. 4.5. Results from the local Getis-Ord G statistics. Warmer colors (red and dark orange, Standard deviation > 1.96) indicate areas where the number of phylogeographic breaks is higher than expected by chance. Light orange and yellow (-1.96 < Std < 1.65) represent areas where the number of breaks is similar to what would be expected by chance alone, whereas blue areas (std < -1.96) are areas where the number of breaks is lower than expected.
Fig. 4.6. Hotspot map of 69 phylogeographic breaks in the Guiana Shield. Warmer colors indicate higher number of breaks (gray=1-5 phylogeographic breaks, yellow= 6-15, orange=16- 30, and dark orange 30 – 67).
88 Two-dimensional null models, as implemented in BIOGEOSIM (Gotelli et al. 2007), also indicated significant spatial clustering in the distribution of phylogeographic breaks. The regression of the observed richness of breaks per cell resulted in a slope =0.052 and an r2 =
0.00002, indicating that the slope is significantly different from 1 (as would be expected if the null model fit the observed data and the simulations explained a small portion of the variation in the distribution of breaks).
To determine whether the distribution of phylogeographic breaks across the study area was a function of the area’s geometry, referred to as the mid-domain effect, I compared the observed distribution of phylogeographic breaks to the null distributions created by BioGeoSim. The number of breaks per cell (observed richness of phylogeographic breaks) was greater than all simulated expected values, with a standardized effect size of 64.72. This means the observed distribution of phylogeographic breaks in the study area is more than 64 standard deviations greater than expected by chance (observed contact zone richness index = 1.77 x 106, mean of expected index = 0.093 x 106). Likewise, the spreading dye model was able to predict only a minute portion of the distribution of phylogeographic breaks in our study (r2 = 0.052).
Location of Contact Zones
At least 31 pairs of taxa come, or are likely to come, into direct contact in the Guiana Shield
(Table 4.2). I mapped a total of 82 contact zones, where some pairs had more than one area of contact (Fig. 4.7). The Global Getis-Ord G statistic yielded positive and significant Z scores (Zg
=11.37; P<0.001) indicating that contact zones cluster spatially and were not randomly distributed throughout the landscape. Local Getis-Ord G statistics identified where these clusters are by depicting in red all regions where Zgi>1.96, and p<0.05 (Fig. 4.8).
89
Fig. 4.7. Position of 80 contact zones identified in the Guiana Shield. Each yellow circle represents a 50-km radius around a location of sympatry, a hybrid, or the mid-point between two localities, where a pair of taxa occurs at less than 100 km apart with no obvious physical barrier between them.
Fig. 4.8. Results from the local Getis-Ord G statistics for 80 contact zones. Warmer colors (red and dark orange, Standard deviation > 1.96) indicate areas where the number of contact zones is higher than expected by chance. Light orange and yellow (-1.96 < Std < 1.65) represent areas where this number is similar to what could be expected by chance alone, whereas blue areas (std < -1.96) are areas where the number of contact zones is lower than expected.
90 Two-dimensional null models also indicated a significant spatial clustering in the distribution of contact zones. The regression of observed richness of breaks per cell resulted in a slope
=2.08, r2 = 0.0098. Slope values close to 1 and high r2 results are expected if the null model fits the observed data. In this case, both values indicate that the null model explains a small portion of the variation in contact zone distribution across the study area, and do not fit the data very well. A hotspot map of contact zones shows they cluster in two particular areas, associated with the Branco and Negro’s headwaters (Fig. 4.9).
To determine whether the distribution of contact zones across the study area was a function of the area’s domain, referred to as the mid-domain effect, I compared the observed distribution of phylogeographic breaks to the null distributions created by BioGeoSim. The number of breaks per cell (observed richness of phylogeographic breaks) was greater than all simulations expected values, with a standardized effect size of 23.5. In other words, the observed distribution of phylogeographic breaks in our study area is more than 23 standard deviations greater than expected by chance (observed contact zone richness index = 1.50 x 104, mean of expected index
= 0.51 x 104).
A similar analysis to assess the effect of the mid-domain effect on the location and distribution of contact zones was performed using BIOGEOSIM and the spreading dye model.
Similar to previous results, I found that the observed number of contact zones per cell was over
21 standard deviations greater than expected by chance (observed contact zone richness index =
1.52 x 104, mean of expected index = 5.34 x 103). The spreading dye model was able to predict an even smaller portion of the observed number of contact zones per cell (r2=0.01). These results indicate that the ‘mid-domain effect’ has a small influence in the spatial patterns observed.
91
Fig. 4.9. Hotspot map of 80 contact zones in the Guiana Shield. Areas in yellow are those with a single contact zone; in orange, areas with 2-4 contact zones; and in dark red, areas with more than 4 contact zones.
Testing for Niche Divergence
Projected distributions from Maxent accurately matched known distributions based on locality points of all 10 taxa examined, and no dramatic over prediction was evident in the models (not shown). In all cases, AUC scores were consistently high (Table 4.3), and models were consistent among replicates, suggesting good performance of the selected algorithm. All pairs showed relatively low values of niche overlap, with a maximum value of 36.8% in
Hypocnemis cantator/H. flavescens (Table 4.4). Niche breadth was consistently higher for eastern than for western taxa, with the exception of both species of Epinecrophylla, which showed comparable values (Table 4.4). The largest difference within a pair was observed in the two species of Brotogeris, although this could be a function of B. cyanoptera having a smaller range in the study region. As expected, the identity test (in which pairs of taxa are lumped as one) showed consistently higher values than the observed niche overlap, showing that the 92 localities where individuals have been recorded are more similar to one another than expected under a random null distribution (Table 4.5). On the other hand, results from the background tests, which do not lump the two taxa, showed that when the environmental background of the allotaxon is taken into account, niche overlaps tend to be less similar than expected by chance.
Results of the niche divergence background test showed taxon-specific results. I found evidence of niche divergence in four of the five pairs investigated, meaning that the niches of most pairs are more different than expected using a null distribution (Table 4.4, Fig. 4.10). In only one of the pairs were these differences valid for both taxa in a pair (see Epinecrophylla gutturalis and E. haematonota). In the three other cases in which niches were different, I only found evidence of niche divergence when comparing niches of eastern taxa with western backgrounds and not when comparing western niches and eastern backgrounds (Table 4.4, Fig. 4.10).
The linear range break test showed that all pair breaks coincide with significant environmentally distinct regions (Table 4.6). This means that the ecological niches of all pairs are more different from one another than if we draw random breaks across the landscape.
Outcome of Secondary Contact
Museum specimens and field data suggest that at least 31 pairs of taxa may come into contact. There is good evidence, based on specimens with intermediate plumage (see chapter 3), that nearly half of these pairs (13) hybridize to some extent (Table 4.7). Mitochondrial data show that the mean corrected genetic distance between pairs of taxa for which hybrids have been found (0.025) is significantly lower than between those for which putative hybrids have not been found (0.084); (t-test, t= 5.01, df= 26.9, p= 0.00003) (Fig. 4.11).
93 Table 4.3. List of 10 taxa (5 pairs) used in the ENMs, including the number of localities used for each model and the Area Under Curve values (AUC), which can be interpreted as the accuracy of the model to predict occurrence data. In general, values over 0.9 represent an excellent model, whereas over 0.8, values can be considered very good. AUC values were obtained by averaging results from 25 different models. Species No. localities AUC Epinecrophylla gutturalis 120 0.89 Epinecrophylla haematonota 119 0.92 Hypocnemis cantator 140 0.87 Hypocnemis flavescens 147 0.90 Brotogeris chrysoptera 50 0.96 Brotogeris cyanoptera 44 0.83 Willisornis p. poecilinotus 195 0.86 Willisornis p. duidae 74 0.92 Pithys a. albifrons 144 0.85 Pithys a. peruvianus 23 0.96
Table 4.4. Results of niche divergence background test (one-sampled, one-tailed t-test). Mean values represent the average result of the 100 replicates to create the null distributions. In all cases, standard deviations from the mean are lower than 0.005. The upper critical value of Student’s distribution for 99 degrees of freedom is 1.66, at a 0.05 acceptance value). Niche Niche 95% conf. Obs. niche Taxon (range background) | divergence? breath Mean interval T value overlap Epinecrophylla gutturalis (haematonota) Yes 47981 0.353 0.350-0.356 17.91* 0.324 Epinecrophylla haematonota (gutturalis) Yes 46819 0.333 0.329-0.336 4.59* Hypocnemis cantator (flavescens) Yes 55716 0.409 0.406-0.413 24.09* 0.368 Hypocnemis flavescens (cantator) No 47792 0.319 0.315-0.322 -20.02 Brotogeris chrysoptera (cyanoptera) Yes 70676 0.393 0.386-0.340 16.94* 0.336 Brotogeris cyanoptera (chrysoptera) No 19511 0.324 0.320-0.329 -5.38 Willisornis p. poecilinotus (duidae) No 65850 0.320 0.314-0.326 -5.38 0.316 Willisornis p. duidae (poecilinotus) | No 32725 0.295 0.293-0.297 -19.04 Pithys a. albifrons (peruvianus) Yes 67178 0.350 0.342–0.358 8.21* 0.316 Pithys a. peruvianus (albifrons) No 24395 0.314 0.311-0.318 -0.81 * p<0.001
Table 4.5. Results of the identity test, which combines locality points form each pair of taxa in a single analysis. Mean values and 95% confidence intervals are the result of 100 replicates. Pair of taxa Mean (std Err.) 95% conf. intervals Epinecrophylla gutturalis/ haematonota 0.821 (0.0020) 0.817-0.826 Hypocnemis cantator/flavescens 0.829 (0.0016) 0.826-0.833 Brotogeris chrysoptera/ cyanoptera 0.766 (0.0036) 0.759-0.774 Willisornis p. poecilinotus/ duidae 0.836 (0.0018) 0.832-0.840 Pithys a. albifrons/ peruvianus 0.777 (0.0029) 0.772-0.783
94 Table 4.6. Results of linear range break test (one-sampled, one-tailed t-test). Mean values represent the average result of the 100 replicates to create the null distributions. The critical value for the t-test is 1.66 (99 degrees of freedom). The upper critical value of Student’s distribution for 99 degrees of freedom is 1.66, at a 0.05 acceptance value). 95% conf. Obs. niche Pair of taxa Mean (std Err.) intervals T value overlap Epinecrophylla gutturalis/ haematonota 0.432 (0.006) 0.420-0.445 16.9237 0.324 Hypocnemis cantator/flavescens 0.473 (0.005) 0.463-0.484 19.6507 0.368 Brotogeris chrysoptera/ cyanoptera 0.417 (0.007) 0.403-0.432 11.1329 0.336 Willisornis p. poecilinotus/ duidae 0.418 (0.009) 0.402-0.436 11.9346 0.316 Pithys a. albifrons/ peruvianus 0.440 (0.009) 0.421-0.459 13.2171 0.316
DISCUSSION
Location of Contact Zones and Phylogeographic Breaks
Data from this study strongly suggest that present-day barriers are important in defining the location of phylogeographic breaks, and that contact zones cluster where these barriers are easiest to overcome. I found that neither avian phylogeographic breaks nor contact zones are randomly distributed throughout the landscape in the Guiana Shield, and that their location cannot be explained by a null expectation based on geographic constraints of the study domain.
Despite using different methodologies, similar results have been reported in the Palearctic, where continental geometry had a poor explanatory power with regard to the location of contact zone hotspots (Aliabadian et al., 2005). Regardless of the results obtained in these two studies (the only ones I am aware of in testing the effect of geometric constraints of the domain in the location of contact zones), the ‘mid-domain effect’ remains as an important variable that need to be considered when analyzing spatial patterns in a geographical context (Colwell et al., 2004). It would be interesting to include within the null models an heterogeneous landscape, where cells would have different probabilities of colonization, where rivers, savannas, and other barriers could be included.
Phylogeographic breaks cluster primarily at the two biggest rivers in the region: the Negro and Branco rivers. Both rivers were shown to represent important biogeographical barriers for 95
Fig. 4.10. Test of niche divergence between pairs of taxa taking background into account. Arrows indicate observed ‘niche overlap values’, which are compared to a null distribution of background divergence. Overlap values smaller than the null distribution support niche divergence, indicated by the stars. Black stars indicate that niches of eastern taxa (black squares) are statistically different from western ranges (backgrounds), whereas white stars indicate that niches of western taxa are statistically different from eastern ranges. No stars indicate that niches are no more different than expected under a null model.
96 Table 4.7. Inference of reproductive isolation based on the presence of putative hybrids between pairs of taxa that come into actual contact within the Branco/Negro interfluvium, in relation to their pairwise mean genetic distance. Pair of taxa in contact Putative Sample size Mean corrected hybrids? E form W form genetic distance Pyrilia caica/P. barrabandi no 7 5 7.86 Phaethornis s. superciliosus/insolitus no 10 12 10.82 Momotus m. momota/microstephanus no 9 4 2.33 Monasa atra/M. morphoeus no 3 4 8.45 Ramphastos t. tucanus/couvieri yes 3 2 0.32 Ramphastos v. vitellinus/culminatus yes 1 2 0.73 Pteroglossus aracari/P. pluricinctus yes 1 1 2.76 Veniliornis cassini/V. affinis no 6 2 8.50 Piculus f. flavigula/magnus yes 9 7 1.97 Celeus undatus/C. grammicus yes 2 10 1.95 Celeus e. elegans/jumanus yes 2 10 0.77 Celeus t. torquatus/ occidentalis yes 7 2 1.12 Epinecrophylla gutturalis/haematonota no 4 6 7.37 Myrmotherula a. axillaris/melaena yes 18 6 1.19 Myrmotherula menetriesisi cinereiventris/pallida no 8 na Na Cercomacra t. tyrannina/saturatior no 17 5 1.88 Hypocnemis c. cantator/H. flavescens no 12 20 9.04 Pithys a. albifrons/peruvianus yes 16 14 1.26 Gymnopithys rufigula/G. leucaspis yes 13 9 2.29 Willisornis p. poecilinotus/duidae yes 10 23 10.82 Dendrocincla f. fuliginosa/phaeochroa no 7 16 6.05 Glyphorynchus s. spirurus/rufigularis no 6 20 8.53 Xiphorhynchus guttatus polystictus/guttatoides yes 12 7 6.36 Xenops minutus ruficaudus/remoratus no 26 5 6.34 Zimmerius acer/ Z. gracilipes no 8 12 12.93 Tyranneutes virescens/T. stolzmanni no 5 5 11.38 Hylophilus ochraceiceps luteifrons/ferrugineifrons no 7 13 13.35 Piprites chloris chlorion/tschudii yes 4 2 1.65 Pheugopedius c. coraya /griseipectus no 7 13 12.89 Euphonia cayennensis/E. rufiventris no 2 3 6.95 the avifauna (see Chapters 2 and 3), so clustering of phylogeographic breaks along them is unsurprising. Above the lower Rio Negro, pairs of taxa follow three distinctive replacement patterns, breaking either along: i) the Rio Branco and the savannas of Roraima and Venezuela, ii) the upper Rio Negro, or iii) along the Branco/Negro interfluvium. These patterns of replacement clearly mirror the geographic patterns observed in Guianan Shield endemics (see chapter 2); therefore, understanding the replacement patterns is necessary for defining the geographic limits of the Guiana area of endemism.
97
Fig. 4.11. Outcome of secondary contact in relation to the pairwise mean genetic distance between allopatric and parapatric pairs of taxa. Left column represents pairs of taxa that do not come into contact and remain allopatric throughout their distributions in the Guiana Shield (n=35); middle column represents those pairs of taxa that come into contact (sympatry or parapatry) and for which putative hybrids have been found in ornithological collections); and right column represents those pairs that come into contact and for which no apparent hybrids have been found or reported.
The first two patterns described (breaks at the Branco and Negro rivers) represent pairs of taxa for which large rivers are their main barriers, but in their headwaters, other barriers play a similar role, particularly at the Branco headwaters. There, additional barriers include: i) non- forested areas such as the Roraima-Rupunnuni savannas and the Gran Sabana; ii) a mountain chain, the Sierra de Lema; and iii) three additional rivers. i.e., the Caroni, the Caura, and the
98 Paragua (See Table 2.3, chapter 2). Thus, although 61 of the 77 pairs have breaks associated with a combination of two large rivers and other physical barriers, 16 pairs (~20% of the sample) have their breaks located within the Branco/Negro interfluvium, away from the influence of those two important rivers (although in most cases, they are still divided by the lower Rio
Negro). Within the Branco/Negro interfluvium, other putative barriers seem to be associated with the location of some breaks, including the Rio Caura and the Sierra de Parima. Therefore, physical barriers may still be important within the interfluvium. On the other hand, at least 9 of those 16 pairs do not appear to have any obvious barrier defining their ranges (Table 4.8).
Competitive exclusion may account for their replacement patterns (see below). Not surprisingly,
66% of these pairs are currently recognized as full species, as opposed to a species/subspecies ratio of ~35% in the entire dataset, suggesting a bias towards well-differentiated pairs being independent of physical barriers. Although I have previously pointed out that the taxonomic level given to a pair of replacing taxa is a subjective decision, it is also clear that, in general, species-level pairs are morphologically more differentiated compared to subspecies-level pairs.
However, this is not necessarily true at the mtDNA level (see Table 4.2).
Table 4.8. Pairs of taxa for which their phylogeographic breaks do not coincide with any obvious geographical feature. Brotogeris chrysoptera/B. cyanoptera Myrmotherula a. axillaris/melaena Monasa atra/M. morphoeus Cercomacra t. tyrannina/saturatior Selenidera culik/S. nattereri Willisornis p. poecilinotus/duidae Veniliornis cassini/V. affinis Zimmerius acer/Z. gracilipes Xiphorhynchus guttatus polystictus/guttatoides
I identified ~80 contact zones between 31 pairs of taxa, but more would probably be identified with further sampling. On the other hand, ‘potential contact zones’ are not documented contact zones, and more subtle barriers, which I have not detected, may stop pairs from coming into contact. My estimates, however, are conservative, and I predict that future
99 sampling will show an even higher number of contact zones in a larger number of pairs. Contact zones, as defined in this study, are also not randomly located throughout the landscape, and they clearly cluster at the headwaters of the Branco and Negro rivers (Figs. 4.9 and 4.10). A closer look at the hotspot contact zones shows that these areas cluster where large rivers cease to act as barriers, suggesting that headwater regions indeed play an extremely important role in the geographic and evolutionary processes in this region (see Fig. 4.12). These data suggest that the headwaters are homologous to mountain passes and foothills in temperate regions, where contact and hybrid zones seem to aggregate (Hewitt, 1999). This suggests that whereas “the challenges faced by organisms in the High Arctic and Wet Tropics are very different” as stated by Hewitt
(2001), their responses and redistribution of geographic boundaries may follow common themes: presence and absence of physical barriers.
The Role of Ecology and Niche Divergence
Although physical barriers may play a fundamental role in both the location of contact zones and phylogeographic breaks in the Guiana Shield, Amazonian biogeography often neglects the possibility that ecology may be responsible for some of the patterns. Almost three decades ago
Endler (1977) put forward the ‘environmental gradient hypothesis’, which proposes that parapatric speciation was possible across steep environmental gradients, without the need for allopatric separation of ancestral populations. Under this hypothesis, Amazonian centers of diversity should correspond to zones of relative environmental uniformity, and regions with population breaks should match zones of environmental change. This hypothesis, although often mentioned in the literature (Haffer, 1997a; Moritz et al., 2000), was rapidly discarded among
Amazonian biogeographers (e.g., Cracraft & Prum, 1988; Haffer, 1997a), but it has never been tested using the appropriate tools, such as those provided by ENMs.
100
Fig. 4.12. Detailed view of the Branco/Negro interfluvium. The two circles represent the headwaters of the Rio Branco (to the right) and the Rio Negro (to the left), which is where I found the contact zones hotspots. The region in between the Upper Orinoco (in orange) and the upper Rio Negro (in black) represents a place where several putative hybrids have been found: Yavita-Pimichim (see Table 3.4, chapter 3, Appendix D) and is the only place in Venezuela where many species otherwise limited by the Rio Negro have been found (e.g., Hylophilus hypoxanthus, Xiphorhynchus ocellatus).
Results from this study, on a very limited number of pairs, suggest that there are environmental gradients in Amazonia, and that the ecological niches of pairs of closely related taxa differ in some instances. The results of the linear range breaking test is particularly interesting, because it shows that breaks in all study cases are located in areas of environmental change, or at least, more different than 100 random breaks. Furthermore, it showed that the difference is lower (although still significant) in the two pairs of taxa that divide in regions without barriers (Brotogeris chrysoptera/B. cyanoptera and Willisornis p. poecilinotus/duidae).
Surprisingly, I found two consistent trends related to eastern (Guianan endemics) and western 101 taxa in this study system: i) niche breath of eastern taxa is generally larger than that of western taxa, and ii) the niches of eastern taxa and western backgrounds are in general significantly different, whereas western niches are not different when compared to eastern backgrounds. At first, a larger niche breath in eastern taxa seems somewhat counterintuitive, given that the
Branco/Negro interfluvium (where western taxa occur) represents one of the most environmental heterogeneous regions of the Amazon and holds some of the tallest mountains in South America outside of the Andes (the tepuis). Nonetheless, the taxa in question are forest-dwelling species, and in most cases rarely occur above 600 m. Thus, although their ranges are in very heterogeneous areas, locality points for these taxa are only in lowland terra firme forests. This may also explain why eastern taxa, when compared to western ranges, show statistically significant differences in their niches. Mountains and large tracks of white-sand forest (where neither eastern of western forms occur) are part of the ‘background’ of western taxa. Against this hypothesis, however, is the fact that a similar trend was found in Pithys albifrons, whose eastern taxon’s range (the nominate form) includes all those tall mountains mentioned above.
Further studies in many more species are necessary to evaluate the generality of these trends.
Of the five pairs investigated, the only one that showed no significant niche divergence were the two subspecies of Willisornis poecilinotus that replace one another along the Branco/Negro interfluvium and are known to hybridize on the slopes of Cerro Duida, in the Venezuelan state of
Amazonas (Zimmer, 1934). Both taxa are morphologically and genetically distinct, and are likely valid biological species. The two forms occur parapatrically in Venezuela and Brazil, without obvious barriers between their ranges (See Fig. 4.3), although distribution limits need to be fine-tuned in Brazil. That their niches, as measured by ENM, do not differ suggests that environmental variables are not responsible for the location of their phylogeographic breaks, supporting the idea that interspecific competition is defining the limits of both forms. This is
102 likely to be the case with many others pairs of taxa, particularly those that replace along the
Branco/Negro interfluvium (Table 4.6). Methodologies such as the ones employed here should help test for interspecific competition to assess the generality of this pattern.
An important issue is that ENMs present idiosyncratic problems of their own (Soberón &
Nakamura, 2009), one of which is that regional variation and environmental gradients often exaggerate ecological divergence in niche models (McCormack, 2009; Godsoe, 2010). The method developed by Warren et al. (2008) and used here takes the variation of the background data into consideration, and by creating null distributions, it allows comparisons to the observed data. On the other hand, the presence of true environmental gradients does not mean that a member of a given pair of taxa could not live in the range of the other taxon if the later were not there, which is one of the main tenets of interspecific competition and competitive exclusion.
Overall, it seems like ENMs can be useful tools to explore differences in the environments that pairs of taxa inhabit, and the results from this study suggest that environmental changes not only do occur in the study area but also seem to be able to explain, in part, the location of some of the breaks found. But the final conclusion from these data is that ecology (through competition, as seems to be the case in Willisornis poecilinotus, or through niche divergence, as apparent in the other cases) cannot be ruled out as a major player in the location of phylogeographic breaks.
Outcome of Secondary Contact
When previously isolated taxa come into secondary contact, at least three outcomes are possible, as outlined by Haffer (1969) and Mayr & O’Hara (1986). i) Geographic overlap: if the speciation process has been completed during the period of isolation, this pair may have developed sufficient differences to avoid competition and to allow them to co-occur in sympatry and remain reproductively isolated. ii) Geographic exclusion: the speciation process has not been completed, and although reproductively isolated, the taxa remain ecologically incompatible,
103 leading to competitive exclusion without hybridization. iii) Hybridization: the speciation process has not been completed and pairs of taxa interbreed when in contact. The formation of broad hybrid zones suggests no barriers to gene flow, whereas a narrow band of hybrids suggests that a certain degree of incompatibility has been reached, although not enough to halt gene flow.
These scenarios are thought to be related to the time since isolation (Mayr & O’Hara, 1986) and can be better studied in suture zones, where multiple interacting lineages share a common environmental setting and show different levels of reproductive isolation (Moritz et al., 2009).
In general, it would be expected that both pre-zygotic and post-zygotic isolation should increase with time of divergence (Coyne & Orr, 2004). A fourth possibility not discussed by Haffer
(1969) but mentioned by Mayr & O’Hara (1986) is that pairs do not come into actual contact, and remain isolated until today. In this case, levels of genetic divergence should not show any distinctive pattern.
Here, I showed that pairs of taxa for which putative hybrids have either been reported in the literature or found in ornithological collections show lower levels of mitochondrial genetic divergence than pairs of taxa that come into contact but do not show signs of hybridization. This suggests that the levels of genetic distance may be good predictors of the likelihood of hybridization in nature. As expected, pairs of taxa that do not come into contact (remain allopatric) do not show any clear pattern and present a larger variance in relation to their genetic distances (Fig. 4.11). Two of the pairs that come into contact and hybridize, Willisornis p. poecilinotus/duidae and Xiphorhynchus guttatus polystictus/guttatoides, are clear outliers, that have large levels of corrected mitochondrial divergence (10.82% and 6.36%, respectively) and are not even sister taxa (see Table 3.3, Chapter 3). The fact that these distantly related pairs are able to interbreed suggests that these levels of genetic divergence are not sufficient to inhibit hybridization. Therefore, premating isolation mechanisms, such as behavioral isolation are more
104 likely to be acting among other pairs, rather than postmating mechanism, of the sort that could affect the development of hybrid individuals (Coyne & Orr, 2004).
Correlated levels of genetic divergence and reproductive isolation have been demonstrated experimentally in Drosophila (Coyne and Orr 1989, 1997). Similarly, Foltz (1997) found that the frequency of hybridization between sea stars was negatively correlated with the divergence time inferred from mtDNA. Several over studies have shown similar trends in different groups of taxa, including frogs (Sasa et al., 1998), Lepidoptera (Presgraves, 2002), ducks (Tubaro &
Lijtmaer, 2002), birds (Price & Bouvier, 2002), and pigeons and doves (Lijtmaer et al., 2003).
These studies relied primarily on experimental crossings under controlled conditions, which have the clear advantage that hybrids can be confirmed and genotyped, and the survival and fertility of
F1 and subsequent generations can be directly assessed. In some cases (e.g. Drosophila), hybrids can be followed over the course of hundreds of generations, providing insights into the genetic mechanisms of pre- and post-zygotic isolation (Edmands, 2002). On the other hand, experimental crossings are not natural outcomes, and these experiments may tell us little about the likelihood of hybridization in natural populations.
This study represents one of the first to use exclusively natural populations, but this has resulted in several limitations: i) putative hybrids have not been confirmed by genetic means; ii) there is no data on the fertility and survival of putative hybrids; iii) there is no information of pre- or post-zygotic isolating mechanisms; and iv) genetic distances rely on a single genetic marker, which is under the stochastic effects of the coalescent (Arbogast et al., 2002; Hudson &
Turelli, 2003), and therefore, may simply represent a very gross characterization of the inferred time of isolation. Finally, I only had access to those putative hybrid individuals that overcame any possible pre-zygotic isolation mechanism but had the misfortune to be collected in the field and deposited in an ornithological collection to which I had access. Therefore, I expect that a
105 larger number of pairs will be found to interbreed. Furthermore, data on contact zones are so limited in the Amazon basin that we don’t even know whether these represent isolated cases of hybridization or samples of broad hybrid zones. Given these limitations, it is surprising that such a clear pattern emerges at all, and hopefully these results will encourage future studies on individual cases, and open new lines of research in the Amazon basin.
Location of Suture Zones and the Diversification Process
Understanding the location of contact zones, phylogeographic breaks, and patterns of hybridization in natural avian populations is also relevant to the debate of both general modes of speciation and Amazonian models of diversification. Although speciation by geographic isolation is the most accepted mode of vertebrate diversification (Mayr, 1942; 1963; Coyne &
Orr, 2004; Price, 2008), considerable debate has surrounded the relative importance of allopatric, parapatric, and sympatric speciation (Lynch, 1989; Chesser & Zink, 1994; Barraclough &
Vogler, 2000; Losos & Glor, 2003; Graham et al., 2004). Results from this chapter agree with one of the main predictions of allopatric speciation proposed by Coyne & Orr (2004) that there should be “geographic concordance of species borders with existing geographic or climatic barriers”. The concordance of significant genealogical partitions across multiple co-distributed taxa has been interpreted as the result of shared historical biogeographic factors in shaping intraspecific phylogenies (Avise, 2000), but also that there is a specific geographic cause for the break (Patton et al., 1998; Avise, 2000; Riddle et al., 2008). As pointed out in the previous chapter, however, it seems highly unlikely that species separated by the Rio Branco are the result of common vicariant events. Therefore, it seems possible that geographical barriers are important meeting points of previously isolated populations. This supports the idea that phylogeographic breaks in multiple species may occur despite the lack of complete isolation between populations, and would tend to cluster in regions of poor habitat quality, ecological
106 gradients, or reduced dispersal (Irwin, 2002). Rivers and other geographical barriers certainly represent areas where dispersal is reduced, and the presence of an ecologically similar allotaxon on the other side of the barrier is likely to stabilize the breaks (Case & Taper, 2005).
The location of suture zones has also played an important role in the discussion of
Amazonian models of diversification. Supporters of the ‘refuge hypothesis’ suggested that contact zones should cluster between putative refugia (Haffer, 1969; Simpson & Haffer, 1987) and found that suture zones in birds (Haffer, 1969; 1974; 1985; 1987), plants (Prance, 1973), lizards (Vanzolini, 1973), and butterflies, were highly concordant (Brown et al., 1974; Simpson
& Haffer, 1978); these patterns were used to help delimit ancestral refugia. That these putative ancestral refugia were then proposed to be useful to delimit priority areas for conservation
(Prance, 1977; Gentry, 1979; Myers, 1982) illustrates the serious practical implications that the location of contact zones may have outside the realm of historical biogeography.
These approaches have been criticized in the past, particularly by Beven and co-workers
(1984), who argued that concordance in the location of refugia for different taxa was not higher than expected by chance, and that there was no evidence that suture zones were not defined by present-day barriers. In a similar fashion, Endler (1977, 1982a) and later Tuomisto &
Ruokolainen (1997) proposed that environmental variables and current ecological conditions could also delimit suture zones, independent of any historical events, such as the persistence of populations in forest refugia. More recently, Whinnett et al. (2005) showed that levels of genetic divergence across a butterfly suture zone in eastern Peru did not support expectations of simultaneous divergence and suggested that ecotones and partial environmental barriers could result in clusters of suture zones in particular regions, despite lineages having independent histories. Here, despite showing the clear importance of physical barriers, I showed that using a limited number of pairs I cannot reject the hypothesis that current environmental variables may
107 coincide also with the location of phylogeographic breaks. In summary, it is clear that the geographic location of contact zones and phylogeographic breaks are related to the presence and absence of physical barriers, but environmental variables may also be responsible in creating the distribution patterns we see today.
108 CHAPTER 5 CONCLUSIONS
This study took advantage of an extraordinary biogeographical setting to study avian distribution patterns in an Amazonian suture zone, using multiple-pairs of taxa representing many independent lineages. Each of the three research chapters used new approaches to address long-standing questions in avian biogeography and speciation, and many of the issues investigated were explored for the very first time in Amazonia.
In the first research chapter (Chapter 2), I investigated avian distribution patterns in the
Guiana Shield. I identified the avian taxa endemic to the lowland terra firme forests of the
Guiana Shield, and performed an analysis of their distributions to define the boundaries of the
Guianan area of endemism. This represents the first attempt since Haffer (1969, 1974), Muller
(1973), and Cracraft (1985) to quantify the distributional patterns of an entire endemic avifauna.
I characterized the Guiana area of endemism by compiling a large distributional database and using a combination of multivariate statistics, spatial analyses, and GIS mapping tools in a quantitative and replicable way that can be applied to any area of endemism.
The Guianan area of endemism has been considered a biogeographical unit since Wallace
(1852). Its boundaries have been delimited traditionally by the Amazon River to the south and the Rio Negro to the southwest. Its western boundary, however, has remained poorly defined until now. The ambiguous delimitation of the western edge can be explained by the fact that
Guianan endemics do not show a single distribution, but follow at least three distinct patterns, which appear to be correlated with the presence of different physical barriers. The Rio Branco, and its associated savannas, represent the main barriers for avian dispersal in the Guiana Shield apart from the Rio Negro, and correspond to the most natural boundary for the Guianan area of endemism to the west. This configuration is consistent with the one suggested by Cracraft
109 (1985), but not with other authors who often neglected this river (Bates et al., 1998; Eberhard &
Bermingham, 2005; Silva et al., 2005). Besides the main rivers in the region (the Negro and the
Branco), I identified eight additional biogeographical barriers for birds, including smaller rivers, non-forested areas (savannas), and mountain ranges. Finally, I found that the area west of the
Rio Branco and east of the upper Negro (the Branco/Negro interfluvium) holds both eastern and western elements, and should not be included within the Guianan area of endemism. I suggest that this area is best described as a transition zone between two distinct avifaunas.
The results of this study suggest that, in contrast to other Amazonian areas of endemism, large rivers alone do not fully explain avian distribution patterns in the Guiana Shield.
Therefore, the longstanding characterization of Amazonia as a mosaic of parapatric areas of endemism, delimited by major rivers, is an oversimplification. This study unveiled a more complex delimitation of areas of endemism than previously recognized, argueing for a re- assessment of other Amazonian areas of endemism, which should be based on newly available distributional data, accurate descriptions of individual species patterns, and replicable methods.
In the second research chapter (Chapter 3), I analyzed the role of Amazonian rivers in generating and maintaining avian diversity by using a comparative approach to explore several aspects of the ‘riverine barrier hypothesis’ for the two most important rivers in the Guiana
Shield: the Branco and Negro Rivers. The research presented in this chapter advanced our understanding in four main aspects relevant to this topic.
First, I confirmed that the two rivers are significant biogeographical barriers to the avifauna.
I showed that phenotypically differentiated populations of birds across the rivers are also mitochondrially distinct (in most cases reciprocally monophyletic), suggesting little or no gene flow is occurring across the rivers. Second, I showed quantitatively that the lower portion of a river represents a stronger barrier to gene flow than its upper reaches. Whereas the lower Rio
110 Negro divides nearly 90 pairs of taxa, the upper Negro acts as a significant barrier for only 19 pairs. The lower portion of this river also presents higher levels of genetic divergence, but there was no evidence of genetic homogenization towards the headwaters, as has been previously suggested. This means that when only the taxa divided by the Rio Negro are considered (n=14), these show similar genetic distances across the lower Rio Negro as across the upper section of the river. Third, I demonstrated that although rivers’ headwaters represent zones of contact between otherwise allopatric populations, only a handful of populations interbreed there. Finally, coalescent simulations offered mixed support for the role of rivers as drivers of simultaneous divergence across multiple co-distributed taxa. Whereas the simulations strongly suggested that several speciation events were responsible for present diversity levels across the Rio Branco, simulations supported one or a few vicariant event for pairs across the Rio Negro.
Together, the evidence obtained in this chapter points to three main conclusions: i) rivers are important mechanism in the maintenance of species diversity by enhancing isolation and inhibiting dispersal; ii) although rivers’ headwaters are regions of contact, only a few cases there is evidence of extensive hybridization; and iii) whereas the Rio Negro may be involved in the diversification process, data strongly argue against such involment of the Rio Branco, suggesting instead that each river has its own idiosyncratic history. Generalizations are not warranted.
In the last research chapter (Chapter 4), I explored the role of physical and ecological factors in the location of contact zones and phylogeographic breaks in the large transition zone of the
Branco/Negro interfluvium. This chapter offered new insights into the processes acting upon the delimitation of species boundaries, particularly where many co-distributed avian taxa cluster along common geographical regions.
First, using spatially explicit and null models, I showed that contact zones and phylogeographic breaks are not randomly distributed across the Guiana Shield and that their
111 location cannot be explained by geographic constraints, such as the ‘mid-domain effect’. I also found that phylogeographic breaks cluster along clear physical barriers, particularly the big rivers of the region, but also along other barriers that include non-forested regions and smaller rivers whose importance has not been properly assessed. Second, by using ecological niche models and comparative null model approaches I compared niche divergence in five pairs that I used as study cases. I found that most of these pairs showed significant levels of niche divergence, and that their breaks partition the habitat into more environmentally distinct regions than expected by chance, suggesting that environmental variables cannot be ruled out as important factors in the location of suture zones. In one pair, I found that hybridizing parapatric taxa do not show significant differences in their niches, supporting the idea that biotic interactions, such as interspecific competition are likely to define their phylogeographic break.
Third, the results suggest that 31 of the 77 pairs of taxa studied come into actual contact, and either occur in sympatry or parapatry. Contact zones clustered at the headwaters of the Branco and Negro rivers (on the Branco/Negro interfluvium), supporting the transitional nature of this region. This pattern also highlights the lack of strong barriers in the clustering of contact zones.
Finally, nearly half of the pairs that come into contact hybridize to some degree, and these interbreeding pairs showed lower levels of mitochondrial genetic divergence than did pairs of taxa that come into contact but do not show any signs of hybridization.
The results from the last chapter points towards some conclusions: i) physical barriers are an important mechanism in the maintenance of species diversity by enhancing isolation and inhibiting dispersal; ii) lack of strong physical barriers allows the formation of contact zones on the rivers headwaters; iii) environmental variables cannot be ruled out as a possible mechanism in the location of suture zones; iv) genetic distance, even though calculated from a single locus, was a good predictor of the probability of hybridization in nature.
112 REFERENCES
Ab’Saber, A.N. (1997) Os dominios morfoclimaticos da America do Sul. Boletim do Instituto de Geografia da Universidade de São Paulo 52: 1-21.
Aleixo, A. (2002) Molecular systematics and the role of the ‘‘várzea’’-‘‘terra firme’’ ecotone in the diversification of Xiphorhynchus woodcreepers (Aves: Dendrocolaptidae). Auk 119: 621–640.
Aleixo, A. (2004) Historical diversification of a terra firme forest bird superspecies: a phylogeographic perspective on the role of different hypotheses of Amazonian diversification. Evolution 58: 1303-1317.
Aleixo, A. & D.F. Rossetti (2007) Avian gene trees, landscape evolution and geology: towards a modern synthesis of Amazonian historical biogeography? Journal of Ornithology 148: 443–453.
Aliabadian, M., Roselaar, C.S., Nijman, V., Sluys, R. & M. Vences (2005) Identifying contact zone hotspots of passerine birds in the Palaearctic region. Biology Letters 1: 21-23.
Anderson, E. (1948) Hybridization of the habitat. Evolution 58: 1303-1317.
Anderson, R.P., Peterson, A.T. & M. Gómez-Laverde (2002) Using niche-based GIS modeling to test geographic predictions of competitive exclusion and competitive release in South American pocket mice. Oikos 98: 3–16.
Álvarez, M., Umaña, A.M., Mejía, G.D., Cajiao, J., von Hildebrand, P. & F. Gast (2003) Aves del Parque Nacional Natural Serranía de Chiribiquete, Amazonia-Provincia de la Guyana, Colombia. Biota Colombiana 4: 49-63.
Araújo, M.B. (2002) Biodiversity hotspots and zones of ecological transition. Conservation Biology 16: 1662–1663.
Araújo, M.B. & P.H Williams (2001) The bias of complementarity hotspots toward marginal populations. Conservation Biology 15: 1710–1720.
Arbogast, B.S. & G.J. Kenagy (2001) Comparative phylogeography as an integrative approach to historical biogeography. Journal of Biogeography 28:819-825.
Arbogast, B.S., S.V. Edwards, J. Wakeley, P. Beerli, & J.B. Slowinski (2002) Estimating divergence times from molecular data on phylogenetic and population genetic timescales. Annual Review of Ecology And Systematics 33: 707-740.
Armenta, J.K., Weckstein, J.D. & D.F. Lane (2005) Geographic variation in mitochondrial DNA sequences of an Amazonian nonpasserine: The Black-spotted Barbet complex. Condor 107:527- 536.
Ávila-Pires, T.C.S. (1995) Lizards of Brazilian Amazonia (Reptilia: Squamata). Zoologische Verhandelingen 299: 1–706.
113
Avise, J.C. (2000) Phylogeography: the history and formation of species. Cambridge, Massachusetts, Harvard University Press.
Ayres, J.M.C. (1986) Uakaris and Amazonian flooded forest. PhD dissertation, University of Cambridge, Cambridge, England.
Ayres, J.M. & T.H. Clutton-Brock (1992) River boundaries and species range size in Amazonian primates. American Naturalist 140: 531-537.
Baloux, F. (2010) The worm in the fruit of the mitochondrial DNA tree. Heredity 104: 419–420.
Barbosa, R.I. (1997) Distribuição das Chuvas em Roraima. Homem, ambiente e ecologia no estado de Roraima (ed. by R.I. Barbosa, E.J.G. Ferreira, and E.G. Castellón), pp. 325-335. INPA, Manaus.
Barlow, J., Mestre, L.A.M., Gardner, T.A. & C.A. Peres (2007) The value of primary, secondary and plantation forests for Amazonian birds. Biological Conservation 136: 212–231.
Barraclough, T.G. & A.P. Vogler (2000) Detecting the geographical pattern of speciation from species-level phylogenies. The American Naturalist 155: 419–434.
Barton N.H. & G.M. Hewitt (1985) Analysis of hybrid zones. Annual Review of Ecology Evolution and Systematics 16: 113–148.
Bates H.W. (1863) The naturalist on the river Amazon. 2 vols., Murray, London.
Bates, J.M. (2001) Avian diversification in Amazonia: evidence for historical complexity and a vicariance model for a basic pattern of diversification. Diversidade biológica e cultural da Amazônia (ed. By I. Vieira, M.A. D’Incão, J.M.C. Silva and D.C. Oren), pp. 119-138. Museu Paraense Emílio Goeldi, Belem.
Bates, J.M., Hackett, S.J. & J. Cracraft (1998) Area-relationships in the Neotropical lowlands: an hypothesis based on raw distributions of passerine birds. Journal of Biogeography 25: 783-793.
Bates, J. M., J. Haffer, & E. Grismer (2004) Avian mitochondrial DNA sequence across a headwater stream of the Rio Tapajós, a major Amazonian river. Journal for Ornithology 145: 199-205.
Beaumont, M.A., Zhang, W. & D.J. Balding (2002) Approximate Bayesian computation in population genetics. Genetics 162: 2025-2035.
Bermingham, E. & C. Moritz (1998) Comparative phylogeography: concepts and applications. Molecular Ecology 7: 367–369.
Beven, S., Connor, E.F. & K. Beven (1984) Avian biogeography in the Amazon basin and the biological model of diversification. Journal of Biogeography 11: 383–399.
Bernard, E. (ed.) (2008) Inventários Biológicos Rápidos no Parque Nacional Montanhas do 114 Tumucumaque, Amapá, Brasil. RAP Bulletin of Biological Assessment 48. Conservation International, Arlington, VA.
Birks, H.J.B. (1987) Recent methodological developments in quantitative descriptive biogeography. Annls Zool. Fenn 24: 165–178.
Borges, S.H. (2007) Análise biogeográfica da avifauna da região oeste do baixo Rio Negro, amazônia brasileira. Revista Brasileira de Zoologia 24: 919–940
Borges S.H., Cohn-Haft, M., Carvalhaes, A.M.P., Henriques, L.M., Pacheco, J.F. & A. Whittaker (2001) Birds of Jaú National Park, Brazilian Amazon: species check-list, biogeography and conservation. Ornitologia Neotropical 12: 109-140.
Brooks, T.M., Hannah, L., Fonseca, G.A.B. & R.A. Mittermeier (2001) Prioritizing hotspots, representing transitions. Trends in Ecology & Evolution 16: 673.
Brown, K.S. (1982) Paleoecology and regional patterns of evolution in Neotropical forest butterflies. Biological diversification in the tropics (ed. by G.T. Prance), pp. 255-308. Columbia University Press, New York.
Brown, K.S., Jr & A.N. Ab'Saber (1979) Ice-age forest refuges and evolution in the neotropics: correlation of paleoclimatological, geomorphological and pedological data with modern biological endemism. Paleoclimas 5: 1-30.
Brown, K.S. & G.T. Prance (1987) Soils and vegetation. Biogeography and Quaternary history in tropical America (ed. by T.C. Whitmore and G.T. Prance), pp. 19–45. Clarendon Press, Oxford, United Kingdom.
Brown, K.S., Sheppard, P.M. & J.R.G. Turner (1974) Quaternary refugia in tropical America: evidence from the race formation in Heliconius butterflies. Proceedings of the Royal Society of London B 187: 369-378.
Brumfield, R.T., Beerli, P. Nickerson, D.A. & S.V. Edwards (2003) The utility of single nucleotide polymorphisms in inferences of population history. Trends in Ecology & Evolution 18: 249-256.
Brumfield, R.T, Tello, J.G., Cheviron, Z.A., Carling, M.D., Crochet, N. & K.V. Rosenberg (2007) Phylogenetic conservatism and antiquity of a tropical specialization: Army-ant-following in the typical antbirds (Thamnophilidae). Molecular Phylogenetics and Evolution 45: 1–13.
Buermann, W., Saatchi, S., Smith, T.B., Zutta, B.R., Chaves, J.A., Mila, B. & C.H. Graham (2008) Predicting species distributions across the Amazonian and Andean regions using remote sensing data. Journal of Biogeography 35: 1160–1176.
Burney, C.W. (2009) Comparative phylogeography of neotropical birds. PhD. Dissertation. Dept. of Biological Sciences, Louisiana State University, Baton Rouge.
Burney, C.W. & R.T. Brumfield (2009) Ecology predicts levels of genetic differentiation in Neotropical birds. The American Naturalist 174: 358-368. 115
Bush, M.B. (1994) Amazonian speciation: a necessary complex model. Journal of Biogeography 21: 5-17.
Cadena, C.D., Álvarez, M., Parra, J.L., Jiménez, I., Mejía, C.A., Santamaria, M., Franco, A.M., Botero, C.A. Mejia, G.D., Umaña, A.M., Calixto, A., Aldana, J. & G. Londoño (2000) The birds of CIEM, Tinigua National Park, Colombia: an overiew of 13 years of ornithological research. Cotinga 13: 46-54.
Campbell, K.E., Frailey, C.D. & L. Romero-Pittman (2006) The pan-Amazonian Ucayali Peneplain, late Neogene sedimentation in Amazonia, and the birth of the modern Amazon River system. Palaeogeography, Palaeoclimate, Palaeoecology 239: 166–219.
Capparella, A.P. (1988) Genetic variation in neotropical birds: implications for the speciation process. Acta Congressus Internationalis Ornithologici 19: 1658-1664.
Capparella, A.P. (1991) Neotropical avian diversity and riverine barriers. Acta Congressus Internationalis Ornithologici 20: 307–316.
Carnaval, A., Hickerson, M.J., Haddad, C.F.B., Rodrigues, M.T. & C. Moritz (2009) Stability predicts genetic diversity in the Brazilian Atlantic Forest Hotspot. Science 323: 785–789.
Carneiro Filho, A., Schwartz, D., Tatumi, S.H. & T. Rosique (2002) Amazonian palaeodunes provide evidence of drier climate phases during the late Pleistocene–Holocene. Quaternary Research 58: 205–209.
Carstens, B.C., Brunsfeld, S.J., Demboski, J.R., Good, J.M. & J. Sullivan (2005). Investigating the evolutionary history of the Pacific Northwest mesic forest ecosystem: hypothesis testing within a comparative phylogeographic framework. Evolution 59: 1639–1652.
Case, T.J., & M.L. Taper (2000) Interspecific competition, environmental gradients, gene flow, and the coevolution of species’ borders. American Naturalist 155: 583–605.
Case, T.J., R.D. Holt, McPeek, M.A. & T.H. Keitt (2005) The community context of species’ borders: ecological and evolutionary perspectives. Oikos 108: 28–46.
Chesser, R.T. & R.M. Zink (1994) Mode of speciation in birds: a test of Lynch’s method. Evolution 48: 490–497.
Cheviron, Z.A., Hackett, S.J. & A.P. Capparella (2005) Complex evolutionary history of a Neotropical lowland forest bird (Lepidothrix coronata) and its implications for historical hypotheses of the origin of Neotropical avian diversity. Molecular Phylogenetics and Evolution 36: 338-357.
Cintra, R., Sanaiotti, T.M. & M. Cohn-Haft (2007) Spatial distribution and habitat of the Anavilhanas Archipelago bird community in the Brazilian Amazon. Biodiversity and Conservation 16: 313-336.
116 Cohn-Haft, M. (2000) A case study in Amazonian biogeography: vocal and DNA-sequence variation in Hemitriccus flycatchers. PhD. Dissertation. Dept. of Biological Sciences, Louisiana State University, Baton Rouge.
Cohn-Haft, M., Whittaker, A. & P.C Stouffer (1997) A new look at the “species-poor” Central Amazon: The avifauna north of Manaus, Brazil. Ornithological Monographs 48: 205-235.
Cohn-Haft, M., Pacheco, A.M.F., Bechtoldt, C.L., Torres, M.F.N.M., Fernandes, A.M., Sardelli, C.H. & I.T. Macêdo (2007) Capítulo 10. Inventário ornitológico. Biodiversidade do Médio Madeira: Bases científicas para propostas de conservação (ed by L. Rapp Py-Daniel, C.P. Deus, A.L. Henriques, D.M. Pimpão & O.M. Ribeiro), pp. 145-178. Editora do INPA: Manaus.
Cohn-Haft, M., Naka, L.N. & A.M. Fernandes (2008) Padrões de distribuição da avifauna da várzea dos rios Solimões e Amazonas. Conservação da várzea: identifcação e caracterização de regiões biogeográficas (ed. by A.L.K.M. Albernaz), pp. 287-323. Ibama/ProVárzea, Manaus.
Colinvaux, P.A. (1997) An arid Amazon? Trends in Ecology & Evolution 12: 318–319.
Colinvaux, P.A., de Oliveira, P.E. & M.B. Bush (2000) Amazonian and neotropical plant communities on glacial time-scales: the failure of the aridity and refuge hypotheses. Quaternary Science Review 19: 141–169.
Colwell, R.K. (2000) A barrier runs through it . . . or maybe just a river. Proceedings of the National Academy of Sciences 97: 13470–13472.
Colwell, R.K. & G.C. Hurtt (1994) Nonbiological gra- dients in species richness and a spurious Rapoport effect. American Naturalist 144: 570–595.
Colwell, R.K. & D.C. Lees (2000). The mid-domain effect: geometric constraints on the geography of species richness. Trends in Ecology & Evolution 15: 70–76.
Colwell, R.K., Rahbek, C. & N.J. Gotelli (2004) The mid-domain effect and species richness patterns: what have we learned so far? The American Naturalist 163: E1–E23.
Conran, J.G. (1995) Family distributions in the Liliiflorae and their biogeographical implications. Journal of Biogeography 22: 1023–1034.
Conservation International (2003) Conservation Priorities for the Guayana Shield: 2002 Consensus.
Costa, G. C., Wolfe, D.B., Shepard, J.P. Caldwell & L.J. Vitt (2008) Detecting the influence of climatic variables on species distributions: a test using GIS niche-based models along a steep longitudinal environmental gradient. Journal of Biogeography 35: 637–646.
Coyne, J.A. & H.A. Orr (1989) Patterns of speciation in Drosophila. Evolution 43: 362-381.
Coyne, J.A. & H.A. Orr (1997). Patterns of speciation in Drosophila revisited. Evolution 51: 295– 117 303.
Coyne, J. A. & H.A. Orr (2004) Speciation. Sinauer Associates, Sunderland, MA
Cracraft, J. (1985) Historical biogeography and patterns of differentiation within the South American avifauna: Areas of endemism. Ornithological Monographs, 36, 49-84.
Cracraft, J. & R.O. Prum (1988) Patterns and processes of diversification: speciation and historical congruence in some Neotropical birds. Evolution, 42, 603-620.
Crawley, M.J. (2009) The R book. John Wiley & Sons, Chichester.
Crease, A. (2009) Avian range extensions from the southern headwaters of the río Caroní, Gran Sabana, Bolívar, Venezuela. Cotinga 12: 5-19. del Hoyo, J., A. Elliot, and J. Sargatal (1992-2008) Handbook of the Birds of the World, vols. 1-13. Lynx Edicions, Barcelona.
Drummond A.J., Ashton B., Buxton S., Cheung M., Heled J., Kearse M., Moir R., Stones-Havas S., Thierer T. & A. Wilson (2010) Geneious version 4.8. Available from http://www.geneious.com.
Eberhard, J. R. & E. Bermingham (2005) Phylogeny and comparative biogeography of Pionopsitta parrots and Pteroglossus toucans. Molecular Phylogenetics and Evolution 36: 288–304.
Edmands, S. (2002) Does parental divergence predict reproductive compatibility? Trends in Ecology & Evolution 17: 520-527.
Edwards, S.V. & P. Beerli (2000) Perspective: gene divergence, population divergence, and the variance in coalescence time in phylogeographic studies. Evolution 54: 1839-1854.
Elith, J., Graham, C.H., Anderson, R.P., Dudı!k, M., Ferrier, S., Guisan, A., Hijmans, R.J., Huettmann, F., Leathwick, J.R., Lehmann, A., Li, J., Lohmann, L.G., Loiselle, B.A., Manion, G., Moritz, C., Nakamura, M., Nakazawa, Y., Overton, J.M., Peterson, A.T., Phillips, S.J., Richardson, K., Scachetti-Pereira, R., Schapire, R.E., Soberón, J., Williams, S., Wisz, M.S. & N.E Zimmermann (2006) Novel methods improve prediction of species’ distributions from occurrence data. Ecography 29: 129-151.
Endler, J.A. (1977) Geographic variation, speciation, and clines. Princeton Univ. Press, Princeton.
Endler, J.A. (1982a) Pleistocene refuges: fact or fancy? In Biological diversification in the tropics (ed. by G.T. Prance) pp. 641–657. Columbia University Press, New York.
Endler J.A. (1982b) Problems in distinguishing historical from ecological factors in biogeography. American Zoologist 22: 441–452.
Environmental Systems Research Institute (2009) ArcView 9.3.1. ESRI, Redlands, CA.
Fernandes, A.M. (2007) Filogeografia comparativa de três espécies de pássaros na bacia do Rio 118 Madeira, Amazônia, Brasil. Master’s Thesis, INPA, Manaus, Brazil.
Ferreira, E., Zuanon, J., Forsberg, B., Goulding, M. & R. Briglia-Ferreira (2007) Rio Branco: peixes, ecología e conservação. Gráfica Biblos.
Foltz, D.W. (1997) Hybridization frequency is negatively correlated with divergence time of mitochondrial DNA haplotypes in a Sea Star (Leptasterias spp.) species complex. Evolution 51: 283-288.
Franzinelli, E. & H. Igreja (2002) Modern sedimentation in the Lower Negro River, Amazonas State, Brazil. Geomorphology 44: 259–271
Gascon, C., Lougheed, S.C. & J.P. Bogart (1996) Genetic and morphological variation in Vanzolinius discodactylus: a test of the river hypothesis of speciation. Biotropica 28: 376-387.
Gascon, C., Lougheed, S.C. & J.P. Bogart (1998) Patterns of genetic population differentiation in four species of Amazonian frogs: a test of the riverine barrier hypothesis. Biotropica 30: 104- 119.
Gascon, C., Malcolm, J.R., Patton, J.L., da Silva, M.N.F., Bogart, J.P., Lougheed, S.C., Peres, C.A., Neckel, S. & P.T. Boag (2000) Riverine barriers and the geographic distribution of Amazonian species. Proceedings of the National Academy of Science 97: 13672-13677.
Gaston, K.J. (2003) The structure and dynamics of geographic ranges. Oxford Univ. Press, Oxford, U.K.
Gelman A. & J. Hill (2007) Data analysis using regression and multilevel/ hierarchical models. Cambridge University Press, New York, NY.
Getis, A. & J.K. Ord (1992) The analysis of spatial association by use of distance statistics. Geographic Analysis 24: 189-206
Gibbs, A.K. & C.N. Barron (1993) The geology of the Guiana Shield. Oxford University Press, Oxford.
Godsoe, W. (2010) Regional variation exaggerates ecological divergence in niche models. Systematic Biology 59: 298–306
Gotelli, N.J., G.L. Entsminger, C. Rahbek, R.K. Colwell, & G.R. Graves (2007) BioGeoSim. Version 1.0. Acquired Intelligence Inc. & Kesey-Bear. Available at http://www.garyentsminger.com/biogeosim/biogeosim.htm
Goulding, M., Barthem, R. & E.J.G. Ferreira (2003) The Smithsonian Atlas of the Amazon. Smithsonian Institution Press, Washington, D.C.
Graham, R.W., Lundelius, E.L., Graham, M.A., Schroeder, E.K., Toomey, R.S., Anderson, E., Barnosky, A.D., Burns, J.A., Churcher, C.S., Grayson, D.K., Guthrie, R.D., Harington, C.R.,
119 Jefferson, G.T., Martin, L.D., McDonald, H.G., Morlan, R.E., Semken, H.A., Webb, S.D., Werdelin, L. & M.C. Wilson (1996) Spatial response of mammals to late Quaternary environmental fluctuations. Science 272: 1601–1606.
Graham, C.H., Ferrier, S., Huettman, F., Moritz, C. & A.T. Peterson (2004) New developments in museum-based informatics and applications in biodiversity analysis. Trends in Ecology & Evolution 19: 497-503.
Graur D. & W. Martin (2004) Reading the entrails of chickens: molecular timescales of evolution and the illusion of precision. Trends in Genetics 20: 80–86.
Guindon S. & O. Gascuel (2003) A simple, fast, and accurate algorithm to estimate large phylogenies by maximum likelihood. Systematic Biology 52: 696–704.
Guisan, A., & N.E. Zimmermann (2000) Predictive habitat distribution models in ecology. Ecological Modelling 135: 147-186.
Hackett, S.J. (1996) Molecular phylogenetics and biogeography of tanagers in the genus Ramphocelus (Aves). Molecular Phylogenetics and Evolution 5: 368–382.
Haffer, J. (1969) Speciation in Amazonian forest birds. Science 165:131-137.
Haffer, J. (1974) Avian speciation in tropical South America. Publications of the Nuttall Ornithological Club 14.
Haffer, J. (1978) Distribution of Amazon forest birds. Bonner zool. Beiträge 29: 38-78.
Haffer, J. (1987) Quaternary history of tropical America. In Biogeography and Quaternary History in Tropical America (ed by. T.C. Whitmore and G.T. Prance) pp. 1-18. Clarendon and Oxford University Press, Oxford.
Haffer, J. (1992) On the “river effect” in some birds of southern Amazonia. Boletim do Museu Paraense Emilio Goeldi, Zoologia 8: 217-245.
Haffer, J. (1993) Time's cycle and Time's arrow in the history of Amazonia. Biogeographica 69: 15- 45.
Haffer, J. (1997a) Alternative models of vertebrate speciation in Amazonia: an overview. Biodiversity and Conservation 6: 451-476.
Haffer J (1997b) Contact zones between birds of southern Amazonia. Ornithological Monographs 48: 281–306
Haffer, J. (2001) Hypotheses to explain the origin of species in Amazonia. Diversidade biológica e cultural da Amazônia (ed. by C. Vieira, J.M.C. Silva, D.C. Oren & M.A. D’Incão), pp. 45-118 Museu Paraense Emílio Goeldi, Belém.
Haffer, J. (2002) A rare hybrid manakin (Aves, Pipridae) and the origin of vertebrate in Amazonia.
120 Rudolstadter nat. hist. Schr. Suppl. 4: 47-73.
Hall, J.P.W. & D.J. Harvey (2001) Phylogenetic revision of the Charis gynaea group (Lepidoptera: Riodinidae) with comments on historical relationships among Neotropical areas of endemism. Annals of the Entomological Society of America 94: 631-647.
Hammond, D.S. (2005a) Ancient land in a modern world. Tropical forests of the Guiana Shield (ed. By D.S. Hammond), pp. 1-14. CABI Publishing, Cambridge.
Hammond, D.S. (2005b) Biophysical features of the Guiana Shield. Tropical forests of the Guiana Shield (ed. By D.S. Hammond), pp. 15-194. CABI Publishing, Cambridge.
Harold, A.S. & M.D. Mooi (1994) Areas of endemism: definition and recognition criteria. Systematic Biology 43: 261-266.
Hausdorf, B. (2002) Units in biogeography. Systematic Biology 51: 648–652.
Hausdorf, B. & C. Hennig (2003) Biotic element analysis in biogeography. Systematic Biology 52: 717–723.
Hawkins, B.A & E.E. Porter (2003) Relative influences of current and historical factors on mammal and bird diversity patterns in deglaciated North America. Global Ecology & Biogeography 12: 475 – 481
Hayes, F. & J.N. Sewlal (2004) The Amazon River as a dispersal barrier to passerine birds: effects of river width, habitat and taxonomy. Journal of Biogeography 31: 1809–1818.
Hellmayr, C.E. (1929-1938) Catalogue of birds of the Americas. Field Museum of Natural History, Zoological Series 13.
Hellmayr, C. E. & B. Conover (1942-1949) Catalogue of birds of the Americas. Field Museum of Natural History, Zoological Series 13.
Hennig, C. & B. Hausdorf (2008). Prabclus: Functions for clustering of presence-absence and abundance data. R package version 2.1-2.
Hershkovitz, P. (1977) Living New World Monkeys (Platyrrhini) With an Introduction to Primates. University of Chicago Press, Chicago.
Hewitt, G.M. (1988) Hybrid zones-natural laboratories for evolutionary studies. Trends in Ecology & Evolution 3: 158-167.
Hewitt, G.M. (1996) Some genetic consequences of the ice ages, and their role on divergence and speciation. Biological Journal of the Linnean Society 58: 247-276.
Hewitt, G.M. (1999) Post-glacial re-colonization of European biota. Biological Journal of the Linnean Society 68: 87–112.
121 Hewitt, G.M. (2004) Genetic consequences of climatic oscillations in the Quaternary. Phil. Trans. R. Soc. B 359: 183–195.
Hickerson, M.J., Dolman, G. & C. Moritz (2006a). Comparative phylogeographic summary statistics for testing simultaneous vicariance across taxon-pairs. Molecular Ecology 15: 209–224.
Hickerson, M.J., Stahl, E. & H.A Lessios (2006b). Test for simultaneous divergence using approximate Bayesian computation. Evolution 60: 2435–2453.
Hickerson M.J., Stahl E. & N. Takebayashi (2007) msBayes: Pipeline for testing comparative phylogeographic histories using hierarchical approximate Bayesian computation. BMC Bioinformatics 8:268.
Hickerson, M.J. & C. Meyer (2008) Testing comparative phylogeographic models of marine vicariance and dispersal using a hierarchical Bayesian approach. BMC Evolutionary Biology 8: 322.
Hickerson, M.J., Carstens, B.C., Cavender-Bares, J., Crandall, K.A., Graham, C.H., Johnson, J.B. Rissler, L., Victoriano, P.F. & A.D. Yoder (2009) Phylogeography’s past, present, and future: 10 years after Avise, 2000. Molecular Phylogenetics and Evolution 54: 291-301.
Hijmans, R.J., Cameron, S.E., Parra, J.L., Jones, P.G. & A. Jarvis (2005) Very high resolution interpolated climate surfaces for global land areas. International Journal of Climatology 25: 1965-1978.
Hilty, S. (2003) Birds of Venezuela, 2nd edn. Princeton University Press, Princeton.
Hoorn, C. & H. Vonhof (2006) Neogene Amazonia: introduction to the special issue. Journal South American Earth Science 21: 1–4.
Hoorn, C. Roddaz, M., Dino, R., Soares, E., Uba, C., Ochoa-Lozano, D. & R. Mapes (2010) The Amazonian Craton and its influence on past fluvial systems (Mesozoic-Cenozoic, Amazonia). Amazonia- landscape and species evolution: a look into the past (ed. by C. Hoorn and F.P. Wesselingh), pp. 103-122. Wiley-Blackwell, Chichester.
Hudson, R.R. & M. Turelli (2003) Stochasticity overrules the ‘three-times rule’: genetic drift, genetic draft, and coalescence times for nuclear loci versus mitochondrial DNA. Evolution 57: 182–190.
Hutchinson, G.E. (1957) Concluding remarks. Cold Spring Harbor Symp Quantitative Biol. 22: 415- 427.
Irion, G. & R. Kalliola (2010) Long-term landscape development processes in Amazonia. Amazonia- landscape and species evolution: a look into the past (ed. by C. Hoorn and F.P. Wesselingh), pp. 185-200. Wiley-Blackwell, Chichester.
Irwin, D.E. (2002) Phylogeographic breaks without geographic barriers to gene flow. Evolution 56: 2383–2394. 122
Jiménez-Valverde, A., Lobo, J.M. & J. Hortal (2008) Not as good as they seem: the importance of concepts in species distribution modelling. Biodiversity Letters 14: 885–890.
Kark, S., Allnutt, T.F., Levin, N., Manne, L.L. & P.H. Williams (2007) The role of transitional areas as avian biodiversity centres. Global Ecology and Biogeography 16: 187–196.
Kidd, D.M. & M.G. Ritchie (2006) Phylogeographic information systems: putting the geography into phylogeography. Journal of Biogeography 33: 1851–1865.
Kozak, K.H., Graham, C.H. & J.J. Wiens (2008) Integrating GIS-based environmental data into evolutionary biology. Trends in Ecology and Evolution 23: 141–148.
Latrubesse, E. (2003) The Late-Quaternary palaeohydrology of large South American river systems. Palaeohydrology: understanding global change (ed. by G. Benito, K.J. Gregory), pp. 193-212. Wiley
Latrubesse, E. & E. Franzinelli (1998) Late Quaternary alluvial sedimentation in the Upper Rio Negro Basin, Amazonia, Brazil: paleohydrological implications. Paleohydrology and Environmental Change (ed. by G. Benito, V.R. Baker and K.J. Gregory), pp. 259–271. Wiley
Latrubesse, E.M & B.W. Nelson (2001) Evidence for Late Quaternary aeolian activity in the Roraima–Guyana Region. Catena 43: 63–80.
Latrubesse, E.M. Stevaux, J.C. & R. Sinha (2005) Tropical Rivers. Geomorphology 70: 187–206.
Leache, A.D., Crews, S.C. & M.J. Hickerson (2007) Two waves of diversification in mammals and reptiles of Baja California revealed by hierarchical Bayesian analysis. Biology Letters 3: 646- 650.
Levins, R. (1968) Evolution In Changing Environments. Monographs in Population Biology, volume 2. Princeton University Press, Princeton, New Jersey, USA.
Lijtmaer, D.A., Mahler, B. & P.L. Tubaro (2003) Hybridization and postzygotic isolation patterns in pigeons and doves. Evolution 57: 1411–1418.
Linder, H.P. (2001) On areas of endemism, with an example from the African Restionaceae. Systematic Biology 50: 892-912.
Lobo, J. M, Jiménez-Valverde, A. & R. Real (2008) AUC: a misleading measure of the performance of predictive distribution models. Global Ecology and Biogeography 17: 145–151.
Long D., Drinkwater, M., Holt, B., Saatchi, S. & C. Bertoia (2002) Global ice and land climate studies using scatterometer image data, EOS, Transaction of American geophysical Union 1-15.
Losos, J.B. & R.B. Glor (200) Phylogenetic comparative methods and the geography of speciation. Trends in Ecology & Evolution 18: 220-227
123
Lougheed, S.C., Gascon, C., Jones, D.A., Bogart, J.P. & P.T. Boag (1999) Ridges and rivers: a test of competing hypotheses of Amazonian diversification using a dart-poison frog (Epipedobates femoralis). Proceedings Royal Society B- 266: 1829–1835.
Lovette, I.J. (2004) Molecular phylogeny and plumage signal evolution in a trans Andean and circum Amazonian avian species complex. Molecular Phylogenetics and Evolution 32: 512-523.
Lynch J.D. (1989) The gauge of speciation. Speciation and its consequences (ed. by D. Otte and J.A. Endler), pp. 527-533. Sinauer, Sunderlands.
Manni, F., Guérard, E. & E. Heyer (2004). Geographic patterns of (genetic, morphologic, linguistic) variation: how barriers can be detected by “Monmonier’s algorithm”. Human Biology 76: 173- 190.
Marks, B.D., Hackett, S.J. & A. Capparella (2002) Historical relationships among Neotropical lowland forest areas of endemism as determined by mitochondrial DNA sequence variation within the Wedge-billed Woodcreeper (Aves: Dendrocolaptidae: Glyphorynchus spirurus). Molecular Phylogenetics and Evolution 24:153-167.
Mast, A.R. & R. Nyffeler (2003) Using a null model to recognize significant co-occurrence prior to identifying candidate areas of endemism. Systematic Biology 52: 271–280.
Mayr, E. (1942) Systematics and the origin of species, from the viewpoint of a zoologist. Cambridge: Harvard University Press.
Mayr, E. (1963) Animal Species and Evolution. Harvard Univ. Press, Cambridge, MA.
Mayr, E. & R. J. O’Hara (1986) The biogeographic evidence supporting the Pleistocene forest refuge hypothesis. Evolution 40: 55–67.
McCarthy, E.M. (2006) Handbook of avian hybrids of the world. Oxford University Press, Oxford, UK.
McCormack, J.E., Zellmer, A.J. & L.L. Knowles (2009) Does niche divergence accompany allopatric divergence in Aphelocoma Jays as redicted under ecological speciation: insights from tests with niche models. Evolution 64: 1231-1244.
McKitrick, M.C. & R.M. Zink (1988) Species Concepts in Ornithology. The Condor 90: 1-14.
Monmonier, M.S. (1973) Maximum-difference barriers: an alternative numerical regionalization method. Geographical Analysis 5: 245–261.
Moore, W.S., Weibel, A.C. & A. Agius (2006) Mitochondrial DNA phylogeny of the woodpecker genus Veniliornis (Picidae, Picinae) and related genera implies convergent evolution of plumage patterns. Biological Journal of the Linnean Society 87: 611–624 Moritz, C., Patton, C.J., Schneider, C.J. & T.B. Smith (2000) Diversification of rainforest faunas: an integrated molecular approach. Annual Review of Ecology and Systematics 31: 533–563. 124
Moritz, C., Hoskin, C. J. MacKenzie, J.B., Phillips, B.L., Tonione, M., Silva, N., VanDerWal, J., Williams, S.E. & C.H. Graham (2009) Identification and dynamics of a cryptic suture zone in tropical rainforest. Proc. R. Soc. B 276: 1235–1244.
Morrone, J.J. (1994) On the identification of areas of endemism. Systematic Biology 43: 438-441.
Muller, P. (1973) The dispersal centers of terrestrial vertebrates in the Neotropical realm. A study in the evolution of the Neotropical Biota and its native landscape. Biogeographica 2: 1-250
Naka, L.N (2004) Structure and organization of canopy bird assemblages in central Amazonia. The Auk 121: 88–102.
Naka, L.N., Cohn-Haft, M., Mallet-Rodrigues, F., Santos, M.P.D. & M.F. Torres (2006) The Avifauna of the Brazilian state of Roraima: bird distribution and biogeography in the Rio Branco basin. Revista Brasileira de Ornitologia 14: 191-232.
Naka, L.N., Cohn-Haft, M., Whittaker, A., Mazar-Barnet, J. & M.F. Torres (2007) Avian biogeography of Amazonian flooded forest in the Rio Branco basin, Brazil. Wilson Journal of Ornithology 119: 439-449.
Naka, L.N., Cohn-Haft, M., & M.P.D Santos (2010) A avifauna de Roraima: ecologia e biogeografia na bacia do Rio Branco. In: Roraima: Homem, Ambiente e Ecologia (ed. by R.I. Barbosa & V. F. Melo) pp. 541-583. Femact, Boa Vista.
Nelson, G. & N. Platnick (1981) Systematics and biogeography: Cladistics and vicariance. Columbia Univ. Press, New York.
Niemiller, M.L., Fitzpatrick, B.M. & B.T. Miller (2008) Recent divergence with gene flow in Tennessee cave salamanders (Plethodontidae: Gyrinophilus) inferred from gene genealogies. Molecular Ecology 17: 2258– 2275.
Nores, M. (1999) An alternative hypothesis of the origin of Amazônian bird diversity. Journal of Biogeography 26: 275-285.
Nosil, P. (2008) Speciation with gene flow could be common. Molecular Ecology 17: 2103-2106.
Nyári, A.S. (2007) Phylogeographic patterns, molecular and vocal differentiation, and species limits in Schiffornis turdina (Aves). Molecular Phylogenetics and Evolution 44: 154-164.
Odum, E.P. (1971) Fundamentals of ecology, 3rd edn. Saunders, Philadelphia.
Ord, J. K., & A. Getis (1995) Local spatial autocorrelation statistics: distributional issues and an application. Geographic Analysis 27: 286–306.
Oren, D.C. & H.G. Albuquerque (1991) Priority areas for new avian collections in Brazilian Amazonia. Goeldiana Zoologia 6: 1-11.
125 Parker, T.A., III, Foster, R.B., Emmons, L.H., Freed, P., Forsyth, A.B. & B.D. Gill (1993) A biological assessment of the Kanuku Mountain Region of southwestern Guyana. Rapid Assessment Program Paper 5, Conservation International, Washington, DC.
Patten, M.A. & Smith-Patten, B. (2008) Biogeographical boundaries and Monmonier’s algorithm: a case study in the northern Neotropics. Journal of Biogeography 35: 407–416.
Patton, J.L. & M.N.F. da Silva (1998) Rivers, refuges, and ridges: the geography of speciation of Amazonian mammals. In: Endless forms: species and speciation (ed. by D.J. Howard and S.H. Berlocher) pp. 202-213. Oxford University Press, Oxford, U.K.
Patton, J.L., da Silva, M.N.F. & J.R. Malcolm (1994) Gene genealogy and differentiation among arboreal spiny rats (Rodentia: Echimyidae) of the Amazon Basin: a test of the riverine barrier hypothesis. Evolution 48: 1314-1323.
Patton, J.L., da Silva, M.N.F. & J.R. Malcolm (1996) Hierarchical genetic structure and gene low in three sympatric species of Amazonian rodents. Molecular Ecology 5: 229-238.
Patton, J.L., da Silva, M.N.F. & J.R. Malcolm (2000) Mammals of the Rio Jurua and the evolutionary and ecological diversification of Amazonia. Bull. Am. Mus. Nat. Hist. 244, 1–306.
Paynter, R.A. & M.A. Traylor (1982) Ornithological Gazetteer of Venezuela. Museum of Comparative Zoology, Cambridge.
Paynter, R.A. & M.A. Traylor (1991) Ornithological Gazetteer of Brazil, vols. 1 and 2. Museum of Comparative Zoology, Cambridge.
Paynter, R.A. Traylor, M.A., Jr & M.A. Traylor (1981) Ornithological Gazetteer of Colombia. Museum of Comparative Zoology, Cambridge.
Peres, C.A., Patton, J.L. & M.N.F. da Silva (1996) Riverine barriers and gene flow in Amazonian saddle-back tamarins. Folia Primatologica 67: 113-124.
Peters, J.L., et al., (1931-1986) Check-list of birds of the world, vols. I-XVI. Museum of Comparative Zoology, Harvard University, Cambridge.
Peters J.L., Gretes W. & K.E. Omland (2005) Late Pleistocene diver- gence between eastern and western populations of wood ducks (Aix sponsa) inferred by the ‘isolation with migration’ coalescent method. Molecular Ecology 14: 3407–3418.
Peterson, A.T. (2001) Predicting species' geographic distributions based on ecological niche modeling. Condor 103: 599-605.
Peterson, A.T., Soberón, J. & V. Sanchez-Cordero (1999) Conservatism of ecological niches in evolutionary time. Science 285: 1265-1267.
Peterson, A.T, Pape!, M. & J. Soberon (2008) Rethinking receiver operating characteristic analysis applications in ecological niche modeling. Ecological Modelling 213: 63–72.
126
Phillimore, A.B. & I.P.F. Owens (2005) Are subspecies useful in evolutionary and conservation biology. Proceedings of the Royal Society B 273: 1049–1053.
Phillips, S.J. & M. Dudik (2008) Modeling of species distributions with Maxent: new extensions and a comprehensive evaluation. Ecography 31: 161-175.
Phillips, S.J., Dudik, M. & R.E. Schapire (2004) A maximum entropy approach to species distribution modeling. In Proceedings of the 21st International Conference on Machine Learning. Banff, Canada.
Phillips, S., Robert, J. Anderson, J. & R.E. Schapire (2006) Maximum entropy modeling of species geographic distributions. Ecological Modelling 190: 231-259.
Pinto, O.M.O. (1978) Novo catálogo das aves do Brasil e lista dos exemplares que as representam no Museu Paulista. Revista Brasileira do Museu Paulista 22: 1-566.
Platnick, N.I. (1991) On areas of endemism. Australian Systematic Botany 4: 1–11.
Porter, A.H. & Johnson, N.A. (2002) Speciation despite gene flow when developmental pathways evolve. Evolution 56: 2103–2111.
Posada, D. & K.A. Crandall (1998) MODELTEST: testing the model of DNA substitution. Bioinformatics 14: 817-818.
Prance, G.T. (1973) Phytogeographic support for the theory of Pleistocene forest refuges in the Amazon basin, based on evidence from distribution patterns in Caryocaraceae, Chrysobalanaceae, Dichapetalaceae and Lecythidaceae. Acta Amazonica 3: 5-28.
Prance, G.T. (1982) Forest refuges: evidence from woody angiosperms. Biological diversification in the tropics (ed. by G.T. Prance), pp. 137–158. Columbia University Press, New York.
Presgraves, D.C. (2002) Patterns of postzygotic isolation in Lepi-doptera. Evolution 56: 1168–1183.
Price, T. (2008) Speciation in Birds. Roberts and Company Publishers, Greenwood.
Price, T. (2010) The roles of time and ecology in the continental radiation of the Old World leaf warblers (Phylloscopus and Seicercus). Phil. Trans. R. Soc. 365: 1749-1762.
Price, T.D., & M.M. Bouvier (2002) The evolution of F1 post-zygotic incompatibilities in birds. Evolution 56: 2083–2089.
Proches, S. & D.J. Marshall (2001) Global distribution patterns of non-halacarid marine intertidal mites: implications for their origins in marine habitats. Journal of Biogeography 28: 47–58.
R Development Core Team (2008). R: A language and environment for statistical computing. R Foundation for Statistical Computing. Vienna, Austria. http://www.R-project.org.
127 Racheli, L. & T. Racheli (2004) Patterns of Amazonian area relationships based on raw distributions of papilionid butterflies (Lepidoptera: Papilioninae). Biological Journal of the Linnean Society 82: 345–357.
Rahbek, C., Gotelli, N.J., Colwell, R. K., Entsminger, G L., Rangel T.F.L.V.B. & G.R Graves (2007) Predicting continental-scale patterns of bird species richness with spatially explicit models. Proceedings of the Royal Society: Biological Sciences 274: 165-174.
Remington, C.L. (1968) Suture zones of hybrid interactions between recently joined biotas. Evolutionary Biology (ed. by T. Dobzhanski, M.K. Hetch, and C. Steere), pp. 321-428. Plenum, New York.
Remsen, J.V., Jr & T.A. Parker, III (1983) Contribution of river-created habitats to bird species richness in Amazonia. Biotropica 15: 223-231.
Remsen, J.V., Jr., Cadena, C.D., Jaramillo, A., Nores, M., Pacheco, J.F., Robbins, M.B. Schulenberg, T.S., Stiles, F.G., Stotz, D.F. & K.J. Zimmer (2010) Version [4 Mar. 2010]. A classification of the bird species of South America. American Ornithologists' Union. http://www.museum.lsu.edu/~Remsen/SACCBaseline.html.
Rheindt, F.E., Norman, J.A. & L. Christidis (2008) DNA evidence shows vocalizations to be a better indicator of taxonomic limits than plumage patterns in Zimmerius tyrant-flycatchers. Molecular Phylogenetics and Evolution 48: 150–156.
Ribas, C.C., Gaban-Lima, R., Miyaki, C.Y. & J. Cracraft (2005) Historical biogeography and diversification within the Neotropical parrot genus Pionopsitta (Aves: Psittacidae). Journal of Biogeography 32: 1409–1427.
Richards, C., Carstens, B.C. & L.L. Knowles (2007) Distribution modelling and statistical phylogeography: an integrative framework for generating and testing alternative biogeographical hypotheses. Journal of Biogeography 34: 1833–1845.
Riddle, B.R., Dawson, M.N., Hadly, E.A., Hafner, D.J., Hickerson, M.J., Mantooth, S.J. & A.D. Yoder (2008) The role of molecular genetics in sculpting the future of integrative biogeography. Progress in Physical Geography 32: 173–202.
Ridgely, R.S. & G. Tudor (1994) The birds of South America, v. ii. The suboscines passerines. University of Texas Press, Austin.
Ridgely, R.S., Agro, D. & L. Joseph (2005) Birds of Iwokrama Forest. Proceedings of the Academy of natural Science of Philadelphia 154: 109–121.
Ridgely, R.S., Allnutt, T.F., Brooks, T., McNicol, D.K., Mehlman, D.W., Young, B.E. & J.R. Zook (2007) Digital Distribution Maps of the Birds of the Western Hemisphere, version 3.0. NatureServe, Arlington, Virginia, USA.
Robbins, M.B., Braun, M.J., Milensky, C.M., Schmidt, B.K., Prince, W., Rice, N.H., Finch, D.W. & B.J. O’Shea (2007) Avifauna of the upper Essequibo River and Acary Mountains, southern 128 Guyana. Ornitologia Neotropical 18: 339–368
Rodrigues, E.B. (2008) Variação morfológica, vocal e genéticaVna espécie politípica Lepidocolaptes albolineatus (Aves – Dendrocolaptidae). Master’s Thesis, MPEG, Belém, Brazil.
Ron, S.R. (2000) Biogeographic area relationship of lowland Neotropical rainforest based on raw distributions of vertebrate groups. Biology Journal of the Linnean Society 71: 379 – 402.
Saatchi , S., Nelson, B., Podest, E. & J. Holt (2000) Mapping land cover types in the Amazon Basin using 1 km JERS-1 mosaic. International Journal of Remote Sensing 21: 1201-1234.
Saatchi, S.S., Houghton, R.A., dos Santos Alvala, R.C., Soares, J.V. & Y. Yu (2007) Distribution of aboveground live biomass in the Amazon basin. Global Change Biology 13: 816-837.
Saatchi, S., Buermann, W., ter Steege, H., Mori, S. & T.B. Smith (2008) Modeling distribution of Amazonian tree species and diversity using remote sensing measurements. Remote Sensing of Environment 112: 2000-2017.
Sardelli, C.H. (2005) Variação genética e geográfica de Hemitriccus minor (Aves-Tyrannidae) na Bacia do Madeira, AM/Brasil. Master’s Thesis, INPA, Manaus, Brazil
Sasa, M.M., Chippindale, P.T. & N.A. Johnson (1998) Patterns of postzygotic isolation in frogs. Evolution 52:1811–1820.
Schaefer C.E.R. & J. Dalrymple (1995) Landscape evolution in Roraima, North Amazonia: planation, paleosols and paleoclimates. Zeitschrift fur Geomophologie 39: 1–28.
Schaefer C.E.R. & J.F. do Vale, Jr. (1997) Mudanças climáticas e evolução da paisagem em Roraima: uma resenha do Cretáceo ao recente. Homem, ambiente e ecologia no estado de Roraima (ed. by R.I. Barbosa, E.J.G. Ferreira, and E.G. Castellón), pp. 231-265. INPA, Manaus.
Schilthuizen, M. (2000) Ecotone: speciation-prone. Trends in Ecology & Evolution 15: 130–131.
Schneider, C.J., Smith, T.B., Larison, B. & C. Moritz (1999) A test of alternative models of diversification in tropical rainforests: Ecological gradients vs. rainforest refugia. Proceedings of the National Academy of Sciences 96: 13869–13873.
Schoener, T.W. (1968) Anolis lizards of Bimini: resource partitioning in a complex fauna. Ecology 49: 704-726.
Short, L. (1972) Hybridization, taxonomy and avian evolution. Ann. Missouri Botanical Garden 59: 447-453.
Sick, H. (1967) Rios e enchentes na Amazônia como obstáculo para a avifauna. Atlas do Simpósio para a biota amazônica 5: 495-520.
Silva, J. M. C. (1998) Birds of the Ilha de Maracá. Maracá: the biodiversity and environment of an Amazonian rainforest (ed. By W. Milliken and J. Ratter), pp. 211-229. John Wiley and Sons, 129 London.
Silva, J.M.C. & D.C. Oren (1996) Application of parsimony analysis of endemicity (PAE) in Amazon biogeography: an example with primates. Biology Journal of the Linnean Society 59: 427– 437.
Silva, J.M.C., Novaes, F.C. & D.C. Oren (2002) Differentiation of Xiphocolaptes (Dendrocolaptidae) across the River Xingu, Brazilian Amazonia: recognition of a new phylogenetic species and biogeographic implications. Bulletin of the British Ornithologists' Club 122: 185-194.
Silva, J.M.C., Rylands, A.B. & G. Fonseca (2005) The fate of the Amazonian areas of endemism. Conservation Biology 19: 689-694.
Silva M.N.F. & J.L. Patton (1998) Molecular phylogeography and the evolution and con- servation of Amazonian mammals. Molecular Ecology 7: 475–86.
Simpson, B. B. & J. Haffer (1978) Speciation patterns in the Amazonian forest biota. Annual Review Ecology and Systematics 9: 497–518.
Smith, T.B., Kark, S., Schneider, C.J., Wayne, R.K. & C. Moritz (2001) Biodiversity hotspots and beyond: the need for preserving environmental transitions. Trends in Ecology & Evolution 16: 431.
Snethlage, E. (1913) Über die Verbreitung der Vogelarten in Unteramazonien. Journal für Ornithologie 61: 469-539
Soberón, J., & M. Nakamura (2009) Niches and distributional areas: Concepts, methods, and assumptions. Proceedings of the National Academy of Sciences 116: 19644–19650.
Sorenson M.D., Ast, J., Dimcheff, D., Yuri, T. & D.P. Mindell (1999) Primers for a PCR-based approach to mitochondrial genome sequencing in birds and other vertebrates. Molecular Phylogenetics and Evolution 12: 105–114.
Spector, S. (2002) Biogeographic crossroads as priority areas for biodiversity conservation. Conservation Biology 16: 1480 –1487.
Stott, P. (1981) Historical plant geography. GeorgeAllen & Unwin, London.
Stotz, D.F, Lanyon, S.M., Schulenberg, T.S., Willard, D.E., Peterson, A.T. & J.W. Fitzpatrick (1997) An avifaunal survey of two tropical forest localities on the middle Rio Jiparaná, Rondônia, Brazil. Ornithological Monographs 48: 763-781.
Swenson N.G. (2008) The past and future influence of geographic information systems on hybrid zone, phylo- geographic and speciation research. Journal of Evolutionary Biology 21: 421–34
Swenson, N. G. & D. J. Howard (2004) Do suture zones exist? Evolution 58: 2391-2397.
130 Swenson, N.G. & D.J. Howard (2005) Clustering of contact zones, hybrid zones, and phylogeographic breaks in North America. The American Naturalist 166: 581-590.
Swofford, D. L. (2001) PAUP*: Phylogenetic analysis using parsimony (*and other methods), version 4.0b10. Sinauer, Sunderland, MA.
Szumik, C.A. & P.A. Goloboff (2004) Areas of Endemism: An improved optimality criterion. Systematic Biology 53: 968-977.
Thiollay, J.M. (1994) Density and rarity in an Amazonian rainforest bird community. Journal of Tropical Ecology 10: 449-481.
Tubaro, P. & D.A. Lijtmaer (2002) Hybridization patterns and the evolution of reproductive isolation in ducks. Biological Journal of the Linnean Society 77: 193–200.
Vanzolini, P.E. (1973) Paleoclimates, relief, and species multiplication in equatorial forests. Tropical Forest Ecosystems in Africa and South America: a Comparative Review (ed. by B.J. Meggers, E.S. Ayensu and W.D. Duckworth), pp. 255-258. Smithsonian Press, Washington.
Venables, W.N. & B.D. Ripley (2002) Modern Applied Statistics with S, 4th edn. Springer, New York.
Wallace, A.R. (1852) On the monkeys of the Amazon. Proceedings of the Zoological Society of London 20: 107–110.
Wallace, A.R. (1876) The geographical distribution of animals, vol. 1. Macmillan, London.
Warren, D.L, Glor, R.E & M. Turelli (2008) Environmental niche equivalency versus conservatism: quantitative approaches to niche evolution. Evolution 62: 2868–2883.
Warren,D.L., Glor, R. & M. Turelli (2010) ENMTools User Manual 1.0.
Weir, J.T., & D. Schluter (2004) Ice sheets promote speciation in boreal birds. Proceedings of the Royal Society of London Series B-Biological Sciences 271: 1881-1887.
Whinnett, A., Zimmerman, M., Willmott, K.R., Herrera, N., Mallarino, R., Simpson, F., Joron, M., Lamas, G. & J. Mallet (2005) Strikingly variable divergence times inferred across an Amazonian butterfly ‘suture zone’. Proceedings of the Royal Society of London B 272: 2525–2533.
Wiens J.J. & M.J. Donoghue (2004) Historical biogeography, ecology and species richness. Trends in Ecology & Evolution 19: 639–44.
Willard, D.E., Foster, M.S., Barrowclough, G.F., Dickerman, R.W., Cannell, P.F., Coats, S.L., Cracraft, J.L. & J.P. O’Neill (1991) The birds of Cerro de la Neblina, Territorio Federal Amazonas, Venezuela. Fieldiana: Zoology, New Series 65: 1-80.
Zimmer, J. T. (1934) Studies of Peruvian Birds Notes on hylophylax, Myrmothera, and Grallaria. American Museum Novitates 703: 1-21. 131
Zimmer, J. T. (1933-1944) Studies of Peruvian Birds. American Museum Novitates.
Zimmer, K.J. & S.L. Hilty (1997) Avifauna of a locality in the upper Orinoco drainage of Amazonas, Venezuela. Ornithological Monographs 48: 865-885.
Zink, R.M. & G.F. Barrowclough (2008) Mitochondrial DNA under siege in avian phylogenetics. Molecular Ecology 17: 2107-2121.
132 APPENDIX A LIST OF 24 GUIANA SHIELD POSSIBLE ENDEMIC TAXA NOT INCLUDED IN THE ANALYSES
Taxon Reason for exclusion Tinamus m. major Taxonomic uncertainty Crax a. alector Taxonomic uncertainty Phaethornis bourcieri whitelyi Taxonomic uncertainty Trogon m. melanurus Taxonomic uncertainty Trogon v. violaceus Taxonomic uncertainty Jacamerops a. aureus Taxonomic uncertainty Sclerurus rufigularis fulvigularis Inadequate geographic sampling Sclerurus c. caudacutus Taxonomic uncertainty Philydor e. erythrocercum Inadequate geographic sampling Deconychura stictolaema clarior Inadequate geographic sampling Dendrexetastes r. rufigula Inadequate geographic sampling Campylorhamphus p. procurvoides Taxonomic uncertainty Thamnophilus m. murinus Taxonomic uncertainty Terenura s. spodioptila Inadequate geographic sampling Formicarius analis crissalis Inadequate geographic sampling Phylloscartes virescens Inadequate geographic sampling Tolmomyias poliocephalus klagesi Taxonomic uncertainty Microbates c. collaris Taxonomic uncertainty Tachyphonus c. cristatus Taxonomic uncertainty Lanio f. fulvus Inadequate geographic sampling Tangara chilensis paradisea Taxonomic uncertainty Tangara v. velia Taxonomic uncertainty Tangara g. gyrola Taxonomic uncertainty Cyanerpes c. cyaneus Taxonomic uncertainty
133 APPENDIX B COMMUNITY COMPOSITION MATRIX DATA FROM 34 LOCALITIES FROM THE GUIANA SHIELD
Presence/absence data matrix of birds used for the MDS and Monmonier’s algorithm analyses (chapter 2). Data sources are present in Table 2.1 Species order, taxonomy, and nomenclature follow Remsen et al. (2010).
1 Taxon/Locality 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 Penelope m. marail/jacupemba 1 0 1 1 1 1 1 1 1 1 1 1 1 1 0 0 0 0 0 0 1 0 0 0 0 0 0 0 0 0 0 0 0 0
Pyrrhura p. picta 0 0 1 1 1 1 1 1 1 0 0 1 1 1 1 0 0 0 0 0 1 0 0 0 0 0 0 0 0 0 0 0 0 0
Pyrrhura melanura 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 1 1 0 1 1 1 1 1 1 1 1 1 1 1 1 1
Brotogeris c. chrysoptera 1 1 1 1 1 1 1 1 1 1 1 1 1 1 0 1 0 0 1 1 1 0 0 0 0 0 0 1 1 1 0 0 0 0
Brotogeris cyanoptera 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 1 0 0 0 0 1 1 1 1 1 0 0 0 1 1 1 1
Gypopsitta caica 1 1 1 1 1 1 1 1 1 1 1 1 0 0 1 0 0 0 0 0 1 0 0 0 0 0 0 0 0 0 0 0 0 0
Gypopsitta barrabandi 0 0 0 0 0 0 0 0 0 0 1 0 1 1 0 1 1 1 1 1 0 0 1 1 1 1 1 1 1 1 1 0 1 1
Phaethornis s. superciliosus 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 0 1 1 1 1 1 0 1 0 0 0 0 0 0 0 Phaethornis superciliosus insolitus 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 1 0 1 1 1 1 1 1 1
Topaza p. pella 1 1 1 1 1 1 1 1 1 1 0 1 0 0 1 0 1 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0
Topaza pyra 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 1 0 1 1 1 1 0 1 1 1 1 1 0
Thalurania f. furcata/fissilis 1 1 1 1 1 1 1 1 1 1 1 1 0 0 0 0 1 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 Thalurania furcata nigrofasciata 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1
Trogon r. rufus 1 1 1 1 1 1 1 1 1 1 1 1 0 0 1 0 1 0 0 0 1 0 0 0 0 0 0 0 0 0 0 0 0 0
Trogon rufus sulphureus 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 1 0 0 0 0 0 0 1 1 1 1 1 1 1 0 1 1
Momotus m. momota 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 0 1 0 0 0 0 0 0 0 0 0 Momotus momota microstephanus 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 1 0 1 1 1 1 1 1 1
Galbula a. albirostris 1 1 1 1 1 1 1 1 0 1 1 1 0 0 0 0 1 0 0 0 1 0 0 0 0 0 0 0 0 0 0 0 0 0 Galbula albirostris chalcocephala 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 1 0 1 1 1 0 1 1 1 1 1 1 1 1 1 1 1 1 1
Galbula d. dea 1 1 1 1 1 1 1 1 0 1 1 1 0 0 1 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0
Galbula dea brunneiceps 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 1 0 1 1 1 1 1 1 1 1 1 1 1 1 1 1 0 Notharchus m. macrorhynchos 0 1 1 1 1 1 1 1 1 1 1 1 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0
Notharchus hyperrhynchus 0 0 0 0 0 0 0 0 0 0 0 0 0 1 0 0 0 1 1 0 0 1 1 1 1 1 1 0 1 1 0 0 1 1
Monasa atra 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 0 1 1 1 1 1 1 1 1 0 0 0 0 0 0 0 0 0 0
Monasa morphoeus 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 1 0 1 1 1 1 1 1 1 1 1 1 1
Capito niger 1 1 1 1 1 1 1 1 1 1 1 1 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0
Capito auratus 0 0 0 0 0 0 0 0 0 0 0 0 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1
Ramphastos t. tucanus 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 0 1 1 1 0 0 0 0 0 0 0 0 0 0 0 0 0
134 1 Taxon/Locality 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34
Ramphastos tucanus couvieri 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 1 0 0 0 1 0 0 1 1 0 1 1 1 1 1 1 1 Ramphastos t. tucanus x couvieri 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 1 0 0 0 0 0 0 0 0 0 0 0
Ramphastos v. vitellinus 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 0 0 1 1 1 0 0 0 0 0 0 0 0 0 0 1 0 0 Ramphastos vitellinus culminatus 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 1 0 0 0 1 0 0 1 1 0 1 1 1 1 0 1 1 Ramphastos v. vitellinus x culminatus 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 1 0 0 0 0 0 0 0 0 0 0 0
Selenidera culik 1 1 1 1 1 1 1 1 1 1 1 1 1 0 1 0 0 0 0 0 0 0 0 1 0 0 0 1 0 0 0 1 0 0
Selenidera reindwartii 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 1 1
Selenidera nattereri 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 1 1 1 0 0 1 1 1 1 1 1 0 0 1 1 0 0 1 0
Pteroglossus viridis 0 0 0 1 1 1 1 1 0 1 1 1 1 1 1 0 1 1 1 0 1 0 1 0 0 0 0 0 0 0 0 0 0 0
Pteroglossus inscriptus 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 1 1 1 0 1 Pteroglossus aracari nigricollis 1 1 1 1 1 1 1 1 1 1 1 1 0 0 0 0 1 0 0 0 1 0 0 0 1 0 0 0 0 0 0 0 0 0
Pteroglossus pluricinctus 0 0 0 0 0 0 0 0 0 0 0 0 1 1 1 0 0 0 0 0 1 0 1 0 1 1 0 0 0 1 1 1 1 1
Veniliornis cassini 1 1 1 1 1 1 1 1 1 1 1 1 0 1 1 0 1 0 0 0 1 0 1 0 0 0 0 0 0 0 0 0 0 0
Veniliornis affinis 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 1 0 1 1 0 0 1 0 1 1 1 1 1 1 1 1 1 1 0
Piculus f. flavigula 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 0 1 1 0 0 0 1 0 0 0 0 0 0 0 0 0
Piculus flavigula magnus 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 1 0 1 0 1 1 1 1 1 1 1 1 1 Celeus u. undatus/amacurensis 1 1 1 1 1 1 1 1 1 1 1 1 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0
Celeus grammicus 0 0 0 0 0 0 0 0 0 0 1 0 1 1 1 1 1 1 1 0 1 1 1 1 1 1 1 1 1 1 1 1 1 1
Celeus e. elegans/hellmayri 1 1 1 1 1 1 1 1 1 1 1 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0
Celeus elegans jumanus 0 0 0 0 0 0 0 0 0 0 0 0 1 1 1 1 1 1 0 0 1 1 1 1 1 1 1 1 1 1 1 0 1 1
Celeus t. torquatus 1 1 1 1 1 1 1 1 1 1 1 1 1 1 0 0 0 1 0 1 1 0 0 0 0 0 0 0 0 0 0 0 0 0
Celeus torquatus occidentalis 0 0 0 0 0 0 0 0 0 0 0 0 0 0 1 0 0 0 1 0 0 1 1 0 1 1 0 1 1 1 1 0 1 1
Synallaxis rutilans dissors 0 0 0 1 1 1 0 1 0 1 1 1 1 1 1 1 1 1 0 0 1 0 1 0 0 0 0 0 0 0 0 0 0 0
Synallaxis rutilans confinis 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 1 0 1 1 0 0 0 1 1 Automolus infuscatus cervicalis 1 1 1 1 1 1 1 1 1 1 1 1 0 1 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0
Automolus infuscatus badius 0 0 0 0 0 0 0 0 0 0 0 0 1 0 1 0 1 1 1 0 1 1 1 1 1 1 1 1 1 1 1 1 0 0 Automolus rubiginosus obscurus 1 1 1 1 1 0 0 1 0 1 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 Automolus rubiginosus venezuelanus 0 0 0 0 0 0 0 0 0 0 0 0 0 0 1 0 0 0 0 0 0 0 1 0 0 0 0 0 1 0 0 0 0 0
Xenops minutus ruficaudus 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 0 1 1 1 0 1 1 1 0 1 0 0 1 0 0 0 0 0 0
Xenops minutus remoratus 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 1 1 0 1 1 1 1 1 1
Dendrocincla f. fuliginosa 1 1 1 1 1 1 1 1 1 1 1 1 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 Dendrocincla fuliginosa phaeochroa 0 0 0 0 0 0 0 0 0 0 0 0 1 1 1 1 1 1 1 0 1 1 1 1 1 1 1 1 1 1 1 1 1 1
Dendrocincla m. merula 1 1 1 1 0 0 1 1 1 1 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0
135 1 Taxon/Locality 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34
Dendrocincla merula bartletti 0 0 0 0 0 0 0 0 0 0 0 0 1 1 0 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1
Deconychura l. longicauda 1 0 1 1 1 1 1 1 0 1 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 Deconychura longicauda connectens 0 0 0 0 0 0 0 0 0 0 0 0 1 1 0 0 1 0 0 0 1 1 0 1 0 0 1 1 1 1 1 1 0 0 Sittasomus griseicapillus amazonus 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 1 0 0 1 0 0 1 1 1 0 1 1 1 1 1 1 Sittasomus griseicapillus axillaris 1 1 1 1 0 0 1 0 0 1 1 1 0 1 1 0 1 0 0 0 1 0 1 0 0 0 0 1 0 0 0 0 0 0
Glyphorynchus s. spirurus 1 1 1 1 1 1 1 1 1 1 1 1 0 1 0 0 0 0 0 0 0 0 1 0 0 0 0 0 0 0 0 0 0 0 Glyphorynchus spirurus rufigularis 0 0 0 0 0 0 0 0 0 0 0 0 1 0 1 1 1 1 1 1 1 1 0 1 1 1 1 1 1 1 1 1 1 1
Hylexetastes perrotii 1 1 1 1 1 1 1 1 1 1 1 1 0 1 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0
Hylexetastes stresemanni 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 1 0 0 1 0 1 0 1 0
Dendrocolaptes c. certhia 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 0 1 1 1 1 1 0 0 0 0 0 0 0 0 0 Dendrocolaptes certhia radiolatus 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 1 0 1 1 1 1 1 1 1
Dendrocolaptes p. picumnus 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 0 0 0 0 0 1 1 1 0 0 0 0 0 0 0 0 0 0 0 Dendrocolaptes picumnus validus 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 1 1 1 1 0 1 1 Xiphorhynchus p. pardalotus/caurensis 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 0 0 0 0 0 0 0 0 0
Xiphorhynchus ocellatus 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 1 1 1 1 1 1 1 1 0 Xiphorhynchus guttatus polystictus 0 0 0 0 0 0 1 1 1 0 1 1 1 1 1 0 1 0 0 0 1 0 1 0 0 0 0 0 0 0 0 0 0 0 Xiphorhynchus guttatus guttatoides 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 1 0 1 1 1 0 1 0 1 1 1 0 1 1 1 1 1 1 1 Xiphorhynchus guttatus connectens 1 1 1 1 1 1 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 Lepidocolaptes a. albolineatus 1 1 1 0 1 1 1 1 1 1 1 1 0 1 0 0 0 0 0 0 0 0 1 0 1 1 0 0 0 0 0 0 0 0 Lepidocolaptes albolineatus duidae 0 0 0 0 0 0 0 0 0 0 0 0 1 0 1 1 0 0 1 1 1 1 0 0 0 0 1 1 1 0 1 0 0 0
Cymbilaimus l. lineatus 1 1 1 1 1 1 1 1 1 1 1 1 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 Cymbilaimus lineatus intermedius 0 0 0 0 0 0 0 0 0 0 0 0 1 0 1 0 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1
Frederickena viridis 1 0 1 1 1 1 1 1 1 1 1 0 1 0 1 1 1 0 0 0 1 0 1 0 0 0 0 0 0 0 0 0 0 0
Frederickena unduligera 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 1 0 1 1 0 0 0 0 0 Thamnophilus amazonicus divaricatus 1 1 1 1 1 1 1 1 1 1 1 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 Thamnophilus amazonicus cinereiceps 0 0 0 0 0 0 0 0 0 0 0 0 1 0 0 1 0 1 0 0 0 1 1 1 1 1 1 1 1 0 1 1 1 1
Epinecrophylla gutturalis 1 1 1 1 1 1 1 1 1 1 1 1 0 0 1 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0
Epinecrophylla haematonota 0 0 0 0 0 0 0 0 0 0 0 0 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1
Myrmotherula guttata 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 0 1 0 1 1 1 1 1 0 1 0 0 0 0 0 0 0
Myrmotherula hauxwelli 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 1 0 0 0 0 0 1 1 1
136 1 Taxon/Locality 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34
Myrmotherula a. axillaris 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 0 0 0 1 0 0 0 0 0 0 1 1 0 0 0 0 0 Myrmotherula axillaris melaena 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 1 1 1 1 1 1 1 1 1 1 0 0 1 1 1 1 1 Myrmotherula menetriesii cinereiventris 1 1 1 1 1 1 1 1 1 1 1 1 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 1 0 0 0 0 0 0 Myrmotherula menetriesii pallida 0 0 0 0 0 0 0 0 0 0 0 0 1 1 1 1 1 1 0 1 1 1 1 1 1 1 1 0 1 1 1 1 1 1
Herpsilochmus stictocephalus 1 1 1 0 1 1 1 1 0 0 0 1 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 Herpsilochmus dorsimaculatus 0 0 0 0 0 0 0 0 0 1 1 0 1 1 1 1 1 1 0 1 1 1 1 1 1 1 1 1 1 1 1 1 1 0 Hypocnemis c. cantator/notaea 1 1 1 1 1 1 1 1 1 1 1 1 0 0 0 0 0 0 0 0 0 0 0 0 1 1 0 0 0 0 0 0 0 0 Hypocnemis cantator flavescens 0 0 0 0 0 0 0 0 0 0 0 0 1 1 1 1 1 1 1 1 1 1 1 1 0 0 1 0 1 0 1 1 1 1 Cercomacra cinerascens cinerascens 0 0 0 0 0 0 0 0 0 0 0 1 1 1 1 0 1 1 1 1 1 1 1 1 1 1 0 1 1 1 1 1 1 1 Cercomacra cinerascens immaculata 1 1 1 1 1 1 1 1 1 1 1 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 Cercomacra tyrannina tyrannina 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 1 0 1 1 1 1 0 1 1 1 0 1 1 1 Cercomacra tyrannina saturatior 1 0 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 Percnostola r. rufifrons/subcristata 1 1 1 1 1 1 0 1 1 1 1 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0
Percnostola minor 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 1 0 0 0 1 0 1 1 1 1 0 1 1 1 1 1 0
Schistocichla l. leucostigma 1 1 1 1 1 1 1 1 1 1 0 1 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 Schistocichla leucostigma infuscata 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 1 0 0 0 0 1 1 1 1 1 1 1 1 1 1 1 1 1
Myrmeciza f. ferruginea 1 1 1 1 1 1 1 1 1 1 1 1 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0
Pithys a. albifrons 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 0 1 0 0 0 0 0 0 0
Pithys albifrons peruvianus 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 1 0 1 1 1 1 1 1 1 Gymnopithys r. rufigula/pallidigula/pallida 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 0 1 0 0 0 0 0 0 0
Gymnopithys leucaspis 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 1 0 1 1 1 1 1 1 1
Willisornis p. poecilinotus 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 0 1 0 0 0 1 0 1 0 0 0 0 0 0 0 0 0 0 0 Willisornis poecilinotus duidae 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 1 0 1 1 1 0 1 0 1 1 1 1 1 1 1 1 1 1 1
Grallaria v. varia 1 0 1 1 1 1 1 0 1 1 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0
Grallaria varia cinereiceps 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 1 0 1 0 0 0
Hylopezus m. macularius 1 0 1 1 1 1 1 1 1 1 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 Hylopezus macularius diversus 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 1 0 0 0 0 0 0 0 0 1 1 1 1 0 0 0
Myrmothera c. campanisona 1 1 1 1 1 1 1 1 1 1 1 1 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 Myrmothera campanisona dissors 0 0 0 0 0 0 0 0 0 0 0 0 1 1 0 1 0 1 1 1 0 0 0 0 1 1 1 1 1 1 0 1 1 1
Conopophaga a. aurita 1 1 1 1 1 1 1 1 1 1 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0
137 1 Taxon/Locality 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 Conopophaga aurita inexpectata 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 1 0 0 1 1 1 1 0 1 1 0
Zimmerius acer 1 1 1 1 1 1 1 1 1 1 1 1 0 0 0 1 0 0 0 0 1 0 0 0 0 0 0 0 0 0 0 0 0 0
Zimmerius gracilipes 0 0 0 0 0 0 0 0 0 0 0 0 1 1 1 0 1 1 1 1 1 0 1 1 1 1 1 1 1 1 1 1 1 1 Lophotriccus vitiosus guianensis 1 1 0 0 0 1 1 1 0 1 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 Hemitriccus zosterops rothschildi 1 1 1 1 1 0 0 1 0 1 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0
Hemitriccus z. zosterops 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 1 0 1 0 1 0 0 1 1 1 1 1 1 1 1 0 0
Todirostrum pictum 1 1 1 1 0 1 1 1 1 1 1 1 1 1 1 1 0 1 1 1 1 1 1 0 1 0 0 0 0 0 0 0 0 0
Todirostrum chrysocrotaphum 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 1 0 1 1 1 0 1 0 1 Tolmomyias assimilis examinatus 1 1 1 1 1 1 1 1 1 1 1 1 0 0 1 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 Tolmomyias assimilis neglectus 0 0 0 0 0 0 0 0 0 0 0 0 1 1 0 1 1 1 0 1 1 0 1 1 1 1 1 1 1 1 1 0 0 0
Platyrinchus coronatus gumia 1 1 1 1 1 1 1 1 1 1 0 1 1 0 1 0 1 0 0 0 1 0 0 0 0 0 0 0 0 0 0 0 0 0
Platyrinchus c. coronatus 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 1 0 1 1 1 0 0 1 1 1 0 1 0
Onychorhynchus c. coronatus 1 1 1 1 1 1 1 1 0 1 1 1 1 0 1 0 1 0 0 1 1 1 0 0 1 0 0 0 0 0 0 0 0 0 Onychorhynchus coronatus castelnaui 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 1 0 0 1 1 0 0 1 1
Terenotriccus e. erythrurus 0 1 0 1 1 1 1 1 1 1 1 1 1 1 1 1 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 Terenotriccus erythrurus venezuelensis 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 1 1 1 1 1 1 1 1 1 1 1 1 1 1 0 1 1 0
Neopipo cinnamomea helenae 0 0 0 1 0 0 1 1 0 1 0 0 0 0 0 0 0 1 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0
Neopipo c. cinnamomea 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 1 0 1 1 1 0 0 1 0 1 0 1 0
Phoenicircus carnifex 1 1 1 1 1 1 1 1 1 1 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0
Phoenicircus nigricollis 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 1 0 0 1 1 0 1 1 1 1 0 1 0
Tyranneutes virescens 1 1 1 1 1 1 1 1 1 1 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0
Tyranneutes stolzmanni 0 0 0 0 0 0 0 0 0 0 0 0 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1
Lepidothrix serena 1 1 1 1 1 0 0 1 1 1 0 1 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0
Lepidothrix suavissima 0 0 0 0 0 0 1 0 0 0 0 1 0 0 1 0 0 0 0 0 0 1 0 0 0 0 0 0 0 0 0 0 0 0
Lepidothrix coronata 0 0 0 0 0 0 0 0 0 0 0 0 0 1 1 1 1 1 1 1 1 1 0 1 1 1 1 1 1 1 1 1 1 1
Chiroxiphia p. pareola 1 0 0 0 1 1 0 0 0 0 0 1 1 1 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0
Chiroxiphia pareola regina 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 1 0 1 1 1 0 0 0 0
Corapipo gutturalis 0 1 1 1 1 0 1 1 0 1 0 1 0 0 1 0 1 0 0 0 1 0 0 0 0 0 0 0 0 0 0 0 0 0 Schiffornis turdina olivacea/wallaci 1 1 1 1 1 1 1 1 1 1 1 1 0 0 0 0 1 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0
Schiffornis turdina amazona 0 0 0 0 0 0 0 0 0 0 0 0 1 0 1 0 0 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1
Iodopleura fusca 0 0 1 0 1 1 1 0 0 1 1 1 0 1 1 0 0 0 0 0 0 0 0 0 1 1 0 0 0 0 0 0 0 0
Iodopleura isabellae 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 1 1 1 0 0 0 0 0 0 1 0 1 1
Piprites chloris chlorion 1 1 1 1 1 1 1 1 1 1 1 1 0 1 1 1 1 1 1 1 1 0 0 0 1 0 0 0 0 0 0 0 0 0
Piprites chloris tschudii 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 1 1 1 1 1 1 1 1 1
138 1 Taxon/Locality 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34
Hylophilus m. muscicapinus 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 0 1 0 0 0 0 0 0 0 0 0
Hylophilus hypoxanthus 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 1 1 1 1 1 1 0 1 0 Hylophilus ochraceiceps luteifrons 1 1 1 1 1 1 1 1 1 1 0 1 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 Hylophilus ochraceiceps ferrugineifrons 0 0 0 0 0 0 0 0 0 0 0 0 1 0 1 1 1 1 1 1 1 0 1 1 1 1 1 1 1 1 0 0 1 1
Cyanocorax cayanus 1 1 1 1 1 1 1 1 1 1 1 1 0 0 1 0 1 0 0 0 1 0 0 0 0 0 0 0 0 0 0 0 0 0
Microcerculus b. bambla 1 1 1 1 1 0 1 1 1 1 1 1 0 1 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 Microcerculus bambla caurensis/albigularis 0 0 0 0 0 0 0 0 0 0 0 0 0 0 1 1 1 1 0 1 0 1 0 1 1 1 0 1 1 1 0 0 0 0
Pheugopedius c. coraya 1 1 1 1 1 1 0 1 0 1 1 0 0 0 0 1 1 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 Pheugopedius coraya caurensis 0 0 0 0 0 0 0 0 0 0 0 0 1 1 1 0 0 1 1 1 1 1 1 1 1 1 1 1 1 1 0 1 1 1 Pheugopedius coraya ridgwayi 0 0 0 0 0 0 1 0 1 0 0 1 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 Cyphorhinus a. aradus/faroensis 1 0 1 1 1 1 0 1 0 1 0 1 0 1 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 Cyphorhinus aradus transfluvialis 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 1 1 1 1 0 1 1
Polioptila g. guianensis 1 0 1 0 0 1 1 1 1 1 1 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0
Polioptila guianensis facilis 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 1 0 1 1 1 1 0 1 0 0 0 0 0
Tachyphonus s. surinamus 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 0 1 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 Tachyphonus surinamus brevipes 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 1 0 1 1 1 1 0 1 1 1 1 0 1
Tangara m. mexicana 1 0 0 1 1 1 1 1 1 1 1 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0
Tangara mexicana media 0 0 0 0 0 0 0 0 0 0 0 0 1 1 1 0 0 0 0 0 1 1 1 0 1 1 0 1 1 1 0 1 1 1
Cyanerpes c. caeruleus 1 0 1 1 1 1 0 1 0 1 1 1 0 1 1 1 1 1 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 Cyanerpes caeruleus microrhynchus 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 1 0 0 1 0 0 1 1 0 1 1 1 1 1 1 1
Hemithraupis f. flavicollis 0 0 1 1 1 1 1 1 1 1 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 Hemithraupis flavicollis aurigularis 0 0 0 0 0 0 0 0 0 0 0 0 1 1 1 0 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 0 1 1 Phaeothlypis rivularis mesoleuca 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 0 1 0 0 0 1 1 0 1 0 0 0 0 0 0 0 0 0 0
Phaeothlypis fulvicauda 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 1 0
Euphonia cayennensis 1 0 1 1 1 1 1 1 1 1 1 1 0 0 1 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0
Euphonia rufiventris 0 0 0 0 0 0 0 0 0 0 0 1 1 0 1 1 1 1 1 1 1 1 1 1 1 1 0 1 1 1 1 1 1 1
139 1 Localities
1. Brazil: Amapá, Porto Grande 2. Brazil: Amapá and Pará, Jari 3. French Guiana: Nouragues field station 4. Brazil: Amapá, Parque Nacional Montanhas do Tumucumaque 5. Suriname: Brownsberg 6. Suriname: Raleigh falls/Voltzberg 7. Guyana: Iwokrama Forest Reserve 8. Guyana: Acary Mountains 9. Guyana: West Kanukus 10. Brazil: Amazonas, Manaus 11. Brazil: Roraima, Parque Nacional do Viruá 12. Venezuela: Bolivar, Reserva Florestal Imataca 13. Brazil: Roraima, Caracaraí 14. Brazil: Roraima, Estação Ecológica Maracá 15. Venezuela: Bolivar, El Pauji 16. Brazil: Amazonas, Rio Demeni/Aracá 17. Venezuela: Bolivar, Cerro Guaiquinima 18. Brazil: Amazonas, Campina do Rio Preto 19. Brazil and Venezuela: Amazonas, Serra do Tapirapecó 20. Brazil: Amazonas, Rio Marauia 21. Venezuela: Bolivar, lower Río Caura 22. Venezuela: Amazonas, Cerro de la Neblina 23. Venezuela: Amazonas, Campamento Junglaven 24. Venezuela: Amazonas, San Carlos del Rio Negro 25. Brazil: Amazonas, São Gabriel da Cachoeira (east bank) 26. Venezuela: Amazonas, Yavita-Pimichin 27. Brazil: Amazonas, São Gabriel da Cachoeira (west bank) 28. Brazil: Amazonas, Manacapuru 29. Brazil: Amazonas, Parque Nacional do Jaú 30. Brazil: Amazonas, Reserva de Desenvolvimento Sustentavel Amanã 31. Brazil: Amazonas, Estação Ecológica Juami-Japura 32. Colombia: Vaupés, Mitu 33. Colombia: Caquetá and Guaviare, Parque Nacional Natural Serranía de Chiribiquete 34. Colombia: Meta, Parque Nacional Tinigua
140 APPENDIX C BEST-FIT MODELS AND PARAMETERS FOR ALL TAXA
Models of DNA evolution selected by the AIC test and parameters estimated in MODELTEST (Posada & Crandal, 1998) for taxa studied. Species or species pair Model Base frequencies1 Rate matrix2 Shape P inv Tinamus major TrN+I 0.3311 0.3510 0.0814 1 64.21 1 1 14.05 0.89 Brotogeris chrysopterus/B. cyanopterus TIM+I 0.3166 0.3524 0.0907 1 13.76 0.25 0.25 7.17 0.48 Pyrilia caica/P. barrabandi TrN 0.3196 0.3437 0.1069 1 62.40 1 1 38.14 - Topaza pella/T. pyra TIM+I 0.3181 0.3437 0.0978 1 160.78 6.62 6.62 91.75 0.67 Phaethornis superciliosus GTR+I 0.3295 0.3312 0.0926 1 10.63 0.26 0.26 10.63 0.65 Phaethornis bourcieri TrN 0.3054 0.3456 0.1091 1 78.12 1 1 31.56 - Thalurania furcata TrN+G 0.3194 0.3181 0.0983 1 27.56 1 1 12.62 - Trogon rufus TrN+I 0.3316 0.3353 0.1012 1 48.17 1 1 20.91 0.2001 0.77 Momotus momota TrN+I* 0.2953 0.3581 0.1058 1 58.07 1 1 21.11 0.69 Galbula albirostris HKY+I 0.2898 0.3466 0.1143 0.82 Galbula dea TrN+I* 0.3063 0.3610 0.1034 1 150.85 1 1 62.80 - Monasa atra/M. morphoeus TrN 0.2770 0.3560 0.0986 1 40.74 1 1 20.43 - Capito niger/C. auratus TrN+I 0.2928 0.3783 0.0891 1 1.4x106 1 1 1.1x105 0.81 Ramphastos tucanus F81 0.2951 0.3785 0.0868 - Ramphastos vitellinus TrN+I 0.2950 0.3838 0.0836 1 56.38 1 1 26.43 0.71 Pteroglossus aracari/P. pluricinctus TrN+I 0.2890 0.3953 0.0891 1 192.22 1 1 42.22 0.80 Veniliornis cassini/V. affinis TrN 0.2860 0.3882 0.1022 1 44.94 1 1 8.24 - Piculus flavigula TrN 0.2922 0.3873 0.0935 1 43.55 1 1 13.48 - Celeus undatus/C. grammicus TrN+I 0.3008 0.3947 0.0853 1 34.93 1 1 18.79 0.80 Celeus elegans TrN 0.2923 0.3803 0.0958 1 59.81 1 1 12.15 - Celeus torquatus HKY 0.3232 0.3454 0.0906 - Synallaxis rutilans GTR 0.3185 0.3305 0.0941 2.6x106 2.1x108 3.3x106 2.6x107 8.6x107 - Automolus infuscatus HKY 0.3232 0.3454 0.0906 - Xenops minutus HKY 0.3250 0.3172 0.0869 - Dendrocincla fuliginosa HKY 0.3179 0.3309 0.0870 - Dendrocincla merula HKY+I 0.3214 0.3170 0.0904 0.88 Sittasomus griseicapillus K81uf 0.3295 0.3312 0.0926 1 10.63 0.26 0.26 10.63 0.65 Glyphorynchus spirurus TrN 0.3227 0.3363 0.0956 1 43.49 1 1 18.00 - Dendrocolaptes certhia GTR 0.3182 0.3335 0.0998 1 8.07 0 0 8.07 - Dendrocolaptes picumnus HKY+I 0.3219 0.3181 0.0912 0.79 Xiphorhynchus pardalotus X. ocellatus HKY+I 0.3178 0.3242 0.0988 0.76 Xiphorhynchus guttatus TVM+I 0.3286 0.3249 0.0893 0.3 3.08 0.07 1.74 3.08 0.84 Lepidocolaptes albolineatus HKY+I 0.3218 0.3411 0.0888 0.89 Cymbilaimus lineatus TVM 0.3219 0.3270 0.0894 1.1x105 1x106 4.3x104 0.0001 1x106 - Frederickena viridis/F. unduligera TrN+I 0.3118 0.3335 0.0940 1 23.38 1 1 13.89 0.77 Thamnophilus murinus TrN+I 0.3166 0.3191 0.0959 1 44.40 1 1 13.54 0.73 Thamnophilus amazonicus HKY 0.3161 0.3516 0.0965 - Epinecrophylla gutturalis/E. haematonota HKY+I 0.3199 0.3399 0.0928 - 141 Species or species pair Model Base frequencies1 Rate matrix2 Shape P inv Myrmotherula guttata /M. hauxwelli GTR+I 0.3139 0.3391 0.0951 0 7.86 0.1456 0.33 2.32 0.73 Myrmotherula axillaris GTR+I 0.3180 0.3351 0.0887 1 34.47 1 1 18.09 0.82 Hypocnemis cantator/H. flavescens HKY 0.3195 0.3307 0.0900 - Cercomacra cinerascens HKY+I 0.3199 0.3287 0.0906 0.90 Cercomacra tyrannina HKY 0.3199 0.3206 0.0892 - Percnostola rufifrons HKY+I 0.3159 0.3489 0.0877 0.85 Schistocichla leucostigma TrN+I 0.3190 0.3410 0.0978 1 13.59 1 1 6.60 0.88 Pithys albifrons TrN+I 0.3122 0.3291 0.1005 1 19.81 1 1 9.67 0.84 Gymnopithys rufigula/G. leucaspis GTR+G 0.3150 0.3304 0.0999 0.19 11.51 1.41 0.46 5.60 - Willisornis poecilinotus GTR+I 0.3161 0.3516 0.0965 1 10.63 0.26 0.26 10.63 0.1907 0.65 Myrmothera campanisona K81uf+I 0.3059 0.3262 0.0945 1 531.17 45.63 45.63 531.17 0.48 Zimmerius acer/Z. gracilipes K81uf 0.2991 0.3214 0.1074 1 2.2x1010 2.1x109 2.1x109 2.2x1010 - Hemitriccus zosterops TrN+I 0.3039 0.3030 0.1057 1 49.59 1 1 16.69 0.64 Todirostrum pictum/T. chrysocrotaphum TrN 0.2875 0.3209 0.1111 1 38.44 1 1 13.28 - Tolmomyias assimilis GTR+I 0.3295 0.3312 0.0926 1 10.63 0.26 0.26 10.63 0.65 Tolmomyias poliocephalus HKY+G 0.3105 0.3224 0.0854 10.47 - Platyrinchus coronatus K81uf 0.3009 0.3396 0.1115 1 5.01 0.42 0.42 5.01 0.1527 - Onychorhynchus coronatus TIM 0.2828 0.3233 0.1107 1 23.85 0.36 0.36 10.49 0.2001 - Terenotriccus erythrurus HKY+I 0.3016 0.3079 0.1067 0.75 Tyranneutes virescens/T. stolzmanni K81uf 0.3129 0.3107 0.0950 1 5.2x1010 4.9x109 4.9x109 5.2x109 - Chiroxiphia pareola HKY 0.3077 0.3186 0.0935 - Schiffornis turdina TrN 0.3156 0.3190 0.0873 1 22.46 1 1 13.31 - Iodopleura fusca/I. isabellae TrN+I 0.3227 0.3226 0.0937 1 132.09 1 1 47.62 0.74 Piprites chloris GTR+I 0.3203 0.3353 0.1067 6.3x108 8.1x109 2.4x108 1.51 2.2x109 0.72 Hylophilus muscicapinus/H. hypoxanthus TrN+I 0.2983 0.3384 0.1174 1 23.37 1 1 9.13 0.62 Hylophilus ochraceiceps TIM+I 0.3152 0.3560 0.1027 1 9.43 0.36 0.36 4.20 0.66 Microcerculus bambla TrN+I 0.2925 0.3712 0.1078 1 25.94 1 1 12.55 0.72 Pheugopedius coraya TIM 0.3136 0.3523 0.1062 1 22.30 0.23 0.23 8.72 0.89 Micorbates collaris TrN 0.2837 0.3569 0.1136 0.48 Tachyphonus cristatus TrN 0.2863 0.3435 0.1129 1 30.13 1 1 15.40 - Tachyphonus surinamus HKY 0.2935 0.3496 0.1091 4.9x1036 0.67 Tangara chilensis TrN 0.2833 0.3774 0.1192 1 47.24 1 1 7.25 0.65 Cyanerpes caeruleus TVM 0.2926 0.3538 0.1051 0.11 5.93 0.16 1.08 5.93 - Hemithraupis flavicollis TrN 0.2941 0.3453 0.1184 1 35.92 1 1 13.44 - Euphonia cayennensis/E. rufiventris TrN+I 0.3311 0.3510 0.0814 1 64.21 1 1 14.05 0.77 1 order of base frequencies is A, C, G, T 2 Order of rate matrices AC, AG, AT, CG, CT, and GT. When a single value is given, this represents the TS/TV (transition/transverion) ratio *Given the low sample size, corrected distances with invariants resulted in artificially high values. Therefore, I used a simpler model without invariants.
142 APPENDIX D EXAMPLES OF PUTATIVE HYBRIDS FROM THE HEADWATERS
All intermediate-looking individuals come from contact zones between parental taxa, farther suggesting introgression. All specimens are from Venezuela and are housed at the Phelps Ornithological Collection, except the specimens of Willisornis poecilinotus, housed at the American Museum of Natural History (AMNH).
Pteroglossus aracari x P. Gymnopithys rufigula x leucaspis pluricinctus
Willisornis p. poecilinotus x Piculus f. flavigula x P. f. magnus W. p. duidae
143 APPENDIX E RESULTS FROM THREE INDEPENDENT JOINT POSTERIOR DENSITIES FOR THE RIO BRANCO USING ABC SIMMULATIONS
Results from three independent joint posterior densities for E(t), !, and ", using ABC simulations (1,000,000 draws form the prior, tolerance=0.0005) using similar prior parameters for 24 pairs of co-distributed taxa across the Rio Branco.
144 APPENDIX F RESULTS FROM THREE INDEPENDENT JOINT POSTERIOR DENSITIES FOR THE RIO NEGRO USING ABC SIMMULATIONS
Results from three independent joint posterior densities for E(t), !, and ", using ABC simulations (1,000,000 draws form the prior, tolerance=0.0005) using similar prior parameters for 14 pairs of co-distributed taxa across the Rio Negro.
145 APPENDIX G LIST OF 26 PAIRS OF TAXA WITH DESCRIBED TAXA REPLACEMENTS WITHIN THE GUIANA SHIELD NOT INCLUDED IN THE ANALYSES
List of 26 pairs of taxa with described phenotypic replacements within the Guiana Shield, not included in the analyses because of presenting either i) weak morphological differentiation ii) no genetic structure, or iii) insufficient distributional or genetic samples to describe general distribution patterns Pair of taxa Distributional data Phylogeographic. data Tinamus m. major/zuliensis weak morph. differentiation No genetic structure Crypturellus erythropus/duidae Poor sampling no genetic data Crax alector alector/erythrognatha Poor sampling no genetic data Psophia c. crepitans/napensis Poor sampling no genetic data Trogon m. melanurus/eumorphus weak morph. differentiation No genetic structure Trogon v. violaceus/crissalis Poor sampling Poor sampling Jacamerops a. aureus/ridwagi weak morph. differentiation No genetic structure Picumnus exilis buffoni/undulatus weak morph. differentiation Poor sampling Sclerurus r. fulvigularis/brunnescens Poor sampling Poor sampling Sclerurus caudacutus brunneus/caudacutus Poor sampling no genetic data Philydor e. erythrocercum/suboles Poor sampling no genetic data Deconychura l. longicauda/connectens Poor sampling Poor sampling Deconychura stictolaema clarior/secunda Poor sampling no genetic data Dendrocolaptes p. picumnus/validus Poor sampling Poor sampling Campylorhamphus p. procurvoides/sanus Poor sampling no genetic data Formicarius analis crissalis/zamorae Poor sampling Poor sampling Grallaria v. varia/cinereiceps Poor sampling no genetic data Hylopezus m. macularius/diversus Poor sampling no genetic data Conopophaga a. aurita/inexpectata Poor sampling Poor sampling Platyrinchus coronatus gumia/coronatus weak morph. differentiation x Chiroxiphia p. pareola/regina Poor sampling Poor sampling Cyphorhinus a. aradus/transfluvialis Poor sampling Poor sampling Tachyphonus c. cristatus/cristatellus weak morph. differentiation No genetic structure Tangara chilensis coelicolor-paradisea/chilensis weak morph. differentiation x Tangara v. velia/iridina weak morph. differentiation Poor sampling Cyanerpes c. cyaneus/dispar weak morph. differentiation Poor sampling
146 VITA
Luciano Nicolás Naka was born in December, 1973, in Buenos Aires, Argentina, to Ignacio
Naka and Sara Salem. As a young boy he developed an early interest in nature, becoming
President of the Asociación de los Animales (with three full members) at the age of 7. His interest in birds got serious in 1986, as he joined the Asociación Ornitológica del Plata. He got engaged in environmental causes and traveled throughout Argentina, from Iguazú (where he saw his first toucan) to Patagonia. During high school, Luciano and Alejandro (his lifetime best friend) devoted weekends and holidays to birdwatching, rock climbing, and hiking, developing an attraction for exotic destinations that eventually took him from the Venezuelan Andes to
Tierra del Fuego and from Morocco to Kathmandu. In 1994, his parents moved to the island of
Santa Catarina, southern Brazil, where he started his undergraduate studies in biology. Rather than sunbathing at the beautiful beaches, he spent most of his time inventorying the island’s avifauna; a research that was later published as a book. During his last year in college, he met
Carolina, the brightest and nicest girl in school, who would become his girlfriend, best friend, and life companion. As they graduated in 1999, they moved to Manaus to pursue a master’s degree in tropical ecology at the Instituto Nacional de Pesquisas da Amazônia (INPA), where he studied canopy birds. In 2002, he joined Mario Cohn-Haft at the INPA bird collection, who introduced him to scientific collecting, bird vocalizations, and Amazonian biogeography. For two years he travelled the Amazon collecting specimens, making recordings, and visiting isolated Indian tribes. In 2004, he joined the Museum of Natural Science at LSU and started his doctoral program at the Remsen-Brumfield lab. His time in Baton Rouge, in particular the year he spent with Carolina and Sofia (their little girl) was the time of his life. He defended his doctoral dissertation in July 2010. Luciano, Carolina, and Sofia live now in Boa Vista, northern
Brazil, and plan on studying the plants and birds of the Amazon basin until the end of their lives.
147