anystyle 1.1.0 → 1.2.0

data/lib/anystyle/utils.rb

@@ -25,18 +25,23 @@ module AnyStyle
       scrub(transliterate(string)).downcase
     end
 
+    def nnum(string, symbol = '#')
+      string.unicode_normalize.gsub(/\d/, symbol)
+    end
+
     def page_break?(string)
       string =~ /\f/
     end
 
     def display_width(string)
-      display_chars(string).rstrip.length
+      display_chars(string).length
     end
 
     def display_chars(string)
       string
         .gsub(/\p{Mn}|\p{Me}|\p{Cc}/, '')
         .gsub(/\p{Zs}/, ' ')
+        .rstrip
     end
 
     def count(string, pattern)
@@ -44,7 +49,7 @@ module AnyStyle
     end
 
     def indent(token)
-      display_chars(token).rstrip[/^(\s*)/].length
+      display_chars(token)[/^(\s*)/].length
     end
 
     def strip_html(string)
@@ -53,11 +58,13 @@ module AnyStyle
     end
   end
 
-  module PdfUtils
+  module PDFUtils
     module_function
 
-    def pdf_to_text(path, layout: true)
-      %x{pdftotext #{layout ? ' -layout' : ''} -eol unix -q "#{path}" -}
+    def pdf_to_text(path, **opts)
+      text = %x{pdftotext #{pdf_opts(path, **opts).join(' ')} "#{path}" -}
+      raise "pdftotext failed with error code #{$?.exitstatus}" unless $?.success?
+      text.force_encoding(opts[:encoding] || 'UTF-8')
     end
 
     def pdf_info(path)
@@ -69,6 +76,43 @@ module AnyStyle
     def pdf_meta(path)
       %x{pdfinfo -meta -isodates "#{path}"}
     end
+
+    def pdf_page_size(path)
+      pdf_info(path)['Page size'].scan(/\d+/)[0, 2].map(&:to_i)
+    end
+
+    private
+
+    def pdf_opts(path, layout: true, encoding: 'UTF-8', **opts)
+      [
+        layout ? '-layout' : '',
+        opts[:crop] ? pdf_crop(path, opts[:crop]) : '',
+        '-eol unix',
+        "-enc #{encoding}",
+        '-q'
+      ]
+    end
+
+    def pdf_crop(path, args)
+      (x, y, w, h) = case args.length
+        when 1
+          [args[0], args[0], -args[0], -args[0]]
+        when 2
+          [args[0], args[1], -args[0], -args[1]]
+        when 4
+          args
+        else
+          raise "invalid crop option: #{args}"
+        end
+
+      if w < 0 || h < 0
+        (width, height) = pdf_page_size(path)
+        w = width - x + w if w < 0
+        h = height - y + h if h < 0
+      end
+
+      "-x #{x} -y #{y} -W #{w} -H #{h}"
+    end
   end
 
   extend Utils

data/lib/anystyle/version.rb

@@ -1,3 +1,3 @@
 module AnyStyle
-  VERSION = '1.1.0'.freeze
+  VERSION = '1.2.0'.freeze
 end

data/res/finder/bb132pr2055.ttx

@@ -241,7 +241,7 @@ blank         |
               | 
               | 
 meta          |                                           ix
-              | Chapter 1
+title         | Chapter 1
 blank         | 
 title         | Introduction
 blank         | 
@@ -261,8 +261,8 @@ blank         |
 text          |       The reason Bernanke took such a public stand defending the Fed’s independence
               | is simple: central bank independence is widely seen as a necessary condition for the
               | attainment of economic and financial stability. Thus, Bernanke and central bankers
-              |   1
-              |    Text of full statement is available at http://www.federalreserve.gov/newsevents/speech/
+meta          |   1
+text          |    Text of full statement is available at http://www.federalreserve.gov/newsevents/speech/
               | bernanke20100203a.htm.
 blank         | 
               | 
@@ -293,8 +293,8 @@ text          | everywhere have a substantial interest in promoting the idea tha
               | thereby ignoring the crucial role that central banks play as lenders of last resort.
               |        Other studies have taken a more targeted approach in studying the impact of
               | politics on policymaking at the Fed by asking whether or not monetary policy becomes
-              |    2
-              |     For a review of the literature investigating this relationship, see Eijffinger and de Haan (1996).
+meta          |    2
+text          |     For a review of the literature investigating this relationship, see Eijffinger and de Haan (1996).
               | Bade and Parkin (1988), Alesina (1988), Grilli, Masciandaro and Tabellini (1991), Cukierman, Web
               | and Neyapti (1992), Alesina and Summers (1993), Eijffinger, Van Rooij and Schaling (1996), Berger,
               | De Haan and Eijffinger (2001), and Keefer and Stasavage (2003), among others, provide additional
@@ -389,7 +389,7 @@ blank         |
 text          | to remain insulated from external political pressure is imperfect.
               |    I conclude in Chapter 5 by discussing the implications of my findings and mapping
               | out some research projects that are natural extensions of this work.
-meta          | Chapter 2
+title         | Chapter 2
 blank         | 
 title         | The Politics Behind the Federal
               | Reserve’s Structure
@@ -1208,8 +1208,8 @@ text          |                                                Paired    Paired
               |                  Republican      6      34     0         3         1          44
               |                  Progressive     1      0      0         0         0          1
               |                  Total           54     34     3         3         1          95
-              |  19
-              |       This small group of midwesterners, led by Lindberg, opposed any sort of central bank.
+meta          |  19
+text          |       This small group of midwesterners, led by Lindberg, opposed any sort of central bank.
 meta          | CHAPTER 2. THE POLITICS BEHIND THE FED’S STRUCTURE                                     35
 blank         | 
               | 
@@ -2744,9 +2744,9 @@ text          |                           Table 2.11: Partisan Preferences in 19
               | R2    Edward         non-eastern            B       Horizontal   cost of moving away from their
               |       Vreeland       bankers                                     desired level of centralization
               | Key: A = politicized central bank
-meta          |      B = decentralized, private bank with autonomous branches
+              |      B = decentralized, private bank with autonomous branches
               |      C = centralized bank controlled by Wall Street
-              | CHAPTER 2. THE POLITICS BEHIND THE FED’S STRUCTURE                             86
+meta          | CHAPTER 2. THE POLITICS BEHIND THE FED’S STRUCTURE                             86
 blank         | 
               | 
               | 
@@ -2765,7 +2765,7 @@ blank         |
               | 
 text          | Figure 2.14: Could the Banking Act have Passed Had the Composition of Congress
               | Not Changed Between 1913 and 1935?
-meta          | Chapter 3
+title         | Chapter 3
 blank         | 
 title         | The Fed as a Political Agent: How
               | Partisan Central Bankers Allow
@@ -3079,7 +3079,7 @@ blank         |
               | 
 title         | 3.2      Do We Know if There is a Political Business
 blank         | 
-text          |          Cycle in Monetary Policy?
+title         |          Cycle in Monetary Policy?
 blank         | 
 text          | Most studies assessing the presence of a political business cycle in monetary policy
               | have decided against pursuing this more fine-grained, individual-level analysis and
@@ -3628,8 +3628,8 @@ text          | partisan identity by using information from press reports at the
               | four are Independents, and five are Republicans. The remaining 15 are not able to be
               | identified, since they either are (or were) not a registered voter or are not included in
               | the Catalist voter file that I have access to through my university’s subscription.34
-              |   33
-              |      Other studies using Catalist for data on voter registrations include Ansolabehere and Hersh
+meta          |   33
+text          |      Other studies using Catalist for data on voter registrations include Ansolabehere and Hersh
               | (2012), Hersh and Nall (2013), and Peterson (2014).
 meta          |   34
 text          |      The subscription is for only a subset of the entire Catalist national voter file.
@@ -4363,7 +4363,7 @@ meta          |      Note:                                     p<0.1;        p<0
 blank         | 
               | 
 text          |  Figure 3.5: Power Analysis: Democratic Elections’ Effect on FOMC Republicans
-meta          | Chapter 4
+title         | Chapter 4
 blank         | 
 title         | The Role of Politics in the Federal
               | Reserve’s Lending during the
@@ -4874,8 +4874,8 @@ text          | officials expending their legislative resources on issues unrela
               | respondence (i.e., letters) issued between the board and Congress pertaining to the
               | Fed’s emergency lending during the financial crisis.19
               |     The over 3,300 pages that were released included 69 letters from congresspersons
-              |    16
-              |       This point is even more true if the campaign contributions are made to politicians that are never
+meta          |    16
+text          |       This point is even more true if the campaign contributions are made to politicians that are never
               | even elected to office. After all, a firm can make a campaign contribution for many reasons, not just
               | to try to influence the Fed’s lending.
 meta          |    17
@@ -5891,8 +5891,8 @@ blank         |
               | 
               | 
               | 
-meta          |                       (c) Political Entity: Letter Writers
-              | CHAPTER 4. THE ROLE OF POLITICS IN THE FED’S LENDING                                     184
+text          |                       (c) Political Entity: Letter Writers
+meta          | CHAPTER 4. THE ROLE OF POLITICS IN THE FED’S LENDING                                     184
 blank         | 
               | 
               | 
@@ -5957,7 +5957,7 @@ text          |  Figure 4.3: Financial Covariates’ Marginal Effects on Income
               | AIC used by Stata            3709.097        3705.628        3702.708      3705.485         3699.858        3696.849         3698.747
 meta          |                                                                                                                                               CHAPTER 4. THE ROLE OF POLITICS IN THE FED’S LENDING
               |                                                                                                                                               186
-              | Chapter 5
+title         | Chapter 5
 blank         | 
 title         | Conclusion
 blank         | 

data/res/finder/bb408gp7470.ttx

@@ -0,0 +1,3919 @@
+title         |  FUNCTIONAL TRAIT MEDIATION OF PLANT-ANIMAL INTERACTIONS:
+              |  EFFECTS OF DEFAUNATION ON PLANT FUNCTIONAL DIVERSITY IN A
+              |                     NEOTROPICAL FOREST
+blank         | 
+              | 
+              | 
+              | 
+text          |                       A DISSERTATION
+              |           SUBMITTED TO THE DEPARTMENT OF BIOLOGY
+              |           AND THE COMMITTEE ON GRADUATE STUDIES
+              |                   OF STANFORD UNIVERSITY
+              | IN PARTIAL FULFILLMENT OF THE REQUIREMENTS FOR THE DEGREE OF
+              |                    DOCTOR OF PHILOSOPHY
+blank         | 
+              | 
+              | 
+              | 
+text          |                        Erin Leigh Kurten
+              |                          August 2010
+              |                   © 2010 by Erin Leigh Kurten. All Rights Reserved.
+              |          Re-distributed by Stanford University under license with the author.
+blank         | 
+              | 
+              | 
+text          |                          This work is licensed under a Creative Commons Attribution-
+              |                          Noncommercial 3.0 United States License.
+              |                          http://creativecommons.org/licenses/by-nc/3.0/us/
+blank         | 
+              | 
+              | 
+              | 
+text          | This dissertation is online at: http://purl.stanford.edu/bb408gp7470
+blank         | 
+              | 
+              | 
+              | 
+meta          |                                           ii
+text          | I certify that I have read this dissertation and that, in my opinion, it is fully adequate
+              | in scope and quality as a dissertation for the degree of Doctor of Philosophy.
+blank         | 
+text          |                                                             Rodolfo Dirzo, Primary Adviser
+blank         | 
+              | 
+              | 
+text          | I certify that I have read this dissertation and that, in my opinion, it is fully adequate
+              | in scope and quality as a dissertation for the degree of Doctor of Philosophy.
+blank         | 
+text          |                                                                                 Peter Vitousek
+blank         | 
+              | 
+              | 
+text          | I certify that I have read this dissertation and that, in my opinion, it is fully adequate
+              | in scope and quality as a dissertation for the degree of Doctor of Philosophy.
+blank         | 
+text          |                                                                                 David Ackerly
+blank         | 
+              | 
+              | 
+              | 
+text          | Approved for the Stanford University Committee on Graduate Studies.
+              |                                  Patricia J. Gumport, Vice Provost Graduate Education
+blank         | 
+              | 
+              | 
+              | 
+text          | This signature page was generated electronically upon submission of this dissertation in
+              | electronic format. An original signed hard copy of the signature page is on file in
+              | University Archives.
+blank         | 
+              | 
+              | 
+              | 
+meta          |                                           iii
+title         | Preface
+              | Dissertation Abstract
+text          |         This dissertation examines how terrestrial vertebrates, as seed dispersers, seed
+              | predators and herbivores, influence plant functional trait composition in tropical forests and
+              | thereby diversity. I conducted this work in the Barro Colorado National Monument (BCNM)
+              | in Central Panama, where a long term mammal exclosure experiment has been ongoing, and in
+              | neighboring Parque Nacional Soberanía (PNS), which together with the BCNM forms a
+              | defaunation gradient driven by hunting.
+              |         I first comprehensively review what is known about how the loss of vertebrates in
+              | tropical forests alters plant-animal interactions, plant demography, and plant diversity.
+              | Defaunation consistently lowers primary dispersal and creates a seed shadow that is more
+              | dense around the parent tree and less dense at sites farther away. However, it also often
+              | lowers seed predation by rodents, and as a consequence, species with rodents as seed predators
+              | and dispersers often benefit from defaunation. While demographic and diversity responses
+              | tend to be more mixed, a few consistent trends emerge. Community dominance tends to
+              | increase in response to defaunation. Often, plants with particular functional traits or abiotic or
+              | unhunted dispersal agents are favored by defaunation.
+              |         I next examined how community-level functional trait composition shifts in seedling
+              | communities (Chapter 2) and sapling communities (Chapter 3) which have experienced
+              | exclosure from terrestrial mammals. Seedling communities in exclosures had higher median
+              | seed mass than paired plots open to the mammal community, but treatments did not differ in
+              | their leaf traits (leaf mass per area and laminar toughness) or wood density. In contrast to the
+              | seedling community, the sapling community did show significant shifts toward higher specific
+              | leaf area and lower leaf toughness in response to herbivore exclosure, primarily due to an
+              | increased dominance of species with those traits, and secondarily due to differences in the
+blank         | 
+              | 
+              | 
+meta          |                                                 iv
+text          | species present in each treatment type.   These data, combined with data from PNS, also
+              | suggest that hunting results in community mean wood density in seedling communities, due to
+              | a disproportionate number of high wood density species relying on hunted animals for their
+              | seed dispersal.
+              |         Finally, I investigated the seed size response to changes in mammal abundance by
+              | measuring vertebrate seed predation rates in a protected and hunted forest (Chapter 4). I
+              | found that in central Panama, seed mass does not correlate well with either body size of the
+              | seed predator, or vertebrate seed predation rates. I suggest that rather than formulate seed
+              | predation rates as a linear function of seed predator abundance, these interactions may be
+              | better modeled as threshold-dependent processes.
+blank         | 
+              | 
+text          | This work suggests that terrestrial vertebrates play an underappreciated role in maintaining
+              | plant diversity and that pan-tropical levels of unsustainable hunting may indirectly lead to
+              | losses of plant biodiversity.
+blank         | 
+              | 
+              | 
+              | 
+meta          |                                              v
+title         | Acknowledgments
+text          | My deep gratitude goes to my two advisors, Rodolfo Dirzo and David Ackerly, for giving me
+              | the flexibility and support to pursue this work. Rodolfo’s generosity as a teacher and advisor,
+              | and his tireless contributions to teaching, research and conservation, both at Stanford and
+              | internationally, are a great inspiration to me. I am thankful to David for graciously and
+              | supportively allowing me to build a network of intellectual support that allowed me to
+              | successfully complete this journey. He has also been an excellent intellectual role model for
+              | me, challenging me to grapple with complexity and nuance and to confront problems from
+              | new perspectives.
+              |         Joe Wright has been a generous and supportive mentor and collaborator throughout
+              | my time at the Smithsonian Tropical Research Institute, and some of this work would not have
+              | been possible without his contributions. Peter Vitousek contributed many insightful
+              | suggestions throughout the course of this work as a committee member, and his service
+              | activities, from facilitating collaboration among scientists in Hawaii, to the First Nations’
+              | Futures Program, have also inspired me. I also thank Fio Micheli for her enthusiasm as a
+              | committee member and for helping me frame my work in a broader context. Walter Carson
+              | very generously permitted me to work on the long-term mammal exclosure experiment he
+              | established in the Barro Colorado National Monument in Panama and shared with me his
+              | long-term datasets.
+              |         The communities at Stanford, Berkeley and STRI were essential in shaping my
+              | graduate school experience. At Stanford, Will Cornwell, Nathan Kraft, Virginia Maztek,
+              | Steve Allison, Stephen Porder, Katie Amatangelo, Jen Funk, Camila Donatti, and Mauro
+              | Galetti provided helpful feedback and support throughout the various stages of my work.
+              | Doug Turner was an essential help in analyzing leaf nutrients, both for chapter three, and work
+              | not included in this thesis. Alex Royo, Allen Herre, Scott Magnan, Liza Comita, Mike Tobin,
+blank         | 
+              | 
+meta          |                                                 vi
+text          | Patrick Jansen, Noelle Beckman, Roland Kays, Jackie Willis, Stephan Schnitzer generously
+              | shared their advice and knowledge of the field site with me, improving the feasibility and
+              | execution of this work.
+              |         Many thanks go to the botanists at BCI. Without their assistance, this work would not
+              | have been possible. Andrés Hernández, Oldemar Valdes, David Brassfield, and Osvaldo
+              | Calderón were always happy to help me identify whatever leaf, fruit, seed, or flower I
+              | brought to their office. Andrés and Oldemar in particular taught me most of what I know
+              | about the BCI flora.
+              |         Many paid and volunteer field and lab assistants helped to make this work possible.
+              | Lissie Jiménez helped immensely with the collection and processing of the thousands of
+              | leaves collected for Chapter three. Ana Patricia Calderón and Rousmery Bethancourt
+              | contributed many early mornings conducting mammal transect surveys. Clare Sherman was
+              | such a dedicated help in the lab and fieldwork for chapter 4, that I sometimes had to remind
+              | her to take a break and have some fun. Susan Rebellon helped with the pilot studies for
+              | chapter four, and also with leaf sample processing at Stanford. Gaspar Bruner, Karen
+              | Kapheim, Adam Roddy, and David Bethancourt, helped me recover (most of) my leaf
+              | samples after they were destroyed in a freezer accident.
+              |         I would also like to thank staff at Stanford and STRI who helped make the logistical
+              | aspects of this work easier. The competence of Pam Hung, Oris Acevedo, Belkys Jiménez,
+              | Orelis Arosemana, and Marcela Paz made them a pleasure to work with. The Falconer library
+              | and copy staff, as well as Allen Smith, did their best to get me the literature I needed, despite
+              | STRI’s rigorous firewall. Valerie Kiszka and Jennifer Mason helped with countless
+              | administrative tasks and advice during my time as a student at Stanford.
+blank         | 
+              | 
+              | 
+              | 
+meta          |                                                 vii
+title         | Contents
+text          | Preface                                                                         iv
+              | Acknowledgements                                                                vi
+              | List of tables                                                                  ix
+              | List of figures                                                                 x
+blank         | 
+text          | Introduction………………………………………………………………………...                                      1
+              | 1 Contemporaneous defaunation and cascading effects on tropical forests…...…    3
+              |       Introduction………………………………………………………………….                                    3
+              |       Scope of Review…………………………………………………………….                                   4
+              |       Methodology………………………………………………………………...                                    5
+              |       Plant-animal interactions…………………………………………………....                          7
+              |             Seed dispersal………………………………………………………....                             7
+              |             Seed Predation…………………………………………………………                                14
+              |             Herbivory & Trampling…………………………………………….…                            17
+              |       Plant Demography……………………………………………………….….                                  17
+              |             Recruitment……………………………………………………………                                  17
+              |             Seedling survival………………………………………………………                              19
+              |             Standing abundance……………………………………………….…..                            20
+              |             Linking Dispersal and Seedling Recruitment…………………………                22
+              |       Community Diversity….……………………………………………………                                 24
+              |             Seedling density……………………………………………………….                              24
+              |             Diversity………………………………………………………………                                   24
+              |             Plant Functional Groups………………………………………………                           25
+              |       What is defaunation? ……………………………………………………….                               27
+              |       Heterogeneity Among Studies………………………………………………                             28
+              |       Conclusions……………………………………………………………….…                                     29
+              |       Appendix 1.1……………………………………………………………..…                                    30
+blank         | 
+text          | 2   Reduced seed dispersal as a consequence of hunting lowers community-level
+              |     wood density in a Neotropical forest ………………………………………                        35
+              |       Abstract……………………………………………………………………                                        35
+              |       Introduction……………………………………………………………..….                                   36
+blank         | 
+              | 
+meta          |                                         viii
+text          |         Methods……………………………………………………………………                                                  38
+              |         Results……………………………………………………………………..                                                41
+              |         Discussion…………………………………………………………………                                                45
+              |         Acknowledgments…………………………………………………………..                                            49
+blank         | 
+text          | 3     Terrestrial mammalian herbivores influence the distribution of defense and
+              |       nutrient traits in a Neotropical forest…………………………………………..                            51
+              |         Abstract………………………..……………………………………………                                               51
+              |         Introduction………………………………………………………………….                                             52
+              |         Methods……………………..………………………………………………                                                54
+              |         Results……………………..………………………………………………..                                              56
+              |         Discussion……………..……………………………………………………                                              61
+              |         Acknowledgments…………………………………………………………..                                            65
+blank         | 
+text          | 4     Hunting does not alter seed predation rates as a function of seed size in a
+              |       Neotropical forest………………………………………………………………                                           67
+              |         Abstract…………………………………………………………………..…                                               67
+              |         Introduction………………………………………………………………...                                            68
+              |         Methods…………………………………………………………………..…                                                72
+              |         Results……………………………………………………………………….                                                76
+              |         Discussion………………………………………………………………..…                                              81
+              |         Conclusion…………………………………………………………………...                                             86
+              |         Acknowledgments…………………………………………………………..                                            86
+blank         | 
+text          | Bibliography……………………………………………………………………….                                                   87
+blank         | 
+              | 
+              | 
+              | 
+title         | List of Tables
+text          | 3.1    The number of species and stems for which traits were measured…………..           57
+              | 3.2    R2 values for pair-wise correlations between species mean trait values ……...   57
+              | 3.3    Effect of species and exclosure on variation in leaf toughness………..            61
+              | 4.1    Mean fresh seed masses of study species……………………………...                          75
+blank         | 
+              | 
+              | 
+              | 
+meta          |                                                ix
+title         | List of Figures
+text          | 1.1  Primary seed dispersal in defaunated sites relative to non-defaunated
+              |      sites……………………………………………………………………                                           9
+              | 1.2 Differences in seed caching in defaunated sites relative to non-
+              |      defaunated sites………………………………………………………                                     10
+              | 1.3 Changes in seedling distribution as a consequence of defaunation……         13
+              | 1.4 Differences in vertebrate seed predation rates as a consequence of
+              |      defaunation…………………………………………………………….                                       16
+              | 1.5 Differences in invertebrate seed predation rates as a consequence of
+              |      defaunation…………………………………………………………….                                       17
+              | 1.6 Plant recruitment responses to defaunation for seedlings and saplings..    19
+              | 1.7 Seedling survival in defaunated sites relative to non-defaunated sites..   20
+              | 1.8 Differences in seedling densities in defaunated sites relative to non-
+              |      defaunated sites………………………………………………………..                                   22
+              | 1.9 Community-level herb densities in defaunation forest comparisons
+              |      and mammal exclosure experiments…………………………………..                           24
+              | 1.10 Differences in species richness, dominance, and diversity for
+              |      defaunation forest comparisons and mammal exclosure experiments...        26
+              | 2.1 Vertebrate activity in open and exclosure plots………………………..                 41
+              | 2.2 Proportion of seedlings by dispersal mode class in open and exclosure      42
+              |      treatments……………………………………………………………..
+              | 2.3 Proportion of seedlings by life form class in open and exclosure           42
+              |      treatments……………………………………………………………..
+              | 2.4 Median seed mass is significantly higher in exclosures………………               43
+              | 2.5 Effects of exclosure and hunting on community median wood density.         44
+              | 2.6 Associations between dispersal agents and species wood specific
+              |      gravity…………………………………………………………………                                          45
+              | 3.1 Changes in plot-level trait means, unweighted by species abundance...      58
+              | 3.2 Abundance-weighted, plot-level trait means in open and exclosure
+              |      plots……………………………………………………………………                                           59
+              | 3.3 Intraspecific differences in species’ mean leaf toughness…………….            60
+              | 4.1 Model of how seed predation should vary with seed mass as a
+              |      function of defaunation intensity……………………………………..                         71
+              | 4.2 Map of Lake Gatun study area………………………………………..                              73
+              | 4.3 Animal abundances in BCI and PNS in 2008…………………………                         77
+blank         | 
+meta          |                                           x
+text          | 4.4   Seed predation rates as a function of seed size………………………..            78
+              | 4.5   Number of palm seed cached in BCI and PNS………………………...                 79
+              | 4.6   Identity of species removing seeds……………………………………                      80
+              | 4.7   Published studies examining seed predation of large-seeded palms…..   85
+blank         | 
+              | 
+              | 
+              | 
+meta          |                                         xi
+              |                                                                                          1
+blank         | 
+              | 
+              | 
+              | 
+title         | Introduction
+text          | Tropical forests are among the most biodiverse ecosystems on the planet. Many
+              | mechanisms have been proposed to explain the maintenance of that diversity,
+              | including negative distance- or density- dependent mortality, niche partitioning, and
+              | neutral processes.   However, little attention has been paid to the role vertebrate
+              | consumers play in maintaining tropical diversity. Evidence from defaunated tropical
+              | forests suggests that these animals play a critical role in diversity maintenance
+              | (Chapter 1). This dissertation examines how terrestrial vertebrates, as seed dispersers,
+              | seed predators and herbivores, influence plant community composition in tropical
+              | forests and thereby levels of diversity. Specifically, I used plant functional traits as a
+              | lens through which to observe changes in seedling and sapling communities, identify
+              | which guilds of consumers were responsible for the changes, and elucidate trait-
+              | mediated mechanisms for the observed change. This work suggests that terrestrial
+              | vertebrates play an underappreciated role in maintaining plant diversity and that pan-
+              | tropical levels of unsustainable hunting may indirectly lead to losses of plant
+              | biodiversity.
+              |        I first examined how community-level functional trait composition shifts in
+              | seedling communities which have been protected from terrestrial mammals (Chapter
+              | 2). I conducted this work in the Barro Colorado National Monument in Central
+              | Panama, where a long term mammal exclosure experiment was established in 1993-
+              | 94, and where terrestrial mammals are relatively abundant. I found that seedling
+              | communities in exclosures did not differ in their dispersal mode or in the relative
+              | abundance of free standing and climbing growth forms, as may be expected in an
+              | experiment that did not manipulate primary dispersal agents. Seedling communities in
+              | exclosures had higher median seed mass than paired plots open to the mammal
+              | community, but treatments did not differ in their leaf traits (leaf mass per area and
+meta          | 2                                          INTRODUCTION
+blank         | 
+              | 
+text          | laminar toughness) or wood density. These results were validated with similar data
+              | from a defaunation gradient in the same region of Central Panama. One key contrast
+              | to the exclosure, however, was that seedling communities in defaunated sites had a
+              | higher representation of species with abiotic dispersal modes, lianas, and species with
+              | lower wood densities, which is consistent with the fact that primary dispersers are
+              | impacted by hunting.
+              |        I next examined the sapling community in the exclosure experiment to evaluate
+              | the effects of herbivores specifically, and to identify the relative contributions of
+              | altered species present, abundance, and trait expression to the differences in functional
+              | trait composition observed (Chapter 3). In contrast to the seedling community, the
+              | sapling community did show significant shifts toward higher leaf nitrogen and lower
+              | leaf toughness in response to herbivore exclosure, primarily due to an increased
+              | dominance of species with those traits, and secondarily due to differences in the
+              | species present in each treatment type.      Interestingly, I also found evidence that
+              | intraspecific differences in leaf traits were also contributing minorly to changes in
+              | community mean leaf toughness, though whether this is the result of differential
+              | mortality among genotypes or microhabitats, or a plastic response to decreased
+              | mammalian herbivory is unknown.
+              |        Finally, I investigated the seed size response to changes in mammal abundance
+              | by measuring vertebrate seed predation rates in a protected and hunted forest (Chapter
+              | 4). I aimed to both test a model of how seed predation rates should vary with seed size
+              | and defaunation intensity, and potentially clarify discrepancies in community level
+              | seed-size responses to hunting at different sites. I found that in central Panama, seed
+              | mass does not correlate well with either body size of the seed predator, or vertebrate
+              | seed predation rates. In fact, I found little difference in seed predation rates between
+              | the protected and hunted forests, despite large differences in key seed predators such
+              | as peccaries and agoutis. I suggest that rather than formulate seed predation rates as a
+              | linear function of seed predator abundance, these interactions may be better modeled
+              | as threshold-dependent processes.
+meta          |                                                                                        3
+blank         | 
+              | 
+              | 
+              | 
+title         | Chapter 1
+blank         | 
+title         | Contemporaneous defaunation and
+              | cascading effects on tropical forests
+              | Erin. L. Kurten, Mauro Galetti
+blank         | 
+              | 
+title         | INTRODUCTION
+blank         | 
+text          |          Large bodied vertebrates have been subject to human hunting for millennia. At
+              | the end of the Pleistocene, a diversity of large mammals became extinct worldwide,
+              | with human overhunting likely being one of the major drivers (Barnosky et al. 2004).
+              | The extinction of mammoth, giant sloths, giant kangaroos, giant deer, and many more
+              | megafauna in such a short time likely changed the structure and composition of their
+              | associated plant communities (Zimov et al. 1995, Guimares et al. 2008, Johnson
+              | 2009).
+              |          The extinction of large vertebrates is not a phenomenon restricted to the past,
+              | but rather continues in the present day. While scientists still debate what caused the
+              | Pleistocene megafaunal extinction (Alroy 2001; de Vivo & Carmignotto 2004; Koch
+              | & Barnosky 2006; Webb 2008), there is little doubt that human activities are
+              | resposible for threatening the persistence of approximately 22 percent of all mammal
+              | and 12 percent of all bird species in the world today (Pimm et al. 2006). Particularly
+              | in tropical forests, animal populations are currently in decline due to both
+meta          | 4                CHAPTER 1: CASCADING EFFECTS OF DEFAUNATION ON TROPICAL PLANTS
+blank         | 
+              | 
+              | 
+text          | unsustainable hunting and habitat fragmentation throughout Asia (Corlett 2007),
+              | Africa (Fa & Brown 2009) and Latin America (Peres & Palacios 2007).
+              |           Whether changes in vegetation structure and composition have occurred as a
+              | result of the extinction of the megafauna or climate change in the past, the effects of
+              | comtemporaneous defaunation on vegetation is measurable. In many parts of the
+              | tropics, plants have lost their major seeds dispersers, seed predators, and herbivores,
+              | likely altering plant demography, spatial distribution, genetic diversity and selection
+              | on seed and plant defense traits within species, with cascading effects on community
+              | composition and diversity.
+blank         | 
+title         | SCOPE OF REVIEW
+blank         | 
+text          |     Because the animals most vulnerable to defaunation, as well as their less vulnerable
+              | competitors, interact with plants as seed dispersers, seed predators, seedling predators,
+              | and herbivores, contemporaneous defaunation is likely to disrupt plant-vertebrate
+              | interactions. Several papers have outlined in detail how species interactions and plant
+              | communities are likely to change as a consequence of these disruptions (Wright 2003,
+              | Dirzo et al. 2007, Muller-Landau 2007). Disruptions of seed dispersal are likely to
+              | have negative effects on plant recruitment by preventing individuals from escaping
+              | distance-dependent and density-dependent mortality (sensu Janzen 1970, Connell
+              | 1971). Reduced seed dispersal may also prevent light-demanding species, or species
+              | with other specific environmental requirements, from reaching sites favorable for
+              | recruitment (Muller-Landau 2007, Brodie et al. 2009). Changes in seed predation,
+              | seedling predation, and herbivory may have positive or negative effects on species
+              | recruitment by altering seed and seedling survival (Wright 2003, Dirzo et al. 2007,
+              | Muller-Landau 2007). Changes in seed predation and seedling predation may have
+              | further indirect benefits for invertebrate seed predators and herbivores.
+              |           Overall, the net effect of defaunation on plant diversity has been hypothesized
+              | to be negative (Wright 2003, Muller-Landau 2007) This is both because it is thought
+              | that species experiencing reduced seed dispersal will not persist if they cannot escape
+meta          |                                                                                               5
+blank         | 
+              | 
+              | 
+text          | distance-dependent or density-dependent mortality, and because seed predators and
+              | herbivores will not suppress populations of competitively dominant species.
+              |        Here we synthesize what is currently known about the indirect effects of
+              | defaunation on tropical plants, in the context of hunting, forest fragmentation, and
+              | animal exclosure. We divide our analysis into three sections, addressing effects on
+              | plant-animal interactions, population demography, and community diversity. Our first
+              | goal was to evaluate the extent to which the hypothesized changes mentioned above
+              | are supported by empirical evidence. Our second goal was to focus on plant species
+              | or community responses that show mixed responses to defaunation, and try to clarify
+              | why such variability may exist. Our third goal was to identify areas of study which
+              | have received little attention and warrant future attention.
+blank         | 
+              | 
+title         | METHODOLOGY
+blank         | 
+              | 
+text          | IDENTIFICATION   OF STUDIES     Here we summarize literature in the field as of 2009.
+              | We identified studies primarily by searching literature databases for publications on
+              | aspects of plant ecology with a “defaunation” or “hunting” component.                      We
+              | supplemented this collection with other studies cited in that literature, as well as
+              | unpublished theses of which we had knowledge.                  In total,   thirty-seven studies
+              | comprise the quantitative portion of this review (Appendix 1).
+blank         | 
+              | 
+text          | STANDARDIZATION       OF RESPONSE VARIABLES              FOR    COMPARISON      Together, the
+              | variation in study design, defaunation intensities compared, species studied, and
+              | response variables measured introduce too much heterogeneity to conduct a formal
+              | meta-analysis of responses (Hedges & Olkin 1985). Yet, we have attempted to present
+              | a review which is quantitative rather than qualitative. We have done so by calculating
+              | effect sizes for each response variable. The effect size estimator we use here is the
+              | percent difference in the response variable reported between defaunated and non-
+              | defaunated sites as follows:
+              |                                                ோವ ିோಿವ
+              |                                Effect size =             ∗ 100,
+              |                                                 ோಿವ
+meta          | 6              CHAPTER 1: CASCADING EFFECTS OF DEFAUNATION ON TROPICAL PLANTS
+blank         | 
+              | 
+              | 
+text          | where RD and RND are the magnitudes of the response variables in the defaunated and
+              | non-defaunated sites, respectively. In cases where the denominator, RND, was zero, we
+              | used the smallest possible non-zero value for RND (e.g. one seed, in the case of
+              | absolute seed dispersal). When the minimum possible response could not be inferred
+              | (e.g., a percent abundance), we conservatively estimated the effect size as 100 percent.
+              | For example, where size class data were reported, sapling to seedling ratios were
+              | calculated to allow for comparison of recruitment rates across studies.
+              |        When studies reported response variables from multiple defaunated or non-
+              | defaunated sites, the data were averaged to derive a single value for defaunated and
+              | non-defaunated states. In studies of defaunation gradients, data on the abundance or
+              | presence-absence of vertebrate species relevant to the study, as well as the authors’
+              | categorical characterizations of the relative levels of defaunation were used to define
+              | sites as “defaunated” or “non-defaunated” for purposes of comparison. In these cases,
+              | sites experiencing the lowest levels of defaunation were often grouped with “no
+              | defaunation” sites, while sites showing medium to high levels of defaunation
+              | comprised the “defaunated” sites.
+blank         | 
+              | 
+text          | QUANTIFICATION OF DEFAUNATION INTENSITY            We attempted to compare effects as a
+              | function of defaunation intensity. We restricted these analyses to Neotropical sites
+              | (70% of studies) because more extensive and comparative data on vertebrate
+              | communities was reported for this region. We selected twelve genera of mammalian
+              | frugivores, granivores, and browsers that differ in body mass and sensitivity to
+              | defaunation pressures: Tapir, Tayassu, Pecari, Odocoileus, Mazama, Ateles, Allouata,
+              | Cebus, Agouti, Dasyprocta, Sciurus and Proechimys. These genera are wide-ranging
+              | in the Neotropics, though the particular species may vary. Because some studies did
+              | not report animal abundances or densities, but rather species presence or absence, we
+              | calculated the degree of defaunation for each site as the percent of focal genera that
+              | appeared to be locally extirpated in that site, though present historically.
+meta          |                                                                                          7
+blank         | 
+              | 
+              | 
+text          | SEED SIZE DATA     In several cases, the perturbations in plant-animal interactions as a
+              | consequence of defaunation are hypothesized to vary as a function of seed size. We
+              | therefore attempted to rank species by seed size in order to present the data in a way
+              | that would address these hypotheses. In most cases, either seed mass or seed length
+              | was reported.     In the case of Celtis durandii, one Astrocaryum species and one
+              | unspecified Dipteryx species for which seed sizes were not reported, an approximate
+              | seed size was estimated by assigning that species the value of a congeneric species
+              | from another site. Species for which both length and mass were reported were used to
+              | approximate a relative size rank for the species for which only mass or length were
+              | reported.   Alternatively, we could have used published seed mass-seed length
+              | correlations to estimate missing size parameters and used one measure of seed size to
+              | assign size ranks. However, this would generally change the positions of only a few
+              | species with unusual seed mass-to-seed length proportions (e.g. Gustavia superba),
+              | and we felt that those species were better assigned a rank by considering the size
+              | parameters in the context of the plant-animal interaction and the natural history of that
+              | species.
+blank         | 
+title         | PLANT-ANIMAL INTERACTIONS
+blank         | 
+              | 
+title         | Seed dispersal
+text          | Seed dispersal is thought to promote plant recruitment in tropical forests by facilitating
+              | escape from natural enemies (Janzen 1970, Connell 1971) and by helping species with
+              | particular environmental requirements for germination and survival (e.g. light-
+              | demanding species) reach favorable microhabitats (Muller-Landau 2007).              Seed
+              | dispersal establishes the spatial distribution and diversity of species in the seed and
+              | seedling banks.
+              |        A single seed can be dispersed multiple times by different agents. Here we
+              | refer to primary seed dispersal as physical removal from the parent tree and deposition
+              | on the ground. Primary seed dispersal can be performed by biotic agents such as
+              | primates, bats, and birds, or by abiotic agents such as wind. Once a seed has dispersed
+              | from the parent tree, it may still disperse further. This we refer to here as secondary
+meta          | 8              CHAPTER 1: CASCADING EFFECTS OF DEFAUNATION ON TROPICAL PLANTS
+blank         | 
+              | 
+              | 
+text          | dispersal. We focus on secondary dispersal by biotic agents such as rodents, though in
+              | some circumstances, water and gravity may also facilitate dispersal on the ground.
+blank         | 
+              | 
+text          | PRIMARY    SEED DISPERSAL         Seed dispersal by biotic agents should decrease as
+              | defaunation intensity increases and the abundance of seed dispersers is reduced. This
+              | should affect large-seeded species to a disproportionate degree for two reasons. First,
+              | their predominate dispersal agents tend to be larger-bodied, and therefore more
+              | vulnerable to defaunation (Peres & van Roosmalen 2002, Holbrook & Loiselle 2009).
+              | There also appears to be less redundancy of seed dispersal agents for large-seeded
+              | species, relative to smaller-seeded species (Peres & van Roosmalen 2002, Nuñez-Iturri
+              | et al. 2008, Donatti et al. 2009). When seed dispersal has been directly measured,
+              | either as the quantity of seeds removed by primary dispersers such as birds or
+              | primates, or as the proportion of seed crop removed from a parent tree, seed dispersal
+              | is lower in defaunated forests in almost all cases examined (Fig. 1.1). The magnitude
+              | of the decrease appears to be moderately correlated with seed size.
+              |        It is important to keep in mind that tropical seed mass distributions span more
+              | than seven orders of magnitude (I.J. Wright et al. 2007). The species reported here
+              | represent only the very upper range of that distribution and were generally selected for
+              | study because they were most likely to show dispersal declines. Almost nothing is
+              | known about how hunting alters biotic dispersal across the broader range of seed sizes
+              | (Muller-Landau 2007).
+              |        Two studies examining how perturbations of bird communities affect the
+              | dispersal of smaller-seeded species (< 1 cm fruit diameter) have shown contrasting
+              | results. Dispersal of Bocageopsis multiflora in defaunated sites was actually more than
+              | 2-fold higher than in non-defaunated sites (Fig. 1.1). The cause of this increase is
+              | unknown, but has been suggested to result from an increase in the relative abundance
+              | of generalist, frugivorous birds in highly fragmented sites (Cramer et al. 2007). In the
+              | case of Celtis durandii, a decrease in avian forest specialists that was not compensated
+              | for by forests generalists appears to be responsible for in an overall decrease in seed
+              | dispersal (Kirika et al. 2008).
+meta          |                                                                                              9
+blank         | 
+              | 
+              | 
+              | 
+text          |               Genus & Source
+              | Large      (Community), Tonga (26)
+              |            Leptonychia, Tanzania (9)
+              |            Dipteryx, Costa Rica (19)
+              |          Duckeodendron, Brazil (10)
+              |            Carapa, Costa Rica (19)
+              |               Attalea, Panama (36)
+              |                 Gustavia, Panama (3)
+              |           Astrocaryum, Panama (36)
+              |         Antrocaryon, Cameroon (35)
+              |                 Virola, Panama (3)
+              |         Choerospondias, Thailand (7)
+              |                 Pouteria, Brazil (2)
+              |                Pourouma, Brazil (2)
+              |                  Virola, Ecuador (20)
+              |                  Euterpe, Brazil (14)
+              |              Bocageopsis, Brazil (10)
+              | Small            Celtis, Uganda (22)
+              |                                          -100     -50      0       50     100    150
+              |                                                  % Difference in Seeds Dispersed
+blank         | 
+              | 
+text          |         FIGURE 1.1 Seed dispersal is lower in defaunated sites relative to non-defaunated
+              |         sites in almost all cases. Species are ranked by seed size, with Leptonychia being
+              |         largest (11.15 cm long) and Bocageopsis and Celtis being smallest (< 1 cm long).
+              |         Numbers in parentheses denote the study number in Appendix 1.
+blank         | 
+text          | SEED CACHING A special case of seed dispersal is the caching of seeds in the ground
+              | by scatter-hoarding rodents. While individual cached seeds may be predated at a later
+              | time, seeds that are not subsequently recovered are afforded protection from
+              | invertebrate seed predators.       For some species, such as large seeded palms with
+              | specialist bruchid beetle seed predators, seed survival is heavily dependent upon seed
+              | caching by rodents.
+meta          | 10               CHAPTER 1: CASCADING EFFECTS OF DEFAUNATION ON TROPICAL PLANTS
+blank         | 
+              | 
+              | 
+text          |         In most cases, seed caching was higher in defaunated sites relative to non-
+              | defaunated sites (Fig. 1.2). This is likely due to the fact that the rodents primarily
+              | responsible for seed caching in these forests, agoutis (Dasyprocta spp.) and squirrels
+              | (Sciurus spp.) may actually benefit from a reduction in the abundance of predators and
+              | larger-bodied competitors at low to medium levels of defaunation (Wright 2003, Dirzo
+              | et al. 2007, Peres & Palacios 2007).             Therefore, while defaunation may have a
+              | negative effect on primary dispersal, this negative effect often does not extend to
+              | secondary seed dispersal by terrestrial rodents.
+blank         | 
+              | 
+text          |                Genus & Source
+              | Large     Carapa, Costa Rica (18)
+              |             Attalea, Panama (24)
+              |        Astrocaryum, Panama (24)
+              |         Lechthis, Costa Rica (18)
+              |     Pentaclethra, Costa Rica (18)
+              |         Astrocaryum, Brazil (17)
+              |         Astrocaryum, Brazil (13)
+              |       Minquartia, Costa Rica (21)
+              |         Hymenaea, Venezuela (4)
+              |           Welfia, Costa Rica (18)
+              |            Otoba, Costa Rica (18)
+              |           Virola, Costa Rica (18)
+              |          Clarisia, Costa Rica (11)
+              | Small     Virola, Costa Rica (11)
+blank         | 
+text          |                                         -200 -100 0      100 200 300 400                         500
+              |                                               % Difference in Seeds Cached
+blank         | 
+              | 
+text          |         FIGURE 1.2 Differences in seed caching rates in defaunated sites relative to non-
+              |         defaunated sites. Species are ranked by size, with Carapa being the largest (20 g) and
+              |         Virola the smallest (2 g).    Absence of bar indicates no difference. Numbers in
+              |         parentheses denote the study number in Appendix 1.
+meta          |                                                                                       11
+blank         | 
+              | 
+              | 
+text          |         Defaunation gradients in Venezuela and Brazil were exceptions to the trend of
+              | higher seed caching rates in defaunated forests (Asquith et al. 1999, Galetti et al.
+              | 2006). In both of these cases, the defaunated sites were small fragments in which
+              | agoutis were virtually absent. Consistent with the loss of this important terrestrial
+              | seed disperser, seed caching in defaunated sites was lower than in non-defaunated sites
+              | in these studies (Fig. 1.2). Squirrels do not compensate for the loss of agoutis (Donatti
+              | et al. 2009).
+              |         In contrast to primary dispersal, no relationship between seed size and
+              | differences in caching rates were apparent between defaunated and non-defaunated
+              | sites. This may again be due to the fact that agoutis are capable of handling large
+              | seeds and fruits much larger than their relatively small gape size might suggest.
+              | Therefore, in the Neotropics secondary dispersal of larger-seeded species may occur
+              | despite some degree of defaunation, provided that agoutis persist in the site.
+              |         Smaller-bodied scatter-hoarding rodents, such as squirrels, exist in Africa and
+              | Asia, however to our knowledge, there is no functional equivalent to the agouti in
+              | Paleotropical forests.    The larger Paleotropical rodents that do exist, such as
+              | Cricetomys species, are larder-hoarding rodents that tend to take seeds to burrows
+              | where they will not survive (Guedje et al. 2003).        Therefore, the seed dispersal
+              | “buffer” that agoutis provide in the Neotropics is less likely to be present in other
+              | systems.
+blank         | 
+              | 
+text          | SPATIAL    DISTRIBUTION OF SEEDS AND SEEDLINGS        The spatial distribution of seeds
+              | and seedlings on the forest floor can also be an indicator of changes in seed dispersal.
+              | In particular, if seed dispersal is lower in defaunated forests, one would expect to find
+              | a greater proportion of seeds or seedlings undispersed under parent trees relative to
+              | what is observed in non-defaunated forests. Likewise, one would expect to see fewer
+              | seeds or seedlings at distances away from parent trees relative to non-defaunated
+              | forests. Indeed for many species, seed or seedling numbers under parent trees were 2-
+              | to 12-fold higher in defaunated sites relative to non-defaunated sites, and were lower
+meta          | 12             CHAPTER 1: CASCADING EFFECTS OF DEFAUNATION ON TROPICAL PLANTS
+blank         | 
+              | 
+              | 
+text          | away from parent trees (Fig. 1.3). These patterns are consistent with the loss of biotic
+              | dispersal agents.
+              |        One notable exception is Hymenaea on land-bridge islands in Venezuela.
+              | Hymenaea shows the highest increases in seed pods remaining under parent trees in
+              | sites without agoutis relative to sites with agoutis. However, it is one of the few cases
+              | in which seedling densities are actually lower under parent trees in defaunated forests
+              | (Fig.1.3).   This is not due to distance or density dependent mortality effects as
+              | mediated by specialist invertebrates or pathogens (Janzen 1970, Connell 1971).
+              | Rather, Hymenaea relies on agoutis to open its seed pods so that seeds may germinate.
+              | In sites lacking agoutis, the seeds cannot escape the seed pod, and seedling densities
+              | are correspondingly low (Asquith et al. 1999).
+              |        One study estimated changes in dispersal distances with defaunation. Cramer
+              | et al. (2007) reported a 60 percent and 80 percent decrease in mean and maximum
+              | dispersal distances respectively for the large-seeded Duckeodendron cestroides in
+              | defaunated fragments, relative to continuous forest. In the same study, the smaller-
+              | seeded Bocageopsis multiflora mean and maximum dispersal distances were not
+              | significantly different. This suggests that large-seeded species may become more
+              | spatially aggregated in defaunated forests.      Those plant species requiring special
+              | microhabitats for germination and recruitment (e.g. light gaps) may have greater
+              | difficultly reaching suitable sites. Meanwhile, smaller seeded species may be less
+              | affected.
+meta          |                                                                                                13
+blank         | 
+              | 
+text          |                   Genus & Source
+              | Large Choerospondias, Thailand (7)
+              |          Leptonychia, Tanzania (9)                                              Under Parent
+              |              Balanites, Gabon (5)
+              |              Attalea, Panama (37)
+              |              Attalea, Panama (37)
+              |              Attalea, Panama (36)
+              |             Dysoxylum, India (31)
+              |            Chisocheton, India (31)
+              |         Astrocaryum, Panama (36)
+              |          Hymenaea, Venezuela (5)
+              |              Polyalthia, India (31)
+              |          Hymenaea, Venezuela (4)
+              |                Syagrus, Brazil (1)
+              | Small Antrocaryon, Cameroon (35)
+              |                                         -250 0       250 500 750 1000 1250 1500 2350 2600
+              | Large           Attalea, Panama (37)
+              |                 Attalea, Panama (37)                                             Away From
+              |                 Attalea, Panama (36)                                               Parent
+              |            Astrocaryum, Panama (36)
+              |           Duckeodendron, Brazil (10)
+              |                Dysoxylum, India (31)
+              |               Chisocheton, India (31)
+              |                 Polyalthia, India (31)
+              | Small        Bocageopsis, Brazil (10)
+blank         | 
+text          |       Seeds                              -100 -50 0       50 100 150 250 300 350
+              |       Seedlings                            % Difference in Seed or Seedling Abundance
+blank         | 
+text          |         FIGURE 1.3 The number or proportion of seeds or seedlings is generally higher
+              |         immediate vicinity of parent trees and generally lower away from parent trees, in
+              |         defaunated sites relative to non-defaunated sites. Species are ordered by seed size.
+              |         Numbers in parentheses denote the study number in Appendix 1.
+blank         | 
+              | 
+              | 
+text          | COMPENSATION       AMONG DISPERSAL AGENTS             An important question regarding seed
+              | dispersal in tropical forests is whether a decrease in dispersal by the principle dispersal
+              | agent may be compensated for by other dispersal agents. Comparisons of primate
+              | diets suggest that smaller species of primates only disperse a nested subset of the plant
+              | species that are dispersed by larger primates. Even when co-occurring dispersers
+meta          | 14              CHAPTER 1: CASCADING EFFECTS OF DEFAUNATION ON TROPICAL PLANTS
+blank         | 
+              | 
+              | 
+text          | appear to have significant dietary overlap on the basis of species consumed, a
+              | quantitative diet analysis reveals that dispersers differ in the fruit species that make up
+              | the major portion of their diets (Poulsen et al. 2002).         This suggests that even
+              | dispersers with diet overlap may not compensate for one another. Similarly, among
+              | frugivores that feed on Virola, alternative dispersers do not shift their diet preferences
+              | enough to compensate for the loss of the principle dispersers (Holbrook & Loiselle
+              | 2009).
+              |          Decreases in primary seed dispersal summarized here (Fig. 1.1) suggest that
+              | this lack of compensation among primary dispersal agents may be fairly general, at
+              | least for larger-seeded species. There are no studies that directly investigate the
+              | degree to which an increase in secondary dispersal by terrestrial vertebrates (e.g. Fig
+              | 1.2) may compensate for reduced dispersal by birds and primates. However, in cases
+              | where species have lost their primary dispersal agent, the number of seeds and
+              | seedlings remaining under adult trees in defaunated sites is often higher than in non-
+              | defaunated sites (Fig. 1.3). These data provide indirect evidence that, at least in the
+              | cases studied, terrestrial seed dispersers do not fully compensate for loss of principle
+              | avian and primate dispersers. It is possible, however, that terrestrial dispersers are
+              | partially compensating for the loss of other dispersers, and that the observed changes
+              | in seed and seedling densities under and away from parents would be even more
+              | extreme in the absence of terrestrial dispersal.
+blank         | 
+title         | Seed Predation
+blank         | 
+text          | PRE-DISPERSAL    SEED PREDATION      Beckman and Muller-Landau (2007) investigated
+              | effects of defaunation on pre-dispersal seed predation for two species of contrasting
+              | seed size in central Panama. They found that pre-dispersal seed predation by
+              | mammals was significantly lower in the defaunated sites for the larger-seeded
+              | Oneocarpus mapora. Authors did not observe mammalian pre-dispersal seed
+              | predation of small-seeded Cordia bicolor. Pre-dispersal seed predation by insects
+              | between sites was not significantly different for either species.
+meta          |                                                                                     15
+blank         | 
+              | 
+              | 
+text          | POST-DISPERSAL       SEED PREDATION   After primary dispersal, seeds may be predated
+              | upon by terrestrial mammals such as rodents, peccaries, pigs and deer. The majority of
+              | studies comparing seed predation rates between defaunated and non-defaunated sites
+              | have performed manipulative experiments, setting out arrays of seeds and monitoring
+              | the fates of those seeds. However a few studies of species with slowly decomposing
+              | endocarps have looked at “standing” rates of seed predation in the field (Wright et al.
+              | 2000, Wright & Duber 2001, Galetti et al. 2006). In 22 of the 31 cases, seed predation
+              | rates were lower in defaunated sites, while in nine cases, seed predation rates were
+              | higher (Fig. 1.4).
+              |        In a conceptual model, larger-seeded species were hypothesized to experience
+              | higher seed predation rates in moderately defaunated forests relative to non-
+              | defaunated forests (Dirzo et al. 2007). At severe intensities of defaunation, however,
+              | large-seeded species were hypothesized to experience lower seed predation rates,
+              | relative to sites with moderate to no defaunation. These changes in seed predation
+              | rates were posited to be in response to changes in the abundance of seed predators of
+              | medium body size.
+              |        Using our estimates of defaunation intensity for Neotropical sites, we
+              | examined how differences in seed predation rates for different plant species varied
+              | with defaunation intensity.     These empirical data do not generally support the
+              | hypothesis that seed predation of larger seeds in moderately defaunated sites is higher
+              | than non-defauanted sites, while being lower in highly defaunated sites. (Fig. 1.4).
+              | Among the largest-seeded species studied, seed predation increased and decreased in
+              | about the same number of cases across all levels of defaunation examined.
+              |        The conceptual model also hypothesized that seed predation rates of smaller
+              | seeded species would increase monotonically with increasing defaunation intensity, as
+              | small rodents experienced release from competition and predation.              However
+              | empirical data suggest that seed predation rates are generally lower for smaller-seeded
+              | species in defaunated sites, regardless of defaunation intensity (Fig. 1.4).
+meta          | 16              CHAPTER 1: CASCADING EFFECTS OF DEFAUNATION ON TROPICAL PLANTS
+blank         | 
+              | 
+              | 
+text          |                                                    250
+blank         | 
+              | 
+              | 
+              | 
+text          |                   % Difference in Seed Predation
+              |                                                    200
+blank         | 
+text          |                                                    150
+blank         | 
+text          |                                                    100
+blank         | 
+text          |                                                     50
+blank         | 
+text          |                                                      0
+blank         | 
+text          |                                                     -50
+blank         | 
+text          |                                                    -100
+              |                                                           0.0      20.0     40.0   60.0   80.0   100.0   120.0
+blank         | 
+text          |                                                                 Defaunation Intensity in Defaunated Site
+blank         | 
+              | 
+text          |        FIGURE 1.4 Differences in seed predation rates as a function of the intensity of
+              |        defaunation in the defaunated sites. Circle sizes indicate the relative seed size of the
+              |        plant species.
+blank         | 
+text          | INVERTEBRATE SEED PREDATION                                               It has been suggested that a reduction in vertebrate
+              | seed predation may benefit invertebrate seed predators in two ways.                                                      First,
+              | invertebrates experience a release from competition for seed resources (Muller-Landau
+              | 2007). Secondly, larvae developing in seeds may experience decreased mortality due
+              | to the reduction of seed predation by vertebrates (Silvius 2002).                                                Through these
+              | mechanisms, defaunation may lead to higher abundances of invertebrates, and thereby
+              | higher rates of invertebrate seed predation.
+              |        Few studies have investigated how defaunation indirectly alters invertebrate
+              | seed predation rates. Consistent with expectation, bruchid beetle seed predation of
+              | palms was higher in defaunated sites for every species investigated (Fig. 1.5).
+              | However, in most cases, the absolute increase in invertebrate seed predation rates only
+              | partially compensated for the decreases in vertebrate seed predation (Wright et al.
+              | 2000, Wright & Duber 2001, but see Galetti et al. 2006).
+meta          |                                                                                                17
+blank         | 
+              | 
+              | 
+text          |                   Genus & Source
+              |                  Attalea, Panama (37)
+              |                  Attalea, Panama (36)
+              |              Astrocaryum, Brazil (17)
+              |             Astrocaryum, Panama (36)
+              |                    Syagrus, Brazil (1)
+              |                    Syagrus, Brazil (1)
+              |                                                 0        500       1000      1500
+              |                                             % Difference in Invertebrate
+              |                                                   Seed Predation
+              |        FIGURE 1.5 Seed predation rates by invertebrates for four species of palms are
+              |        higher in defaunated sites relative to non-defaunated sites. Species are ranked by
+              |        size, with Attalea being the largest and Syagrus the smallest. Numbers in parentheses
+              |        denote the study number in Appendix 1.
+blank         | 
+              | 
+title         | Herbivory & Trampling
+blank         | 
+              | 
+text          | Few observational studies have evaluated differences in herbivory or trampling as a
+              | consequence of defaunation.         Dirzo & Miranda (1991) an absence of vertebrate
+              | herbivory in Los Tuxtlas, a defaunated site. Alves-Costa (2004) categorized herbivory
+              | of the palm Syagrus romanzoffiana into classes by percent damage and also found that
+              | the frequency individuals experiencing little to no herbivory was 66% higher in
+              | defaunated sites.
+              |        Mortality by trampling has not been evaluated for live seedlings in the context
+              | of defaunation, however trampling has been evaluated with seedling models.                     In
+              | Bolivia, “trampling” of seedling models by vertebrates was 75 percent lower in
+              | defaunated forests (Roldán and Simonetti 2001).
+blank         | 
+              | 
+title         | POPULATION DEMOGRAPHY
+blank         | 
+              | 
+text          | RECRUITMENT         Reduced seed dispersal in defaunated forests has been hypothesized
+              | to have negative impacts on plant recruitment by preventing individuals from escaping
+              | distance-dependent     and density dependent             mortality (Muller-Landau 2007).
+              | Empirically, plant recruitment has been reported either in terms of new seedlings
+meta          | 18             CHAPTER 1: CASCADING EFFECTS OF DEFAUNATION ON TROPICAL PLANTS
+blank         | 
+              | 
+              | 
+text          | entering the population in a determined length of time, or as the ratio of juvenile plants
+              | to seedlings. Authors could define seedlings variously as first year germinants or by
+              | size class. In about half of cases, recruitment was about 2-fold higher in defaunated
+              | forests than in non-defaunated forests, both for seedlings and saplings (Fig. 1.6). In
+              | the other half of cases, recruitment is lower in defaunated forests.
+              |        There have been virtually no investigations into why recruitment rates change.
+              | The variation in changes in recruitment do not seem to be consistently related to the
+              | degree of defaunation across studies, loss of particular dispersers or loss of particular
+              | seed predators. Changes in vertebrate seedling predation and herbivory, as well as
+              | distance and density dependent mortality as mediated by invertebrate herbivores and
+              | pathogens (Janzen 1970, Connell 1971), can be responsible for altered seedling
+              | recruitment. However, no studies have recorded herbivory rates or causes of plant
+              | mortality, so no definitive data exists on how each of these mechanisms contribute to
+              | the changes in recruitment observed. One study of Syagrus found that sapling
+              | recruitment near the parent tree was slightly lower in defaunated forests, while
+              | recruitment far from the parent tree was more than 2-fold higher in defaunated forests
+              | (Alves-Costa 2004). This may be because non-vertebrate natural enemies are more
+              | than compensating for the loss of vertebrate herbivores near the parent trees, but not at
+              | far distances, for this species. However, no data exist with which to evaluate this
+              | explanation.
+              |        It is also possible, in some cases, that differences in recruitment are not
+              | actually due to defaunation.     In some studies of seedling recruitment, only one
+              | defaunated site and one non-defaunated site were compared.              This opens the
+              | possibility that some differences in recruitment were due to other environmental
+              | factors that vary across the landscape, such as precipitation or edaphic conditions. The
+              | driver of defaunation is another potential source of variation. In some cases, hunting
+              | was the cause of defaunation, altering primarily the animal community. However in
+              | other studies fragmentation or a combination of fragmentation and hunting was the
+              | cause of defaunation. As fragmentation alters many aspects of the microclimate, it is
+              | possible that changes in seedling recruitment are due to the effects of fragmentation on
+meta          |                                                                                            19
+blank         | 
+              | 
+              | 
+text          | microclimate, rather than or in addition to the effects of a reduction in herbivore
+              | abundance.
+blank         | 
+              | 
+text          |                 Genus & Source
+              |                Attalea, Panama (36)                                     Seedlings
+              |          Dipteryx, Peru/Panama (34)
+              |            Astrocaryum, Brazil (17)
+              |           Astrocaryum, Panama (36)
+              |                  Syagrus, Brazil (1)
+              |                                                                          Saplings
+              |          (Community), Malaysia (31)
+              |            Leptonychia, Tanzania (9)
+              |                 Balanites, Gabon (5)
+              |             Astrocaryum, Brazil (17)
+              |            Syagrus (near), Brazil (1)
+              |              Syagrus (far), Brazil (1)
+              |                                        -100    0 100 200 300 400 500
+              |                                               % Difference in Recruitment
+blank         | 
+              | 
+text          |        FIGURE 1.6 Plant recruitment does not respond consistently to defaunation across
+              |        species for either seedlings or saplings. Numbers in parentheses denote the study
+              |        number in Appendix 1.
+blank         | 
+              | 
+text          | SEEDLING SURVIVAL As with recruitment, responses of seedling survival to
+              | defaunation are quite variable (Fig. 1.7). Asquith et al. (1997) found that differences
+              | in seedling survival could vary among species within the same study system. Survival
+              | was lowest on highly defaunated islands in Lake Gatun, Panama for two of the three
+              | species tested. Exclosure experiments verified that spiny rats, free of competition
+              | from larger mammals, reach population sizes 10-fold greater on these islands relative
+              | to the non-that sites (Adler 1996), and are responsible for the increased mortality. The
+              | third species, which did not show a difference in survival between sites, Virola
+              | surinamensis, experienced 100% mortality at both defaunated and non-defaunated
+              | sites (Asquith et al. 1997).
+              |        Differences in survival rates can also be variable within a species in the same
+              | region, likely due to differences in the defaunation intensities being compared. Two
+meta          | 20               CHAPTER 1: CASCADING EFFECTS OF DEFAUNATION ON TROPICAL PLANTS
+blank         | 
+              | 
+              | 
+text          | studies have compared the survival of Gustavia superba in defaunated sites in the
+              | Lake Gatun region of central Panama. In one case, the defaunated site was hunted
+              | relative to the non-defaunated site, but the animal community was not so impoverished
+              | as to lead to a competitive release of small rodents. In that study, G. superba had
+              | higher survival rates in the defaunated sites. However, when the defaunated sites were
+              | islands with only high populations of spiny rats present, survival of G. superba was
+              | lower in the defaunated site. This difference in the degree of defaunation intensity
+              | evaluated in the two studies, and compensatory changes in abundance of non-hunted
+              | species, are likely to account for the contrasting responses in seedling survival, and
+              | illustrate that changes in seedling survival as a consequence of defaunation may not
+              | change monotonically.
+blank         | 
+              | 
+text          |              Genus & Source
+              |             Dipteryx, Panama (3)
+              |             Gustavia, Panama (3)
+              |       Dipteryx, Peru/Panama (33)
+              |                Syagrus, Brazil (1)
+              |                Virola, Panama (3)
+              |            Gustavia, Panama (32)
+              |      Pentaclethra, Costa Rica (18)
+              |                                     -75    -50     -25    0     25     50     75    100 125
+              |                                                  % Difference in Seedling Survival
+blank         | 
+              | 
+text          |          FIGURE 1.7 Seedling survival does not show consistent changes as a consequence of
+              |          defaunation, even within sites (e.g. study 4) or species (e.g. Gustavia superba in
+              |          Panama). Numbers in parentheses denote the study number in Appendix 1.
+blank         | 
+              | 
+text          | STANDING ABUNDANCE Seedling densities do not respond uniformly to defaunation.
+              | Species that are dispersed and predated upon by scatterhoarding rodents tend to
+              | increase in abundance in defaunated sites (Fig. 1.8). Likewise the increase in seedling
+              | abundance of the elephant dispersed Balanites wilsoniana, despite reduced seed
+              | dispersal, was attributed to low seed and seedling predation in the defaunated site
+meta          |                                                                                          21
+blank         | 
+              | 
+              | 
+text          | (Babweteera et al. 2007). Meanwhile, species that are bird dispersed have lower
+              | abundances in non-defaunated sites.
+              |        These data suggest, for species which are rodent dispersed and predated, any
+              | increases in distance-dependent or density-dependent mortality by invertebrates or
+              | pathogens, as a consequence of reduced dispersal, do not compensate for reductions in
+              | vertebrate seed and seedling predation, resulting in a net increase in survival for these
+              | species. The same does not appear to be true for bird dispersed species, although there
+              | is no theoretical reason why these species should not experience high reductions in
+              | vertebrate seed predation.
+              |        All studies involving rodent dispersed species were conducted on palm species,
+              | most of which are preferred food species for agoutis, a dominant Neotropical seed
+              | predator. More studies in Neotropical forests are required to determine if these
+              | species are unique in experiencing a net benefit from defaunation, or if other species
+              | preyed upon by agoutis also increase in abundance in defaunated sites. These studies
+              | should include species that have birds or primates as their primary dispersal agents.
+              |        In contrast to the rodent-dispersed palms, little is known about the seed
+              | predators of the bird-dispersed species studied. However, at least two of the three
+              | studies of bird-dispersed species were conducted in Paleotropical forests with different
+              | assemblages of seed predators. Studies of vertebrate seed and seedling predation in
+              | the Paleotropics may determine whether differences observed between rodent-
+              | dispersed species and bird-dispersed species may be due to differences in the seed and
+              | seedling predator community, rather than differences between species with different
+              | dispersal agents per se.
+meta          | 22             CHAPTER 1: CASCADING EFFECTS OF DEFAUNATION ON TROPICAL PLANTS
+blank         | 
+              | 
+text          |             Genus & Source
+              |        Astrocaryum, Brazil (17)                                 Rodent Dispersed
+              |             Dipteryx, Peru (33)
+              |              Syagrus, Brazil (1)
+              |       Astrocaryum, Bolivia (29)
+              |              Syagrus, Brazil (1)
+              |            Attalea, Panama (36)
+              |            Attalea, Panama (37)
+              |       Astrocaryum, Panama (36)
+blank         | 
+text          |                                                                   Bird Dispersed
+              |          Chisocheton, India (31)
+              |            Polyalthia, India (31)
+              |           Dysoxylum, India (31)
+              |             Euterpe, Brazil (14)
+              |        Leptonychia, Tanzania (9)
+              |                                 -100          0         100        200     300
+              |                                                               Elephant Dispersed
+              |             Balanites, Gabon (5)
+              |                                  -400       0        400        800      1200
+              |                                          % Difference in Seedling Densities
+              |        FIGURE 1.8 Differences in seedling densities in defaunated sites relative to non-
+              |        defaunated sites. Numbers in parentheses denote the study number in Appendix 1.
+blank         | 
+              | 
+text          | LINKING   DISPERSAL AND SEEDLING RECRUITMENT              It is clear, by several measures,
+              | that seed dispersal decreases as a consequence of defaunation (Figs. 1.1 & 1.3), though
+              | effects on seed caching are variable (Fig. 1.2). What is less clear is whether decreased
+              | dispersal translates to decreases in plant recruitment.           Investigators focusing on
+              | decreased seed dispersal as a consequence of defaunation often suggest that decreased
+              | dispersal will result in decreases in seed and seedling survival and plant recruitment.
+              | This is because undispersed seeds are assumed to be unable to escape distance-
+              | dependent and/or density-dependent mortality (Janzen 1970, Connell 1971).
+              |        However, there are several problems with this assumption. First, there are
+              | many species for which distance- and/or density-dependent mortality have not been
+              | detected (Hyatt et al. 2003, Uriarte et al. 2005). Secondly, spatial data on seedling
+              | densities also show that reduced seed dispersal does not generally lead to great
+              | decreases in seedling abundances under parent trees in defaunated forests (Fig. 1.3).
+              | This suggests that though seed and seedling mortality may be higher under parent
+meta          |                                                                                     23
+blank         | 
+              | 
+              | 
+text          | trees, the higher number of seeds that fall below the parent tree in defaunated forests
+              | compensate for this higher mortality and seedling densities under parent trees remain
+              | relatively similar to those observed in non-defaunated forests. Finally, as seen above,
+              | vertebrate seed and seedling predation rates may also vary with defaunation (Figs. 1.4
+              | & 1.5). Seedling density data suggest that reduced rates of vertebrate seed predation
+              | result in increased survival rates that more than compensate for any increases in
+              | distance-dependent or density-dependent mortality by non-vertebrates, at least for
+              | rodent-dispersed species and species like B. wilsoniana (Fig. 1.6). Taken together,
+              | these data suggest that the assumed negative effects of reduced seed dispersal on plant
+              | recruitment as a consequence of hunting may not be general, but rather manifest
+              | themselves only in species which (1) experience strong natural enemy-mediated,
+              | negative density-dependent mortality (2) rely exclusively upon a few, large-bodied,
+              | hunted vertebrates for seed dispersal, and (3) do not experience a demographic benefit
+              | of reduced seed predation by rodents with increasing defaunation intensity.
+              |        Interestingly, there are few studies that actually investigate changes in
+              | dispersal and seedling recruitment simultaneously in a single species. In the case of
+              | the large seeded palm Attalea butyracea in Central Panama, seed dispersal is reduced
+              | in defaunated forests (Wright & Duber 2001). In addition, the intensity of bruchid
+              | beetle seed predation does increase with proximity to the parent tree, and the effect
+              | intensifies with defaunation. However, increased bruchid beetle seed predation did
+              | not fully compensate for the decrease in vertebrate seed predation associated with the
+              | defaunation, and seedling densities of the palm were higher in defaunated sites,
+              | despite the decrease in seed dispersal.
+              |        Brodie et al. (2009) modeled the effects of decreased dispersal on seedling
+              | recruitment for Choerospondias axillaris in Thailand and did find evidence that
+              | decreased dispersal would translate to decreased recruitment in this species. However
+              | this was because defaunation deprived C. axillaris of favorable regeneration sites,
+              | since C. axillaris seems to depend on light gaps for regeneration and could not
+              | disperse to gaps without its vertebrate disperser (Brodie et al. 2009). It was not a
+              | consequence of distance-dependent or density-dependent mortality.
+meta          | 24               CHAPTER 1: CASCADING EFFECTS OF DEFAUNATION ON TROPICAL PLANTS
+blank         | 
+              | 
+              | 
+title         | COMMUNITY DIVERSITY
+blank         | 
+              | 
+text          | SEEDLING    DENSITY      Seedling or herb density at the community level is consistently
+              | higher where exclosure experiments (Ickes et al. 2001, Royo & Carson 2005) or
+              | defaunation due to hunting and fragmentation (Dirzo & Miranda 1991) completely
+              | eliminate vertebrate herbivory (Fig. 1.9).               Interestingly, at many moderately
+              | defaunated sites, community seedling densities are lower, relative to non-defaunated
+              | sites.   It may be that terrestrial herbivores initially increase in abundance as a
+              | consequence of a release from predation, and perhaps competition.                        However,
+              | decreased seedling densities at the community level are in contrast to higher seedling
+              | densities of many individual species in defaunated sites (Fig. 1.8). This discrepancy
+              | further highlights the need to understand how defaunation alters plant recruitment
+              | across a broader range of species, and to reconcile changes in recruitment at the
+              | species level with changes at the community-level.
+              |                  Site & Source
+              |                     Uganda (8)                                Forest comparison
+              |                     Uganda (8)                                Exclosure study
+              |                    Panama (38)
+              |                    Bolivia (29)
+              |                    Panama (30)
+              |                       Peru (27)
+              |                   Malaysia (21)
+              |                    Panama (30)
+              |                   Malaysia (21)
+              |                    Mexico (12)
+              |                               -100      -50      0      50     100     150
+              |                                       % Difference in Seedling Density
+blank         | 
+              | 
+text          |          FIGURE 1.9 Community level seedling and herb densities generally increase in
+              |          cases of high defaunation but decrease in others. Numbers in parentheses denote the
+              |          study number in Appendix 1.
+blank         | 
+              | 
+text          | DIVERSITY Several short-term exclosure studies have shown that species richness
+              | initially increases with herbivore removal (Fig. 1.10a). However, this appears to be an
+              | effect of increased stem densities increasing the likelihood that rare species will be
+              | present inside the exclosures. In an exclosure experiment in Panama, rare herbs
+meta          |                                                                                       25
+blank         | 
+              | 
+              | 
+text          | increased 2-fold (Royo & Carson 2005) (Fig. 1.10b). However, defaunation also
+              | appears to increase species dominance as well (Fig. 1.10b). Diversity indices appear
+              | to mask the changes in community composition that occur in exclosure experiments,
+              | because the effects of increased species richness and increased dominance tend to
+              | cancel each other out, such that exclosures show only small decreases in indices of
+              | diversity.
+              |         In forest comparisons, species richness was generally lower. Unlike short-term
+              | exclosure experiments, defaunated forests have experienced decades of defaunation,
+              | likely allowing competitively dominant plant species to outcompete rarer and less
+              | dominant species, reducing diversity.       Consistent with this hypothesis, species
+              | dominance was higher in the cases that reported it (Fig. 1.10b). Responses in diversity
+              | indices were mixed (Fig. 1.10c), again suggesting that these indices may be less
+              | informative than directly evaluating changes in species richness and dominance.
+blank         | 
+              | 
+text          | PLANT   FUNCTIONAL GROUPS        Nuñez-Iturri et al. (2007, 2008) and Wright et al.
+              | (2007) both found little difference in total species richness between defaunated and
+              | non-defaunated sites. However, the lack of differences in species richness belied
+              | important changes in plant community composition. Species with abiotic or unhunted,
+              | biotic dispersal agents consistently show higher abundance in hunted sites (Wright et
+              | al. 2007, Nuñez-Iturri et al. 2008, Terborgh et al. 2008). In contrast, species dispersed
+              | by game animals show decreased abundance in defaunated sites. Consistent with this
+              | finding, lianas, which tend to be abiotically dispersed (Hardesty & Muller-Landau
+              | 2005, Wright et al. 2007), were higher in relative abundance in defaunated sites in
+              | central Panama (Wright et al. 2007).
+meta          | 26                  CHAPTER 1: CASCADING EFFECTS OF DEFAUNATION ON TROPICAL PLANTS
+blank         | 
+              | 
+text          |        Site & Source                              A    Site & Source                             B
+              |           Mexico (12)                                    Panama (30)
+              |       (game) Peru (28)                                      Brazil (1)
+              |       (game) Peru (27)                                   Panama (30)
+              |             Uganda (8)                                   Panama (30)
+              |               Peru (28)                                   Mexico (12)
+              |             Uganda (8)                                                   100 200 500 600 700
+              |                                                                          0
+              |            Bolivia (29)                                          % Difference Species Dominance
+              |            Bolivia (29)
+              |                                                         Site & Source                            C
+              |            Panama (38)
+              |                                                          Mexico (12)
+              |               Peru (27)
+              |                                                         Malaysia (21)
+              |          Malaysia (21)
+              |                                                         Colombia (25)
+              |            Panama (30)
+              |                                                          Panama (38)
+              |      (abiotic) Peru (28)
+              |                                                                      -80    -40       0       40
+              |  (non-game) Peru (27)                                                  % Difference in Diversity
+              |        Colombia (25)
+              |      (abiotic) Peru (27)
+blank         | 
+text          |                       -75 -50 -25 0 25 50 75                                       Forest comparison
+              |                       % Difference Species Richness                                Exclosure study
+blank         | 
+text          |           FIGURE 1.10 Diversity responses to defaunation. (a) species richness, (b) species
+              |           dominance, and (c) species diversity (Shannon-Weiner, Simpson, or Fisher’s α
+              |           indices). Peruvian richness differences are generally for groups with different dispersal
+              |           agents; dominance for the Panama study is similarly for subsets of common and rare
+              |           herbs. Numbers in parentheses denote the study number in Appendix 1.
+blank         | 
+              | 
+text          |           The studies found important differences in community seed mass responses.
+              | In Peru, the relative abundance of large seeded species was lower in defaunated sites,
+              | while in Panama, the community mean seed mass was higher in defaunated sites
+              | (Wright et al. 2007, Nuñez-Iturri et al. 2008). The discrepancy appears to be due to
+              | differences in faunal communities between Peru and Panama. In the Peruvian study,
+              | defaunation greatly reduces populations of large primate dispersers, while not greatly
+              | affecting the relatively low populations of seed predators such as agoutis. In contrast,
+              | defaunation in Panama, defaunation greatly reduces agouti populations, but does not
+              | greatly affect seed dispersal by large primates, as these animals are rare or absent even
+              | in the non-defaunated sites.
+              |           Studies of the effects of defaunation effects on other functional traits are rare.
+              | Both a forest comparison study in Panama and an exclosure study in Panama found
+meta          |                                                                                       27
+blank         | 
+              | 
+              | 
+text          | that community median leaf mass per area and leaf laminar toughness in the seedling
+              | community did not change in response to defaunation (Ch. 2). However, saplings (40-
+              | 100 cm) in the exclosure experiment did have lower community mean leaf toughness,
+              | and higher community mean leaf nitrogen relative to controls (Ch. 3). Wood density
+              | was the only functional trait that had different responses to defaunation in the forest
+              | comparison and exclosure studies. This is likely due to the fact that primary dispersal
+              | is altered in the forest comparison, but not in the exclosure study (Ch. 2).
+blank         | 
+              | 
+title         | WHAT IS DEFAUNATION?
+text          | The heterogeneity in response to defaunation, particularly at the level of plant
+              | population dynamics and community composition, suggest that more precision is
+              | needed in characterizing exactly what is meant by defaunation, including its drivers
+              | and the effects on the particular mammal community being studied. Among the
+              | studies reported there, there was variation in the levels of defaunation in both “non-
+              | defaunated” and “defaunated” forests across studies. In some cases, sites with similar
+              | species compositions would be considered “non-defaunated” in one study, and
+              | “defaunated” in another.
+              |        Unfortunately, terms such as “hunted”, “fragmented,” or “disturbed” are
+              | usually used to classify sites into treatments, but do not correspond to a specific
+              | change in a vertebrate community. This problem could be addressed by reporting a
+              | robust characterization of the vertebrate community. Often, studies did not actually
+              | report differences in either abundances of key species (e.g. a particular seed disperser)
+              | or the vertebrate community as a whole.      In other cases, the presence or absence of
+              | species were reported, but not animal abundances. Reporting presence/absence data
+              | will not capture situations in which animals such as rodents or smaller primates
+              | actually increase in population size as defaunation reduces the predation and
+              | competition they experience. In other cases, a species may be present, but in such rare
+              | numbers so as to be functionally extinct.
+              |        Though estimating animal abundances is labor-intensive, we suggest that
+              | further clarity and progress in understanding the effects of defaunation on plant
+meta          | 28             CHAPTER 1: CASCADING EFFECTS OF DEFAUNATION ON TROPICAL PLANTS
+blank         | 
+              | 
+              | 
+text          | communities will rely on exactly this level of precision in characterizing sites used in
+              | comparative studies.
+blank         | 
+              | 
+title         | HETEROGENEITY AMONG STUDIES
+blank         | 
+              | 
+text          | Understanding the heterogeneity and sources of potential bias in the studies examined
+              | here is important both for understanding the limitations and generality of the data, as
+              | well as identifying areas for further research. Research on effects of defaunation on
+              | plants has predominantly focused on Neotropical mammals.           Over 70 percent of
+              | studies were conducted in the Neotropics, while only 15 percent and 10 percent of
+              | studies were conducted in Africa and Asia respectively, with the remaining 5 percent
+              | conducted in Australia and Polynesia. As this review has demonstrated, responses to
+              | defaunation can be quite variable even within the Neotropics, where faunal
+              | communities are relatively similar. More studies in the Paleotropics are needed to
+              | understand how plant responses to defaunation in these forests may differ from those
+              | observed in Neotropical forests.
+              |        Only 15 percent of studies reported effects of avian defaunation, with the
+              | remainder examining effects of mammalian defaunation.          Many of those studies
+              | focused on rodents and/or primates.       Despite the fact that many of the larger
+              | vertebrates such as peccaries, pigs, deer, tapir and elephants are most vulnerable to
+              | defaunation, only a few studies investigated the effects of losing these animals from
+              | the community.     Future studies should seek to understand the impacts of losing
+              | relatively understudied groups of vertebrates, as well as understanding how losses of
+              | dispersers, seed predators, and herbivores may synergize.
+              |        It is also important to note that forest comparisons generally had low
+              | replication. The majority of studies only had one non-defaunated site, and sometimes
+              | only one defaunated site. We suggest that a better understanding of patterns and
+              | mechanisms could come from increasing the number of study sites examined, and
+              | including a range of defaunation intensities within a single study, with better
+              | quantification of the differences in vertebrate communities. We suspect that many of
+meta          |                                                                                        29
+blank         | 
+              | 
+              | 
+text          | the interactions and patterns discussed in this review do not vary linearly with animal
+              | abundance, but rather, change in a step-wise or threshold dependent manner as key
+              | interaction partners are removed from a system. Studying gradients of defaunation
+              | intensity would allow scientists to better elucidate the levels of defaunation at which
+              | changes in key interactions occur, and perhaps allow for the reconciliation of
+              | discrepancies in responses among studies.
+blank         | 
+              | 
+title         | CONCLUSIONS
+blank         | 
+              | 
+text          | It is now well documented that larger-seeded species experience reduced dispersal,
+              | and increased aggregation around parent trees as a result of defaunation. However the
+              | overall consequences for species recruitment appear variable. More work is required
+              | to link reduced dispersal to subsequent stages of the seedling recruitment process to
+              | determine whether reduced dispersal poses a general threat to the persistence of large-
+              | seeded species. Responses to defaunation at the community level are also variable,
+              | and future studies should work to reconcile the changes observed that the species
+              | interaction or population levels with changes in composition at the community level.
+meta          | 30                 CHAPTER 1: CASCADING EFFECTS OF DEFAUNATION ON TROPICAL PLANTS
+blank         | 
+              | 
+              | 
+text          | APPENDIX 1.1 Reference information and raw data for response variables for studies
+              | referenced in figures. Ndef = value for non-defaunated site(s); Def = Value for
+              | defaunated site(s); % ∆ = Percent change from non-defaunated site to defaunated site.
+blank         | 
+              | 
+text          | No   Reference                        Site        Response Variable                     Ndef   Def     %∆
+              | 1    Alves-Costa (2004)               Brazil      Diaspores predated (%)                31.2   12.6    -60
+              | 1    Alves-Costa (2004)               Brazil      Insect seed predation (near)          10.7   23.4    119
+              | 1    Alves-Costa (2004)               Brazil      Insect seed predation (far)           1.6    24.2    1376
+              | 1    Alves-Costa (2004)               Brazil      Seed density                          0.16   0.18     8
+              | 1    Alves-Costa (2004)               Brazil      Seedlings under parent                0.06   0.72    1187
+              | 1    Alves-Costa (2004)               Brazil      Juvenile:Seedling (near)              2.33   1.16    -50
+              | 1    Alves-Costa (2004)               Brazil      Juvenile:Seedling (far)               2.65   9.51    259
+              | 1    Alves-Costa (2004)               Brazil      Seedling density                      26.8   46.8     75
+              | 1    Alves-Costa (2004)               Brazil      No. species/No. plants                0.23   0.14    -37
+              | 1    Alves-Costa (2004)               Brazil      Frequency of low herbivory            24.1   40.0     66
+              | 1    Alves-Costa (2004)               Brazil      No. new plants                        41.2   118.4   187
+              | 1    Alves-Costa (2004)               Brazil      Seedling mortality (%)                40.1   42.7     6
+              | 2    Andresen (2003)                  Brazil      % Pouteria seeds buried                40    32.5    -19
+              | 2    Andresen (2003)                  Brazil      % Pourouma seeds buried                36     19     -47
+              | 3    Asquith et al. (1997)            Panama      Seed removal                          0.01   0.05    900
+              | 3    Asquith et al. (1997)            Panama      Seedling survival                     0.21   0.07    -69
+              | 3    Asquith et al. (1997)            Panama      Seed removal                          0.01   0.08    700
+              | 3    Asquith et al. (1997)            Panama      Seed dispersal                        0.56   0.24    -58
+              | 3    Asquith et al. (1997)            Panama      Seedling survival                     0.49   0.22    -56
+              | 3    Asquith et al. (1997)            Panama      Seed removal                          0.00   0.00     0
+              | 3    Asquith et al. (1997)            Panama      Seed dispersal                        0.54   0.19    -65
+              | 3    Asquith et al. (1997)            Panama      Seedling survival                     0.00   0.00     0
+              | 4    Asquith et al. (1999)            Venezuela   Pod removal                           1.00   0.14    -86
+              | 4    Asquith et al. (1999)            Venezuela   Seed removal                          0.92   0.60    -35
+              | 4    Asquith et al. (1999)            Venezuela   Seed caching                          0.17   0.00    -100
+              | 4    Asquith et al. (1999)            Venezuela   Seed density under parent             4.0    100.0   2400
+              | 4    Asquith et al. (1999)            Venezuela   Seedling density under parent         22.0    1.8    -92
+              | 5    Babweteera et al. (2007)         Gabon       Juvenile density                       85    1169    1275
+              | 5    Babweteera et al. (2007)         Gabon       Sapling:Seedling                      0.21   0.00    -98
+              | 5    Babweteera et al. (2007)         Gabon       Pole:Seedling                         0.08   0.01    -92
+              | 5    Babweteera et al. (2007)         Gabon       Prob. Juvenile's under conspecific    58.0   89.0     53
+              | 6    Beckman & Muller-Landau (2007)   Panama      Primary seed removal                  42.0   24.1    -43
+              | 6    Beckman & Muller-Landau (2007)   Panama      Predispersal seed predation           55.0   25.0    -55
+              | 6    Beckman & Muller-Landau (2007)   Panama      Secondary seed removal                0.0    10.6    100
+              | 6    Beckman & Muller-Landau (2007)   Panama      Primary seed removal                  21.6   14.6    -32
+              | 6    Beckman & Muller-Landau (2007)   Panama      Predispersal seed predation           7.0     7.0     0
+              | 6    Beckman & Muller-Landau (2007)   Panama      Predispersal seed predation           20.0    0.0    -100
+              | 6    Beckman & Muller-Landau (2007)   Panama      Secondary seed removal                80.0   10.8    -87
+              | 7    Brodie et al. (2009)             Thailand    Proportion fruits under adults        0.19   0.68    252
+              | 7    Brodie et al. (2009)             Thailand    Seeds dispersed                       0.15   0.01    -91
+              | 8    Chapman & Onderdonk (1998)       Uganda      Seedling density, non-valley sites    12.5    5.5    -56
+              | 8    Chapman & Onderdonk (1998)       Uganda      Seedling richness, non-valley sites   5.0     3.1    -39
+              | 8    Chapman & Onderdonk (1998)       Uganda      Seedling density, valley bottom       2.5     1.3    -50
+meta          |                                                                                                          31
+blank         | 
+              | 
+              | 
+text          | 8    Chapman & Onderdonk (1998)   Uganda       Seedling richness, valley bottom    1.3     1.0    -23
+              | 9    Cordeiro and Howe (2003)     Tanzania     Seed dispersal                     12.8     3.0    -76
+              |                                                Proportion indiv under adults
+              | 9    Cordeiro and Howe (2003)     Tanzania     (<10 m)                            57.3    88.5    54
+              | 9    Cordeiro and Howe (2003)     Tanzania     Juvenile:Seedling                   0.4     0.3    -27
+              | 9    Cordeiro and Howe (2003)     Tanzania     Seedling abundance                 163.7   96.7    -41
+              | 10   Cramer et al. (2007)         Brazil       Seeds dispersed                    48.0    16.0    -67
+              | 10   Cramer et al. (2007)         Brazil       Seeds dispersed                     5.0    12.0    140
+              | 10   Cramer et al. (2007)         Brazil       Ave. dispersal distance             7.5     3.0    -60
+              | 10   Cramer et al. (2007)         Brazil       Ave. dispersal distance            23.0    18.0    -22
+              | 10   Cramer et al. (2007)         Brazil       Max. dipersal distance             20.0     4.0    -80
+              | 10   Cramer et al. (2007)         Brazil       Max. dipersal distance             17.0    18.0     6
+              | 10   Cramer et al. (2007)         Brazil       Seeds away from parent             27.0     1.0    -96
+              | 10   Cramer et al. (2007)         Brazil       Seeds away from parent              8.0     6.0    -25
+              | 11   DeMattia et al. (2004)       Costa Rica   Seeds destroyed                    100.0   93.6     -6
+              | 11   DeMattia et al. (2004)       Costa Rica   Seeds destroyed                    68.8    65.6     -5
+              | 11   DeMattia et al. (2004)       Costa Rica   Seeds destroyed                    100.0   84.1    -16
+              | 11   DeMattia et al. (2004)       Costa Rica   Seeds destroyed                    31.2    15.6    -50
+              | 11   DeMattia et al. (2004)       Costa Rica   Seeds destroyed                    93.8    75.0    -20
+              | 11   DeMattia et al. (2004)       Costa Rica   Seeds destroyed                    28.0    12.2    -57
+              | 11   DeMattia et al. (2004)       Costa Rica   Seeds destroyed                    18.5    24.5    33
+              | 11   DeMattia et al. (2004)       Costa Rica   Seeds cached                       25.0     3.1    -88
+              | 11   DeMattia et al. (2004)       Costa Rica   Seeds cached                       3.07    9.22    200
+              | 12   Dirzo & Miranda (1991)       Mexico       Herbivory                          0.27    0.00    -100
+              | 12   Dirzo & Miranda (1991)       Mexico       Seedling density                   22.6    52.8    134
+              | 12   Dirzo & Miranda (1991)       Mexico       Seedling species richness          6.65    2.30    -65
+              | 12   Dirzo & Miranda (1991)       Mexico       Seedling diversity                 3.71    1.07    -71
+              | 12   Dirzo & Miranda (1991)       Mexico       Seedling species dominance         14.0     2.0    -86
+              | 13   Donatti et al. (2009)        Brazil       Seeds predated                     0.18    0.03    -83
+              | 13   Donatti et al. (2009)        Brazil       Seeds cached                       0.33    0.13    -61
+              | 14   Fadini et al. (2009)         Brazil       Seed dispersal                      1.6     0.9    -42
+              | 14   Fadini et al. (2009)         Brazil       Seed predation                     99.7     7.4    -93
+              | 14   Fadini et al. (2009)         Brazil       Seedling density                   1000     80     -92
+              | 15   Farwig et al. (2006)         Kenya        seed removal                       0.97    1.57    62
+              | 16   Fleury & Galetti (2006)      Brazil       Seed predation                     15.8    48.0    203
+              | 16   Fleury & Galetti (2006)      Brazil       Seed predation                     15.8     0.0    -100
+              | 17   Galetti et al. (2006)        Brazil       Seed predation (%)                  6.9     0.3    -96
+              | 17   Galetti et al. (2006)        Brazil       Seed caching (%)                   12.2     3.4    -72
+              | 17   Galetti et al. (2006)        Brazil       Seed predation (%)                 38.4     2.0    -95
+              | 17   Galetti et al. (2006)        Brazil       Seed predation (%)                 34.4    69.6    102
+              | 17   Galetti et al. (2006)        Brazil       Seedling density                   148.2   34.5    -77
+              | 17   Galetti et al. (2006)        Brazil       Seedling:Adult                     6.96    0.35    -95
+              | 17   Galetti et al. (2006)        Brazil       Sapling:Seedling                   1.98    10.07   409
+              | 18   Guariguata et al. (2000)     Costa Rica   Secondary seed removal             0.41    0.13    -67
+              | 18   Guariguata et al. (2000)     Costa Rica   Secondary seed removal             0.31    0.37    20
+              | 18   Guariguata et al. (2000)     Costa Rica   Secondary seed removal             0.88    0.94     7
+              | 18   Guariguata et al. (2000)     Costa Rica   Secondary seed removal             0.59    0.91    56
+              | 18   Guariguata et al. (2000)     Costa Rica   Secondary seed removal             0.66    0.88    33
+              | 18   Guariguata et al. (2000)     Costa Rica   Secondary seed removal             0.87    0.38    -56
+              | 18   Guariguata et al. (2000)     Costa Rica   Secondary seed removal             0.98    0.40    -59
+meta          | 32                  CHAPTER 1: CASCADING EFFECTS OF DEFAUNATION ON TROPICAL PLANTS
+blank         | 
+              | 
+              | 
+text          | 18   Guariguata et al. (2000)     Costa Rica   Secondary seed predation          0.41    0.13    -67
+              | 18   Guariguata et al. (2000)     Costa Rica   Secondary seed predation          0.31    0.37    20
+              | 18   Guariguata et al. (2000)     Costa Rica   Secondary seed predation          0.77    0.91    18
+              | 18   Guariguata et al. (2000)     Costa Rica   Secondary seed predation          0.55    0.91    64
+              | 18   Guariguata et al. (2000)     Costa Rica   Secondary seed predation          0.57    0.78    38
+              | 18   Guariguata et al. (2000)     Costa Rica   Secondary seed predation          0.73    0.38    -48
+              | 18   Guariguata et al. (2000)     Costa Rica   Secondary seed predation          0.95    0.40    -58
+              | 18   Guariguata et al. (2000)     Costa Rica   Seed caching                      0.00    0.00    0
+              | 18   Guariguata et al. (2000)     Costa Rica   Seed caching                      0.00    0.00    0
+              | 18   Guariguata et al. (2000)     Costa Rica   Seed caching                      0.03    0.11    300
+              | 18   Guariguata et al. (2000)     Costa Rica   Seed caching                      0.01    0.03    400
+              | 18   Guariguata et al. (2000)     Costa Rica   Seed caching                      0.10    0.09    -7
+              | 18   Guariguata et al. (2000)     Costa Rica   Seed caching                      0.00    0.13    100
+              | 18   Guariguata et al. (2000)     Costa Rica   Seed caching                      0.00    0.03    100
+              | 18   Guariguata et al. (2000)     Costa Rica   Tertiary seed removal/predation   0.50    0.81    63
+              | 18   Guariguata et al. (2000)     Costa Rica   Tertiary seed removal/predation   1.00    0.80    -20
+              | 18   Guariguata et al. (2000)     Costa Rica   Tertiary seed removal/predation   1.00    1.00    0
+              | 18   Guariguata et al. (2000)     Costa Rica   Seedling survival                 0.00    0.07    100
+              | 19   Guariguata et al. (2002)     Costa Rica   Seeds dispersed                   0.11    0.03    17
+              | 19   Guariguata et al. (2002)     Costa Rica   Seeds dispersed                   0.00    0.02    -67
+              | 20   Holbrook & Loiselle (2009)   Ecuador      Seeds removed (primary)           89.4    66.8    -25
+              | 21   Ickes, K., et al. 2001.      Malaysia     Seedlings                         75.5    117.5   56
+              | 21   Ickes, K., et al. 2001.      Malaysia     Sapling density                   107.1   142.3   33
+              | 21   Ickes, K., et al. 2001.      Malaysia     Mortality                         10.0     8.7    -13
+              | 21   Ickes, K., et al. 2001.      Malaysia     Recruits                          14.9    48.8    228
+              | 21   Ickes, K., et al. 2001.      Malaysia     Species richness                  50.6    55.7    10
+              | 21   Ickes, K., et al. 2001.      Malaysia     Diversity                         45.8    39.7    -13
+              | 21   Ickes, K., et al. 2001.      Malaysia     Liana proportion of stems         25.9    30.5    18
+              | 22   Kirika et al. (2008)         Uganda       Seed dispersal                    14124   3714    -74
+              | 23   Kurten (2010a)               Panama       seed mass                         0.21    0.32    58
+              | 23   Kurten (2010a)               Panama       SLA                               20.0    22.4    12
+              | 23   Kurten (2010a)               Panama       Leaf toughness                    1.49    1.49    0
+              | 23   Kurten (2010a)               Panama       Wood density                      0.65    0.64    -2
+              | 24   Kurten (2010b)               Panama       Secondary seed predation          0.50    0.81    63
+              | 24   Kurten (2010b)               Panama       Secondary seed predation          0.90    0.85    -5
+              | 24   Kurten (2010b)               Panama       Secondary seed predation          0.53    0.11    -79
+              | 24   Kurten (2010b)               Panama       Secondary seed predation          0.00    0.00    0
+              | 24   Kurten (2010b)               Panama       Seed caching                      1.16    1.33    15
+              | 24   Kurten (2010b)               Panama       Seed caching                      0.83    1.16    40
+              | 25   Lizcano (2006)               Colombia     Richness                          8.02    11.24   40
+              | 25   Lizcano (2006)               Colombia     Shannon Index                     0.82    0.75    -9
+              | 26   McConkey & Drake (2006)      Tonga        Seed dispersal                    0.34    0.03    -90
+              | 27   Nuñez-Iturri et al. (2007)   Peru         Sapling richness - total          47.4    50.4    6
+              | 27   Nuñez-Iturri et al. (2007)   Peru         Sapling richness -large mammal    16.6     7.8    -53
+              | 27   Nuñez-Iturri et al. (2007)   Peru         Sapling richness - nongame        27.1    36.6    35
+              | 27   Nuñez-Iturri et al. (2007)   Peru         Sapling richness - abiotic        3.70    6.00    62
+              | 27   Nuñez-Iturri et al. (2007)   Peru         Sapling density - total           78.3    90.0    15
+              | 27   Nuñez-Iturri et al. (2007)   Peru         Sapling density -large mammal     26.8    11.3    -58
+              | 27   Nuñez-Iturri et al. (2007)   Peru         Sapling density - nongame         45.5    67.8    49
+meta          |                                                                                                      33
+blank         | 
+              | 
+              | 
+text          | 27   Nuñez-Iturri et al. (2007)   Peru       Sapling density - abiotic         5.91   10.98    86
+              | 28   Nuñez-Iturri et al. (2008)   Peru       Species richness (primate disp)   1.60   0.60    -63
+              | 28   Nuñez-Iturri et al. (2008)   Peru       Species richness (abiotic disp)   1.40   1.80     29
+              | 28   Nuñez-Iturri et al. (2008)   Peru       Species richness                  145    100     -31
+              | 28   Nuñez-Iturri et al. (2008)   Peru       Density (primate disp)            1.90   0.90    -53
+              | 28   Nuñez-Iturri et al. (2008)   Peru       Density (abiotic)                 3.60   11.30   214
+              | 29   Roldan & Simonetti (2001)    Bolivia    Seed predation                    0.43   0.64     49
+              | 29   Roldan & Simonetti (2001)    Bolivia    Trampling survival                0.78   0.90     15
+              | 29   Roldan & Simonetti (2001)    Bolivia    Seedling density                  4.21   3.92     -7
+              | 29   Roldan & Simonetti (2001)    Bolivia    Species richness                  35.0   34.0     -3
+              | 29   Roldan & Simonetti (2001)    Bolivia    Seedling spp:Adult spp            0.61   0.59     -5
+              | 29   Roldan & Simonetti (2001)    Bolivia    Seedling composition              4.21   3.92     -7
+              | 30   Royo & Carson (2005)         Panama     Species richness                  16.8   13.1    -22
+              | 30   Royo & Carson (2005)         Panama     Herb Density                      1.21   1.19     -2
+              | 30   Royo & Carson (2005)         Panama     Density dominants                 0.42   0.35    -17
+              | 30   Royo & Carson (2005)         Panama     Density rare                      0.34   0.16    -53
+              | 30   Royo & Carson (2005)         Panama     Cover                             7.90   5.90    -25
+              | 30   Royo & Carson (2005)         Panama     Cover of dominants                2.65   1.50    -43
+              | 31   Sethi & Howe (2009)          India      Total seedlings                   2.55   0.76    -70
+              | 31   Sethi & Howe (2009)          India      Seedlings under parent            0.37   0.60     64
+              | 31   Sethi & Howe (2009)          India      Seedlings away from parent        2.19   0.16    -93
+              | 31   Sethi & Howe (2009)          India      Total seedlings                   2.17   0.81    -62
+              | 31   Sethi & Howe (2009)          India      Seedlings under parent            0.99   0.35    -65
+              | 31   Sethi & Howe (2009)          India      Seedlings away from parent        1.18   0.46    -61
+              | 31   Sethi & Howe (2009)          India      Total seedlings                   4.16   0.81    -80
+              | 31   Sethi & Howe (2009)          India      Seedlings under parent            2.96   0.58    -81
+              | 31   Sethi & Howe (2009)          India      Seedlings away from parent        1.20   0.24    -80
+              | 32   Sork (1987)                  Panama     Secondary seed predation          0.94   0.41    -56
+              | 32   Sork (1987)                  Panama     seedling survival                 0.69   0.97     42
+              |                                   Peru/
+              | 33   Terborgh & Wright (1994)     Panama     Seed predation                    0.35   0.82    136
+              |                                   Peru/
+              | 33   Terborgh & Wright (1994)     Panama     Seedling density                  38.3   29.3    -23
+              |                                   Peru/
+              | 33   Terborgh & Wright (1994)     Panama     Seedling recruitment              0.25   0.96    276
+              |                                   Peru/
+              | 33   Terborgh & Wright (1994)     Panama     Seedling mortality                0.20   0.29     43
+              | 34   Terborgh et al. (2008)       Peru       Recruit. Bird dispersed           0.23   0.24     4
+              | 34   Terborgh et al. (2008)       Peru       Recruit. Bat dispersed            0.10   0.11     16
+              | 34   Terborgh et al. (2008)       Peru       Recruit. Lg. primate dispersed    0.25   0.19    -24
+              |                                              Recruit. Sm. arboreal mammal
+              | 34   Terborgh et al. (2008)       Peru       dispersed                         0.33   0.24    -26
+              |                                              Recruit. Terrestrial mammal
+              | 34   Terborgh et al. (2008)       Peru       dispersed                         0.04   0.02    -50
+              | 34   Terborgh et al. (2008)       Peru       Recruit. Wind disp.               0.05   0.17    278
+              | 35   Wang et al. (2007)           Cameroon   Diaspore under adult              10.0   50.0    400
+              | 35   Wang et al. (2007)           Cameroon   Seed dispersal                    0.98   0.58    -41
+              | 36   Wright et al. (2000)         Panama     Seed predation                    90.8   47.7    -47
+              | 36   Wright et al. (2000)         Panama     Seed predation                    0.03   0.34    1033
+              | 36   Wright et al. (2000)         Panama     Seed density under parent         180    2749    1427
+              | 36   Wright et al. (2000)         Panama     Seed density away from parent     2.65   1.79    -32
+              | 36   Wright et al. (2000)         Panama     Prop. seeds dispersed             0.93   0.39    -58
+meta          | 34                  CHAPTER 1: CASCADING EFFECTS OF DEFAUNATION ON TROPICAL PLANTS
+blank         | 
+              | 
+              | 
+text          |  36   Wright et al. (2000)        Panama    Seedling density                      0.0    0.0    300
+              |  36   Wright et al. (2000)        Panama    Seedlings:Adult                      11.0    28.9   163
+              |  36   Wright et al. (2000)        Panama    Seed predation                       91.7    21.7   -76
+              |  36   Wright et al. (2000)        Panama    Seed predation                        0.1    0.3    460
+              |  36   Wright et al. (2000)        Panama    Seed density under parent            910     1290   42
+              |  36   Wright et al. (2000)        Panama    Seed density away from parent         4.4    2.8    -37
+              |  36   Wright et al. (2000)        Panama    Prop. seeds dispersed                 0.9    0.1    -93
+              |  36   Wright et al. (2000)        Panama    Seedling density                      0.1    0.2    129
+              |  36   Wright et al. (2000)        Panama    Seedlings:Adult                      113.8   92.5   -19
+              |  37   Wright et al. (2001)        Panama    Seed density under parent tree       24.6    35.7   46
+              | 37    Wright et al. (2001)        Panama    Seed density away from parent        0.9     0.2    -81
+              |                                             tree
+              | 37    Wright et al. (2001)        Panama    Seedling density under parent         0.0    0.1    282
+              |                                             Seedling density away from
+              |  37   Wright et al. (2001)        Panama    parent                                0.0    0.1    282
+              |  37   Wright et al. (2001)        Panama    Seedling density (total)              0.0    0.1    282
+              |  37   Wright et al. (2001)        Panama    Seed predation-rodents                0.8    0.4    -46
+              |  37   Wright et al. (2001)        Panama    Seed predation-bruchid                0.1    0.4    266
+              |  38   Wright et al. (2007)        Panama    Community mean seed mass              90     100    11
+              |  38   Wright et al. (2007)        Panama    Prop. seedlings (hunted disperser)    0.7    0.4    -40
+              | 38    Wright et al. (2007)        Panama    Prop. seedlings (unhunted             0.2    0.4    75
+              |                                             disperser)
+              | 38    Wright et al. (2007)        Panama    Proportion lianas                     0.1    0.4    150
+              | 38    Wright et al. (2007)        Panama    Seedling density                     15.2    9.4    -38
+              | 38    Wright et al. (2007)        Panama    Seedling Richness                    70.7    69.8   -1
+              | 38    Wright et al. (2007)        Panama    Seedling S-W Diversity                2.6    3.2    23
+meta          |                                                                                     35
+blank         | 
+              | 
+              | 
+              | 
+title         | Chapter 2
+blank         | 
+text          | Reduced seed dispersal as a
+              | consequence of hunting lowers
+              | community-level wood density in a
+              | Neotropical forest
+              | Authors: Erin L. Kurten, S. Joseph Wright, Andrés Hernandéz , Walter P. Carson
+blank         | 
+title         | Abstract
+blank         | 
+text          | Defaunation alters interactions between plants and their seed dispersers and predators,
+              | and is sometimes associated with decreased diversity at the community level. A more
+              | cryptic plant response to defaunation is an altered distribution of life histories and
+              | functional traits in the community. In this study, we censused 12,769 seedlings in a
+              | long-term terrestrial mammal exclosure experiment (LTEE) in Central Panama to
+              | examine how vertebrate seed dispersers and seed predators shape relative abundances
+              | of species classified by dispersal mode, growth forms, seed mass, leaf mass per area
+              | (LMA), leaf toughness, and wood density. We then compared exclosure results with
+              | data from a natural comparison of 38,250 seedlings in nearby hunted and unhunted
+              | sites. The exclosure experiment excludes terrestrial seed predators and herbivores but
+              | not seed dispersers such as primates, birds, and bats. In contrast, the hunting
+meta          | 36         CHAPTER 2: CONSEQUENCES OF HUNTING FOR SEED DISPERSAL AND WOOD DENSITY
+blank         | 
+              | 
+              | 
+text          | comparison reduces the abundance of all three groups. We tested the hypotheses that
+              | (1) where seed dispersal is disrupted, species with unhunted dispersal modes, and
+              | lianas, which are abiotically dispersed, are favored, (2) where seed predation is
+              | reduced, larger-seed plant species are favored, and (3) where herbivory is reduced,
+              | species with lower LMA, leaf toughness, and wood density are favored. Consistent
+              | with our expectations, dispersal-driven changes in the relative abundance of growth
+              | forms or dispersal mode classes were not observed in the terrestrial mammal exclosure
+              | experiment, while higher median seed mass as a consequence of reduced seed
+              | predation was observed. In contrast to our hypothesis, leaf functional traits did not
+              | differ as a consequence of defaunation either in the forest comparison or in the LTEE.
+              | Interestingly, wood density responses differed between our two experiments. Species
+              | dispersed by mammals and large birds had significantly higher mean wood density
+              | than wind dispersed species, suggesting that seed dispersal, rather than seedling
+              | herbivory, may mediate decreases in community median wood density in hunted
+              | forests.
+blank         | 
+title         | Introduction
+blank         | 
+text          | Over-hunting has caused severe declines, and even local extirpations, of hunted
+              | wildlife populations in tropical forests throughout Latin America (Peres & Palacios
+              | 2007), Africa (Fa & Brown 2009), and Asia (Corlett 2007). Defaunation perturbs
+              | plant interactions with the frugivores, granivores, and herbivores that play critical
+              | roles in establishing the composition and spatial distribution of seedlings in tropical
+              | forest seedling banks. Reduced seed dispersal in hunted forests results in higher
+              | proportions of undispersed seeds under parent trees of species reliant on hunted biotic
+              | dispersal agents (Wright & Duber 2001, Cordiero & Howe 2003, Babweteera et al
+              | 2007, Sethi & Howe 2009). Vertebrate seed predation rates may increase or decrease,
+              | depending on the severity of defaunation and the size of the seeds and seed predators
+              | affected (Guariguata et al. 2000, Galetti et al. 2006, Dirzo et al. 2007). Vertebrate
+              | herbivory rates can also be drastically reduced (Dirzo and Miranda 1991). The net
+              | effect of these perturbations can be the loss of plant diversity (Dirzo and Miranda
+meta          |                                                                                      37
+blank         | 
+              | 
+              | 
+text          | 1991, Chapman & Onderdonk 1998), with up to 66% of species failing to recruit in the
+              | most extreme case documented.
+              |        Other important changes may also be occurring in the plant community even if
+              | species richness is unaffected. Species with hunted dispersal agents, such as primates
+              | or large birds, consistently decrease in abundance in hunted forests, while species with
+              | abiotic or unhunted dispersal agents increase (Wright et al. 2007b, Nuñez-Iturri et al.
+              | 2008, Terborgh et al. 2008). Lianas, the majority of which are abiotically dispersed in
+              | this forest (Hardesty & Muller-Landau 2005), also increase in abundance in hunted
+              | sites (Wright et al. 2007b). Where dispersal of large, primate-dispersed seeds is
+              | disrupted by local extirpation of ateline monkeys, the relative abundance of those plant
+              | species is lower than in sites where ateline monkeys are present (Nuñez-Iturri et al.
+              | 2008). In sites where hunting decreases rodent seed predator populations, the median
+              | seed mass of the seedling community is higher than in unhunted sites (Wright et al.
+              | 2007b).
+              |        In this study, our first goal was to move beyond investigations of dispersal
+              | mode classes and seed mass to examine the effect of defaunation on other functional
+              | traits. Lower abundances of browsers and other seedling predators in hunted sites
+              | should lead to reduced vertebrate herbivory and seedling predation. Leaf mass per
+              | unit area (LMA), low leaf laminar toughness (hereafter referred to as “toughness”),
+              | and wood density are correlated with palatability to vertebrates or seedling survival
+              | (Wardle et al. 2002, Alvarez-Clare & Kitajima 2007). We hypothesized that species
+              | with low leaf area per unit mass, low leaf laminar toughness, and lower wood density
+              | should be favored in hunted sites.
+              |        A weakness of many forest comparison studies in defaunation work is that they
+              | lack replication of hunted and unhunted forests. Even when there is replication,
+              | differences in climate, soils, forest age, fragmentation or other anthropogenic factors
+              | may covary with hunting intensity in comparative studies, making it difficult to
+              | attribute the patterns observed to hunting alone. Therefore, a second goal of this study
+              | was to validate results from forest comparisons with data from a long-term exclosure
+              | experiment (LTEE). We conducted this study in Central Panama, where a hunting
+meta          | 38         CHAPTER 2: CONSEQUENCES OF HUNTING FOR SEED DISPERSAL AND WOOD DENSITY
+blank         | 
+              | 
+              | 
+text          | defaunation gradient has been previously documented (Wright et al. 2007b). Within
+              | the unhunted, protected sites, a terrestrial mammal exclosure experiment has been
+              | ongoing for the past fifteen years (Royo & Carson 2005).
+              |        Because hunting typically disrupts multiple plant-animal interactions
+              | simultaneously, it can be difficult to determine the relative influence of lost seed
+              | dispersers, seed predators, or herbivores on resulting seedling communities in
+              | comparative studies. A third goal of this study was to improve the extent to which
+              | changes in the seedling community could be attributed to changes in particular
+              | interactions. Unlike the hunted forest comparison, our LTEE excludes terrestrial seed
+              | predators and herbivores from exclosure plots, but does not alter the abundance of
+              | primary dispersal agents such as primates and most large birds. By comparing results
+              | between the forest comparison and exclosure approaches, we can identify which
+              | changes in the seedling community are due to changes in seed dispersal and which are
+              | due to changes in seed or seedling predation. .
+              |        We predicted that seedling communities in exclosures would have higher
+              | median seed massas a consequence of lower seed predation, consistent with the forest
+              | comparison (Wright et al. 2007). In contrast to the forest comparison experiment, we
+              | predicted the LTEE would not show differences in the relative abundances of dispersal
+              | mode classes or growth forms. We expected that median LMA, toughness and wood
+              | density would be lower in exclosure plots relative to paired plots open to terrestrial
+              | mammals, and in hunted forests relative to unhunted forests, as a consequence of
+              | reduced herbivory.
+blank         | 
+              | 
+title         | Methods
+blank         | 
+              | 
+text          | Study sites. The Barro Colorado National Monument (BCNM) is a group of islands
+              | and mainland peninsulas located in Lake Gatun in the Republic of Panama. Because
+              | of forest guard patrol and researcher presence, activity of poachers is virtually
+              | nonexistent on Barro Colorado Island (BCI) itself and is highly reduced on Gigante
+              | (GIG) and other peninsulas. The Parque Nacional Soberanía (PNS) is a protected area
+meta          |                                                                                    39
+blank         | 
+              | 
+              | 
+text          | contiguous with the BCNM, which is not actively patrolled. Intensity of poaching
+              | within the PNS varies with accessibility, but is much greater than within the BCNM
+              | (Wright et al. 2000).
+blank         | 
+              | 
+text          | Exclosure experimental design.      Between late-1993 and mid-1994, eight pairs of
+              | fenced, exclosure plots and open, control plots were established within the BCNM
+              | (Royo & Carson 2005). Paired fenced and unfenced plots are 30 m x 45 m and are
+              | approximately 5 m apart.       Exclosure fences are constructed of 12.7 x 12.7 cm
+              | galvanized steel fencing 2.2 m tall and buried 0.25 m deep. A secondary 1.3 x 1.3 cm
+              | mesh surrounds the lower 70 cm and also extends 0.25 m below ground. Twenty-
+              | eight 0.5 x 0.5 m subplots were established in each of the 16 plots in a stratified
+              | random fashion to census seedlings.        To avoid fence effects, no subplots were
+              | established within a 5-7 m buffer zone of a plot edge. Within each subplot, each plant
+              | less than 50 cm in height was tagged, measured, and identified. Seedling census data
+              | were collected in 2006.
+blank         | 
+              | 
+text          | Exclosure effectiveness. To assess differences in the animal communities between
+              | open and exclosure treatments, Reconyx RC-55 infrared cameras (Reconyx, Inc.,
+              | Holmen, WI) were used to monitor mammal activity in the 16 plots between August
+              | and October 2008. Each plot was monitored at 6-7 locations, for a total of 41 camera
+              | trap days per plot. A camera trigger was counted as a new “visit” if (1) it was a
+              | different species than the prior trigger, or (2) if 60 minutes had elapsed between
+              | triggers (Di Bitetti et al. 2008). We chose this method of quantification because
+              | unmarked individuals of most mammals in this community cannot be distinguished by
+              | photograph, and because this measure of activity likely describes animal impact on
+              | each plot more accurately than abundance. Because it was difficult to accurately
+              | count individuals for social animals such as peccaries (Tayassu tajacu) and coatis
+              | (Nasua narica), activity by these species was analyzed as the number of visits by
+              | social groups rather than visits by individuals.
+meta          | 40         CHAPTER 2: CONSEQUENCES OF HUNTING FOR SEED DISPERSAL AND WOOD DENSITY
+blank         | 
+              | 
+              | 
+text          | Hunting intensity comparison. At 9 protected sites (BCNM) and 11 intensely hunted
+              | sites (PNS), all adult trees larger than 20 cm diameter at breast height (dbh) within a 1
+              | ha plot were mapped and identified by Condit et al. (2002) and Wright et al. (2007b).
+              | Between June and December 2004, all woody plants less than 50 cm tall in an 8 x 8 m
+              | subplot were censused by Wright et al. (2007b). Seedling plots were at the center of
+              | each tree plot, or as close as possible in three plots in which tree falls occurred
+              | between adult and seedling censuses. A total of 38250 individuals comprising 312
+              | identified species were recorded (Wright et al. 2007b).
+blank         | 
+              | 
+text          | Plant traits. Wright et al. (2010) determined life history traits, dry seed mass, leaf
+              | mass per area (LMA), laminar toughness, and wood density for species in central
+              | Panama. In this study, dry seed mass was measured as endosperm plus embryo mass
+              | after oven drying to constant mass at 60ºC. Leaf mass per unit area was measured on
+              | shade leaves as the mass of a standard-sized leaf disc after oven drying to constant
+              | mass at 60ºC divided by the area of the disc. Laminar toughness was measured as the
+              | force necessary to cut across a leaf, perpendicular to the midvein (Lucas et al. 2000).
+              | Wood specific gravity, or mass after drying at 100ºC per unit volume, was used as the
+              | unit of wood density. Data on dispersal agents have been assembled from published
+              | studies and long-term personal observations.
+blank         | 
+              | 
+text          | Analyses. We used a MANOVA to test whether mammal activity differed between
+              | open and control plots for each vertebrate species observed. Paired t-tests were used
+              | to test for differences in the relative abundance of growth forms, dispersal mode
+              | classes, and functional traits. For the hunting comparison, an ANCOVA was used to
+              | determine whether the correspondence of functional traits between seedling and adult
+              | communities differed with hunting. Statistical analyses were performed with SYSTAT
+              | © 11.0 (Richmond, CA, U.S.A.) and R version 2.7.2 (R Development Core Team,
+              | 2008).
+meta          |                                                                                                         41
+blank         | 
+              | 
+              | 
+title         | Results
+blank         | 
+              | 
+text          | EXCLOSURE     EFFECTIVENESS.                   Fenced plots excluded non-climbing, terrestrial
+              | granivores and herbivores from exclosures, including agoutis (Dasyprocta punctata),
+              | paca (Agouti paca), deer (Mazama Americana and Odocoileus virginianus), and
+              | peccaries (T. tajacu) (Fig. 1). Most climbing animals appeared to be undeterred by the
+              | exclosure treatment.           Spiny rats (Proechimys steerei) showed higher activity in
+              | exclosures than controls, perhaps responding to a lack of competition for food, greater
+              | cover, or a reduction in ocelet abundance (Felis pardalis) in the exclosures. Most
+              | birds were not affected by the exclosures, a notable exception being tinamous, which
+              | are seed predators (Erand et al. 1991).
+blank         | 
+              | 
+text          |                                    0
+              |                     Agouti                                                                  ***
+blank         | 
+text          |                                0           20            40            60             80          100
+blank         | 
+text          |                   Anteater
+              |                                    0
+              |                  Armadillo
+              |                                    0
+              |                   Capuchin
+              |                      Coati†
+              |                                    0
+              |                       Deer
+              |                                    0
+              |                     Mouse
+              |                                    0
+              |                      Ocelot                *
+              |                  Opposum
+              |                                    0
+              |                                                      *
+              |                        Paca                                                 **
+              |                                    0
+              |                   Peccary†                                                                  **
+              |                                    0
+              |                      Rabbit
+              |                   Spiny Rat
+              |                    Squirrel
+blank         | 
+text          |                                    0
+              |                   Tinamou
+              |                Birds (other)
+blank         | 
+text          |                                0       2         4            6       8          10        12     14
+              |                                                               Visits/plot
+blank         | 
+              | 
+              | 
+              | 
+text          |        FIGURE 2.1. Total visits per plot by species for control plots (dark grey) and exclosure
+              |        plots (light grey). Zeros denote no individuals observed in exclosures. *** p < 0.001,
+              |        ** p < 0.01, * p < 0.05, 
 p < 0.1; † groups, not individuals
+meta          | 42                             CHAPTER 2: CONSEQUENCES OF HUNTING FOR SEED DISPERSAL AND WOOD DENSITY
+blank         | 
+              | 
+              | 
+              | 
+text          | DISPERSAL                          AGENTS                           & LIFEFORMS. A total of 12,769 seedlings of 322 species were
+              | recorded in the 2006 LTEE seedling census. Of those, dispersal agents were known
+              | for 247 species (76.7%) and 12575 stems (98.5%), and growth form was known for
+              | 241 species (74.8%) and 12468 stems (97.6%). Consistent with our expectations,
+              | there was no difference in the proportions of seedlings with different dispersal agents
+              | (hunted: t = 1.161, P = 0.2651; unhunted t = 1.861, P = 0.089) (Fig. 2) or growth
+              | forms (free-standing: t = 1.7131, P = 0.1096) (Fig. 3) in the LTEE.
+meta          |                                     1.0
+blank         | 
+              | 
+              | 
+              | 
+text          |                                                                            Hunted           Mixed          Unhunted
+              |                                     0.8
+              |          Proportion of seedlings
+blank         | 
+text          |                                     0.6
+              |                                     0.4
+              |                                     0.2
+              |                                     0.0
+blank         | 
+              | 
+              | 
+              | 
+text          |                                                                        Open Excl          Open Excl        Open Excl
+blank         | 
+text          |                                                                                     Treatment by Disperser
+              |        FIGURE 2.2. Proportion of seedlings by dispersal mode class is not significantly
+              |        different between treatments in exclosure experiment.
+              |                                                                      1.0
+blank         | 
+              | 
+              | 
+              | 
+text          |                                                                                 Climbing        Free Standing
+              |                                           Proportion of seedlings
+blank         | 
+text          |                                                                      0.8
+              |                                                                      0.6
+              |                                                                      0.4
+              |                                                                      0.2
+              |                                                                      0.0
+blank         | 
+              | 
+              | 
+              | 
+text          |                                                                                 Open Excl           Open Excl
+blank         | 
+text          |                                                                                Treatment by Growth form
+blank         | 
+text          |        FIGURE 2.3. Proportion of seedlings by life form is not significantly different
+              |        between treatments in exclosure experiment.
+meta          |                                                                                                             43
+blank         | 
+              | 
+              | 
+              | 
+text          | SEED MASS. In the LTEE, seed dry mass was known for 166 (51.6 %) of species and
+              | 10866 (85.1%) of stems. Exclosures removed all but the smallest seed predators. The
+              | reduction in agouti activity was particularly dramatic (Fig. 4). Consistent with these
+              | findings, as well as results from the forest comparison (Wright et al. 2007b), median
+              | seed mass was 44% higher in exclosure plots, relative to controls (t = 2.315, P =
+              | 0.02261).
+blank         | 
+text          |                                              250
+              |                  Diaspore Dry mass (mg)
+blank         | 
+text          |                                              200
+              |                                              150
+              |                                              100
+blank         | 
+              | 
+              | 
+              | 
+text          |                                                            Open                 Excl
+blank         | 
+              | 
+text          |                                                                   Treatment
+blank         | 
+text          |        FIGURE 2.4.                        (A) Median seed mass is significantly higher in exclosures (p <
+              |        0.02261).
+blank         | 
+              | 
+text          | LEAF TRAITS.    LMA and laminar toughness were known for 11,173 stems (87.5%)
+              | and 10,904 stems (85.4%) respectively in the LTEE, and for 33,839 stems (87.5 %)
+              | and 29,962 stems (77.5 %) respectively in the hunting experiment. Exclosure and
+              | hunting treatments were expected to reduce the number of terrestrial vertebrate
+              | seedling predators in a way analogous to seed predators. A reduction in vertebrate
+              | herbivores was expected to favor species with low LMA and small laminar toughness.
+              | Contrary to expectation, neither the exclosure nor the hunting experiment showed
+              | significant differences in median LMA or laminar toughness, nor was there a
+              | significant difference between protected and hunted sites when the correspondence
+meta          | 44            CHAPTER 2: CONSEQUENCES OF HUNTING FOR SEED DISPERSAL AND WOOD DENSITY
+blank         | 
+              | 
+              | 
+text          | between seedling and adult communities was evaluated by ANCOVA (data not
+              | shown).
+blank         | 
+              | 
+text          | WOOD DENSITY. Adult wood specific gravity was known for 208 species (64.6% %)
+              | and 11,620 stems (91.0 %) in the LTEE, and for 239 species (63.2 %) and 28,488
+              | stems (73.3 %) in the hunting experiment. Stem tissue density is strongly positively
+              | correlated with seedling survival in this forest (Alvarez-Clare & Kitajima 2007). If
+              | greater stem tissue densities confer a protective effect against herbivores, we would
+              | expect higher wood density species to be favored in open plots relative to exclosures,
+              | and protected sites relative to hunted sites. Median wood densities were greatly
+              | reduced in hunted site seedling communities, relative to both protected site seedling
+              | communities (pooled variance t=3.86, P = 0.0011, one-tailed test) (Fig. 5a), and their
+              | surrounding adult communities (F1,18 = 16.2, p < 0.001) (Fig. 5b). However, the
+              | LTEE showed no differenced in median wood density (t = 0.024, P = 0.49) (Fig. 5a).
+              |                                                                                                                                  0.65
+              |                                                                                         Wood density of tree seedlings (g/cm3)
+              |                                               0.65
+              |               Wood Specific Gravity (g/cm3)
+blank         | 
+              | 
+              | 
+              | 
+text          |                                                      A                                                                                    B
+              |                                                          Exclosure           Hunting
+              |                                                                                                                                  0.60
+              |                                               0.60
+blank         | 
+              | 
+              | 
+              | 
+text          |                                                                                                                                  0.55
+              |                                               0.55
+blank         | 
+              | 
+              | 
+              | 
+text          |                                                                                                                                  0.50
+blank         | 
+              | 
+text          |                                                                                                                                  0.45
+              |                                               0.50
+blank         | 
+              | 
+              | 
+              | 
+text          |                                                                                                                                         0.45   0.50   0.55   0.60   0.65
+              |                                                          Open   Excl   Protect   Hunt                                             Wood density of canopy trees (g/cm3)
+              |                                                                  Treatment
+blank         | 
+text          |      FIGURE 2.5. (A) Median wood density is not significantly different between treatments in the
+              |      exclosure experiment (p = 0.0238), but is significantly lower in hunted sites relative to
+              |      protected sites (p = 0.0011). White and grey boxes indicate analogous treatments (B) Seedling
+              |      median wood density is lower than adult median wood density for tree species in hunted sites
+              |      (dark circles) but not in protected sites (light triangles) (F1,18 = 16.2, p < 0.001). The dashed
+              |      line represents is the 1:1 line at which adult and seedling communities have the same median
+              |      values.
+blank         | 
+              | 
+text          |          We tested whether differences in primary dispersal may be responsible for the
+              | discrepancy in wood density responses, as it was for differences in dispersal mode and
+              | growth form. We used an ANOVA to compare mean wood densities across dispersal
+meta          |                                                                                                                                                                           45
+blank         | 
+              | 
+              | 
+text          | mode classes. Interestingly, species with unhunted dispersal agents had significantly
+              | lower mean wood density than either species with hunted dispersal modes, or species
+              | dispersed by agents in both categories (ANOVA, F2,                                                                                      277,   p = 0.0017) (Fig. 6a).
+              | Specifically, wind dispersed species had significantly lower wood densities compared
+              | to both mammal and large bird dispersed species (ANOVA, F6, 461, p = 0.0057) (Fig.
+              | 6b).
+blank         | 
+              | 
+text          |                                                                                                                    0.7
+              |                                       1.0
+              |  Mean Wood Specific Gravity (g/cm3)
+blank         | 
+              | 
+              | 
+              | 
+text          |                                                                                                                          B
+blank         | 
+              | 
+              | 
+              | 
+text          |                                                                               Mean Wood Specific Gravity (g/cm3)
+              |                                              A   a
+              |                                                           a
+              |                                                                    b                                               0.6          ab
+              |                                                                                                                                          a
+              |                                                                                                                                                 ab
+              |                                                                                                                                                         a     ab
+              |                                       0.8
+blank         | 
+              | 
+              | 
+              | 
+text          |                                                                                                                           ab                                        b
+              |                                                                                                                    0.5
+              |                                       0.6
+blank         | 
+              | 
+              | 
+              | 
+text          |                                                                                                                    0.4
+              |                                       0.4
+blank         | 
+              | 
+              | 
+              | 
+text          |                                                                                                                    0.3
+              |                                       0.2
+blank         | 
+              | 
+              | 
+              | 
+text          |                                                                                                                    0.2
+              |                                       0.0
+blank         | 
+              | 
+              | 
+              | 
+text          |                                                                                                                    0.1
+blank         | 
+text          |                                               Hunted   Mixed    Unhunted                                           0.0
+              |                                                                                                                          Bat   Bird     L.Bird Explo   Mam   Water Wind
+              |                                                  Dispersal Mode Class
+              |                                                                                                                                       Dispersal Agent
+blank         | 
+              | 
+              | 
+              | 
+text          |                                             FIGURE 2.6. (A) Species with unhunted dispersal agents have lower WSGs. (B)
+              |                                             Wind dispersed species have significantly lower mean wood density than both
+              |                                             mammal and large bird dispersed species. Bars with same letter were not
+              |                                             significantly different in post hoc comparisons.
+blank         | 
+              | 
+title         | Discussion
+blank         | 
+              | 
+title         | Hunting, vertebrate abundance, and plant functional diversity
+blank         | 
+              | 
+text          | This study demonstrates that hunting effects change in the seedling community of
+              | tropical forests primarily by perturbing biotic seed dispersal and vertebrate seed
+              | predation. Increases in lianas and reductions in species with hunted biotic dispersal
+              | agents were observed in the hunting experiment, where primary dispersers were
+              | reduced (Wright et al. 2007b), but not in the LTEE, where primary dispersers were not
+              | manipulated (Figs. 2 and 3). Activity of seed predators was dramatically reduced in
+              | the LTEE, and here we observed increases in median seed mass (Fig. 4). This result is
+meta          | 46         CHAPTER 2: CONSEQUENCES OF HUNTING FOR SEED DISPERSAL AND WOOD DENSITY
+blank         | 
+              | 
+              | 
+text          | consistent with larger-seed species experiencing an escape from vertebrate seed
+              | predation (Dirzo et al., 2007), and lends strong support to the view that changes in
+              | seed predator abundance are responsible for the increases in median seed mass
+              | observed in hunted forests relative to protected forests (Wright et al. 2007b).
+              |        Leaf traits showed no differences in both the LTEE and the hunting
+              | experiment. This suggests that the reduction of vertebrate herbivores by hunting does
+              | not have a strong impact on changes in seedling community composition at early
+              | stages of plant recruitment.
+              |                Wood density responses to herbivore reduction differed between the
+              | LTEE and hunting experiment. The likeliest explanation for the discrepancy is that, as
+              | with differences in dispersal mode and growth form responses between our
+              | experiments, differences in wood density responses are dispersal-driven. To test this
+              | post hoc hypothesis, we evaluated whether dispersal mode classes differ in their mean
+              | wood densities. Consistent with our hypothesis that experimental differences in wood
+              | density responses are due to differences in dispersal, wood density was significantly
+              | lower for species with unhunted dispersal agents than for species having either hunted
+              | dispersal agents, or a combination of the two (Fig. 6).
+              |        In contrast to the hunting experiment, the LTEE shows higher spiny rat activity
+              | in the defaunated treatment. As spiny rats are known to be significant seedling
+              | predators in the absence of other competitors (Asquith et al. 1997), it is also possible
+              | that they may be compensating for the reduction in other seedling predators inside the
+              | exclosures. However several lines of evidence contradict this putative explanation.
+              |        First, if spiny rats were compensating for other herbivores inside the exclosure
+              | as in the study of Asquith et al. (1997), we would expect overall seedling densities to
+              | be similar in open and exclosure plots, but instead, seedling densities in exclosures are
+              | about 2-fold higher than in open plots (data not shown). The herbivory study on
+              | islands which indicated that spiny rats alone can achieve seedling predation rates as
+              | high as communities with relatively intact mammal fauna were conducted on islands
+              | in which spiny rat densities are among the highest rodent densities ever documented
+              | (Adler 1996), but our study does not replicate these high densities. In addition,
+meta          |                                                                                         47
+blank         | 
+              | 
+              | 
+text          | herbivory by spiny rats cannot explain why wood density responses differ between
+              | experiments, but leaf traits did not.
+blank         | 
+              | 
+text          |        Our evidence for a link between seed dispersal and community wood density is
+              | an interesting result because Brodie and Gibbs (2009) recently proposed that
+              | reductions in seed dispersal as a consequence of defaunation may result in shifts in
+              | mean wood density at the community level.           However, the mechanism that they
+              | proposed assumed that seed mass was positively correlated with wood density across
+              | species in tropical forests, which is true neither generally, nor for our site specifically
+              | (Wright et al. 2010).      Here we show that dispersal likely mediates changes in
+              | community wood density distributions via an alternative mechanism, namely that low
+              | and high wood density plant species are disportionately associated with dispersal
+              | mode agents that are favored and disfavored respectively by hunting.
+blank         | 
+              | 
+text          | Implications of defaunation for carbon sequestration in tropical forests.
+blank         | 
+              | 
+text          |        Our results suggest that there may be important links between hunting,
+              | frugivore abundance, and the ability of future tropical forests to sequester carbon.
+              | Disruption of seed dispersal by game animals has the potential to alter the ability of
+              | forests to store carbon by (1) decreasing the abundance of trees relative to lianas
+              | (Wright et al. 2007b), and (2) shifting wood density distributions in the community
+              | toward lower wood density species (Fig 5a).
+              |        Higher liana relative abundance may reduce carbon storage at the ecosystem
+              | level in two ways. Lianas compete with trees for light, displace tree leaf mass (Kira &
+              | Ogawa 1971), and lower tree growth.         One recent study from Amazonia estimated
+              | that above ground competition from lianas was responsible for reductions in tree
+              | growth that reduced carbon sequestration by the equivalent of 0.25 Mg C ha-1 yr-1 (van
+              | der Heijden & Phillips, 2009). Growth of the lianas themselves amounted to 0.07 Mg
+              | C ha-1 yr-1, which did not compensate for reductions in tree growth. The authors
+meta          | 48         CHAPTER 2: CONSEQUENCES OF HUNTING FOR SEED DISPERSAL AND WOOD DENSITY
+blank         | 
+              | 
+              | 
+text          | therefore estimated that competition from lianas reduced carbon sequestration at the
+              | ecosystem level by net amount of 0.18 Mg C ha-1 yr-1 or 71%.
+              |        The consequences of lowering community mean wood density for carbon
+              | storage depend on the relationship between wood density and tree volume in a
+              | particular forest. Three Panamanian forests in the region, including BCI, exhibit a
+              | weak, but negative relationship between plot mean wood density and plot biomass
+              | (Stegen et al. 2009). This suggests that the reduction in community wood density in
+              | our study site may actually lead to higher forest biomass, and partially compensate for
+              | any reduction resulting from the relative reduction in free-standing stems. However,
+              | forests exhibiting positive wood density-biomass relationships could experience a
+              | decrease in carbon sequestration potential under the same scenario of reduced plot
+              | wood density.
+              |        The phenomenon of over-hunting in tropical forests is widespread and severe.
+              | If the changes in carbon storage potential represented by shifting seedling
+              | communities in this study are as wide-spread as over-hunting itself, the potential
+              | impact may be great. However, many questions must be answered before we can
+              | understand the magnitude of the impact that hunting may have on the ability of
+              | tropical forests to sequester carbon.
+              |        First, to our knowledge, this is the first study demonstrating that tropical tree
+              | wood densities are associated with dispersal mode classes, and in a way that would
+              | respond to hunting. Studies investigating the generality of this association would
+              | represent the first step in predicting whether the changes observed in our site would
+              | apply pan-tropically. Secondly, here we report changes in the seedling community,
+              | but better models of forest dynamics are needed to understand the extent to which
+              | those changes will propagate to the adult community. Third, how lianas alter carbon
+              | storage at the ecosystem level is poorly understood, and in particular we do not
+              | understand how liana competition with trees varies as a function of liana abundance.
+              | Finally, as noted by Stegen et al. (2009), there are many unanswered questions
+              | pertaining to how changes in wood density distributions translate to changes in
+              | biomass at the plot level.
+meta          |                                                                                    49
+blank         | 
+              | 
+              | 
+text          |            Hunting is not the only cause of defaunation. Frugivores, granivores, and
+              | herbivores may also decrease in abundance as a result of habitat fragmentation and
+              | loss (Asquith et al. 1999, Galetti et al. 2006, Terborgh et al. 2006, Cramer et al.
+              | 2007).      As with hunting, decreases in abundance with intensity of habitat loss
+              | disproportionately affect larger-bodied animals. Therefore, even in the absence of
+              | hunting, other sources of defaunation may be expected to have similar indirect effects
+              | on plant community functional trait composition and ecosystem services in tropical
+              | forests.
+blank         | 
+              | 
+title         | Acknowledgements
+blank         | 
+              | 
+text          | We extend our gratitude to the Autoridad Nacional del Ambiente (ANAM) for
+              | permitting us to conduct research within the Parque Nacional Soberanía.        David
+              | Brassfield, Didimo Ureña, and Joana Balbuena conducted the seedling censuses in the
+              | hunting experiment. A competitive research grant from the Smithsonian Tropical
+              | Research Institute funded seedling censuses in hunted and protected forests. The
+              | Smithsonian Tropical Research Institute, the U.S. Department of Defence Legacy
+              | Fund, and the U.S. Agency for International Development funded canopy tree
+              | censuses at hunted and protected forests. NSF grant DEB 9527729 funded the LTEE.
+              | NSF DEB-0808338, the Theresa Heinz Environmental Scholars program, and the
+              | STRI short-term fellowship program provided support to E. Kurten.
+              | 50   CHAPTER 2: CONSEQUENCES OF HUNTING FOR SEED DISPERSAL AND WOOD DENSITY
+meta          |                                                                                      51
+blank         | 
+              | 
+              | 
+              | 
+title         | Chapter 3
+blank         | 
+title         | Terrestrial mammalian herbivores
+              | influence the distribution of defense
+              | and nutrient traits in a Neotropical
+              | forest
+              | Erin. L. Kurten, Walter P. Carson
+blank         | 
+title         | ABSTRACT
+blank         | 
+text          | Mammalian herbivores are known to be important regulators of plant community
+              | composition in tropical savannas and temperate systems. However, their impact on
+              | tropical communities beyond their impact as seedling predators has remained largely
+              | unexplored, perhaps because of the assumed importance of insect herbivory for
+              | tropical plants beyond the seedling stage. In this study, we censused 35,069 saplings
+              | in a long-term terrestrial mammal exclosure experiment in Central Panama to
+              | determine whether mammalian herbivores had the ability to regulate the abundances
+              | of species differing in wood density, specific leaf area (SLA), leaf toughness, and leaf
+              | nitrogen. We found that community mean leaf toughness decreased in exclosures over
+              | time, while community mean leaf nitrogen increased.           These differences were
+              | primarily due to an increased abundance of species possessing favored traits, rather
+              | than species turnover as a consequence of herbivore exclusion. No community-level
+              | differences were observed for mean wood density or specific leaf area. We also found
+meta          | 52               CHAPTER 3: EFFECTS OF MAMMALIAN HERBIVORES ON SAPLING TRAITS
+blank         | 
+              | 
+              | 
+text          | that intraspecific leaf toughness was lower in exclosures. Our results suggest that
+              | mammalian herbivores are important for shaping plant functional trait distributions
+              | well beyond the seedling stage, and that their effects on the sapling community are
+              | distinct from their effects on the seedling community.
+blank         | 
+              | 
+title         | INTRODUCTION
+blank         | 
+text          | Mammalian herbivores are known to be important top-down regulators of plant
+              | community composition in many ecosystems (Olff & Ritchie 1998, Parker et al.
+              | 2006). In tropical forests, mammals are recognized as significant seedling predators
+              | (Barone and Coley 1996). However their importance as herbivores in later stages of
+              | plant recruitment has been less studied.
+              |        One study in peninsular Malaysia found that native pigs (Sus scrofa) uproot or
+              | snap 500 saplings per female pig in their reproductive nest building (Ickes et al. 2001).
+              | Experimental exclusion of pigs resulted in sapling density increasing by 32 percent,
+              | relative to open controls in just two years (Ickes et al. 2001). Lizcano (2006) excluded
+              | large mammalian herbivores, including mountain tapir (Tapirus pinchaque) and
+              | brocket deer (Mazama spp.), in montane tropical forests in Columbia and discovered a
+              | 2-fold increase in understory plant biomass in two years.
+              |        These two studies found higher species richness in exclosures, as may be
+              | expected with higher stem densities, but no differences in species diversity indices.
+              | The discrepancy between species richness and species diversity indices is explained by
+              | the fact that species dominance often increases in mammal exclosures as well. The
+              | positive and negative influences of increased species richness and dominance
+              | respectively on the diversity index of an exclosure community can cancel each other
+              | out, leading to the diversity index of an exclosure community being similar to that of a
+              | control community.
+              |        In other ecosystems, herbivore exclosure not only results in changes in plant
+              | density and diversity, but also in plant functional trait composition. This is because,
+              | although mammals are generalist herbivores relative to many insects, they still exhibit
+              | diet preferences. In one temperate forest, browsers selected species lower in defense
+meta          |                                                                                        53
+blank         | 
+              | 
+              | 
+text          | compounds (e.g. leaf phenolics, condensed tannins, fiber, lignin) (Wardle et al. 2002),
+              | and in a sub-humid grassland, grazers preferred taller-growing species (Diaz et al.
+              | 2001).
+              |          However, at the community level, evolutionary history and present day
+              | intensity of herbivory interact with diet preferences to produce mixed effects on
+              | functional trait composition.      In cases where species have evolved naïve of
+              | mammalian herbivores, introduced browsers change species composition in favor of
+              | species with lower average tissue quality (Wardle et al. 2002). The same occurs in
+              | intensely grazed grasslands (Chaneton et al. 1988). In cases where plant communities
+              | have evolved with mammalian herbivores, herbivory results in communities of species
+              | with higher average tissue quality, both in the presence of native herbivores
+              | (McNaughton 1985), as well as domestic herbivores that are not grazing at intensive
+              | levels (Cingolani et al. 2002).
+              |          In contrast to other ecosystems, mammals in tropical forests are perceived to
+              | be relatively unimportant herbivores. Mammals are responsible for about one quarter
+              | of all herbivory in tropical forests. As generalists, they are perceived to have less of a
+              | selective influence in shaping plant community composition and plant anti-herbivore
+              | defense than specialist invertebrate herbivores (Coley and Barone 2003). However,
+              | tropical forest mammals are important seedling predators. Therefore, the primary aim
+              | of this study was to test whether mammalian herbivores could indeed have effects on
+              | plant functional trait composition in tropical forests similar to those seen in temperate
+              | forests by influencing the composition of the younger cohorts of the plant community.
+              |          We conducted this study in the Barro Colorado National Monument in central
+              | Panama, where a terrestrial mammal exclosure experiment has been ongoing for the
+              | past fifteen years (Royo & Carson 2005). We predicted that if mammalian herbivores
+              | were important for determining sapling functional trait distributions, exclosure would
+              | favor species that are faster growing (typically species with low wood density and
+              | high specific leaf area, or SLA), less defended (low leaf toughness) and which have
+              | higher nutrient leaves (higher leaf nitrogen). Our second goal was to evaluate the
+              | hypothesis that differences in traits between treatments developed both as a
+meta          | 54               CHAPTER 3: EFFECTS OF MAMMALIAN HERBIVORES ON SAPLING TRAITS
+blank         | 
+              | 
+              | 
+text          | consequence of species turnover as well as increases in abundance of favored species
+              | already common to both treatments.
+              |        A final goal of this study was to examine whether the traits of the species
+              | themselves changed intraspecifically with mammal exclusion. In temperate systems,
+              | changes in mammalian herbivory induce changes in photosynthetic rates (Oleksyn et
+              | al. 1998, Nabeshima et al. 2001), photosynthetic efficiency (Retuerto et al. 2004), leaf
+              | toughness (Shimazaki & Miyashita 2002). In tropical savanna, megaherbivores
+              | influence the concentrations of condensed tannins in Acacia (Ward & Young 2002).
+              | Though herbivory rates in tropical forests are much higher overall in comparison to
+              | temperate forests (Coley & Barone 1996), suggesting that perhaps constitutive anti-
+              | herbivore defenses should be ubiquitous. However, several tropical species alter their
+              | defense in response to herbivory, and it has been suggested that the lack of more
+              | tropical examples of plasticity in defense are due to lack of study rather than
+              | biological differences between temperate and tropical plants (Karban & Baldwin
+              | 1997). We therefore examined whether the defense trait we examined, leaf toughness,
+              | may decrease intraspecifically in response to mammal exclusion.
+blank         | 
+              | 
+title         | METHODS
+blank         | 
+              | 
+text          | Study sites. The forest in Central Panama surrounding Lake Gatun is semi-deciduous
+              | moist forest. We conducted our experiment on Barro Colorado Island (BCI) and on
+              | the Gigante Peninsula, within the Barro Colorado National Monument (BCNM). The
+              | BCI has relatively high abundances of vertebrate herbivores affected by mammal
+              | exclosure, compared to other Neotropical forests, including deer (Mazama Americana
+              | and Odocoileus virginianus), collared peccary (Tayassu tajacu), agoutis (Dasyprocta
+              | punctata), pacas (Agouti paca), rabbits (Silvilagus brasiliensis). This is due to a
+              | combination of factors, including a lack of resident large felines on the island,
+              | protection from poaching, and other environmental factors. Tapir (Tapirus bairdii) are
+              | also present on BCI, but are rare.
+meta          |                                                                                        55
+blank         | 
+              | 
+              | 
+text          | Exclosure experimental design. Between late-1993 and mid-1994, eight pairs of
+              | fenced, exclosure plots and open, control plots were established within the BCNM
+              | (Royo & Carson 2005). Four pairs of plots are on (BCI) and the remaining four pairs
+              | are on the mainland Gigante Penninsula. Fenced and unfenced plots are 30 m x 45 m
+              | and are approximately 5 m apart. Exclosure fences are constructed of 12.7 x 12.7 cm
+              | galvanized steel fencing 2.2 m tall and buried 0.25 m deep. A secondary 1.3 x 1.3 cm
+              | mesh surrounds the lower 70 cm and also extends 0.25 m below ground. A 5-7 m
+              | buffer zone at the plot edge, and bisecting the plot, allowed for access and the
+              | avoidance of fence effects. Exclosures excluded the species mentioned above, but did
+              | not exclude climbing mammals such as squirrels and spiny rats, nor did they exclude
+              | arboreal mammals such as monkeys (Fig. 2.1). Monitoring data suggest that birds
+              | were not altered by the exclosures, with the possible exception of tinamous (Fig. 2.1).
+              | Though tortoises occur on our study site and are likely to be excluded by the fences,
+              | they are small and rare, and generally not considered to have an effect on our study
+              | results.
+blank         | 
+text          | Plant censuses. In 1994 and 2003, all woody juveniles (“saplings”) taller than 40 cm
+              | or larger than 3 mm diameter at 20 cm in height and not in the buffer zone were
+              | identified, measured and marked.
+blank         | 
+              | 
+text          | Plant traits. We sampled leaves from up to six individual saplings of each tree
+              | species in each of the eight plots occurring on BCI. Specific leaf area was measured
+              | on shade leaves of saplings as the fresh lamina area divided by the mass of a whole
+              | leaf with petiole after oven drying to constant mass at 60ºC (Cornelissen et al. 2003).
+              | Leaf samples were dried at 65ºC, ground using a ball grinder, and analyzed on an
+              | NA1500 elemental analyzer (Carlo Erba Instruments, Italy) for total N and C. Leaf
+              | toughness was measured as the force necessary to punch a 3mm diameter hole the leaf
+              | lamina with a force gauge, avoiding leaf veins to the extent possible (Sagers &Coley
+              | 1995). Wright and colleagues (2007b) determined dry seed mass and adult tree wood
+              | specific gravity (2010) for both free standing species and lianas in central Panama.
+meta          | 56                CHAPTER 3: EFFECTS OF MAMMALIAN HERBIVORES ON SAPLING TRAITS
+blank         | 
+              | 
+              | 
+text          | Analyses. Trait correlations were conducted to assess whether traits in this study were
+              | orthogonal to one another, and thus likely to respond independently to herbivory.
+              | Community mean trait values were calculated in three ways. First, we calculated the
+              | grand means of the traits at the species level, and used these species means to
+              | determine the average trait value of the species occurring in each plot. We next
+              | repeated this analysis, but weighted the species’ trait means by species’ abundances in
+              | the plots. Finally, we calculated species trait means separately for exclosure and open
+              | communities, and recalculated the abundance-weighted plot means using these
+              | treatment-specific species trait values.
+              |        Linear mixed models were used to evaluate whether changes in community
+              | mean trait values over time differed between treatments. Exclosure and time were
+              | treated as fixed factors and the exclosure-control pair were treated as a random factor
+              | nested within time. Our experiment tested for an exclosure-time interaction, and
+              | significance values reported here refer to that interaction. All analyses were conducted
+              | in R (R Development Core Team, 2008).
+blank         | 
+title         | RESULTS
+blank         | 
+text          | SPECIES TRAITS Functional traits were measured for 38.5-76.4 percent of species in
+              | the experiment and 58.8-91.6 percent of stems in the experiment (Table 3.1). Traits
+              | correlations were largely not significant, with the exceptions of SLA-leaf toughness
+              | and SLA-leaf nitrogen (Table 3.2). However, the correlation coefficients for these two
+              | relationships were still low (r2 = 0.325 and r2 = 0.231 respectively), and our results did
+              | not show these trait pairs responding in a coordinated fashion (Fig. 3.1 & 3.2).
+meta          |                                                                                       57
+blank         | 
+              | 
+text          |        TABLE 3.1. The number of species and stems for which traits were measured.
+blank         | 
+text          |                                   No. Spp      % Spp      No. Stems      % Stems
+              |                   Total            369           -         35,069           -
+              |           Wood specific gravity    282          76.4       32,113         91.6
+              |            Specific Leaf Area      154          41.7       23,342         66.6
+              |             Leaf Toughness         152          41.2       23,322         66.5
+              |              Leaf Nitrogen         142          38.5       20,633         58.8
+blank         | 
+              | 
+text          |        TABLE 3.2. R2 values for pair-wise correlations between species mean trait
+              |        values evaluated in this study. The sign in parenthesis denotes whether the
+              |        relationship was positive or negative. All correlations are significant (p <
+              |        0.001).
+blank         | 
+text          |                        Wood specific gravity        SLA         Toughness
+              |              SLA            0.031 (-)
+              |           Toughness         0.004 (-)             0.325 (-)
+              |            Nitrogen          0.11 (-)             0.231 (+)       0.081 (-)
+blank         | 
+              | 
+              | 
+              | 
+text          | EFFECTS   OF   SPECIES OCCURRENCE In 1993, approximately 87 percent of species
+              | occurring in the experiment were present in both treatments. In 2003, 75 percent of
+              | species occurring in the experiment were present in both treatments. Therefore, when
+              | we averaged the trait means of the species occurring in each plot, we observed that the
+              | changes in plot trait means over time did not differ between treatments for any of the
+              | traits measured (Fig. 3.1).
+meta          | 58                                             CHAPTER 3: EFFECTS OF MAMMALIAN HERBIVORES ON SAPLING TRAITS
+blank         | 
+              | 
+              | 
+              | 
+text          |         Wood Specific Gravity (g/cm3)
+              |                                         0.60                                                   22
+              |                                                  1994          2003                                    1994        2003
+blank         | 
+              | 
+              | 
+              | 
+text          |                                                                         Mean SLA (mg/mm2)
+              |                                         0.59                                                   21
+              |                                         0.58                                                   20
+              |                                         0.57                                                   19
+              |                                         0.56                                                   18
+              |                                         0.55                                                   17
+              |                                         0.54                                                   16
+              |                                                Open Excl    Open Excl                                Open Excl   Open Excl
+              |                                          A           Treatment                                 B           Treatment
+              |         Mean Leaf Toughness (g/mm2)
+blank         | 
+              | 
+              | 
+              | 
+text          |                                          34
+              |                                                  1994          2003                            2.6     1994        2003
+              |                                                                         Mean % Leaf Nitrogen
+              |                                          33
+              |                                                                                                2.4
+              |                                          32
+              |                                                                                                2.2
+              |                                          31
+              |                                                                                                2.0
+              |                                          30
+              |                                                                                                1.8
+              |                                          29
+              |                                                                                                1.6
+              |                                          28
+              |                                                Open Excl    Open Excl                                Open Excl   Open Excl
+              |                                          C           Treatment                                 D           Treatment
+blank         | 
+text          |        FIGURE 3.1. Changes in plot-level trait means, not weighted by species
+              |        abundance, did not differ significantly between treatments for (A) wood
+              |        density, (B) SLA, (C) leaf toughness or (D) leaf nitrogen.
+blank         | 
+text          | EFFECTS             OF                    SPECIES ABUNDANCE        Stem densities increased by 46% in exclosures
+              | from 1994 to 2003, but did not increase in open plots (data not shown). When we
+              | averaged the trait means of the species occurring in each plot, and weighted traits by
+              | species abundance, we did observe differences between treatments for some traits.
+              | We predicted that exclosures would favor species with lower wood density, lower
+              | SLA, lower leaf toughness, and higher leaf nitrogen. We did not see significant
+              | differences between treatments in wood density (Fig. 3.2A) or SLA (Fig. 3.2B).
+              | However, changes in leaf toughness (Fig. 3.2C) and leaf nitrogen (Fig. 3.D) over time
+              | were significantly different between treatments. Consistent with our expectations,
+              | exclosures had lower community mean leaf toughness and higher community mean
+meta          |                                                                                                                                  59
+blank         | 
+              | 
+              | 
+text          | leaf nitrogen in 2003, relative to open plots. However these differences did not exist
+              | when the experiment was initiated in 1994 (Fig. 3.2 C & D).
+blank         | 
+              | 
+              | 
+text          |         Wood Specific Gravity (g/cm3)
+              |                                         0.60                                                  24
+              |                                                  1994        2003                                     1994         2003
+blank         | 
+              | 
+              | 
+              | 
+text          |                                                                        Mean SLA (mg/mm2)
+              |                                         0.59                                                  22
+              |                                         0.58                                                  20
+              |                                         0.57                                                  18
+              |                                         0.56                                                  16
+              |                                         0.55                                                  14
+              |                                         0.54                                                  12
+              |                                                Open Excl   Open Excl                                Open Excl    Open Excl
+              |                                          A           Treatment                                B           Treatment
+              |         Mean Leaf Toughness (g/mm3)
+blank         | 
+              | 
+              | 
+              | 
+text          |                                          38                                                   2.6
+              |                                                  1994        2003      Mean % Leaf Nitrogen           1994         2003
+              |                                          36                                                   2.4
+              |                                                                  *                                                    *
+              |                                          34                                                   2.2
+              |                                          32                                                   2.0
+              |                                          30                                                   1.8
+blank         | 
+text          |                                          28                                                   1.6
+blank         | 
+text          |                                          26                                                   1.4
+              |                                                Open Excl   Open Excl                                Open Excl    Open Excl
+              |                                          C           Treatment                                D           Treatment
+blank         | 
+              | 
+text          |        FIGURE 3.2. Changes in abundance-weighted, plot-level trait means did not
+              |        differ significantly between treatments for (A) wood density or (B) SLA, but
+              |        did differ significantly between treatments for (C) leaf toughness and (D) leaf
+              |        nitrogen. (* p = 0.06 , time x treatment interaction)
+blank         | 
+              | 
+text          | INTRASPECIFIC DIFFERENCES IN                               LEAF TRAITS                        Because we sampled leaf traits in both
+              | open and exclosure plots, we could examine how intraspecific differences in trait
+              | means contributed to the differences between treatments at the community level. For
+              | one trait, leaf toughness, we found that determining species trait means separately for
+              | open and exclosure treatments, and using those treatment-specific species trait means
+              | to calculate plot-level means, resulted in an enhanced difference between treatments at
+meta          | 60               CHAPTER 3: EFFECTS OF MAMMALIAN HERBIVORES ON SAPLING TRAITS
+blank         | 
+              | 
+              | 
+text          | the plot-level in the later time point (Fig. 3.3), relative to using grand trait means for
+              | species (Fig. 3.2C).
+blank         | 
+text          |                                                       38
+blank         | 
+              | 
+              | 
+text          |                         Mean Leaf Toughness (g/mm2)
+              |                                                              1994               2003
+              |                                                       36
+              |                                                       34                         *
+              |                                                       32
+              |                                                       30
+              |                                                       28
+              |                                                       26
+              |                                                            Open   Excl     Open Excl
+              |                                                                     Treatment
+blank         | 
+text          |        FIGURE 3.3. Intraspecific differences in species’ mean leaf toughness between
+              |        open and exclosure plots increase the difference in treatment-level mean leaf
+              |        toughness in 2003. (* p = 0.004, time x treatment interaction)
+blank         | 
+text          |        For species having at least two leaf samples from each treatment type, 61.8%
+              | of species had lower leaf toughnesses in exclosures relative to controls (binomial test,
+              | N = 76, p = 0.025). This bias was even stronger when the sample was limited to
+              | species with treatment mean toughnesses based on at least eight leaf samples, with
+              | 73.9% of species having lower leaf toughnesses in exclosures (binomial test, N = 23,
+              | p = 0.017).
+              |        Because sample sizes among species and between treatments was unbalanced,
+              | a MANOVA approach could not be taken to assess whether treatment means within a
+              | species were significantly different. Therefore, using the latter subset of species
+              | above, a two-way ANOVA was applied. The association of exclosure treatment with
+              | lower mean leaf toughness with species was also significant using this approach
+              | (Table 3.3).
+meta          |                                                                                      61
+blank         | 
+              | 
+              | 
+text          |            TABLE 3.3. Effect of species and exclosure on variation in leaf
+              |            toughness. N = 23 species, each of which had at least eight leaf
+              |            toughness samples from each of the two treatment types.
+blank         | 
+text          |             Effect                  df      SS      MS        F        p
+              |             Species                 22    207.4    9.425    165.2   < 0.001
+              |             Exclosure               1     0.305    0.305    5.352    0.021
+              |             Species x Exclosure     22    1.292    0.059    1.029    0.425
+              |             Residuals              604    34.467   0.057
+blank         | 
+              | 
+              | 
+title         | DISCUSSION
+blank         | 
+text          | Regulation of plant functional trait composition by mammalian herbivores
+              | In this study, we demonstrate that terrestrial mammals influence understory functional
+              | trait distributions in tropical forests. They do so primarily by suppressing populations
+              | of species with traits that are correlated with high nutrition and low defense.
+              | Consistent with our predictions that reduced herbivory should favor species with low
+              | defense and high nutrient content, community mean leaf toughness decreased over
+              | time in exclosures over time (Fig. 3.2C), and community mean leaf nitrogen increased
+              | in exclosures over time (Fig 3.2D), while open control plots did not show equivalent
+              | changes.
+              |        Wood density is associated with seedling survival in the presence of herbivores
+              | in this forest (Alvarez-Claire & Kitajima 2007, 2009). Browsers and grazers have also
+              | been shown to select species with higher SLA in other systems (Wardle et al. 2002,
+              | Cingolani et al. 2002).     However, neither of these traits changed in response to
+              | mammal exclosure. This could be because, at the sapling stage in tropical forests, the
+              | low light conditions of the forest understory also favor low SLA, high wood density
+              | species and perhaps this environmental factor constrains the community response to
+              | herbivory with respect to those traits.
+              |        We did not find that changes in functional traits at the community-level were
+              | the result of divergences in species presence-absence between exclosure and open
+              | plots. The sapling community established prior to the imposition of the exclosure
+              | treatments, and therefore two treatments started out with very similar communities.
+meta          | 62               CHAPTER 3: EFFECTS OF MAMMALIAN HERBIVORES ON SAPLING TRAITS
+blank         | 
+              | 
+              | 
+text          | Over longer time scales, competitively dominant species may outcompete less
+              | competitive species in the exclosure treatment, resulting in lower species richness,
+              | particularly of species which are better defended but slower growing. However, we
+              | did not observe decreases in species richness in the time-frame of this experiment
+              | (data not shown).
+              |        Seed and seedling predators are also excluded from the exclosure communities.
+              | Therefore, it is reasonable to question whether the effects we observed are due
+              | exclusively to herbivore exclusion. Two lines of evidence allow us to attribute our
+              | results to herbivores. First, long-term studies of seedling growth in central Panama
+              | suggest that all individuals in this size class established before the experimental
+              | treatments were imposed (S.J. Wright, pers. comm.). Therefore, saplings in both
+              | treatments should have been exposed to the same seed and seedling predation agents.
+              |        Consistent with the assumption that saplings established prior to exclusion of
+              | seed and seedling predators, community mean seed mass does not differ between
+              | treatments for the sapling community (data not shown). This is in contrast to the
+              | seedlings in the exclosure communities, which did establish under conditions of seed
+              | and seedling predator exclusion, and as a result have higher community mean seed
+              | mass than seedlings in open control communities (Fig. 2.4).
+              |        These two lines of evidence suggest that exclusion of seed and seedling
+              | predators are not responsible for the exclosure effects observed in the sapling
+              | community, leaving only herbivory as a possible explanation.
+blank         | 
+              | 
+title         | Intraspecific differences in leaf toughness
+text          | The majority of species in the experiment showed lower mean leaf toughness inside
+              | the exclosure, relative to open plots. When these intraspecific differences in species
+              | leaf toughness were accounted for (Fig. 3.3), treatment differences in plot level leaf
+              | toughness in 2003 were 66 percent larger than when a grand mean for the species leaf
+              | toughness was applied (Fig. 3.2C). This result suggests that in the case of leaf
+              | toughness, the trait values measured for juvenile tropical plants may vary with the
+              | degree of herbivory experienced.
+meta          |                                                                                         63
+blank         | 
+              | 
+              | 
+text          |        There are several possible explanations for the intraspecific divergence in leaf
+              | toughness. First, it is possible that, within species, genotype or environment produces
+              | variation in leaf toughness, and individuals with lower leaf toughness survive at a
+              | lower rate in open plots than in exclosures . Support for this mechanism can be found
+              | in studies of insect herbivory on plant genotypes which differ in their expression of
+              | qualitative defense traits (Kessler & Baldwin, 2001, Silfer et al. 2009).
+              |        An alternative explanation is that plants are relaxing their investment in
+              | mechanical defenses when they no longer experience mammalian herbivory.
+              | Hundreds of plant species are known to increase investment in defense when exposed
+              | to herbivory, and subsequently attenuate that defense when herbivory levels are lower
+              | (Karban & Baldwin 1997).           However few studies have documented reduced
+              | investment in defense upon reduction of mammalian herbivory.                   In Acacia
+              | drepanolobium decreases in spine length or spinescence are detectable after 22 months
+              | (Young & Okello 1998), decreases in condensed tannin concentrations are detectable
+              | after 24-36 months (Ward & Young 2002) and reduced investment in indirect defenses
+              | occur within several years (Huntzinger et al. 2004).
+              |        Regardless of the mechanism, the fact that mammalian herbivores may play a
+              | role in altering leaf toughness intraspecifically in tropical plants is unexpected. Insects
+              | are thought to be responsible for three quarters of herbivory in tropical forests, and
+              | also the source of the strongest selection pressures plants receive to develop anti-
+              | herbivore defenses in this system (Barone & Coley 2002). The experimental plots in
+              | which we measured leaf toughness did not manipulate insects, and it is plausible that
+              | plants in both treatments could have experience high rates of invertebrate herbivory,
+              | and therefore shown no attenuation in defense traits.        Yet, it appears that insect
+              | herbivores could not compensate for mammalian herbivores in maintaining high
+              | intraspecific leaf toughness when mammals were excluded.
+blank         | 
+              | 
+title         | Implications for tropical defaunation
+text          |        In many temperate forests, populations of native or introduced large
+              | mammalian herbivores are not regulated in a top-down fashion by predators. As a
+meta          | 64               CHAPTER 3: EFFECTS OF MAMMALIAN HERBIVORES ON SAPLING TRAITS
+blank         | 
+              | 
+              | 
+text          | result, these communities currently experience intense levels of herbivory, low rates of
+              | plant recruitment and often decreases in species richness and diversity (Wardle et al.
+              | 2002, Rooney 2001).        However, tropical forests are undergoing the opposite
+              | perturbation. Throughout the tropics, mammalian herbivores are being unsustainably
+              | hunted (Peres & Palacios 2007, Corlett 2007, Fa & Brown 2009). Preferred game
+              | species of animals are often large-bodied herbivores, such as elephants, tapir, pigs,
+              | peccaries and deer.
+              |        Understanding how and why such perturbations in herbivore abundance alter
+              | plant species diversity and functional composition will help conservation biologists
+              | identify potential interventions to conserve plant species threatened indirectly by
+              | defaunation.    However, because most studies examining changes in functional
+              | diversity have been conducted in temperate forests, where the perturbation
+              | experienced by plant communities is increased rates of herbivory, little empirical data
+              | exists with which to evaluate how tropical forests may respond to decreased rates of
+              | herbivory by native herbivores as a consequence of defaunation.
+              |        The biomass of terrestrial mammalian herbivores in tropical forests is quite
+              | low, relative to other ecosystems (Bodmer 1989, Leigh et al. 1982). In addition,
+              | insects are thought to account for approximately 75 percent of overall herbivory in
+              | tropical forests. Taken together, these data might suggest that observed reductions in
+              | terrestrial mammalian herbivore populations in tropical forests may not have large
+              | impacts on sapling community composition.           However, this study demonstrates
+              | reduced herbivory as a consequence of tropical defaunation is likely to cause an
+              | increase in dominance of sapling species with higher leaf nitrogen and lower leaf
+              | toughness. Leaves with higher leaf nitrogen and lower concentrations of defense
+              | compounds typically have higher decomposition rates. Therefore, it is conceivable
+              | that if changes in the sapling community propagate to alter canopy composition,
+              | defaunation may alter nutrient cycling in tropical forests.
+meta          |                                                                                   65
+blank         | 
+              | 
+              | 
+title         | ACKNOWLEDGEMENTS
+blank         | 
+text          | D. Ackerly, L. Curran, R. Dirzo and P. Vitousek provided constructive criticism which
+              | improved this work. The Smithsonian Tropical Research Institute granted permission
+              | to conduct this work, and provided logistical support. D. Turner provided assistance
+              | with leaf nitrogen analysis. J. Wright provided seed mass and wood density data.
+              | Many thanks go to L. Jimenez, C. Sarmiento, S. Rebellon, A. Calderón, R.
+              | Bethancourt for help in the field and lab. A. Hernandez and O. Valdes provided
+              | indispensible help with species identification. NSF DEB-0808338, the Theresa Heinz
+              | Environmental Scholars program, and the STRI short-term fellowship program
+              | provided support to E. Kurten.
+              | 66   CHAPTER 3: EFFECTS OF MAMMALIAN HERBIVORES ON SAPLING TRAITS
+meta          |                                                                                      67
+blank         | 
+              | 
+              | 
+              | 
+title         | Chapter 4
+blank         | 
+text          | Hunting does not alter seed
+              | predation rates as a function of seed
+              | size in a Neotropical forest
+              | Erin L. Kurten
+blank         | 
+title         | ABSTRACT
+blank         | 
+text          | This study tested the hypothesis that vertebrate seed predation rates vary with seed
+              | size and with hunting intensity in tropical forests, corresponding to direct impacts of
+              | hunting on vertebrate species of differing size classes. Seed fate of seeds from species
+              | with mean fresh seed masses ranging from 0.002 to 62.4 g was compared between a
+              | protected and nearby hunted forest in central Panama. Transect survey data verified
+              | that higher hunting intensity corresponded with lower abundances of key mammalian
+              | seed predators. Seed arrays of a subset of species were monitored with camera traps
+              | to verify the identities of animals responsible for seed removal. In this system, there
+              | was only weak support for a relationship between seed predation rates and seed size,
+              | and no evidence that this relationship changed at high hunting intensity. There was
+              | also no evidence that seed predator body size varied with seed size, as agoutis
+              | (Dasyprocta punctata) were responsible for most of the verified removal for five plant
+              | species examined, ranging from 1.95 to 62.4 g in fresh seed mass.
+meta          | 68       CHAPTER 4: SEED PREDATION RATES AS A FUNCTION OF SEED SIZE AND DEFAUNATION
+blank         | 
+              | 
+              | 
+title         | INTRODUCTION
+blank         | 
+text          | Overharvesting of bushmeat is a pervasive threat to biodiversity across the tropics (Fa
+              | & Peres 2001, Peres & Palacios 2007, Corlett 2007). This defaunation not only
+              | directly lowers vertebrate abundance in hunted forests, but initiates trophic cascades
+              | which perturb the plant, insect, and pathogen communities as well. Preferred game
+              | species of animals are usually large-bodied vertebrates, such as ateline monkeys, tapir,
+              | peccary and deer, as well as large birds, and in some cases reptiles, in the context of
+              | Neotropical forests. Understanding how and why such cascading perturbations alter
+              | species diversity and functional composition in the plant community is important for
+              | understanding the role mammals play in maintaining the diversity of tropical forests,
+              | as well as identify potential interventions to conserve plant species threatened by the
+              | negative indirect effects of defaunation.
+              |        Several studies have documented reduced seed dispersal as a consequence of
+              | hunting in tropical forests, particularly for larger-seed species (e.g. Wright et al. 2000,
+              | Guariguata et al. 2000, Guariguata et al. 2002, Galetti et al. 2006, Wang et al. 2007,
+              | Brodie et al. 2009, Fadini et al. 2009, Holbrook & Loiselle 2009). These studies
+              | suggest that reduced dispersal should result in reduced recruitment of large-seeded
+              | species, in particular those that rely on large-bodied mammals vulnerable to hunting
+              | for their seed dispersal (Peres & van Roosmalen 2002, Nuñez-Iturri et al. 2008,
+              | Terborgh et al. 2008).
+              |        However, in most cases, reduced dispersal due to hunting has not been
+              | explicitly linked to subsequent reductions in seedling recruitment. While one study
+              | modeled the potential reduction in seedling recruitment, based on estimates of reduced
+              | dispersal to and differential survival in different microhabitats (Brodie et al. 2009),
+              | only one set of studies of which I am aware measured seed dispersal, seed predation,
+              | and seedling recruitment for the same species in both protected and hunted sites
+              | (Wright et al. 2000, Wright & Duber 2001). Wright and colleagues (2000, 2001)
+              | focused on two large-seeded palm species, members of genera which are widespread
+              | in the Neotropics. They found seed dispersal to be indeed reduced, and seedling
+              | densities near parent trees to be higher, as one would predict to be associated with
+meta          |                                                                                       69
+blank         | 
+              | 
+              | 
+text          | hunting. However, contrary to what has been hypothesized in the seed dispersal
+              | literature, recruitment of Attalea and Astrocaryum palms was actually higher in the
+              | hunted sites (Wright et al. 2000, 20001). The reason for this unexpected increase in
+              | recruitment was that hunting reduced the intensity of vertebrate seed predation, and
+              | this reduction was not fully compensated for by invertebrates.     (Cramer et al. 2003)
+              | suggested a similar reduction in seed predation was responsible for higher recruitment
+              | of the African species Balanites wilsoniana in defaunated sites, despite lower dispersal
+              | by elephants, however seed predation rates were never measured (Babweteera et al.
+              | 2007).
+              |          It is clear from this example that a better understanding of the consequences of
+              | hunting for seed predation is necessary for providing context to the studies
+              | documenting reduced seed dispersal. In studies which have measured seed predation
+              | in sites with differing intensities of hunting, rates of vertebrate seed predation have
+              | often been lower in hunted sites (Terborgh & Wright 1994, Wright et al. 2000, Wright
+              | & Duber 2001, Guariguata 2000, Beckman & Muller-Landau 2007, but see Roldán &
+              | Simonetti 2001).
+              |          Even less clear is how perturbations in seed predation rates as a consequence
+              | of hunting may vary with seed size. Because larger-seeded species are expected to
+              | disproportionately experience reduced seed dispersal, understanding how seed
+              | predation changes with hunting across a range of seed sizes is important for
+              | understanding whether or not the potentially negative demographic consequences of
+              | reduced seed dispersal of larger-seeded species is intensified, maintained, or
+              | compensated for by alterations in seed predation across species (Stoner et al. 2007).
+              | However, most studies which have measured seed predation rates under various
+              | hunting regimes have focused on one or two plant species (e.g. Terborgh & Wright
+              | 1994, Wright et al. 2000, Roldán & Simonetti 2001, Wright & Duber 2001, Galetti et
+              | al. 2006, Beckman & Muller-Landau 2007, Donatti et al. 2009). Due to the variation
+              | in location, mammal community composition, and degree of defaunation in the sites
+              | studied, these data cannot be directly compared to assess how seed predation rates
+              | change with defaunation intensity as a function of seed size.           Guariguata and
+meta          | 70       CHAPTER 4: SEED PREDATION RATES AS A FUNCTION OF SEED SIZE AND DEFAUNATION
+blank         | 
+              | 
+              | 
+text          | colleagues (2000) measured seed predation rates of seven species, but focused
+              | primarily on small-seeded timber species. .
+              |        In 2007, Dirzo and colleagues published a model of how vertebrate seed
+              | predation could be expected to vary with seed size and intensity of vertebrate
+              | defaunation in tropical forests (Fig. 4.1). At low levels of hunting, vulnerable, large-
+              | bodied animals drastically decrease in abundance or become locally extirpated. In
+              | response to this competitive release, as well as a likely reduction in predation by
+              | felines, medium-sized seed predators initially become more abundant, until they
+              | become the primary game species at moderate levels of hunting. At the highest
+              | intensities of defaunation, even medium-sized seed predators have become rare. Small
+              | rodents increase steadily with increasing hunting intensity, as they experience a
+              | reduction in competition and predation.
+              |        In this model, predation intensity of larger-seeded species mirrors the changes
+              | in abundance of medium-bodied seed predators (e.g. 4-20 kg body mass in
+              | Neotropical communities), while predation of smaller-seeded species increases as
+              | small bodied seed predator (< 1 kg) biomass increases. As a result, the relative
+              | proportion of larger and smaller seeds comprising the seedling bank is expected to
+              | shift as defaunation intensity increases. This model was important for clarifying how
+              | seed predation is likely to vary by seed-size, and how this relationship is dependent
+              | upon the defaunation context.
+meta          |                                                                                                                                          71
+blank         | 
+              | 
+text          |                                                       Large vertebrates
+              |                                         A
+blank         | 
+              | 
+text          |                Seed predator Biomass
+              |                                                       Mid-sized vertbrates
+              |                                                       Small rodents
+blank         | 
+text          |                                              BCI
+              |                                                                  PNS
+              |                                                                                                          C           No Def.
+blank         | 
+              | 
+              | 
+              | 
+text          |                                                                               Seed predation Intensity
+              |                                                                                                                     High Def.
+blank         | 
+              | 
+              | 
+              | 
+text          |                                         B                     Large Seeds
+              |              Seed predation Intensity
+blank         | 
+              | 
+text          |                                                               Small Seeds
+blank         | 
+              | 
+              | 
+              | 
+text          |                                                                                                          Small                  La rge
+              |                                                                                                                  Seed Size
+blank         | 
+              | 
+              | 
+text          |                                 None               Moderate            High
+              |                                             Defaunation Intensity
+              |        FIGURE 4.1. Model of how seed predation should vary with seed
+              |        mass as a function of defaunation intensity in a Neotropical forest
+              |        (modified from Dirzo et al. 2007). (A) Changes in large, medium and
+              |        small vertebrate seed predator abundance with increasing defaunation
+              |        intensity. The approximate positions of BCI and PNS along the
+              |        defaunation gradient shown here are based on mammal transect
+              |        surveys. (B) Seed predation intensity of large and small seeded species
+              |        on a biomass basis is correlated with the abundance of medium- and
+              |        small-bodied seed predators respectively. (C) As defaunation
+              |        intensifies, the relationship between seed predation rates and seed size
+              |        goes from positive to negative.
+blank         | 
+text          |        With this study, I aimed to test several assumptions and hypotheses inherent in
+              | this model. The first assumption I tested was that larger-seeded species experience
+              | higher seed predation intensity than smaller seeded species at all but the highest levels
+              | of defaunation intensity. The second assumption of the model I tested was that larger-
+              | seeded species tend to be consumed by large- and medium-bodied vertebrates.
+              | Likewise, smaller-seeded species tend to be consumed by smaller-bodied vertebrates.
+              | A third hypothesis implicit in the model is that at high levels of defaunation intensity,
+              | where the number of large- and medium-bodied vertebrates is highly reduced, larger-
+              | seeded species experience reduced seed predation, relative to less defaunated sites. A
+meta          | 72       CHAPTER 4: SEED PREDATION RATES AS A FUNCTION OF SEED SIZE AND DEFAUNATION
+blank         | 
+              | 
+              | 
+text          | fourth prediction of the model is that smaller-seeded species experience increased seed
+              | predation as defaunation intensity increases, as populations of small bodied vertebrates
+              | increase in response to competitive release and reduced predation. There is a great
+              | interest in determining the extent to which other vertebrates or natural enemies can
+              | compensate for the loss of a primary seed predator.
+              |        To test these hypotheses, seeds of 13 species spanning the range of 0.002 to
+              | 64.2 g mean fresh seed mass were monitored in arrays in both a protected forest and a
+              | nearby hunted forest. Fate of seeds was monitored for four weeks. To           examine
+              | whether larger- and smaller-seeded species indeed had larger- and smaller-bodied
+              | seed predators respectively, the seed arrays on BCI were monitored with camera traps
+              | for five species ranging from 1.95 to 64.2 g fresh mean seed mass for one week, and
+              | the the agents of seed removal were recorded.
+              |        To assess whether other vertebrates might compensate for the loss of a
+              | principle seed predator, I monitored two large-seeded palm species with camera traps
+              | in both sites. This allowed me to determine if and how the species of vertebrate seed
+              | disperser differed between the protected and hunted sites, in particular if smaller
+              | vertebrates compensated for the removal of larger vertebrates. I also recorded the
+              | source of mortality for all species when seeds were killed by invertebrates and
+              | pathogens, to examine whether those agents could also compensate for the loss of
+              | larger bodied seed predators.
+blank         | 
+              | 
+title         | METHODS
+blank         | 
+              | 
+text          | Study sites. The forest in Central Panama surrounding Lake Gatun is semi-deciduous
+              | moist forest. I examined the effects of defaunation by hunting by comparing protected
+              | and hunted forests in this area (Fig. 4.2). Barro Colorado Island (BCI) was chosen as
+              | the protected forest site, and could be regarded as a moderately defaunated site. Due
+              | to its size (15 km2), the island no longer supports white-lipped peccary, and has
+              | visiting, rather than resident, jaguar (Panthera onca) and puma (Puma concolor).
+              | Other large mammals, such as tapir (Tapirus bairdii) and spider monkeys (Ateles
+meta          |                                                                                      73
+blank         | 
+              | 
+              | 
+text          | geoffroyi), were locally extirpated and historically reintroduced (Enders 1939,
+              | Terwilliger 1978, Milton & Hopkins 2005). However today, activity of poachers is
+              | virtually nonexistent on BCI due to the monitoring by forest guards and Panamanian
+              | police. The last confirmed incident of poaching on BCI occurred in 1989 (Wright et
+              | al. 2007).   Consistent with an absence of poachers, researchers on BCI do not
+              | encounter evidence of hunting, and wildlife is common to see.
+              |        Adjacent to the Barro Colorado National Monument (BCNM) is the Parque
+              | Nacional Soberanía (PNS). PNS is a protected area, however the region of the park in
+              | which the study was conducted is not actively patrolled. This is due to both fewer
+              | forest guards at the park, a lack of vehicles to access the study area, which is more
+              | than 20 km from the park headquarters. Consistent with a lack of protection, various
+              | evidence of hunting activity was encountered while conducting this study, including
+              | spent shells, campfires, litter, gun shots, hunting dogs, experimental vandalism, and in
+              | one case, encountering a hunting party.
+blank         | 
+              | 
+              | 
+text          |                         BCNM
+              |                         PNS
+              |                         trails
+blank         | 
+              | 
+              | 
+              | 
+text          |                        N
+              |                                                 0             5 km
+blank         | 
+              | 
+text          |        FIGURE 4.2. Map of Lake Gatun study area in central Panama. Seed
+              |        predation experiments were conducted in areas demarcated by trails.
+meta          | 74       CHAPTER 4: SEED PREDATION RATES AS A FUNCTION OF SEED SIZE AND DEFAUNATION
+blank         | 
+              | 
+              | 
+text          | Mammal census. Diurnal mammal transect surveys were conducted from mid-August
+              | to mid-December in 2008. Once per week in each site, an observer surveyed a 5 km
+              | transect between the hours of 6:30 AM and 11:30 AM, walking at a speed of 1 km/hr.
+              | Surveys were not conducted during rain. Observers recorded the species and age
+              | (adult or juvenile) of all mammals sighted, and measured initial detection distance
+              | with a Nikon ProStaff 550 range finder (Nikon Inc., Melville, NY). Because of the
+              | time and difficulty involved in counting individuals in troupes of primates, these
+              | species were recorded in units of groups. Because most species had few observations
+              | in PNS, and the parameters that density estimators require could not be accurately
+              | estimated for that site, I report animal abundances (individuals/km or groups/km),
+              | rather than densities.
+meta          |                                                                                       75
+blank         | 
+              | 
+              | 
+text          |      TABLE 4.1. Mean fresh seed masses of study species. BCI and PNS
+              |      columns denotes species used in each site. Most species are free standing
+              |      growth forms. C. turczanninowii and T. richardii are lianas.
+              |        Species                       Family          Seed mass (g)   BCI    PNS
+              |        Apeiba tibourbou              Tiliaceae          0.0057         X      X
+              |        Protium tenuifolium           Burseraceae         0.18          X
+              |        Cupania latifolia             Sapindaceae         0.19                 X
+              |        Chamaedorea tepejilote        Arecaceae           0.49          X      X
+              |        Connorus turczanninowii       Conneraceae         0.49          X
+              |        Thevetia ahouai               Apocynaceae         1.11                 X
+              |        Oneocarpus mapora             Arecaceae           1.95          X
+              |        Virola surinamensis           Myristicaeae        2.43          X
+              |        Calophyllum longifolium       Clusiaceae          5.48          X
+              |        Astrocaryum standleyanum      Arecaceae           9.84          X      X
+              |        Gustavia superb               Lecythidaceae       16.0          X
+              |        Attalea butyracea             Arecaceae           16.2          X      X
+              |        Tontelea richardii            Celastraceae        64.2          X
+blank         | 
+              | 
+text          | Seed preparation. Thirteen woody plant species ranging from 0.002 to 64.2 g mean
+              | fresh seed mass were selected for this study (Table 4.1), representing 76 % of the
+              | range in fresh seed mass for this area on a log scale. While these species are common
+              | to both BCI and PNS, heavy insect and/or pathogen infestation of some species in
+              | PNS prevented collection of viable seeds in this site. Because of protected area rules,
+              | seeds could not be moved between sites. Therefore, eleven species were examined on
+              | BCI and six species in PNS, with four species common to both sites, spanning almost
+              | the entire range of seed size in the study.
+              |           Seeds appearing rotten, insect-damaged, or otherwise unviable were not
+              | included in the experiment. To remove confounding effects of differing seed pulps, all
+              | seeds were cleaned of pulp or mucilage. After pulp removal, mass, width and length
+              | were measured for seeds of all species but Apeiba tibourbou. Apeiba seed mass for
+              | BCI was previously recorded (Wright et al. 2007). A thread was attached to all but
+              | the smallest seeds (A. tibourbou and P. tenuifolium), so that I could recover the seeds
+              | and record their fate.
+meta          | 76       CHAPTER 4: SEED PREDATION RATES AS A FUNCTION OF SEED SIZE AND DEFAUNATION
+blank         | 
+              | 
+              | 
+text          | Seed predation experiment. A randomized block design was used to compare seed
+              | predation rates in the two sites.    Within each forest, six blocks were delineated.
+              | Within each block, each species was placed on the forest floor in an array of 6-8 seeds
+              | in a different random location. For G. superba on BCI and C. tepelijote in PNS, only
+              | four arrays were used, as few seeds unattacked by insects were available. Because A.
+              | tibourbou and P. tenuifolium, were small and not marked with thread, these species
+              | were set out in shallow mesh trays with 6-8 seeds per tray. Data reported here is from
+              | a total of 732 seeds.
+              |        Experiments were conducted between mid-July and mid-September 2009.
+              | Seeds were monitored for four weeks, during which seed fate and agent of mortality
+              | were recorded. Classes of seed fate were: unmoved (intact), cached, moved but not
+              | buried (intact), predated by vertebrates, predated by insects, and killed by pathogens.
+              | Seeds which were removed and only the thread recovered were presumed predated. .
+blank         | 
+              | 
+text          | Assessment of seed predator identity. The arrays of 5 species with masses ~2 g or
+              | larger were monitored with Reconyx RC-55 infrared cameras on BCI for the first
+              | week. Only half of the Virola arrays were monitored due to camera limitation. In
+              | addition, the arrays of two of the larger-seeded species, A. standleyanum and A.
+              | butyracea were monitored in PNS. More species were not monitored in PNS due to
+              | concerns over equipment loss. As a conservative assignment of seed predator identity,
+              | the species of seed predator was only recorded when the removal of a marked seed
+              | was clearly visible in the image. Instances of probable seed removal, but in which
+              | camera angle, position of the animal, or clarity of the image prevented a clear
+              | determination of the event, were not assigned to a particular animal species.
+blank         | 
+              | 
+text          | Analyses. The relationships between mean seed mass and mean seed mortality in the
+              | two sites was assessed via an ANCOVA, with site as the fixed factor and log-
+              | transformed mean seed mass as the covariate. Differences in seed caching rates
+              | between sites were evaluated with a t-test for both palm species. All analyses were
+              | conducted in R (R Development Core Team, 2008).
+meta          |                                                                                                 77
+blank         | 
+              | 
+              | 
+title         | RESULTS
+blank         | 
+text          | Mammal census. Diurnal mammal transect surveys, conducted from mid-August to
+              | mid-December in 2008, verified differences in degree of defaunation between BCI and
+              | PNS, which approximately correspond to moderate and intense degrees of defaunation
+              | respectively (Fig. 4.3.)
+blank         | 
+              | 
+title         |                     Medium-bodied
+              |                   Dasyprocta punctata
+              |                   Cebus capuchinus
+              |                   Alouatta palliata
+              |         Species
+blank         | 
+              | 
+              | 
+              | 
+text          |                     Large-bodied
+              |                   Ateles geoffroyi ‡
+              |                   Mazama americana
+              |                   Odocoileus virginianus *
+              |                   Pecari tajacu
+              |                                                                                     BCI
+              |                   Tapirus bairdii ‡                                                 PNS
+blank         | 
+text          |                                              0.0   0.2    0.4       0.6       0.8         1.0
+              |                                                          Individuals per km
+blank         | 
+              | 
+text          |        FIGURE 4.3. Animal abundances in BCI and PNS as assessed by trail
+              |        census in 2008. Above are species of medium body size known to
+              |        maintain or increase their population density with moderate hunting,
+              |        but decrease with intense hunting (Peres and Palacios 2007). Below
+              |        are large-bodied animals that have been historically present in the sites
+              |        and which are known to be most vulnerable to hunting. At both BCI
+              |        and PNS, most large-bodied mamals are rare or locally extinct.
+              |        However, medium-sized mammals are highly abundant in BCI,
+              |        whereas they are also quite rare in PNS, indicating the two sites are
+              |        moderately and intensely hunted, respectively. † locally extirpated in
+              |        PNS; ‡ locally extirpated in BCI, reintroduced, and currently rare;
+              |        *present and rare at both sites, not observed in this census.
+blank         | 
+              | 
+text          | Seed predation intensity. Relationships between seed mortality and seed size were
+              | evaluated by ANCOVA, with site being a fixed factor.                  Seed mortality over all
+              | species, when all sources of mortality were pooled, was not correlated with seed mass
+              | (F1,13 = 1.47, p = 0.25), and this did not differ between sites (F1,13 = 0.69, p = 0.42)
+meta          | 78                          CHAPTER 4: SEED PREDATION RATES AS A FUNCTION OF SEED SIZE AND DEFAUNATION
+blank         | 
+              | 
+              | 
+text          | (Fig. 4.4a). For the subset of species with verified vertebrate seed predation , there
+              | was not significant (F1,12 = 1.45, p = 0.25) (Fig 4.4b) Apeiba, which appeared to have
+              | no vertebrate seed predators at either site. was excluded from the model. ). Also,
+              | contrary to the model (Fig. 4.1), the effect of site was not signficant (F1,13 = 0.11, p =
+              | 0.74) (Fig. 4.4b).                             Mortality caused by invertebrates and fungal pathogens was
+              | negatively correlated with seed mass across all species (F1,13 = 21.1, p > 0.001).
+              | Again, this relationship did not differ by site (F1,13 = 0.19, p = 0.67).
+              |                                                                          Mortality by
+              |                                             All Sources                  Vertebrates        Invertebrates & pathogens
+              |  Proportion seeds killed
+blank         | 
+              | 
+              | 
+              | 
+text          |                            1.0
+              |                                                            A                            B                          C
+              |                            0.8
+blank         | 
+text          |                            0.6
+blank         | 
+text          |                            0.4
+blank         | 
+text          |                            0.2
+blank         | 
+text          |                            0.0
+              |                                  -6    -4     -2   0   2   4   -6   -4     -2   0   2   4 -6    -4   -2   0   2    4
+              |                                       BCI                            Log(Seed mass)
+              |                                       PNS
+              |                            FIGURE 4.4. Seed predation rates as a function of seed size on BCI and in PNS.
+              |                            (A) Across all species there was a negative, but non-significant, relationship
+              |                            between seed mass and seed mortality. (B) Seed mass and vertebrate seed
+              |                            predation were not correlated when analysis was restricted to species with verified
+              |                            vertebrate seed predators. (Points for Apeiba slightly offset for visualization.) (C)
+              |                            Seed mass and seed mortality by invertebrates and fungal pathogens were
+              |                            negatively correlated (includes Apeiba). No differences were observed between
+              |                            sites.
+blank         | 
+              | 
+text          |                            Scatter-hoarding rodents do not just predate seeds. For two of the large seeded
+              | palms, seed caching by scatter-hoarding rodents is critical for escaping seed predation
+              | by bruchid beetles. I predicted that caching of Attelea and Astrocaryum would be
+              | lower in the hunted site, as the primary seed caching agent, the agouti (Dasyprocta
+              | punctata), was less abundant at that site (Fig. 4.3), unless compensated for by
+              | squirrels. Contrary to this prediction, the number of seeds cached did not differ
+              | between sites for either species (Attalea: t = -0.129, p = 0.55, one-tailed test;
+              | Astrocaryum: t = -0.412, p = 0.66, one-tailed test). (Fig. 4.5). This equivalence
+meta          |                                                                                       79
+blank         | 
+              | 
+              | 
+text          | between sites appears not to be due compensation by squirrels, in as far as every seed
+              | caching event for which the agent could be verified by camera trap was committed by
+              | an agouti (data not shown).
+blank         | 
+              | 
+text          |                                     2.5
+blank         | 
+              | 
+text          |                                     2.0
+blank         | 
+text          |                      Seeds Cached
+              |                                     1.5
+blank         | 
+              | 
+text          |                                     1.0
+blank         | 
+              | 
+text          |                                     0.5
+blank         | 
+              | 
+text          |                                     0.0
+              |                                           BCI   PNS           BCI   PNS
+              |                                            Attalea           Astrocaryum
+              |                                                  Sites by Species
+              |         FIGURE 4.5. Number of seeds cached per array of eight seeds for two
+              |         large seeded palms did not differ between BCI and PNS.
+blank         | 
+text          | Seed predator identity. Camera trap monitoring of seeds in the first week on BCI
+              | revealed D. punctata, to be the primary agent of seed removal for all species (Fig.
+              | 4.6).   Seed removal by spiny rats (Proechimys steerei) and squirrels (Sciurus
+              | granatensis) accounted for little of the seed predation and varied by plant species (Fig.
+              | 4.6). Data from BCI do not appear support the hypothesis that seed predator size is
+              | correlated with seed size.
+meta          | 80                        CHAPTER 4: SEED PREDATION RATES AS A FUNCTION OF SEED SIZE AND DEFAUNATION
+blank         | 
+              | 
+              | 
+text          |                     50                                                                                50
+              |                                                                         Not taken
+              |                                                                         Unknown
+              |                                                                         Squirrel
+              |                                                                         Spiny rat                     40
+              |                     40                                                  Agouti
+blank         | 
+              | 
+              | 
+              | 
+text          |                                                                                     Number of seeds
+              |   Number of seeds
+blank         | 
+              | 
+              | 
+              | 
+text          |                                                                                                       30
+              |                     30
+blank         | 
+              | 
+text          |                                                                                                       20
+              |                     20
+blank         | 
+              | 
+text          |                                                                                                       10
+              |                     10
+blank         | 
+text          |                                                                                                        0
+              |                     0
+              |                                                                                                            BCI     PNS     BCI      PNS
+              |                          Oneocarpus   Virola   Astrocaryum   Attalea   Tontelea
+              |                                                                                                              Attalea        Astrocaryum
+              |                                       Species ranked by mass                                                     Sites by Species
+blank         | 
+              | 
+text          |                           FIGURE 4.6. Identity of species removing seeds during first week of
+              |                           experiment. (Left) On BCI five species ranging in mass from 1.95 g
+              |                           (Oneocarpus) to 64.2 g (Tontelea) were primarily removed by
+              |                           agoutis, and showed no relationship between seed size and seed
+              |                           predator body size. (Right) Seed removal of large-seeded palms at
+              |                           BCI and PNS. Removal of Attelea was higher in PNS, but
+              |                           Astrocaryum removal did not differ between sites. Seed predation
+              |                           by small mammals was higher in PNS for Attalea, and higher in BCI
+              |                           for Astrocaryum.
+blank         | 
+text          |                          A comparison of palm seed dispersal in BCI and PNS revealed that the seed
+              | removal of Attelea in PNS was more than 4-fold higher in the hunted site, contrary to
+              | expectation. Verified Attelea predation by D. punctata was similar in both sites, while
+              | seed predation by small mammals, primarily squirrels, was much higher in the hunted
+              | site. Astrocaryum removal in the first week was 100% in both sites, with confirmed
+              | removals by D. punctata being slightly higher in the hunted site, and removals by P.
+              | steerei being slightly higher in the protected site (Fig. 4.6).
+meta          |                                                                                      81
+blank         | 
+              | 
+              | 
+title         | DISCUSSION
+blank         | 
+text          | Hunting is a pervasive threat to biodiversity in tropical forests. Decreases in plant
+              | diversity as high as 66% have been documented as an indirect consequence of hunting.
+              | Yet the mechanisms by which that diversity is lost are poorly understood, as is their
+              | context dependence. This study aimed to test some basic assumptions about plant-
+              | seed predator interactions in the context of seed size, as well as hypotheses as to how
+              | those interactions may change as a consequence of defaunation. This information is
+              | useful for understanding how vertebrate seed predators shape community level
+              | distributions of seed size, and to predict how and why plant community composition
+              | may change with increasing hunting intensity.
+              |        This discussion focuses primarily on plant-seed predator interactions.
+              | However, as several of the animals relevant to this discussion are also recognized to be
+              | secondary seed dispersers, much of the discussion which follows is equally relevant to
+              | seed dispersal.
+blank         | 
+              | 
+text          | Seed size and vertebrate seed predation intensity. Overall, this study found little
+              | support for the model assumption that vertebrate seed predation rates are correlated
+              | with seed mass, either positively in moderately defaunated sites, or negatively in
+              | highly defaunated sites. Other seed traits, as well as natural history, may be more
+              | important for determining seed predation rates than seed mass per se. For instance, of
+              | the three of largest seed-species in this study, the Astrocaryum and Attelea exhibited
+              | the highest rates of seed removal, whereas Gustavia exhibited one of the lowest.
+              | Relative to Gustavia, the palm seeds have an endosperm that is high in fat content, and
+              | therefore energy rich for their weight. The rodents are well adapted to the primary
+              | defense of these palms against seed predators: a woody endocarp. While the thin seed
+              | coat of Gustavia would appear to make it more vulnerable to vertebrate seed
+              | predators, it may exhibit chemical defenses that make them less desirable to
+              | rodents(e.g. Forget 1992). Indeed, it was not uncommon on BCI and PNS to see bits
+              | of Gustavia pulp with a pile of intact seeds of the forest floor, the result of agoutis
+              | consuming the fruit flesh and discarding the seeds (E. Kurten, pers. obs.). However,
+meta          | 82       CHAPTER 4: SEED PREDATION RATES AS A FUNCTION OF SEED SIZE AND DEFAUNATION
+blank         | 
+              | 
+              | 
+text          | there are circumstances in which Gustavia seed predation has been higher (Sork 1987,
+              | Forget 1992).
+              |         The natural history of the two palm species also make them preferable for
+              | hoarding.    A. standleyanum and A. butyracea germinate after 8 and 12 months
+              | respectively. Therefore, they are likely to persist intact as hoarded seeds until the
+              | period of low fruit availability from November through February (Foster 1982). In
+              | contrast, Gustavia germinates within 1-2 weeks of fruit fall (E. Kurten, pers. obs.).
+              | While the seeds remain attached to the establishing seedling for many more months,
+              | seeds and cotyledons are vulnerable to consumption by other seedling predators, and
+              | seed energy stores become reduced over time. By November, Gustavia seedlings are
+              | no longer of interest to agoutis on BCI (Forget 1992). Therefore, in contrast to the
+              | palms, Gustavia is not a species useful for provisioning rodents through the period of
+              | low fruit availability.
+              |         Comparison of the large-seeded palm species and Gustavia illustrates how
+              | variability in seed consumption by vertebrates among seeds of similar size is likely to
+              | be introduced by a variety of differences in nutritive value, defense traits, and
+              | germination strategy. Variation in these traits will likely cause deviations in
+              | consumption rates from what may be predicted by seed size and optimal foraging
+              | theory alone.
+blank         | 
+              | 
+text          | Seed predator identity. Camera-based observations for five species preyed upon by
+              | vertebrates showed little support for the model assumptions that the seed predator
+              | body size is correlated with the size of the seed. While enclosure and exclosure
+              | experiments have demonstrated that small rodents prefer small seeded species to
+              | taxonomically related larger seeds, and can exert a large, and disproportionate seed
+              | predation pressure on smaller seeded species (Dirzo et al. 2007, Mendoza & Dirzo
+              | 2007), this study suggests that when seeds are available to the entire community of
+              | potential seed predators, the importance of smaller mammals, even for relatively
+              | smaller seeds, is diminished. In particular, agoutis appear to be a highly important
+              | seed predator and secondary seed disperser in this system.
+meta          |                                                                                       83
+blank         | 
+              | 
+              | 
+text          |         There may be other circumstances, however, in which the relationship between
+              | seed size and seed predator biomass may be stronger. It is notable that, despite the
+              | fact that peccaries are well-documented seed predators of palms such as Astrocaryum
+              | and Attelea in many parts of the Neotropics (Beck 2006), no observations of predation
+              | of these two species by peccaries were observed in this study. This is likely due to the
+              | fact that larger-bodied seed predators such as peccaries and brocket deer tend to
+              | consume fruits and seeds under fruiting trees, where fruit and seed densities are
+              | highest and foraging most efficient. In contrast, the seed arrays in this experiment best
+              | mimicked the situation in which a few seeds have already been dispersed away from
+              | the parent tree. Rodents such as agoutis and squirrels, will also forage under fruiting
+              | trees. However, as scatter-hoarding species, they are behaviorally more likely than
+              | larger seed predators to search for, find and consume small, isolated patches of seeds
+              | on the forest floor.
+              |         In addition to dispersal context, the composition of the mammal community is
+              | likely influential in determining the strength of any relationship between seed size and
+              | seed disperser size. For example, in forests such as Cocha Cashu, Peru, agoutis are
+              | much less abundant (Terborgh & Wright 1994), and therefore less likely to
+              | demonstrate a high importance as seed predators and seed dispersers across a range of
+              | species, as seen on BCI. BCI is also lacking an abundance of terrestrial vertebrates at
+              | the two extremes of seed predator body size. White-lipped peccaries (T. pecari) have
+              | been locally extinct for more than half a century on BCI, and tapirs are rare. At the
+              | other extreme, in parts of Mexico and Belize, mice such as Peromyscus and
+              | Heteromys can be significant seed predators of some species (Coates-Estrada &
+              | Estrada 1988, Klinger and Rejmanek 2009), whereas rodents smaller than spiny rats
+              | (P. steerii) are rare on BCI.
+blank         | 
+              | 
+text          | Seed predation and defaunation. Despite large differences in abundances of key seed
+              | predators such as collared peccaries and agoutis, decreased vertebrate seed predation
+              | rates of larger-seeded species were not generally observed between BCI and PNS, nor
+              | did smaller-seeded species generally show increases in seed predation rates in the
+meta          | 84       CHAPTER 4: SEED PREDATION RATES AS A FUNCTION OF SEED SIZE AND DEFAUNATION
+blank         | 
+              | 
+              | 
+text          | hunted site relative to the protected site. Rates of seed caching of large-seeded palms,
+              | critical for seedling recruitment in those species, also did not differ between sites.
+              | These data corroborate the work of Wright and colleagues (2001) with respect to
+              | large-seeded palms, and extend it to a broader spectrum of species.
+              |        Rather than seed predation, and secondary seed dispersal, varying linearly with
+              | agouti abundance in this system, it may be useful to consider the possibility of seed
+              | predator thresholds. In particular, lower populations of agoutis as a consequence of
+              | hunting may serve to decrease competition for fruit and seed resources among agoutis
+              | locally, but within certain limits, may not interfere with seed dispersal and seed
+              | predation rates from the perspective of the plant. However, if hunting reduces agouti
+              | populations below a particular threshold in abundance, there are no longer enough
+              | animals to adequately maintain levels of seed predation and dispersal.
+              |        Conceptualizing the disruption of plant-seed predator interactions in terms of
+              | thresholds is similar to the model proposed by Galetti and colleagues (2006), in which
+              | Astrocaryum seed predation rates are a sigmoidal or logistic, rather than linear,
+              | function of agouti abundance. Data from two defaunation gradients suggest this may
+              | be the case (Fig. 4.7). In Panama, sites with agouti densities 60%-80% lower than
+              | BCI still had high rodent seed predation rates equivalent to or higher than those
+              | observed on BCI, but where agouti densities were further reduced, seed predation rates
+              | dropped dramatically (Wright et al. 2001). Likewise in Brazil, sites with agouti
+              | abundances 80% lower than the maximum density observed still maintained agouti
+              | seed predation rates approximately equal to those observed in the sites with the highest
+              | agouti abundance (Donatti et al. 2009).
+meta          |                                                                                            85
+blank         | 
+              | 
+              | 
+text          |                                                    100   Above threshold
+blank         | 
+              | 
+              | 
+              | 
+text          |                        Rodent seed predation (%)
+              |                                                          Below threshold
+blank         | 
+text          |                                                    80
+blank         | 
+              | 
+text          |                                                     60
+blank         | 
+              | 
+text          |                                                     40
+blank         | 
+              | 
+text          |                                                     20
+blank         | 
+              | 
+text          |                                                      0
+              |                                                           A         B      C    D      E
+              |                                                                     Literature Study
+              |          FIGURE 4.7. Published studies examining seed predation of
+              |          large-seeded palms by rodents, primarily agoutis, across
+              |          defaunation gradients consistently find that high rates of seed
+              |          predation are maintained despite reductions in agouti densities up
+              |          to 80% of maximum observed densities. Below this threshold,
+              |          seed predation rates drop 48-96%. (A) Galetti et al. (2006),
+              |          Astrocaryum aculeatissimum, under parent trees; (B) Donatti et
+              |          al. (2009), A. aculeatissimum, controlled experiment; (C) Galetti
+              |          et al.(2006), ), A. aculeatissimum, controlled experiment; (D)
+              |          Wright et al. (2000), Astrocaryum standleyanum, “dispersed”
+              |          seeds; (E) Wright et al. (2000), Attalea butyracea, “dispersed”
+              |          seeds.
+              |        At a community level, evidence has been found both for large, mammal- or
+              | primate-dispersed species decreasing in abundance in association with hunting
+              | (Nunez-Iturri et al. 2008, Terborgh et al. 2008), and community mean seed mass
+              | increasing in association with hunting (Wright et al. 2007, Ch. 2). Differences in the
+              | mammalian community context, in the relative levels of defaunation being compared,
+              | and differences in the traits and natural histories of plant species driving the
+              | community level patterns may all be contributing to such discrepancies in community
+              | level response to hunting.
+meta          | 86      CHAPTER 4: SEED PREDATION RATES AS A FUNCTION OF SEED SIZE AND DEFAUNATION
+blank         | 
+              | 
+              | 
+title         | CONCLUSIONS
+blank         | 
+              | 
+text          | Overall, this empirical test of a theoretical model (Dirzo et al. 2007) of how seed
+              | predation rates should vary as a function of seed size and defaunation intensity found
+              | little general support for the model. In this system, there was only weak support for a
+              | relationship between seed mass and vertebrate seed predation rate, and no evidence
+              | that seeds were increasingly predated upon by larger-bodied seed predators as seed
+              | size increased by species. Furthermore, despite large differences in the abundances of
+              | key seed predators between a hunted and protected site, no differences were seen in
+              | vertebrate seed predation, or seed caching by agoutis.      This work highlights the
+              | importance of considering both plant natural history and mammal community context
+              | when trying to predict the indirect effects of hunting on plant recruitment. It also is
+              | consistent with accumulating evidence that the functional roles that seed predators and
+              | seed dispersers play in tropical forests may not be linearly correlated with their
+              | abundance in a system, but rather, may be threshold-dependent.
+blank         | 
+title         | ACKNOWLEDGEMENTS
+blank         | 
+text          | The author is grateful to the Autoridad Nacional del Ambiente (ANAM) of Panama
+              | for permission to conduct work in Parque Nacional Soberanía, and to the Smithsonian
+              | Tropical Research Institute (STRI) for permission to conduct this work on BCI, and
+              | for providing logistical support. I would like to thank J. Wright for seed mass data
+              | and advice about sites. N. Beckman also provided useful advice in the planning stages
+              | of this experiment. The Fondo Peregrino–Panamá, in particular A. Muela, provided
+              | logistical support at the site in the Parque Nacional Soberanía. Many thanks go to C.
+              | Sherman, R. Bethancourt and R. Acosta for their hard work in the field and the lab,
+              | and to S. Rebellon for help with pilot work. D. Brassfield and O. Calderón provided
+              | help with species identification. I would like to thank O. Arosemena at STRI for her
+              | help with permitting, and all the staff on BCI whose help made conducting this work
+              | easier. R. Dirzo and C. Donatti provided comments which improved the manuscript.
+              | This project was made possible with funding from NSF DEB-0808338.
+meta          |                                                                                       87
+blank         | 
+              | 
+              | 
+              | 
+title         | Bibliography
+ref           | Alroy, J. 2001. A musltispecies overkill simulation of the end-Pleistocene megafaunal
+              |        mass extinction. Science 292:1893-1896.
+              | Ackerly, D. D., C. A. Knight, S.B. Weiss, K. Barton, K.P. Starmer. 2002. Leaf size,
+              |        specific leaf area and microhabitat distribution of chaparral woody plants:
+              |        Contrasting patterns in species level and community level analyses. Oecologia
+              |        130(3): 449-457.
+              | Adler, G.H. 1996. The island syndrome in isolated populations of a tropical forest
+              |        rodent. Oecologia. 108:694-700.
+              | Alvarez-Clare S. & K. Kitajima 2007. Physical defence traits enhance seedling
+              |        survival of Neotropical tree species. Func. Ecol. 21: 1044-1054.
+              | Alvarez-Clare S. & K. Kitajima 2009.
+              | Alves-Costa, C.P. 2004. Efeitos da defaunação de mamíferos herbívoros na
+              |        comunidade vegetal. Dissertation thesis. Instituto de Biologia, Universidade
+              |        Estadual de Campinas, Brazil.
+              | Andresen, E. 2003. Effect of forest fragmentation on dung beetle communities and
+              |        functional consequences for plant regeneration. Ecography. 26: 87–97.
+              | Andresen, E. and S.G.W. Laurance. 2003. Possible Indirect Effects of Mammal
+              |        Hunting on Dung Beetle Assemblages in Panama. Biotropica. 39(1): 141–146.
+meta          | 88      CHAPTER 4: SEED PREDATION RATES AS A FUNCTION OF SEED SIZE AND DEFAUNATION
+blank         | 
+              | 
+              | 
+ref           | Asquith, N. M. Wright, S.J. and M.J. Clauss. 1997. Does mammal community
+              |        composition control recruitment in neotropical forests? Evidence from
+              |        Panama. Ecology. 78(3): 941-946.
+              | Asquith N.M., J. Terborgh, A.E. Arnold, and C.M. Riveros. 1999. The fruits the
+              |        agouti ate: Hymenaea courbaril seed fate when its disperser is absent. J. Trop.
+              |        Ecol. 15: 229-235.
+              | Babweteera F. P. Savill and N. Brown. 2007. Balanites wilsoniana: Regeneration with
+              |        and without elephants. Biol. Cons. 134: 40-47.
+              | Babweteera F. and N. Brown. 2009. Can remnant frugivore species effectively
+              |        disperse tree seeds in secondary tropical rain forests? Biodivers Conserv. 18:
+              |        1611–1627.
+              | Barnosky, A. D., P. L. Koch, R. S. Feranec, S. L. Wing, and A. B. Shabel. 2004.
+              |        Assessing the causes of Late Pleistocene extinctions on the continents. Science
+              |        306: 70-75.
+              | Barone, J.A. and P.D. Coley. 2002. Herbivorismo y las defensas de las plantas. Pages
+              |        465-492.in M.R. Guariguata and G.H. Kattan, editors. Ecología y conservación
+              |        de Bosques Neotropicales. Libro Universitario Regional, Cartago, Costa Rica.
+              | Beck H. and J. Terborgh. 2005. Groves versus isolates: How spatial aggregation of
+              |        Astrocaryum murumuru palms affects seed removal. J. Trop. Ecol. 18(2): 275-
+              |        288.
+              | Beck H. 2006. A review of peccary-palm interactions and their ecological
+              |        ramifications across the Neotropics. J. Mam., 87(3): 519–530.
+              | Beckman, N.G. and H. Muller-Landau. 2007. Differential effects of hunting on pre-
+              |        dispersal seed predation and primary and secondary seed removal of two
+              |        Neotropical tree species. Biotropica 39(3): 328-339.
+              | Brodie J.F., O.E. Helmy, W.Y. Brockelman, and J.L. Maron. 2009. Bushmeat
+              |        poaching reduces the seed dispersal and population growth rate of a mammal-
+              |        dispersed tree. Ecol. App. 19(4): 854–863.
+              | Brodie, J.F. and H. Gibbs. 2009. Bushmeat hunting as climate threat. Science. 326:
+              |        364-365.
+meta          |                                                                                       89
+blank         | 
+              | 
+              | 
+ref           | Bunker, D.E. and Carson, W.P. 2005. Drought stress and tropical forests woody
+              |        seedlings: effect on community structure and composition. J. Trop. Ecol. 93:
+              |        794-806.
+              | Chaneton E.J., J.M. Facelli and R.J.C. Leon.1988. Floristic changes induced by
+              |        flooding on grazed and ungrazed lowland grasslands in Argentina. J. Range
+              |        Manag. 41(6); 495-499
+              | Chapman C.A. & D.A. Onderdonk. 1998. Forests Without Primates: Primate/Plant
+              |        Codependency. Am. J. Primatol. 45:127–141.
+              | Cingolani, A. M., G. Posse, and M.B. Collantes. 2005. Plant functional traits,
+              |        herbivore selectivity and response to sheep grazing in Patagonian steppe
+              |        grasslands. J. Ecol. 42(1): 50-59.
+              | Coley, P. D. and J. A. Barone. 1996. Herbivory and plant defenses in tropical forests.
+              |        Ann. Rev. Ecol. Sys. 27: 305-335.
+              | Condit, R., N. Pitman, E.G. Leigh Jr., J. Chave, J. Terborgh, R.B. Foster, P. Nuñez, S.
+              |        Aguilar, R. Valencia, G. Villa, H.C. Muller-Landau, E. Losos, S.P. Hubbell.
+              |        2002. Beta-diversity in tropical forest trees. Science. 295: 666-669.
+              | Connell, J. H. 1971. On the role of natural enemies in preventing competitive
+              |        exclusion in some marine animals and in rain forest trees. Pages 298-312 in
+              |        R.G. Den Boer, editor. Dynamics of numbers in populations. Proceedings of
+              |        the Advanced Study Institute on dynamics of numbers in populations,
+              |        Oosterbeck, 1971. Centre for Agricultural Publishing and Documentation,
+              |        Wageningen..
+              | Cordiero N. and H. Howe. 2003. Forest fragmentation severs mutualism between seed
+              |        dispersers and an endemic African tree. PNAS. 100(24): 14052–14056.
+              | Corlett, R.T. 2007. The impact of hunting on the mammalian fauna of tropical Asian
+              |        forests. Biotropica 39(3): 292-303.
+              | Cornelissen, J.H.C., S. Lavorel, E. Garnier, S. Diaz, N. Buchmann, D.E. Gurvich, P.B.
+              |        Reich, H. ter Steeg, H.D. Morgan, M.G.A. van der Heijden, J.G. Pausas, and
+              |        H. Poorter. 2003. A handbook of protocols for standardized and easy
+              |        measurement of plant functional traits worldwide. Aust. J. Bot. 51: 335-380.
+meta          | 90       CHAPTER 4: SEED PREDATION RATES AS A FUNCTION OF SEED SIZE AND DEFAUNATION
+blank         | 
+              | 
+              | 
+ref           | Cramer J.M, R.C.G. Mesquita, G.B. Williamson. 2007. Forest fragmentation
+              |        differentially affects seed dispersal of large and small-seeded tropical trees.
+              |        Biol. Cons. 1 3 7 : 4 1 5 –4 2 3.
+              | DeMattia, E.A., Curran, L.M. and B.J. Rathcke. 2004. Effects of small rodents and
+              |        large mammals on Neotropical seeds. Ecology. 85(8): 2161–2170.
+              | DeMattia, E.A. and B.J. Rathcke. 2006. Effects of Small Rodent and Large Mammal
+              |        Exclusion on Seedling Recruitment in Costa Rica. Biotropica. 38(2): 196–202.
+              | de Vivo, M. and A. P. Carmignotto. 2004. Holocene vegetation change and the
+              |        mammal faunas of South America and Africa. Journal of Biogeography
+              |        31:943-957.
+              | Diaz, S., I. Noy-Meir, M. Cabido. 2001. Can grazing response of herbaceous plants be
+              |        predicted from simple vegetative traits? J. App. Ecol. 38(3): 497-508.
+              | Di Bitetti M.S., A. Paviolo, C.A. Ferrari, C. De Angelo, and Y. Di Blanco. 2008.
+              |        Differential responses to hunting in two sympatric species of Brocket deer
+              |        (Mazama americana and M. nana). Biotropica 40(5): 636-645.
+              | Dirzo, R. and A. Miranda. 1991. Altered Patterns of herbivory and diversity in the
+              |        forest understory: A case study of the possible consequences of contemporary
+              |        defaunation. Pages 273-287 in P.W. Price, T.M. Lewinsohn, G.W. Fernandes,
+              |        & W.W. Benson, editors. Plant-Animal Interactions: Evolutionary Ecol. in
+              |        Tropical and Temperate Regions. John Wiley & Sons, Inc.
+              | Dirzo, R., E. Mendoza, and P. Ortiz. 2007. Size-related differential seed predation in a
+              |        heavily defaunated Neotropical rain forest. Biotropica. 39(3): 355-362.
+              | Donatti, C.I., P.R. Guimarães Jr., and M. Galetti. 2009. Seed dispersal and predation
+              |        in the endemic Atlantic rainforest palm Astrocaryum aculeatissimum across a
+              |        gradient of seed disperser abundance. Ecol. Res. DOI 10.1007/s11284-009-
+              |        0601-x.
+              | Erard, E., M. Théry, and D. Sabatier. 1991. Régime alimentaire de Tinamus major
+              |        (Tinamidae), Crax alector (Cracidae) et Psophia crepitans (Psophiidae), en
+              |        Forêt Guyanaise. Gibier Faune Sauvage 8: 183–210.
+meta          |                                                                                       91
+blank         | 
+              | 
+              | 
+ref           | Fa J.E. and D. Brown. 2009. Impacts of hunting on mammals in African tropical
+              |        moise forests: a review and synthesis. Mamm. Rev. 39(4): 321-264.
+              | Fa, J.E. and Peres C.A. 2001. Game vertebrate extraction in African and Neotropical
+              |        forests: A intercontinental comparison. In J.D. Reynolds, G.M. Mace, K.H.
+              |        Redford, and J.G. Robinson (Eds.) Conservation of Exploited Species.
+              |        Cambridge University Press, Cambridge, UK. Pages 203-241.
+              | Fadini, R.F., M. Fleury, C.I. Donatti, and M. Galetti. 2008. Effects of frugivore
+              |        impoverishment and seed predators on the recruitment of a keystone palm.
+              |        Acta Oecolo. 35(2): 188-196.
+              | Farwig, N. K. Böhning-Gaese, B. Bleher. 2006. Enhanced seed dispersal of Prunus
+              |        africana in fragmented and disturbed forests? Oecologia. 147: 238–252.
+              | Fleury, M. and M. Galetti. 2006. Forest fragment size and microhabitat effects on
+              |        palm seed predation. Biol. Cons. 131: 1-13.
+              | Forget, P.M. 1992. Seed removal and seed fate in Gustavia superb (Lecythidaceae).
+              |        Biotropica. 24(3): 408-414.
+              | Foster, R. B. 1982. Famine on Barro Colorado Island. Pages 201–212 in Leigh,
+              |        E.G.,Wilson, D. M. & Rand, A. S. (eds). The ecology of a tropical forest:
+              |        seasonal rhythms and long-term change. Smithsonian Institution Press,
+              |        Washington, DC.
+              | Galetti M., C.I., Donatti, C.I., A.S. Pires, P.R. Guimaraes Jr., R. Jordano. 2006. Seed
+              |        survival and dispersal of an endemic Atlantic forest palm: the combined effects
+              |        of defaunation and forest fragmentation. Bot. J. Linn. Soc. 151: 141–149.
+              | Garwood, N.C. 1983. Seed Germination in a Seasonal Tropical Forest in Panama: A
+              |        Community Study. Ecol. Monographs 53(2): 159-181.
+              | Gehring, C.A., J.E. Wolf and T.C. Theimer. 2002. Terrestrial vertebrates promote
+              |        arbuscular mycorrhizal fungal diversity and inoculums potential in a rain forest
+              |        soil. Ecol. Lett. 5: 540–548.
+              | Guariguata, M.R., J.J. Rosales-Adame and B. Finegan. 2000. Seed Removal and Fate
+              |        in Two Selectively Logged Lowland Forests with Constrasting Protection
+              |        Levels. Cons. Biol. 14(4): 1046–1054.
+meta          | 92        CHAPTER 4: SEED PREDATION RATES AS A FUNCTION OF SEED SIZE AND DEFAUNATION
+blank         | 
+              | 
+              | 
+ref           | Guariguata, M.R., H. Arias-Le Claire and G. Jones. 2002. Tree Seed Fate in a Logged
+              |         and Fragmented Forest Landscape, Northeastern Costa Rica. Biotropica.
+              |         34(3): 405–415.
+              | Guimaraes, P. R., Jr., M. Galetti, and P. Jordano. 2008. Seed Dispersal Anachronisms:
+              |         Rethinking the Fruits Extinct Megafauna Ate. PLoS ONE 3:e1745.
+              | Hardesty B.D. and H.C. Muller-Landau. 2005. Seed dispersal of woody plants in
+              |      tropical forests: concepts, examples, and future directions. Pages 267-309 in
+              |      D.F.R.P. Burslem, M. A. Pinard and S. E.Hartley, editors. Biotic interactions in
+              |      the Tropics: their role in the maintenance of species diversity. Cambridge Univ.
+              |      Press., New York, NY, U.S.A.
+              | Harms, K. E., S. J. Wright, O. Calderon, A. Hernandez, and E.A. Herre. 2000.
+              |         Pervasive density-dependent recruitment enhances seedling diversity in a
+              |         tropical forest. Nature 404(6777): 493-495.
+blank         | 
+ref           | Hedges L.V. and I. Olkin. (1985). Statistical Methods for Meta-Analysis. Academic
+              |         Press. San Diego, CA, USA.
+              | Holbrook, K.M. and B.A. Loiselle. 2009. Dispersal in a Neotropical tree, Virola
+              |         flexuosa (Myristicaceae): Does hunting of large vertebrates limit seed
+              |         removal? Ecology. 90(6): 1449–1455.
+              | Ickes, K., S. J. Dewalt, S. Appanah. 2001. Effects of native pigs (Sus scrofa) on
+              |         woody understorey vegetation in a Malaysian lowland rain forest. J. Trop.
+              |         Ecol. 17(2): 191-206.
+              | Janzen, D. H. 1970. Herbivores and the number of tree species in tropical forests. Am.
+              |         Nat. 104(940): 501-528.
+              | Johnson, C. N. 2009. Ecological consequences of Late Quaternary extinctions of
+              |         megafauna. Proceedings Of The Royal Society B-Biological Sciences
+              |         276:2509-2519.
+              | Kankam B.O. and W. Oduro . (2009). Frugivores and fruit removal of Antiaris
+              |         toxicaria (Moraceae) at Bia Biosphere Reserve, Ghana. J Trop. Ecol. 25: 201-
+              |         204
+meta          |                                                                                        93
+blank         | 
+              | 
+              | 
+ref           | Karban, R. and I.T. Baldwin. 1997. Induced responses to herbivory,: University of
+              |        Chicago Press, London, England.
+              | Kira T. and H. Ogawa. 1971Assessment of primary production in tropical and
+              |        equatorial forests. Pages 309–321 in P. Duvigneaud, editor. Productivity of
+              |        Forest Ecosystems, UNESCO, Paris.
+              | Kirika J.M., N. Farwig, and K. Böhning-Gaese. 2008. Effects of Local Disturbance of
+              |        Tropical Forests on Frugivores and Seed Removal of a Small-Seeded
+              |        Afrotropical Tree. Cons. Biol. 22(2): 318–328.
+              | Kitajima, K. 2003. Impact of cotyledon and leaf removal on seedling survival in three
+              |        tree species with contrasting cotyledon functions. Biotropica 35(3): 429-434.
+              | Koch, P. L. and A. D. Barnosky. 2006. Late quaternary extinctions: State of the
+              |        debate. Annual Review of Ecology Evolution and Systematics 37:215-250.
+              | Leigh, E.G. Jr. G., Rand, and D.M. Winsor (Eds.) The Ecology of a Tropical Forest:
+              |        Seasonal Rhythms and Long-Term Changes, 2nd Ed. Smithsonian Institution
+              |        Press, Washington, D.C.
+              | Lizcano, D.J. 2006. Ecology and conservation of large mammals in the Northern
+              |        Andes. Dissertation. Durrell Institute of Conservation and Ecology,
+              |        University of Kent, U.K.
+              | Lucas P.W., I.M. Turner, N.J. Dominy and N. Yamashita. 2000. Mechanical defenses
+              |    to herbivory. Ann. Bot. 86: 913-920.
+              | Mayfield, M.M., Boni, M.E., Daily G.C. and Ackerly, D.D. 2005. Species and
+              |        functional diversity of native and human-dominated plant communities.
+              |        Ecology 86(9): 2365-2372.
+              | McConkey, K.M. and D.R. Drake. 2006. Flying foxes cease to function as seed
+              |        dispersers long before they become rare. Ecology. 87(2): 271-286.
+              | McNaughton, S. J. 1985. Ecology of a grazing ecosystem: The Serengeti. Ecological
+              |        Monographs 55(3): 259-294.
+              | Mendoza E. and R. Dirzo. 2007. Seed-size variation determines interspecific
+              |        differential predation by mammals in a neotropical rain forest. Oikos. 116(11):
+              |        1841-1852.
+meta          | 94       CHAPTER 4: SEED PREDATION RATES AS A FUNCTION OF SEED SIZE AND DEFAUNATION
+blank         | 
+              | 
+              | 
+ref           | Moran C., C.P. Catterall and J. Kanowski. 2009. Reduced dispersal of native plant
+              |        species as a consequence of the reduced abundance of frugivore species in
+              |        fragmented rainforest. Biol. Cons. 142: 541-552.
+              | Myers, J.A. and K. Kitajima. 2007. Carbohydrate storage enhances seedling shade
+              |        and stress tolerance in a neotropical forest. J. Ecol. 95: 383-395.
+              | Nabeshima, E., M. Murakami, T. Hiura. 2001. Effects of herbivory and light
+              |        conditions on induced defense in Quercus crispula. Journal of Plant Research
+              |        114(1116): 403-409.
+              | Nuñez-Iturri, G. and H.F. Howe. 2007. Bushmeat and the fate of trees with seeds
+              |        dispersed by large primates in a lowland rain forest in Western Amazon.
+              |        Biotropica 39(3): 348-354.
+              | Nuñez-Iturri, G., O. Olsson, H.F. Howe. 2008. Hunting reduces recruitment of
+              |        primate-dispersed trees in Amazonian Peru. Biol. Cons. 141: 1536-1546.
+              | Oleksyn, J., P. Karolewski, M.J. Giertych, R. Zytkowiak, P.B. Reich, M.G. Tjoelker.
+              |        1998. Primary and secondary host plants differ in leaf-level photosynthetic
+              |        response to herbivory: Evidence from Alnus and Betula grazed by the alder
+              |        beetle, Agelastica alni. New Phytologist 140(2): 239-249.
+              | Osunkoya, O.O., J.E. Ash, A.W. Graham, and M.S. Hopkins. 1993. Growth of tree
+              |        seedlings in tropical rain forests of North Queensland, Australia. J. Trop. Ecol.
+              |        9(1): 1-18.
+              | Olff, H. and M. E. Ritchie. 1998. Effects of herbivores on grassland plant diversity.
+              |        Trends Ecol. Evol. 13(7): 261-265.
+              | Pacheco L.F. and J.A. Simonetti. 2000. Genetic Structure of a Mimosoid Tree
+              |        Deprived of Its Seed Disperser, the Spider Monkey. Cons. Biol. 14(6): 1766-
+              |        1755.
+              | Pacheco, M.A.W. 2001. Effects of flooding and herbivores on variation in recruitment
+              |        of palms between habitats. J. Ecol. 89:, 358–366.
+              | Parker, J.D., D.E. Burkepile, and M.E.Hay. 2006. Opposing effects of native and
+              |        exotic herbivores on plant invasions. Science. 311(5766): 1459-1461.
+meta          |                                                                                          95
+blank         | 
+              | 
+              | 
+ref           | Peres, C. A., and M. van Roosmalen. 2002. Patterns of primate frugivory in Amazonia
+              |        and the Guianan shield: Implications to the demography of large-seeded plants
+              |        in overhunted tropical forests. Pages 407–423 in D. Levey,W. Silva, and M.
+              |        Galetti (Eds.). Seed dispersal and frugivory: Ecology, evolution and
+              |        conservation, CAB International, Oxford, UK.
+              | Peres C.A. and E. Palacios. 2007. Basin-wide effects of game harvest on vertebrate
+              |        population densities in Amazonian Forests: Implications for animal-mediated
+              |        seed dispersal. Biotropica 39(3): 304-315.
+              | Pimm, S., P. Raven, A. Peterson, C. H. Sekercioglu, and P. R. Ehrlich. 2006. Human
+              |        impacts on the rates of recent, present, and future bird extinctions. Proc Natl
+              |        Acad Sci U S A 103:10941-10946.
+              | Pizo, M.A. and Vieira, E.M. 2004. Granivorous birds as potentially important post-
+              |        dispersal seed predators in a Brazilian forest fragment. Biotropica 36(3):
+              |        417:423.
+              | Poulsen, J. R., C. J. Clark, E. F. Connor, and T. B. Smith. 2002. Differential resource
+              |        use by primates and hornbills: Implications for seed dispersal. Ecology 83:
+              |        228–240.
+              | Rodriguez, C., E. Leoni, F. Lezama, and A. Altesor. 2003. Temporal trends in species
+              |        composition and plant traits in natural grasslands of Uruguay. J. of Veg. Sci.
+              |        14(3): 433-440.
+              | R Development Core Team. 2008. R: A language and environment for statistical
+              |        computing. R Foundation for Statistical Computing, Vienna, Austria. ISBN 3-
+              |        900051-07-0, URL http://www.R-project.org.
+              | Retuerto R., B. Fernandez-Lema, R. Rioloa, and J.R. Obeso. 2004. Increased
+              |        photosynthetic performance in holly trees infested by scale insects. Func.
+              |        Ecol. 18: 664-669.
+              | Roldan, A.I. and J.A. Simonetti. 2001. Plant-mammal interactions in tropical Bolivian
+              |        forests with different hunting pressures. Cons. Bio. 15(3): 617-623.
+meta          | 96      CHAPTER 4: SEED PREDATION RATES AS A FUNCTION OF SEED SIZE AND DEFAUNATION
+blank         | 
+              | 
+              | 
+ref           | Royo, A. A. and W. P. Carson. 2005. The herb community of a tropical forest in
+              |        central Panama: dynamics and impact of mammalian herbivores. Oecologia
+              |        145(1): 66-75.
+              | Sagers C.L. and P.D. Coley. 1995. Benefits and costs of defense in a Neotropical
+              |        shrub. Ecology. 76(6) 1835-1843.
+              | Sethi P. and H.F. Howe. 2009. Recruitment of Hornbill-Dispersed Trees in Hunted
+              |        and Logged Forests of the Indian Eastern Himalaya. Cons. Biol. 23(3): 710-
+              |        718.
+              | Shimazaki, A. and T. Miyashita. 2002. Deer browsing reduces leaf damage by
+              |        herbivorous insects through an induced response of the host plant. Ecol. Res.
+              |        17(5): 527-533
+              | Schipper, J. and J. S. Chanson and F. Chiozza and N. A. Cox and M. Hoffmann and V.
+              |        Katariya and J. Lamoreux and A. S. L. Rodrigues and S. N. Stuart and H. J.
+              |        Temple and J. Baillie and L. Boitani and T. E. Lacher and R. A. Mittermeier
+              |        and A. T. Smith and D. Absolon and J. M. Aguiar and G. Amori and N.
+              |        Bakkour and R. Baldi and R. J. Berridge and J. Bielby and P. A. Black and J. J.
+              |        Blanc and T. M. Brooks and J. A. Burton and T. M. Butynski and G. Catullo
+              |        and R. Chapman and Z. Cokeliss and B. Collen and J. Conroy and J. G. Cooke
+              |        and G. A. B. da Fonseca and A. E. Derocher and H. T. Dublin and J. W.
+              |        Duckworth and L. Emmons and R. H. Emslie and M. Festa-Bianchet and M.
+              |        Foster and S. Foster and D. L. Garshelis and C. Gates and M. Gimenez-Dixon
+              |        and S. Gonzalez and J. F. Gonzalez-Maya and T. C. Good and G. Hammerson
+              |        and P. S. Hammond and D. Happold and M. Happold and J. Hare and R. B.
+              |        Harris and C. E. Hawkins and M. Haywood and L. R. Heaney and S. Hedges
+              |        and K. M. Helgen and C. Hilton-Taylor and S. A. Hussain and N. Ishii and T.
+              |        A. Jefferson and R. K. B. Jenkins and C. H. Johnston and M. Keith and J.
+              |        Kingdon and D. H. Knox and K. M. Kovacs and P. Langhammer and K. Leus
+              |        and R. Lewison and G. Lichtenstein and L. F. Lowry and Z. Macavoy and G.
+              |        M. Mace and D. P. Mallon and M. Masi and M. W. McKnight and R. A.
+              |        Medellin and P. Medici and G. Mills and P. D. Moehlman and S. Molur and A.
+meta          |                                                                                       97
+blank         | 
+              | 
+              | 
+ref           |        Mora and K. Nowell and J. F. Oates and W. Olech and W. R. L. Oliver and M.
+              |        Oprea and B. D. Patterson and W. F. Perrin and B. A. Polidoro and C. Pollock
+              |        and A. Powel and Y. Protas and P. Racey and J. Ragle and P. Ramani and G.
+              |        Rathbun and R. R. Reeves and S. B. Reilly and J. E. Reynolds and C.
+              |        Rondinini and R. G. Rosell-Ambal and M. Rulli and A. B. Rylands and S.
+              |        Savini and C. J. Schank and W. Sechrest and C. Self-Sullivan and A.
+              |        Shoemaker and C. Sillero-Zubiri and N. De Silva and D. E. Smith and C.
+              |        Srinivasulu and P. J. Stephenson and N. van Strien and B. K. Talukdar and B.
+              |        L. Taylor and R. Timmins and D. G. Tirira and M. F. Tognelli and K.
+              |        Tsytsulina and L. M. Veiga and J. C. Vie and E. A. Williamson and S. A.
+              |        Wyatt and Y. Xie and B. E. Young. 2008. The status of the world's land and
+              |        marine mammals: Diversity, threat, and knowledge. Science 322:225-230.
+              | Sork, V.L. 1987. Effects of predation and light on seedling establishment in Gustavia
+              |        superba. Ecology 68(5):1341-1350.
+              | Strauss, S. Y. and A. A. Agrawal. 1999. The ecology and evolution of plant tolerance
+              |        to herbivory. Trends Ecol. Evol. 14(5): 179-185.
+              | Stoner, K.E., K. Vulinec, S.J. Wright, and C.A. Peres. 2007. Hunting and plant
+              |        community dynamics in tropical forests: A synthesis and future directions.
+              |        Biotropica 39(3): 385–392.
+              | Terborgh, J. and S.J. Wright. 1994. Effects of mammalian herbivores on plant
+              |        recruitment in two Neotropical forests. Ecology 75(6): 1829-1833.
+              | Terborgh J., K. Feeley, M. Silman, P. Nuñez, B. Balukjian. 2006. Vegetation
+              |        dynamics of predator-free land-bridge islands. J. Ecol. 94: 253–263.
+              | Terborgh, J., G. Nuñez-Iturri, N.C.A. Pitman, F.H. Cornejo Valverde, P. Alvarez, V.
+              |        Swamy, E.G. Pringle, and C.E.T. Paine. 2008. Tree recruitment in an empty
+              |        forest. Ecology. 89(6):1757–1768.
+              | Wang, B. C., V.L. Sork, M.T. Leong, T.B. Smith. 2007. Hunting of mammals reduces
+              |        seed removal and dispersal of the Afrotropical tree Antrocaryon klaineanum
+              |        (Anacardiaceae). Biotropica 39(3): 340-347.
+meta          | 98      CHAPTER 4: SEED PREDATION RATES AS A FUNCTION OF SEED SIZE AND DEFAUNATION
+blank         | 
+              | 
+              | 
+ref           | Ward, D. and T. P. Young. 2002. Effects of large mammalian herbivores and ant
+              |        symbionts on condensed tannins of Acacia drepanolobium in Kenya. J. Chem.
+              |        Ecol. 28(5): 921-937.
+              | Wardle, D. A., K. I. Bonner, and G.M. Barker. 2002. Linkages between plant litter
+              |        decomposition, litter quality, and vegetation responses to herbivores. Func.
+              |        Ecol. 16(5): 585-595.
+              | Webb, C. O. and D. R. Peart. 1999. Seedling density dependence promotes
+              |        coexistence of Bornean rain forest trees. Ecology 80(6): 2006-2017.
+              | Webb, S. 2008. Megafauna demography and late Quaternary climatic change in
+              |        Australia: A predisposition to extinction. Boreas 37:329-345.
+              | Weiher, E. and P. Keddy. 1995. Assembly rules, null models, and trait dispersion:
+              |        New questions from old patterns. Oikos 74(1): 159-164.
+              | Wilson, D.E., F.R. Cole, J.D. Nichols, R. Rudran, and M.S. Foster. 1996. Measuring
+              |        and Monitoring Biological Diversity: Standard Methods for Mammals.
+              |        Smithsonian Institution Press, Washington, D.C.
+              | Wright, I.J, D.D. Ackerly, F. Bongers, K.E. Harms, G. Ibarra-Manriquez, M. Matinez-
+              |        Ramos, S.J. Mazer, H.C. Muller-Landau, H. Paz, N.C. Pitman, L. Poorter,
+              |        M.R. Silman, C.F. Vriesendorp, C.O. Webb, M. Westoby, S.J. Wright. 2007.
+              |        Relationships among ecologically important dimensions of plant trait variation
+              |        in seven Neotropical forests. Ann. Bot. 99: 1003-1015.
+              | Wright, S.J., H. Zeballos, I. Dominguez, M.M. Gallardo, M.C. Moreno and R. Ibanez.
+              |        2000. Poachers alter mammal abundance, seed dispersal, and seed predation in
+              |        a Neotropical forest. Cons. Bio. 14(1): 227-239.
+              | Wright, S.J and H.C. Duber. 2001. Poachers and forest fragmentation alter seed
+              |        dispersal, seedling survival, and seedling recruitment in the palm Attalea
+              |        butyracea, with implications for tropical tree diversity. Biotropica 33(4):
+              |        583:595.
+              | Wright, S.J. 2003. The myriad consequences of hunting for vertebrates and plants in
+              |        tropical forests. Pers. in Plant Ecol. Evol. Sys. 6(1-2): 78-86.
+meta          |                                                                                        99
+blank         | 
+              | 
+              | 
+ref           | Wright, S.J., A. Hernandéz, and R.Condit. 2007. The bush meat harvest alters seedling
+              |        banks by favoring lianas, large seeds and seeds dispersed by bats, birds and
+              |        wind. Biotropica 39(3): 363-371.
+              | Wright, S.J., K. Kitajima, N.J.B. Kraft, P.B. Reich4, I.J. Wright, D.E. Bunker, R.
+              |    Condit, J.W. Dalling, S.J. Davies, S. Díaz, B.M.J. Engelbrecht, K.E. Harms, S.P.
+              |    Hubbell, C.O. Marks, Maria C. Ruiz-Jaen, C.M. Salvador and A.E. Zanne. 2010.
+              |    Functional traits and the growth-mortality tradeoff in tropical trees. In press.
+              |    Ecology. [doi:10.1890/09-2335.1]
+              | Young, T. P. and B. D. Okello. 1998. Relaxation of an induced defense after exclusion
+              |        of herbivores: Spines on Acacia drepanolobium. Oecologia 115(4): 508-513.
+              | Zimov, S. A., V. I. Chuprynin, A. P. Oreshko, F. S. Chapin, J. F. Reynolds, and M. C.
+              |        Chapin. 1995. Steppe-tundra transition: a herbivore-driven biome shift at the
+              |        end of the Pleistocene. American Naturalist 146:765-794.
+blank         |