Distributional Ecology and Diversity Patterns of Tropical Montane Birds

Material Information

Distributional Ecology and Diversity Patterns of Tropical Montane Birds
Jankowski, Jill
Place of Publication:
[Gainesville, Fla.]
University of Florida
Publication Date:
Physical Description:
1 online resource (261 p.)

Thesis/Dissertation Information

Doctorate ( Ph.D.)
Degree Grantor:
University of Florida
Degree Disciplines:
Committee Chair:
Robinson, Scott K.
Committee Co-Chair:
Levey, Douglas J.
Committee Members:
Holt, Robert D.
Phelps, Steven M.
Sieving, Kathryn E.
Graduation Date:


Subjects / Keywords:
Birds ( jstor )
Climate change ( jstor )
Cloud forests ( jstor )
Congeners ( jstor )
Ecology ( jstor )
Forests ( jstor )
Modeling ( jstor )
Species ( jstor )
Vegetation ( jstor )
Vegetation structure ( jstor )
Biology -- Dissertations, Academic -- UF
abiotic, aggression, altitude, andes, asymmetric, avian, behavior, biotic, climate, cloud, community, competition, congener, conservation, costa, ecotone, egg, elevation, forest, gradient, guild, habitat, interaction, interspecific, limit, manu, model, mountain, mutualism, neotropical, nestedness, niche, overlap, peru, physiology, playback, predation, range, rarity, replacement, rica, richness, song, sorensen, species, survey, symmetric, territoriality, thermal, tilaran, tolerance, tropics, turnover, vegetation
bibliography ( marcgt )
theses ( marcgt )
government publication (state, provincial, terriorial, dependent) ( marcgt )
born-digital ( sobekcm )
Electronic Thesis or Dissertation
Zoology thesis, Ph.D.


The distributions that species exhibit along environmental gradients are geographical expressions of the range of ecological conditions that allow species persistence. My dissertation research used elevational gradients in the Peruvian Andes and in Central America to study range boundaries of birds along mountainsides and to understand implications of ecological reinforcement of range boundaries for landscape patterns of diversity. This includes description and modeling of elevational range size, position, and response shapes, community-level analysis of species turnover and composition, and experimental approaches with single species. Bird survey data collected along a 2700-m elevational gradient in Peru showed that most species exhibited symmetric response curves with elevation; however, 16% of species showed asymmetric response curves. Congeners typically showed little range overlap, and many were separated by elevational gaps. The prominence of gaps and symmetric responses suggests that biotic interactions such as competition rarely enforce range boundaries along this gradient. Bird surveys matched to plots with measured vegetation structure and tree inventory plots showed that birds had lower richness, broader distributions, and lower turnover compared to trees, but patterns of turnover were largely congruent between taxa. Tree composition, vegetation structure and elevation together explained 82% of variation in bird dissimilarity. Covariation among vegetation structure, trees, and elevation made the largest contribution to explained variation, suggesting that vegetation structure and composition are highly linked to bird composition and diversity. Single-species approaches to range limits used heterospecific song playback experiments to test the hypothesis that interspecific competition between congeners with elevational replacements constrains range boundaries. Playbacks conducted in the Tilaran Mountains of Costa Rica showed that individuals at replacement zones exhibited aggressive territorial behavior in response to songs of congeners. As distance from replacement zones increased, aggression towards congener song decreased, suggesting a learned component to interspecific aggression and an important role of species' densities in development of these interactions. While competitive interactions appear weak in determining range limits in the Andes, these interactions are apparently important among species tested in Central America. These results invite future investigation into potentially varying conditions along elevational gradients that influence the strength of biotic interactions in limiting species. ( en )
General Note:
In the series University of Florida Digital Collections.
General Note:
Includes vita.
Includes bibliographical references.
Source of Description:
Description based on online resource; title from PDF title page.
Source of Description:
This bibliographic record is available under the Creative Commons CC0 public domain dedication. The University of Florida Libraries, as creator of this bibliographic record, has waived all rights to it worldwide under copyright law, including all related and neighboring rights, to the extent allowed by law.
Thesis (Ph.D.)--University of Florida, 2010.
Adviser: Robinson, Scott K.
Co-adviser: Levey, Douglas J.
Electronic Access:
Statement of Responsibility:
by Jill Jankowski.

Record Information

Source Institution:
Rights Management:
Copyright Jankowski, Jill. Permission granted to the University of Florida to digitize, archive and distribute this item for non-profit research and educational purposes. Any reuse of this item in excess of fair use or other copyright exemptions requires permission of the copyright holder.
Embargo Date:
Resource Identifier:
706489330 ( OCLC )
LD1780 2010 ( lcc )


This item has the following downloads:

Full Text




2 2010 Jill Emily Jankowski


3 To my parents, Mike and Laverne, and my loving and adorable partner in crime, Aaron


4 ACKNOWLEDGMENTS This dissertation would not have been possible wi thout the unwavering dedication and guidance from my coadvisors and committee, the l ong months of work contributed by my assistants, the logistical and financial support of many people and organizations, both in the United States and abroad and the patience and encouragement from loved ones throughout this six year endeavor My advisors, Doug Levey and Scott Robinson, have been terrific role models in academics and teaching, and their dedication to family serves as an example of how to keep the most important things in perspective. The members of my committee, including R. Holt, S. Phelps, and K. Sieving, were extremely support ive and gave constructive feedback on all aspects of the dissertation. Many people worked in the field with me and assisted in d ata collection, often for months on end including T. Boza, R. Butrn, L. M. Cabrera, R. Cruz, K. Elliot, V. Huamn, M. Jurado, M. Libsch, G. Londoo, J. Olano, Z. Peterson, N. Quisiyupanqui, R. Quispes, M. Salazar, S. J. Socolar, A. Spalding, J. Ungvari M artin, and C. Zurita. I thank the Campbell, Cruz, Guindon, Leitn, Lowther, Morrison, Rockwell and Stuckey families of Monteverde for allowing me to work on their properties, and the Monteverde and Santa Elena cloudforest preserves, the Monteverde Conserv ation League, Bosque Eterno, and Monteverde Biological Station for granting permission to work in these protected areas K. Rabenold helped enormously with logistics in Monteverde, and I am very grateful for the academic and emotional support he has conti nued to provide over the years, as a friend and colleague The staff of Parque Nacional Manu, Asociacin par a la Conservacin de la Cuenca Amazonica (ACCA ) and Peru Verde made field research possible in Peru I thank the Direccin General de Areas Naturales Protegidas y Flora y Fauna Silvestre of INRENA, which


5 provided permits for research and exportation of samples. M. Silman, N. Salinas W. Farfan, K. Garcia, and L. Imunda blazed the trails, literally, for much of our work in Peru, and let me tag along on adventures to other beautiful corners of Manu. Many thanks to D. Willard and the staff at the Field Museum of Chicago for use of the bird collections, and to D. Stotz, D. Lane, S. Herzog, J. Tobias and R. Yaber for help with identification of Andean bird songs. A. Spalding built microphones, repaired mistreated electronic equipment, and at one point recovered data from failed hard drives. Thanks to K Rabenold C. Stracey G. Londoo and J. Terborgh for comments on various dissertation chapters and t o B. Bolker, J. Poulsen and C. Clark for statistical advice. Funding was received from a University of Florida Alumni Fellowship, the Animal Behavior Society Sigma Xi, the American Ornithologists Union, ACCA, the National Aeronautical and Space Administ ration, and the National Science Foundation.


6 TABLE OF CONTENTS ACKNOWLEDGMENTS ...................................................................................................... 4 page LIST OF TABLES ................................................................................................................ 9 LIST OF FIGURES ............................................................................................................ 10 ABSTRACT ........................................................................................................................ 11 CHAPTER 1 INTRODUCTION ........................................................................................................ 13 2 A MECHANISTIC APPROACH TO ELEVATIONAL RANGES OF TROPICAL BIRDS ......................................................................................................................... 16 Introduction ................................................................................................................. 16 Defining the Physiological Niche for Mon tane Species ............................................. 20 Physiological Constraints on Range Limits at Different Life History Stages ...... 23 Predicted Adaptations in Physiological Tolerances along Elevational Gradients ........................................................................................................... 24 Biotic Challenges for Montane Species ..................................................................... 26 Predation Risk along Elevational Gradients ........................................................ 26 Competitive Interactions and Species Replacements ........................................ 28 Mutualistic Interactions in MixedSpecies Flocks ............................................... 32 Habitat Structure and Vegetation ........................................................................ 33 Modeling Species Distributions to Test Abiotic and Biotic Effects on Ranges ......... 35 Scenarios of Climate Change Impacts for Tropical Montane Species ..................... 37 Conclusions ................................................................................................................ 40 3 RANGE SIZ ES AND RESPONSE SHAPES OF TROPICAL BIRDS ALONG AN ANDEAN ELEVATIONAL GRADIENT ....................................................................... 42 Introduction ................................................................................................................. 42 Methods ...................................................................................................................... 45 Study Site ............................................................................................................. 45 Bird Surveys ......................................................................................................... 46 Data Analyses ...................................................................................................... 48 Results ........................................................................................................................ 50 Elevational Range Sizes ...................................................................................... 50 Species Response Curves and Position of Range Boundaries ......................... 52 Discussion ................................................................................................................... 55 Elevational Range Sizes ...................................................................................... 56 Species Response Shapes an d Positions of Range Boundaries ...................... 58 Conclusion ............................................................................................................ 62


7 4 THE RELATIONSHIP OF TROPICAL BIRD COMMUNITIES TO TREE COMPOSITION AND VEGETAT ION STRUCTURE ALONG AN ANDEAN ELEVATIONAL GRADIENT ....................................................................................... 68 Introduction ................................................................................................................. 68 Methods ...................................................................................................................... 71 Study Region ........................................................................................................ 71 Tree Censuses ..................................................................................................... 73 Bird Surveys ......................................................................................................... 73 Vegetation Structure ............................................................................................ 74 Avian Feeding Guild Classification ...................................................................... 75 Data Summary and Analyses .............................................................................. 75 Results ........................................................................................................................ 78 Species Richness ................................................................................................. 78 Patterns of Species Occurrence, Nestedness and Turnover ............................. 79 Taxonomic Congruence in Birds and Trees with Elevation ................................ 80 Explanation of Bird Dissimilarity by Vegetation Structure, Elevation and Trees ................................................................................................................. 82 Discussion ................................................................................................................... 84 Species Richness Patterns in Birds and Trees ................................................... 84 Species Turnover, Nestedness and Patterns of Compositional Similarity ......... 86 Importance of Tree Composition, Vegetation Structure and Elevation .............. 89 Prospects for Montane Birds and Trees with Climate Change .......................... 90 5 SQUEEZED AT THE TOP: INTERSPECIFIC COMPETITION MAY CONSTRAIN ELEVATIONAL RANGES IN TROPICAL BIRDS ............................. 101 Introduction ............................................................................................................... 101 Methods .................................................................................................................... 103 Study Area and Target Species ......................................................................... 103 Territory Mapping, Playback Stimuli and Experiments ..................................... 104 Statistical Analysis ............................................................................................. 107 Results ...................................................................................................................... 108 Discussion ................................................................................................................. 110 6 CONCLUSION .......................................................................................................... 119 APPENDIX A ELEVATIONAL RANGES OF SPECIES IN MANU ................................................. 123 B SUMMARY OF RARE SPECIES ............................................................................. 137 C SPECIES RESPONSE CURVES ALONG THE MANU GRADIENT ...................... 144 D RESPONSE CURVES OF CONGENERS ALONG THE MANU GRADIENT ........ 226


8 E VARIATION PARTITIONING METHODS BASED ON MULTIPLE REGRESSION ON DISTANCE M ATRICES ........................................................... 239 LIST OF REFERENCES ................................................................................................. 242 BIOGRAPHICAL SKETCH .............................................................................................. 261


9 LIST OF TABLES Table page 3 -1 Summary of elevational range overlap of species within multi -species genera. .................................................................................................................... 63 4 -1 Plot names and elevation study transects, and number of bird auditory visual survey points and vegetation structure areas associated with each plot. ............ 93 4 -2 Pairwise dissimilarity values for sim and nes across bird survey sites wi thin 650-m elevational zones along the gradient. ........................................................ 93 4 -3 Proportion of explained variation (R2) in bird species dissimilarity ( sim) in models from multiple regression on distance matrices (MRM ). ........................... 94 4 -4 Percent of variation in bird dissimilarity ( sim) f or all birds and for each guild explained uniquely by each predictor or co explained by multiple predictors.. .... 94 4 -5 Proportion of explained variation (R2) in bird species dissimilarity (using sim and nes) in models from multiple regression on distance matrices (MRM). ........ 95 4 -6 Percentage of variation in bird dissimilarity (using sim and nes) explained uniquely by each predictor or coexplained by multiple predictors.. ..................... 95 5 -1 Number of playback t rials conducted per species (includ es control, congener, and conspecific trials) and number of individuals tested per species. ............... 114 5 -2 Mixed model results for each species pair for closest approach to speaker and latency to approach speaker for individuals close to th e replacement zone ( wrens, thrushes).. .................. 115 5 -3 Mixed model results for each species pair for closest approach to speaker and latency to approach speaker for individuals at varying distance from the replacement zone (shown as Distance below). ................................................ 116


10 LIST OF FIGURES Figure page 3 -1 Elevational range profiles for birds netted or detected during point counts within the Manu study area. ................................................................................... 64 3 -2 The elevational ranges of species that are completely encompassed within the study area are plotted against species elevational midpoints. ...................... 65 3 -3 Frequency histogram for the number of sites where species were detected during netting (n = 51 sites) and point counts (n= 185 sites). .............................. 66 3 -4 Examples of Huissman-Olff -Fresco (HOF) model types used to estimate species responses (as probability of occurrence) with elevation. ........................ 67 4 -1 Map of the study area in Manu National Park, with locations of 1ha tree plots along the elevational gradient (triangles). ............................................................. 96 4 -2 ( a) Species richness for trees and birds with elevat ion for 15 1ha plots and (b) the relationship between species richness of birds (SB) and trees (ST). ..... 97 4 -3 Species richness of avian foraging guilds across plots varying in (a) elevation and (b) average canopy height. .......................................................................... 98 4 -4 Dissimilarity values of sim (a) and nes (b) for adjacent plots along the elevational gradient for birds (circles) and trees (diamonds). ............................... 99 4 -5 Cluster dendrograms for birds (a) and trees (b) across 1hectare plots (labels show plot code and elevation), using the average linkage method. ................... 100 5 -1 Response to playback trials by individuals with territories near replacement zone s ( wrens, -thrushes). ................. 117 5 -2 Closest approach to the speaker in response to congener and conspecific stimuli for individuals at increasing distances from the replacement zone. ....... 118


11 Abstract of Dissertation Presented to the Graduate School of the University of Florida in Partial Fulfillment of the Requirements for the Degree of Doctor of Philosophy DISTRIB UTIONAL ECOLOGY AND DIVERSITY PATTERNS OF TROPICAL MONTANE BIRDS By Jill Emily Jankowski August 2010 Chair: Scott K. Robinson Cochair: Douglas J. Levey Major: Zoology The distributions that species exhibit along environmental gradients are geographical expressions of the range of ecological conditions that allow species persistence. My dissertation research used elevational gradient s in the Peruvian Andes and in Central America to study range boundaries of birds along mountainsides and to understand i mplications of ecological reinforcement of range boundaries for landscape patterns of diversity. This includes description and modeling of elevational range size, position, and response shapes, community -level analysis of species turnover and composition, and experimental approaches with single species. Bird survey data collected along a 2700-m elevational gradient in Peru showed that most species exhibited symmetric response curves with elevation; however, 16% of species showed asymmetric response curves Congeners typically showed little range overlap, and many were separated by elevational gaps. The prominence of gaps and symmetric responses suggests that biotic interactions such as competition rarely enforce range boundaries along this gradient. Bir d surveys matched to plots with measured vegetation structure and tree inventory plots showed that birds had lower richness,


12 broader distributions, and lower turnover compared to trees, but patterns of turnover were largely congruent between taxa. Tree composition, vegetation structure and elevation together explained 82% of variation in bird dissimilarity. Covariation among vegetation structure, trees, and elevation made the largest cont ribution to explained variation, suggesting that vegetation structur e and composition are highly linked to bird composition and diversity. Single -species approaches to range limits used heterospecific song playback experiments to test the hypothesis that interspecific competition between congeners with elevational replacements constrains range boundaries. Playbacks conducted in the Tilarn Mountains of Costa Rica showed that individuals at replacement zones exhibited aggressive territorial behavior in response to songs of congeners. As distance from replacement zones inc reased, aggression towards congener song decreased, suggesting a learned component to interspecific aggression and an important role of species densities in development of these interactions. While competitive interactions appear weak in determining rang e limits in the Andes, these interactions are apparently important among species tested in Central America. These results invite future investigation into potentially varying conditions along elevational gradients that influence the strength of biotic int eractions in limiting species.


13 CHAPTER 1 INTRODUCTION Ecologists have long been challenged to understand the ecological factors that maintain species range limits. Tropical mountain ranges, because of their high species diversity and density of range li mits at small spatial scales, have enormous potential as study systems for understanding species distributions and the ecological factors that maintain them. My dissertation research uses elevational gradient s in the Peruvian Andes and in Central America to explore and test several alternative ecological mechanisms that could determine range limits of montane birds. Chapter two is an essay that places the study of range limits of montane birds in a context of alternative mechanisms and explores ways to ind ependently assess the effects of abiotic and biotic factors on bird species distributions. This discussion centers on considering elevational distributions of species as a representation of species fundamental and realized niches and outlines a species distribution modeling framework that can be applied to test the combination of factors that best predict dist ributions along gradients. Chapter three uses an extensive database of bir d surveys collected along the 27 00m elevational gradient in Manu National Park Peru to examine the ranges and response curves of breeding birds with elevation. Huismann-Olff -Fresco models are employed to evaluate the shape of species responses and assess the long-held assumption of species exhibiting symmetric, bell -shaped distributions along ecological gradients. Modeled species responses are also used to make inferences on the influence of abiotic versus biotic interactions, with particular attention to the role of competitive interactions, in maintaining species distr ibu tions in this landscape.


14 Chapter four explores the association between bird communities and vegetation along the elevational gradient in Manu to determine whether patterns of species richness and turnover in bird and tree taxa are congruent and whether tree composition, vegetation structure, elevation, or a combination of variables best predict bird composition. These analyses are conducted for all birds and also within avian foraging guilds. The dataset consists of b ird surveys and vegetation structure m easurements conducted at a subset of sites in Manu that were spatially matched to 15 1ha tree inventory plots. The analysis describes species richness and occurrence patterns for trees, all birds, and separately for avian guilds. Dissimilarity matrices and cluster dendrograms are used to present patterns of species turnover and nestedness with elevation and to examine cross -taxon congruence. Multiple regression on distance matrices and variation partitioning were employed to determine the unique and cov arying contributions of trees, vegetation structure, and elevation to explaining variation in bird composition Chapter five presents a direct test of the role of biotic processes in maintaining range boundaries of tropical montane birds. One historically popular hypothesis, especially for birds, is that interspecific competition constrains ranges of closely related species that replace each other along elevational gradients. Supporting evidence, however, is based on patterns of occurrence and does not reveal potential mechanisms. This study, located in the Tilarn M ountains of Costa Rica, uses heterospecific playback experiments to test a prediction of this hypothesis in two genera of tropical songbirds Catharus (Turdidae) and Henicorhina (Troglodytid ae), in which species have nonoverlapping elevational distributions along the Pacific slope. Behavioral responses to


15 song playbacks are evaluated to determine whether species respond aggressively to songs of congeners where their ranges meet. Additional ly, species reactions to congener songs were analyzed at increasing distances from replacement zones to determine whether responses were learned or, alternatively, whether they might reflect misdirected intraspecific aggression. Finally, the level of asy mmetry in aggressive responses of species pairs at the replacement zone was evaluated to assess the likelihood of interspecific behavioral dominance. Tropical mountains have been targeted as regions where climate change will have high impacts on narrowly distributed flora and fauna. Over the next century, tropical montane species may be forced to undergo upslope range shifts as large as the width of their elevational distributions. Any constraints on species ability to shift their ranges, whether abiotic or biotic in nature, will likely reduce population sizes and increase their risk of extinction. This dissertation introduces a variety of approaches that can be used to understand ecological constraints on species range boundaries in these landscapes so that the potential risks to species with climate change can be better evaluated.


16 CHAPTER 2 A MECHANISTIC APPROACH TO ELEVATIONAL RANGES OF TROPICAL BIR DS Introduction Ecologists have often used environmental gradients to understand the range of conditi ons associated with species occurrence and the reinforcement of species boundaries (Whittaker 1975, Brown et al. 1996, Case and Taper 2000, Case et al. 2005, Holt and Keitt 2005 Holt et al. 2005). Elevational gradients, in particular, have been widely a dopted as model systems for understanding range limits, in part because they provide large changes in abiotic conditions and habitats over relatively small spatial scales (Whittaker 1975, Terborgh 1971, Diamond 1973). Such changes go hand -in hand with hig h beta diversity or species turnover which describes the rate at which species composition changes across space, and high regional richness, or the total number of species occurring in a larger area ( Magurran 2004, McKnight et al. 2007, Jankowski et al. 2 009, see also Tuomisto 2010). Elevational gradients worldwide differ considerably in both the typical breadth of species ranges and levels of beta diversity, with gradients in tropical regions generally being characterized by species with narrower distrib utions than those of temperate species (McCain 2009). Likewise, elevational gradients in the tropics typically have higher beta diversity than those in the temperate zone (Buckley and Jetz 2008, Melo et al. 2009). These patterns are found across vertebrate taxa, although vagile, endothermic taxa (e.g., birds) tend to have lower rates of species turnover than sedentary, ectothermic taxa (e.g., amphibians; Buckley and Jetz 2008). This feature of tropical mountains high species turnover generated by narrow ranges can lead to levels of richness that rival tropical lowland rainforest at regional scales, making tropical


17 montane landscapes hotspots of global biodiversity (Myers et al. 2000). Revealing the mechanisms that underlie the narrow ranges of montane sp ecies in the tropics can help reveal mechanisms that maintain regional diversity. The study of mechanisms underlying tropical montane species distributions take s on a new, practical importance in the face of climate change. M ontane landscapes are expected to undergo vast changes and reorganization of community structure as species are pushed upslope in response to warming temperatures (Parmesan 2006). Even under moderate climate change scenarios (e.g., 3C increase in the next century; Solomon et al. 2007 ), simply tracking the thermal environment would require species to shift hundreds of meters upward in elevation, given typical adiabatic lapse rates in the tropics of ~5 -6C/1000m (Colwell et al. 2008, Bush et al. 2004, J. M Rapp, pers. comm. ). In many cases, these distances are larger than the current width of species elevational distributions. Moreover, there is mounting evidence that tropical species could be intolerant of even relatively small shifts in thermal environments, at least in ectothermic animals (e.g., Deutsch et al. 2008, Tewksbury et al. 2008, Huey et al. 2009). It has also been suggested that the tropical montane climate itself may be endangered, as it is predicted to be among the climates that may no longer exist after a century of projected climate change (Williams et al. 2007). The combination of these three strikes restricted elevational ranges of species, high sensitivity of species to change, and major predicted shifts in environmental conditions suggest that these regions of im mense biodiversity will be especially vulnerable to changes in climate. Yet, for most taxa, we lack the fundamental data (see below) that will allow us to predict how communities will respond to shifting climate regimes, or what the key drivers may be


18 On e response to the growing urgency of predicting effects of future climate change on species has been the development and widespread application of species distribution models (SDMs; Peterson et al. 2002, Guisan and Thuiller 2005, Elith et al. 2006, Hijmans and Graham 2006, Guisan et al. 2007). One class of SDMs called bioclimatic envelope models are built using climate variables that represent the range of sites where species occur. Climate envelopes are then projected onto present or future climate landscapes to generate predicted species distributions. Though widely employed, bioclimatic envelope models have been called into question for their ability to accurately predict species ranges (see Pearson and Dawson 2003). The most widely held criticism is that the climate models do not explicitly incorporate biotic interactions, even though many such interactions could constrain species ranges (e.g., competition, predation, mutualisms; Arajo and Luoto 2007, Cunningham et al. 2009, Chapter 5). When appl ied to species distributions in montane regions, however, an important limitation is the mismatch in spatial scale on which bioclimatic envelope models are most successful (e.g., regional spatial scales > 200 km: Pearson and Dawson 2003) and the scales imp ortant to mapping distributions of montane birds (e.g., site, local or landscape scales of 100 m 10 km; Graves 1988, Stotz et al. 1996). Bioclimatic envelope models therefore may not perform well for species with either exceedingly small geographic rang es, or species with narrow dimensions to their range, such as those exhibited by many montane species. Moreover, the approach adopted by bioclimatic models is not designed to reveal potential drivers of species range boundaries Y et, distinguishing among alternative underlying abiotic and biotic factors is critical to understanding why elevational ranges of species vary among gradients,


19 how this influences variation in beta diversity, and how we should expect montane species to respond to climate change. More recently developed techniques in species distribution modeling may offer promising avenues for testing the roles of abiotic and biotic forces on species ranges. At their core, SDMs are necessarily linked to ecological niches of species. As Hutchins on described, the niche space of an organism includes the range of all environmental conditions and factors that permit the persistence of that organism over time ( Brown 1984, Holt 2009). The fundamental niche includes the range of abiotic environments suitable for a species in the absence of competition and other biotic interactions, whereas the presence of both abiotic and biotic factors in the environment restricts a species to its realized niche (Silvertown 2004). As the literature on species distribu tion modeling has expanded (see Guissan and Thuiller 2005 and Elith and Leathwick 2009 for reviews), so has discussion of the representation of species fundamental and realized niches in SDMs (e.g., Pearson and Dawson 2003, Kearney and Porter 2009) Most have concluded that SDMs largely reveal species realized niches because the models are based on observed occurrences of species, which are influenced by both abiotic conditions and biotic interactions (Guisan and Thuiller 2005). By contrast, a more recently developed approach to species distribution modeling, called mechanistic niche modelling, determines the sites suitable for species based on species measured physiological responses to a range of environmental conditions, rather than the values of c limate variables where species occur (e.g., Kearney et al. 2008). Such mechanistic models of species physiological tolerances may therefore reflect the fundamental niche rather than the realized niche (Kearney and Porter 2009,


20 see also Holt 2009). The c ombination of such mechanistic models with more traditional SDMs, by predicting species distributions based on aspects of their fundamental and realized niches, respectively, may provide new opportunities for testing hypotheses of the role of abiotic facto rs and biotic factors in species distributions. In this essay, we propose an approach to understanding range boundaries that considers the interplay of both abiotic and biotic constraints on avian elevational distributions Our aim is to develop a framework that incorporates empirical data as predictors within SDMs so that the relative influence of abiotic and biotic factors can be evaluated. The abiotic aspects of species distributions, representing the range of conditions within the fundamental niche o f a species, are considered through identifying physiological responses of species to changes in ambient conditions with elevation. A species hypothesized physiological niche can be further modified or constrained by the biotic environment in various w ays forcing species to occupy only a subset of the locations that are physiologically suitable. We begin by outlining a method to estimate the physiological niche of montane birds, with special considerations for physiological limitations during differen t life history stages and possible physiological adaptations of species at different altitudes. This is followed by a discussion of potential biotic forces that could limit species to regions within their physiological niche, and a description of how thes e attributes can be incorporated into SDMs. We conclude by exploring various scenarios for responses of species to climate change in montane regions, given these potential abiotic and biotic forces on range boundaries. Defining the Physiological Niche for Montane S pecies The physiological response of organisms to their environment, and the ability of organisms to maintain homeostasis, is perhaps the most important consideration in


21 defining a species fundamental niche (Kearney and Porter 2009). On mountainsides, the most apparent and consistent changes with increasing altitude are declining ambient temperatures and reduced oxygen availability due to declines in barometric pressure, air density, and partial pressure of oxygen. Elevational trends in solar radiation, wind speeds and vapor pressure (i.e., a measure of atmospheric moisture that is independent of temperature) can be highly var iable across seasons (e.g., J. M Rapp, pers comm. ), and elevational patterns in annual precipitation are highly depend ent upon the geographic position of montane slopes (e.g., windward versus leeward gradients; Lomolino et al. 2006). All of these conditions can combine to influence basic physiological processes of respiration, temperature regulation and water balance, as well as behaviors such as foraging, reproduction, and, in volant animals, flight physiology and mechanics (McNab 2002, see also Altshuler and Dudley 2006). Efforts to understand the range of physiological conditions that tropical species can tolerate have stemmed, in part, from the classic hypothesis proposed by Janzen (1967) This hypothesis posits that mountain passes are higher in the tropics because tropical species have limited exposure to varying thermal regimes across seasons, and therefore shoul d be adapted to a narrower range of physiological conditions compared to temperate zone species. This hypothesis further predicts that tropical species should have narrower elevational ranges and tropical mountains should show higher species turnover (see also Ghalambor et al. 2006). Additionally, because daily temperatures undergo greater fluctuations at higher elevations in the tropics (e.g., above 3000m; Soobramoney et al. 2003) highelevation species should have broader physiological tolerances than low elevation species. As predicted by this hypothesis, elevational


22 ranges of species match location -specific thermal regimes (Ghalambor et al. 2006, McCain 2009) and at least ectotherms have narrower thermal tolerances in the tropics (Tewksbury et al. 2008, Huey et al. 2009). Few studies, however, have tested the actual physiological tolerances of tropical species ( e.g., McNab 2001, Huey et al. 2009), with even fewer examining tolerances in tropical montane species (Soriano et al. 2002). If species elev ational distributions were governed by their physiological capacity to tolerate the breadth of ambient conditions they experience, then species distributions along gradients should reflect measured traits of their metabolic functions in varying climatic c onditions (Kearney and Porter 2009, Monahan 2009). Energy expenditure under different thermal conditions, for example, is a commonly used physiological metric to evaluate the ability of organisms to persist in or colonize thermally distinct environments ( Alexander 1999, McNab 2002, Angilletta 2009). For endotherms, the range of temperatures in which individuals do not need to expend additional energy to regulate body temperature is called the thermal neutral zone (TNZ; Schmidt -Nielsen 1997). If the physi ological niche, as defined by the TNZ (e.g., Lindsay et al. 2009), matches the thermal range of species observed elevational distributions, then it would suggest that elevational distributions are determined largely by thermal tolerances. Alternatively, if local habituation or acclimation to thermal environments occurs within species, the TNZ of a species could vary within its elevational range. The thermal tolerance of ectotherms, because their body temperature is subject to changes in ambient temperature, can instead be measured using thermal performance curves (Huey and Bennet 1987, Angilletta 2009), which plot the the relationship between some


23 fitness -related measure (e.g., survival, growth, reproduction, or capacity for movement) and body temperature (see Kearney and Porter 2009). In montane environments, temperature may act in combination with reduced oxygen availability and lower air density, which vary systematically with altitude, to determine species physiological tolerances. Because changing air densities alter the power requirements of flight, species may face important energetic or morphological constraints at these elevations, especially those species with high activity requirements (e.g., hummingbirds: Altshuler et al 2004, Altshuler and Dudley 2006). While birds demonstrate particularly efficient respiratory and circulatory systems, highelevation species can possess additional adaptations for enhanced gas exchange (e.g., higher baseline affinities of hemoglobin for oxygen; Altshuler and Dudley 2006). Thus, lower oxygen availability may constrain upper range boundaries of low elevation species if the adaptive plasticity for anatomical or physiological traits to cope with hypoxia is constrained. Furthermore, limitations in dealing with oxygen concentrations may act independently or in combination with thermal tolerances to restrict species metabolic performance. Perhaps, for endotherms, a temperatureoxygen concentration neutral zone, which reflects both the changes in temperature and oxygen availability with altitude, would be a better descriptor of a species fundamental physiological niche. Physiological Constraints on Range Limits at Different Life History Stages For organisms with complex life histories, we may also need to consider physiological constraints at different stages of the life cycle. For example, although adult birds are endotherms, for much of their development, in particular during the egg and most of the nestling stages, avian embryos are essentially ectothermic. Embryonic development in birds is unique among oviparous vertebrates because it typically


24 requires direct transfer of body heat from incubating adults (Deeming 2002, Turner 2002). The energetic costs of incubation have generally been overlooked because they are thought to be far lower than the more obvious costs of feeding young. Recent studies suggest, however, that incubation is more energetically costly than previously realized (Williams 1966, Thomson et al. 1998, Tinbergen and Williams 2002). Because tim e and energy invested in incubation translate to less time and energy for foraging and self maintenance, incubating birds face a tradeoff between parental care and self care (Williams 1996, Tinbergen and Williams 2002, Turner 2002, Londoo et al 2008). El ucidating this life history tradeoff requires understanding what influences incubation rhythms (i.e., the length and temporal pattern of incubation bouts; Martin 2004), which, in turn, may shed light on the ability of different species to exist at certain elevations or habitats (Webb 1987). McCain (2009), for example, found that elevational ranges of tropical birds appear more constrained by temperature during nesting than during the rest of the year, which suggests that the physiological needs of young an d limitations on incubation behavior of adults may contribute to limitations of elevational ranges in tropical birds. More generally, we predict that the elevational distributions of birds, especially at the upper limit of the elevational range where phys iological conditions are most extreme, will be constrained to areas where species can no longer regulate egg temperature by varying any of the parameters that birds use to ensure proper embryonic development ( e.g., nest insulation, location, incubation rhy thm, egg size ; Collias and Collias 1984, Martin et al. 2007) Predicted Adaptations in Physiological Tolerances along Elevational Gradients Because physiological pressures change with elevation, we predict that species with different elevational ranges sh ould vary predictably in their capacity to tolerate


25 environmental conditions. For example, when considering species at their resting metabolic rate (RMR), we might expect highelevation species to use energy more efficiently when exposed to lower oxygen c oncentrations than species that only occur at lower elevations. Because reduced air density at high elevation may also constrain species, especially in foraging capacity, we might further expect high elevation species that rely heavily on flight to exhibi t morphological adaptations that enhance flight performance and reduce wing loading (e.g., Altshuler et al. 2004). In the case of e mbryos, species occurring at higher elevations may be better able to develop at lower ambient temperatures than those occurri ng at lower elevations. The cooler the ambient environment, the more quickly eggs will cool during parental recesses from incubation (Webb 1987, Cresswell et al. 2004). Although embryonic exposure to cold temperatures during incubation is not as lethal a s excessive upper temperatures (Webb 1987), embryos that grow at temperatures below optimum experience slower growth. Accordingly, embryos developing under cool temperatures should have low metabolic rates (Olson et al. 2006, Nikkilson 2007), which will allow slow transformation of egg resources into embryonic tissue rather than using them for self -maintenance. In this way, embryos in cold temperatures may avoid depletion of their resources and successfully complete development before hatching. Alternati vely, parents of embryos that develop in cool temperatures could provide additional resources for the embryo (e.g., proportionately more yolk), to avoid the probability of resource depletion (Martin 2008), or modify their behavior to sustain an egg temperature closer to the optimum (e.g., by reducing length of trips or increasing nest insulation). Of course, the length of development is also affected by vulnerability to nest


26 predation ( e.g., Ghalambor and Martin 2002 ), which also varies strongly with elevation (see below). If oxygen concentrations constrain species distributions at the level of embryonic development, then embryo performance curves should be affected by altered oxygen concentrations, and highelevation species should perform better under l ower oxygen levels. At high elevations, a larger functional pore area of egg shells would allow more oxygen diffusion into the shell, but would also increase water vapor loss from the egg, which occurs more readily at high elevations ( Rahn et al. 1977, Carey et al. 1983). There is evidence that pore area of shells is lower in highelevation species (Carey et al 1989), suggesting that preventing desiccation may be more important than increasing oxygen transfer. Oxygen concentrations, however, may still constrain metabolic performance. T esting whether highelevation species outperform those at lower elevations in reduced oxygen environments would address this possibility. Biotic Challenges for Montane Species The underlying premise that the realized niche of species can differ from the fundamental niche was that biotic interactions (e.g., competition) may shape a species distribution within its fundamental niche (Sobern 2007). In this section, we outline a variety of biotic interactions that, based on pr eliminary evidence, could be important determinants for range boundaries of montane birds. The evidence for each of these interactions is presented, followed by predictions for how observed species distributions may be constrained within their physiologi cal niche. Predation Risk along Elevational G radients In birds nest predation is the primary source of mortality and is an important selective pressure on avian life history traits and nesting behavior (Ricklefs 1969, Martin


27 1993, Conway and Martin 2000). High levels of nest predation, for example, favor species with long incubation bouts to minimize activity around the nest that could attract predators and possibly shorter overall incubation periods (Ghalambor and Martin 2002, Martin et al. 2006, Massaro et al. 2008). Experimental evidence suggests that nest predation risk can influence nesting habitat choice (Fontaine and Martin 2006). The characteristically high rate of nest predation in the tropics, relative to the temperate zone, is also suspected t o select for smaller clutch sizes in tropical birds (Skutch 1985) or smaller eggs (Martin et al. 2006), such that investment in any given brood is reduced and the ability to re-nest is increased. Given the strong selection pressure imposed by nest predat ors, the distributions of montane birds could be determined by elevational variation in the risk of nest predation and in the kinds of predators present. Little is known about changes in nest predation risk along elevational gradients in the tropics, but existing studies suggest that predation pressure declines linearly with increasing elevation as diversity and abundance of avian nest predators decrease (Skutch 1985, Boyle 2008). Preliminary data from southeastern Peru (G. Londoo, unpubl. data ) also sug gest that rates of nest predation decline with elevation; however, change in daily survival rate across elevations is not linear. Rather, rates of nest predation decline dramatically over a short elevational range, from over 15% daily nest loss near 1000m to approximately 5% daily nest loss near 1500 m. Given a typical 25 -35 day nesting period, these rates translate into losses > 97% near 1000m, but only 50-80% above this elevation. This abrupt change in nest predation rates may correspond with dramatic decreases in some of the major nest predators such as monkeys and snakes. Given these patterns in nest


28 predation, it is likely that the extremely high levels of nest loss found at lower elevations may prevent highelevation species from extending their ranges down slope because their suite of life history characteristics would not have been evolutionarily shaped by such severe nest predation pressures. Moreover, if nest predation decreases nonlinearly with elevation, then we may predict that highelevatio n species show realized distributions with truncated lower boundaries, despite being physiologically capable of occupying lower portions of the gradient. Low elevation species, being tolerant of the highest of predation levels, should occupy their entire physiological niche, unless limited by other biotic interactions at high elevations. Competitive Interactions and Species R eplacements One of the most comprehensive studies of montane birds in the tropics concluded that competition is an overriding force l imiting species distributions along elevational gradients (Terborgh 1971, Terborgh and Weske 1975). Direct competitive interactions, in particular, were used to explain observations of species range limits that coincide along elevational gradients. This pattern of species replacement has been most commonly noted for closely related, usually congeneric species. The hypothesis that interspecific competitive interactions prevent the coexistence of strong competitors along elevational gradients was suppor ted by the observation of species range expansion along replicate gradients in which one of the competitors was absent (Terborgh and Weske 1975); however, community wide patterns suggesting competition have not been experimentally verified. For territor ial animals such as birds, interspecific competition is often expressed as interspecific territoriality (i.e., territorial defense against individuals of other species; Orians and Willson 1964). Interspecific territoriality has been experimentally shown t o


29 reflect the outcome of interspecific competition, such that the habitat segregation of species resulting from this behavior confers higher fitness compared to individuals whose territories overlap with those of competitors (Martin and Martin 2001). Nume rous studies have used song playback experiments to elicit aggressive responses in territory holders and evaluate the strength of territorial behaviors (Martin et al. 1996, Trillo and Vehrencamp 2005, de Kort et al 2009). Likewise, heterospecific song pl aybacks (e.g., playing songs of one species to another) have been used to detect aggressive interactions between species. In the lowland tropics of Peru, heterospecific playback experiments demonstrated interspecific aggression between species that partit ion early and late-succession habitats, shaping local distributions in multiple genera of birds (Robinson and Terborgh 1995). Heterospecific song playbacks experiments have recently been applied to species showing replacements along elevational gradient s, allowing measurements of behavioral responses of interspecific aggression and evaluation of interspecific interactions that may underlie abrupt replacements. Recent work in the Tilarn mountain range in Costa Rica found that interspecific territorialit y occurs at range boundaries separating congeners along large-scale elevational gradients (Chapter 5). For four of five species tested, individuals located near replacement zones (i.e., where territories of congeners are found within 100m of each other) s howed aggressive territorial responses to congener song playbacks. These results support the hypothesis that interspecific aggression maintains range limits in congeners with elevational replacements.


30 Two other results from the same study system provide further insight into how interspecific aggression can constrain species distributions. First, two of three species pairs tested in the Tilarn Mountains showed asymmetry in their responses toward congener songs, in which one species responded more aggres sively than the other. This is common in many interspecifically aggressive species (e.g., Robinson and Terborgh 1995, Martin and Martin 2001). Second, species with interspecific territorial responses only expressed this behavior near replacement zones; wh en playbacks were conducted with individuals at increasing distances from replacement zones, the aggression of territory holders towards congener songs decreased. These results suggest that interspecific aggression has a learned component that depends upon levels of contact with heterospecific competitors; if species rarely encounter competing congeners, they do not develop a territorial response towards these species. These biotic interactions should have consequences for the realized distributions of s pecies along elevational gradients. Specifically, species that have interspecific territorial responses toward congeners at replacement zones should exhibit narrower, truncated elevational distributions compared to the range predicted by their physiologic al tolerances. Additionally, when considering elevational distributions of all species along the gradient, species that co occur with congeners on a mountainside should, on average, occupy smaller portions of their physiological niches than species of mon otypic genera, assuming that congeners are more often competitors than species from other genera. Symmetric and asymmetric interspecific aggression will be expressed differently in the degree to which realized distributions of species match their physiol ogical niches.


31 If aggressive interactions between species are symmetric, then both species should show narrower realized distributions than what their physiology would allow. Alternatively, if aggressive interactions between species at replacement zones are asymmetric, dominant species should be able to successfully exclude congeners from sections of the gradient that it prefers. We would therefore expect dominant species to occupy a greater portion of their physiological niche along gradients, whereas s ubordinate species should occupy a smaller fraction of the areas where they could persist based on their physiology alone. Local densities of interacting species at range boundaries will influence the strength of aggressive interactions and the degree to w hich realized distributions of species reflect their physiological niches. If the development of interspecific aggression depends upon the level of contact with heterospecific competitors, then we predict that any environmental factors or traits that infl uence species densities would also affect the strength of interspecific aggression in maintaining range limits. When both interacting species are abundant at their replacement zone, they should exhibit strong territorial responses to congener songs, resulting in limited range overlap. When both species have low densities, territorial responses by either species should be comparatively weak and range overlap should be greater, perhaps allowing for closer matching between physiological and realized elevati onal ranges. When one species is more common than the other, the less common species should show stronger interspecific territorial responses, because it experiences a higher encounter rate of its competing congener, but it is unclear how much range overlap should be observed in these cases.


32 Mutualistic Interactions in Mixed-Species Flocks Birds are players in a wide variety of mutualistic interactions, with some spanning different trophic levels, as in the roles of birds as pollinators and seed dispersers of flowering plants, and others operating within trophic levels, as in the case of mixed species foraging flocks. In the temperate zone, mixed -species flocks are largely transitory, but in lowland tropical forests, such flocks can be impressively stable, maintaining the same composition of individuals throughout the year. These flocks are organized around nuclear species, and other core species, which actively defend a joint flock territory against conspecifics, resulting in flocks primarily composed o f single mated pairs of each species (Munn and Terborgh 1979, Powell 1979, Munn 1985). Tropical species differ in the extent to which they participate in such flocks. Previous studies of mixed -species flocks have ranked species by flocking propensity, including basic categories of obligate (>85% of the detections of a species), frequent (50 -85%), and occasional flockers (Buskirk et al. 1972, Munn and Terborgh 1979, Jullien and Thiollay 1998). While studies of high elevation flocks suggest a more dynam ic structure than flocks in the lowlands (Buskirk et al. 1972, Poulsen 1996), no study has yet examined patterns of change in flock composition or the associations of flocking species ranges along elevational gradients If flocking species in montane fore sts are dependent upon mixed-species flocks for their foraging needs then flocking behaviors may impose constraints on species distributions along gradients, causing interacting species to reach their range boundaries at the same elevations and, potentia lly, limiting them to subsets of their physiological ranges. Generally, we should expect species that depend most on mixedspecies flocks to be those most likely limited by these mutualistic interactions Obligate


33 flocking species, for example, should ha ve range boundaries that coincide along gradients, whereas facultative or occasional flocking species should show independent range limits that are independent of other flock member species. In terms of the realized distributions of species along gradient s, nuclear species would be expected to express elevational ranges that are constrained compared to what their physiological tolerances would allow. Species with lower flock propensities would not be expected to have range limits that coincide with nuclear or core species boundaries. Data from Weske (1972) fro m the Vilcabamba range of Peru provide preliminary evidence that discrete communities where species have similar range boundary locations along elevational gradients may occur in tanager flocks. M any of the same insectivorous species that depend upon flocks in the lowlands have upper range limits in foothills or cloud forest, but their reliance on mixed -species flocks at these elevations has not been investigated. Habitat Structure and V egetation The relationship between birds and vegetation has been a natural starting point for understanding the influence of biotic interactions on bird species distributions (Wiens 1989, Block and Brennan 1993). Vegetation structure provides important cues that guide habitat selection in terrestrial birds (Rotenberry 1985, Lee and Rotenberry 2005) and also provides food resources and substrates used for shelter, nesting and foraging (Robinson and Holmes 1982, Cody 1985, Lee and Marsden 2008) Likewise many pla nts rely upon birds for pollination and seed dispersal (Stiles 1985, van Schaik et al. 1993). In tropical lowland and montane forests, the variation in type and structural complexity of habitats has been shown to influence bird diversity and composition


34 (MacArthur and MacArthur 1961, Karr and Freemark 1983, Terborgh 1985), especially for some foraging guilds (e.g., insectivores; Terborgh et al. 1977; Chapter 4). In tropical montane landscapes, vegetation undergoes striking changes from low to high elevations, shifting from lower montane rainforest to cloud forest to puna grassland above treeline (Richards 1996, Terborgh 1971, Patterson et al. 1998). Distinctive patches of vegetation can be found within each of these forest classifications. For example, bamboos, including Guadua spp. at low elevations and Chusquea spp at high elevations, can form dense stands within forest. At higher elevations, exposed, windswept ridges are covered in dense, shrublike v egetation with a short canopy, and landslides leave scars in various stages of succession across the landsc a pe. Thus, even tropical montane landscapes that are largely unaltered by human activity can be extremely disturbance-mediated, resulting in a pa tchwork of low and high -canopy forest with distinct st ructural elements. Effects of this variation are readily seen in patterns of site richness in birds, and, at higher elevations, vegetation type influences levels of nestedness in bird composition (i.e., the degree to which species occur in nested subsets across sites) depending on canopy cover, canopy height, and ground cover (Chapter 4). Vegetation structure and composition can also influence patchiness of species within their elevational ranges. For example, bamboo stands and regenerating landslides ar e micro habitats that harbor specialist species in montane landscapes (Stotz et al. 1996). Such associations to particular vegetation types or to a range of structural elements in forest may shape the realized distributions of spec ies along elevational gradients, limiting species to a subset of the range where species should occur based on physiological tolerances.


35 Modeling Species Distributions to Test Abiotic and Biotic Effects on Ranges Incorporating only physiology -based traits of a species into SDMs is useful because doing so portrays predicted distributions uninfluenced by the multitude of biotic factors that can affect species occurrence. If all physiological responses of a species to climatic conditions are represented, then SDMs should provide an accurate assessment of a species fundamental niche (Kearney 2006, Kearney et al. 2008, La Sorte and Jetz 2010). The se mechanistic niche models can be used, by themselves, to test hypotheses of the relevance of physiological tolerances to species range l imits. For example, if mechanistic models underpredict occurrence (i.e., if a species is found at sites where the model predicts absence), then perhaps behavioral or genetic flexibility within a species may allow it to occupy a wider range of climates or, alternatively, spatial variation in climate parameters may create patches suitable for species persistence. If mechanistic models overpredict species occurrence (i.e., if a species is not found at sites where the model predicts presence), then additional variables may need to be incorporated to improve model results. This may include information on previously unmeasured physiological response parameters (improving the mechanistic model) or information on biotic pressures in the environment. A primary dif ficulty with prediction of species distributions relying on mechanistic modeling of species physiology is in how well the measured responses to abiotic conditions by species can be matched to conditions species would experience in the landscape. Spatial variation in abiotic conditions may be particularly important. For example, along elevational gradients, temperature decreases predictably with elevation, but microclimatic variation in the landscape at similar elevations can be caused by differences in s olar radiation, forest structure, and cloud cover. The influence of daily as


36 well as seasonally varying conditions in temperature and moisture on species should also be considered when outlining areas that, climatically, should lie within a species distr ibution. Moreover, behavioral and habitat use patterns may restrict species to a narrower range of microclimates than what is available in the landscape. The challenge is therefore to translate and map the physiological tolerances of species to the lands cape such that the resulting model is useful for interpreting the role of physiology in species distributions. Combining mechanistic niche modeling with variables that represent biotic interactions offers a robust approach to hypothesis testing of species range boundaries. Until recently, SDMs rarely explicitly included variables reflecting biotic interactions, despite widespread acknowledgment of their potential role in limiting species distributions (Pearson and Dawson 2003, Guisan et al. 2005, Elith and Leathwick 2009). Several studies have recently attempted to include such effects by incorporating predictors of the occurrence or abundance of interacting species, such as a critical host plant (Araujo and Luoto 2007) or a potential competitor (Ritchie et al 2009), and determining whether including these variables improves model predictions. Other studies have inferred biotic interactions such as competition by demonstrating a lack of overlap in the distributions of potential competitors when models bas ed solely on climate variables predicting cooccurrence (Anderson et al. 2002, Costa et al. 2008). Ideally, independently collected empirical data or experiments should support the role of biotic predictor variables on species occurrence (e.g., Cunningham et al 2009). It is this data driven framework that we recommend for examining the distributions of


37 montane birds, where the role of physiological and biotic response variables can be evaluated for both upper and lower range boundaries. As they are descr ibed elsewhere, the output of mechanistic niche models and more traditional SDMs differ (Kearney and Porter 2009). Mechanistic niche models provide some measure of species fitness (e.g., survival, performance, reproductive capacity) in response to variat ion in environmental conditions; thereafter, portions of the area of interest can be included within or excluded from the species range based on whether site -specific climate variables allow suitable levels of the species response. SDMs by contrast, ty pically produce output in terms of probability of occurrence or abundance given values of environmental parameters across sites. Combining these techniques to test hypotheses on the roles of abiotic and biotic processes can be achieved in at least two way s. First, predicted distributions from mechanistic niche modeling and models built on biotic variables could be compared directly in their predictive accuracy (e.g., see also Hijmans and Graham 2006). Alternatively, a species physiological response coul d be represented as a predictor that represents the abiotic conditions (e.g., metabolic rates associated with site thermal environments), and multiple models including abiotic and biotic predictors, or interactions of these, could be compared in their abil ity to predict species occurrence (Kearney and Porter 2009). Scenarios of Climate Change Impacts for Tropical Montane Species Species may respond to future climate change in numerous ways in montane landscapes, and predicting the most likely scenario is important for prudent planning of biodiversity conservation. Overall, montane species are expected to show upslope range shifts to track optimal climatic conditions (Hughes 2000, Wilson et al 2005, Parmesan 2006), which in tropical montane areas, should equate to movements of 150


38 200 m elevation for each 1C increase (using adiabatic estimates from Bush et al. 2004, Colwell et al. 2008). This may be more plausible for birds, which have the capacity to shift quickly in response to changing climates (Thomas and Lennon 1999, Hickling et al 2006, Tingley et al. 2009, Zuckerberg et al 2009) than for more sedentary taxa such as trees, which may shift more slowly due to their shorter dispersal ranges and longer generations times (Huntley 1991, Iverson et al 2004, Thuiller et al 2008, Ibanez et al 2009). These shifts also assume that temperature changes of this magnitude (e.g., changes in maximum and minimum temperatures, daily and seasonal variation and relative humidity) will make highelevation, unoccupied sites suitable for a species and make previously occupied sites at lower elevations unsuitable. Alternatively, if appropriate habitat microclimates can still be found within the current elevational ranges of species, then species may respond successf ully by changing the spatial use of their current environment ( e.g., Huey et al. 2009) rather than exhibiting elevational shifts. Another possibility is that elevational shifts, if governed by climate, will be asymmetric, with species expanding into new s uitable environments and seeking appropriate microclimates through behavior in locations already occupied. Potentially, the range of thermal environments that species already experience may dictate the extent of range shifts that species will undergo. Fo r example, highelevation species in the tropics, which experience a large range of temperatures on a daily basis (Ghalambor et al 2006; McCain 2009), may be less subject to range shifts compared to low elevation species. Whether shifts, either locally o r with elevation, are likely from a climate tracking perspective, is contingent upon the range of physiological tolerances of species.


39 Colonization of new suitable sites by species tracking changing climates may effectively shuffle the composition of local biotas, bringing species into contact that have not coexisted in the recent past. In this way, a wide range of species may experience shifting biotic pressures. Understanding what the current biotic pressures on species ranges are may make the consequences of novel combinations of species more predictable. For example, we expect that changes in the nest predator community along elevational gradients could introduce particularly intense selection pressures on nesting birds, in part because the current di sparity in nest predation is so drastic between foothill and cloud forest elevations (see above). Upslope shifts in nest predators with climate change could have wide-ranging effects on the reproductive success of cloud forest birds. Additionally, direct competitive interactions between closely related species that currently show elevational replacements present a biotic hurdle for species range expansions that will likely depend on whether changes in the abiotic environment alter species resource acquis ition, the capacity of individuals to hold territories, and whether interspecific aggression is symmetric or asymmetric (Chapter 5). Dominant low elevation species could effectively track climate changes, relegating subordinate species to higher elevations, whereas dominant highelevation species may prevent upward range expansion of subordinate low elevation species. Vegetation structure represents an important environmental feature used in avian habitat selection (e.g., Lee and Rotenberry 2005), but it is unclear to what extent bird distributions, if unhindered by other biotic interactions, will track rapidly changing abiotic conditions versus potentially slow changes in vegetation structure and plant distributions. If empirical data can be collected to map physiological tolerances of


40 species and assess biotic interactions at range boundaries, then more informed scenarios of range shifts can be explored. While we have focused much of our discussion on abiotic and biotic pressures along intact elevational gradients, anthropogenic alteration of montane landscapes is widespread in the tropics and could present major barriers to range expansion of species in response to climate change. In the Andes, one of the most formidable barriers, even along most conti nuously forested gradients, is the humanmaintained treeline at high elevations (Feeley and Silman 2010). In birds and many other taxa, this inflexible shift from highelevation cloud forest to fire generated puna grassland will likely squeeze cloud fores t species into narrower elevational distributions. Though some studies show that cloud forest birds have broad elevational distributions in the Andes, there is a suite of species with narrow distributions in upper cloud forest that might be particularly s usceptible to a treeline-climate change pinch (see Chapter 4 Patterson et al. 1998). Conclusions The narrow distributions of montane species in the tropics make these communities exceptionally susceptible to global climate changes (Parmesan 2006), where s pecies may be faced with upslope range shifts to track warming temperature regimes (Colwell et al. 2008). In total, nearly 30% of the worlds threatened birds are restricted to narrow elevational ranges in tropical montane regions. Projections of such ra nge shifts have typically taken the form of climateenvelope models, but these models make critical assumptions that species will track shifting temperature regimes. The degree to which we can rely on these models depends upon understanding how species wi ll physiologically tolerate changing temperatures. Measurements of metabolic


41 traits in tropical birds worldwide are currently limited to fewer than 100 species (e.g., Wiersma et al. 2007), and only 13 are montane (McNab 2009). Yet, not even these studies have measured thermal or oxygen tolerances; they are virtually unknown for any tropical montane birds. Biotic interactions also likely serve as important limiting factors to tropical species distributions, constraining species in ways not predicted by phy siological tolerances alone. Empirical data on species distributions, physiological traits and species interactions will allow multiple mechanisms potentially influencing range limits to be tested simultaneously for tropical birds. In addition to allowi ng more informed predictions of how species are likely to respond to climate change, these data will allow testing of numerous questions on the elevational distributions of birds: Are physiological tolerances of species greater at high elevations compared to low elevations? Are upper and lower range boundaries of montane species affected differently by abiotic and biotic pressures? Accordingly, are biotic pressures on species ranges more likely at low elevations, or are abiotic pressures more common at high elevations? Where along elevational gradients are particular biotic pressures more intense? Can disturbance mediated heterogeneous landscapes of tropical mountains affect patchiness of species distributions, or, alternatively, select for species tha t tolerate a broader range of habitats and microclimates? Are high and low elevation tropical species equally likely to show upslope range shifts with climate change? The possibility of addressing these and other questions continue to make elevational gr adients exciting frontiers for future research on the ecology of species distributions.


42 CHAPTER 3 RANGE SIZES AND RESPONSE SHAPES OF TROPICAL BIRDS ALONG AN ANDEAN ELEVATIONAL GRADIENT Introduction Species distributions along environmental gradients repr esent the interaction of multiple ecological and evolutionary processes and have long served as sy stems to understand patterns in species ranges and boundaries. Elevational gradients in particular, have been central to the study of species distributions and community organization (Terborgh 1971, Diamond 1973, Whittaker 1975) and, especially in the last decade, have been important venues for studies of diversity (Patterson et al. 1998, Navas 2003, Brehm et al. 2005, Herzog et al. 2005, McCain 2005, Jankows ki et al 2009, Romdal and Rahbek 2009), constraints on and shifting of range boundaries (Chapter 5, Wilson et al. 2005, Merrill et al. 2008, Moritz et al. 2008, Tingley et al. 2009), and determinants of patterns in range size (Ghalambor et al. 2006, McCain 2009). Each of these areas is pertinent t o predicting how climate change in montane landscapes will affect regional biodiversity (Shoo et al. 2006, Colwell et al 2008, Sekercioglu et al. 2008). In particular, determining the forces acting to shape the di stributions of individual species will be key to making informed predictions for how those species will respond to future climate change (Chapter 2). Hypotheses for the abiotic and biotic forces that influence species distributions along elevational gradients can be informed by three basic characteristics of species distributions: (1) the size of the elevational distribution, which indicates the extent of species occurrence and the range of conditions species can tolerate; (2) the shape of the distribution, showing the way that species respond to conditions along the gradient through changes in abundance; and (3) the position of the distribution with respect to


43 other species, which can reveal potential interspecific interactions, when informed by species natural history. Numerous mechanisms have been proposed to predict patterns of species elevational range sizes along gradients. Janzen (1967) suggested greater seasonal variation and overlap in climate conditions across altitudes in temperate mountains compared to tropical mountains and proposed that temperate species should have broader physiological tolerances and wider elevational range sizes than tropical species. An extension of this hypothesis can be applied to single elevational gradients, predic ting that elevational range sizes of species should increase with elevation to reflect the larger variation in daily temperature patterns at exposed upper elevations compared to lowlands (Ghalambor et al. 2006, McCain 2009). An alternative hypothesis for the same pattern is that mountain tops exhibit an island effect, such that lower diversity and greater isolation at high elevations results in broader niche s and larger elevational distributions through ecological release ( McNaughton and Wolf 1970, MacAr thur 1972) Finally, the mid domain effect hypothesis best known for its predictions of a hump-shaped pattern in species richness along environmental gradients, predicts that range sizes will be largest for species in the center of a bounded gradient due to constraints on the maximum attainable range size of species with midpoints close to gradient boundaries (Colwell and Lees 2000, Colwell et al. 2004). Wh ereas elevational range sizes describe the extent of species occurrence along gradients and patter ns in environmental tolerance across species, they provide little information about how species respond to gradients within their range through changes in probability of occurrence or abundance. Ecological niche theory predicts


44 Gaussian unimodal distribut ions along environmental gradients ; species reach their highest abundance at the center of their range and decline gradually toward the periphery (Austin 2005, 2007). This view of a bell -shaped, symmetric species response has become a practical assumption for statistical treatment of species -environment relationships (e.g., ordination techniques : Palmer 1993, ter Braak 1985, Oksanen and Minchin 2002) and a conceptual foundation for testing ecological and evolutionary hypotheses related to species ranges, referred to as the abundant -center assumption (Brown et al. 1996, Sagarin and Gaines 2002, Sagarin 2006). It is now recognized, however, that species interactions and extreme environmental stress along gradients may result in asymmetric species response curves (e.g., Austin and Smith 1989, Austin and Gaywood 1994). Describing the shape of these curves and relating curves of different species to each other can reveal when and why deviations from symmetric bell -shaped distributions occur and can direct fur ther testing on the role of abiotic and biotic forces on species boundaries. Furthermore, knowing the underlying response shapes is of practical importance when model ing speci es distributions in relation to the environment (Oksanen and Minchin 2002, Austi n 2007, Santika and Hutchinson 2009). The position of a species distribution with respect to other species along gradients can be used as an additional line of evidence to test hypotheses on the abiotic or biotic factors that enforce range limits (Whitaker 1975). This aspect goes hand-in hand with properties of a species response shape. For example, species with elevational replacements, if maintained by competitive interactions, would be expected to show segregated distributions along gradients and perh aps truncated boundaries in the direction of a competitors range (Terborgh 1971, Whittaker 1975). Alternatively,


45 comparisons of species with similar range positions and coincident range boundaries may indicate groups of species that are limited by simila r underlying conditions (e.g., habitat boundaries, ecotones) or species with strong mutualistic interactions (see Chapter 2). In this study, we provide survey data of 395 species of breeding birds along a 2700m elevational gradient in the tropical Andes t o describe the size of species elevational ranges evaluate the shape of species response curves with elevation, paying special attention to their symmetry, and, assess the position of species distributions with respect to other species along the gradient To our knowledge, this is the first statistical analysis of species responses along an elevational gradient for any tropical vertebrate. From these analyses, we infer the role of abiotic and biotic interactions in maintaining species distributions alon g the gradient In particular, we address the potential for competitive interactions to maintain boundaries between closely related species and the role of habitat association in determining species distributions Methods Study Site The study area is located on the eastern slope of the Andes in the Department of Cuzco, southeastern Peru (S130320, W0713249) along an elevational gradient from 800 3500 m. Temperature decreases with elevation along this gradient following an adiabatic lapse rate of approximately 5.2C/km, and annual precipitation declines monotonically with elevation at a rate of approximately 1250 mm km1 (J. M. Rapp, pers. comm. ). Wh ereas temperatures show little seasonal variation, there is strong seasonality in precipitation, w ith a dry season lasting from April through October (driest


46 periods in June and July) and a rainy season from November through March (with precipitation peaks in January and February; J. M. Rapp, pers. comm. ). Avian nesting behaviors, including territory establishment and singing, generally begin during the late dry season (August) and continue at least through December Vegetation along the gradient shifts from montane rainforest to cloud forest to puna grassland (Terborgh 1971, Patterson et al. 1998), wi th patches of distinctive vegetation occurring within each of these broad habitat classifications (Chapter 4). Most notably, these vegetation patches include exposed ridge tops with stunted vegetation, patches of bamboo ( Guadua spp below 1500 m and Chusquea spp above 1500 m) and vegetation in various stages of recovery following landslide s which are particularly common in midto high elevations where topography is steepest The elevation of the cloud base, which marks the transition from montane fores t to cloud forest, is located between 1400 2100 m, al though this depends on season and time of day ( J. M. Rapp, pers. comm. ). Above ~2100m biomass of mosses and other epiphytes increases markedly, serving as a biotic indicator of the saturation of fore st by clouds and mist. Above ~ 2800 m, the cloud forest gives way to a stunted canopy, classified as elfin forest elsewhere (Patterson et al. 1998). In Manu, the transition between elfin forest and puna grassland varies from 3000 3500 m and is largely set by anthropogenic factors, specifically the history and extent of burning in puna grasslands for cattle grazing by highelevation farming communities (Feeley and Silman 2010c). We sampled forest habitats along this gradient stopping at treeline. Bi rd Surveys The regional species pool within Manu National Park including lowland and montane forest birds, non-forest species, and migrants, comprises nearly 1100 species


47 (Patterson et al. 1998, Walker et al. 2006). In this study, we focused on resident montane forest birds. Audio visual point -count surveys were conducted during field seasons in 2006 2009 at 185 forested points along the elevational gradient, with an average of 18 19 points within each 250 -m elevational zone. Surveys took place duri ng the first half of the breeding season (July November), when birds were most easily detected by song, in 2006 2009. Points were separated by at least 1 3 0 m along narrow trails and marked with a Garmin GPSMap 60CSx GPS unit and flagging tape. Sets of ten ten minute counts were conducted between 5:00 and 9:00 hours on mornings without heavy winds or rain. Sites were revisited a total of four times throughout the study period (2006 2009) usually with two or more visits occurring within a single breeding season. R evisiting points a fourth time yielded 14% new species detections on average, whereas visiting more than four times resulted in only 6% new species detections, even in the most diverse parts of the gradient (J. Jankowski, unpubl data). Th us, four visits to a given point typically provide d a representative sample of detectable species while maximizing efficiency. O rder of visitation to each set of points was reversed between visits to reduce biases due to temporal variation in species det ectability. During each count, all individuals detected were identified and their distances from the point were estimated. Individuals detected >100 m from the center of the point were not included in analyses because birds detected at longer distances m ay be at substantially different elevations from the elevation at the point in this steep landscape. All counts were recorded using an Edirol R -09 WAV/MP3 digital recorder with a built -in omnidirectional microphone for later review of species identificati ons.


48 W e supplemented audiovisual counts with mist netting of birds to provide presence absence information on cryptic species or species that vocalize infrequently. Netting was conducted using ten nets per group. A n average of 56 groups were run within each 250-m elevational zone throughout the breeding seasons from 2005 to 2009 (total mist net hours = ~10,000) Mist nets were placed along or near trails at forested sites and were run for three days from approximately 6:00 to 17:00 hrs during favorable weather conditions (Chapter 4). Data Analyses Species occurrence data from audiovisual survey points and from mist netting sites were combined to describe elevational ranges of each species. A species was considered present at a site if it was detected during any visit, or absent if it was never detected during the entire sampling history of that site. To visualize trends in range size along the gradient, species were ranked by the midpoint of their elevational range and plotted with elevation. Spec ies response curves along the elevational gradient were fitted to presence absence data using Huissman -Olff -Fresco (HOF) models (Oksanen and Minchin 2002, Wilson et al. 2007). HOF models enable five hierarchical models to be fit to species presence absenc e or abundance data along environmental gradients These include, from most complex to simplest functions, V) skewed, or asymmetric, IV) symmetric, III) plateau, II) monotone, and I) uniform response forms. The most complex model, which fits a skewed or asymmetric response, is written : ) exp( 1 1 ) exp( 1 1 dx c bx a M


49 where, is the expected value (e.g., probability of occurrence, when using presence absence information), M is the maximum possible value (e.g., equal to 1 when using probability of occurrence), x i s the gradient of interest (elevation), which can be scaled from 0 to 1 for analysis, and a b c and d are estimated parameters of the function (Oksanen and Minchin 2002) The simpler response forms can be derived from the asymmetric model by fixing one or more of the estimated parameters to constant values. HOF models were fitted to species presence absence data using program JUICE v. 7.0 (Tich 2002, Tich and Holt 2006), which calculates species responses within the R software environment utilizing t he gravy package (R development Core Team 2010). A non -linear maximum likelihood procedure was used to estimate the parameters for the set of hierarchical models (Oksanen and Minchin 2002), and the final HOF model (I V) was selected using AIC. JUICE allows fitting of species response curves using alternative models, including general additive models (GAM) using R package mgcv, general linear models (GLM) and bell -shaped, Gaussian responses. For visual comparison of species response curves, data we re fitted using GAM and Gaussian responses, in addition to HOF models (see Oksanen and Minchin 2002) All species detected at ten or more point -count locations or mist netting sites were modeled. Species responses were calculated using either audiovisua l count data or mist netting data on occurrence, and relied on data from the survey method that resulted in a larger number of detections across sites. R esponse curves with elevation were summarized in terms of shape, elevational range size and the placement of upper and lower range boundaries. Several species groups were examined for general trends to allow inferences about potential ecological


50 forces operating on species ranges. These include lowland, cloud forest, and highland species, bamboo specialists, and species that share the elevational gradient with congeners versus monotypic genera. Two methods were used to assess evidence for direct competitive interactions among congeners: compression of elevational range size in multi-species genera comp ared to monotypic genera; and in the case of species replacement patterns, evaluation of asymmetric shapes to response curves with range boundary truncations adjacent to congeners. Results Elevational Range Sizes Our surveys generated 13,415 audiovisual detections and 4,298 mist -net captures for 421 resident breeding bird species, 395 of which were found in forested sites (Appendix A). Of these, 351 species (89%) were detected during audio visual surveys, whereas 269 species (68%) were detected during mi st -netting. The mean elevational range size of species detected at ten or more sites within the study area was 865 m, and ranged between 83 m and 2525 m. This mean value, however, is biased toward s smaller ranges by lowland (< 800 m) and highland species (> 3500 m) with elevational ranges marginally included within the study area. When considering only species whose complete elevational range falls within 800 3500 m (Walker et al. 2006, Schulenberg et al. 2007), the mean range size was 1021 m (ranging from 322 2173 m). Species with narrower elevational ranges were found mostly at low and high extremes of the gradient, whereas most species occupying the middle portions of the gradient had relatively large range sizes (Figures 3 -1a). This pattern rem ains even after species whose ranges extend beyond the study area were removed (Figure 31b). In


51 this restricted set of species, average elevational ranges of species (detected at sites) with midpoints in the lowest third (800 1700 m) and highest t hird (2600 3500 m) of the gradient were 728 and 815 m, respectively, whereas the average elevational range of species with midpoints between 1700 2600 m was at least 500 m larger (1352 m). Likewise, a plot of the elevational range sizes of species ver sus their midpoints shows a lack of species with narrow ranges in the central portion of the gradient (Figure 32). This plot typically exhibits a pyramid shape due to geometric constraints on the attainable range sizes of a species given its elevational midpoint. For example, a species whose midpoint is located near a gradient boundary necessarily has a smaller range size along the gradient, whereas a species whose midpoint lies in the middle of the gradient could exhibit a variety of range sizes. Most s pecies were detected at very few sites (Figure 33). For example, no species occurred at even 50% of the 236 forest survey sites (including 185 audio visual survey points and 51 net group sites). Eighty -three percent of species (n = 326) were detected at 30 sites or fewer, and 54% of species (n = 215) o ccurred at 10 or fewer sites. Rarity found within this group of 215 species can be attributed to several potential causes (Appendix B). Nearly half (47%) of these species had lowland distributions that ex tend ed upslope no more than 300 m into the study area. One species had a broad highland range above the study area and just reached our high elevation sites. Twenty four species (11%) had rather inconspicuous behavior or resided in forest strata that wer e difficult to sample (e.g., high canopy) and were likely rare due to low detection probability. Another 17% of species were either habitat specialists with patchy distributions within the study area (n = 21; e.g., streamside residents, bamboo


52 specialis ts), or were common in open or heavily disturbed habitats and rarely sampled in forest (n = 14). One -fourth of the rare species (26%, n = 56) were generally detectable with netting or were conspicuous when present during counts, but nevertheless occur red at low densities across the landscape or had an especially narrow elevational zone where they are found. Much of this group was composed of small insectivores (mostly suboscines), hummingbirds, and large bodied birds belonging to various foraging guilds. Species Response Curves and Position of Range Boundaries Response curves were modeled for 164 species; 144 of these were modeled using count data and 20 were modeled using netting data (Appendix C) Of the total species modeled, 98 species (60%) exhibited symmetric response curves with elevation (HOF model IV ; Fig 3 4 ), and 26 species (16%) showed asymmetric responses, in which one elevational boundary was more sharply truncated than the other (HOF model V). Truncations were equally likely to occur on the lower or upper elevational range boundary for species with asymmetric responses (13 upper and 13 lower truncations). Plateau responses from the upper or lower gradient boundary (HOF model III) were exhibited by only seven species, five highland and two lowland. Thirty one species (18%) showed monotonic responses (HOF model II), in which probability of occurrence decline d from one boundary of the gradient; 28 of these were lowland or lower foothill species and three were highland species. Only two specie s showed flat response shapes along the gradient (HOF model I) Response curves were generally similar between HOF and GAM models across species, showing similar range sizes, boundary locations, and symmetry for 85% of the 164 modeled species. In cases w here the two


53 approaches differed, GAM models typically fitted more complex shapes than possible through the hierarchical HOF models (Appendix C). Lowland species (i.e., species occurring down to 250 m; Walker et al. 2006) with ranges extending upslope int o our study area always reached their upper elevational limits within the lower third of the gradient (800 1700 m). Of 43 lowland species with modeled ranges, 58% dropped out in a narrow elevational zone between 1000 1300 m, and 84% reached their uppe r range boundaries below 1400 m. There were comparatively few highland species (i.e., species occurring in forest or puna above 3500 m) that show ed similar declines in occurrence moving downslope. Of the six modeled species with highland distributions, f our had plateau response curves that d id not reach lower boundaries until 2000 2500 m Only two species show ed a monotonic decline in occurrence from the highlands, one of which did not reach its lower boundary until 2000 m. Cloud forest species includ ed 75 modeled species with range optima, (i.e., the peaks of their modeled response curves) between 1400 2800 m, following the designation of Manu cloud forest in Patterson et al. (1998). Upper and lower range boundaries of these species were found betw een 1000 3300 m. Over half of cloud forest species showed lower range boundaries between 1100 1700 m (29% from 11001400 m and 28% from 1400 1700 m). Another 23% of species showed lower range boundaries farther upslope, between 2100 2500 m. Likewise, upper range boundaries of cloud forest birds were concentrated in two areas, with 50% occurring in lower cloud forest, between 1700 2200 m, and another 30% of upper limits occurring at high elevations, between 3000 3300 m. These patterns tend to characterize two


54 groups of cloud forest birds: those with optima occurring above and below approximately 2200 m. One-fourth of cloud forest birds (n = 18) showed asymmetric response shapes, and there was an equal proportion of species with upper and lower range boundary truncations. The locations of these range truncations were not concentrated at any particular elevation. Bamboo habitat specialists that were modeled included 12 species found in Guadua and one species found in Chusquea Seven of the Gu adua species showed upper range boundaries coincident or within 100 m elevation of the upper limits of Guadua, and three of these showed asymmetric response shapes with truncated upper range boundaries. The single Chusquea specialist that could be modeled (Drymophila caudata) showed a lower range limit within 50 m elevation of the lower limits of Chusquea. A comparison of range sizes between species with and without congeners along the elevational gradient showed that the median range size for species sha ring the gradient with congeners was only slightly narrower than the median range size for species in monotypic genera (689 m versus 757 m), and this difference between the groups was not statistically significant (MannWhitney test, Z = 0.8, p = 0.21). Visualizing modeled species response curves of congeners along the gradient allows additional assessment of the potential role of competitive interactions at range boundaries. Generally, congeners partitioned the gradient showing a rather uniform distribu tion of species optima with elevation (e.g., species in Grallaria ; Appendix D Figure D -13 ). Unexpectedly, most congeners that were modeled were separated by elevational gaps or show ed little overlap (Table 3-1 ; see example in Figure D -21 ). Gaps


55 between congeners ranged from 50 700 m elevation, and were usually over 250 m elevation. Many of these congeners are easily detected by audiovisual counts (e.g., Formicarius Chamaeza, Anisognathus ), so the gaps are likely not due to low detection probability. Repulsion interactions at range boundaries between congeners, where the range boundary that is adjacent to a congener range shows truncation, were uncommon. Species in only five of 26 genera with multiple species exhibited asymmetric distributions with a truncated boundary abutting that of a congener. Truncations were reciprocal among these congeners in only one genus ( Myioborus ; Figure D -25 ). Discussion Our results show several overriding features of montane bird distributions along this Andean elevati onal gradient. We find that elevational range sizes of species generally increase from foothill to cloud forest elevations, but decline toward the upper end of the gradient. The considerable majority of species have symmetric, bell -shaped responses to el evation, consistent with ecological theory. Asymmetric responses, however, were not uncommon, and at least a portion of these may be attributable to biotic interactions. Our analyses of range boundary position and response shapes across species groups su ggest that both habitat association and competitive interactions may set range boundaries for selected species, but for many others, these particular biotic factors appear to be weak in determining elevational range limits. Congeners that partition the gra dient, for example, have ranges separated by elevational gaps in the majority of cases. Below we discuss of each of these results in relation to predictions from the literature.


56 Elevational Range Sizes On average, range sizes of species were close to a ki lometer wide (1021 m). This is generally larger than the average range sizes found by other avian studies on this and nearby gradients (e.g., averages between 710807 in Patterson et al. 1998 and approximately 741 m in Terborgh 1971). Both of these studi es, however, sampled lower elevations more extensively than in this study, and the inclusion of lowland species would certainly reduce the average range size. No species along the Manu gradient had an elevational range of more than 2200 m. Species with t he most restricted ranges tend ed to occur in foothills and highland elevations, wh ereas species with the broadest ranges tended to occur in cloud forest. This pattern of larger range sizes in cloud forest is consistent with that reported by Patterson et al (1998). This range -size trend is also consistent with patterns that would be generated by a mid domain effect, in which species with range midpoints found near gradient boundaries necessarily have limited range sizes. Still, the nearly complete lack of species with narrow ranges in the middle of the gradient, and the association of these elevations with cloud forest habitat may also reflect biological phenomena For example, it is possible that cloud forests, due to their frequent saturation by clouds and mist contain key structural characteristics in vegetation that a core group of montane birds use when foraging (e.g., epiphytic growth) and the large elevational extent of these structural characteristics allows montane bird species to attain corresp ondingly large elevational distributions T o be conclusive, t he frequency distribution of range sizes could be tested against expectations under a middomain effect (MDE; Colwell and Lees 2000) to assess the degree to which observed patterns in elevational range distributions reflect


57 biological processes versus geometric constraints on species ranges by gradient boundaries. Our results do not support Janzens (1967) hypothesis that species exposed to a wider range of climatic conditions (e.g., greater diur nal temperature variation at high elevations) show broader elevational distributions except when comparing low elevation and foothill species to cloud-forest species. Species in the highest third of the study area had relatively narrow distributions whi ch is contrary to this hypothesis It is possible however, that the reduced range size of species in the highest third of the gradient is a product of the severe habitat alteration above treeline. Anthropogenic disturbance, in the form of deforestation, fire, and grazing pressures maintains the high elevation puna grassland throughout much of the high Andes and present s a hard boundary for forest species that require more than scattered patches of montane scrub for persistence (Stotz et al. 1996, Feeley and Silman 2010c) Th us, anthropogenic modifications confound the evaluation of range size patterns in forest birds along these gradients. Independent measurements of species physiological tolerances (e.g., thermal tolerances) w ould be a more direct tes t of the mechanisms underlying Janzens hypothesis (Chapter 2). This assessment would also allow this physiology -based hypothesis for increased range size at higher elevations to be distinguished from alternative mechanisms that are expected to produce si milar patterns, such as the competition -based island effect proposed to affect species residing in low diversity, highelevation communities (MacArthur 1972). A large proportion of the species detected in our study region were uncommon or rare. Although describing species rarity was not an initial objective of this study, it is


58 noteworthy that approximately half of species detected could not be modeled effectively, despite extensive sampling; fifty -four percent of species (n=215) were detected at ten or f ewer sampling points. Whereas some of these species are rare because of some sampling artifact (e.g., incompletely sampling the species elevational range or low detection probability), a substantial number of species 77, occurred at low densities, had h ighly restricted elevational distributions, or were habitat specialists. Furthermore, another 102 forest species that are known to occur within the elevational range of our study area were never detected (Chapter 4; Walker et al. 2006). This suggests that rarity is a prevailing characteristic of the bird communities along this Andean gradient, much as it is in the neighboring Amazonian lowlands (Terborgh et al. 1990). Practically, the rarity of Andean birds makes the study of their natural history, range attributes and abundance patterns extremely challenging. Species Response Shapes and Positions of Range Boundaries The majority of species (60%) had symmetric, unimodal responses to the elevational gradient, consistent with assumptions of ecological niche theory ( Austin 2005, Austin 2007, Heikkinen and Mkip 2010). Asymmetric responses however, were not uncommon (16% of species modeled) and response shapes in this more limited group may reflect biotic interactions or physiological stresses at range bo undaries Truncated boundaries of species with asymmetric responses were not confined to a particular elevation nor were they more likely to occur at upper or lower limits of species distributions Our analysis of low elevation species indicates that man y species in this group reach upper limits at a similar elevational zone. At the community level, this transition from lowland to montane avifauna pinpoints the region of highest species turnover along


59 the gradient (Chapter 4). Given that along latitudin al gradients, southern range boundaries of species are often determined by biotic factors whereas northern boundaries are set by abiotic factors (MacArthur 1972, Gross and Price 2000), we may expect that along elevational gradients, biotic pressures should be more influential at lower limits of species elevational distributions and abiotic or physiological pressures more significant at upper boundaries. Overall, physiological tolerances of tropical birds are still poorly known (but see Wiersma et al. 2007), but it has been shown that metabolic capacity for cold tolerance is more restricted for suboscines, which dominate the lowland bird community in Manu, in comparison to oscines, which represent a greater portion of the community at higher elevations (see McKechnie and Swanson 2010). Such limited flexibility in thermal tolerances, if widespread, may play a role in setting the upper limits of this lowland avifauna. Analysis of range boundary positions for cloud forest birds, including species with optima b etween 1400-2800 m, indicate that these species can generally be divided into two groups with range optima below and above the location of the cloud base in the breeding season. This result is also consistent with community -level analyses along the gradient, which show increased species turnover in bird composition near 2250 m (Chapter 4). Within this general division of upper and lower cloud forest birds, species generally had scattered range boundaries, rather than concentrated boundaries at any single elevation. This suggests that, to some extent, montane species distributions reflect the physical changes in climate or vegetation (or both) along the elevational gradient. The cloud base effect in Manu, however, does not impose any sharp boundaries or t runcations on species ranges. Such cloud forest ecotone effects on


60 bird communities have been found along other elevational gradients in the tropics (e.g., Terborgh 1985a, Jankowski et al. 2009), though the degree to which avian elevational limits are affected is variable. Perhaps elevational gradients with more drastic changes in moisture over small spatial scales (e.g., rainshadow gradients; Jankowski et al. 2009) would be more likely to show montane species with truncated and spatially coincident ra nge boundaries located at the cloud base. Among the most distinctive microhabitats along the Manu gradient are the stands of bamboo that grow within forest, which are known to host avian specialists in the lowlands (Kratter 1997) and highlands (Stotz et al 1996, Schulenberg et al. 2007). In montane areas, these stands often occur as a successional stage following landslides or treefalls. The upper limits of Guadua bamboo in Manu occur near 1500 m, and Chusquea extends from this elevation upward to treeli ne. Our results show that approximately half of the common Guaduaassociated species reached their upper range limits well below the limits of this habitat type. It is possible that patches of Guadua found near its upper limits are very limited in extent and only attract species that can persist in smaller stands (e.g., Hypocnemis cantator Myrmotherula longicauda, Basileuterus bivittatus all of which drop out at 1500m). The only common Chusquea specialist seems to have its lower limit set by the existence of appropriate habitat. Congeners that share the elevational gradient collectively show ed little evidence for competitive interactions in maintaining range boundaries Species within multi species genera did not exhibit compressed elevational ranges c ompared to species in monotypic genera, and the majority of distributions of congeners along the gradient were separated by elevational gaps or cooccurred at range boundaries at low densities


61 (< 10% overlap of congener ranges). Repulsion interactions, ev aluated by asymmetric response shapes with range truncation toward congeners, were relatively uncommon, occurring decidedly in three of 26 genera. Two other genera each contained one species that exhibited this pattern. Evidence of interspecific competit ive interactions at range boundaries in mountain ranges elsewhere suggests that higher densities and sufficient contact between congeners may be a prerequisite for active or behavioral exclusion at range boundaries (Chapter 5, see also Terborgh 1971). Bas ed on this, it is unlikely that the congeners partitioning the Manu gradient that were too rare to be modeled are limited by direct (interference) competitive interactions. While these results suggest that direct competitive interactions are not likely to maintain range boundaries at present, one cannot rule out the influence of historical interactions nor can we assess the impacts of diffuse competition in these communities (see Terborgh and Weske 1975). Indeed, the well defined segregation and overdisp ersion in the range optima of most of these congeners along the gradient is highly suggestive of patterns resulting from selective niche divergence and resource partitioning described by ecological theory (e.g., Whittaker 1975). Patterns of elevational gaps between avian congeners have been found elsewhere (Terborgh 1971), yet few hypotheses have been proposed to explain them. MacArthur (1972) suggested that along low productivity resource gradients, competing species that partition this gradient may be se parated by a zone where neither species could occur if levels of species primary resources were sufficiently low. An alternative explanation, proposed by Cornell (1974), suggest s that if species showing such replacements are more susceptible to parasites or pathogens of their congeners, and


62 fitness consequences are sufficiently negative, then elevational ranges of congeners could be separated by an intermediate region that neither species could colonize, and the size of this gap would correspond to the di spersal distance of the parasite via its vectors. Any local selection in the host for genetic variants for increased tolerance of the congeners parasites would likely be swamped by gene flow from the central portion of each species range. It would also be useful to test physiological alternatives for the range boundaries of congeners exhibiting this elevational separation to determine whether these species boundaries may be better explained by physiological specialization to the climatic conditions exp erienced within their ranges (see Chapter 2). Conclusion Our study makes an important step toward defining the underpinnings of species range size s response shapes and the potential roles of abiotic and biotic interactions along tropical elevational g radients. Range sizes of species show trends toward larger ranges at m iddle (cloud forest) elevations, and future analyses, including measurements of species physiolgocial tolerances and tests against mid -domain effects can be used to test alternative hy potheses for this pattern. Our analysis of species response shapes along elevational gradients in the tropics calls attention to several notable patterns in both the shape and position of species range boundaries that may present fruitful areas of focused research. These include understanding the mechanisms that cause lowland species to drop out within a narrow elevational w indow and the existence of elevational gaps separating congeners that partition the gradient. Finally, species -habitat associations and direct competitive interactions between closely related species may influence the overall position of species distributions with elevation, but they do not appear to maintain species range boundaries along this gradient.


63 Table 3 1. Summary of elevat ional range overlap of species within multi -species genera. Only genera with >1 modeled species are shown Gap indicates an elevational gap separates congeners along the gradient. Low Density indicates overlap with < 10% co occurrence of congener ranges. Overlap indicates congener range overlap with occurrence. Repulsion denotes an asymmetric response shape of at least one congener and range truncation in the direction of the other congener. Congeners separated by elevational gaps are assumed not to have repulsion due to comp etitive interactions (even if they exhibit asymmetric response shapes). Genus Gap Low Density Overlap Overlap Repulsion Tinamus X Patagioenas X Geotrygon X Phaethornis X Trogon x x Pharomachrus X Aulacorhynchus x Colaptes X Xiphorhynchus X Thamnophilus X Chamaeza X Formicarius X Grallaria x Scytalopus x x Mecocerculus x Mionectes x Ochthoeca x Pipreola x Turdus x Hemispingus X Anisognathu s X Diglossa x Chlorospingus x Arremon X Myioborus x x Basileuterus x genera with > 3 species, where only one species shows asymmetry in response shape


64 750 1250 1750 2250 2750 3250 0 50 100 150 200 250 300 350 400 Species Ranked by Elevational Midpoint Elevation (m) 750 1250 1750 2250 2750 3250 0 25 50 75 100 125 150 175 Species Ranked by Elevational Midpoint Elevation (m) Figure 31. Elevational range profiles for birds netted or detected dur ing point counts within the Manu study area. Species are ranked along the x axis by their elevational midpoint. Bars indicate the maximum and minimum elevation for a) all species detected and b) the subset of montane species whose complete elevational ranges lie within the study area. a) b)


65 0 500 1000 1500 2000 2500 3000 750 1250 1750 2250 2750 3250 Elevational Midpoint of Species Elevational Range (m) Figure 32. The elevational ranges of species that are completely encompassed within the study area are plotted against species elevational midpoints. The pyramid shape of points demonstrates constraints on range size a ccording to species locations along the gradient, a product of the middomain effect. Empty regions within this pyramid indicate range sizes that are not exhibited by species with those elevational midpoints.


66 0 50 100 150 200 250 10 20 30 40 50 60 70 80 90 100 110 Number of Sites Detected Number of Species Figure 33. Frequency histogram for the nu mber of sites where species were detected during netting ( n = 51 sites) and point counts ( n= 185 sites)


67 Figure 34. Examples of Huissman-Olff -Fresco (HOF) model types used to estimate species responses (as probability of occurrence) with elevation The model types include: Asymmetric (V); Symmetric (IV); Plateau (III); Monotonic (II); and Flat (I). S pecies used to demonstrate each model type in the figure are listed below the x axis. II V I III IV


68 CHAPTER 4 T HE RELATIONSHIP OF T ROPICAL BIRD COMMUNI TIES TO TR EE COMPOSITION AN D VEGETATION STRUCTU RE ALONG AN ANDEAN ELEVATIONAL G RADIENT I ntroduction Tropical montane species are especially vulnerable to future climate change (Parmesan 2006, Williams et al 2007, Colwell et al. 2008, Tewksbury et al. 2008, Feeley and Silman 2010a). In tropical montane regions, plants and animals are often confined to narrow elevational ranges, producing belt -like distributions that may extend hundreds to thousands of kilometers in distance, but span only a few hundred meters in elevation (Stotz et al. 1996, Vzquez G. and Givnish 1998, Haber 2000, Jankowski et al. 2009). In response to warming climates, montane species are generally expected to show upslope range shifts in order to track optimal abiotic conditions (Hughes 2000, Wilson et al. 2005, Parmesan 2006 Colwell et al. 2008, Feeley and Silman 2010b). In birds, such shifts in geographic distributions could potentially occur rapidly, as has been noted in some Palearctic species ( Thomas and Lennon 1999 Hickling et al 2006 P armesan 2006 Visser et al. 2009, Zuckerberg et al. 2009). Trees, on the other hand, may shift distributions more slowly than birds, due to substrate specificity, shorter dispersal distances, and longer generation times ( Huntley 1991 Clark 1998 Iverson et al 2004 Takahashi and Kamitani 2004, Thuiller et al. 2008, Ibanez et al. 2009). Even under scenarios of rapid plant migration, given that bird and plant distributions are set by factors that differ in intensity and kind, it is unclear to what extent bird distributions will track rapidly changing abiotic conditions versus changing tree distributions, which at least in part determine features of the environment used in avian habitat selection (e.g., Lee and Rotenberry 2005).


69 The relationship between birds and vegetation has been a natural starting point for understanding the influence of biotic interactions on bird species distributions (Weins 1989a,b Block and Brennan 1993). Vegetation forms a fundamental component of terrestrial avian habitats and pr ovides important cues that guide habitat selection in birds (Wiens 1989a,b, Rotenberry 1985 Lee and Rotenberry 2005). Plants provide birds with food resources as well as habitat structure and substrates used for shelter and foraging (Holmes and Robinson 1981, Robinson and Holmes 1982, Cody 1985, Lee and Marsden 2008) and, likewise many plants rely upon birds for pollination and fruit dispersal (Terborgh 1977, Stiles 1985, van Schaik et al. 1993, Burns 2004). In temperate regions, previous studies have shown a strong correspondence between avian assemblages and both the composition and structure of vegetation (Rotenberry 1985, Fleishmann et al. 2003, Lee and Rotenberry 2005 Fleishmann and Mac Nally 2006). While both aspects of vegetation are likely imp ortant, it has been suggested that structu ral differences best explain variation in bird species occurrence at larger spatial scales, while floristic differences best explain variation at smaller scales ( Bersier and Meyer 1994, Lee and Rotenberry 2005). S uch assessment of plant composition within structurally similar environments may occur if birds preferentially forage in particular tree species (e.g., Gabbe et al. 2002), perhaps responding to fine-scale structural differences that favor specialized avian foraging strategies (Robinson and Holmes 1984). In contrast to temperate regions, the level of correspondence between tropical bird and pl ant communities is poorly studied, though generally, the type and structural complexity of habitats are known to infl uence bird species diversity and composition


70 (MacArthur and MacArthur 1961, Karr and Freemark 1983, Terborgh 1985b). Most studies addressing tropical avian vegetation associations have adopted broad classification schemes for vegetation, using distinctive structures or dominant plant species to examine bird distributions (Terborgh 1985a,b, Robinson and Terborgh 1995 Kratter 1997). It has been proposed that different guilds of birds should be more sensitive to structural versus compositional changes in vegetation (Terborgh 1985b), with insectivorous birds that rely on specific foraging substrates responding to structural changes and frugivores being more directly linked to plant community composition (Hasui et al. 2007) Along tropical elevational gradients, vegetation changes dramatically in both structure and composition (Grubb et al. 1963, Grubb 1977 Richards 1996). The coincidence of range limits in birds with distinctive habitat shifts along gradients (e.g., ecotones) suggests that vegetation may i nfluence avian species distributions ( Terborgh and Weske, 1975 Terborgh 1985a J ankowski et al. 2009). In their survey of avian distributions along an elevational gradient in the Vilcabamba range of Peru, Terborgh and Weske (1975) reported that one-sixth of elevational range limits coincided with vegetation ecotones, affecting either upper or lower range boundaries of 28% of bird species in the study. This work suggests that, at least to some degree, birds respond to recognizable changes in vegetation al ong elevational gradients, but th e level of correspondence of birds and plants, and the relative influence of vegetation structure and plant species composition on avian communities remains unknown. Here we provide the first analysis, to our knowledge, o f the correspondence of bird and tree assemblages for an environmental gradient in the tropics focusing on a


71 forested elevational gradient from 800 3500 m in the Andes of southeastern Peru, one of the worlds biodiversity hotspots. Using datasets of bi rd and tree species composition and information on vegetation structure along the same gradient, we describe the change in species richness and community composition for these taxa with elevation and determine the extent to which their patterns of species turnover are congruent. Then we ask how well species composition of birds across sites can be predicted by composition of trees, vegetation structure, elevation, or a combination of these factors. We also examine particular feeding guilds of birds, testi ng the hypothesis that foraging guilds that rely on plants directly for food resources (i.e., frugivores, nectarivores, granivores) are more closely linked to plant c omposition, whereas insectivor ous birds are more influenced by structural changes in veget ation. If relationships between birds and vegetation are weak in these environments, which would be indicated by incongruent patterns of species turnover or low explanatory power of bird species composition by vegetation structure or floristic composition, then we would expect greater flexibility in bird-tree interactions, and range shifts of birds with climate change would be likely to occur more independently of vegetation changes on tropical mountains. If we find evidence for high correspondence of bir d and tree communities, then range shifts in bird communities might be constrained by selective forces in birds for particular characteristics of vegetation. Methods Study Region Our survey sites are distributed along an elevational gradient on the eastern slope of the Andes in the Department of Cuzco, southeastern Peru (S130320, W0713249). The study area is accessible from the Cuzco-Pilcopata highway, an


72 unpaved road on the eastern border of Manu National Park which descends from above treeline through the Andean montane forest of the Kosipata Valley and into the Amazon basin. Our highest survey sites at ~3500 m elevation are located 15 km northwest of Acjanaco pass (S131149.0, W0713711.5) and the lowest sites are found at ~800 m elevation in the Andean foothills, 5 km from the park guard station of Tono (S125722.4, W0712853.9). Survey sites are located on trails in primary forest or older secondary growth forest contiguous with primary forest, accessed either by foot or by wading throu gh stream beds (Figure 4 -1) Vegetation along the gradient undergoes striking changes within this elevational range, including montane rainforest, cloud forest, and puna grassland as one proceeds from lower to higher elevations ( sensu Terborgh 1971; Patter son et al. 1998). Within each of these broad forest classifications are patches of distinctive vegetation. Bamboos, including Guadua spp. at elevations below 1500 m and Chusquea spp. above 1500 m, are common across the landscape and in many areas form dense stands within forest. At middle and high elevations, some exposed, windswept ridges are covered in elfin forest -short, dense shrub-like v egetation with a canopy less than 3 m high. Landslides are ubiquitous, leaving scars in various stages of succe ssion across the landsc a pe. Their incidence and return frequency vary with slope and geological substrate, which also contributes to overall habitat heterogeneity. The result is a pa tchwork of low and high-canopy forest with distinct elements of vegetati on structur e even at similar elevations. While logistical constraints prevented us from surveying all of these vegetation types within each elevational zone, we were able to represent most of them in our survey sites.


73 Tree Census es Trees were censused in 15 1ha plots, established approximately every 250 m elevation within primary forest. Plots at 1500 m, 1750 m and 3000 m had two replicates on different transects that were separated by approximately 3 9 k m. Within each 1 -ha plot, every individual 0 cm diameter at breast height (dbh) was measured, tagged and identified to species or to morphospecies (Silman et al. in review ). Bird Surveys Prior to bird surveys, we used field seasons from 2004 2005 for mist netting ( > 10,000 net -hours see below ) and for training in identification of bird vocalizations using a combination of personal song recordings, an online song database for Neotropical birds (, published recordings, and expert knowledge (D. Lane, R. Yaber, pers. comm. ). During f ield seasons from 2006 2008, we established 172 permanent sites to conduct audio visual counts along the elevational gradient, with an average of 17 (ranging from 7 to 34) forest bird survey sites per 250m elevation. Due to steepness of the terrain and restricted accessibility, some elevational zones were sampled more broadly than others. Approximately half of these sites were associated with the 1 -hectare tree census plots (see below). Sites of audio visual counts were placed at least 130 m apart ( horizontal distance) along narrow trails and marked with a Garmin GPSMap 60CSx or 76Cx GPS unit and flagging tape. All counts were conducted between 5:00 and 9:00 hours by C. L. M erkord or myself on mornings without heavy wind or rain. Sites were revisit ed between four and eight times throughout th e study period (2006 2008) with most visits occurring in the avian breeding season (July November). Sets of 10 12 counts were conducted each morning, and the order of visitation to points was reversed between


74 visits to reduce biases due to temporal variation in species detectability. Counts lasted for either 5 or 10 minutes, depending on the observer. During each count, all individuals detected were identified and their distances from the point were es timated. We did not use detections beyond 100 m from the point because birds detected at longer distances may be at substantially different elevations from the elevation at the point. Counts were recorded using a digital recorder with a built in omnidire ctional microphone (Edirol R 09 WAV/MP3 digital recorder) for later review of species identifications. We supplemented audiovisual counts with mist netting of birds to provide presence absence information on cryptic species or species that vocalize infre quently. Mist netting surveys were conducted within or immediately adjacent to all 1ha tree census plots. Mist nets (2.5 by 6 to 12 m, 34 mm mesh) were placed both along and off trail and were run for three days from approximately 6:00 to 17:00 hrs duri ng favorable weather conditions Vegetation Structure Vegetation structure was measured at each bird audiovisual count site within square plots of 10 by 10 m usin g a protocol adapted from Martin et al. (1997) Variables summarized from vegetation structure data generally describe aspects of forest vertical structure or understory vegetation density. Forest vertical structure variables included average canopy height, number of trees > 10 cm dbh (tallied within a 20 by 20 m plot surrounding each 10 by 10 m plot), total tree basal area, and percent canopy cover, obtained from hemispheric photos taken 1 m above ground level using a Nikon D 50 digital SLR camera and fisheye lens mounted on a tripod and analyzed using Gap Light Analyzer v.2 (Frazer et al. 1999 ). Understory vegetation variables included density of


75 small stems (< 2.5 cm dbh; percent cover by shrubs or saplings and bamboo ( Chusquea spp. and Guadua spp.), and percent vegetation cover < 50 cm. Avian Feeding Guild Classification Birds were classified into feeding guilds using references on diets of tropical birds, first -hand observations of foraging in the field, and fecal samples of birds captured with mist nets. Our classification primarily follows Terborgh et al. (1990) for species occurring in lowland and foothill tropical forest and a number of studies with diet information of highland species (Parker and ONeill 1980 Poulin et al 1994 Restrepo and Gomez 1998, Herzog et al 2003). We designated species as frugivorous, nectarivorous, granivorous, omnivorous, or insectivorous. Carnivorous species, such as raptors, and piscivorous species, such as kingfishers, were not included in foraging guild analyses. Such guild classifications are admittedly crude, because very few species are strictly limit ed to a single food type. F or example, even specialized nectarivores such as hummingbirds (Trochilidae) supplement their diets with insects for protein (Stiles, 1995). We expect that this gui ld classification, however, generally reflects each species principal foraging method and dietary requirement. Data Summary and Analyses Analyses were performed at the level of 1ha tree plots (N = 15; hereafter plot level) and, for particular question s, at the level of bird survey sites (N = 172; site level). For plot -level analyses, bird survey data and vegetation structure data collected at bird survey sites were assigned to a 1-ha tree plot if the sites were located within 500 m horizontal distance or within 150 m elevation of the tree plot. If a survey site could be assigned to more than one tree plot by those criteria, then it was assigned to the plot


76 closest in elevation. Mist -netting locations were assigned to a tree plot if the netting occu rred on or within 100 m of the edge of the tree plot. Bird species presence absence data from associated survey sites and mist netting locations were then combined to create a plot -species matrix. Tree occurrence data for each plot were used to create a presence absence plot -species matrix of tree species composition. To summarize plot -level vegetation structure, values of variables were averaged across bird survey sites associated with each plot. Site-level analyses used bird audio visual survey and vegetation structure data only. Bird occurrences at survey sites were used to create a presence absence site -species matrix. We examined patterns of species richness in trees and in birds (overall and within each guild) across plots. We then used several ap proaches based on species dissimilarity matrices to examine variation in bird species composition and to evaluate correspondence of bird and tree communities. Measurements of change in species composition across sites (e.g., beta diversity) that are expressed as dissimilarities can reflect two separate processes: change due to species loss (i.e., nestedness) and change due to replacement of one set of species by another (i.e., turnover; Koleff et al. 2003, Baselga 2010). Because our sites are distributed along an elevational gradient with a large disparity in richness across sampling locations, it is important to distinguish between these processes. We accomplish this by adopting an approach proposed by Baselga (2010) that partitions total beta diversity into contributions by spatial turnover and nestedness. In this approach, Sorensons dissimilarity index ( sor), a common measure of total beta diversity, is broken down into two additive indices: Simpson dissimilarity ( sim; Lennon et al. 2001), which describes spatial turnover without the


77 influence of richness gradients, and nestedness -resultant dissimilarit y ( nes), which is derived as the difference between sor and sim (see Baselga, 2010). In the absence of nestedness, sor is equal to sim. Each index varies between 0 and 1, with lower values indicating a greater proportion of shared species and larger values indicating greater dissimilarity between two locations. Dissimilarity matrices for sim and nes were calculated in R v. 2.11.0 (R Development Core Team 2010) using functions provided by Baselga (2010). At the site level, we summarized pairwise bird dissimilarity values of species turnover ( sim) and nestedness ( nes) for survey sites within four 650m elevational zones to determine whether either beta-diversity metric changed along the elevational gradient. At the plot level, we examined dissim ilarity values between adjacent plots for birds and trees to assess congruency in species turnover and nestedness between taxa. In cases with replicate tree plots at each elevation, the dissimilarity values of those plots with adjacent higher and lower pl ots were averaged. We also performed separate cluster analyses of bird and tree dissimilarity matrices using the sim index and an agglomerative hierarchical clustering algorithm (Legendre and Legendre 1998) Several grouping methods (e.g., single, complete, average, Wards) were used to create cluster dendrograms. The dendrogram with cophenetic distances (i.e., th e distance at which sites are clustered) that best corresponded to the original distances among sites was selected Finally, w e addressed the statistical relationship of bird species composition to tree composition, elevation and vegetation structure throu gh multiple regression on distance matrices (MRM; Legendre and Legendre 1998, Lichstein 2007), where the dissimilarity matri x of tree species composition ( sim), and distance matrices of elevation


78 and vegetation structure variables were used as predictors to explain variation in bird species sim). Models with all possible combinations of predictor matrices were examined to determine which combination explained the most variation in bird species composition across plots. We then used variance partitioning methods to divide the variation explained in bird dissimilarity into fractions representing unique contributions by each predictor alone an d variation coexplained by a combination of predictors ( Appendix E ; Maca et al. 2007, Linares Palomino and Kessler 2009). In the case of vegetation structure, we began with all structural variables in the model, then used backward elimination to remove v ariables without a statistically significant ( p > 0.05) contribution to explaining variation in bird dissimilarity. These structural variables were thereafter retained in more complicated models that included vegetation structure (e.g., vegetation structure and elevation). MRM analyses at the plot level were performed for all birds and independently for each foragingguild. MRM analysis was also performed at the site level for the four 650m elevational zones along the gradient to determine the variation of sim and nes explained in each zone, where predictor matrices included vegetation structure and elevation. All analyses were performed in R 2.11.0 using functions within packages vegan (Oksanen et al. 2008), stats and ecodist (R Development Core Team 2010). Results Species Richness Our surveys revealed high regional richness across the 15-ha plots in both bird and tree communities (Table 4 1). Overall, 894 species of trees were identified from the plots, while 353 species of birds were identified fr om survey sites and mist netting locations associated with plots. Per plot richness for trees was nearly 25% higher than


79 that of birds; the average number of tree species per plot was 99.4 (ranging from 38 to 171), and the average number of bird species p er plot was 75.1 (ranging from 27 to 130). Species richness across plots was highly correlated between birds and trees (Fig ure 4 2 a,b; r = 0.83; p < 0.0001), and the number of species per plot generally decreased with elevation for both taxa, although the re was a low elevation plateau in tree richness between 800 1750 m, above which the number of species dropped considerably (Figure 4 -2 a). When richness within foraging guilds of birds was analyzed, we found that the pattern of decreasing richness with elevation was driven primarily by loss of insectivores along the gradient (Fig ure 4 3 a). Species richness in frugivores, granivores, nectarivores and omnivores also decreased with elevation, but only marginally in comparison with the loss of insectivores. Similar patterns across guilds are shown when species richness is plotted with canopy height (Figure 4 3 b). C anopy height decreases with elevation ( r = -0.68; p < 0.0001) and correspondingly, species richness increases with increasing canopy height, espe cially in insectivores. Patterns of Species Occurrence, Nestedness and Turnover Individual species in both taxa showed narrow distributions along the elevational gradient, generating high species turnover at the community level. On average, individual tr ee species occurred on 1.75 of the 15 plots (12%), and bird species, on average, were associated with 3.1 plots (21%). Over half of tree species (55%) and one-third of bird species (30%) in our study were restricted to a single plot, and 82% of tree speci es and 50% of bird species were restricted to two or fewer plots. Only 5% of bird species, and not a single tree species, occurred on half or more of the plots. At the plot level, pairwise dissimilarity values of species turnover sim) ranged from 0.1 5 to 0.98 in birds (mean = 0.65 0.28 s.d.) and from 0.49 to 0.98 in trees (mean


80 = 0.86 0.13). For both t nes) made up a much smaller component of change in species composition, compared to species turnover. Values of n es across plots ranged from 0 to 0.38 (mean = 0.07 0.08) in birds and from 0 to 0.12 (mean = 0.03 0.02) in trees. At the site level, pairwise bird dissimilarity values varied more than at the plot level sim ranging from 0 to 1 (mean = 0. 76 nes ranging from 0 to 0.86 (mean = 0.05 0.09). Mean pairwise dissimilarity values within 650m elevational zones were not constant across zones (Table 4 -2). Dissimilarity attributed to turnover sim) declined with elevation, such that pairw ise poi nts in high elevation zones, on average, reached only two-thirds of the levels of dissimilarity found in the low elevation zone. Dissimilarity nes), however, was between two to four times greater in the highelevation zones and negligible between points in low elevation zones. This indicates that variation in species composition at high elevations to a greater extent, is determined by differences in richness, where the birds found at less diverse sites are subsets of the species found at more diverse sites. At low elevations, differences in composition across sites are largely due to replacement of one set of species by another, and richness across sites varies less. Taxonomic Congruence in Birds and Trees with Eleva tion Analysis using dissimilarity indices of adjacent plots revealed several patterns in species turnover of bird and tree communities with elevation. Across 250m intervals, trees showed much higher turnover, with adjacent -zone dissimilarity ranging from 0.5 to 0.75, compared to birds, whose dissimilarity values ranged from 0.2 to nearly 0.5 (Fig ure 4 4 ). Points of particularly high species turnover are apparent in both communities and occur in similar locations: (1) foothill elevations between 1000 1250 m ;


81 (2 ) the middle of the gradient or cloud forest between 2000 2500 m (this turnover point occurs between 2000 2250 m for trees and between 2250 2500 m for birds); a nd (3) at treeline between 3250 3450 m (this zone also shows high turnover in t rees, with the loss of all stems > 10 cm dbh above 3450 m). With the highest elevation plots comparison removed (due to spatial mismatch of habitat types sampled between birds and trees at treeline), dissimilarity in sim between adjacent plots was positively correlated between taxa ( r = 0.65, p = 0.02). Dissimilarity of adjacent plots due to nestedness was generally higher in birds than in trees, with relatively high nestedness in cloud forest elevations (1750 2250 m) and in high elevation forest > 3000 m. Dissimilarity due to nestedness in trees was consistently low along the elevational gradient, with slightly higher values in lower cloud forest (1750 1850 m). Dissimilarities in nes were not significantly ass ociated in birds and trees in comparisons of adjacent plots ( r = 0.17, p = 0.30) Cluster analyses of bird and tree communities were constructed using the average grouping method, which produced dendrograms with cophenetic correlations of 0.84 and 0.85 for birds and trees, respectively. Bird and tree dendrograms contained three and four clusters, respectively, that generally separated plots by elevation (Fig ure 4 -5 ). Birds and trees each contained a foothill elevation group (800 1000 m), but plots wi thin middle elevations (i.e., lower cloud forest; 1250 2250 m) were clustered differently between birds and trees. Both taxa showed clusters identifying higher elevation plots (2500 3450 m) in birds, groupings within this highelevation cluster were n ot easily distinguished, compared to trees. Generally, compositional dissimilarities among tree plots were greater than among bird plots, as shown by the


82 distance at which clusters formed in these taxa. For example, in trees, no tree plots shared more than 60% of species (i.e., no pair of plots was joined with a distance below 0.40; Fig ure 4 5 ), whereas in birds, the majority of plots were paired at 40% dissimilarity or lower, indicating, overall, that birds shared more species between plots, compared to trees. Because bird survey sites assigned to each plot are located up to 150 m elevation above or below the plot itself, our bird sites sample a larger area and slightly broader elevational range compared to trees. It is therefore possible that highe r species turnover in trees is generated by a sampling artifact, where narrower spatial sampling makes adjacent plots less likely to share species. To address this, we restricted our analysis to bird sites within 50 m horizontal distance and 25 m elevation of the tree plot. This resulted in slightly higher bird species turnover, as expected, with adjacent -plot dissimilarities ranging from 0.3 to 0.7 (compared to a range of 0.2 to 0.45 in the original dataset). Turnover in trees, however, was still substantially higher than in birds, with an average of 15% greater dissimilarity between adjacent zones in trees compared to birds, suggesting that taxonomic differences are also biological. We proceed with our original sampling design, as it better represents the regional bird diversity. Explanation of Bird Dissimilarity by Vegetation Structure, Elevation and Trees Multiple regression on distance matrices (MRM) showed that a combination of vegetation structure, elevation, and tree dissimilarity best explained variation in bird dissimilarity across plots (Table 4 -3). For the group all birds, and for all but one guild, this combination included all three predictors; however, models with only two of the three predictors often performed nearly as well. In granivores, the best combination included only elevation and vegetation structure variables. Overall, a large proportion of


83 the total variation in bird species dissimilarity was explained across groups (ranging from 44% in granivores to 82% in all birds). Al l groups except granivores had > 70% of variation explained. Variation partitioning of MRM showed that co -contributions by vegetation structure, elevation, and trees explained between 14 31% of variation in bird dissimilarity across groups (Table 4 4). For all birds, frugivores, insectivores, and nectarivores, the remaining variation was largely explained by co -contributions of elevation and trees (29 32%), and uniquely by elevation (10 21%). In granivores, the remaining variation was explained by vegetation structure (13%) as well as covariation in vegetation structure and elevation (10%). In omnivores, unique contribution by trees (10%) as well as covariation in trees and elevation (26%) explained most of the remaining variation. Bird species turnover ( sim) and nestedness ( nes) at the site level within 650 m elevational zones along the gradient was always best explained by a combination of vegetation structure and elevation (Table 4 -5). Variation in species turnover patterns across sites was generally better explained than nestedness patterns; the total variation explained in turnover ranged from 12 74% across zones while the variation explained in nestedness ranged from 6 23%. The unique contribution by elevation and vegetation structure varied a cross elevational zones (Table 4 -6). Elevation explained more variation in species turnover and nestedness in low elevation zones and explained relatively little variation in high elevation zones. Conversely, the unique contribution of vegetation structure to species turnover and nestedness was greater at high elevations.


84 Discussion Our surveys along an elevational gradient in Southeastern Peru include nearly 900 tree species and 350 bird species. For trees, there is little information beyond this stud y on regional diversity in the area, except in lowland communities (Terborgh et al. 1996, Pitman et al 1999, 2001). For birds, a published list from this landscape by Walker et al. (2006) indicates that 582 resident species occur between 800 3500 m on the eastern slope of the Andes, with 562 species found within forest or along forest edges. Of the 209 species missing from our dataset, 58 were detected at other sites that were not associated with tree or vegetation plots (J. E. Jankowski and C. L. Merk ord, unpubl. data ). Threequarters (74%) of the remaining missing species come from lower elevations (800-1700 m), and were likely missed due to differences in location and sampling intensity between this study and Walker et al (2006). In particular, lo wland species might show higher upper elevational range limits on slopes in closer proximity to the Amazon basin compared to their distributions in our study area (D. Stotz, pers. comm ). Additionally, Walker et al. (2006) present an accumulation of recor ds from expeditions and surveys spanning three decades Nevertheless, our avian dataset includes 63% of known resident species in this elevational range. Species Richness Patterns in Birds and Trees Our results show that, g enerally, species richness in tr ees and birds decreased with elevation resulting in a strong positive correlation of richness between taxa. A notable deviation from this correlation occurs at middle elevations, where trees maintain high, lowland-level richness and bird richness continues to decline. This discrepancy could be explained by several changes that occur independently of tree diversity at these plots. First, the foothill elevation of 1250 m marks the loss of canopy emergent


85 trees, greatly reducing the complexity of forest canopy structure, with consequences for richness of canopy birds and foraging flocks. In addition, this elevation marks a decrease in overall tree growth rates and fruit crop production (M. Silman, unpubl. data), reflecting decreases in above-ground net pri mary productivity. Reduced productivity could have strong consequences for the resource base of many avian guilds, and especially for insectivores, which can have specific, inflexible, and time -consuming foraging behaviors for seeking out cryptic prey (Naoki 2007). From an energetic perspective, increasing the foraging area to compensate for lower productivity should be comparably easier for frugivores and granivores (often large-bodied), and traplining nectarivores, which can travel widely in search of f ood (Stile s 1975). Our results also clearly indicate that the loss of insectivores dominates the pattern of decreasing bird richness with elevation; changes in richness of other guilds are relatively slight. These patterns match those found by Terborgh (1977) in the Vilcabamba range of Peru In addition to potential effects of decreasing productivity, discussed above, declining vertical forest structure with elevation may explain reduced diversity at high elevations, following the widely accepted associat ion between habitat complexity and bird spec ies diversity (e.g., MacArthur and MacArthur 1961; Cody 1985, Kissling et al 2008). Canopy height may be important for insectivore diversity, in particular, with less space in forest vertical structure at higher elevations to support a large range of specialized foraging strategies. Understanding mechanistic ecological correlates of vegetation structure and elevation and their impacts on richness of foraging guilds needs further investigation, but at least in N eotropical birds, any


86 explanation for decreasing richness with elevation should address the overwhelming loss of insectivores. Species Turnover, Nestedness and Patterns of Compositional Similarity While species richness of trees is generally higher than birds within each plot, these differences alone do not account for the much larger discrepancy in regional diversity found between montane trees and birds; regional richness in trees is more than 2.5 times that of birds. These differences may instead be ex plained by differences in beta diversity of trees and birds along the gradient. Measures of dissimilarity between adjacent plots in bird and tree communities show that, overall, trees experience higher turnover with elevation than do birds. Because of th e short distances between plots along this steep elevational gradient, and the high dispersal capacity of trees (Clark et al 1999), the higher turnover in trees is not likely a result of dispersal limitation. Rather, increased turnover may result from hi gher diversity in tree communities combined with greater specialization to environmental conditions along gradients (e.g., temperature and rainfall), limiting establishment and growth to a narrow elevational range. Indeed, patterns of floristic composition in tree communities have been linked to edaphic characteristics in numerous studies (Potts et al. 2002 Ruokolainen et al. 2007, Bohlman et al. 2008), and specialization to edaphic characteristics has been identified as a likely driver of divergence amon g closely related species (Fine et al. 2005). Across our study area, one line of evidence suggests that soil and tree composition are associated. The lowest dissimilarity (or highest similarity) observed between adjacent plots occurred between those at 1 850 and 2000 m ( sim = 0.49), with values equivalent or more similar than pairs of plots at the same elevation (e.g., replicate plots at 1500m and 3000m have values of 0.68 and 0.49, respectively).


87 These two plots are located on soils derived from granite, whereas other plots lie on Paleozoic shales and slates (M. Silman, unpubl. data) These results are consistent with results from elevational transects on distinct geologies from Mount Kinabalu, Borneo ( Aiba and Kitayama 1999). Despite differences in lev els of species turnover in birds and trees, we found similar regions of punctuated turnover between these taxa along the gradient, suggesting that birds and trees are either associated, or both communities respond in parallel fashion to some other aspect o f the elevational gradient. The matching peaks of species turnover of birds and trees in the foothills (10001250 m) are especially notable (Figure 4 4 a ); adjacent -plots are approximately 10% more dissimilar, followed by a steady decline in adjacent plot dissimilarity until 2000 m. In birds, the foothill peak in turnover is likely driven by the loss of species in insectivorous families (e.g., Furnariidae, Thamnophilidae, Tyrannidae) and increased representation from other families (e.g., Trochilidae). A second region of elevated turnover occurs in the middle of the gradient. In trees, the high turnover between 20002250 m is likely influenced by the soil type differences, mentioned above. In birds, high turnover between 22502500 m is likely influenced by the location of the 2250 m plot on an exposed ridge with truncated vegetation. The cluster dendrogram, however, also suggests a prominent division between low and high elevation flora and fauna at this location. In birds, for example, this zone marks the appearance of genera with highelevation diversity centers (e.g., Ochthoeca, Diglossa, Conirostrum ) and replacements within genera (e.g., Grallaria Scytalopus Myioborus ). This zone roughly corresponds to the elevation of the cloud base during the dr y season, which may be responsible for shifts in moisture


88 levels, with consequences for vegetation structure and composition. The third region of higher turnover occurs at treeline, where bird composition turnover is equivalent to the foothill peak in dis similarity. This marks a primary ecotone transition in vegetation that birds appear to recognize, although this treeline shift could not be detected in the tree censuses due to the loss of stems Compared to species turnover, nestedness contributed much less to dissimilarity in birds and trees across plots or to dissimilarity in birds across survey sites. There was, however, variation in the relative contributions of turnover and nestedness to site level bird dissimilarity along the gradient. Specifically, nestedness was more prominent at higher elevations (e.g., > 2100 m), whereas dissimilarity due to turnover was greater at low elevations. Such a pattern in birds may be linked to variation in landscape characteris tics along the elevational gradient. Exposed, windswept ridges and landslides are more frequent at middle to high elevations, creating patches of truncated or regenerating vegetation across the landscape. Accordingly, our results show that vegetation str ucture uniquely explained more site level variation in bird dissimilarity (for both nestedness and turnover) in high elevation zones. This suggests that high elevation bird communities, at least to some degree, are composed of species specialized to struc tural aspects of vegetation, and generalist species that inhabit a wider range of vegetation types. Low elevation bird communities, on the other hand, tend to show a larger component of species turnover (with little nestedness) across sites, which is attr ibutable to elevational differences or co varying effects of elevat ion and vegetation structure.


89 Importance of Tree Composition, Vegetation Structure and Elevation Multiple regression on distance matrices ( MRM) and variation partitioning analysis revealed that vegetation structure, tree composition and elevational position of plots are all important predictors of bird community composition along the gradient. Moreover, these predictors covary for all bird groups, the largest fraction of variation in bir d dissimilarity was either co explained by vegetation, trees and elevation, or co explained by trees and elevation. Thus, while patterns of species turnover in birds were successfully predicted across study sites (up to 82% of variation explained), it is difficult to attribute the major fraction of explained variation to one variable over another. Within avian guilds, however, there was a moderate fraction of additional variation explained (e.g., and elevation, and this varied across groups. Granivore composition, more than other guilds, contained the most variation explained uniquely by vegetation structure, particularly by density of large stems, basal area of trees, and canopy height. Most granivores in our dataset forage on the forest floor (e.g., tinamous, woodquail) or in the canopy, moving over large distances (e.g., parrots). It is possible that understory vegetation structure could influence occurrence of ground-foraging species, while canopy height likely influence s occurrence of parrots. That said, most variation in this group remained unexplained, perhaps due to the rarity of granivores. All but two species occurred at survey sites (well below the average of 21% for al l birds). Omnivore composition was more associated with tree dissimilarity compared to other guilds. Notably, this guild contained the greatest proportion of canopy species (87% versus 2368% in other guilds). Based on these results, we find little supp ort for the hypothesis that avian foraging guilds with diets based on plants are more closely associated with plant


90 composition. Moreover, variation in insectivore composition was not more associated with vegetation structure compared to other guilds. Pro spects for Montane Birds and Trees with Climate Change Most projections of species range shifts with climate change have used abiotic variables (e.g., temperature and moisture) to predict future climate landscapes in which species will likely be distributed with new temperature regimes. Given the climate -change predictions of a change of 4-6 C for the study region over the next century (Cramer et al 2004 Urrutia and Vuille 2009), and given an adiabatic lapse rate of approximately 5.6 C per 1000 m elevation for tropical montane landscapes (e.g., Bush et al 2004), we should predict that within a hundred years, a given elevation should reflect the present day temperature environment of areas 500 m lower in elevation. Given their high mobility and dispersal abilities, we might expect birds to keep up with changing climates better than sessile taxa such as trees, but only provided that birds disperse and undergo habitat selection independent of tree composition or structural characteristics of vegetation. Our analysis has demonstrated a strong statistical correspondence between birds and aspects of both vegetation structure and tree species composition, although these effects were heavily confounded with each other and with elevation. Still, the congruence between bird and tree communities suggests that these taxa respond to the elevational gradient in similar ways, such that changes in birds may not occur independently of trees or vegetation. If this correspondence is driven by avian habitat selection, tropical birds may not show expected upslope range shifts in montane landscapes for a long time after climate change, perhaps at the expense of tolerating lower reproductive success within suboptimal thermal environments.


91 Patterns of vegetation structure w ith elevation may also impose community wide constraints on shifts in bird distributions, especially if a multitude of bird species shift upslope in response to warming. Our results show that bird diversity was positively associated to canopy height along the elevational gradient. If structural complexity is indeed linked to bird diversity, there may not be sufficient space for numerous lower elevation species to shift their ranges into smaller forests. Lower canopies at higher elevations may impose an i nflexible ceiling on the number and types of species that can share more truncated forests, and birds may collectively face a fundamental constraint on the structural space available to accommodate increasing diversity. In conclusion, our analyses sugges t that understanding birds relationship to vegetation structure and composition will be important for predicting responses to climate change. Avian range shifts will likely rely on how climate change affects forest dynamics, successional pathways, and major physiognomic vegetation types found along these mountainsides. Where information on vegetation structure is available, models of species projected distributions could incorporate vegetation parameters of future habitats that species will likely occup y, allowing alternative projections of species responses to climate change to be explored under varying scenarios of habitat constraints. In addition, understanding the variation in vegetation types occupied by bird species could be useful for projections For example, vegetation generalists will likely be the best responders to climate change, and may reliably track physiological optima as thermal and moisture regimes change, whereas vegetation specialists will be more constrained by specific habitat req uirements. With the narrow elevational distributions that so many Andean species exhibit in these heterogeneous montane


92 landscapes, such local scale projections that include information on habitat association will be necessary for precise evaluations of t he threats that tropical montane species face.


93 Table 4 1. Plot names and elevation, study transects, and number of bird auditory visual survey points and vegetation structure ar eas associated with each plot. Number of species of birds ( detected by counts and netting or by counts only) and trees found on each plot is shown. Plot Name Study Region Elevation (m) Bird survey points/plot No. bird species (counts & n etting) No. bird species (counts only) No. tree species TO_800 Tono 850 5 121 87 171 TO_1000 Tono 950 5 126 66 151 SI_1250 San Isidro 1250 8 130 95 163 SP_1500 San Pedro 1500 4 83 61 166 SI_1500 San Isidro 1500 8 99 61 168 SP_1750 San Pedro 1750 4 79 60 148 TU_1850 Trocha Union 1850 3 50 40 90 TU_2000 Trocha Union 2000 7 72 54 70 TU_2250 Tr ocha Union 2250 7 51 43 73 TU_2500 Trocha Union 2500 5 77 47 62 TU_2750 Trocha Union 2750 7 66 55 56 TU_3000 Trocha Union 3000 4 48 38 39 WA_3000 Wayqecha 3000 8 70 51 52 TU_3250 Trocha Union 3250 3 28 21 44 TU_3450 Trocha Union 3450 3 27 17 38 T able 4 2 Pairwise dissimilarity values for sim and nes across bird survey sites within 650-m elevational zones along the gradient. Number of sites per zone, mean (with standard deviation), minimum and maximum dissimilarity values are shown for each index. Simpson's index ( sim) Nestedness res ultant index ( nes) Elevational Zone No. of Sites Mean (sd) Min Max Mean (sd) Min Max 800 1450 m 42 0.70 (0.17) 0.24 1 0.03 (0.03) 0 0.24 1450 2100 m 38 0.52 (0.12) 0.12 1 0.07 (0.07) 0 0.48 2100 2750 m 52 0.44 (0.15) 0 0.83 0.14 (0.15) 0 0.8 5 2750 3400 m 36 0.44 (0.15) 0 0.83 0.13 (0.12) 0 0.72


94 Table 4 3 Proportion of explained variation (R2) in bird species dissimilarity ( sim) in models from multiple regression on distance matrices (MRM) Mod els were performed for all birds and sep arately for each guild. Significance was evaluated based on 1,000 permutations. Predictors in model All birds Frugivores Insectivores Nectarivores Granivores Omnivores Vegetation Structure (VS) 0.36 ** 0.20 0.34 0.28 0.37 ** 0.33 ** Elevation 0.76 ** 0.7 2 ** 0.71 ** 0.70 ** 0.31 ** 0.55 ** Trees 0.65 ** 0.52 ** 0.62 ** 0.60 ** 0.18 0.61 ** VS + Elevation 0.78 ** 0.73 ** 0.73 ** 0.74 ** 0.44 ** 0.61 ** VS + Trees 0.70 ** 0.53 ** 0.67 ** 0.63 ** 0.41 ** 0.68 ** Elevation + Trees 0.80 ** 0.73 ** 0.76 ** 0.74 ** 0.31 ** 0.66 ** VS + Elevation + Trees 0.82 ** 0.74 ** 0.77 ** 0.78 ** 0.44 ** 0.71 ** P < 0.01, ** P = 0.001 Tab le 4 4. Perce nt of variation in bird dissimilarity ( sim) for all birds and for each guild explained uniquely by each predictor or co explained by multiple predictors. Values were obtained by v ariation partitioning of model R2 values from multiple regression on distance matrices. Predictors All birds Frugivores Insectivores Nectarivores Granivores Omnivores Vegetation structure (VS) 2 1 1 4 13 5 Elevation 12 2 1 10 15 3 3 Trees 4 1 4 4 0 10 VS + Elevation 3 1 4 0 10 2 VS + Trees 0 0 0 1 0 1 Elevation + Trees 30 32 29 32 4 26 VS + Elevation + Trees 31 19 28 24 14 24 Unexplained 18 26 23 21 56 29


95 Table 4 5. Proportion of explained variation (R2) in bird species dissimilarity ( using sim and nes) in models from multiple regression on distance matrices (MRM) Mod els were performed at the site level for all birds within four elevational zones along the gradient. Significance was evaluated based on 1,000 p ermutations. Simpsons index ( sim) Predictors 800 1450 m 14502100 m 21002750 m 27503400 m Vegetation structure 0.22*** 0.31*** 0.05** 0.27*** Elevation 0.73*** 0.44*** 0.07*** 0.11*** VS + Elevation 0.74*** 0.56*** 0.12*** 0.30*** Nestednessresultant index ( nes) Predictors 800 1450 m 14502100 m 21002750 m 27503400 m Vegetation structure 0.04 ** 0.04* 0.10* 0.17*** Elevation 0.15*** 0.04*** 0.01 0.09*** VS + Elevation 0.18*** 0.06* 0.12* 0.23*** P < 0.05, ** P < 0.01 *** P = 0.001 Table 4 6. Percentage of variation in bird dissimilarity (using sim and nes) explained uniquely by each predictor or coexplained by multiple predictors. Models were performed at the site level, for all birds within four elevational zones. Values were obtained by variation partitioning of model R2 values from multiple regression on distance matrices. Simpsons index ( sim) Predictors 800 1450 m 14502100 m 21002750 m 27503400 m Vegetation structure 2 11 5 19 Elevation 52 25 7 3 VS + Elevation 20 20 0 8 Nestedness resultant index ( nes) Predictors 800 1450 m 14502100 m 21002750 m 27503400 m Vegetation structure 3 3 11 14 Elevation 14 3 0 6 VS + Elevation 2 1 0 3


96 Figure 4 1 Map of the study area in Manu National Park, with locations of 1ha tree plots along the elevational gradient (triangles). Bird survey sites (circles; see magnification) were located within 500 m horizontal distance (or 150 m elevation) of each tree plot.


97 Figure 42 ( a) Species richnes s for trees and birds with elevation for 15 1ha plots and (b) the r elationship between species richness of birds (SB) and trees (ST). An estimate of tree species richness for lowland terra firme in Manu National Park (approximately 400 m elevation) is gi ven as a reference (from Pitman et al 2001; averaged across 20 1-ha plots). The equation for the linear relationship, Pearson correlation coefficient (r) and pvalue are shown.


98 Figure 4 3 Species richness of avian foraging guilds across plots varyin g in (a) elevation and (b) average canopy height. Best fit lines are shown for each guild, including nectarivores, granivores, omnivores, frugivores and insectivores.


99 a) 0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1 800 1000 1250 1500 1750 1850 2000 2250 2500 2750 3000 3250 Elevational zone (m) Simpson's Dissimilarity Bird Simpson's Tree Simpson's b) 0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1 800 1000 1250 1500 1750 1850 2000 2250 2500 2750 3000 3250 Elevational zone (m) Nestedness Dissimilarity Bird Nestedness Tree Nestedness Figure 4 4 Dissimilarity values of sim (a) and nes (b) for adjacent plots along the elevational gradient for birds ( circles ) and trees (diamonds). The x axis indicates the name of the lower of the two adjacent plots compared.

PAGE 100

100 a) b) Figure 4 5 Cluster dendrograms for birds (a) and trees (b) across 1 hectare plots (labels show plot code and elevation), using the average linkage method.

PAGE 101

101 CHAPTER 5 SQUEEZED AT THE TOP: INTERSPECIFIC COMPET ITION MAY CONSTRAIN ELEVATIONAL RANGES I N TROPICAL BIRDS Introduction T ropical montane plants an d animals are typically restricted to narrow, belt like elevational distributions, often only a few hundred meters wide (Stotz et al 1996, Vsquez & Givnish 1998, Jankowski et al. 2009). This elevational specialization of species leads to high species turnover or beta diversity along tropical elevational gradients, making these landscapes important centers of global biodiversity and endemism ( Stotz et al. 1996, Myers et al. 2000). Although distributions of tropical montane species are becoming increasi ngly well documented through investigation of species richness and turnover patterns ( Patterson et al. 1998, Herzog et al. 2005, Jankowski et al. 2009, Romdal and Rahbek 2009), there is little understanding of the processes acting at range boundaries to maintain these species narrow elevational distributions. Research focused on understanding species range limits, however, is of immediate importance in the context of climate change; our ability to predict how climate changes will drive range shifts and possible extinctions hinges on understanding the factors that determine where a given species occurs. In particular, species with restricted distributions along elevational gradients may be highly vulnerable (Parmesan 2006, Colwell et al. 2008, Sekercioglu et al. 2008). Range boundaries are generated by multiple abiotic and biotic factors that influence the persistence of populations (see Holt and Keitt 2005); a combination of such factors could lead to narrow distributions of tropical montane species. Recent studies have emphasized abiotic, physiologically based explanations, in which small seasonal variation in thermal regimes leads to physiological specialization, as

PAGE 102

102 classically proposed by Janzen (1967; Ghalambor et al. 2006, McCain 2009). Tropical ectotherms, for example, appear less able than temperate ectotherms to tolerate widely varying thermal conditions and may be at greater risk from global climate change (Tewksbury et al. 2008, Huey et al. 2009). Likewise, range boundaries may be influenc ed by biotic interactions (Case et al. 2005, Price and Kirkpatrick 2009). In the tropics, direct interspecific competitive interactions at range boundaries constitute another classic hypothesis of what limits species distributions in montane landscapes ( Terborgh 1971, Diamond 1973). Such interactions should result in elevational gradients that are partitioned into nonoverlapping ranges of competing species, compressing the distributions of any single competitor. Along many tropical mountainsides, the re is strong evidence for such replacements between closely related species ( Bull 1991), especially in birds (Terborgh 1971 ). Demonstration of range compression of species in areas with competitors present, and range expansion where competitors are absent, is consistent with this hypothesis (Terborgh and Weske 1975, Remsen and Graves 1995), but experimental support for competitive interactions between species in replacement zones is still lacking. Here we test a prediction of the hypothesis that competit ive interactions, in the form of interspecific aggression, determine species range limits along an elevational gradient in the Tilarn Mountains of Costa Rica. We focus our study on two passerine genera with species that show distinct elevational replace ments: woodw rens (Henicorhina Troglodytidae) and nightingale-thrushes ( Catharus Turdidae). Using a series of song playback experiments to detect aggressive territorial behaviors, which

PAGE 103

103 may reflect underlying competitive interactions (Martin and Martin 2001) we tested whether territorial individuals in species with elevational replacements respond aggressively to songs of congeners where their ranges meet. We analyzed each species reaction to congener songs at increasing distances from replacement zon es to determine whether responses were learned or, alternatively, whether they might reflect misdirected intraspecific aggression (the mistaken identity hypothesis of Murray 1971). Finally, we evaluated whether interspecific responses of species pairs at the replacement zone were asymmetric in their level of aggression, a likely indication of interspecific behavioral dominance ( Robinson and Terborgh 1995). Methods Study Area and Target Species The Tilarn M ountains of n orthwestern Costa Rica (1018N, 8445W; max altitude 1850 m) run northwest to southeast, with the continental divide separating the leeward Caribbean slope from the drier Pacific slope. The 36 km2 study area ranges from 10001700 m altitude on the Pacific slope within 3.6 km of the cont inental divide. Th e study area shows dramatic changes in moisture, averaging 4000 -6000 mm rainfall annually at the ridgetop cloud forest and declining approximately 1000 mm km1 downslope (see Jankowski et al. 2009 for more details) C ongener ic species wi th adjacent, nonoverlapping elevational distributions in the study area were selected for their patterns of replacement, relatively high local population densities, and accessibility of their replacement zones along the mountainside. Our focal species were two wood wrens, Henicorhina leucosticta (White breasted Wood-Wren) and H. leucophrys (Gray -breasted WoodWren) and three nightingale thrushes, Catharus aurantiirostris (Orange-billed Nightingale -Thrush), C.

PAGE 104

104 mexicanus (Black headed Nightingale -Thrush) and C. fuscater (Slaty -backed Nightingale -Thrush) All of these s pecies are largely restricted to forest understory and are partially to entirely insectivorous (Stiles et al. 1989). Along the elevational gradient, H. leucosticta and C. aurantiirostris i nhabit drier more seasonal low elevation forest, whereas H. leucophrys and C. fuscater inhabit high elevation moisture-saturated cloud forest. C. mexicanus has a narrow elevational distribution on the Pacific slope of 75125 m (max. 600 m horizontal dist ance) between the other two nightingale-thrushes. Territory Mapping, Playback Stimuli and Experiments Territories of target species were located at their replacement zones, where individuals of each species are in daily contact with congeners and at varyi ng distances from replacement zones, up to 1.5 km (horizontal distance) within species elevational ranges. For C. mexicanus no territories were more than 600 m from the replacement zone because of this species extremely narrow range. Territories of C. fuscater were not studied away from the replacement zone because of the lack of behavioral responses toward its lower elevation congener, C. mexicanus at the replacement zone (see Results) Prior to playback experiments, transects within the study site w ere walked daily, and the locations of singing individuals of target species were marked using a GPSMap 60CSx GPS unit (with an accuracy of 6 m in forest). These territory locations were revisited on multiple days thereafter. During this time, individuals movements within their territory and simultaneous singing events with neighbors were noted to estimate the territory s center and to approximate its boundaries. Because our target species regularly sing throughout the morning from well-defined areas, we are confident that our method of territory mapping by simultaneous singing events with neighbors, coupled with observations of individuals movements, is a reliable approach

PAGE 105

105 to distinguish territories of unbanded birds. Each territory along the gradie nt was assigned a proximity value to the replacement zone, defined by the distance between the territory center and that of the nearest congener using ArcGIS 9.2 (ESRI 2007). Songs used for playback stimuli were recorded within the study area in May o f 2007 and 2008. Recordings were made from within 20 m of singing individuals not involved in interactions with neighbors using a parabolic dish and microphone and an Edirol R -09 digital recorder. Songs were filtered to remove low -frequency noise, gener ally below 750 Hz, and other unwanted noises such as other s inging birds using Raven Pro 1.3 (Cornell Lab of Ornithology 2003). W aveforms in some recordings were amplified so that all recordings could be broadcast at equal volume. Each recorded song was presented as a playback stimul us to only one individual per species following recommendations by Kroodsma et al. (2001) Playback experiments were conducted in May -June of 2007 -2008 during the peak of the local breeding season when birds are active ly sing ing and defend ing established territories. Each individual was tested on two days usually within a five -day period Tests consisted of an 8minute observation period of the focal bird, during which no stimulus was given, followed immediately by e ither an 8-minute heterospecific (hereafter congeneric ) or conspecific playback broadcast by a speaker within the territory. The second test consisted of another 8minute observation period followed by whichever playback was not performed on the first v isit. The order of presentation of congener and conspecific playback stimuli was randomized. O bservation and playback periods for each trial were recorded using an Edirol R 09 digital recorder and Sennheiser ME -66 microphone mounted on a tripod. Playbac k songs were played from

PAGE 106

106 a second recorder connected to a playback speaker (Dean Markley GT1000 Micro Amp) placed 5 m from the observers and approximately 15 m from the focal individual. S ong stimul i w ere broadcast at approximately 70-80 dB SPL at 5 m (varying slightly across species) for up to 3 min or until the focal bird approached to within 5 m of the speaker. Playbacks were conducted from 05:30 to 14:00, avoiding periods of heavy wind or rain. First, t erritorial male s were located, usually by song. If the individual was involved in a counter -singing bout with a neighbor, we delayed the trial for several hours. If disputes persisted, no trials with that individual were attempted that day If non-target individuals ( congener ic or conspecific neighbor s ) approached the territory in response to the playback stimulus, we were thereafter unable to distinguish whether the target individual was responding to our playback or the presence of a non-target bird. In these situations, the playback was aborted and attempted on a different day. During the 8 -minute observation and playback periods, all movements of the focal bird were mapped by noting the distances of the bird from the speaker and the amount of time spent in each location. B ehavioral variables summ arized from these data includ ed closest approach to speaker and latency to approach the speaker to within 10 m. In 2008, we used the same protocol as in 2007, except that we also conducted control playbacks for target individuals to evaluate use of pre playback observation periods as a negative control Stimuli for control playbacks used locally recorded songs from one of two common species: Basileuterus culicivorous (Goldencrowned Warbler, Family Parulidae) or Hylophilus decurtatus (Lesser Greenlet Family Vireonidae). These species have wide elevational distributions and are sympatric but not known to

PAGE 107

107 interact with the target species. For all behavioral variable s examined, we found no differences between control playbacks and observation periods prior to control playbacks nor between control playbacks and observation periods prior to congener or conspecific playbacks (Kruskal Wallis non -parametric ANOVA; for all species and variables, values range from 0.28 6.0, d.f. = 3, Ps range from 0.11 0.96). Therefore, we pooled the behavioral data recorded during observation periods prior to congener and conspecific playbacks to use as a comparative stimulus type ( hereafter control) in statistical models of behavioral response to playback. Statistical Analysis We analyzed behavioral response to playbacks for each species pair using both general and generalized linear mixed models performed in SAS/STAT v. 9.2 (SAS Institute 2008) for two groups: 1) individuals close to the replacement zone (woodwrens 100 m; nightingale -thrushes 200 m) and 2) all individuals at varying distances from the replacement zone. Each response variable (i.e., closest approach to speaker and latency to approach to within 10 m) was analyzed separately for each pair of cong eners sharing a range boundary The response variable c losest approach to the speaker was square-root transformed to achieve normality and was modeled using a g eneral linear m ixed m odel for individuals close to the replacement zone and for all individuals The response variable l atency to approach the speaker was analyzed using a logistic regression which modeled the probability of a response during the playback period, given the explanatory variables. We used this procedure because many individuals did not approach to within 10 m (i.e., no response). Then, using only responding individuals close to the replacement zone, we performed a second analysis using a general linear

PAGE 108

108 mixed model to examine the time to approach to within 10 m (log transformed) given the explanatory variables. We report mixed m odel results for time to approach for individuals close to the replacement zone, and we report logistic regression results for the probability of a response during the playback period when including all individuals. Explanatory variables for all models included stimulus (congener, conspecific or control ), species of the target individual distance of the target individuals territory from the replacement zone (only for models including all individuals), and interaction terms Individual was included as a repeated subject. For models with all individuals, if species was a significant term in the model ( P < 0.05), each species was then analyzed separately to avoid modeling threeway interactions Post ho c tests for direct effects or interactions with categorical explanatory variables used least squares mean difference with a Tukey adjustment to test for significant differences among levels. Contrast statements were used to test for significant difference s among levels in the distance stimulus interaction term with distance as a continuous variable. See Table 5 1 for sample sizes of individuals tested for each species. Results Individuals of each genus ( Henicorhina and Catharus ) responded aggressively to playbacks of congener songs. In woodwrens, individuals with territories close to the replacement zone responded aggressively to playbacks of their congeners songs, approaching on average 71% closer to the speaker and responding 85% faster compared to t he control period (Figure 5 -1; Table 5 2 ; t11 = 6.8 and 7.8, respectively, p s < 0.0001). Woodwrens responded to congener songs by approaching the speaker quietly with short flights through the understory, then singing near the speaker for several minute s. These responses to congener songs were not significantly different

PAGE 109

109 from responses to conspecific songs (closest approach: t11 = 2.2, p = 0.11; latency to approach: t11 = 1.04, p = 0.32). When considering individuals at varying distances from the repla cement zone, we found that woodwrens located farther from the replacement zone showed a decreased response to congener playbacks (Figure 5 -2; Table 53 ; contrasts of congener to control and conspecific playback s : for closest approach, F1,49 = 39.01, p < 0 .0001; for latency: F1,141 = 12.2, p < 0.0006). At distances > 1 km from the replacement zone, responses to congener playback did not differ from the control period (for closest approach and latency, t8 = 1.7, p s > 0.26). The low elevation C. aurantii rostris and its middleelevation congener, C. mexicanus likewise responded to congener playbacks (Fig ure 5 -1 ; Table 5 -2 ). These responses were strongest for C. aurantiirostris which approached 69% closer to the speaker and responded 75% faster compared to control periods (t11 = 5.8 and t22 = 3.5, respectively, p s < 0.02). In this species, responses to congener and conspecific playbacks were similar in both the closest approach and latency to respond (t12 = 0.83 and t22 = -0.06, respectively, p s > 0.95 ). C. mexicanus on average, approached 27% closer and responded 57% faster to congener playbacks compared to controls (t11 = 4.0 and t22 = 3.2, respectively, p s species was not as aggressive as the response to conspecific playback ( closest approach: t12 = 4.0, p = 0.02; latency: t22 = 4.27, p = 0.004). When considering individuals at varying distances from t he replacement zone, we found that these two species of nightingale -thrushes showed weaker responses to congener playbacks farther from the replacement zone (Figure 5 -2; Table 5 3 ; closest approach C.

PAGE 110

110 aurantiirostris : F1,25 = 10.8, p = 0.0003, C. mexicanus : F1,25 = 7.8, p = 0.01; latency to approach significant for C. aurantiirostris only: F1,50 = 7.1, p = 0.01). The middleelevation C. mexicanus and its highelevation congener, C. fuscater responded differently to congener playbacks. Compared to control periods, C. mexicanus approached on average 28% closer and responded 51% faster to congener playbacks (Figure 5 -1; Table 5 2 ; t21 = 3.6 and t41 = 4.5, respectively; p s response to congener playbacks was nonetheless weaker than its response to conspecific playbacks (Fig ure 5 1; closest approach: t21 = 4.9, p = 0.001; latency: t41 = 7.5, p < 0.0001). In this species, there was no change in the strength of the response to congener playbacks farther from the replacement zone ( Table 5 3 ). In C fuscater response to congener playbacks did not differ from the control period ( Fig ure 5 -1; Table 5 -2 ; closest approach: t21 = 0.6; latency: t41 = 0.28; p s = 0.99) indicating no detectable aggressive response of this spe cies toward C. mexicanus Disc ussion These results demonstrate aggressive interactions between species that replace each other along elevational gradients, consistent with the hypothesis that such biotic interactions are important in determining species range limits in tropical montan e landscapes ( Terborgh and Weske 1975). The influence of interspecific competition on the spatial arrangement of species in sympatry has been suggested for many taxa, including fish ( Bay et al. 2001), amphibians ( Cunningham et al. 2009 ), reptiles (Langkil de and Shine 2004 ) and mammals ( Brown 1971 ). In birds, such interactions can determine local habitat selection ( Martin and Martin 2001), settlement patterns of migrants arriving to breeding grounds ( Fletcher 2007), and spatial partitioning of closely rela ted species along successional gradients ( Robinson and Terborgh 1995). In

PAGE 111

111 particular, song playback experiments in birds have been shown to be an appropriate method to assess underlying competitive interactions between species; indeed, habitat segregation of species resulting from this behavior can confer higher fitness compared to individuals whose territories overlap with those of congeneric competitors (Martin and Martin 2001). Our observations of interspecific territorial aggression at range boundarie s of species with replacements support the presence of competitive interactions and the importance of this biotic interaction in maintaining segregated distributions. Our study is the first to provide experimental support for the hypothesis that such inte rspecific competitive interactions could reinforce range boundaries that segregate species along large -scale elevational gradients in the tropics Our results are also unique in demonstrating that interspecific aggression varies greatly over small spatial scales. In particular, we discovered that interspecific aggression occurs primarily when individuals are in close contact with congeners and weakens with increasing distance from zones of replacement. This pattern is not consistent with the alternative hypothesis of response to congener song that the responses simply reflect mistaken identity of congeners and misdirected intraspecific aggression (e.g., Murray 1971). Rather, a ssuming that our populations are not genetically divergent within the 4-km g radient of our site (which would be unlikely at this spatial scale in a primarily forested landscape lacking major dispersal barriers ), our results suggest a learned component to aggressive interactions at range boundaries and a behavioral flexibility that allows finely tuned responses corresponding to the likelihood of encountering heterospecifics (see also Richards 1979) In this situation,

PAGE 112

112 the development of such behavioral interactions likely depends upon local densities of interacting species along gr adients. Species pairs tested at their replacement zone did not respond equally to playbacks of their congeners songs. Whereas wood wrens exhibited symmetric levels of interspecific aggression, nightingale -thrushes showed asymmetry in their territorial r esponses to congeners. The highelevation C. fuscater showed no response toward songs of the middleelevation C. mexicanus whereas C. mexicanus responded aggressively to C. fuscater songs. Such asymmetries suggest that interspecific dominance in the agg ressive species could limit some species to portions of the elevational gradient. In light of climate change, interspecific aggression could have important implications for species with behavior mediated elevational range limits, especially if highelevati on species were subordinate. If warming in montane climates allow s upslope range expansion by d ominant competitors, then high elevation species could be forced to still higher elevations and become dependent on progressively smaller land areas to sustain viable populations. If interspecific aggression between congeners is widespread, m any species could face such a scenario. Considering only species that are currently threatened with extinction, we estimate that 108 of 334 tropical montane species occurri ng at high elevations (approximately 9% of the 1184 threatened species, worldwide) have elevational ranges that are bordered by widespread low elevation congeners ( data from BirdLife International 2000). Dominant congeners at higher elevations could simil arly prevent upslope expansion of subordinate species, squeezing middle elevation species between an expanding suboptimal abiotic environment at the

PAGE 113

113 lower bounda ry and a resistant biotic upper boundary. In this scenario, dominant high elevation species would be able to hold off upslope movement of lower elevation species as true kings of the hill for much longer than predicted by climate models alone. In conclusion, our results point to the importance of including biotic interactions in predicting community responses to climate change; doing so may be particularly important for diverse tropical systems. In the Tilarn Mountains, there is already evidence for climate driven population declines and elevational range shifts across many taxa, including birds (Pounds et al. 1999). Under a moderate warming scenario of 3 C over t he next century ( Solomon et al. 2007 ), montane species can be expected to shift their ranges 500 m upslope in response to rising temperatures (assuming an adiabatic lapse rate of 6 C/ 1000 m altitude; Colwell et al. 2008; Gasner et al. 2010). For the high elevation species in our study, this corresponds to shifts nearly as large as their elevational distributions. While range shifts in tropical montane regions will undoubtedly be affe cted by a changing abiotic environment and species physiological tolerances (Tewksbury et al. 2008), the constraints imposed by biotic interactions will likely limit the ability of m any species to track optimal abiotic conditions.

PAGE 114

114 Table 5 1. Number of playback trials conducted per species (includes control, congener, and conspecific trials) and number of individuals tested per species. Two groups of individuals are distinguished: individuals close to the replacement zone (wood wrens thrushes individuals at all distances from the replacement zone. Playback trials for C. mexicanus are divided into playbacks associated with C. aurantiirostris (low elevation congener) and those associated wi th C. fuscater (highelevation congener). Playbacks with C. fuscater were not conducted more than 200m from the replacement zone. Number of Trials Number of Individuals Tested Target species Close to Replacement Zone All Distances from Replacement Zone Close to Replacement Zone All Distances from Replacement Zone Henicorhina leucosticta 28 100 8 25 Henicorhina leucophrys 24 108 6 27 Catharus aurantiirostris 16 108 5 27 Catharus mexicanus ( vs. C. aurantiirostris ) 40 108 10 27 Catharus mexicanus ( vs. C. fuscater ) 72 88 18 22 Catharus fuscater 24 --6 --

PAGE 115

115 Table 5 2. Mixed model results for each species pair for closest approach to speaker and latency to approach speaker for individuals close to the replacement zone ( wrens, m for nightingale thrushes). Variables in the model include Stimulus, or the song stimulus delivered to the target individual, and Species, which is the species of the target individual. Total N shows the total sample size followed by the number of in dividuals tested (including both species). Statistics from post hoc tests are provided in text. Closest Approach Latency to Approach H. leucosticta/H. leucophrys H. leucosticta/H. leucophrys Stimulus F2,11 = 142.0; P < 0.0001 F2,11 = 49.4; P < 0.0001 Species F1,11 = 1.93; P = 0.19 F1,11 = 1.05; P = 0.33 Total N 50 obs, 13 ind 50 obs, 13 ind C. aurantiirostris/C. mexicanus C. aurantiirostris/C. mexicanus Stimulus F2,12 = 42.1; P < 0.0001 F2,22 = 29.7; P < 0.0001 Species F1,12 = 20.7; P = 0.0007 F1,11 = 0.06; P = 0.81 Stimulus x Species F2,12 = 4.3; P = 0.04 F2,22 = 3.0; P = 0.07 Total N 56 obs, 14 ind 52 obs, 13 ind C. mexicanus/C. fuscater C. mexicanus/C. fuscater Stimulus F2,21 = 71.4; P < 0.0001 F2,41 = 63.5; P < 0.0001 Species F1,21 = 1 .5; P = 0.24 F2,21 = 7.0; P = 0.02 Stimulus x Species F2,21 = 3.1; P = 0.07 F2,41 = 2.9; P = 0.06 Total N 88 obs, 23 ind 90 obs, 23 ind

PAGE 116

116 Table 5 3. Mixed model results for each species pair for closest approach to speaker and latency to approach speak er for individuals at varying distance from the replacement zone (shown as Distance below). Other variables in the model include Stimulus, the song stimulus delivered to the target individual, and Species, which is the species of the target individual. Total N shows the total sample size followed by the number of individuals tested (including both species). Statistics from post hoc tests are provided in text. Closest Approach Latency to Approach H. leucosticta / H. leucophrys H. leucosticta / H. leuco phrys Stimulus F 2,49 = 24.7; P < 0.0001 F 1,141 = 8.9; P = 0.0002 Species F 1,49 = 0.1; P = 0.73 F 1,141 = 2.2; P = 0.14 Distance F 1,49 = 14.0; P = 0.0005 F 1,141 = 0.14; P = 0.70 Stimulus x Distance F2,49 = 19.5; P < 0.0001 F2,141 = 8.20; P = 0.0004 To tal N 198 obs, 52 ind 198 obs, 52 ind C. aurantiirostris C. mexicanus C. aurantiirostris C. mexicanus Stimulus F 2,25 = 43.2; P < 0.0001 F 2,25 = 25.5; P < 0.0001 F1,50 = 9.4; P = 0.003 F1,52 = 5.8; P = 0.02 Distance F1,25 = 4.7; P = 0.04 F 1,25 = 0.02 ; P = 0.89 F1,50 = 2.0; P = 0.16 F1,52 = 0.02; P = 0.9 Stimulus x Distance F2,25 = 9.0; P = 0.001 F 2,25 = 4.1; P = 0.03 F1,50 = 7.1; P = 0.01 F1,52 = 1.5 ; P = 0.2 Total N 105 obs, 27 ind 108 obs, 27 ind 79 obs, 27 ind 81 obs, 27 ind C. mexicanus ( C. f uscater ) C. mexicanus ( C. fuscater ) Stimulus F 2,20 = 25.5; P < 0.0001 ---F 1,40 = 12.1; P < 0.00 0 1 ---Distance F 1,20 = 2.2; P = 0.15 ---F 1,41 = 0.08; P = 0.78 ---Stimulus x Distance F2,20 = 0.34; P = 0.72 ---ns* ---Total N 84 obs, 22 ind ---64 obs, 22 ind ---* Interaction term not significant; without this term stimulus is significant in simpler model

PAGE 117

117 C. fuscater C. mexicanus C. aurantiirostris X X H. leucosticta H. leucophrys Figure 51. Response to playback trials by individuals with territories near replacement zone s ( wrens, -thrushes). Bars show closest approach to the speaker (means SEM) during control (black), congener (gray) and conspecific (white) trials for each pair of species tested. Values with different letters are significantly different for that species ( p < 0.02; least squares mean difference with Tukey adjustment). Schematics to right depict the elevational location of species, with arrows of different thickness indicating the relative strength of response to c ongener playbacks between species pairs at the replacement zone and x to indicate species with no response to congeners.

PAGE 118

118 H. leucosticta 0 500 1000 1500 2000 Closest Approach (m) 0 5 10 15 20 25 H. leucophrys 0 200 400 600 800 0 5 10 15 20 25 Congener Conspecific C. mexicanusDistance from Replacement Zone (m) 0 100 200 300 400 500 600 700 0 5 10 15 20 C. aurantiirostrisDistance from Replacement Zone (m) 0 500 1000 1500 2000 2500 Closest Approach (m) 0 5 10 15 20 Figure 5 2. Closest a pproach to the speaker in response to congener and conspecific stimuli for individuals at increasing distances from the replacement zone. Filled circles show responses of each individual to congener playbacks (with solid best fit lines), and hollow squares show responses of individuals to conspecific playback trials (with da shed best fit lines). Species pairs shown have significant stimulus -distance interactions, where post hoc contrasts of congener response to combined control and conspecific response for each species a re significant with p s

PAGE 119

119 CHAPTER 6 CONCLUSION The overarching goal of this dissertation is to develop ways to study the causes and maintenance of range boundaries along elevational gradients, and to understand the implications of ecological reinforcement of ranges f or patterns of diversity at the landscape level. The mechanistic approach to understanding range boundaries outlines a target for future work. The underlying conceptual framework of this approach is based on classic concepts of the fundamental and realiz ed niches of species. It proposes a way to estimate the physiological niche of bird species in response to elevation, based on metabolic rates and thermal tolerances. This working physiological niche describes the species idealized distribution if it wer e limited only by the range of physical conditions it could tolerate. Using these hypothesized distributions as a baseline, it is then possible to determine whether species show different realized distributions along gradient s U sing a combination of field experiments and intensive data collection on life history and behavioral traits, researchers can begin to evaluate additional biotic constraints that could be acting on species range limits. We highlight the forces of interspecific competition, mutualis tic interactions, and nest predation and consider how each of these factors may be important in maintaining range boundaries of Neotropical montane birds Analysis of elevational range sizes, response shapes and position of species distributions along the gradient revealed that e levational range sizes were largest for cloud forest birds, and these were nearly twice as large as those of species found in foothill and highland elevations. Additional tests against middomain effect models should be conducted t o determine whether this pattern is biological. M ost species

PAGE 120

120 exhibited symmetric response curves with elevation, consistent with the expectations of ecological niche theory. A moderate proportion of species showed asymmetric response curves which were t runcated at one range boundary; however there was no tendency for truncations to occur at lower or upper boundaries. Our analyses suggest that both habitat association and competitive interactions may set range boundaries for selected species, but for ma ny others, these particular biotic factors appear to be weak in determining elevational range limits along the Andean gradient. Congeners found along the gradient for example, typically showed little range overlap, and many congeners were separated by elevational gaps Only two pairs of congeners showed evidence of repulsion interactions through range truncations at replacement zones. The lack of strong competitive interactions maintaining the range boundaries of congeners in the Andes contrasts sharply with results from field experiments conducted in the Tilarn Mountain range of Cost Rica. Results of heterospecific playbacks along this gradient with Henicorhina and Catharus spp showed that individuals at replacement zones exhibit aggressive territori al behavior in response to songs of congeners. As distance from replacement zones increased, aggression towards congener song decreased, suggesting a learned component to interspecific aggression. Additionally, aggressive responses in Catharus were asymm etric, indicating interspecific dominance. These results provide experimental evidence consistent with the hypothesis that interspecific competitive i nteractions restrict ranges of N eotropical birds. They also suggest that densities of congeners at range boundaries play an important role in the development of these interactions. The contrasting results of the strength of competition in maintaining range boundaries between Central America and the Andes

PAGE 121

121 invites future investigation into the conditions along elevational gradients that influence these biotic interactions. Few assessments of diversity patterns and congruence in species composition across different taxonomic groups exist, especially for such diverse biological systems. Furthermore, examination of associations of bird species to vegetation structure and composition has rarely been attempted in the tropics. The assessment of cross -taxon congruence in bird and tree communities along the Manu gradient found that s pecies richness was generally high er in trees than in birds. While diversity in both taxa decreased with elevation, tree richness showed a low elevation plateau before declining at higher elevations. Tree species had narrower distributions and higher turnover compared to birds, but patterns of turnover along the gradient were congruent between taxa. Nestedness contributed much less to bird and tree dissimilarity, though in birds, the nestedness component of dissimilarity increased at higher elevations and was best explained by vegetation structure. Multiple regression of distance matrices showed that tree composition, vegetation structure and elevation were all important predictors of bird composition, explaining 82% of variation in bird dissimilarity across plots. For all groups of bir ds, covariation between vegetation structure, trees, and elevation made the greatest cont ribution to explained variation, though additional variation was explained uniquely by tree composition (in omnivores), vegetation structure (in granivores) and by elevation (in frugivores, insectivores and nectarivores). While trees appear to be more specialized than birds along the elevational gradient, the patterns of change with elevation in birds and trees are similar.

PAGE 122

122 This research on the distributions and range boundaries of species along montane gradients is highly relevant to understanding the response of these montane communities to future climate change. N early 30% of the worlds threatened birds are restricted to narrow elevational ranges in tropical mountains and species are expected to shift their ranges upslope to follow optimal climate conditions as warming continues Projections of such species range shifts typically assum e that species will track changing thermal regimes. The degree to which we can r ely on these projections, however, also depends upon understanding how species respond physiologically to changing temperatures, but such thermal tolerances are virtually unknown for tropical birds. Species interactions also likely impose limits to species elevational distributions, constraining ranges in ways not predicted by physiological tolerances alone. Future research in distributional modeling should be directed not only toward collection of data on species occurrence or abundance but also on phy siological traits and species interactions While filling the data void will be a difficult task for tropical birds, and even more so for poorly known taxa, increasing data availability through focused empirical research will allow multiple mechanisms pot entially influencing range limits to be assessed simultaneously

PAGE 123

123 APPENDIX A ELEVATIONAL RANGES O F SPECIES IN MANU Table A -1. Species detected in forested sites along the elevational gradient in the Manu study area. The number of sites where each specie s was detected (by netting or counts) is indicated, with each species observed elevational minimum, maximum, midpoint and range. For species with sufficient detections to model response shapes, the modeled minimum, maximum, optimum, probability of occurr ence, elevational range, and HOF model type is given. The optimum indicates the elevation where species show the ir maximum probability of occurrence. Species are listed in taxonomic order with nomenclature following the American Ornithologists Union South American species classification (Remsen et al. 2010). No. Sites Detected Observed Elevational Range Modeled Elevational Range Species Netted Counts Midpoint Min. Max. Range Optimum Prob. of Occurrence Min. Max. Range HOF Type Nothocercus nigrocapill us 0 8 2377 1852 2901 1049 Tinamus tao 0 14 933 805 1061 256 975 0.79 881 1069 188 IV Tinamus osgoodi 0 12 1423 1194 1651 457 1476 0.36 1251 1702 451 IV Tinamus guttatus 0 8 968 923 1013 90 Crypturellus soui 0 8 894 818 969 151 Crypturellus obsoletus 0 35 1994 975 3013 2038 1590 0.51 1246 2394 1148 V Penelope montagnii 0 17 2334 1676 2992 1316 2511 0.18 1877 3143 1265 IV Penelope jacquacu 0 4 895 829 960 131 Aburria aburri 0 9 1639 1332 1945 613 Ortalis guttata 0 1 1331 1331 1331 0 Odontophorus speciosus 0 13 1504 1061 1947 886 1397 0.28 1063 1729 665 IV Odontophorus balliviani 0 4 2664 2537 2791 254 Odontophorus stellatus 0 2 901 893 909 16 Elanoides forficatus 0 1 818 818 818 0 Accipiter striatus 0 1 2010 2010 2010 0 Buteo magnirostris 0 1 823 823 823 0 Micrastur ruficollis 0 16 1418 823 2012 1189 805 0.28 805 1877 1072 II Ibycter americanus 0 1 895 895 895 0 Claravis mondetoura 0 1 2673 2673 2673 0 Patagioenas fasciata 0 28 2715 2087 3342 1255 2650 0.36 2175 3122 947 IV Patagioenas plumbea 0 48 1382 818 1945 1127 1369 0.75 946 1791 845 IV

PAGE 124

124 Table A -1. Continued No. Sites Detected Observed Elevational Range Modeled Elevational Range Speci es Netted Counts Midpoint Min. Max. Range Optimum Prob. of Occurrence Min. Max. Range HOF Type Patagioenas subvinacea 0 4 1021 805 1236 431 Leptotila rufaxilla 0 1 892 892 892 0 Geotrygon frenata 1 20 2164 1501 2828 1327 1995 0.27 1371 26 16 1244 IV Geotrygon montana 7 10 1136 877 1395 518 1225 0.90 1126 1322 196 IV Ara ararauna 0 1 829 829 829 0 Ara militaris 0 4 1169 1124 1213 89 Ara severus 0 15 1027 818 1236 418 805 0.69 805 1105 300 II Primolius couloni 0 5 899 837 9 60 123 Aratinga mitrata 0 22 2663 1911 3414 1503 3414 0.63 2699 3414 715 II Aratinga leucophthalma 0 6 842 805 878 73 Bolborhynchus lineola 0 14 2467 1551 3383 1832 3414 0.12 2029 3414 1385 III Pionus menstruus 0 20 1021 805 1236 431 80 5 0.68 805 1251 446 II Pionus tumultuosus 0 5 2345 1657 3032 1375 Amazona mercenaria 0 33 1793 826 2760 1934 805 0.24 805 2297 1492 III Amazona farinosa 0 2 828 818 837 19 Piaya cayana 0 9 1328 876 1780 904 Dromococcyx pavoninus 0 4 1367 1279 1454 175 Ciccaba albitarsis 0 1 2908 2908 2908 0 Glaucidium bolivianum 1 2 2725 2458 2992 534 Lurocalis rufiventris 0 1 2575 2575 2575 0 Eutoxeres condamini 20 1 1461 923 1998 1075 1373 0.65 982 1767 785 I V Threnetes leucurus 4 1 1072 893 1250 357 Phaethornis ruber 5 11 890 805 975 170 805 0.70 805 980 175 II Phaethornis hispidus 1 1 892 837 946 109 Phaethornis guy 22 11 1305 946 1663 717 1280 0.85 1002 1560 558 IV Phaethornis koepckeae 3 0 976 853 1100 248 Phaethornis superciliosus 4 2 1048 893 1202 309 Doryfera ludovicae 15 1 1382 1100 1663 563 1461 0.64 1194 1731 538 IV Doryfera johannae 3 0 1215 930 1501 571 Schistes geoffroyi 13 4 1293 805 1781 976 1499 0 .46 1098 1898 800 IV Colibri thalassinus 3 34 2194 1486 2901 1415 1755 0.90 1562 2243 681 V

PAGE 125

125 Table A -1. Continued No. Sites Detected Observed Elevational Range Modeled Elevational Range Species Netted Counts Midpoint Min. Max. Range Optimum Prob. of O ccurrence Min. Max. Range HOF Type Colibri coruscans 3 0 2151 1309 2992 1683 Heliothryx auritus 1 0 1129 1129 1129 0 Heliangelus amethysticollis 7 56 2921 2458 3383 925 2968 0.82 2553 3385 832 IV Phlogophilus harterti 17 25 1278 893 166 3 770 1100 0.95 972 1355 384 V Adelomyia melanogenys 25 13 1942 1129 2756 1627 1936 0.93 1338 2531 1194 IV Aglaiocercus kingi 5 1 1961 1396 2527 1131 Lesbia nuna 0 1 2685 2685 2685 0 Chalcostigma ruficeps 3 33 2643 2184 3101 917 2540 0.5 2 2183 2897 715 IV Metallura tyrianthina 7 21 2936 2458 3414 956 3414 0.33 2712 3414 702 III Haplophaedia assimilis 2 0 1960 1393 2527 1134 Aglaeactis cupripennis 0 1 2991 2991 2991 0 Coeligena coeligena 22 4 1878 1202 2554 1352 1746 0.9 0 1302 2193 891 IV Coeligena torquata 4 6 2673 2458 2887 429 Coeligena violifer 9 10 2918 2458 3377 919 2931 0.20 2608 3257 650 IV Boissonneaua matthewsii 1 4 2740 2507 2972 465 Ocreatus underwoodii 25 40 1557 930 2184 1254 1713 0.88 136 9 2060 691 IV Heliodoxa schreibersii 4 1 1300 1100 1501 401 Heliodoxa aurescens 1 0 1129 1129 1129 0 Heliodoxa leadbeateri 22 15 1500 1100 1899 799 1434 0.80 1110 1757 646 IV Chaetocercus mulsant 1 1 2091 1401 2781 1380 Chloros tilbon mellisugus 1 1 1105 1007 1202 195 Klais guimeti 1 0 1202 1202 1202 0 Campylopterus largipennis 3 0 1039 877 1202 325 Thalurania furcata 8 2 1101 853 1350 498 Taphrospilus hypostictus 4 0 1265 1129 1401 272 Amazilia viridicauda 1 0 1501 1501 1501 0 Chrysuronia oenone 3 0 1151 1100 1202 102 Pharomachrus auriceps 0 21 2474 1899 3049 1150 2535 0.25 1963 3106 1143 IV Pharomachrus antisianus 0 11 1766 1541 1991 450 1721 0.50 1525 1919 394 IV T rogon melanurus 0 6 852 826 878 52 Trogon viridis 0 2 872 866 878 12

PAGE 126

126 Table A -1. Continued No. Sites Detected Observed Elevational Range Modeled Elevational Range Species Netted Counts Midpoint Min. Max. Range Optimum Prob. of Occurrenc e Min. Max. Range HOF Type Trogon violaceus 0 6 948 805 1090 285 Trogon curucui 0 3 984 878 1090 212 Trogon collaris 1 10 1178 866 1489 623 1079 0.29 844 1311 467 IV Trogon personatus 1 60 2171 1350 2992 1642 1734 0.60 1575 3015 1440 V C hloroceryle aenea 1 0 930 930 930 0 Baryphthengus martii 3 13 1257 892 1621 729 1251 0.39 1003 1499 496 IV Momotus aequatorialis 2 0 1337 1309 1365 56 Galbula cyanescens 1 3 1030 805 1255 450 Nystalus striolatus 0 2 997 980 1013 33 Malacoptila fulvogularis 4 8 1497 1213 1781 568 Micromonacha lanceolata 0 2 1459 1139 1779 640 Nonnula ruficapilla 1 0 946 946 946 0 Capito auratus 0 6 1025 960 1090 130 Eubucco versicolor 4 13 1485 1110 185 9 749 1580 0.31 1267 1893 626 IV Ramphastos tucanus 0 1 818 818 818 0 Aulacorhynchus prasinus 0 8 1128 826 1429 603 Aulacorhynchus derbianus 2 11 1435 1090 1779 689 1554 0.27 1249 1859 611 IV Aulacorhynchus coeruleicinctis 0 15 2085 1454 2716 1262 2464 0.33 2175 2754 579 IV Andigena hypoglauca 0 17 2844 2528 3159 631 2874 0.34 2582 3164 582 IV Selenidera reinwardtii 1 8 1264 826 1702 876 Pteroglossus azara 1 0 950 950 950 0 Picumnus aurifrons 0 1 837 837 837 0 Melanerpes cruentatus 0 3 953 892 1013 121 Veniliornis affinis 0 1 951 951 951 0 Colaptes rubiginosus 1 32 1567 1124 2010 886 1781 0.91 1431 1958 527 V Colaptes rivolii 1 22 2525 2067 2982 915 2788 0.36 2415 3158 744 IV Dryocopus linea tus 0 4 1064 934 1194 260 Campephilus rubricollis 0 1 878 878 878 0 Campephilus melanoleucos 0 3 1044 893 1194 301 Sclerurus mexicanus 1 0 930 930 930 0 Schizoeaca helleri 1 6 3193 2972 3414 442

PAGE 127

127 Table A -1. Cont inued No. Sites Detected Observed Elevational Range Modeled Elevational Range Species Netted Counts Midpoint Min. Max. Range Optimum Prob. of Occurrence Min. Max. Range HOF Type Synallaxis azarae 18 78 2225 1401 3049 1648 2081 0.94 1682 2809 1127 V Sy nallaxis cabanisi 1 3 1101 805 1397 592 Cranioleuca marcapatae 1 1 2705 2636 2774 138 Cranioleuca curtata 3 23 1391 969 1813 844 1355 0.45 998 1713 715 IV Premnornis guttuligera 2 0 2080 1702 2458 756 Premnoplex brunnescens 22 6 1480 1100 1859 759 1487 0.82 1153 1817 664 IV Margarornis squamiger 4 20 2670 1998 3342 1344 3414 0.34 2754 3414 660 III Pseudocolaptes boissonneautii 2 35 2438 1716 3159 1443 2063 0.34 1877 3255 1378 V Anabacerthia striaticollis 20 24 1476 1100 1852 7 52 1525 0.60 1243 1807 564 IV Syndactyla rufosuperciliata 7 17 1645 1279 2010 731 1783 0.72 1567 2003 436 IV Simoxenops ucayalae 4 3 1118 923 1313 390 Ancistrops strigilatus 0 4 903 826 980 154 Hyloctistes subulatus 3 5 991 853 1129 276 Philydor ruficaudatum 1 4 1070 938 1202 264 Philydor erythrocercum 1 1 865 853 878 26 Philydor erythropterum 0 3 922 892 951 59 Philydor rufum 2 1 1129 1007 1250 243 Anabazenops dorsalis 1 14 1021 805 1236 431 8 05 0.69 805 1055 250 II Thripadectes melanorhynchus 15 13 1245 930 1561 631 1204 0.77 974 1431 457 IV Thripadectes holostictus 5 6 1682 1354 2010 656 Automolus ochrolaemus 7 24 1028 805 1250 445 805 0.87 805 1115 310 II Automolus rubiginosus 2 2 1105 930 1279 349 Automolus rufipileatus 0 7 899 829 969 140 Lochmias nematura 5 1 1093 818 1367 549 Xenops minutus 4 2 1049 818 1279 461 Xenops rutilans 0 2 958 826 1090 264 Dendrocincla tyrannina 3 8 2688 252 7 2849 322 Dendrocincla fuliginosa 4 12 1142 829 1454 625 1082 0.48 894 1269 376 IV Deconychura longicauda 0 1 975 975 975 0 Glyphorynchus spirurus 9 7 1169 837 1501 664 Dendrexetastes rufigula 0 10 1167 938 1395 457 1186 0.60 10 35 1337 303 IV

PAGE 128

128 Table A -1. Continued No. Sites Detected Observed Elevational Range Modeled Elevational Range Species Netted Counts Midpoint Min. Max. Range Optimum Prob. of Occurrence Min. Max. Range HOF Type Xiphocolaptes promeropirhynchus 3 26 1963 876 3049 2173 Dendrocolaptes certhia 0 3 1084 951 1217 266 Dendrocolaptes picumnus 2 9 1153 805 1501 696 Xiphorhynchus obsoletus 1 0 853 853 853 0 Xiphorhynchus ocellatus 3 20 1159 818 1501 683 Xiphorhynchus gut tatus 3 12 997 866 1129 263 959 0.66 868 1053 185 IV Xiphorhynchus triangularis 14 20 1549 1100 1998 898 1786 0.71 1429 1914 485 V Lepidocolaptes lacrymiger 2 10 2052 1309 2795 1486 2634 0.25 2477 2790 313 IV Campylorhamphus pucherani 0 1 2542 2542 2542 0 Campylorhamphus trochilirostris 5 16 1117 805 1429 624 1076 0.39 818 1337 519 IV Cymbilaimus sanctaemariae 0 12 1051 823 1279 456 805 0.59 805 1074 269 II Thamnophilus palliatus 1 4 1352 1173 1531 358 Thamnophilus schistaceus 2 23 102 1 805 1236 431 805 0.87 805 1121 316 II Thamnophilus caerulescens 6 18 1828 1507 2148 641 1739 0.85 1601 2000 399 V Thamnophilus aethiops 1 2 851 826 877 51 Dysithamnus mentalis 13 27 1260 818 1702 884 1069 0.57 805 1371 566 IV Thamnomanes schi stogynus 2 20 1011 805 1217 412 805 0.93 805 1016 211 II Epinecrophylla spodionota 5 9 1162 823 1501 678 Epinecrophylla ornata 8 11 1117 805 1429 624 805 0.38 805 1282 477 II Epinecrophylla erythrura 1 0 877 877 877 0 Myrmotherula brach yura 0 3 1031 826 1236 410 Myrmotherula longicauda 2 12 1133 837 1429 592 1301 0.47 1016 1405 389 V Myrmotherula axillaris 1 0 1501 1501 1501 0 Myrmotherula schisticolor 14 6 1545 1309 1781 472 1522 0.71 1282 1762 480 IV Myrmotherula me netriesii 0 2 907 904 909 5 Herpsilochmus axillaris 0 19 1256 980 1531 551 1209 0.77 1022 1400 378 IV Microrhopias quixensis 2 16 1042 805 1279 474 805 0.66 805 1134 329 II Drymophila caudata 4 20 2247 1541 2952 1411 1820 0.63 1522 2115 592 IV Hypocnemis subflava 4 27 1117 805 1429 624 1262 0.71 909 1366 457 V Terenura sharpei 1 3 1242 934 1550 616

PAGE 129

129 Table A -1. Continued No. Sites Detected Observed Elevational Range Modeled Elevational Range Species Netted Counts Midpoint Min. Max. R ange Optimum Prob. of Occurrence Min. Max. Range HOF Type Cercomacra cinerascens 0 1 823 823 823 0 Cercomacra serva 3 11 1110 805 1415 610 805 0.55 805 1076 271 II Cercomacra manu 1 7 870 805 934 129 Pyriglena leuconota 12 33 1478 946 2 010 1064 1846 0.84 1335 1963 629 V Myrmoborus leucophrys 4 24 933 805 1061 256 805 0.99 805 1006 201 II Myrmoborus myotherinus 2 9 964 818 1110 292 Percnostola lophotes 1 15 1070 805 1335 530 805 0.54 805 1217 412 II Schistocicla leucostigma 2 2 1080 818 1341 523 Myrmeciza hemimelaena 8 32 1100 805 1395 590 805 0.97 805 1173 368 II Myrmeciza atrothorax 0 4 817 805 829 24 Myrmeciza goeldii 1 9 948 805 1090 285 Myrmeciza fortis 0 2 997 980 1013 33 Rhegmatorhina melanosticta 4 2 913 876 950 74 Hylophylax naevius 7 18 1030 805 1255 450 805 0.93 805 982 177 II Phlegopsis nigromaculata 2 2 937 923 950 27 Formicarius analis 3 10 933 805 1061 256 805 0.81 805 925 120 II Formicarius rufipectus 1 30 1459 1139 1779 640 1316 0.88 1183 1632 449 V Chamaeza campanisona 2 13 1338 1124 1551 427 1280 0.52 1095 1468 373 IV Chamaeza mollissima 0 14 2392 1972 2811 839 2462 0.22 2003 2926 924 V Grallaria squamigera 0 3 2990 2566 3414 848 Grallaria gua timalensis 0 11 1340 1173 1506 333 1267 0.63 1131 1405 274 IV Grallaria albigula 0 26 1845 1506 2184 678 1859 0.89 1596 2123 527 IV Grallaria erythroleuca 1 50 2489 1965 3013 1048 2637 0.81 2198 2864 665 V Grallaria rufula 0 23 3077 2739 3414 675 3414 1 .00 2981 3414 433 II Myrmothera campanisona 0 3 899 893 904 11 Grallaricula flavirostris 3 1 1819 1640 1998 358 Grallaricula ferrugineipectus 4 12 2727 2458 2995 537 2921 0.35 2704 3137 433 IV Conopophaga ardesiaca 15 20 1320 960 1679 7 19 1310 0.59 1022 1598 575 IV Liosceles thoracicus 1 16 964 818 1110 292 805 0.73 805 1074 269 II Scytalopus parvirostris 5 79 2713 2012 3414 1402 3414 0.86 2282 3414 1132 III Scytalopus atratus 4 41 1668 1324 2012 688 1744 0.97 1444 2044 600 IV

PAGE 130

13 0 Table A-1. Continued No. Sites Detected Observed Elevational Range Modeled Elevational Range Species Netted Counts Midpoint Min. Max. Range Optimum Prob. of Occurrence Min. Max. Range HOF Type Scytalopus schulenbergi 0 1 3414 3414 3414 0 Phyllomyia s cinereiceps 0 2 1445 1373 1517 144 Phyllomyias plumbeiceps 0 1 1578 1578 1578 0 Elaenia albiceps 1 6 1971 950 2991 2041 Elaenia pallatangae 1 19 2788 2527 3049 522 2965 0.65 2767 3164 397 IV Mecocerculus stictopterus 1 53 2924 2465 3383 918 2837 0.78 2558 3310 751 V Mecocerculus leucophrys 1 14 2797 2553 3040 487 2801 0.39 2621 2981 360 IV Anairetes parulus 0 1 2972 2972 2972 0 Serpophaga cinerea 1 0 1357 1357 1357 0 Pseudotriccus ruficeps 1 12 2920 2463 3377 914 3059 0.21 2634 3414 780 IV Corythopis torquatus 2 4 962 948 975 27 Zimmerius bolivianus 4 54 2118 1335 2901 1566 2000 0.66 1361 2639 1278 IV Phylloscartes poecilotis 1 0 1392 1392 1392 0 Phylloscartes ophthalmicus 6 21 1329 1007 16 51 644 1301 0.59 1045 1554 509 IV Phylloscartes orbitalis 1 0 946 946 946 0 Phylloscartes ventralis 6 3 1770 1541 1998 457 Phylloscartes parkeri 0 4 1260 1124 1395 271 Mionectes striaticollis 47 13 2115 853 3377 2525 2115 0.92 1 105 3125 2019 I Mionectes olivaceus 17 0 1177 853 1501 649 853 1.00 853 1254 401 II Mionectes oleagineus 6 0 914 853 975 123 Leptopogon amaurocephalus 2 2 922 893 950 57 Leptopogon superciliaris 16 27 1291 866 1716 850 1353 0.64 1063 16 42 579 IV Myiotriccus ornatus 5 9 1010 818 1202 384 Myiornis ecaudatus 0 3 918 876 960 84 Lophotriccus pileatus 15 40 1372 930 1813 883 1395 0.88 1100 1689 590 IV Hemitriccus flammulatus 0 5 909 866 951 85 Hemitriccus granadens is 5 55 2737 2258 3215 957 2783 0.74 2355 3210 856 IV Hemitriccus rufigularis 0 1 1090 1090 1090 0 Poecilotriccus albifacies 0 2 887 823 951 128 Poecilotriccus plumbeiceps 3 7 1816 1610 2023 413 Poecilotriccus latirostris 0 1 89 5 895 895 0

PAGE 131

131 Table A -1. Continued No. Sites Detected Observed Elevational Range Modeled Elevational Range Species Netted Counts Midpoint Min. Max. Range Optimum Prob. of Occurrence Min. Max. Range HOF Type Rhynchocyclus fulvipectus 4 3 1360 11 24 1597 473 Tolmomyias assimilis 1 8 1097 826 1367 541 Tolmomyias flaviventris 1 0 1324 1324 1324 0 Platyrinchus mystaceus 10 2 1518 1255 1781 526 1537 0.48 1249 1822 573 IV Myiophobus inornatus 7 0 1474 950 1998 1048 M yiophobus ochraceiventris 0 5 2969 2723 3215 492 Myiobius villosus 6 0 1258 853 1663 811 Terenotriccus erythrurus 3 1 906 837 975 138 Pyrrhomyias cinnamomeus 5 54 2233 1365 3101 1736 3414 0.43 1807 3414 1607 III Lathrotriccus eu leri 7 16 1267 818 1716 898 805 0.55 805 1243 438 II Contopus fumigatus 0 28 2016 930 3101 2171 2770 0.45 2436 3104 668 IV Knipolegus aterrimus 0 1 2685 2685 2685 0 Myiotheretes striaticollis 0 21 2182 1332 3032 1700 2110 0.11 1066 3153 2087 I Ochthoeca frontalis 2 1 3306 3235 3377 142 Ochthoeca pulchella 8 45 2562 2023 3101 1078 2790 0.76 2490 3090 600 IV Ochthoeca cinnamomeiventris 3 8 2725 2458 2992 534 Ochthoeca rufipectoralis 2 15 3007 2637 3377 740 2986 0.60 2817 3156 339 IV Legatus leucophaius 0 7 964 818 1110 292 Conopias cinchoneti 1 4 1380 1331 1429 98 Myiodynastes chrysocephalus 1 6 2034 1334 2734 1400 Tyrannus melancholicus 0 8 2015 1307 2723 1416 Rhytipterna simplex 0 2 872 866 8 78 12 Myiarchus tuberculifer 2 31 2111 1007 3215 2208 2420 0.28 1658 3182 1524 IV Ramphotrigon megacephalum 0 6 878 805 951 146 Ramphotrigon fuscicauda 1 3 937 923 951 28 Attila spadiceus 0 2 1149 1124 1173 49 Pipreola intermedia 4 38 2405 1709 3101 1392 2858 0.57 2279 3077 798 V Pipreola arcuata 1 43 2581 1947 3215 1268 2918 0.68 2530 3310 780 IV Pipreola frontalis 1 0 1781 1781 1781 0 Ampelion rubrocristatus 0 3 2786 2685 2887 202 Rupicola peruvianus 7 8 1265 866 1663 797

PAGE 132

132 Table A -1. Continued No. Sites Detected Observed Elevational Range Modeled Elevational Range Species Netted Counts Midpoint Min. Max. Range Optimum Prob. of Occurrence Min. Max. Range HOF Type Cephalopterus ornatus 0 1 1307 1307 1307 0 Tyranneutes stolzmanni 0 1 866 866 866 0 Machaeropterus pyrocephalus 9 16 1124 818 1429 611 805 0.61 805 1186 381 II Lepidothrix coronata 0 2 1018 975 1061 86 Lepidothrix coeruleocapilla 22 12 1281 853 1709 857 853 0.98 853 1406 553 II Manacus manacus 2 0 963 950 975 25 Chiroxiphia boliviana 31 45 1401 930 1872 942 1428 0.96 1032 1825 793 IV Pipra fasciicauda 3 4 902 853 951 99 Pipra chloromeros 4 2 921 829 1013 184 Schiffornis turdin a 3 10 954 818 1090 272 805 0.51 805 1071 266 II Pachyramphus versicolor 0 2 2965 2828 3101 273 Piprites chloris 1 0 930 930 930 0 Vireolanius leucotis 0 5 1059 1007 1110 103 Vireo leucophrys 1 7 1541 1373 1709 336 Vire o olivaceus 0 2 828 818 837 19 Hylophilus hypoxanthus 0 5 852 826 878 52 Hylophilus ochraceiceps 3 15 981 823 1139 316 995 0.83 891 1100 209 IV Cyanolyca viridicyanus 1 37 2738 2463 3013 550 2710 0.62 2420 2997 577 IV Cyanocorax violace us 0 9 878 805 951 146 Cyanocorax yncas 0 19 1716 1398 2033 635 1823 0.86 1619 2026 407 IV Pygochelidon cyanoleuca 2 0 1395 1392 1397 5 Microcerculus marginatus 7 33 1069 805 1332 527 805 0.97 805 1183 378 II Odontorchilus branickii 1 6 1456 1334 1578 244 Troglodytes aedon 0 2 2987 2982 2991 9 Troglodytes solstitialis 8 45 2559 1883 3235 1352 2754 0.65 2365 3140 775 IV Campylorhynchus turdinus 0 8 1051 823 1279 456 Pheugopedius genibarbis 4 15 1151 805 1496 691 1136 0.33 836 1434 597 IV Cinnycerthia fulva 8 34 2642 1998 3286 1288 2780 0.46 2321 3203 882 V Henicorhina leucophrys 19 52 2044 1324 2764 1440 1742 0.97 1447 2261 814 V Cyphorhinus thoracicus 10 23 1394 1124 1663 539 1251 0.81 1121 1525 404 V Myades tes ralloides 25 23 1902 975 2828 1853 1416 0.80 1017 1815 798 IV

PAGE 133

133 Table A -1. Continued No. Sites Detected Observed Elevational Range Modeled Elevational Range Species Netted Counts Midpoint Min. Max. Range Optimum Prob. of Occurrence Min. Max. Range H OF Type Catharus fuscater 1 0 1350 1350 1350 0 Catharus dryas 7 7 1261 1124 1398 274 Entomodestes leucotis 10 43 2161 1541 2781 1240 1956 0.98 1721 2438 717 V Turdus leucops 2 15 1536 1124 1947 823 1684 0.36 1337 2029 691 IV Turdus hau xwelli 1 0 1392 1392 1392 0 Turdus ignobilis 7 8 1235 829 1640 811 Turdus fuscater 1 7 3025 2673 3377 704 Turdus chiguanco 1 1 2983 2974 2992 18 Turdus serranus 6 36 2355 1550 3159 1609 2780 0.54 2250 3044 793 V Creurgops dentatus 2 3 1712 1541 1883 342 Hemispingus atropileus 3 10 2465 1947 2982 1035 2715 0.14 2292 3135 843 IV Hemispingus superciliaris 1 5 2681 2542 2820 278 Hemispingus frontalis 3 3 2363 1872 2854 983 Hemispingus melanotis 13 16 2300 1365 3235 1870 1739 0.45 1423 2052 629 IV Hemispingus xanthophthalmus 0 1 2781 2781 2781 0 Hemispingus trifasciatus 0 3 3378 3342 3414 72 Cnemoscopus rubrirostris 1 7 2635 2458 2811 353 Thlypopsis ruficeps 0 3 2825 2668 29 81 313 Trichothraupis melanops 18 1 1396 1129 1663 534 1497 0.83 1259 1736 477 IV Lanio versicolor 2 17 964 818 1110 292 805 0.89 805 1027 222 II Ramphocelus carbo 7 3 1369 1309 1429 120 Thraupis episcopus 1 2 1362 1331 1392 61 Thraupis cyanocephala 4 27 2184 1335 3032 1697 2616 0.34 2104 3127 1023 IV Buthraupis montana 1 63 2821 2426 3215 789 2905 0.97 2563 3247 684 IV Anisognathus igniventris 0 22 2963 2542 3383 841 3119 0.75 2866 3372 506 IV Anisognathus somptuosus 2 24 19 66 1621 2310 689 1943 0.91 1684 2198 514 IV Chlorornis riefferrii 4 54 2837 2458 3215 757 2809 0.80 2443 3174 731 IV Delothraupis castaneoventris 0 8 2804 2507 3101 594 Iridosornis analis 16 23 1636 1124 2148 1024 1763 0.63 1447 2081 634 IV Iri dosornis jelskii 1 3 2880 2383 3377 994 Pipraeidea melanonota 3 0 1405 1324 1486 162

PAGE 134

134 Table A -1. Continued No. Sites Detected Observed Elevational Range Modeled Elevational Range Species Netted Counts Midpoint Min. Max. Range Optimum P rob. of Occurrence Min. Max. Range HOF Type Chlorochrysa calliparaea 12 10 1440 1100 1780 680 1406 0.41 1030 1782 752 IV Tangara cyanicollis 6 1 1298 1194 1401 207 Tangara xanthogastra 1 0 877 877 877 0 Tangara punctata 3 2 1487 1194 17 81 587 Tangara vassorii 2 3 2745 2458 3032 574 Tangara nigroviridis 6 1 1547 1392 1702 310 Tangara chilensis 3 33 1365 818 1911 1093 1142 0.63 818 1465 647 IV Tangara gyrola 1 1 1163 1124 1202 78 Tangara chrysotis 1 1 1 492 1392 1591 199 Tangara xanthocephala 1 2 1682 1392 1972 580 Tangara parzudakii 1 0 1998 1998 1998 0 Tangara schrankii 2 5 903 826 980 154 Tangara arthus 12 1 1376 1100 1651 551 1269 0.51 1002 1540 538 IV Cyanerpes ca eruleus 2 1 1239 1124 1354 230 Chlorophanes spiza 1 1 963 826 1100 274 Iridophanes pulcherrimus 1 0 1663 1663 1663 0 Hemithraupis flavicollis 0 1 826 826 826 0 Conirostrum albifrons 0 1 2542 2542 2542 0 Diglossa mystacalis 1 4 3203 2991 3414 423 Diglossa brunneiventris 2 9 2788 2527 3049 522 Diglossa glauca 15 32 1926 1061 2791 1730 1885 0.99 1622 2149 527 IV Diglossa caerulescens 4 7 1788 1392 2184 792 Diglossa cyanea 10 84 2403 1392 341 4 2022 3414 1.00 2464 3414 950 II Catamblyrhynchus diadema 2 2 2808 2636 2981 345 Chlorospingus ophthalmicus 4 35 1977 1392 2561 1169 1919 0.87 1554 2284 731 IV Chlorospingus parvirostris 7 0 1686 1350 2023 673 Chlorospingus flavigulari s 17 18 1410 1007 1813 806 1272 0.60 1053 1491 438 IV Chlorospingus canigularis 1 0 1100 1100 1100 0 Coereba flaveola 6 7 1134 866 1401 535 Saltator grossus 0 10 870 805 934 129 805 0.93 805 894 89 II Saltator maximus 5 7 1124 818 1429 611

PAGE 135

135 Table A -1. Continued No. Sites Detected Observed Elevational Range Modeled Elevational Range Species Netted Counts Midpoint Min. Max. Range Optimum Prob. of Occurrence Min. Max. Range HOF Type Ammodramus aurifrons 2 0 1399 1397 1401 4 Haplospiza rustica 3 4 2208 1597 2820 1223 Sporophila schistacea 0 7 842 805 878 73 Sporophila luctuosa 1 0 1392 1392 1392 0 Oryzoborus angolensis 3 0 1135 946 1324 378 Arremon taciturnus 6 12 909 837 980 143 922 0. 70 849 993 143 IV Arremon brunneinucha 5 14 1727 1367 2087 720 1765 0.58 1538 1989 451 IV Arremon torquatus 0 12 2786 2539 3032 493 2725 0.35 2577 2877 300 IV Arremon castaneiceps 1 0 1354 1354 1354 0 Atlapetes melanolaemus 15 56 2318 1401 3235 1834 2535 0.67 1987 3083 1096 IV Piranga flava 0 3 1093 1013 1173 160 Piranga leucoptera 1 2 1270 1124 1415 291 Chlorothraupis carmioli 5 21 958 826 1090 264 933 0.93 834 1032 198 IV Cyanocompsa cyanoides 4 6 1016 818 1213 395 Parula pitiayumi 1 32 1329 1007 1651 644 1361 0.82 1105 1616 511 IV Myioborus miniatus 17 68 1408 818 1998 1180 1802 0.97 1074 1937 864 V Myioborus melanocephalus 7 83 2392 1401 3383 1982 2316 0.97 2073 3111 1038 V Basileuterus bivittatus 15 33 1242 89 3 1591 698 1311 0.89 993 1462 470 V Basileuterus chrysogaster 1 21 965 805 1124 319 805 0.93 805 1029 224 II Basileuterus luteoviridis 7 28 2837 2458 3215 757 2770 0.47 2456 3080 624 IV Basileuterus signatus 9 49 2422 1852 2992 1140 2292 0.81 1856 2728 871 IV Basileuterus coronatus 20 34 1896 1334 2458 1124 1718 0.95 1457 1979 522 IV Basileuterus tristriatus 15 10 1681 1340 2023 683 1605 0.84 1348 1860 512 IV Phaeothlypis fulvicauda 0 1 818 818 818 0 Psarocolius angustifrons 4 20 1419 805 203 3 1228 805 0.54 805 1397 592 II Psarocolius atrovirens 3 10 1696 1358 2033 675 1637 0.25 1314 1961 647 IV Psarocolius decumanus 0 9 1020 823 1217 394 Psarocolius bifasciatus 0 10 934 892 975 83 943 0.78 894 995 102 IV Cacicus chrysonotus 1 36 25 56 2010 3101 1091 2736 0.53 2352 3117 764 IV Cacicus cela 0 1 934 934 934 0 Amblycercus holosericeus 2 1 2599 2458 2739 281

PAGE 136

136 Table A -1. Continued No. Sites Detected Observed Elevational Range Modeled Elevational Range Species Netted Co unts Midpoint Min. Max. Range Optimum Prob. of Occurrence Min. Max. Range HOF Type Euphonia mesochrysa 7 3 1401 1139 1663 524 Euphonia xanthogaster 32 70 1402 805 1998 1193 805 0.82 805 1760 955 III Chlorophonia cyanea 2 50 1443 876 2010 1134 16 56 0.89 1149 1924 775 V

PAGE 137

137 APPENDIX B SUMMARY OF RARE SPEC IES Table B -1 Species detected at ten sites or fewer (with netting and counts) in the Manu study area. The total number of detections is given for each species with the proposed cause for its rar ity across surveys. A species i s assigned to only on e of the categories described. Lowland species are those no t detected above 1200 m and whose distribution extends below 500 m; Highland species are those not found below 3200 m and whose distribution extends above 3800 m. Detection indicates that species rarely sings, has inconspicuous behavior, or resides in areas that are difficult to survey Habitat specific denotes species that a re found within easily identifiable habitat types within fores t (see Habitat Notes for details). Non -forest species are those common along forest edges, second growth, and clearings within the study area. Low density species is a catch all category for easily detected forest species with no known habitat specif icity. Species No. Sites Detected Lowland Highland Detection Habitat Specific Non forest Low Density Habitat notes Nothocercus nigrocapillus 8 x Tinamus guttatus 8 x Crypturellus soui 8 x Penelope jacquacu 4 x Aburria aburri 9 x Ortalis guttata 1 x Odontophorus balliviani 4 x Odontophorus stellatus 2 x Elanoides forficatus 1 x Accipiter striatus 1 x Buteo magnirostris 2 x Ibycter americanus 1 x Claravis mondetoura 1 x Nomadic, Chusquea Patagioenas subvinacea 4 x Leptotila rufaxilla 1 x Ara ararauna 1 x Ara militaris 4 x Primolius couloni 5 x Aratinga leucophthalma 6 x Pionus tumultuosus 5 x Amazona farinosa 2 x P icumnus aurifrons 1 x

PAGE 138

138 Table B -1 Continued Species No. Sites Detected Lowland Highland Detection Habitat Specific Non forest Areas Low Density Habitat notes Piaya cayana 9 x Dromococcyx pavoninus 4 x Ciccaba albitarsis 1 x Gla ucidium bolivianum 3 x Lurocalis rufiventris 1 x Threnetes leucurus 5 x Gaps, River edges Phaethornis hispidus 2 x Phaethornis koepckeae 3 x Phaethornis superciliosus 6 x Doryfera johannae 3 x Colibri corusca ns 3 x Heliothryx auritus 1 x Aglaiocercus kingi 6 x Lesbia nuna 1 x Haplophaedia assimilis 2 x Aglaeactis cupripennis 1 x Montane scrub Coeligena torquata 10 x Boissonneaua matthewsii 5 x Heliodoxa sc hreibersii 5 x Heliodoxa aurescens 1 x Chaetocercus mulsant 2 x Chlorostilbon mellisugus 2 x Klais guimeti 1 x Campylopterus largipennis 3 x Thalurania furcata 10 x Taphrospilus hypostictus 4 x Amazil ia viridicauda 1 x Chrysuronia oenone 3 x Trogon melanurus 6 x Trogon viridis 2 x Trogon violaceus 6 x Trogon curucui 3 x

PAGE 139

139 Table B -1 Continued Species No. Sites Detected Lowland Highland Detection Habitat Specific Non forest Areas Low Density Habitat notes Chloroceryle aenea 1 x Momotus aequatorialis 2 x Streamsides Galbula cyanescens 4 x Nystalus striolatus 2 x Micromonacha lanceolata 2 x Nonnula ruficapilla 1 x Capito aur atus 6 x Ramphastos tucanus 1 x Aulacorhynchus prasinus 8 x Selenidera reinwardtii 9 x Pteroglossus azara 1 x Melanerpes cruentatus 3 x Veniliornis affinis 1 x Dryocopus lineatus 4 x River edge forest Campephilus rubricollis 1 x Campephilus melanoleucos 3 x Sclerurus mexicanus 1 x Schizoeaca helleri 7 x Montane scrub Treeline Synallaxis cabanisi 4 x Cranioleuca marcapatae 2 x Premnornis guttuligera 2 x Si moxenops ucayalae 7 x Guadua Ancistrops strigilatus 4 x Hyloctistes subulatus 8 x Philydor ruficaudatum 5 x Philydor erythrocercum 2 x Philydor erythropterum 3 x Philydor rufum 3 x Automolus rubiginosus 4 x Automolus rufipileatus 7 x Lochmias nematura 6 x Streamsides Xenops minutus 6 x Xenops rutilans 2 x

PAGE 140

140 Table B-1 Continued Species No. Sites Detected Lowland Highland Detection Habitat Specific Non forest Areas Low Density H abitat notes Deconychura longicauda 1 x Dendrexetastes rufigula 10 x Dendrocolaptes certhia 3 x Xiphorhynchus obsoletus 1 x Campylorhamphus pucherani 1 x Thamnophilus palliatus 5 x Thamnophilus aethiops 3 x Epinecrophylla erythrura 1 x Myrmotherula brachyura 3 x Myrmotherula axillaris 1 x Myrmotherula menetriesii 2 x Terenura sharpei 4 x Cercomacra cinerascens 1 x Cercomacra manu 8 x Schistocicla leucostigma 4 x Streamsides Myrmeciza atrothorax 4 x Myrmeciza goeldii 10 x Myrmeciza fortis 2 x Rhegmatorhina melanosticta 6 x Phlegopsis nigromaculata 4 x Grallaria squamigera 3 x Myrmothera campanisona 3 x Grall aricula flavirostris 4 x Scytalopus schulenbergi 1 x Treeline Phyllomyias cinereiceps 2 x Phyllomyias plumbeiceps 1 x Elaenia albiceps 7 x Anairetes parulus 1 x Montane scrub Serpophaga cinerea 1 x River edge forest, Islands Corythopis torquatus 6 x Phylloscartes ventralis 9 x Phylloscartes orbitalis 1 x Phylloscartes parkeri 4 x

PAGE 141

141 T able B -1 Continued Species No. Sites Detected Lowland Highland Detection Habitat Specific Non forest Ar eas Low Density Habitat notes Phylloscartes poecilotis 1 x Mionectes oleagineus 6 x Leptopogon amaurocephalus 4 x Myiornis ecaudatus 3 x Hemitriccus flammulatus 5 x Hemitriccus rufigularis 1 x Outlying ridges Poeci lotriccus latirostris 1 x Poecilotriccus plumbeiceps 10 x Rhynchocyclus fulvipectus 7 x Tolmomyias assimilis 9 x Tolmomyias flaviventris 1 x Poecilotriccus albifacies 2 x Myiophobus inornatus 7 x Myiophobus ochraceiventris 5 x Myiobius villosus 6 x Terenotriccus erythrurus 4 x Knipolegus aterrimus 1 x Montane scrub Ochthoeca frontalis 3 x Legatus leucophaius 7 x Conopias cinchoneti 5 x Myiodynastes chrysocephalu s 7 x Tyrannus melancholicus 8 x Rhytipterna simplex 2 x Ramphotrigon megacephalum 6 x Ramphotrigon fuscicauda 4 x Guadua Attila spadiceus 2 x Piprites chloris 1 x Pipreola frontalis 1 x Ampelion rubrocristatus 3 x Cephalopterus ornatus 1 x Tyranneutes stolzmanni 1 x Lepidothrix coronata 2 x Manacus manacus 2 x

PAGE 142

142 Tabl e B -1 Continued Species No. Sites Detected Lowland Highland Detection Habitat Specific Non forest Areas Low Density Habitat notes Pipraeidea melanonota 3 x Pipra fasciicauda 7 x Pipra chloromeros 6 x Pachyramphus versicolor 2 x Piranga flava 3 x Vireolanius leucotis 5 x Vireo leucophrys 8 x Vireo olivaceus 2 x Hylophilus hypoxanthus 5 x Cyanocorax violaceus 9 x Pygochelidon cyanoleuca 2 x Odontorchilus branickii 7 x Troglodytes aedon 2 x Campylorhynchus turdinus 8 x Catharus fuscater 1 x Streamsides Turdu s hauxwelli 1 x Turdus fuscater 8 x Turdus chiguanco 2 x Creurgops dentatus 5 x Hemispingus superciliaris 6 x Hemispingus frontalis 6 x Hemispingus xanthophthalmus 1 x Hemispingus trifasciatus 3 x Treel ine Cnemoscopus rubrirostris 8 x Thlypopsis ruficeps 3 x Ramphocelus carbo 10 x Thraupis episcopus 3 x Delothraupis castaneoventris 8 x Iridosornis jelskii 4 x Tangara cyanicollis 7 x Tangara xanthogastra 1 x Tangara punctata 5 x Tangara vassorii 5 x

PAGE 143

143 Table B -1 Continued Species No. Sites Detected Lowland Highland Detection Habitat Specific Non forest Areas Low Density Habitat notes Tangara nigroviridis 7 x Tangara gyrola 2 x Tangara chrysotis 2 x Tangara xanthocephala 3 x Tangara parzudakii 1 x Tangara schrankii 7 x Cyanerpes caeruleus 3 x Chlorophanes spiza 2 x Iridophanes pulcherrimus 1 x Hemithraupis flavicollis 1 x Conirostrum albifrons 1 x Diglossa mystacalis 5 x Montane scrub Catamblyrhynchus diadema 4 x Chlorospingus parvirostris 7 x Chlorospingus canigularis 1 x Saltator grossus 10 x Ammodramus aurifrons 2 x Hapl ospiza rustica 7 x Landslides, Chusquea Sporophila schistacea 7 x Sporophila luctuosa 1 x Oryzoborus angolensis 3 x Arremon castaneiceps 1 x Streamsides Piranga leucoptera 3 x Cyanocompsa cyanoides 10 x Phaeot hlypis fulvicauda 1 x Streamsides Psarocolius decumanus 9 x Psarocolius bifasciatus 10 x Cacicus cela 1 x Amblycercus holosericeus 3 x Chusquea Euphonia mesochrysa 10 x

PAGE 144


PAGE 145

145 Figure C -3. Figure C -4.

PAGE 146

146 Figure C -5 Figure C -6.

PAGE 147

147 Figure C -7. Figure C -8.

PAGE 148

148 Figure C -9. Figure C -10.

PAGE 149

149 Figure C -11. Figure C -12.

PAGE 150

150 Figure C -13. Figure C -14.

PAGE 151

151 Figure C -15. Figure C -16.

PAGE 152

152 Figure C -17. Figure C -18.

PAGE 153

153 Figure C -19. Figure C -20.

PAGE 154

154 Figure C -21. Figure C -22.

PAGE 155

155 Figure C -23. Figure C -24.

PAGE 156

156 Figure C -25 Figure C -26.

PAGE 157

157 Figure C -27. Figure C -28.

PAGE 158

158 Figure C -29. Figure C -30.

PAGE 159

159 Figure C -31. Figure C -32.

PAGE 160

160 Figure C -33. Figure C -3 4.

PAGE 161

161 Figure C -35. Figure C -36.

PAGE 162

162 Figure C -37. Figure C -38.

PAGE 163

163 Figure C -39. Figure C -40.

PAGE 164

164 Figure C -41. Figure C -42.

PAGE 165

165 Figure C -43. Figure C -44.

PAGE 166

166 Figure C -45. Figure C -46.

PAGE 167

167 Figure C -47. Figure C -48.

PAGE 168

168 Figure C -49. Figure C -50.

PAGE 169

169 Figur e C -51. Figure C -52.

PAGE 170

170 Figure C -53. Figure C -54.

PAGE 171

171 Figure C -55. Figure C -56.

PAGE 172

172 Figure C -57. Figure C -58.

PAGE 173

173 Figure C -59. Figure C -60.

PAGE 174

174 Figure C -61. Figure C -62.

PAGE 175

175 Figure C -63. Figure C -64.

PAGE 176

176 Figure C -65. Figure C -66.

PAGE 177

177 Figure C -67 Figure C -68.

PAGE 178

178 Figure C -69. Figure C -70.

PAGE 179

179 Figure C -71. Figure C -72.

PAGE 180

180 Figure C -73. Figure C -74.

PAGE 181

181 Figure C -75. Figure C -76.

PAGE 182

182 Figure C -77. Figure C -78.

PAGE 183

183 Figure C -79. Figure C -80.

PAGE 184

184 Figure C -81. Figure C -82.

PAGE 185

185 Figure C -83. Figure C -84.

PAGE 186

186 Figu re C -85. Figure C -86.

PAGE 187

187 Figure C -87. Figure C -88.

PAGE 188

188 Figure C -89. Figure C -90.

PAGE 189

189 Figure C -91. Figure C -92.

PAGE 190

190 Figure C -93. Figure C -94.

PAGE 191

191 Figure C -95. Figure C -96.

PAGE 192

192 Figure C -97. Figure C -98.

PAGE 193

193 Figure C -99. Figure C -100.

PAGE 194

194 Figure C -101. Fi gure C -102.

PAGE 195

195 Figure C -103. Figure C -104.

PAGE 196

196 Figure C -105. Figure C -106.

PAGE 197

197 Figure C -107. Figure C -108.

PAGE 198

198 Figure C -109. Figure C -110.

PAGE 199

199 Figure C -111. Figure C -112.

PAGE 200

200 Figure C -113. Figure C -114.

PAGE 201

201 Figure C -115. Figure C -116.

PAGE 202

202 Figure C -117. Fi gure C -118.

PAGE 203

203 Figure C -119. Figure C -120.

PAGE 204

204 Figure C -121. Figure C -122.

PAGE 205

205 Figure C -123. Figure C -124.

PAGE 206

206 Figure C -125. Figure C -126.

PAGE 207

207 Figure C -127. Figure C -128.

PAGE 208

208 Figure C -129. Figuer C -130.

PAGE 209

209 Figure C -131. Figure C -132.

PAGE 210

210 Figure C -133. Fi gure C -134.

PAGE 211

211 Figure C -135. Figure C -136.

PAGE 212

212 Figure C -137. Figure C -138.

PAGE 213

213 Figure C -139. Figure C -140.

PAGE 214

214 Figure C -141. Figure C -142.

PAGE 215

215 Figure C -143. Figure C -144.

PAGE 216

216 Figure C -145. Figure C -146.

PAGE 217

217 Figure C -147. Figure C -148.

PAGE 218

218 Figure C -149. Fi gure C -150.

PAGE 219

219 Figure C -151. Figure C -152.

PAGE 220

220 Figure C -153. Figure C -154.

PAGE 221

221 Figure C -155. Figure C -156.

PAGE 222

222 Figure C -157. Figure C -158.

PAGE 223

223 Figure C -159. Figure C -160.

PAGE 224

224 Figure C -161. Figure C -162.

PAGE 225

225 Figure C -163. Figure C -164.

PAGE 226


PAGE 227

227 Figure D -3. Figure D -4.

PAGE 228

228 Figure D -5. Figure D -6.

PAGE 229

229 Figure D -7. Figure D -8.

PAGE 230

230 Figure D -9. Figure D -10.

PAGE 231

231 Figure D -11. Figure D -12.

PAGE 232

232 Figure D -13. Figure D -14.

PAGE 233

233 Figure D -15. Figure D -16.

PAGE 234

234 Figure D -17. Figure D -18.

PAGE 235

235 Figure D -19. Figure D -20.

PAGE 236

236 Figure D -21. Figure D -22.

PAGE 237

237 Figure D -23. Figure D -24.

PAGE 238

238 Figure D -25. Figure D -26.

PAGE 239

239 APPENDIX E VARIATION PARTITIONI NG METHODS BASED ON MULTI PLE REGRESSION ON DISTANCE MATRICES Variation partitioning of the multiple regression on distance matrices (MRM) analysis was used to determine the percentage of the variation in bird dissimilarity matrices that was explained uniquely by each of the predic tor variables, and the variation that was co explained by multiple predictors as a result of intercorrelation (e.g., Maca et al. 2007) This procedure examines variation explained in a response matrix by a series of models using all possible combinations of predictor variables. For cases with three predictors, the total variation can be partitioned into eight fractions: a ) uniquely explained by vegetation structure b ) uniquely explained by elevation c ) uniquely explained by tree dissimilarity d ) co explained by veg etation structure and elevation e ) co explained by vegetation structure and tree dissimilarity f ) co explained by elevation and tree dissimilarity g ) co explained by vegetation structure, elevation, and tree dissimilarity h ) unexplained variation The following models were performed (shown below), representing all possible combinations of the three predictors The variation explained by each model can be represented by the sum of a subset of lettered categories above. For example, variation explained by a model relat ing bird composition to elevation (Model II below) is the sum of the variation uniquely explained by elevation (b), co explained by elevation

PAGE 240

240 and vegetation structure (d), co explained by elevation and trees (f), and co explained by all three (g). Model I included effects of vegetation structure on bird dissimilarity. This model included all vegetation structure variables initially, after which backwards elimination was used to remove variables that were not significant at the P < 0.05 level. The significant structure variables retained in the model were thereafter used in more complex models with the other predictors of elevation and tree dissimilarity (see below). Variation explained by this model can be represented by R2 = a+d+e+g. Model II included e ffects of elevation on bird dissimilarity. Variation explained by this model can be represented by R2 = b+d+f+g. Model III included effects of tree dissimilarity on bird dissimilarity. Variation explained by this model can be represented by R2 = c+e+f+g. Model IV (vegetation structure and elevation) included variables retained in models I and II, giving R2 = a+b+d+e+f+g. Model V (vegetation structure and trees) included variables retained in models I and III, giving R2 = a+c+d+e+f+g. Model VI (elevation and trees) included variables retained in models II and III, giving R2 = b+c+d+e+f+g. Model VII (vegetation structure, elevation and trees) included all variables retained in models I III, giving R2 = a+b+c+d+e+f+g.

PAGE 241

241 By subtracting the R2 values of one m odel from another, the percentage of variation attributed to each lettered category can be calculated: a ) unique contribution by vegetation structure: Model VII Model VI b ) unique contribution by elevation: Model VII Model V c ) unique contribution by tree diss imilarity: Model VII Model IV d ) co -contribution by vegetation structure and elevation: Model V Model III (a) e ) co -contribution by vegetation structure and tree dissimilarity: Model IV Model II (a) f ) co -contribution by elevation and tree dissimilarity: Model IV Model I (b) g ) co -contribution by vegetation structure, elevation, and tree dissimilarity: Model VII (a + b + c + d + e + f) h ) unexplained variation: 100 (a + b + c + d + e + f + g) Letter categories should sum to 100%. See Maca et al. (2007) for a similarly structured analysis of variation partitioning based on multiple regression on distance matrices

PAGE 242

242 LIST OF REFERENCES Aiba, S. and K. Kitayama. 1999. Structure, composition and species diversity in an altitude -substrate matrix of rain forest tree communities on Mount Kinabalu, Borneo. Plant Ecology 140:139-157. Alexander, R. M. 1999. Energy for animal life. Oxford University, Cambridge University Press, USA. Altshuler, D. L., F. G. Stiles, and R. Dudley. 2004. Of hummingbirds and helicopters: Hovering costs, competitive ability, and foraging strategies American Naturalist 163:1625. Altshuler, D. L., and R. Dudley. 2006. The physiology and biomechanics of avian flight at high altitude. Integrative and Comparative Biology 46:6271. Anderson, R. P., A. T. Peterson, and M. Gmez 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. Angilleta, M. 2009. Thermal Adaptation: A theoretical and empirical synthesis. Univeristy of Chicago Press. Arajo, M. B., and M. Luoto. 2007. The importance of biotic interactions for modelling species distributions under climate change. Global Ecology and Biogeography 16:743753. Austin, M P., and T. M. Smith. 1989. A new model for the continuum concept. Vegetatio 83:34 -47. Austin, M. P., and M. J. Gaywood. 1994. Current problems of environmental gradients and species response curves to continuum theory. Journal of Vegetation Science 5:473 -482. Austin, M. P. 2005. Vegetation and environment: Discontinuities and continuities. Pages 52 84 in E. van der Maarel editor. Vegetation Ecology Blackwell Science Ltd. Oxford, UK Austin. M. 2007. Species distribution models and ecological theory: A critical assessment and some possible new approaches. Ecological Modelling 200:119. Baselga, A. 2010. Partitioning the turnover and nestedness components of beta diversity. Global Ecology and Biogeography 19:134143. Bay, L. K., G. P. Jones, and M. I McCormick 2001. Habitat selection and aggression as determinants of spatial segregation among damselfish on a coral reef. Coral Reefs 20:289298.

PAGE 243

243 Bergen, K. M., A. M. Gilboy, and D. G. Brown. 2007. Multi dimensional vegetation struct ure in modeling avian habitat. Ecological Informatics 2 :9 22. Bersier, L -F. and D. R. Meyer. 1994. Bird assemblages in mosaic forests: the relative importance of vegetation structure and floristic composition along the successional gradient. Acta Oecologica 15:561576. Bir dLife International. 2000. Threatened birds of the world. Lynx Editions and BirdLife International, Barcelona and Cambridge, UK. Block, W. M., and L. A. Brennan. 1993. The habitat concept in ornithology: theory and application. Current Ornithology 11:359 1. Bohlman, S. A., W. F. Laurance, S. G. Laurance, H. E. M. Nacimento, P. M. Fearnside, and A. Andrade. 2008. Importance of soils, topography and geographic distance in structuring central Amazonian tree communities. Journal of Vegetation Science 19:863-8 74. Boyle, A. W. 2008. Can variation in risk of nest predation explain altitudinal migration in tropical birds? Oecologia 155:397 -403. Brehm, G., L. M. Pitkin, N. Hilt, and K. Fiedler. 2005. Montane Andean rain forests aer a global hotspot of geometrid m oths. Journal of Biogeography 32:16211627. Brown, J. H. 1971. Mechanisms of competitive exclusion between two species of chipmunks. Ecology 52:305 -311. Brown, J. H. 1984. On the relationship between abundance and distributions of species. American Natur alist 124:255 -279. Brown, J. H., G. C. Stevens, and D. M. Kaufman. 1996. The geographic range: Size, shape, boundaries and internal structure. Annual Reviews in Ecology and Systematics 27:597-623. Buckley, L. B. and W. Jetz. 2008. Linking global turnover of species and environments. Proceedings of the National Academy of Sciences 105:1783617841. Buermann, W., S. Saatchi, T. B. Smith, B. R. Zutta, J. A. Chaves, B. Mil, and C. H. Graham. 2008. Predicting species distributions across the Amazon and Andean regi ons using remote sensing data. Journal of Biogeography 35:1160-1176. Bull, C. M. 1991. Ecology of parapatric distributions. Annual Review of Ecology and Systematics 22:1936.

PAGE 244

244 Burns, K. C. 2004. Scale and macroecological patterns in seed dispersal mutualisms. Global Ecology and Biogeography 13:289 293. Bush, M. B., M. R. Silman, and D. H. Urrego. 2004. 48,000 Years of Climate and Forest Change in a Biodiversity Hot Spot. Science 303:827829. Buskirk, W. H., G. V. N. Powell, J. F. Wittenberger, R. E. Buskirk, and T. U. Powell. 1972. Interspecific bird flocks in tropical highland Panama. Auk 89:612 -624. Carey, C., S. D. Garber, E. L. Thompson, and F. C. James. 1983. Avian reproduction over an altitudinal gradient. II. Physical characteristics and water loss of eggs. Physiological Zoology 56:340 -352. Carey, C., F. LeonVelarde, O. DuninBorkowskit, and C. Monge-C. 1989. Shell conductance, daily water loss, and water content of Puna Teal eggs. Physiological Zoology 62:83 95. Case, T. J. and M. L. Ta per. 2000. Interspecific competition, environmental gradients, gene flow, and the coevolution of species borders. American Naturalist 155:583605. Case, T. J., R. D. Holt, M. A. McPeek, and T. H. Keitt 2005. The community context of species borders: ec ological and evolutionary perspectives. Oikos 108:28 46. Chambers, J. Q., G. P. Asner, D. C. Morton, L. O. Anderson, S. S. Saatchi, F. D. B. Esprito -Santo, M. Palace, and C. Souza, Jr. 2007. Regional ecosystem structure and function: ecological insights from remote sensing of tropical forests. Trends in Ecology and Evolution 22:414423. Chase, J. M., and M. A. Leibold. 2003. Ecological Niches: Linking Classical and Contemporary Approaches. University of Chicago Press, Chicago, Illinois, USA Clark, J. S 1998. Why trees migrate so fast: Confronting theory with dispersal biology and the paleorecord. American Naturalist 152:204-224. Clark, D. B., M. W. Palmer, and D. A. Clark. 1999. Edaphic factors and the landscape scale distributions of tropical rain fo rest trees. Ecology 80:26622675. Cody, M. L. 1985. Habitat selection in birds. Academic Press, New York New York, USA. Collias, N. E. and E. C. Collias.1984. Nest Building and Bird Behavior Princeton University Press, Princeton, N ew J ersey, USA Col well, R. K., and D. C. Lees. 2000. The mid -domain effect: geometric constraints on the geography of species richness. Trends in Ecology and Evolution 15:70 -76.

PAGE 245

245 Colwell, R. K., C. Rahbek, and N. J. Gotelli. 2004. The middomain effect and species richness patterns: What have we learned so far? American Naturalist 163:E1E23. Colwell, R. K., G. Brehm, C. L. Cardels, A. C. Gilman, and J. T. Longino. 2008. Global warming, elevational range shifts, and lowland biotic attrition in the wet tropics. Science 322: 258 -261. Conway, C. J. and T. E. Martin. 2000. Effects of ambient temperature on avian incubation behavior. Behavioral Ecology 11:178188. Cornell, H. 1974. Parasitism and distributional gaps between allopatric species. American Naturalist 108:880-883. Cornell Lab of Ornithology, Bioacoustics Research Program. 2003 Raven Pro 1.3. Cornell University, Ithaca, New York. Costa, G. C., C. Wolfe, D. B. Shepard, J. P. Caldwell, and L. J. Vitt. 2008. Detecting the influence of climatic variables on species distributions: a test using GIS niche based models along a steep longit udinal environmental gradient. Journal of Biogeography 35:637646. Coudun, C., and J. Ggout. 2006. The derivation of species response curves with Gaussian logistic regression is sensit ive to sampling intensity and curve characteristics. Ecological Modelling 199:164 175. Cramer, W., A. Bondeau, S. Schaphoff, W. Lucht, B. Smith, and S. Sitch. 2004. Tropical forests and the global carbon cycle: Impacts of atmospheric carbon dioxide, clima te change and rate of deforestation. Philosophical Transactions of the Royal Society of London: Series B 359:331343. Cresswell, W., S. Holt, J. M. Reid, D. P. Withfield, R. J. Mellanby, D. Norton, and S. Waldron. 2004. The energetic costs of egg heating constraint incubation attendance but do not determine daily energy expenditure in the pectoral sandpiper. Behavioral Ecology 15:498 -507. Cunningham, H. R., L. J. Rissler, and J. J. Apodaca. 2009. Competition at the range boundary in the slimy salamander: using reciprocal transplants for studies on the role of biotic interactions in spatial distributions. Journal of Animal Ecology 78:5262. Deeming, D. C. 2002. Embryonic development and utilization of egg components. Pages 43 53 in D. C. Deeming, editor. A vian incubation b e haviour, environment, and ev olution. Oxford University Press, New York New York USA.

PAGE 246

246 de Kort, S. R., E. R. B. Eldermire, E. R. A. Cramer, and S. L. Vehrencamp. 2009. The deterrent effect of bird song in territory defense. Behavioral Ec ology 20:200206. Deutsch, R. B., J. J. Tewksbury, R. B. Huey, K. S. Sheldon, C. K. Ghalambor, D. C. Haak, and P. R. Martin. 2008. Impacts of climatic warming on terrestrial ectotherms across latitude. Proceedings of the National Academy of Sciences 105:6 6686672. Diamond, J. 1973. Distributional ecology of New Guinea birds. Science 179:759769. Elith, J., C. H. Graham, R. P. Anderson, M. Dudk, S. Ferrier, A. Guisan, R. J. Hijmans, F. Huettmann, J. R. Leathwick, A. Lehmann, J. Li, L. G. Lohmann, B. A. L oiselle, G. Manion, C. Moritz, M. Nakamura, Y. Nakazawa, J. M. Overton, A. T. Peterson, S. J. Phillips, K. Richardson, R. Scachetti -Pereira, R. E. Schapire, J. Sobern, S. Williams, M. S. Wisz, and N. E. Zimmermann. 2006. Novel methods improve prediction o f species distr ibutions from occurrence data. Ecography 29:129151. Elith, J. and J. R. Leathwick. 2009. Speces distribution models: Ecological explanation and prediction across spa ce and time. Annual Review of Ecology Evolution and Systematics 40:677-6 97. Environmental Systems Research Institute. 2007. ArcView GIS, V. 9.2. Redlands, California. Feeley, K. J. and M. R. Silman. 2010a. Modelling the responses of Andean and Amazonian plant species to climate change: the effects of georeferencing errors a nd the importance of data filtering. Journal of Biogeography 37:733740. F eeley, K. J. and M. R. Silman. 2010b. Biotic attrition from tropical forests correcting for truncated temperature niches. Global Change Biology 16:1830-1836. Feeley, K. J. and M. R. Silman. 2010c. Land use and climate change effects on population size and extinction risk of Andean plants. Global Change Biology doi:10.1111/j.13652486.2010.02197.x Fine, P. V. A., D. C. Daly, G. V. Muoz, I. Mesones, and K. M. Cameron. 2005. The co ntribution of edaphic heterogeneity to the evolution and diversity of Burseraceae trees in the Western Amazon. Evolution 59:14641478. Fleishman, E., N. McDonal, R. Mac Nally, D. D. Murphy, J. Walters, and T. Floyd. 2003. Effects of floristics, physiognomy and non native vegetation on riparian bird communities in a Mojave Desert watershed. Journal of Animal Ecology 72:484490.

PAGE 247

247 Fleishman, E. and R. Mac Nally. 2006. Patterns of spatial autocorrelation of assemblages of birds, floristics, physiognomy, and primary productivity in the central Great Basin, USA. Diversity and Distributions 12:236243. Fletcher, R. J. 2007. Species interactions and population density mediate the use of social cues for habitat selection. Journal of Animal Ecology 76:598606. Fon taine, J. J. and T. E. Martin. 2006. Experimental test of nest predation influences on adult habitat selection in a breeding bird community. American Naturalist 168:811 -818. Frazer, G. W., C. D. Canham, and K. P. Lertzman. 1999. Gap Light Analyzer (GLA): Imaging software to extract canopy structure and gap light transmission indices from true -colour fisheye photographs. Simon Fraser University, Burnaby, British Columbia, and the Institute of Ecosystem Studies, Millbrook, New York USA Gabbe, A. P ., S. K Robinson, and J. D. Brawn. 2002. Tree-Species Preferences of Foraging Insectivorous Birds: Implications for Floodplain Forest Restoration. Conservation Biology 16:462470. Gasner, M. R., J. E. Jankowski, A. L. Ciecka, K. O. Kyle, and K. N. Rabenold. 2010 Projecting impacts of climate change on Neotropical montane forests. Biological Conservation 143:1250-1258 Ghalambor, C. K. and T. E. Martin. 2002. Comparative manipulation of predation in incubating birds reveals variability in the plasticity of res ponses. Behavioral Ecology 13:101-108. Ghalambor, C. K., R. B. Huey, P. R. Martin, J. J. Tewksbury, and G. Wang. 2006. Are mountain passes higher in the tropics? Janzens hypothesis revisited. Integrative and Comparative Biology 46:5 -17. Goetz, S., D. St einberg, R. Dubayah, and B. Blair 2007. Laser remote sensing of canopy habitat heterogeneity as a predictor of bird species richness in an eastern temperate forest, USA. Remote Sensing of Environment 108:254-263. Graves, G. R. 1988. Linearity of geographi c range and its possible effect on the population structure of Andean birds. Auk 105:47-52. Gross, S. J., and T. D. Price. 2000. Determinants of the northern and southern range limits of a warbler. Journal of Biogeography 27:869878. Grubb, P. J. 1977. C ontrol of forest growth and distribution on wet tropical mountains with special references to mineral nutrition. Annual Review of Ecology and Systematics 8:83 -107.

PAGE 248

248 Grubb, P. J., J. R. Lloyd, T. D. Pennington, and T. C. Whitmore. 1963. A comparison of mont ane and lowland rain forest in Ecuador. 1. The forest structure, physiognomy and floristics. Journal of Ecology 51:567-601. Guisan, A. and W. Thuiller. 2005. Predicting species distribution: offering more than simple habitat models. Ecology Letters 8:9931009. Guisan, A., C. H. Graham, J. Elith, F. Huettmann, and NCEAS Species Distribution Modelling Group. 2007. Sensitivity of predictive species distribution models to change in grain size. Diversity and Distributions 13:332 -340. Haber, W. A. 2000. Plant s and vegetation. Pages 39 -70 in N. M. Nadkarni and N. T. Wheelwright, editors. Monteverde: E cology and conservation of a tropical cloud forest Oxford University Press, New York New York, USA Hasui, ., V. Sousa da Mota Gomes, and W. R. Silva. 2007. Ef fects of vegetation traits on habitat preferences of frugivorous birds in Atlantic rain forest. Biotropica 39:502509. Heikkinen, J., and R. Mkip. 2010. Testing hypotheses on shape and distribution of ecological response curves. Ecological Modelling 221:388399. Herzog, S. K., A. R. Soria, and E. Matthysen. 2003. Seasonal variation in avian community composition in a high -Andean Polylepis (Rosaceae) forest fragment. Wilson Bulletin 115:438447. Herzog, S. K., M. Kessler, and K. Bach. 2005. The elevati onal gradient in bird species richness at the local scale: a foothill peak and a highelevation plateau. Ecography 28:209 -222. Hickling, R., D. B. Roy, J. K. Hill, R. Fox, and C. D. Thomas. 2006. The distributions of a wide range of taxonomic groups are expanding polewards. Global Change Biology 12:450 -455. Hijmans, R. J. and C. H. Graham. 2006. The ability of climate envelope models to predict the effect of climate ch ange on species distributions. Global Change Biology 12:2272 2281. Holmes, R. T. and S. K. Robinson. 1981. Tree species preferences of foraging insectivorous birds in a northern hardwoods forest. Oecologia 48:3135. Holt, R. D. and T. H. Keitt. 2005. Species borders: a unifying theme in ecology. Oikos 108:3 -6. Holt, R. D., T. H. Keitt, M. A. Lewis, B. A. Maurer, and M. L. Taper. 2005. Theoretical models of species borders: Single species approaches. Oikos 108:18 -27.

PAGE 249

249 Holt, R. D. 2009. Bringing the Hutchinsonian niche into the 21st century: Ecological and evolutionary perspectives. Proc eedings of the National Academy of Sciences 106:19659 -19665. Huey, R. B., and A. F. Bennett. 1987. Phylogenetic studies of coadaptation: Preferred temperatures versus optimal performance temperatures of lizards. Evolution 41:1098 -1115. Huey, R. B., C. A. Deutsch, J. J. Tewksbury, L. J. Vitt, P. E. Hertz, H. J. lvarez Prez, and T. Garland, Jr. 2009. Why tropical forest lizards are vulnerable to climate warming. Proceedings of the Royal Society B -Biological Sciences 276:19391948. Hughes, L. 2000. Biological consequences of global warming: is the signal already apparent? Trends in Ecology and Evolution 15:5661. Huntley, B. 1991. How plants respond to climate change. Migration rates, individualism and the consequences for plant communities. Annals of Bot any 67:15-22. Ibanez, I., J. S. Clark, and M. C. Dietze. 2009. Estimating colonization potential of migrant tree species. Global Change Biology 15:11731188. Iverson, L. R., M. W. Schwartz, and A. M. Prasad. 2004. How fast and far might tree species migr ate in the eastern United States due to climate change? Global Ecology and Biogeography 13:209-219. Jankowski, J. E., A. L. Ciecka, N. Y. Meyer, and K. N. Rabenold. 2009. Beta diversity along environmental gradients: implications of habitat specialization in tropical montane landscapes. Journal of Animal Ecology 78:315-327. Janzen, D. H. 1967. Why mountain passes are higher in the tropics. American Naturalist 101:233 -249. Jullien, M. and J. Thiollay. 1998. Multi -species territoriality and dynamic of Neo tropical forest understorey bird flocks. Journal of Animal Ecology 67:227252. Karr, J. R. and K. E. Freemark. 1983. Habitat selection and environmental gradients: dynamics in the "stable" tropics. Ecology 64:1481 -1494. Kearney, M. 2006. Habitat, environment and niche: what are we modelling? Oikos 115:187 -191. Kearney M., B. L. Phillips, C. R. Tracy, K. A. Christian, G. Betts, and W. P. Porter. 2008. Modelling species distributions without using species distributions: the cane toad in Australia unde r cu rrent and future climates. Ecography 31:423434.

PAGE 250

250 Kearney, M. and W. Porter. 2009. Mechanistic niche modelling: combining physiological and spatial data to predict species ranges. Ecology Letters 12:334350. Kissling, W. D., R. Field, and K. Bhning -Gaes e. 2008. Spatial patterns of woody plant and bird diversity: functional relationships or environmental effects? Global Ecology and Biogeography 17 :327-339. Koleff, P., K. J. Gaston, and J. J. Lennon. 2003. Measuring beta diversity for presenceabsence dat a. Journal of Animal Ecology 72:367382. Kratter, A. W. 1997. Bamboo specialization by Amazonian birds. Biotropica 29:100110. Kroodsma, D. E., B. E. Byers, E. Goodale, S. Johnson, and W. C. Liu. 2001. Pseudoreplication in playback experiments, revisited a decade later. Animal Behavior 61:1029 1033. Langkilde, T. and R. Shine. 2004. Competing for crevices: interspecific conflict influences retreat -site selection in montane lizards. Oecologia 140:684-691. La Sorte, F. A. and W. Jetz. 2010. Avian distri butions under climate change: toward improved projections. Journal of Experimental Biology 213:862 -869. Lee, D. C. and S. J. Marsden. 2008. Increasing the value of birdhabitat studies in tropical forests: choice of approach and habitat measures. Bird Conservation International 18:S109-S124. Lee, P Y. and J. T. Rotenberry. 2005. Relationships between bird species and tree species assemblages in forested habitats of eastern North America. Journal of Biogeography 32:1139 -1150. Legendre, P. and L. Legendre. 1998. Numerical Ecology. 2nd edition. Elsevier Science, Amsterdam Lennon, J. J., P. Koleff, J. J. D. Greenwood, and K. J. Gaston. 2001. The geographical structure of British bird distributions: diversit y, spatial turnover and scale. Journal of Animal Ecology 70:966979. Lichstein, J. W. 2007. Multiple regression on distance matrices: a multivariate spatial analysis tool. Plant Ecology 188:117 -131. Linares Palomino, R. and M. Kessler. 2009. The role of dispersal ability, climate, and spatial separat ion in shaping biogeographical patterns of phylogenetically distant plant groups in seasonally dry Andean forests of Bolivia. Journal of Biogeography 36:280290.

PAGE 251

251 Lindsay, C., C. Downs, and M. Brown. 2009. Physiological variation in Amethyst Sunbirds ( Chal comitra amethystina) over an altitudinal gradient: A seasonal comparison. Comparative biochemistry and physiology A molecular and integrative physiology 152:22 -45. Lomolino, M. V., B. R. Riddle, and J. H. Brown. 2006. Biogeography. Sinauer Associates Inc. Sunderland, Massachusetts, USA. Londoo, G. A., D. J. Levey, and S. K. Robinson. 2008. Effects of temperature and food on incubation behavior of the Northern Mockingbird ( Mimus polyglottos ). Animal Behavior 76:669 -677. MacArthur, R. H., and J. W. MacAr thur. 1961. On Bird Species Diversity. Ecology 42:594598. MacArthur, R. H. 1972. Geographical e cology: Patterns in the distribution of species. Harper & Row, Publishers, Inc., NewYork, NY, USA. Maca, M. J., K. Ruokolainen, H. Tuomisto, J. Quisbert, and V. Cala. 2007. Congruence between floristic patterns of trees and lianas in a so uthwest Amazonian rain forest. Ecography 30:561 -577. Magurran, A. E. 2004. Measuring Biological Diversity. Blackwell Science Ltd., Oxford, UK. Martin, P. R., J. R. Fothering ham, L. Ratcliffe, and R. J. Robertson. 1996. Response of American redstarts (suborder Passeri) and least flycatchers (suborder Tyranni) to heterospecific playback: the role of song in aggressive interactions and interference competition. Behavioral Ecolog y and Sociobiology 39:227-235. Martin, P. R., and T. E. Martin. 2001. Behavioral interactions between coexisting species: song playback experiments with wood warblers. Ecology 82:207-218. Martin, T. E. 1993. Nest predation and nest sites: New perspectives on old patterns. BioScience 43:523532. Martin, T. E., C. R. Paine, C. J. Conway, W. M. Hochachka, P. Allen, and W. Jenkins. 1997. BBIRD field protocol. Montana Cooperative Wildlife Research Unit, University of Montana, Missoula, Montana, USA. Martin, T. E. 2004. Avian life -history evolution has an eminent past: does it have a bright future? Auk 121:289 301.

PAGE 252

252 Martin, T. E., R. D. Bassar, S. K. Bassar, J. J. Fontaine, P. Lloyd, H. A. Mathewson, A. M. Nikklison, and A. Chalfoun. 2006. Life history and ecological correlates of geographic variation in egg and clutch mass among passerine species. Evolution 60:390 398. Martin, T. E., S. K. Auer, R. D. Bassar, A. M. Niklison, and P. Lloyd. 2007. Geographic variation in avian incubation periods and parental i nfluences on embryonic temperature. Evolution 61:25582569. Martin, T. E. 2008. Egg size variation among tropical and temperate songbirds: An embryonic temperature hypothesis. Proceedings of the National Academy of Sciences USA 105:92689271. Massaro, M. A. Starling Windhof, J. V. Briskie, and Martin, T. E. 2008. Introduced mammalian predators induce adaptive shifts in parental behaviour in an endemic New Zealand bird. PLoS O NE 3:e2331. doi:10.1371 /journal.pone.0002331. McCain, C. M. 2005. Elevational gradients in diversity of small mammals. Ecology 86:366372. McCain, C. M. 2009. Vertebrate range sizes indicate that mountains may be higher in the tropics. Ecology Letters 12:550560. McKechnie, A. E., and D. L. Swanson. 2010. Sources and significance of variation in basal summit and maximal metabolic rates in birds. Current Zoology in press. McKnight, M. W., P. S. White, R. I. McDonald, J. F. Lamoreux, W. Sechrest, R. S. Ridgely, and S. N. Stuart. 2007. Putting beta -diversity on the map: Broad -scale congruence and coincidence in the extremes. PLoS Biology 5:e272.doi:10.1371/journal.pbio.0050272. McNab, B. K. 2001. The energetics of toucans, a barbet, and a hornbill: Implications for avian frugivory. Auk 118:916-933. McNab, B. K. 2002. The physiological ecology of vertebrates. Cornell University press, Ithaca, New York, USA. McNab, B. K. 2003. Ecologi cal factors affect the level and scaling of avian BMR Comparative biochemistry and physiology A molecular and integrative physiology 152:2245. McNaughton, S. J., and L. L. Wolf. 1970. Dominance and the niche in ecological systems. Science 167:131139.

PAGE 253

253 Melo, A. S., T. F. L. V. B. Rangel, and J. A. F. Diniz -Filho. 2009. Environmental drivers of beta diversity patterns i n New -World birds and mammals. Ecography 32:226236. Merrill, R. M., D. Gutirrez, O. T. Lewis, J. Gutirrez, S. B. Dez, and R. J. Wils on. 2008. Combined effects of climate and biotic interactions on the elevational range of a phytophagous insect. Journal of Animal Ecology 77:145-155. Monahan, W. B. 2009. A mechanistic niche model for measuring species distributional responses to s eason al temperature gradients. PLoS ONE 4:e7921. doi:10.137/journal.pone.0007921. Moritz, C. M., J. L. Patton, C. J. Conroy, J. L. Parra, G. C. White, and S. R. Beissinger. 2008. Impact of a century of climate change on small mammal communities in Yosemite Nat ional Park, USA. Science 322:261264. Mller, J., C. Moning, C. Bssler, M. Heurich, and R. Brandl. 2009. Using airborne laser scanning to model potential abundance and assemblages of forest passerines. Basic and Applied Ecology 10:671-681. Munn, C. A., and J. W. Terborgh. 1979. Multi -species territoriality in Neotropical foraging flocks. Condor 81:338347. Munn, C. A. 1985. Permanent canopy and understor y flocks in Amazonia: Species composition and population density. Pages 683-712 in P. A. Buckley, M. S. Foster, E. S. Morton, R. S. Ridgel y, and F. G. Buckley, editors. Neotropical Ornithology Ornithological Monographs No. 36. The American Ornithologists Union, Washington D.C., USA. Murray, B. G. 1971. The ecological consequences of interspecific terri torial behavior in birds. Ecology 52: 414-423. Myers, N., R. A. Mittermeier, C. G. Mittermeier, G. A. B. da Fonseca, and J. Kent. 2000. Biodiversity hotspots for conservation priorities. Nature 403:853-858. Naoki, K. 2007. Arthropod resource partitioning among omnivorous tanagers (Tangara spp.) in Western Ecuador. Auk 124:197-209. Navas, C. A. 2003. Herpetological diversity along Andean elevational gradients: links with physiological ecology and evolutionary physiology. Comparative Biochemistry and Physiology Part A 133:469-485. Nikkilson, A. M. 2007. Influence of embryonic metabolic rate and incubation temperature on incubation length variation in Neotropical passerines. M.Sc. T hesis. University of Montana, Missoula, Montana, USA.

PAGE 254

254 Oksanen, J., and P. R Minchin. 2002. Continuum theory revisited: What shape are species responses along ecological gradients? Ecological Modelling 157:119 129. Oksanen, J., R. Kindt, P. Legendre, R. B. OHara, G. L. Simpson, M. H. H. Stevens, and H. Wagner. 2008. Vegan: com munity ecology package, version 1.172. Available at: http://vegan.r -forge.r Olson, C. R. C. M. Vleck, and D. Vleck. 2006. Periodic cooling of birds eggs reduces embryonic growth efficiency. Physiological and Biochemical Zoology 79:927936. Orians, G. H. and M. F. Willson. 1964. Interspecific territories of birds. Ecology 45:736745. Palmer, M. W. 1993. Putting things in even better order: The advantages of canonical correspondence anal ysis. Ecology 74:2215 -2230. Parker, T. A., III and J. P. O'Neill. 1980. Notes on little known birds of the upper Urubamba valley, southern Peru. Auk 97:167176. Parmesan, C. 2006. Ecological and evolutionary response to recent climate change. Annual R eview of Ecology Evolution and Systematics 37:637-669. Patterson, B. D., D. F. Stotz, S. Solari, J. W. Fitzpatrick, and V. Pacheco. 1998. Contrasting patterns of elevational zonation for birds and mammals in the Andes of southeastern Peru. Journal of Biog eography 25:593 -607. Pearson, R.G. and T. P. Dawson. 2003. Predicting the impacts of climate change on the distribution of species: are bioclimate envelope models useful? Global Ecology & Biogeography 12:361371. Peterson A. T., M. A. Ortega-Huerta, J. Bartley, V. Snchez -Cordero, J. Sorbern, R. H. Buddemeier, and D. R. B. Stockwell. 2002. Future projections for Mexican faunas under global climate change scenarios. Nature 416:626629. Pitman, N. C. A., J. W. Terborgh, M. R. Silman, and P. Nez V. 199 9. Tree species distributions in an upper Amazonian forest. Ecology 80:26512661. Pitman, N. C. A., J. W. Terborgh, M. R. Silman, P. Nez V. D. A. Neill, C. E. Cern, W. A. Palacios, and M. Aulestia. 2001. Dominance and distribution of tree species in u pper Amazonian terra firme forests. Ecology 82:21012117. Potts, M., P. S. Ashton, L. S. Kaufman, and J. B. Plotkin. 2002. Habitat patterns in tropical rain forests: A comparison of 105 plots in Northwest Borneo. Ecology 83:2782 -2797.

PAGE 255

255 Poulin, B., G. Lefe bvre, and R. McNeil. 1994. Characteristics of feeding guilds and variation in diets of bird species of three adjacent tropical sites. Biotropica 26:187197. Poulsen, B. O. 1996. Structure, dynamics, home range and activity pattern of mixed species bird f locks in a montane Alder dominated secondary forest in Ecuador. Journal of Tropical Ecology 12:333 343. Pounds, J. A., M. P. L. Fogden, and J. H. Campbell. 1999. Biological response to climate change on a tropical mountain. Nature 398:611-615. Powell, G. V. N. 1979. Structure and dynamics of interspecific flocks in a Neotropical mid elevation forest. Auk 96:375390. Price, T. D. and M. Kirkpatrick. 2009. Evolutionarily stable range limits set by interspecific competition. Proceedings of the R oyal Society B -Biological Sciences 276:14291434. R D evelopment Core Team. 2010. R: A language and environment for statistical computing. R Foundation for Statistical Computing, Vienna, Austria. Available at: http: //www.r / (accessed April 10, 2010). Rahn, H., C. Carey, K. Balmas, B. Bhatia, and C. Paganelli. 1977. Reduction of pore area of the avian eggshell as an adaptation to altitude. Proceedings of the National Academy of Sciences USA 74:3095-3098. Remsen, J. V., Jr. and W. S. Graves. 1995. Distribution patterns and zoogeography of Atlapetes BrushFinches (Emberizinae) of the Andes. Auk 112:225 -236. Remsen, J. V., Jr., C. D. Cadena, A. Jaramillo, M. Nores, J. F. Pacheco, M. B. Robbins, T. S. Schu lenberg, F. G. Stiles, D. F. Stotz, and K. J. Zimmer. Version [June 2010]. A classification of the bird species of South America. American Ornithologists' Union. Restrepo, C. and N. Gmez. 1998. Respons es of understory birds to anthropogenic edges in a neotropical montane forest. Ecological Applications 8:170183. Richards, D. G. 1979. Recognition of neighbors by associative learning in Rufous -sided Towhees. Auk 96: 688-693. Richards, P. W. 1996. The t ropical rain forest: An ecological study. Cambridge University Press, Cambridge, UK. Ricklefs, R. E. 1969. An analysis of nesting mortality in birds. Smithsonian Contribution to Z oology Number 9.

PAGE 256

256 Ritchie, E. G., J. K. Martin, C. N. Johnson, and B. J. Fox 2009. Separating the influences of environment and species interactions on patterns of distribution and abundance: competition between large herbivores. Journal of Animal Ecology 78:724731. Robinson, S. K. and R. T. Holmes. 1984. Effects of plant spec ies and foliage structure on the foraging behavior of forest birds. Auk 101:672684. Robinson, S. K. and J. Terborgh. 1995. Interspecific aggression and habitat selection by Amazonian birds. Journal of Animal Ecology 64:1-11. Romdal, T. S., and C. Rahbek. 2009. Elevational zonation of afrotropical forest bird communities along an homogeneous forest gradient. Journal of Biogeography 36:327336. Rotenberry, J. T. 1985. The role of habitat in avian community composition: physiognomy or floristics? Oecologi a 67:213217. Ruokolainen, K., H. Tuomisto, M. J. Maca, M. A. Higgins, and M. Yli Halla. 2007. Are floristic and edaphic patterns in Amazonian rain forests congruent for trees, pteridophytes and Melastomataceae? Journal of Tropical Ecology 23:1325. Saatchi, S., R. Houghton, R. Avala, Y. Yu, and J -V. Soares. 2007. Spatial Distribution of Live Aboveground Biomass in Amazon Basin. Global Change Biology 13:816 837. Sagarin, R. D., and S. D. Gaines. 2002. The abundant centre distribution: To what extent is it a biogeographic rule? Ecology Letters 5:137 -147. Sagarin, R. D., S. D. Gaines, and B. Gaylord. 2006. Moving beyond assumptions to understand abundance distributions across the ranges of species. Trends in Ecology and Evolution 21:524530. Santika, T., and M. F. Hutchinson. 2009. The effect of species response form on species distribution model prediction and inference. Ecological Modelling 220:23652379 SAS Institute. 2008. SAS/STAT Users Guide, version 9.2. SAS Institute, Cary, North Carolina. Schmidt Nielsen, K. 1997. Animal Physiology: Adaptation and Environment. Cambridge University Press, Cambridge, UK. Schulenberg, T. S., D. F. Stotz, D. F. Lane, J. P. ONeill, and T. A. Parker III. 2007. Birds of Peru. Princeton University Press, Princet on, New Jersey, USA.

PAGE 257

257 Sekercioglu, C. H., S. H. Schneider, J. P. Fay, and S. R. Loarie. 2008. Climate change, elevational range shifts, and bird extinctions. Conservation Biology 22:140 -150. Silvertown, J. 2004. Plant coexistence and the niche. Trends in Ecology and Evolution 19:605611. Skutch, A. F. 1985. Clutch size, nesting success, and predation on nests of tropical birds, reviewed. Pages 575 5 94 in P. A. Buckley, M. S. Foster, E. S. Morton, R. S. Ridgely, and F. G. Buckley, editors. Neotropical Ornithology. Ornithological Monographs No. 36. The American Ornithologists Union, Washington D.C., USA. Sobern, J. 2007. Grinnellian and Eltonian niches and geographic distrib utions of species. Ecology Letters 10:1115 1123. Solomon, S., D. Qin, M. Manning, Z. Chen, M. Marquis, K. B. Averyt, M. Tignor, and H. L. Miller. 2007. A Contribution of Working Group I to the Fourth Assessment Report of the Intergovernmental Panel on Climate Change. Cambridge University Press, New York, New York, USA. Soobramoney, S ., C. T. Downs, and N. J. Adams. 2003. Physiological variability in the Fiscal Shrike ( Lanius collaris) along an altitudinal gradient in South Africa. Journal of thermal biology 28 :581-594. Soriano, P. J., A. Ruiz, and A. Arends. 2002. Physiological responses to ambient temperature manipulation by three species of bats from Andean cloud forests. Journal of Mammalogy 83:445457. Stiles, F. G. 1975. Ecology, flowering phenology, and hummingbird pollination of some Costa Rican Heliconia species. Ecology 56:2 85 -301. Stiles, F. G. 1985. Seasonal patterns and coevolution in the hummingbird-flower community of a Costa Rican subtropical rainforest. Pages 757787 in P. A. Buckley, M. S. Foster, E. S. Morton, R. S. Ridgely, and F. G. Buckley, editors. Neotropical O rnithology. Ornithological Monographs No. 36. The American Ornithologists Union, Washington D.C., USA. Stiles, F. G., A. Skutch, and D. Gardner. 1989. A Guide to the Birds of Costa Rica. Cornell University Press, Ithaca. Stiles, F. G. 1995. Behavioral, ecological and morphological correlates of foraging for arthropods by the hummingbirds of a tropical wet forest. Condor 97:853-878. Stotz, D. F., J. W. Fitzpatrick, T. A. Parker III, and D. K. Moskovits. 1996. Neotropical Birds. Ecology and Conservation. University of Chicago Press, Chicago.

PAGE 258

258 Takahashi, K. and T. Kamitani. 2004. Effect of dispersal capacity on forest plant migration at a landscape scale. Journal of Ecology 92:778785. Terborgh, J. 1971. Distributions on environmental gradients: Theory and a preliminary interpretation of distributional patterns in the avifauna of the Cordillera Vilcabamba, Peru. Ecology 52:23-40. Terborgh, J. and J. S. Weske. 1975. The role of competition in the distribution of Andean birds. Ecology 56 :562 576. Terbor gh, J. 1977 Bird species diversity on an Andean elevational gradient. Ecology 58:1007 -1019. Terborgh, J. 1985a. The role of ecotones in the distribution of birds. Ecology 66:12371246. Terborgh, J. 1985b. Habitat selection in Amazonian birds. Pages 311 -3 38 in M. L. Cody, editor. Habitat selection in birds Academic Press, New York New York, USA Terborgh, J., S. K. Robinson, T. A. Parker III, C. A. Munn, and N. Pierpont. 1990. Structure and organization of an Amazonian forest bird community. Ecological Monographs 60:213 -238. Terborgh, J., R. B. Foster, and P. Nez V. 1996. Tropical tree communities: A test of the nonequilibrium hypothesis. Ecology 77:561 567. Ter Braak, C. J. F. 1985. Correspondence analysis of incidence and abundance data: Propertie s in terms of a unimodal response model. Biometrics 41:859 -873. Tewksbury, J. J., B. H. Raymond, and C. A. Deutsch. 2008. Putting the heat on tropical animals. Science 320:12961297. Thomas, C. D. and J. J. Lennon. 1999. Birds extend their ranges northw ards. Nature 399:213. Thomson, D. L., P. Monaghan, and R. W. Furness. 1998. The demands of incubation and avian clutch size. Biological review of the Cambridge philosophical society 73:293304. Thuiller, W., C. Albert, M. B. Arajo, P. M. Berry, M. Cabez a, A. Guisan, T. Hickler, G. F. Midgley, J. Paterson, F. M. Schurr, M. T. Sykes, and N. E. Zimmerman. 2008. Predicting global change impacts on plant species' distributions: Future challenges. Perspectives in Plant Ecology Evolution and Systematics 9:137-1 52. Tich, L. 2002. JUICE, software for vegetation classification. Journal of Vegetation Science 13:451-453.

PAGE 259

259 Tich, L., and J. Holt. 2006 JUICE: Program for management, analysis and classification of ecological data. Vegetation Science Group, Masaryk Uni versity Brno, Czech Republic. Tinbergen, J. M. and J. B. Williams. 2002. Energetics of incubation. Pages 300 313 in Avian incubation Behaviour, Environment, and Evolution (D. C. Deeming, Editor). Oxford University Press, New York, USA Tingley, M. W., W B. Monahan, S. R. Beissinger, and C. Moritz. 2009. Birds track their Grinnellian niche through a century of climate change. Proceedings of the National Academy of Sciences 106:19637-19643. Trillo, P. A. and S. L. Vehrencamp. 2006. Song types and thei r structural features are associated with specifi c contexts in the banded wren. Animal Behavior 70:921935. Tuomisto, H. 2010. A diversity of beta diversities: straightening up a concept gone awry. Part 2. Quantifying beta diversity and related phenomena. Ecography 33:23 -45. Turner, J. S. 2002. Maintenance of egg temperature. Pages 119 142 in D. C. Deemin, editor. Avian incubation Behaviour, Environment, and Evolution. Oxford University Press, New York, USA. Turner, W., S. Spector, N. Gardiner, M. Flade land, E. Sterling, and M. Steininger. 2003. Remote sensing for biodiversity science and conservation. Trends in Ecology and Evolution 18:306314. Urrutia, R. and M. Vuille. 2009. Climate change projections for the tropical Andes using a regional climate model: Temperature and precipitation simulations for the end of the 21st century. Journal of Geophysical Research 114:D02108. van Schaik, C. P., J. W. Terborgh, and S. J. Wright. 1993. The phenology of tropical forests: adaptive significance and consequences for primary consumers. Annual Review of Ecology and Systematics 24:353377. Vsquez G., J. A. and T. J. Givnish 1998. Altitudinal gradients in tropical forest composition, structure, and diversity in the Sierra de Manantln. Journal of Ecology 86 : 99 9 -1020. Vierling, K. T., L. A. Vierling, W. A. Gould, S. Martinuzzi, and R. M. Clawges. 2008. LIdar: Shedding new light on habitat characterization and modeling. Frontiers in Ecology and the Environment 60 :9098. Visser, M. E., A. C. Perdeck, J. H. van B alen, and C. Both. 2009. Climate change leads to decreasing bird migration distances. Global Change Biology 15:18591865.

PAGE 260

260 Walker, B., D. F. Stotz, T. Pequeo, and J. W. Fitzpatrick. 2006. Birds of the Manu Biosphere Reserve. Fieldiana Zoology 110:2349. Webb, D. R. 1987. Thermal tolerance of avian embryos: a review. Condor 89:874 898. Weske, J. S. 1972. The distribution of the avifauna in the Apurimac valley of Peru with respect to environmental gradients, habi tat, and related species. PhD D issertation, University of Oklahoma, Norman, Oklahoma, USA. Whittaker, R. H. 1975. Communities and Ecosystems. MacMillan, New York, USA. Wiens, J. A. 1989a. The ecology of bird communities: volume 1: foundations and patterns. Cambridge University Press, Cambridge, UK Wiens, J. A. 1989b. The ecology of bird communities: volume 2: processes and variations. Cambridge University Press, Cambridge UK Wiersma, P., A. Muoz -Garcia, A. Walker, and J. B. Williams. 2007. Tropical birds have a slow pace of life. Proceedings of the National Academy of Sciences USA 104:93409345. Williams, G. C. 1966. Natural selection, the cost of reproduction, and a refinement of Lacks principle. American Naturalist 100:687 690. Williams, J. W., S. T. Jackson, and J. E. Kutzbach. 2007. Pr ojected distributions of novel and disappearing climates by 2100 AD. Proceedings of the National Academy of Sciences 104:5738-5742. Wilson, R. J., J. Gutirrez, D. Martnez, R. Agudo, and V. J. Monserrat. 2005. Changes to the elevational limits and extent of species ranges associated with climate change. Ecology Letters 8:1138 -1146. Zuckerberg, B., A. M. Woods, and W. F. Porter. 2009. Poleward shifts in breeding bird distributions in New York State. Global Change Biology 15:18661883.

PAGE 261

261 BIOGRAPHICAL SKET CH Jill Emily Jankowski was born in June of 1980 and grew up in Southwestern Indiana near the Ohio River basin. Her parents own property adjacent to a large tract of forest, which served as her favorite stomping grounds throughout childhood. The close ti es she developed with the natural world later inspired her to study ecology in far away places and close to home. Jill graduated in 1998 from F.J. Reitz High School on the west side of Evansville, and she attended Purdue University for her undergraduate e ducation. She began as a student in civil engineering and played soccer on the Womens Boilermaker varsity team, but her academic interests turned to biology in her second year. She graduated from Purdue with a Bachelor of Science in ecology, evolution and population biology in 2002. Jill was accepted into graduate school at Purdue, where she continued research begun with Dr. Kerry Rabenold as an undergraduate examining patterns of diversity and endemism of bird communities in the cloud forests of Costa Rica. In the spring of 2004, she obtained her Master of Science in the Department of Biological Sciences upon the completion of her thesis: Patterns of avian diversity and the interspecific abundance distribution relationship in the Tilarn Mountain Rang e, Costa Rica. Jill began her doctoral dissertation work in the fall of 2004 in the Department of Zoology at the University of Florida with her co advisors, Dr. Doug Levey and Dr. Scott Robinson. There she continued her studies of montane birds in the t ropical Andes, honing in on some of the ecological forces that determine where species live along large -scale elevational gradients. This dissertation is the product of over 20 months of data collection from field sites in Costa Rica and Peru. Outside of her field seasons, Jill has resided in Gainesville, Florida, with her fianc, Aaron Spalding, and her two cats, Missy and Duchess.