Non-Random Seed Dispersal by Lemur Frugivores

RICE UNIVERSITY

Nonrandom Seed Dispersal by Lemur Frugivores: Mechanism, Patterns and Impacts

Onja Harinala F.E. Razafindratsima

A THESIS SUBMITTED IN PARTIAL FULFILLMENT OF THE REQUIREMENTS FOR THE DEGREE

Doctor of Philosophy

APPROVED, THESIS COMMITTEE:

ArnyU! Dunham, Chair Assistant Professor of BioScienees

Volker H. Associate Professor of BioScienees

Evan Siemann Harry C. and Olga K. Wiess Professor of BioScienees

Caroline A. Masiello Associate Professor of Earth Science_ Elizabeth M. Erhart Department Chair and Associate Professor of Anthropology TSU San Marcos

HOUSTON, TEXAS March 2015

ABSTRACT

Nonrandom Seed Dispersal by Lemur Frugivores: Mechanism, Patterns and Impacts

by Onja Harinala F.E. Razafindratsima

Frugivores act as seed-dispersal agents in many ecosystems. Thus, understanding the roles and impacts of frugivore-mediated seed dispersal on plant spatial structure and plant-plant associations is important to understand the structure of plant communities.

Frugivore-mediated seed dispersal is behaviorally driven, generating nonrandom patterns of seed dispersion; but, we know relatively little about how this might affect plant populations or communities. I examined how frugivores affected plants from the individual level to the population and community levels. I used modeling, trait-based and phylogenetic approaches combined with field observations and experiments, focusing on seed dispersal by three frugivorous lemur species in the biodiverse rainforest of

Ranomafana National Park, Madagascar. An analysis of traits suggested that 84% of trees in Ranomafana are adapted for animal dispersal, of which 74% are dispersed by these three lemur species, indicating their role as generalist dispersers. The distribution of fruit and seed size of bird-dispersed species was nested within the wide spectrum of size distribution associated with lemur dispersal. Nonrandom seed dispersal by these three frugivores increased per capita recruitment of the seeds of a long-lived canopy tree,

Cryptocarya crassifolia, by four-fold compared with no dispersal, even though it was not an overall advantage compared to random dispersal. The three frugivores not only

iii dispersed seeds away from parent and conspecific adult-trees, but also biased seed dispersal toward certain microhabitats. By using fruiting trees as seed dispersal foci, lemur frugivores structure the spatial associations between dispersed seeds and adult- trees nonrandomly, in terms of fruiting time, dispersal mode and phylogenetic relatedness. However, lemur-dispersed tree species were not more likely to be each other’s neighbors as adults. Interestingly, co-fruiting neighboring trees sharing lemur dispersers were more phylogenetically distant than expected by chance despite a phylogenetic signal in lemur dispersal mode, although there was no phylogenetic signal in the timing of fruiting among lemur-dispersed tree species. Results suggest an important link between frugivore foraging behavior and the spatial, temporal and phylogenetic patterning of seed dispersal. The important role of frugivores in structuring seed dispersion and seed-adult plant associations has critical implications for plant-plant interactions, biodiversity patterns and community structure. iv

Acknowledgments

First of all, I would like to express my gratitude to my advisor, Amy Dunham, for being an excellent mentor. Her wisdom, intelligence, guidance, support, encouragement, enthusiasm, and patience have tremendously helped me grow professionally and personally over the years. Thank you in particular for accepting me to work in your lab, letting me explore my passion for lemur ecology, pushing me to think outside the box and be better in everything I do, working closely with me to publish my research findings and to prepare proposals for grants and fellowships, providing me with an excellent atmosphere and guidance for doing research, always making the time for me, and for being my “hiking buddy” along the crazy path of Ranomafana terrain. I have learned a lot from you.

I am also thankful to my committee, Evan Siemann, Volker Rudolf, Elizabeth

Erhart and Caroline Masiello for their guidance and advice throughout my study, and for their substantial comments and insights on my field research plans and analyses. Thanks also to Thomas Jones for his help with analyzing lemur movement, and for discussions about statistics. I am also grateful to Jean Aroom for her helpful assistance with the use of GIS and associated programs. I have also benefited from advice and assistance from other faculty outside of Rice, including Kenneth Whitney, Clive Nuttman, Joerg

Ganzhorn, Andrea Baden, Jonah Ratsimbazafy and Kerry Brown.

I am grateful to Barbara Martinez, Nirilanto Ramamonjisoa, Haldre Rogers and

Patricia Wright for being such great role models and mentors. Your words of encouragements and wisdom, your friendship and your support have helped me a lot; in many ways, you have been great examples to me of successful women researchers.

My PhD studies and field research would not have been possible without the generous financial support of Philanthropic Educational Organization (International

Peace Scholarship), The Leakey Foundation (Franklin Mosher Baldwin Fellowship),

Schlumberger Foundation (Faculty For The Future fellowship), Garden Club of America

Award in Tropical Botany, International Foundation for Science (grant no. D/4985-1 and

D/4985-2), The Rufford Foundation (grant no. 5748-1, 11288-2, and 13874-B), Primate

Conservation Inc. (No. 908), The Explorers Club Exploration Fund, Idea Wild, Primate

Society of Great Britain, Centre ValBio, and Rice University including the Nettie S.

Autrey Fellowship, Wagoner Foreign Study Fellowship and Rice Centennial Award.

MICET (Madagascar Institut pour la Conservation des Ecosystèmes Tropicaux) and Centre ValBio were extremely helpful with logistics while I was in Antananarivo and in Ranomafana. I would like to thank all of their staff for making my field experience smooth and for going above and beyond in helping me and my research team. Special thanks to Eileen Larney, John Caddle, Prisca Andriambinintsoa, Pascal Rabeson, Desiré

Randrianarisata, Jean de Dieu Ramanantsoa, Paul Rakotonirina, Jean-Claude

Razafimahaimodison, Benjamin Andriamiahaja, Benjamin Randrianambinina, Jean

Rakotoarison, and Tiana Razafindratsita. I would also like to thank Madagascar National

Parks and the Malagasy Ministry of Forest and Water for research permission. Thanks also to the Missouri Botanical Garden in St-Louis for access to their herbarium collections, and to the “Parc Botanique et Zoologique de Tsimbazaza” in Antananarivo for help with plant species identifications. Thanks also to Coke Smith, Stacey Tecot, Noel

Rowe and Volker Rudolf for providing awesome lemur pictures. vi

I could not have completed my fieldwork without the support and assistance of many people, including field technicians, local villagers, and undergraduates. I am grateful to Jean-Claude Ramanandraibe, Nerée Beson, Paul Rasabo, Justin Solo, Georges

Razafindrakoto, Auguste Pela, Armand Razafitsiafajato, Jocelyn Mamiharilala, Victor

Rasendry, Marolahy, Leroa, Velomaro, and Letsara for their valuable assistance with fieldwork. Your amazing knowledge of the fauna and flora in Ranomafana has been extremely valuable for my research, and your help greatly improved my skills in botanical identifications and lemur observations. Thanks also for showing me all the wonders of Ranomafana forest and its surroundings, and for teaching me how to survive in the forest. I enjoyed very much your company in the field and learning about your culture. I would also like to thank the undergraduates who helped me in collecting data both in field and in lab: Miora Andriamalala, Raoby Solohely, Parfait Rafalinirina,

Savina Venkova, Syia Mehtani, Zo Andriampenomanana and Jake Krauss. Thank you for your valuable contribution. Thanks also to Irene Tomaschewski for help with the biochemical analyses.

I am also thankful to all the local villagers in Ranomafana who helped my research team as porters, cooks and messengers during our field expeditions. Thanks also to Dave, José, Claude, Tahiana, Adafi and Sitraka for being great drivers for my team and my humongous field gears making the 11-hours difficult ride to/from Ranomafana less painful, and for being patient with my multiple stops. Thanks also to Franck Rabenahy and the other undergraduate and graduate students working at Centre ValBio for making my stay in Ranomafana a memorable and pleasant one, and for making me laugh during the difficult times of fieldwork. vii

I would also like to thank the graduate students and post-docs, in the Ecology and

Evolutionary Biology program at Rice University, with whom I have interacted for the past 6 years. Thanks to Brian Maitner, Ben van Allen, Andrew Bibian, Scott

Chamberlain, and Patrick Clay for helpful discussions with data analyses. Thanks also to

Maria Meza-Lopez, Chris Roy, Ching-Hua Shih, Therese Lamperty, Nick Rasmussen,

Chris Dibble, and all the other graduate students in the EEB program for making my experience at Rice enjoyable. I am also thankful to Rice Office of International Students and Scholars for keeping me on the right and legal path from the start of my stay in the

US; and to Laura Gibson, Diane Hatton, Lauren Hanson and Andrea Zorbas of the EEB program at Rice University for their logistical and administrative support.

The support of my family and friends has also been great through these years. I am so grateful to my husband, Mamy-Fy Rakotondrainibe, for his support, encouragement, and understanding. Even though he is not an outdoor person, he agreed to camp with me in the jungle of Madagascar for three months to chase after lemurs every day, and showed great interest in lemur observations despite being pooped on his head by a lemur. Thank you for being patient with me, and for willing to listen to all my practice talks and my endless discussions about seed dispersal and the Janzen-Connell hypothesis.

I also thank my family for their complete understanding and support. Special thanks to my mom for being very supportive, despite her fears and concerns when I left the homeland to go to a country she knew nothing about and then came back to spend time in the jungle full of leeches. I would also like to express my gratitude to Jared Day, Ryan

Gorczycki, Voahangy Ramariavelo-Grenier and the Malagasy community in Houston, for their loving support and for making me feel like I had an extended family in Houston. viii

Contents

Acknowledgments ...... iv

Contents ...... viii

List of Figures ...... xi

List of Tables ...... xiv

Introduction and Summary ...... 1

Patterns of Movement and Seed Dispersal by Three Lemur Species...... 7 2.1. Introduction ...... 8 2.2. Methods ...... 10 2.2.1. Study site and species ...... 10 2.2.2. Data collection ...... 11 2.2.3. Parameter estimation ...... 13 2.2.4. Data analysis ...... 15 2.3. Results ...... 16 2.3.1. Movement and foraging patterns ...... 16 2.3.2. Seed-dispersal patterns ...... 18 2.4. Discussion ...... 20 2.4.1. Seed dispersal patterns ...... 21 2.4.2. Seed dispersal distance relative to other primates ...... 23 2.5. Conclusions and future directions ...... 24 Assessing the Impacts of Nonrandom Seed Dispersal by Multiple Frugivore Partners on Plant Recruitment ...... 27 3.1. Introduction ...... 28 3.2. Methods ...... 30 3.2.1. Study site and system ...... 30 3.2.2. Sampling microhabitats of seed deposition ...... 31 3.2.3. Seed fate experiment ...... 33 3.2.4. Data analyses ...... 34 3.3. Results ...... 35

3.3.1. Patterns of seed-deposition into different microhabitats ...... 35 3.3.2. Seed fate across microhabitats ...... 37 3.3.3. Sapling recruitment ...... 38 3.4. Discussion ...... 39 Frugivores Bias Seed-Adult Tree Associations through Nonrandom Dispersal: a Phylogenetic Approach ...... 43 4.1. Introduction ...... 44 4.2. Materials and Methods ...... 47 4.2.1. Study site and system ...... 47 4.2.2. Direct observations of seed dispersal by lemurs ...... 48 4.2.3. Lemur-mediated seed dispersal intro seed traps ...... 48 4.2.4. Fruiting phenology...... 49 4.2.5. Phylogenetic patterns of seed dispersal ...... 50 4.2.6. Adult-tree distribution ...... 51 4.2.7. Statistical analyses ...... 52 4.3. Results ...... 53 4.3.1. Patterns of seed dispersal relative to fruiting ...... 53 4.3.2. Phylogenetic signal of lemur-mediated dispersal ...... 55 4.3.3. Phylogenetic patterns of lemur-dispersal seeds and nearest adult neighbor .... 55 4.3.4. Adult-tree distributions ...... 57 4.4. Discussion ...... 57 4.4.1. Seed-dispersal bias under fruiting feeding trees ...... 58 4.4.2. Phylogenetic patterns of seed dispersal ...... 59 4.4.3. Adult-tree distribution ...... 60 4.5. Conclusions ...... 61 Patterns of Trait Dispersion and Phylogenetic Structure of Co-fruiting Plant Assemblages ...... 62 5.1. Introduction ...... 63 5.2. Materials and methods ...... 66 5.2.1. Study site and frugivore guild ...... 66 5.2.2. Fruit trait data ...... 67 x

5.2.3. Dispersal mode ...... 69 5.2.4. Monitoring of fruiting phenology ...... 70 5.2.5. Phylogenetic tree reconstruction ...... 71 5.2.6. Phylogenetic signal ...... 72 5.2.7. Trait dispersion among co-fruiting plant assemblages ...... 73 5.2.8. Phylogenetic structure of co-fruiting plant assemblages ...... 74 5.2.9. Test of randomness for trait and phylogenetic metrics ...... 75 5.3. Results ...... 75 5.3.1. Fruit traits and seed-dispersal strategies ...... 75 5.3.2. Fruiting phenology...... 77 5.3.3. Phylogenetic signal ...... 79 5.3.4. Patterns of trait dispersion of co-fruiting assemblages ...... 80 5.3.5. Phylogenetic structure of co-fruiting assemblages ...... 81 5.4. Discussion ...... 83

References ...... 89

Appendices ...... 109

List of Figures

Figure 1.1 – Location of study sites in Ranomafana National Park, Madagascar ...... 3

Figure 1.2 – The three seed dispersers in our study were (a) Eulemur rubriventer, (b) Eulemur rufifrons, and (c) Varecia variegata editorum; and fruits of (d) Cryptocarya crassifolia in Ranomafana National Park, Madagascar. Photo credits: (a) Coke Smith, (c) Mamy-Fy Rakotondrainibe, (b & d) O. H. Razafindratsima. (Plate 1 in Razafindratsima and Dunham, 2005) ...... 4

Figure 2.1 – Representative examples of a daily path for one selected group of each lemur species within the group’s home range (dashed line): (a) E. rubriventer, (b) E. rufifrons, and (c) V. v. editorum. The star symbol indicates location every 15 min, and the grey tree- symbols represent feeding trees. The daily path from start in the morning until the end of observations in late afternoon is indicated by the thick black line...... 17

Figure 2.2 – Circular distribution of the turning angle between consecutive steps. E. rufifrons (middle panel) tends to do more backtracking (moving in a direction counter to the prior point) than E. rubriventer (left panel) and V. v. editorum (right panel), which progress in relatively straight lines. The grey numbers are scales of the number of counts in the various angle classes. Arrows represent vector means...... 18

Figure 2.3 – Frequency distribution of estimated distances that seeds are dispersed from most likely parent trees by (a) E. rubriventer, (b) E. rufrifrons, and (c) V. v. editorum of all the plant species they disperse in RNP. Seed count excludes plant species with seeds < 1 mm in length such as Ficus spp. and Psidium spp. which were found in large quantities in fecal samples...... 19

Figure 2.4 – Significant relationship between female body mass and seed dispersal distance across arboreal lemur (black circles) and non-lemur (open circles) primates. Lemurs have among the shortest seed dispersal distances measured in primates (data and citations in Appendix B) ...... 23

Figure 3.1 – (A) Quantity of Cryptocarya seeds dispersed by each frugivore species (E. rub: E. rubriventer, E. ruf: E. rufifrons, V. v. ed: V.v. editorum) per tree during the plant species’ fruiting season, and for all dispersers combined (All). (B) Dispersal events by the three frugivore species under heterospecific trees with different microhabitat categories (gap: <55% cover, medium-shaded: 55-75%, shaded: > 75%) relative to availability in their habitats. 1: proportional dispersal; > 1: more seeds deposited in the microhabitat category than expected based on availability; < 1: less seeds dispersed in the microhabitat than expected. (C) Mean proportions of dispersal of Cryptocarya seeds

xii under conspecific trees relative to tree availability in the habitats of each disperser. Error bars represent standard error across groups, and composite standard error for the “All” category. Labels above bars indicate significance levels (n.s.: non-significant, *: P < 0.05)...... 36

Figure 3.2 – Survival and recruitment of Cryptocarya in microhabitats with different canopy covers (gap: <55% cover, medium-shaded: 55-75%, shaded: > 75%) and under conspecific crowns. Error bars represent standard errors ...... 37

Figure 3.3 – Probability recruitment model results. The estimated per capita seed recruitment probabilities are presented for seeds dispersed by each frugivore species (A: E. rubriventer, B: E. rufifrons and C: V. v. editorum) vs. seeds fallen under parent tree (no dispersal) and randomly dispersed seeds (relative to microhabitat availability). Boxes and error bars show mean, 95% bootstrap confidence intervals and SD respectively. .... 38

Figure 4.1 – Patterns of seed-dispersal events near fruiting trees by three frugivore species relative to availability of trees fruiting in their habitats. Error bars represent standard errors across months...... 54

Figure 4.2 – (A) More seeds are dispersed by lemurs into seed traps under their food trees that are fruiting than under other trees in the environment. (B) More seeds are dispersed by lemurs under lemur-dispersed trees when fruiting than non-fruiting, but this pattern does not hold for trees with other dispersal mechanisms. Error bars represent standard errors. *: statistical significance P < 0.05, n.s.: non-significant ...... 55

Figure 4.3 – Phylogenetic relationships (NTI) between dispersed seeds (that do not fall under conspecifics) and their nearest adult neighbor trees fluctuate across months for each of the three disperser species. Values < 0 indicate that the species pairs are less closely related than expected by chance; > 0 suggest that they are more closely related than expected by chance, and values near zero indicate a random relatedness...... 56

Figure 5.1 – Differences in fruit and seed traits among the three major known dispersal mode. This figure presents two axes of nonmetric multidimensional scaling (nMDS) plots. Each point correponds to one species, and the ellipse polygons represents standard error with 0.95 confidence limit ...... 76

Figure 5.2 – Absolute frequency distribution of fruiting species (a-d) and individual trees (e-h) (averaged across years), considering all species (a) and all individuals (e), and by dispersal syndromes (b,f: bird-dispersed; c,g: lemur-dispersed; and d,h: mixed)...... 78

Figure 5.3 – Trait dispersion over time of co-fruiting assemblages, as represented by the standardized effect size of functional diversity (ZFDis) for each fruit trait type, from xiii clustered (positive values) to even (negative values). (A) 2-week sampling period (B) monthly fruiting assemblages. P-values correspond to runs test used to assess temporal independence of the trait-dispersion index. Observations were from July 2012-June 2014...... 82

Figure 5.4 – Temporal structure of phylogenetic relatedness of co-fruiting assemblages, as represented by NRI (Net Relatedness Index) and NTI (Nearest Taxa Index). Positive value: clustering, negative values: overdispersion. (A) 2-week sampling period, (B) Monthly fruiting assemblages. P-values correspond to runs test used to assess temporal independence of NRI and NTI. Observations were from July 2012-June 2014...... 83

xiv

List of Tables

Table 2.1 – Mean ± standard deviation of home-range size, daily path lengths, speed, and absolute turning angles of the three frugivorous lemur species in Ranomafana National Park, Madagascar...... 17

Table 2.2 – Generalized Linear Model full-factorial test examining differences in dispersal distances versus the principal component score and species identity...... 20

Table 5.1 – Average value and standard deviation of each morphological and biochemical trait used for each dispersal mode. Dispersal mode was determined based on the consumption of the tree species by each frugivore group. Length and width are in millimeters; biochemical components are given in percentages dry weight...... 77

Table 5.3 – Eigenvalues and percentage of variance accounted by each principal component for phenology variables and each trait type, resulting from PCA and CATPCA (for defense traits with discrete variable), and phylogenetic signal in phenology and each trait type using Pagel’s λ. Value of λ equals 0 if there is no phylogenetic signal, and approaches 1 with increasing phylogenetic pattern...... 80

Chapter 1

Introduction and Summary

A central goal in ecology has been to understand the processes that govern the assembly and structure of ecological communities (Elton 1946, Connor and Simberloff

1979, Cavender-Bares et al. 2009). Recognizing these processes is critical for understanding current and historical mechanisms that determine the origins, maintenance, and distribution of biodiversity (Webb et al. 2002). In terrestrial systems, understanding the impacts of seed dispersers on plant spatial structure and plant-plant associations is an important step towards this goal given (a) the prevalence of frugivores as seed-dispersal agents in many ecosystems (up to 90% of trees, Herrera 2002), and (b) the importance of seed rain as the initial template from which plant community patterns develop (Fort and

Richards 1998, Wang and Smith 2002). Thus, seed dispersal has gained a central place in many theories of community ecology (Hubbell 2001, Terborgh et al. 2002, Wang and

Smith 2002), but it is often considered as an uniform or random process. However, seed dispersal by frugivores is behaviorally driven, generating nonrandom patterns of seed

2 dispersion (Wenny 2001, Westcott et al. 2005). Yet, we know relatively little about how this might affect plant populations and communities.

One of the main advantages of seed dispersal by frugivores is the escape of seeds from distance- and density-dependent mortality associated with proximity near parent trees, known as “Janzen-Connell hypothesis” (Janzen 1970, Connell 1971, Hubbell

1979). However, nonrandom movement of seeds into microsites that are favorable for their germination and establishment, known as “directed dispersal hypothesis” (Howe and Smallwood 1982, Wenny 2001), is also crucial for seeds that do escape such effects.

These mechanisms are not mutually exclusive, and both may be important in determining the contribution of a disperser to plant fitness. Another factor that may influence seed establishment is the identity of adult-tree neighbor near which they land and with which they will interact through facilitation, competition or shared enemies (Carlo 2005,

Blendinger et al. 2011). In other words, even if a seed escapes the proximity of a parent tree, and lands in a potentially favorable microhabitat, the identity of its neighbor could affect its establishment. Landing near phylogenetically similar heterospecific adult-tree neighbors may have negative effects on seedling survival, a mechanism recently defined as “Phylogenetic Janzen-Connell effect” (Liu et al. 2012), because closely related species tend to have similar traits associated with certain biotic and abiotic factors (Gilbert and

Webb 2007, Cavender-Bares et al. 2009, Bagchi et al. 2010). Conversely, being near closely related adult-neighbor may also improve seedling recruitment if closely related species share habitat preferences that favor seedling community levels.

I conducted this study in the rainforest of Ranomafana National Park, Madagascar at four study sites (Figure 1.1). I focused on seed dispersal by three frugivorous lemur 3 species: Eulemur rubriventer, Eulemur rufifrons and Varecia variegata editorum (Figure

1.2). Descriptions of both the study sites and lemur species are detailed in any of the chapters of my thesis.

47°24'0"E 47°28'0"E 47°320"E

2r4'0"S- ’4'0"S

21°8'0"S- °8'0"S

21°12'0"S- °12'0"S

"alatakely 21°16'0"S- °16'0"S itoharanana, Valohoaka

Legend 2r20,0,,S- "20'0"S I IRanomafana NP • Study sites - - - Highway Mangevo* Elevation High : 1401

21°24'0"S' Low : 536 °24'0"S 0 12 4 Km

47°24'0"E 47°28'0"E 47°32'0"E Figure 1.1 – Location of study sites in Ranomafana National Park, Madagascar

The first step to understand the importance of nonrandom seed dispersal by frugivores is to examine how they bias seed dispersal (see Chapter 2-4). In Chapter 2, I examined how the foraging behavior and movement of the three lemur frugivores in this study might result in biased seed dispersal patterns. The three lemur frugivores not only dispersed seeds away from parent and conspecific adult trees, but also biased seed dispersal toward certain microhabitats (Chapter 3) and near fruiting trees upon which 5 they fed (Chapter 4). In Chapter 3, I use a long-lived canopy tree, Cryptocarya crassifolia

(Figure 1.2), which is commonly dispersed by these three dispersers, to investigate the impacts of biased seed dispersal toward different microhabitats on plant recruitment (i.e., directed dispersal). I used probability recruitment models to compare recruitment success under dispersal by each and all frugivores, no dispersal, and random dispersal. Models were parameterized using a three-year recruitment experiment and observational data of dispersal events. Directed dispersal was not an overall advantage of having traits for animal dispersal for Cryptocarya crassifolia. However, the three lemur species increased per capita seed recruitment by four fold vs. no dispersal. In Chapter 4, I discuss how nonrandom patterns of seed dispersal, resulting from frugivore attractions to fruiting trees, influence the phylogenetic relatedness between seeds and adult neighbors, and how these patterns relate to adult-tree distributions. By using fruiting trees as seed dispersal foci, lemur frugivores structure the spatial associations between dispersed seeds and neighboring adult-trees nonrandomly in a given month in terms of fruiting time, dispersal mode and phylogenetic relatedness. However, lemur-dispersed tree species were not more likely to be each other’s neighbors as adults. Interestingly, co-fruiting neighboring trees sharing lemur dispersers were found to be more phylogenetically distant than expected by chance despite a phylogenetic signal in lemur dispersal mode. Finally in

Chapter 5, I examine how fruit and seed traits among co-fruiting species are structured under a temporal framework, which may have implications for understanding the processes underlying mechanisms of seed dispersal by frugivores.

This research suggests an important link between frugivore foraging behavior and the spatial, temporal and phylogenetic patterning of seed dispersal. This research also 6 provides new perspectives about the importance of frugivores in structuring seed dispersion and seed-adult plant associations, with important implications for plant-plant interactions, biodiversity patterns and community structure.

Chapter 2

Patterns of Movement and Seed Dispersal by Three Lemur Species

This chapter has been edited, reformatted, and reprinted from: Razafindratsima, O.H., T.A. Jones and A.E. Dunham. 2014. Patterns of movement and seed dispersal by three lemur species. American Journal of Primatology 76:84-96

Abstract

We combined data on gut-passage times, feeding, and movement to explore the patterns of seed dispersal by Eulemur rubriventer, Eulemur rufrifrons, and Varecia variegata editorum lemurs in Ranomafana National Park, Madagascar. These lemur species deposited less than half of their consumed seeds > 100 m away from conspecific trees (40-50%). Long-distance dispersal (> 500 m) was rare and average dispersal distances were short relative to those reported of similar-sized haplorrhine primates. The three lemur species showed no significant differences in mean seed-dispersal distances.

However, they differed in the shape of their frequency distributions of seed-dispersal distances as a result of differences in how they moved through their habitats. The short 8 distances of seed dispersal we observed and the depauperate frugivorous fauna in

Madagascar suggest seed-dispersal may be more limited than in other tropical forests with important implications for plant-community dynamics, biodiversity maintenance, and restoration efforts in Madagascar.

Keywords: gut-passage time, local dispersal, Madagascar, movement, primate, seed dispersal

2.1. Introduction

In tropical forests, primates constitute between 25 and 40% of the frugivorous biomass (Eisenberg and Thorington Jr. 1973, Chapman 1995) and are critical for seed dispersal of plant communities across the tropics (Fleming 1979, Chapman 1995,

Chapman and Russo 2006, Norconk et al. 2011). Primates play an even larger role in

Madagascar because of the island’s relatively depauperate frugivorous bird and bat communities (Langrand 1990, Hawkins and Goodman 2003). Since the spatial distribution of seed dispersal is determined by the movement of dispersers, the nonrandom way primates move while utilizing resources within their habitats (Garber 1989,

2000, Garber and Jelinek 2006, Janmaat et al. 2006) may have important consequences for the spatial patterning of plant diversity (Wehncke et al. 2003, Chapman and Russo

2006, Cousens et al. 2010). However, studies exploring both primate movement patterns and seed-deposition patterns have been limited (see Wehncke et al. 2003, Chapman and

Russo 2006, Russo et al. 2006), particularly for Madagascar’s primates (but see Spehn and Ganzhorn 2000, Moses and Semple 2011). 9

The temporal and spatial patterning of frugivorous-primate movement combined with seed passage-time in the gut determine how far they disperse seeds from parent trees

(Westcott and Graham 2000, Westcott et al. 2005, Russo et al. 2006, Karubian and

Duraes 2009). Dispersal distance is important because it can strongly affect plant recruitment patterns at different scales (Spiegel and Nathan 2007). At the local scale, seed dispersal allows seeds and seedlings to escape density-dependent mortality resulting from higher levels of natural enemies (pathogens, seed-predators and herbivores), and competition in the vicinity of parent or conspecific adult trees (Janzen 1970, Connell

1971, Howe and Smallwood 1982). At larger scales, long-distance dispersal events can facilitate colonization of newly opened habitats and increase the spread rate of plant populations (Cain et al. 2000, Higgins et al. 2003, Nathan 2006).

Our objective in the present study was to examine movement and seed-dispersal patterns of a guild of seed-dispersing primates in the southeastern rainforests of

Madagascar. Lemurs may differ from other primate species in their role as seed dispersers because of their relatively small home and day ranges (Harvey and Clutton-

Brock 1981, Crowley et al. 2011), and their low metabolic rates relative to body size

(Ross 1992) that might constrain daily movement (Sparrow and Newell 1998). If so, primate-driven dispersal dynamics in Madagascar’s forests may differ substantially from other tropical forests.

Resolving the functional role of primate seed-dispersers in forest communities is becoming urgent given the increasing population declines of primates globally (Chapman

1995, Stoner et al. 2007), and the reliance of up to 90% of tropical trees on seed dispersal by frugivorous vertebrates (Terborgh et al. 2002). In Madagascar, this is especially 10 critical given that lemurs are thought to be the dominant seed dispersers in the ecosystem

(Ganzhorn et al. 1999, Bollen et al. 2004, Wright et al. 2011), and that 91% of lemurs are currently classified as at risk of extinction (IUCN 2012), making them one of the most endangered groups of vertebrates in the world. As habitat loss, hunting and climatic changes continue to put their populations at risk (Dunham et al. 2008, 2011, Barrett and

Ratsimbazafy 2009, Dewar and Richard 2012, Parga et al. 2012, Ratsimbazafy et al.

2012), studies of their role in ecosystem processes are becoming even more urgent.

We describe the movement patterns of three frugivorous Malagasy primates in

Ranomafana National Park (RNP) and explore the resulting patterns of seed dispersal.

Using data on foraging, defecations, movement and estimations of gut-passage duration, we estimated the distances that lemurs carried seeds away from parent trees. We then created frequency distributions of estimated seed-dispersal distances for each lemur species. Finally, we discuss the relevance of lemur seed-dispersal to the maintenance of tropical forests in the region, and how their role compares to non-lemur primates.

2.2. Methods

2.2.1. Study site and species

This study was conducted in Ranomafana National Park (RNP) located in the south-eastern rainforest of Madagascar (21º 16’S and 047º 20’ E), which encompasses

41000 ha of montane forest (Wright et al. 2012). Mean monthly rainfall ranges from 10-

1200 mm (Dunham et al. 2011) and mean annual temperature ranges from 4-32 ºC

(Wright et al. 2011). The precipitation in RNP is highly variable with a peak wet season 11 in January to March (average monthly rainfall of 508 mm), and the dry season peaks in

June-October (average monthly rainfall of 143 mm) (Dunham et al. 2011). The elevation in Ranomafana ranges between 600-1500 m (Wright and Andriamihaja 2002).

The species used in our study constitute a guild of three closely related (Family

Lemuridae), diurnal and highly frugivorous lemur species (Figure 1.2): red-bellied lemur

(Eulemur rubriventer, body mass: 1.6-2.1 kg), red-fronted brown lemur (Eulemur rufifrons, body mass: 2.2-2.3 kg), and southern black-and-white ruffed lemur (Varecia variegata editorum, body mass: 2.5-4.8 kg) (body masses from Glander et al. 1992,

Baden et al. 2008). These species are the largest frugivores in the rainforest of

Ranomafana and are suggested to be important seed dispersers because of their ability to swallow large-sized seeds, which they pass undamaged and with increased germination success (Dew and Wright 1998, Wright et al. 2011). Their populations have declined in several sites within RNP (Wright et al. 2012) and are currently declining throughout their range (IUCN 2012).

Proper permits and authorizations were obtained from Madagascar’s government prior to data collection and we adhered to the American Society of Primatologists (ASP) principles for the ethical treatment of primates. Research did not involve any direct contact or manipulation of animals.

2.2.2. Data collection

Data on movement and foraging patterns were collected between June 2010 and

June 2011 with a total of 1572 observation hours on seven groups of E. rubriventer

(average group size: 3.44 ± SD 0.55), eight groups of E .rufifrons (average group size: 12

7.13 ± SD 3.67) and nine groups of V. v. editorum (average group size: 2.82 ± SD 1.21) accessed from four research sites within the park (Figure 1.1: Mangevo, Talatakely,

Valohoaka and Vatoharanana). The majority of these groups have at least one collar- tagged individual from previous studies (Wright et al. 2012) that we used to identify the focal group. For groups without tagged individuals, we were able to identify individuals based on their physical characteristics, which allowed us to track the same group

(Wehncke and Dominguez 2007).

We located a focal group, starting at about 0700 hrs, and attempted to continuously track them until they were no longer active (about 1700 hrs.) or until we could no longer locate them. We used observations of a focal individual within each group as the basis for our group movement data. Locations were recorded every 15 min using a hand-held Global Positioning System (GPS) unit (Garmin 60 series). Our GPS coordinate points had a mean accuracy of 7.52 ± SD 3.99 m (N = 830, 64% of points were within accuracy of less than 5 m), which allowed us to track relatively exact paths.

Every time the group stopped to feed, we recorded the location and identity of each feeding-tree species with the help of local botanical experts. Specimens from unknown plant species were collected and dried for later identification.

During lemur observations, we collected all observed fecal depositions by group members and recorded the geographic coordinate points of defecations. We then extracted, identified and counted all seeds of size > 1 mm in each fecal sample

(Stevenson 2000). Local research technicians familiar with the local flora were able to identify passed seeds to at least their vernacular names. 13

2.2.3. Parameter estimation

All location records were transformed into a metric x and y coordinate system in

ArcGIS 10.0 (ESRI) and MATLAB 7.12.0 (The Mathworks, Inc.) for parameter estimation including home-range size, daily path length, speed, turning angles, and seed- dispersal distance. We used MATLAB for initial error-checking, to manipulate the database and to calculate movement parameters. In cases where a series of location points of less than 45 min was missing because an animal was out of sight, we interpolated the missing points based on the location points before and after the missing values. Data with missing points of more than 45 min were excluded from analyses. We also excluded path segments that suggested obvious errors in location data (Di Fiore and Suarez 2007) based on visual examination of initial movement plots.

We used the “Minimum Bounding Geometry” function in ArcGIS to calculate the home-range sizes of each group with a convex-hull polygon method based on location points of individuals, feeding sites, and defecation sites collected over the duration of the study (Wehncke et al. 2003). Daily paths were measured by joining successive point location records within a day that constituted more than 7 h of observation. Location data recorded every 15 minutes was used to calculate step lengths (defined as distance between 15 min location points), speed (during a movement bout) and turning angles

(Westcott and Graham 2000). Turning angles (-180° to 180°) represented the angle of direction lemurs moved relative to their previous location (Zimmerman 1979, Garber

1989). A value of 0° indicates no change of direction, a positive value indicates movement to the right, negative value indicates movement to the left, and values near - 14

180° or +180° indicate a reversal of movement (i.e. backtracking) (Erhart and Overdorff

2008a, Will and Tackenberg 2008).

We estimated seed-dispersal distance of each dispersal event by measuring the straight-line distance between the location of the seed(s) in a fecal deposition and a likely parent tree. Multiple conspecific seeds in the same fecal deposit were counted as one dispersal event. We determined probable parent trees by selecting trees of the same species as the dispersed seed(s), for which feeding events occurred before dispersal but within the estimated range of gut passage times for the disperser (description below). If there was more than one candidate parent-tree, we took an average of potential dispersal distances as our estimate. While this introduced some uncertainty to our estimates, it allowed us a large enough sample size to test hypotheses on the social-group level and avoid pseudo-replication, common in many primate studies.

The ranges of gut-passage times used in selecting potential parent trees for dispersal events were based on minimum and maximum values of both our observations and from published direct measurements from captive lemurs. Specifically, in captive lemurs, the gut-passage times from feeding experiments using solute or plastic markers were 155-270 min for E. rubriventer, 60-155 min for E. rufifrons (E. fulvus in original paper) (Overdorff and Rasmussen 1995), and 30-210 min for V. v. editorum (V. variegata in original paper) (Cabre-Vert and Feistner 1995). We estimated gut-passage times from our observational records based on fruits from plant species eaten only once the day of observation for which a seed dispersal event was later observed and prior to defecation, which occurred at least 5 h after the start of observations (Stevenson 2000, Wehncke et al. 2003, Moses and Semple 2011). Following these criteria, the gut-passage times for 15 wild E. rubriventer ranged between 96-433 min (mean = 214.16 min, N = 12), for E. rufifrons 72-371 min (mean = 190.08 min, N = 12), and for V. v. editorum 42-468 min

(mean = 194.81 min, N = 32). To our best knowledge, these are the first accounts of gut- passage times for wild E. rubriventer and E. rufifrons. The mean gut-passage time for V. v. editorum is less than the findings of Moses and Semple (2011) in wild V. variegata

(260 ± SD 160 min) in Manombo rainforest, located southeast of RNP.

2.2.4. Data analysis

We used SPSS 20.0 (IBM Inc.) to perform all statistical analyses. We used

Pearson Chi-square tests to examine how lemur species differed in the frequency distributions of their turning angles (bins of 30°) and dispersal distances (bins of 25 m).

We performed a Multivariate General Linear Model (GLM), evaluated with Pillai’s trace test, to evaluate how lemur species differed simultaneously in home range size and movement variables, including daily path length and speed. Replicates of dependent variables were based on means of individual social-group estimates. Data were normally distributed, thus meeting the assumptions for a multivariate GLM.

To explore how lemur movement was associated with seed-dispersal distance, we used a Generalized Linear Model with lemur species as a fixed factor. Group-level estimates of home range and movement variables, including daily path length and speed, were reduced to independent, uncorrelated variables through a principal component analysis (PCA). We retained one principal component with an eigenvalue > 1 (Peres-

Neto et al. 2005), which explained 57.23% of the total variation. We examined the influence (loadings) of our movement variables on the remaining PCA score and then 16 tested if it could explain significant variation in mean seed-dispersal distances among social-groups with lemur species as a fixed factor in a full factorial analysis.

We also used a nested Generalized Linear Model to explore the relationship between gut passage time and associated seed-dispersal distances, with lemur species as a fixed factor. Variables in this analysis were derived from seed-dispersal events used to calculate gut-passage time of the three species (see methods above). We did not include dispersal events for which dispersal distance was estimated based on a range of measured gut passage times. Distances were nested within day of observation which was nested within group.

2.3. Results

2.3.1. Movement and foraging patterns

The estimated mean group home-ranges of these lemur species are presented in

Table 1. Since the home ranges were estimated based on our limited study period, they may underrepresent year-long range sizes because ranges may shift over seasons in relation to resource availability (Erhart and Overdorff 2008b). Indeed, cumulative estimates of daily home-range sizes did not reach an asymptote by the end of the sampling period. The observed daily path of the three lemur species consisted of movement bouts (travelling between periods of foraging and resting) of variable length

(e.g. Figure 2.1) and speed (Table 2.1). 17

(a) E. rubriventer (c) V. v. editorium

0 35 70 140m I ■ ■ ■ »

Figure 2.1 – Representative examples of a daily path for one selected group of each lemur species within the group’s home range (dashed line): (a) E. rubriventer, (b) E. rufifrons, and (c) V. v. editorum. The star symbol indicates location every 15 min, and the grey tree-symbols represent feeding trees. The daily path from start in the morning until the end of observations in late afternoon is indicated by the thick black line.

Table 2.1 – Mean ± standard deviation of home-range size, daily path lengths, speed, and absolute turning angles of the three frugivorous lemur species in Ranomafana National Park, Madagascar.

Lemur species Home range (ha) Daily path length (m) Speed (m/s)

E. rubriventer 13.70 ± 04.56 ( N = 7) 519.33 ± 367.16 (N = 17) 0.075 ± 0.069 (N = 114)

E. rufifrons 22.06 ± 12.19 (N = 8) 796.91 ± 475.01 (N = 33) 0.093 ± 0.092 (N = 233)

V. v. editorum 29.10 ± 21.89 (N = 9) 1079.95 ± 692.65 (N = 75) 0.079 ± 0.078 (N = 504)

The three studied lemur species differed significantly in how they moved through their habitat (multivariate-GLM: Pillai’s Trace, F6,40 = 3.090, P = 0.003). However, tests of between-subjects effects revealed no significant differences in the lemurs’ speed (F2,21

= 2.222, P = 0.133) or home range-size (F2,21 = 1.964, P = 0.165), and marginally non- significant differences in mean daily path-lengths (F2,21 = 3.122, P = 0.065). V. variegata editorum appear to have the longest average daily path length, which was more than double that of E. rubriventer, which was the shortest of the three lemurs (Table 2.1). The 18 three species showed a non-uniform distribution of observed turning angles (Figure 2.2).

They often returned to the same feeding trees and resting areas, and tended to repeatedly use the same pathways between resting and feeding trees. The frequency distribution of their turning angles differed significantly among species (Pearson Chi-square: X2 =

46.063, df = 22, P = 0.002). The two Eulemur spp. were observed to backtrack more often than V. v. editorium, which had a higher tendency for more direct and linear routes between points Figure 2.2. This was especially more apparent for E. rufifrons.

E. rubriventer E. rufifrons V. v. editorum

2.3.2. Seed-dispersal patterns

Eulemur rubriventer was observed to disperse 23 out of 44 fruit species (plus 3 species not observed in their diet), while E. rufifrons was observed to disperse the seeds of 23 out of 64 fruit species that they consumed during our study period (plus 2 species not observed being consumed), and for V. v. editorum 20 out of 53 fruit species (Appendix

A). During our study, 21 of the 42 fruiting plant species dispersed by the lemurs were only dispersed by one of the three lemur species (i.e., not dispersed by multiple lemur editorum nter rive rub (a) E. m

species). However, our sample size was CD CD

05 limited and more sampling over a longer

seeds time period is needed to get a better

understanding of both dietary and seed- ro 3 dispersed CO of

ro dispersal overlap. All lemur species (c)

deposited a large proportion of seeds to' Proportion away from conspecific trees, and at ro co seeds distances of more than 100 m (42.4% for cn cn cn cn cn cn cn cn cn cn cn cn cn cn cn O 05 ro E. rubriventer, 43.8% for E. rufifrons, co v. dispersed of -t*.

05 and 50% for V. v. editorum). Eulemur •vl -o -o -o -o -o CD o Proportion ro ro ro ro ro ro ro ro ro o P rubriventer was estimated to disperse b CD

ro seeds away from parent trees by a mean o V. g O 4ÿ ■>1 cn b o seeds •-* o

05 distance of 119.81 m (max: 358.7 m), ro cn cn cn cn co of w‘ cn cn cn cn cn 95.59 m (max: 417.1 m) for E. rufifrons, dispersed -c* -&ÿ cn cn cn cn co N> U1 N5 N5 and 116.86 m (max: 630.3 m) for V. v. O K5 Q) Proportion o NJ O ro ro ro 3 o o editorum. The frequency of seed

Figure 2.3 – Frequency3 distribution of estimated O O

IO distances that seeds are dispersed from most O) O g deposition at different distance -0 likely parent trees by (a) E. rubriventer, (b) E. rufrifrons, and (c) V. v. editorum of all the plant -vj species they disperse in RNP. Seed count categories from potential parent trees o o o o excludes plant species with seeds < 1 mm in 3 length such as Ficus spp. and Psidium spp. which differed significantly among the lemur were found in large quantities in fecal samples. 2 r* species (Figure 2.3, Pearson: X =

"j 518.82, df = 40, P < 0.0001). All three lemur species showed a peak frequency of seed

distribution within 25 m of the parent tree. Seed deposition frequency then declined

slowly with distance for V. v. editorum; however, the patterns for Eulemur spp. appear 20 bimodal with a second smaller peak occurring at about 150-175 m from the seed origin for E. rubriventer, and a second peak at 125-150 m for E. rufifrons (Figure 2.3).

In our PCA of movement variables, all three movement variables had similar loadings of same sign (mean speed: 0.778; mean daily path length: 0.763; and mean home range-size: 0.728). When the principal component score was used in a full- factorial, generalized linear model with lemur species as a fixed factor, PC-1 explained significant variation in mean seed-dispersal distance, while species identity and interaction terms were not significant (Table 2.2). At the seed-level, gut-passage time was associated with seed-dispersal distance, independent of species identity (Wald Chi-square

= 320.717, df = 22, P < 0.0001).

Table 2.2 – Generalized Linear Model full-factorial test examining differences in dispersal distances versus the principal component score and species identity Factors Wald Chi-square df P Species 3.525 2 0.172 PC-1 4.502 1 0.034 Species * PC-1 0.884 2 0.643

2.4. Discussion

This study demonstrates that frugivorous lemurs may disperse less than half (42-

50%) of all consumed seeds more than 100 meters from their parent trees. Long-distance dispersal (> 500m) was rare and only observed for V. v. editorum. The three species differed in the way that they moved through their habitats resulting in interspecific differences in the frequency distributions of seed dispersal distances from likely parent trees. However, mean seed-dispersal distance was not significantly different across species. Although the principal component score of movement variables was found to be 21 an important predictor of dispersal distance, the variation found between groups was greater than between lemur species.

Results also showed that lemurs repeatedly used certain paths and feeding trees, which is likely to confer aggregated patterns of seed dispersion and constrain seed dispersal distance (especially if the disperser tends to reverse their direction and return to the same feeding trees). Eulemur rufifrons, in particular, uses a lot of backtracking behaviour (as seen in Figure 2.2), which may increase their probability of depositing seeds in a clustered pattern. Backtracking behaviour in E. rufifrons is suggested to be due to the large size of their groups, and their dependence on large fruiting trees as food resources, to which they frequently return over the course of a day and on successive days (Erhart and Overdorff 2008a). As the three largest frugivores in the system, the dispersal patterns conferred by these primates may help shape the recruitment patterns of many plant species in the forest and affect the colonization and expansion of plant populations (Cain et al. 2000, Higgins et al. 2003, Muller-Landau and Hardesty 2005,

Nathan 2006).

2.4.1. Seed dispersal patterns

The distance that a seed is dispersed from parent trees is critical for seed survival and recruitment probabilities because of density-dependent mortality associated with proximity to conspecific adult trees (Janzen 1970, Connell 1971, Johnson et al. 2012, Liu et al. 2012). This conspecific density-dependent effect may disappear at a distance of 15 m away (Hubbell et al. 2001); thus dispersal distances of more than 100 m are often considered long-distance dispersal (Cain et al. 2000). The three lemur species in this 22 study disperse a majority of seeds well beyond 15 m from potential parent trees, and often at a distance of more than 100 m. However, dispersal beyond 500 m was rare in our study and only observed by V.v. editorum. Because tails of dispersal distributions consist of rare events and are inherently difficult to quantify, they are not well resolved in our study. However, we expect that occasional long-distance dispersal of seeds may be facilitated by these lemurs when sub-adult or adult lemurs transfer groups or immigrate to new areas and thus move long distances rapidly. Theoretical studies have suggested that even very rare long distance dispersal events can enhance species range expansion and survival of species in dynamic landscapes (see Nathan 2006 for review). Events that are rare on the time-scale of one year may still be quite important over the demographic time-scale relevant for a long lived rainforest tree, some of which may live hundreds of years. These three lemur species were observed to emigrate from their natal groups at least once in their lifetime (Erhart and Overdorff 1999, Overdorff and Tecot 2006, Baden

2011); therefore they may contribute to long-dispersal of their food trees that are fruiting during their emigration period. Primates with more fluid social structures, such as group fission-fusion strategy observed in Eulemur rufifrons (Erhart and Overdorff 1999,

Overdorff et al. 1999), can also cover large ranges leading to a higher probability of seed dispersal across different microsites and scattered seed dispersal pattern (Chapman and

Russo 2006).

The observed frequency distribution of estimated seed-dispersal distances by V. v. editorium showed a peak close to the parent tree, followed by a rapid decline and a long tail. However, the distributions of seed-dispersal distances by the two Eulemur species appear bimodal. Most notably, E. ruffifrons made equally likely seed depositions near the 23 parent and at around 150 m from the parent tree, resulting in a high probability that seeds would be transported to that distinct distance class. The first peak may be explained by the common lemur behavior of feeding, resting and then defecating in the same tree

(Overdorff and Strait 1998). The second peak might be explained by the mean distance moved after foraging bouts when they do travel immediately after foraging, and/or may be related to the spatial distribution of favored feeding trees in their habitat, however more research is necessary. Backtracking may also be a factor leading to these patterns.

2.4.2. Seed dispersal distance relative to other primates

The lemur species in our study are clearly effective at depositing a large portion of ingested seeds beyond the risk of density-dependent mortality. However, the dispersal distances estimated from our study and from other studies of lemurs (Spehn and

Ganzhorn 2000, Moses and Semple 2011) are notably shorter than those of primate frugivores in other tropical regions

(Figure 2.4, Appendix B). Small body size of lemurs may account for some of these differences.

Indeed, the mean seed dispersal distance of arboreal frugivorous primate species is strongly associated with body mass (Figure Figure 2.4 – Significant relationship between female body mass and seed dispersal distance across arboreal 2 2.4; R = 0.422, F1,12 = 8.765, P = lemur (black circles) and non-lemur (open circles) primates. Lemurs have among the shortest seed dispersal 0.012). However, both Cebus distances measured in primates (data and citations in Appendix B) 24 capucinus (Wehncke et al. 2003) and Cebus apella (Zhang and Wang 1995) disperse seeds at least twice as far, on average, as the similar-sized V. v. editorum. We thus suspect that the relatively small home-range sizes and day-range lengths of lemurs relative to other primates (Harvey and Clutton-Brock 1981, Crowley et al. 2011) may contribute to the relatively short distances they disperse seeds. This also means that short dispersal distances by lemurs may be common in other forested regions of Madagascar as well, and warrants further study.

It is also important to note that Madagascar’s lemur communities have changed with the large-scale extinction of large-bodied lemurs that disappeared after the Holocene

(Godfrey et al. 2012, Razafindratsima et al. 2013), and today’s forests may be missing an important part of their frugivore community. While it is uncertain which extinct lemurs, if any, were once present at our study site, some extinct species are suspected to have been important for long-range seed dispersal in Madagascar (Crowley et al. 2011).

2.5. Conclusions and future directions

Together, the extinction of Madagascar’s largest primate-frugivores, the naturally depauperate bird and bat communities (Langrand 1990, Hawkins and Goodman 2003), and the relatively short range of seed dispersal by extant lemur frugivores may be indicative of restricted seed dispersal of Madagascar’s trees relative to other tropical forests. A cross-site comparison of congeneric Commiphora trees in Madagascar and

South Africa (dispersed primarily by birds), showed much shorter dispersal distances in

Madagascar relative to South Africa (Bleher and Böhning-Gaese 2001). Similar cross-site comparisons of dispersal differences on a tree community scale are required to determine 25 if Madagascar forests do indeed differ in dispersal regimes through limited local dispersal relative to other regions.

The implications of local dispersal may be large. Short dispersal distances may result in more local clustering of adult-tree distributions (Bleher and Böhning-Gaese

2001), thus increasing the heterogeneity of the forest. Limited dispersal has also been emphasized in the coexistence literature as a potentially important mechanism for reducing competitive exclusion and maintaining tree biodiversity within diverse tropical forests (Hubbell and Foster 1986, Tilman 1994, Snyder and Chesson 2003). Local seed- dispersal has even been suggested to enhance plant speciation (Baak 2005). Thus, if narrow distribution patterns of dispersed seeds are common and have existed throughout

Madagascar’s history, they may have been a contributing factor to the incredible diversity of plants on the island.

If limited local dispersal is wide-spread in Madagascar, it may have important implications for restoration as it may slow recovery of the already highly fragmented and disturbed forests in Madagascar. Such information may be critical for informing approaches for current and future forest restoration activity in Madagascar. This does not imply that lemurs do not play a critical role in seed dispersal in Madagascar’s forests. In contrast, the loss of the local dispersal that does exist could be devastating to plant communities. Given the low diversity and potential for redundancy of frugivores in this system, impacts of lemur extinction may be tremendous for Madagascar’s forests. With

91% of lemurs currently facing high risks of extinction (IUCN 2012), Madagascar’s plant communities may face changes as major dispersers are lost from the system. 26

We are not aware of any other taxa in RNP that may act as alternative seed dispersers for the majority of plant species dispersed by these lemurs. Among the frugivores in RNP, the large-sized lemur species, Propithecus edwardsi (mean body mass: 5.6 Kg) (King et al. 2005) has been suggested to be a seed predator given its mastication of the majority of seeds it consumes (Dew and Wright 1998), but is able to pass intact seeds when it ages and teeth become overly worn (King et al. 2005). Birds and nocturnal frugivorous-primates such as Microcebus rufus and Cheirogaleus crossleyi may also contribute to the seed dispersal of plant species in Ranomafana like their congeners in the littoral rainforest of south-east of Madagascar (Lahann 2007), but they may not be as effective for the many large-seeded species in the forest. However, studies on seed dispersal services by these frugivores in RNP are lacking (but see Rakotomanana et al. 2003).

Given the limited seed-dispersal distance by lemurs observed in our study, we recommend that conservation action should incorporate ways of encouraging the movement and dispersal of seed-dispersing lemurs via corridors and regenerating habitats across anthropogenic barriers (such as roads) to promote regeneration of native habitats

(Razafindratsima and Martinez 2012). Although we lack data on seed-dispersal services by other frugivores in RNP forests, we recommend consideration of the whole frugivore community in conservation actions because the richness and abundance within a frugivore assemblage can have positive effects on the effectiveness of seed dispersal

(Garcia and Martinez 2012). 27

Chapter 3

Assessing the Impacts of Nonrandom Seed Dispersal by Multiple Frugivore Partners on Plant Recruitment

This chapter has been edited, reformatted, and reprinted from: Razafindratsima, O.H and A.E. Dunham. 2015. Assessing the impacts of multiple frugivore partners on plant recruitment. Ecology 96:24-30

Abstract

Directed dispersal is defined as enhanced dispersal of seeds into suitable microhabitats resulting in higher recruitment than if seeds were dispersed randomly.

While this constitutes one of the main explanations for the adaptive value of frugivore- mediated seed dispersal, the generality of this advantage has received little study, particularly when multiple dispersers are involved. We used probability recruitment models of a long-lived rainforest tree in Madagascar, to compare recruitment success under dispersal by multiple frugivores, no dispersal, and random dispersal. Models were parameterized using a three-year recruitment experiment and observational data of

28 dispersal events by three frugivorous lemur species that commonly disperse its seeds.

Frugivore-mediated seed dispersal was nonrandom with respect to canopy cover and increased modeled per-seed sapling recruitment four-fold compared to no dispersal.

Seeds dispersed by one frugivore, Eulemur rubriventer, had higher modeled recruitment probability than seeds dispersed randomly. However, as a group, our models suggest that seeds dispersed by lemurs would have lower recruitment than if dispersal were random.

Results demonstrate the importance of evaluating the contribution of multiple frugivores to plant recruitment for understanding plant population dynamics, and the ecological and evolutionary significance of seed dispersal.

Keywords: demography, directed dispersal, lemurs, primates, seed dispersal, tropical forest

3.1. Introduction

Frugivores act as seed-dispersal agents in many ecosystems (Herrera 2002). In tropical systems, for example, up to 90% of trees have traits adapted for animal-mediated seed dispersal (zoochory) (Howe and Smallwood 1982). A critical step for understanding the adaptive value of zoochory is quantifying the effect of seed deposition patterns on seed fate. A large body of work has suggested that escape of seeds from distance and density-dependent effects of falling under a parent tree (escape hypothesis) (Janzen 1970,

Connell 1971, Howe and Smallwood 1982) may be a primary selective advantage of zoochory for plants (see reviews in Wright 2002, Freckleton and Lewis 2006). However, the enhanced dispersal of seeds into microhabitats favorable for germination and recruitment (directed dispersal) (Howe and Smallwood 1982, Wenny 2001) could also be 29 critical for seeds that do escape such effects. For directed dispersal to be an adaptive strategy for zoochorous plants, the advantage of dispersal into specific microhabitats must be greater than expected under random dispersal (i.e., in which seeds reach microhabitats in proportion to their availability).

Studies exploring directed dispersal have generally focused on the behavior of a single disperser species (e.g., Green et al. 2009, Hirsch et al. 2012) or extreme situations in which seeds have very limited viability unless deposited in specific and very limited microhabitats (Carlo and Aukema 2005, Spiegel and Nathan 2007, Green et al. 2009).

However, most fruiting plants are dispersed by multiple frugivores (who may differ in the locations to which they disperse seeds) and seeds are exposed to a gradient of suitable microhabitats (Brodie et al. 2009). Given the inherent difficulty in tracking seeds dispersed by multiple seed-dispersers, empirical evidence supporting the generality of the directed-dispersal hypothesis is limited. One such study by Wenny and Levey (1998) demonstrated that one of five frugivorous bird species distributed seeds of a Neotropical tree nonrandomly with respect to microhabitat characteristics resulting in higher plant recruitment. However, there is little understanding of how general such effects might be or how a suite of frugivores may together contribute quantitatively to plant demography through directed dispersal. Nonrandom dispersal could be a selective advantage for plants with multiple frugivores if plants have fruit traits to attract dispersers sharing certain foraging behaviors, habitat preferences or probabilistic patterns of seed dispersal that favor recruitment (Wenny 2001), or if one or a few dominant dispersers drive net- advantageous dispersal patterns. Conversely, if trees attract dispersers that have nonrandom dispersal patterns, there would be evolutionary pressure to adapt to these 30 microhabitats. Such mechanisms would allow directed dispersal to arise through evolutionary adaptations of the plant despite the diffuse interactions between individual dispersers and their host-plants (Wenny 2001).

We focused on a long-lived rainforest tree, Cryptocarya crassifolia, in southeastern Madagascar and three frugivorous lemur species that commonly disperse its seeds to assess the advantages of directed dispersal by multiple frugivores to plant recruitment. We tested three main hypotheses: 1) seed dispersal by frugivores would be distributed nonrandomly across microhabitats with varying canopy covers, 2) recruitment would be higher for seeds nonrandomly dispersed than seeds fallen under the parent tree

(no dispersal) or randomly dispersed, and 3) that the three species of frugivores would differ in their contribution to plant recruitment. To do this, we carried out direct observations of dispersal events by the three frugivores and parameterized probability recruitment models using dispersal observations and results of a three-year recruitment experiment across different microhabitats.

3.2. Methods

3.2.1. Study site and system

This study was conducted in the evergreen rainforest of Ranomafana National

Park (RNP), Madagascar (47°18’- 47°37’E, 21°02’- 21°25’S). Established as a protected area in 1991, RNP comprises 41,600 ha (Wright and Andriamihaja 2002). With an average annual rainfall of 2830 mm, the precipitation in RNP is seasonal with a peak wet season in January to March (average monthly rainfall of 508 mm), and a dry season in

June-October (average monthly rainfall: 143 mm) (Dunham et al. 2011). Elevation in 31

RNP ranges between 600 - 1500 m (Wright and Andriamihaja 2002). Our work was conducted in four study sites (Mangevo, Talatakely, Valohoaka and Vatoharanana) within the southern block of forest in the park (Figure 1.1), which is home to 330 known tree and large shrub species.

Our study tree species, Cryptocarya crassifolia Baker (syn. Ravensara crassifolia

Danguy, family: Lauraceae), locally known as Tavolomanitra (see Figure 1.2), is a long- lived canopy tree endemic to the southeastern rainforests in Madagascar. It is a successional-climax species that produces large brown-colored fleshy fruits (length: 20 mm, diameter: 19 mm) with one large, ovoid seed (length: 18 mm, diameter: 17 mm), and sets fruit annually from March to October (our data). In RNP, Cryptocarya seeds are commonly dispersed through defecation by three frugivorous lemur species

(Razafindratsima et al. 2014), which are also the largest-bodied frugivores in the system

(2.0-3.4kg, Razafindratsima et al. 2013): Eulemur rubriventer (red-bellied lemur),

Eulemur rufifrons (red-fronted brown lemur), and Varecia variegata editorum (southern black-and-white ruffed lemur, see Figure 1.2). These species’ reported densities in RNP are 5.46 individuals / km2 for E. rubriventer, 6.70 individuals / km2 for E. rufifrons, and

2.23 individuals / km2 for V. v. editorum (Wright et al. 2012). They are arboreal and disperse seeds while travelling, resting, or feeding in the canopy; however, they differ in their patterns of foraging and movement through their habitats (Razafindratsima et al.

2014). Surveys of V. v. editorum suggest they prefer areas with closed canopy covers

(Herrera et al. 2011).

3.2.2. Sampling microhabitats of seed deposition

We determined patterns of seed dispersal by each lemur species into different 32 microhabitats through direct observations of defecation events. Our sample included observations of eight pre-habituated groups of each species (average group size: E. rubriventer: 3.05±0.71, E. rufifrons: 6.16±3.48, V. v. editorum: 2.94±1.23) which were distributed across four sites (Mangevo, Talatakely, Valohoaka and Vatoharanana) from

June 2010-June 2011. There was extensive overlap among groups and between species at each site. Groups and species were alternated daily (when locating another group was possible) to ensure comparable data were collected across all groups under study. During lemur group follows, a team of 2-4 observers collected as many fecal samples as they could from all group individuals (total: 1340 samples, Appendix C). All seeds within each fecal sample were identified to species. Conspecific seeds in the same fecal deposition were considered as one dispersal event. For each fecal deposition, we also identified the nearest adult neighboring tree and measured canopy cover above the fecal deposition using a spherical densiometer (Model C, Forestry Densiometers USA). We categorized canopy cover as gap (<55% canopy cover), medium-shaded (55-75%), and shaded (>75%). The “gap” category was representative of observed recent tree fall gaps in the park. The “shaded” category (>75%) was chosen such that it was the minimum interval that was represented by at least 10% of the forest within each of the study sites.

We tested for nonrandom patterns of seed deposition by comparing observed deposition microhabitats with availability in each study site. To quantify availability, we established nine straight-line transects in each of the four study sites that overlapped with the ranges of the lemurs. Transects were 500 m long x 4 m wide (Brodie et al. 2009), ran perpendicular to a main trail system at each site, and were spaced at least 100 m apart.

Total area sampled per site was 18,000 m2. To estimate the relative availability of adult 33

Cryptocarya trees, we surveyed tree composition in each transect by identifying and counting all adult trees (defined as minimum size of reproduction, estimated from known stems by trained local botanists). We estimated the availability of each canopy-cover category by measuring canopy cover at points every 10 m along each transect using a densiometer (for microhabitat availability, see Appendix D).

3.2.3. Seed fate experiment

We conducted an experiment with Cryptocarya seeds to assess seed germination, removal/predation and seedling survivorship across different canopy covers and under conspecific trees. This experiment was conducted in Talatakely, because its accessibility allowed for biweekly monitoring. Seeds used in the experiment were extracted from fresh ripe fruits collected on the ground or from fruiting trees of Cryptocarya trees rather than from lemur defecations because of logistical challenges in collecting enough seeds in lemur scat. However, our germination rates for Cryptocarya seeds were similar to rates found for seeds passed through the gut of the three lemur species in a previous study

(Dew and Wright 1998). Seeds were thoroughly mixed and then planted under conspecific trees or in one of three microhabitat treatments: 1) under canopy comprised of gap habitat (<55%), 2) medium-shaded habitat (55-75%), and 3) shaded habitat

(>75%). Each treatment was replicated 10 times. All heterospecific sites were located

>15m away from a Cryptocarya tree. The locations of experimental plots were selected in close proximity to small trails for ease of access and were at least 100 m apart.

To separate the effects of seed removal/predation from germination, each plot was comprised of paired subplots containing a “rodent exclosure” and an “open” treatment in each of our replicated microhabitats described above. Exclosure subplots consisted of 34 wire cages (mesh size: 1 cm2) measuring 30 x 30 x 15 cm pinned to the ground with wooden stakes in each corner (Brodie et al. 2009). Open subplots were located ~ 25 cm from exclosure subplots and marked with wire-flags. Five seeds were placed on the surface of each paired subplot (4 microhabitat types × 10 locations × 2 exclosure treatments × 5 seeds = 400 seeds total). For the open subplots, seeds were each tethered to a buried garden staple with 150 cm thread glued to each seed with scent-free, non-toxic superglue to enable tracking of seed fate (Forget and Cuijpers 2008). Since seed removal and predation were difficult to disentangle in many cases, we combined them into a single category (“seed removal/predation”).

Plots were monitored every two weeks for three months (June to September 2010) for seed removal/predation, germination, and survivorship. To quantify seed removal/predation, we counted the number of seeds missing or with obvious signs of predation. No signs of rodents entering the cages were evident during the experiment.

Unfortunately, after an initial three months of monitoring, new park regulations required that we remove cages from our exclosure subplots. All plots were then monitored to assess one-year and three-year survivorship.

3.2.4. Data analyses

We used a General Linear Model in SPSS 20.0 (IBM Inc., Armonk, NY) to examine how the three disperser species differ in the estimates of the quantity of

Cryptocarya seeds dispersed per adult tree in their habitat per fruiting season (see

Appendix E). Sampling level was per group with ‘species’ treated as a fixed effect. When site was included as a random factor, it was not a significant predictor (F3,2.33 = 0.45, P =

0.74), so was excluded from the analysis. Data met assumptions of a parametric test. 35

Patterns of seed deposition bias into specific microhabitat categories (based on availability, see Appendix E), were compared between disperser species and relative to a null expectation with a chi-square test. We also used a paired t-test to determine if social groups of dispersers deposit Cryptocarya seeds under conspecific trees more frequently than expected based on tree abundances in the groups’ habitats. Replicates were based on means of individual lemur group estimates.

We performed a full-factorial generalized linear model (GLM) with negative binomial distribution and log-link function in SPSS 20.0 to assess the effects of canopy cover and nearest adult-tree conspecificity on Cryptocarya recruitment success and vital rates. The methods used for the estimations of the seed-dispersal parameters and vital rates are detailed in Appendix E.

We assessed the relative contribution of each disperser to the 3-year recruitment of their host plant relative to random dispersal and no dispersal with a probability recruitment model performed in MATLAB R2011a (The MathWorks Inc., Natick, MA) parameterized by our observational and experimental data (Appendix E).

3.3. Results

3.3.1. Patterns of seed-deposition into different microhabitats

Frugivore species differed significantly in the quantity of seeds dispersed per adult tree in the habitat (F2, 18 = 3.77, P = 0.04; Figure 3.1A). Eulemur rubriventer dispersed about three times more Cryptocarya seeds per tree than E. rufifrons (LCD post- hoc test, P = 0.03) or V. v. editorum (P = 0.02). There was no significant difference between the quantity of seeds dispersed per tree by E. rufifrons and V. v. editorum (P = 36

0.10). All three lemur species in our study 1.0 -I

Q Q. 0.6 TD with respect to availability of canopy cover Z ® 2 §.0.4 E » categories. This was true when we only 3 13

ro considered dispersal events under heterospecific > < 0.0 E.rub E.ruf V.v.ed All 2 trees (Figure 3.1B; E. rubriventer: χ = 166.9, df Disperser

= 23, P < 0.0001; E. rufifrons: χ2 = 288.9, df = B 1.5 □ E. rub □ E.ruf 2 1.0 23, P < 0.0001; V .v. editorum: χ = 456.6, df = □ V.v.ed

ro 23, P < 0.0001), or when conspecific trees were 3 0.5 included in the analysis (Appendix F). All three 0.0 lemur species dispersed seeds into medium- -0.5 -|

medium -1.0 gap shaded shaded areas and away from shaded areas shaded Canopy cover categories significantly more than expected by chance c 0.6 □ Dispersal under Cryptocarya □ Cryptocarya trees in habitat (Figure 3.1B). The two Eulemur spp. also had a 0.5 ■| 0.4 o tendency to differentially disperse seeds into gaps Q. 2 0.3 c ro while there was no bias for V. v. editorum (Figure ® 0.2 0.1 3.1B). E. rufifrons dispersed Cryptocarya seeds nR 0.0 E.rub E.ruf V.v.ed under the crowns of conspecific adult-trees more Disperser Figure 3.1 – (A) Quantity of Cryptocarya seeds dispersed by each frugivore species (E. rub: E. rubriventer, E. ruf: than expected by chance (Figure 3.1C) (t5 = 3.07, E. rufifrons, V. v. ed: V.v. editorum) per tree during the plant species’ fruiting season, and for all dispersers combined (All). (B) Dispersal events by the three P = 0.03), but there was no significant difference frugivore species under heterospecific trees with different microhabitat categories (gap: <55% cover, medium- shaded: 55-75%, shaded: > 75%) relative to availability in for E. rubriventer (t6 = 0.88, P = 0.41) or V. v their habitats. 1: proportional dispersal; > 1: more seeds deposited in the microhabitat category than expected based on availability; < 1: less seeds dispersed in the editorum (t3 = 0.91, P = 0.43). microhabitat than expected. (C) Mean proportions of dispersal of Cryptocarya seeds under conspecific trees relative to tree availability in the habitats of each disperser. Error bars represent standard error across groups, and composite standard error for the “All” category. Labels above bars indicate significance levels (n.s.: non-significant, *: P < 0.05).

3.3.2. Seed fate across microhabitats

The three-year sapling recruitment differed significantly with canopy cover category (Wald χ2 = 8.62, df = 2, P = 0.01). Recruitment rates were >3 times higher in gaps than in medium and heavily shaded sites (0.28 vs. 0.09; Figure 3.2). Recruitment was also significantly lower near conspecific adult neighbors when controlling for canopy cover (Wald χ2 = 3.99, df = 1, P = 0.046). Interaction terms (canopy cover category and conspecificity) were not quite significant (Wald χ2= 2.53, df = 1, P = 0.11).

Both canopy cover and conspecificity had significant effects on one to three year transition probabilities (Wald χ2= 9.52, df = 2, P = 0.01; Wald χ2= 9.81, df = 1, P = 0.002 respectively) with a significant interaction between variables (Wald χ2= 6.50, df = 1, P =

0.01). However, there was -gap no significant effect on -• germination rates (canopy cover: Wald χ2= 0.40, df = 2,

P = 0.82, conspecificity:

Wald χ2= 0.01, df = 1, P =

0.91), three month to one year transition probabilities

(canopy cover: Wald χ2=

4.63 , df = 2, P = 0.10, conspecificity: Wald χ2= Figure 3.2 – Survival and recruitment of Cryptocarya in 0.60, df = 1, P = 0.44), or microhabitats with different canopy covers (gap: <55% cover, medium-shaded: 55-75%, shaded: > 75%) and under removal/predation rates conspecific crowns. Error bars represent standard errors 38

(canopy cover: Wald χ2= 0.57, df = 2, P = 0.75, conspecificity: Wald χ2= 0.30, df = 1, P

= 0.58).

3.3.3. Sapling recruitment

Results of our probability recruitment model suggest that Cryptocarya seeds

dispersed by lemurs will have on average a four times higher three-year recruitment rate

than seeds falling directly under the canopy of the parent tree. For each lemur species,

this difference was supported by non-overlapping 95% confidence intervals (Figure 3.3).

0.20 seed 0.16 per 0.12 recruitment 0.08 of

Probability 0.04

0.00 nonrandom random no Dispersal type

The model suggested that directed dispersal of Cryptocarya seeds was performed

by E. rubriventer but not E. rufifrons or V. v. editorum. Seeds dispersed by E. rubriventer

were estimated to have 1.3 times higher recruitment success than randomly dispersed

seeds. However, random dispersal was estimated to be more advantageous for 39 recruitment than lemur dispersal by E. rufifrons and V. v. editorum (random dispersal was

2.2 and 2.5 times higher respectively; non-overlapping 95% confidence intervals).

3.4. Discussion

Understanding the effects of nonrandom frugivore-mediated seed dispersal on plant recruitment is critical for understanding plant population dynamics and the ecological and evolutionary consequences of seed dispersal. Nonrandom, directed movement of seeds towards suitable microhabitats (directed dispersal) is frequently discussed as an adaptive advantage of seed dispersal by frugivores (Farwig and Berens

2012, Beckman and Rogers 2013). Unfortunately, the difficulty in tracking seeds dispersed by multiple frugivores has meant we have limited understanding of such advantage to plant recruitment. Demographic analyses incorporating variation of microhabitats into which seeds are dispersed, are important for assessing the population- level consequences of nonrandom seed dispersal (Brodie et al. 2009, Loayza and Knight

2010). Our model allowed us to address the relative contribution of different dispersers to the partial recruitment of their host-plant, and to predict the value of nonrandom dispersal for seed recruitment success.

Results from our probability recruitment models suggest that nonrandom dispersal is not currently an overall advantage of having traits associated with zoochory by lemur frugivores for Cryptocarya crassifolia, a long-lived tree species in a Madagascar rainforest. While all frugivores in our study dispersed seeds nonrandomly with respect to microhabitat, their patterns of dispersal differed from each other. One of the plant’s three frugivores (Eulemur rubriventer) may provide directed dispersal for the plant, with 40 modeled sapling recruitment of dispersed seeds 1.3 times higher than for seeds dispersed randomly; however, recruitment of seeds dispersed by the other two frugivorous lemurs

(Eulemur ruffifrons and Varecia variegatta editorum) were 2-2.5 times lower than for seeds dispersed randomly. The overall contribution to sapling recruitment, by the three dispersers as a group, was modeled to be lower than if those seeds were dispersed randomly in the habitat. This was true even though the most effective disperser (E. rubriventer) was estimated to disperse more seeds per adult tree than the other two frugivore species combined. However, all three frugivorous lemurs did provide

Cryptocarya trees an advantage by moving seeds away from the crowns of parent trees, resulting in a four-fold higher recruitment probability.

Frugivores differed in the microhabitats where they tended to deposit seeds; resulting in differences in recruitment outcomes relative to random dispersal models.

Both Eulemur spp., biased dispersal into gaps but V. v. editorum did not. Eulemur ruffifrons biased dispersal under conspecifics where recruitment was low. These differences are likely due to the way the three species use their habitat. For example, V. v. editorum is not frequently observed in habitat with open canopies (Herrera et al. 2011), whereas E. ruffifrons frequently backtracks to the same fruiting trees during the day

(Razafindratsima et al. 2014), which may lead to increased seed deposition under conspecifics.

One hypothesis for the adaptive value of animal-mediated seed dispersal is the benefit of directed dispersal to recruitment (Wenny 2001). Animal-mediated dispersal is often nonrandom, and if seeds are directed by animals towards areas that enhance recruitment, a clear advantage is obtained. In our study, lemurs, as a group, do not 41 contribute to a directed dispersal advantage to the plant in the sense that recruitment was lower than if the total number of lemur-dispersed seeds were instead distributed randomly in the environment. However, the fact that lemurs moved some of the seeds away from conspecifics did act to greatly increase modeled seedling recruitment relative to the no- dispersal scenario. Our results stress the importance of evaluating the contribution of multiple dispersers of a shared host-plant species to recruitment to determine if nonrandom dispersal could be providing an overall advantage to plant demography.

There have been few studies that have evaluated the impact of nonrandom seed dispersal by multiple frugivores on host-plant recruitment. These few studies are supportive of our findings that frugivores often differ in their patterns of dispersal and contribution to plant recruitment (Wenny and Levey 1998, Brodie et al. 2009). Wenny et al. (1998) demonstrated that only one of five species of birds directed seeds nonrandomly into gaps where recruitment success was high. Brodie et al. (2009) found that three frugivore species sharing a host-plant species differentially dispersed seeds into different microhabitats resulting in variable impacts on plant recruitment; however, it is unclear if the dispersal patterns qualified as directed dispersal.

While lemur-mediated dispersal is less effective than random dispersal, it still results in four times higher recruitment than no dispersal for their host-plant species. The functional loss of seed dispersers has become a growing problem worldwide as a result of anthropogenic factors such as hunting and habitat fragmentation (Farwig and Berens

2012, Vidal et al. 2013). Our model contributes to other recent work (Cordeiro and Howe

2003, Brodie et al. 2009, Levi and Peres 2013) in suggesting that the absence of animal dispersers can have negative consequences for plant demography. If plant-animal 42 dispersal webs tend to be modular, such that plants tend to specialize on animal dispersers by disperser type (e.g., Donatti et al. 2011), then there may be important consequences of disperser loss for plant populations in Madagascar where almost all frugivorous lemurs are currently facing high risks of extinction (Schwitzer et al. 2014).

Chapter 4

Frugivores Bias Seed-Adult Tree Associations through Nonrandom Dispersal: a Phylogenetic Approach

Abstract

Frugivores are the main seed dispersers in many ecosystems making behaviorally driven, nonrandom patterns of seed dispersal a common process. Yet, we lack a clear understanding of how frugivores may structure spatial associations between seeds and adult-tree neighbors. We used a temporal and phylogenetic approach to address this. Data show that by using fruiting trees as seed dispersal foci, lemur frugivores structure the spatial associations between dispersed seeds and neighboring adult-trees nonrandomly in a given month in terms of fruiting time, dispersal mode and phylogenetic relatedness.

However, lemur-dispersed tree species were not more likely to be each other’s neighbors as adults. Interestingly, neighboring trees sharing lemur dispersers and with overlapping 44

fruiting time were found to be more phylogenetically distant than expected by chance despite a phylogenetic signal in lemur dispersal mode. Results suggest the importance of frugivore-mediated seed dispersal in structuring seed-adult plant associations with important implications for plant-plant interactions within a community.

Keywords: seed dispersal, nonrandom dispersal, spatial patterns, contagious dispersal, phylogeny, fruiting timing, lemurs, Madagascar, primates, tropical forest

4.1. Introduction

Spatial and temporal processes of seed dispersal may play an important role in structuring plant-plant interactions within communities (Hubbell et al. 2001, Terborgh et al. 2002, Rosindell et al. 2011). These processes set the template for spatial proximities within a community, which in turn determine plant-plant interactions (Fort and Richards

1998, Wang and Smith 2002). Importantly, the patterns of spatial proximities resulting from seed dispersal may be nonrandom if seed-dispersing frugivores alter their movements in response to the locations of resources (i.e. fruits) in their environment at a given time (Herrera 2002, Boyer et al. 2006). Since frugivores are the main agents of seed dispersal in many ecosystems (up to 90%, Herrera 2002), understanding these processes is a critical step toward understanding the structure and diversity of plant communities (Terborgh et al. 2002).

Frugivores, which tend to be generalists (Herrera 1982, Bascompte and Jordano

2007, McConkey and Brockelman 2011), spend a large proportion of their time foraging for fruits, and may have a tendency to defecate seeds while feeding (Pratt and Stiles 45

1983, Clark et al. 2004). For instance, Clark et al. (2004) report that birds in Dja Reserve,

Cameroon, frequently dispersed seeds under bird-dispersed tree species during their fruiting season. Such behavior could result in nonrandom associations between dispersed seeds and their adult neighbors in terms of fruiting phenology (seasonal timing of fruiting), dispersal mode, and/or evolutionary history (e.g., if fruiting phenologies or dispersal modes are evolutionarily conserved). While there has been increasing recognition that frugivores often use fruiting trees as dispersal foci (a process often referred to as ‘contagious dispersal’) (Kwit et al. 2004, Clark et al. 2004, Carlo et al.

2007, Blendinger et al. 2011), the resulting pattern of nonrandom associations between seeds and adult heterospecifics remain understudied, particularly at the community level.

Given the potential consequences of these nonrandom spatial proximities for recruitment success, understanding how frugivores and timing of fruiting can mediate plant-plant spatial relationships may have important implications for our understanding of processes that shape plant communities.

One of the main hypothesized advantages of vertebrate seed dispersal is movement away from a parent tree to avoid the associated effects of distance- and density-dependent mortality (Janzen 1970, Connell 1971, Howe and Smallwood 1982). However, once conspecific neighbors are avoided, seed and seedling recruitment success near heterospecific neighbors may vary according to similarities in traits and/or evolutionary history. For example, recent experimental and observational studies suggest that the probabilities of seedling establishment may be predicted by the phylogenetic relatedness between seeds and their adult-neighbors (Vamosi et al. 2009, Burns and Strauss 2011,

Liu et al. 2012, Lebrija-Trejos et al. 2014). Species that are closely related may share 46

susceptibilities and defenses against natural enemies due to trait conservatism (Webb et al. 2006, Gilbert and Webb 2007, 2007, Cavender-Bares et al. 2009), and tend to have similar responses to abiotic factors as well (i.e., habitat filtering, Vamosi et al. 2009,

Burns and Strauss 2011). If phylogenetic patterns are predictive of outcomes of plant- plant interactions, it is important to understand the role frugivores may play in structuring these interactions. Integrating an evolutionary component into studies of seed dispersal can provide a better understanding of the consequences of frugivore-mediated seed dispersal in a community context.

The spatial patterns of seed dispersal often initiate subsequent life-history processes such as recruitment and survival (Nathan and Muller-Landau 2000, Beckman and Rogers

2013), but it is unclear how nonrandom seed dispersal patterns might translate into spatial proximities of adult plants. By laying the template for plant recruitment, nonrandom seed dispersal could influence neighborhood patterns of adult plant distributions (Levine and

Murrell 2003, Rodríguez-Pérez et al. 2012). Alternatively, post-dispersal processes may be so strong that they override initial spatial patterns such that associations found at the seed stage are not predictive of adult distributions and associations.

In this study, we examined how nonrandom patterns of seed dispersal, resulting from frugivore attraction to fruiting trees (contagious dispersal), affects spatial proximities of seeds and adult trees in a diverse tropical forest. Specifically, we investigated (1) whether fruiting trees are used as dispersal foci by three generalist primate frugivores, (2) how this behavior may structure nonrandom heterospecific associations between dispersed seeds and adult neighbors in terms of fruiting phenology 47

(seasonal timing of fruiting), dispersal mode, and evolutionary history, and (3) how nonrandom patterns of neighbor association at the dispersal stage may translate into adult-tree distributions. We applied a combination of direct and indirect dispersal observations, tree community surveys, and a phylogenetic approach to a diverse community of rainforest trees in Madagascar.

4.2. Materials and Methods

4.2.1. Study site and system

This study was carried out in the evergreen montane rainforest of Ranomafana

National Park (RNP), located in southeastern Madagascar (47°18’- 47°37’E, 21°02’-

21°25’S). There is a highly seasonal pattern of rainfall in RNP with monthly averages ranging from 10-1200 mm; a peak wet season occurring in January to March (average monthly rainfall of 508 mm), and the dry season peaks in June-October (average monthly rainfall: 143 mm) (Dunham et al. 2011). Elevation in RNP ranges between 600-1500 m

(Wright and Andriamihaja 2002). RNP is home to 330 known tree and large shrub species (Razafindratsima and Dunham 2015).

We focused our study on the seed dispersal patterns of three arboreal, highly frugivorous lemur species: Eulemur rufifrons (red-fronted brown lemur), Eulemur rubriventer (red-bellied lemur) and Varecia variegata editorum (Southern black and white ruffed-lemur). Eulemur and Varecia species are the largest frugivores in the system

(2.0-3.4kg, Razafindratsima et al. 2013), at least 1 kilo larger than all other frugivores in

Ranomafana. These genera are generalist frugivores, known to disperse seeds of many 48

plant species in Ranomafana (Dew and Wright 1998, Razafindratsima et al. 2014,

Razafindratsima and Dunham 2015) and elsewhere throughout the island (Moses and

Semple 2011, Razafindratsima and Martinez 2012, Sato 2012). Fruit constitutes about 57-

99% of their monthly diet (Wright et al. 2011).

4.2.2. Direct observations of seed dispersal by lemurs

We quantified seed deposition patterns of the three frugivores through direct observations of lemur-defecation events. We sampled defecations of seeds from eight pre-habituated groups of each of the three lemur species (average group size: E. rubriventer: 3.05 ± SD 0.71, E. rufifrons: 6.16 ± SD 3.48, V. v. editorum: 2.94 ± SD

1.23) from June 2010 to June 2011, which were distributed across four study sites within the southern parcel of the park (Mangevo, Talatakely, Valohoaka and Vatoharanana; map in Figure 1.1). We alternated observed groups and species daily to ensure that we sample comparable data across all groups. During lemur observations, a team of 2-4 observers attempted to collect all fecal depositions by group members. We identified the species of all seeds within each fecal deposition, and considered conspecific seeds in the same fecal sample as one dispersal event. Also, for each fecal deposition, we identified the species and fruiting status of the nearest adult neighboring tree (i.e., the tree whose canopy was most directly above the defecated seed or whose trunk was nearest to the seed).

4.2.3. Lemur-mediated seed dispersal intro seed traps

We also sampled lemur-mediated seed dispersal indirectly using seed traps. Each trap was made of durable window screening stapled on a ring of polyvinyl tubing (total 49

trap area ~0.5m2) and hung on trees to a height of 1-1.5 m to reduce seed removal by granivores; however, suspension of traps may not eliminate seed removal completely

(Clark et al. 2004). Traps were placed every 25 m along 20 linear pre-established transects located in two established field sites within RNP (Vatoharanana and Valohoaka, see map in Figure 1.1). Transects were 100 m long, ran perpendicular to main research trails and were spaced at least 100 m apart. We recorded the identity of all tree species with crowns overhanging each trap. Tree species were categorized as lemur-dispersed if seeds were observed in lemur scats, and if the tree species has been observed consumed and/or dispersed by the studied lemur species based on the literature and expert knowledge of local botanical and primatological research technicians in RNP (Appendix

G). Species with fruit traits associated with dispersal by large lemurs (large-sized, indehiscent, dull-colored and with edible fleshy pulp) were also considered as lemur- dispersed species (Scharfe and Schlund 1996, Birkinshaw 2001, Bollen et al. 2004,

2005). Every two weeks for 16 months, we monitored seed traps, and identified and recorded their contents including the presence, identity and number of seeds in lemur scat, and recorded the fruiting status of all trees with crowns overhanging each seed trap.

4.2.4. Fruiting phenology

The patterns of fruiting phenology in RNP were determined through monitoring of fruiting status for all canopy trees ≥ 5 cm DBH (Diameter at Breast Height), and understory trees ≥ 2 cm DBH, within 2 m on either side of each of the transects described above. The DBH thresholds were chosen because many trees in Madagascar fruit below

10 cm DBH (pers. obs.). Individuals were marked with numbered aluminum tags. 50

Phenological monitoring occurred every two weeks from July 2012 - July 2014, during which we recorded the presence or absence of immature and mature fruits for each marked individual (Morellato et al. 2000).

4.2.5. Phylogenetic patterns of seed dispersal

We constructed a local phylogenetic tree of all tree species in RNP using

Phylomatic v3 and the Angiosperm Phylogeny Group III (AGP3) derived megatree

R20120829 (Webb and Donoghue 2005). Phylogenetic branch lengths were calibrated using the bladj algorithm in Phylocom 4.2 (Webb et al. 2008) with estimated molecular and known fossil ages (Wikström et al. 2001).

To estimate the phylogenetic dispersion of tree species categorized as dispersed by lemurs within the community of tree species in RNP, we used Fritz and Purvis’ D statistic for discrete traits (Fritz and Purvis 2010). Phylogenetic dispersion (D) was calculated using the package caper, by comparing the observed phylogenetic pattern with

1000 permutations, in R 3.0.3 (R Core Team 2012) based on random or Brownian motion patterns of evolution. A value of D > 1 suggests phylogenetic overdispersion, D = 1 indicates a random trait distribution, D ~ 0 indicates a Brownian motion mode of evolution, and D < 0 indicates a highly clustered trait.

We then explored how seed dispersal by lemurs may bias the phylogenetic relationship between dispersed seeds and their adult-tree neighbors. We assessed the phylogenetic relationship between the tree where the lemur frugivore was perched during a dispersal event and the seed(s) dispersed during the event. We used a phylogenetic 51

index of relatedness, Net Relatedness Index (NRI) (Webb et al. 2002), to determine if the associations were random or biased with respect to phylogeny. NRI is a standardized effect size of the mean phylogenetic distance between all taxa (MPD) in the sample (s) relative to a random distribution (r). In this case, we had only two taxa so it represented the standardized effect size of the actual phylogenetic distance. NRI was estimated in

Phylocom 4.2 (Webb et al. 2008) using the following equation: NRI = -1 x [(MPDs -

MPDr)/sd(MPDr)]. To generate null communities, seed and tree species pairs were randomly drawn without replacement (for a total of 1000 iterations), from two separate species pools using MATLAB R2013b (The MathWorks Inc., Natick, MA). For each lemur group, the pool for seed species was comprised of species being dispersed during lemur observations. The tree species pool consisted of tree species in each of the group’s home range, surveyed from transects of 500 x 4 m in each group’s territory (details on transect in Razafindratsima and Dunham 2015). The mean phylogenetic distances (MPD) between species pairs were calculated in Phylocom 4.2 (Webb et al. 2008).

4.2.6. Adult-tree distribution

We examined how patterns of dispersal of seeds near fruiting trees may be reflected in the distribution of adult trees by examining the tendency of lemur-dispersed trees to have nearest neighbors that are also lemur-dispersed with an overlapping fruiting time. We mapped all individual trees within each of the phenology transects described above. For each tree (referred as “focal”), we determined its nearest neighbors in four cardinal directions using the “point-centered quarter” method by Cottam & Curtis (1956).

This method consists of dividing the area around the focal tree into four 90° quarters and 52

measuring the distance from the focal to the closest tree in each quarter. We measured the distances between all trees within each plot, and the angles between each pair of trees using the package spatstat in R 3.0.3 (R Core Team 2012). Nearest neighbor trees within each quarter angle of the tree focal were then simply sorted. These observed patterns were compared with null expectations. To obtain the null estimations, we randomly shuffled the locations of trees using the algorithm rlabel with the package spatstat in R.

We also tested the distribution of phylogenetic relationships of focal lemur- dispersed tree with neighbor co-fruiting adult tree relative to a random expectation using

NRI described above, and the Nearest Taxon Index (NTI) which is calculated in the same way except it uses the Mean Nearest Neighbor Distance (MNND), rather than MPD.

MNND is simply the mean phylogenetic distances between nearest relatives within samples.

4.2.7. Statistical analyses

We used a Generalized Linear Model (GLM) in SPSS 22.0 (IBM Inc. Armonk,

NY) to examine whether the three frugivore species deposit seeds near fruiting, lemur- dispersed trees more frequently than expected by chance (i.e., relative to availability of fruiting trees in the habitat). Because co-fruiting assemblages vary over time, we used monthly percentage of dispersal events under fruiting trees as our dependent variable, with groups nested within species. To obtain null expectations, we ran multinomial randomizations of seed depositions relative to neighbor trees sampled based on availability of fruiting lemur-dispersed trees in the habitat. Fruiting data were from our 2- year phenological monitoring. We assigned a tree as fruiting if it fruited during that 53

month. While some trees may not have fruited the entire month, this method made our test of biased-dispersal conservative. Randomizations were run 1000 times in MATLAB

R2013b (The MathWorks Inc., Natick, MA). With our data from seed traps; we used a

Generalized Linear Model with Poisson log-link function to examine whether dispersal

(count of seeds dispersed per month in each trap) was biased toward seed traps with a fruiting, lemur-dispersed tree overhead. We also performed a Friedman’s test to examine whether the dispersal rate of lemur-dispersed seeds was higher under lemur-dispersed tree species during the fruiting versus non-fruiting period.

For our analyses examining biases in phylogenetic distribution, we used one- sample t-tests to determine whether the mean of the estimated values of NRI / NTI differed from zero, demonstrating a nonrandom phylogenetic association.

For our adult tree analysis, we performed a Friedman’s test to compare the observed and expected mean proportion of neighbor trees that are lemur-dispersed and have overlapping fruiting time with the focal trees.

4.3. Results

4.3.1. Patterns of seed dispersal relative to fruiting

Overall, the three frugivore species in our study dispersed seeds (through defecation) near fruiting trees upon which they fed, significantly more than expected by chance (Figure 4.1; GLM, Wald χ2 = 89.97, df = 1, P < 0.0001). This tendency differed significantly among lemur species (Figure 4.1; GLM, Wald χ2 = 39.56, df = 23, P = 54

0.017); however, this result was driven mainly by the greater tendency of E. rufifrons to deposit seeds under conspecific trees (Appendix H). Seed dispersal rates under heterospecific, lemur-dispersed trees were similar across frugivore species (Appendix H).

40.00

- observed

expected

30.00

20.00

10.00

dispersed trees that are fruiting are that trees dispersed Percent of dispersal events under lemur under events dispersal of Percent 0.00 E. rubriventer E. rufifrons V. v. editorum Disperser species

We also examined the rate of seeds actively dispersed by lemurs into our seed traps, determined from seeds found in lemur scat. Our seed trap data supported our direct dispersal observations. The rate of lemur-mediated seed dispersal into a seed trap was significantly higher if there was a lemur-dispersed tree bearing fruits overhead (Figure

4.2A; Wald χ2 = 145.503, df = 1, P < 0.0001). The rate of seed dispersal by lemurs under lemur-dispersed tree species was significantly higher when the tree was fruiting (Figure

4.2B; Friedman’s test, χ2 = 5.23, df = 1, P = 0.02); but there was no significant difference in dispersal of seeds under non-lemur-dispersed tree species during fruiting vs. non- fruiting periods (Friedman’s test, χ2 = 1.67, df = 1, P = 0.20). 55

1 80 1.80 □fruiting □non-fruiting •k 1.50 rate 1.20 dispersal 090 seed mean 0.60 0.30

0.00 lemur-dispersed trees other trees trees overhead of seed traps Figure 4.2 – (A) More seeds are dispersed by lemurs into seed traps under their food trees that are fruiting than under other trees in the environment. (B) More seeds are dispersed by lemurs under lemur-dispersed trees when fruiting than non-fruiting, but this pattern does not hold for trees with other dispersal mechanisms. Error bars represent standard errors. *: statistical significance P < 0.05, n.s.: non-significant

4.3.2. Phylogenetic signal of lemur-mediated dispersal

Within our tree community of Ranomafana, we found a phylogenetic signal of having fruits dispersed by the three large-bodied lemurs in our study, which differed from the random expectation (D = 0.160, P < 0.001). In other words, closely related species were more likely to have fruit dispersed by lemurs than expected by chance. However, the phylogenetic distribution did not differ significantly from a Brownian model of evolution (P = 0.27), in which the phenotypic differences between species are proportional to the time since divergence.

4.3.3. Phylogenetic patterns of lemur-dispersal seeds and nearest adult neighbor

We examined the pattern of relatedness (NTI) between seeds dispersed by lemurs under heterospecific trees and the adult neighbor trees they fell under, and found that results were no different than expected by chance for each of the three lemur species (E. editorum riven rub ter 1 — (a) if] \ I E. — H-coot]— l-CH

56 H N>

co adult-tree i

rubriventer: mean = 0.003, t7 = 0.14, P I between I

= 0.99; E. rufifrons: mean = 0.02, t7 = dispersed

0.27, P = 0.80; V. v. editorum: mean = •— neighbor — □ D-jH and -0.17, t7 = -1.46, P = 0.19). When CD hDi relatedness INJ seed

conspecifics were included in the heterospecific — phylognetic analysis, the phylogenetic relationships >

of dispersed seeds with nearest adult adult-tree i □-CH dispersed n v. CO trees showed a pattern of phylogenetic m neighbor between

clustering for those dispersed by E. ► I— ht-»— i. (c) ► rufifrons and V. v. editorum, and a ► and II— relatedness ► O seed J ►»! heterospecific pattern no different than random ► phylognetic expectation for those dispersed by E V. I o o — •fH o o rubriventer (Appendix I). > N3 adult-tree 8 dispersed Since the dispersal network between O neighbor

changes over time with differing o relatedness

N) fruiting phenologies, we also looked at o i CO and heterospecific •+ I § b

the monthly temporal pattern of NTI of seed phylognetic § I

o dispersed seeds and their adult wmm > > ro c_ c_ c_

o neighbors. We found that this pattern

Figure 4.3 – Phylogenetic relationshipsO (NTI) i. O b

- between dispersed seeds (that do not fall under o o - varied over time, idiosyncratically for conspecifics) and their nearest adult neighbor trees fluctuate across months for each of the three each species, ranging from patterns of disperser species. Values < 0 indicate that the - species pairs are less closely related than expected overdispersed in some months to by chance; > 0 suggest that they are more closely related than expected by chance, and values near zero indicate a random relatedness. 57

clustered patterns in others (Figure 4.3).

4.3.4. Adult-tree distributions

We found that the tendency of lemur-dispersed trees to have nearest neighbors that are also lemur-dispersed with an overlapping fruiting time were no different than expected by chance (observed mean proportion: 0.498 ± SE 0.017, expected mean proportion: 0.496 ± SE 0.017; Friedman’s test χ2 = 0.346, df = 1, P = 0.78). However, we found that lemur-dispersed adult trees were more distantly related, than expected by chance, to their nearest neighbors that were dispersed by lemurs and had overlapping fruiting phenologies (mean NTI = -0.35, t17 = -3.56, P = 0.002).

4.4. Discussion

Since trees are sessile beyond the seed stage, patterns of spatial proximity at a local neighborhood scale set the stage for interactions between plants. Seeds represent the mobile stage, which, in many ecosystems, rely on frugivores for dispersal (Howe and

Smallwood 1982, Herrera 2002). Thus, one way that frugivores may impact plants, in a community context, is by structuring the spatial proximities of seeds they disperse with adult plants in the community. Increasing evidence has suggested that fruiting trees may be used as seed-dispersal foci by frugivores (Clark et al. 2004, Blendinger et al. 2011), which have a tendency to defecate seeds during subsequent feeding bouts. Despite the potential importance of nonrandom associations experienced by seeds for our understanding of community structure and diversity of plant communities (Terborgh et al

2002), the role frugivores play in structuring patterns of seed-adult associations in terms 58

of fruiting phenology, dispersal mode and evolutionary relatedness remain understudied, particularly at the community level. We showed that by biasing seed dispersal near fruiting trees upon which they fed, frugivores strongly influenced the phylogenetic relatedness of dispersed seeds to their nearest adult neighbors with which seeds will interact. However, we found that the observed patterns were not maintained through the adult stage. Interestingly, adult heterospecific neighbors of lemur-dispersed trees that shared dispersers and with overlapping timing of fruiting were more distantly related than expected by chance given the phylogenetic make-up and abundances of lemur-dispersed trees in the community. Future work on post-dispersal mechanisms may be important for better understanding patterns of spatial proximities within adult plant communities.

4.4.1. Seed-dispersal bias under fruiting feeding trees

The three frugivores in this study nonrandomly dispersed seeds toward fruiting trees where they were foraging, resulting in nonrandom associations of dispersed seeds with heterospecific co-fruiting plant species. By landing under a tree species that is in fruit, a seed is likely to experience a high density of heterospecific seeds. For the dispersed seed, this may be better than landing under a parent tree if it is a superior competitor or if seed predators are host-specific and target the seeds of the parent tree

(Kwit et al. 2007). However, this situation may also undermine the quality of dispersal because of increased attraction of natural enemies or competitive interactions (Schupp et al. 2002, Russo and Augspurger 2004, Kwit et al. 2007).

The spatial pattern of frugivore-mediated seed dispersal beneath fruiting trees may be a common phenomenon among frugivore-dispersed species because of the broad 59

diet of frugivores and the frequent overlap in timing of fruiting (Schupp et al. 2002, Kwit et al. 2004). This is especially important in tropical forests, where frugivores are the main seed dispersers (Howe and Smallwood 1982, Terborgh et al. 2002). Work by Clark et al.

(2004) in Dja Reserve, Cameroon found very different dispersal regimes for plants dispersed by birds and primates, with birds depositing a majority of seeds under fruiting trees, but not primates. In contrast, in our study, Malagasy primates biased seed dispersal toward fruiting trees in their diet. This may be due to their tendencies to defecate while feeding (Overdorff and Strait 1998). In addition, their broad diet (Razafindratsima et al.

2014) combined with a large proportion of time spent in fruiting trees and a directed movement between fruiting trees (OHR personal observation) may together affect the biased dispersal of heterospecific seeds under lemur-dispersed trees.

4.4.2. Phylogenetic patterns of seed dispersal

The nonrandom patterns of seed dispersal influenced the patterns of phylogenetic relatedness between pairs of dispersed seeds and associated neighbor adult-trees. We observed that seeds dispersed by E. rufifrons and V. v. editorum were more likely to fall under trees that were more closely related to the seeds, but not for seeds dispersed by E. rubriventer. The phylogenetic relatedness between seeds and their nearest neighbor may affect outcomes of their interactions, and by extension, plant recruitment and community structure, given that closely related species often share similar traits (Webb et al. 2006,

Gilbert and Webb 2007, Cavender-Bares et al. 2009). However, despite the increased recognition of the importance of evolutionary history in influencing plant recruitment

(Webb et al. 2006, Gilbert and Webb 2007), only a handful studies have explored such 60

effect. The limited work on this topic to date has shown contrasting results of the phylogenetic effects on seedling recruitment. For instance, Liu et al. (2012) found negative effects of phylogenetically similar neighboring adult trees on the survival of

Castanopsis fissa seedlings, a phenomenon they defined as “Phylogenetic Janzen-Connell effect”. However, Lebrija-Trejos et al. (2014) reported improved seedling survival near phylogenetically similar heterospecific neighbors for many species. Further research exploring the important role of phylogenetic relatedness in plant recruitment is much needed.

4.4.3. Adult-tree distribution

The primary dispersion of seeds is often recognized as an initial template for the establishment of plant communities (Nathan and Muller-Landau 2000, Levine and

Murrell 2003). This initial pattern might initiate subsequent life-history processes, and therefore affect the dynamics of tree recruitment, and indirectly determine the structure of plant populations and communities. Our results suggest that the patterns of dispersal generated by frugivores were not maintained at the adult stage. Co-fruiting lemur- dispersed trees in close proximity in each month were more distantly related than expected from a random draw of co-fruiting lemur dispersed trees in the same month.

This pattern might arise through density-dependent effects with closely related species

(i.e., phylogenetic Janzen-Connell effects) (Liu et al. 2012), acting on the important life- history processes that occur between dispersal and transition to adult plants; and it could significantly change the community structure template driven by frugivore seed dispersal. 61

4.5. Conclusions

Our results provide new perspectives regarding the roles of frugivores in mediating plant interactions and community structure. We also show an important link between frugivore foraging behavior and the spatial, temporal and phylogenetic patterning of seed dispersal. Further research is however needed to understand the potential connections between variations in phylogenetic structure and the dynamics of plant recruitment.

Chapter 5

Patterns of Trait Dispersion and Phylogenetic Structure of Co-fruiting Plant Assemblages

Abstract

Fruit and seed traits mediate mutualistic interactions with frugivorous seed dispersers and, therefore, may play a role in competition and facilitation of seed dispersal services between co-fruiting species within a community. However, there is little known about how these traits are structured in a phenological framework. Understanding how traits and evolutionary history of species in a community are structured via phenological processes can shed light on potential mechanisms important for community assembly.

We used a combined phylogenetic and trait-based approach, focusing on fruit and seed traits, to examine the temporal patterns of fruiting within a biodiverse plant community with year-round fruiting. We found that the fruit and seed trait distributions and phylogenetic patterns of fruiting varied over time, and depended on the temporal scale of 63

observation. At a temporal scale of fortnightly phenological observations, co-fruiting species tended to be more similar than expected by chance across all measured fruit and seed traits; however, the phylogenetic pattern of fruiting fluctuated non-randomly in time between clustered and overdispersed. When phenological observations were lumped into monthly fruiting observations, the overall clustered trait patterns and the non-random temporal patterns of trait and phylogenetic dispersion were lost. Our results suggest that it may be beneficial for co-fruiting plant species to share similar fruit and seed traits, which could be a result of facilitation or environmental filtering; however further work is required to determine mechanisms. These results also highlight the importance of a temporal scale in examining structural patterns of communities.

Keywords: competition, dispersal, facilitation, fruiting, fruit traits, Madagascar, phylogenetic signal, phylogenetic structure, plant phenology, tropical forest

5.1. Introduction

Fruit and seed traits are thought to play an important role in the attraction of frugivores, which act as main seed dispersers in many ecosystems (Whitney 2009,

Lomáscolo and Schaefer 2010, Schaefer 2011). These traits may therefore mediate competition and facilitation among co-fruiting plant species for seed-dispersal services

(Sargent 1990, Saracco et al. 2005, Aslan and Rejmanek 2012). Differences in traits among co-fruiting species may also be linked to similarities in environmental preferences or competitive abilities of stages post-dispersal. While some studies have examined the trait distributions of fruits in a community context (e.g., Mayfield et al. 2005, Ding et al. 64

2012), we know little about how fruit traits within a community are distributed across time. Because species interactions vary with phenology, considering a temporal context may be important for understanding mutualistic interactions, and processes and patterns of community structure (Weiher and Keddy 1995, Webb et al. 2002, Wiens et al. 2010).

Multiple and competing factors may influence similarities among fruit and seed traits, and thus can affect trait distributions within co-fruiting assemblages. Sharing fruit and seed traits with co-fruiting heterospecifics may result in increased competition for frugivores and reduced fruit removal for some species (Carlo et al. 2003, Saracco et al.

2005). Alternatively, fruit traits involving competitive ability to attract seed dispersers could also result in clustered trait dispersion (Chesson 2000, Mayfield and Levine 2010).

For example, among co-fruiting trees that share dispersers, coexistence may be more likely if their ability to attract the same frugivores differs little between species. Given that fruits tend to have generalized relationships with their animal mutualists and sharing seed dispersers does not have the same costs to fitness as other mutualistic interactions

(Blüthgen et al. 2007, Whitney 2009, Staggemeier et al. 2010), co-fruiting species may benefit from converging in fruit and seed traits, through facilitation or by attracting the same high-quality seed dispersers (Burns 2002, Whitney 2009). For example, some studies have shown that heterospecific co-fruiting neighbors sharing dispersers can enhance fruit removal (Saracco et al. 2005, Carlo and Morales 2008).Since frugivores have a tendency to disperse seeds while feeding (Chap 4, Clark et al. 2004, Carlo et al.

2007), sharing dispersers might reduce seed fall under conspecifics where distance- and density-dependent mortality is often high (Janzen 1970, Connell 1971). Sharing dispersers may also avoid extensive co-deposition of conspecific seeds, thus reducing 65

intraspecific competition among seedlings (Whitney 2009, Beltrán et al. 2012). In addition to of the mechanism, understanding patterns of trait dispersion among co- fruiting plant assemblages may have important implications for understanding plant community structure.

While often used as a proxy for trait similarities, phylogenetic patterns may or may not be reflective of patterns of trait dispersion within a community (Gerhold et al. in press); for example, if fruit and seed traits are labile and lack a phylogenetic signal or if there is evolutionary convergence in traits among distant lineages (Cavender-Bares et al.

2009). Therefore, combining trait-based and phylogenetic approaches is useful for studies of community structure. Previous studies that have examined community patterns of fruit and seed traits, without considering a temporal context, have shown that co-fruiting assemblages appear to have phylogenetic patterns that either do not differ from random, or have strong clustering patterns, suggesting a limited role of competition for seed dispersers in structuring plant communities (Staggemeier et al. 2010, Silva et al. 2011).

While these studies have provided important information about plant communities, it is possible that the patterns of traits and phylogenetic structure vary by season and across time because of the temporally dynamic environment and biotic interactions. Pooling all temporally interacting communities or averaging across time ignores that only species fruiting in synchrony can interact with each other via frugivores, and that species often have distinct temporal niches within a season (Silvertown 2004, Levin 2006). The temporal composition and abundance of seed dispersers and the seasonal variations in rainfall may also affect the composition of co-fruiting plant assemblages (van Schaik et al. 1993, Oberrath and Bohning-Gaese 2002, Staggemeier et al. 2010). 66

In this study, we examined the associations between traits and phylogenies to understand potential mechanisms underlying the structure of co-fruiting plant assemblages. Using a species-rich community of fleshy fruit plants in the southeastern rainforest of Ranomafana National Park (RNP), Madagascar, we specifically ask: (1) how fruit/seed traits and fruiting phenology are related to evolutionary history? (2) how do patterns of trait distributions, and evolutionary relatedness of co-fruiting plants vary over time?

5.2. Materials and methods

5.2.1. Study site and frugivore guild

This study was conducted in the evergreen montane rainforests of Ranomafana

National Park (RNP), located in southeastern Madagascar (47°18’- 47°37’E, 21°02’-

21°25’S), which comprises an area of 41,600 ha with an elevation of 600-1500 m

(Wright and Andriamihaja 2002). The climate in Ranomafana is highly seasonal with average monthly rainfall ranging from 10-1200 mm, a peak wet season in Jan-Mar, and a peak dry season in Jun-Oct (Dunham et al. 2011). Mean monthly temperature ranges from 4-32º C (Wright et al. 2011). RNP is home to more than 330 species of trees and large shrubs (Razafindratsima and Dunham 2015).

The frugivore assemblage in RNP consist of four large-bodied diurnal lemur species, two small-sized nocturnal lemur species, and seven bird species (Wright et al.

2011, Razafindratsima 2014). Three of these arboreal lemur species (Eulemur rubriventer, Eulemur rufifrons and Varecia variegata editorum) and two of the bird 67

species (Philepitta castanea and Alectronas madagascariensis) are known to act as seed dispersers in Ranomafana (Dew and Wright 1998, Rakotomanana et al. 2003, Wright et al. 2011, Razafindratsima 2014, Razafindratsima et al. 2014, Razafindratsima and

Dunham 2015).

5.2.2. Fruit trait data

For each plant species in RNP, we attempted to sample the following trait categories: color, size, reward and defense. We collected 5-30 individual fruits per species from multiple adult trees, depending on fruit availability.

Fruit colors are recognized as communicative traits enhancing the attraction of seed dispersers, because colors can indicate the nutritional contents of the fruits (Schaefer and Schmidt 2004, Cazetta et al. 2012). In this study, we described fruit color using three attributes (Hue, Value and Chroma) following Munsell’s color system for plant tissues

(Baltimore 1972, Wheelwright and Janson 1985, Villafuerte et al. 1998). “Hue” corresponds to 17 subdivisions of the four principal chromatic colors (i.e., red, yellow, blue and purple). Each subdivision of hue included either the principal color or a combination of these principal hues and a 2.5 step relative to the intensity of color; for example: 2.5 Yellow-Red, 5 Yellow-Red, 7.5 Yellow-Red, 10 Yellow-Red, 2.5 Green-

Yellow, etc. (Baltimore 1972). In our data analyses, the hue subdivisions was converted into a continuous variable relative to color wavelengths, such that pure red is represented by zero and colors of shorter wavelength (up to blue purple) are represented by increasing values (Touchon and Warkentin 2008). Color “Value” (0-10) indicates the level of lightness or darkness of a color relative to a neutral grayscale (hereafter referred to as 68

“lightness” to avoid confusion with the use of the word “value” in other context).

“Chroma” (or saturation) refers to the degree of departure of a given Hue from a neutral gray of the same value (zero for a neutral gray to 10, 12, 14 for more saturation).

For size, we measured length and width of fruits and seeds using a caliper to the nearest millimeter. We only considered these two dimensions because length and width are often the major constraints for frugivores to handling and swallowing seeds (Mazer and Wheelwright 1993, Forget et al. 2007). For some species for which we were unable to get fruit samples in the field, fruit size and color were recorded from the herbarium archives of the Missouri Botanical Garden (MBG in St-Louis), and from descriptive traits of genera in the Generic Tree Flora of Madagascar (Schatz 2001).

Fruit reward and defense were based on biochemical contents of the fruits, which often contain both palatable and unpalatable components (Cazetta et al. 2012). The pulp of ripe fruits was dried at 40-50°C either in a drying oven or in aluminum foil above a campfire in field. Samples were then stored individually in a plastic bag with a small sack of silica gel prior to analyses. Biochemical analyses were performed at the Institute of

Zoology, Department of Ecology and Conservation, University of Hamburg, Germany.

Samples were analyzed for nitrogen, soluble proteins, sugar, lipid, condensed tannins, polyphenols, and for neutral (NDF) and acid (ADF) detergent fiber following protocols described in Bollen et al. 2004a. The presence of alkaloids also was tested using

Dragendorff’s, Mayer’s and Wagner’s reagent (Cromwell 1955). The sample was considered to contain alkaloids when at least two of the reagents were positive (Ganzhorn

1988). Fruit nutrients associated with a reward for frugivores were composed of the amount of nitrogen, soluble proteins, sugar and lipid in the fruit. Fruit biochemical 69

defense was defined by the amount of tannins, polyphenols, fiber, and alkaloids, which may act as feeding deterrents and thus may have negative effects on food selection by frugivores (Schaefer et al. 2003, Bollen et al. 2004, 2005, Cazetta et al. 2007).

5.2.3. Dispersal mode

We inferred the dispersal mode of each plant species, as bird-dispersed (i.e., being primarily dispersed by birds), lemur-dispersed (i.e., being primarily dispersed by lemurs), and with mixed dispersal syndrome (i.e., dispersed by both lemurs and birds), based on the identity of frugivore consuming their fruits. Consumer identity were from data of direct observations of feeding and defecation from literature (Chap 4, Overdorff 1993,

White et al. 1995, Dew and Wright 1998, Rakotomanana et al. 2003, Razafindratsima et al. 2014) and from long-term knowledge of the local technicians familiar with the fauna and flora in RNP. Since animal-dispersed species are usually defined as having edible parts that encourage swallowing by frugivores (Howe and Smallwood 1982, Seidler and

Plotkin 2006), plant species with fruits consumed by either of both groups of frugivores were considered as being dispersed. We initially separated animal-dispersed plant species from those that are dispersed by abiotic means by examining the fruit morphology and any derivation (e.g., presence of plumes and wings, dehiscence, absence of edible aril or pulp) (Howe and Smallwood 1982, Seidler and Plotkin 2006).

We analyzed differences in traits among the three major dispersal modes, using nonparametric, permutational multivariate analysis, PERMANOVA (McArdle and

Anderson 2001, Anderson 2001), based on Bray-Curtis similarity metrics using PRIMER v7 (Clarke and Gorley 2015). Permutations were run 999 times. To visualize the 70

separation of species relative to different dispersal mode, we used two axes of nonmetric multidimensional scaling (nMDS) plots of traits using functions in the ‘vegan’ package in

R 3.0.3 (R Core Team 2012). Species with any missing trait value were excluded from these analyses.

5.2.4. Monitoring of fruiting phenology

We monitored fruiting phenology in two sites within the southern forest in RNP

(Valohoaka and Vatoharanana; see map in Figure 1.1). In each site, we set-up 10 transects of 100 m long and 4 m wide, which were spaced at least 100 m apart. All individual canopy trees (≥ 5 cm Diameter at Breast Height, DBH) and understory trees (≥

2 cm DBH) were identified in the field by local research technicians familiar with the local flora, and marked with numbered aluminum plant-tags. In total, we marked 3,320 individuals of 155 species belonging to 43 families and 89 genera (see Appendix I).

Monitoring occurred every two weeks for two years (July 2012 - June 2014), during which we recorded the presence / absence of immature and mature fruits for each tagged individual (Morellato et al. 2000, Cortés-Flores et al. 2013). Co-fruiting plant assemblages were defined at two scales: (1) at the time scale of our phenological sampling, which recorded co-fruiting species every 2 weeks, i.e. species in an assemblage were actually fruiting at the same time, (2) at a larger timescale that groups all fruiting species within each month as sample units, in which species that use the same seasonal resources are potentially fruiting together. The co-fruiting assemblages were composed of

7-37 plant species. 71

To analyze fruiting phenology, we performed circular statistics in MATLAB

R2013b (The MathWorks Inc., Natick, MA) using CircStat toolbox (Berens 2009).

Months were first converted into angles so that the year is represented as a circle of 360º with an arbitrary origin (i.e. Jul, 0º or 360º; Aug, 30º; Sep, 60º etc.). We then calculated the average angle α, which corresponds to the point in year where there is a peak in fruiting. The length of the mean vector r, which is a concentration around the mean, was also calculated; the closer the value of r is to 1, the more concentrated the data sample is around the mean direction. We used the Rayleigh test (Zar 1999) to test for the statistical significance of α, (i.e., whether fruiting is uniformly distributed around the circle). We did the analyses both with the number of species fruiting and with the number of individual trees fruiting, for each dispersal mode and for all of them combined. The abundances of species and individual trees fruiting used in these circular statistics were averaged across years. We also measured the average duration of fruiting for each plant species.

5.2.5. Phylogenetic tree reconstruction

We built a local master phylogeny with all plant species in Ranomafana, using

Phylomatic v3 (Webb and Donoghue 2005), which is an online interface used to reconstruct a phylogeny based on a set of species taxonomic names defined by the user.

We used the Angiosperm Phylogeny Group III (AGP3) derived megatree R20120829

(Webb and Donoghue 2005) as a base tree to build the plant phylogeny used in this study.

Branch lengths of our phylogenetic tree were estimated using the bladj algorithm 72

implemented in Phylocom 4.2 (Webb et al. 2008) with estimated molecular and known fossil ages (Wikström et al. 2001).

5.2.6. Phylogenetic signal

To determine how phenology and fruit traits are related to evolutionary history, we tested for a phylogenetic signal in phenology and in each of the fruit trait categories

(color, fruit size, seed size, reward and defense). A phylogenetic signal is the tendency of related species to resemble each other more than species drawn at random from the phylogenetic tree (Lord et al. 1995, Prinzing 2001). To test for a phylogenetic signal, we used pruned phylogenies of the master tree, containing only species for which we had traits. We reduced multiple variables of each trait types and the phylogeny variables

(peak of fruiting, concentration of fruiting around the peak and duration of fruiting) into uncorrelated variables using PCA. We retained principal components with an eigenvalue

> 1 (Peres-Neto et al. 2005). For fruit biochemical defense, which contained both discrete and continuous variables, we performed a categorical PCA (CATPCA) instead of a PCA.

We used the resulting PC scores to test for phylogenetic signal, which was estimated with

Pagel’s λ (Pagel 1999) using the geiger package in R 3.0.3 (R Core Team 2012). Pagel’s

λ is a scaling parameter for the correlations between species, assessing the phylogenetic signal in trait against a Brownian motion model from 1000 randomizations. Value of λ equals 0 if there is no phylogenetic signal, and approaches 1 with increasing phylogenetic pattern (Pagel 1999, Fritz et al. 2009).

We also tested for phylogenetic signal for dispersal modes (bird-dispersed, lemur- dispersed and mixed) by calculating Fritz and Purvis’s statistic D for binary variable 73

(Fritz and Purvis 2010) from R package caper, with 1000 permutations. The statistic D is based on the sum of sister clade differences, and compares trait evolution with a

Brownian motion model. A value of D = 1 indicates random trait distribution across the tips of the phylogeny, and D = 0 indicates a Brownian motion mode of evolution.

Decreasing value of D from 1 indicates increasing phylogenetic clumping such that the estimate of D varies around 0 for highly phylogenetically clumped trait. This method also tests for significant departure of D from 1 (random expectation, i.e. D < 1 suggests significant phylogenetic signal) and from Brownian threshold model.

5.2.7. Trait dispersion among co-fruiting plant assemblages

For each trait type (color, size, seed size, reward, and defense), we measured its dispersion within each co-fruiting assemblage using a measure of functional dispersion

(FDis, Laliberté and Legendre 2010). FDis computes the mean distance of a species in a multidimensional trait space from the mean of all species, in which the space is determined by a set of traits (Laliberté and Legendre 2010, Ding et al. 2012, Chamberlain et al. 2014). To determine whether the FDis of the co-fruiting assemblage was different from the null expectation, we compared the observed FDis with mean FDis of 1000 randomly generated assemblages from a pool of all species with known traits across all assemblages. We accounted for differences in sample size between assemblages by standardizing the metric; i.e. subtracting the mean distance of the null assemblages and dividing by the standard deviation. The standardized effect sizes of the FDis for each co- fruiting assemblages were calculated in R (R Core Team 2012) with the package “FD” 74

(Laliberté and Legendre 2010) and using the following equation (Ding et al. 2012):

ZFDis = -1 x [(FDisobs- FDisnull)/sd(FDisnull)]

The subscript obs and null refer to the observed means for our assemblages and the means for null assemblages respectively. A negative ZFDis indicate that the dispersion of traits among co-fruiting species is larger than expected by chance

(evenness), while a positive value indicate that the trait dispersion is smaller than expected by chance (clustering).

5.2.8. Phylogenetic structure of co-fruiting plant assemblages

We used two indices to quantify the community-wide phylogenetic dispersion of species that have overlap in fruiting time: nearest taxon index (NTI) and net relatedness index (NRI) (Webb et al. 2002). NTI is a standardized effect size for the mean nearest phylogenetic neighbor distance (MNND) for all species in each community; and NRI is a standardized effect size of the mean pairwise phylogenetic distance (MPD) for all species in each community. We calculated NTI and NRI in Phylocom 4.2 (Webb et al. 2008) using the following equations: NTI = -1 x [(MNNDobs - MNNDnull)/sd(MNNDnull)] and

NRI = -1 x [(MPDobs - MPDnull)/sd(MPDnull)].

MNNDobs and MPDobs correspond to the observed means calculated for our communities, while MNNDnull and MPDnull are means for null communities; sd refers to the standard deviations of the null distribution. To generate null communities, we used the null model #1 in Phylocom, in which case all species present in at least one assemblage were used as the species pool, and the species richness of each assemblage 75

was maintained. For each assemblage, species names were randomly shuffled without replacement 999 times, and the values of MNND and MPD were calculated each time for each null assemblage. Negative values of NTI or NRI indicate that co-fruiting species are less phylogenetically related than expected by chance (‘phylogenetic overdispersion’), whereas positive values of NTI or NRI indicates phylogenetic clustering (i.e. co-fruiting species are more phylogenetically related than expected by chance) (Webb et al. 2002).

5.2.9. Test of randomness for trait and phylogenetic metrics

We used serial runs test (Zar 1999) to assess temporal independence of ZFDis, NRI, and

NTI indices over time, testing the null hypothesis that the distribution of these metrics over time is random. If the tendency for clustering or evenness/overdispersion exists in more than one unit of time, the runs test will show significant difference from randomness (Santorelli et al. 2014).

5.3. Results

5.3.1. Fruit traits and seed-dispersal strategies

The three major dispersal modes significantly differed in fruit and seed traits

(PERMANOVA; pseudo-F2,81 = 4.085, P = 0.008; Error! Reference source not found.).

The nonmetric multidimensional scaling (nMDS) analysis on the variation of fruit traits relative to the different dispersal modes (Error! Reference source not found.) show that the first nMDS axis was positively related to fiber content in fruits (percentages of ADF:

R2 = 0.519; and NDF: R2 = 0.4673) and negatively correlated with protein content (R2 = 76

0.3809). The second axis was negatively correlated with size (seed length R2 = 0.4958, fruit width R2 = 0.4890, fruit length R2 = 0.4077, seed width R2 = 0.4214).Visual representation of the nMDS plots revealed a cluster of points for bird-dispersed species and more dispersed patterns for lemur-dispersed species (Error! Reference source not found., as represented by the centroid of the trait distribution with an ellipse of standard error at 0.95 confidence). The traits associated with lemur-dispersed species were largely different from bird-dispersed species; however bird-dispersed species were nested within the wide spectrum of size distribution associated with lemur dispersal. Bird-dispersal tree species appear to be smaller in size, richer in sugar content and lower in protein, fiber and polyphenols than lemur-dispersed species (Table 5.1). Mixed dispersal mode showed an overlap in traits with bird- and lemur-dispersed species.

2 MDS

MDS 1 Figure 5.1 – Differences in fruit and seed traits among the three major known dispersal mode. This figure presents two axes of nonmetric multidimensional scaling (nMDS) plots. Each point correponds to one species, and the ellipse polygons represents standard error with 0.95 confidence limit 77

Dispersal mode Traits bird lemur mixed color hue 4.89 ± 3.35 7.33 ± 3.95 7.00 ± 5.43 color chroma 4.89 ± 1.73 5.38 ± 1.72 4.50 ± 1.41 color lightness 5.33 ± 3.4 6.25 ± 2.68 6.88 ± 3.08 fruit length 7.21 ± 3.12 23.65 ± 19.38 12.66 ± 7.18 fruit width 7.38 ± 3.31 18.82 ± 13.08 12.20 ± 7.28 seed length 5.45 ± 2.74 12.64 ± 6.64 8.97 ± 5.01 seed width 3.70 ± 2.13 9.29 ± 6.06 6.96 ± 5.01 nitrogen 1.37 ± 0.43 1.37 ± 0.67 1.37 ± 0.61 soluble protein 3.06 ± 2.19 3.36 ± 1.98 3.57 ± 2.03 Sugar 22.22 ± 17.80 9.48 ± 9.36 12.43± 12.75 Lipids 3.74 ± 4.54 6.91 ± 7.54 3.15±2.78 NDF % 39.11 ± 13.85 41.40 ± 14.43 42.99±11.84 ADF % 29.78 ± 13.29 32.19 ± 12.71 31.64±12.30 condensed tannins 0.06 ± 0.13 0.15 ± 0.37 0.18±0.37 polyphenols 0.75 ± 0.76 1.32 ± 1.62 1.36±1.15

Among species with known disperser mode within our transects, zoorchorous plant species accounted for 82.44% (n = 131) of the species; of which 8.33% are primarily dispersed by birds, 74.07% dispersed by lemurs, and 17.59% dispersed by both birds and lemurs. Looking at individual trees in transects, on average 91% (n = 3087) were zoochorous, among which 6.80% were bird-dispersed, 67.78% were lemur-dispersed and

25.42% were with mixed dispersal mode.

5.3.2. Fruiting phenology

Overall, fruiting occurred year-round (Figure 5.2). Looking at the number of species fruiting per month, fruiting was uniformly distributed across months without 78

significant peaks of fruiting (a, Rayleigh test p = 0.84). We also did not find any

significant peak of fruiting for any dispersal mode (b-d, Table 5.2). However, when

considering the number of individual trees fruiting per month, the distributions of fruiting

were not uniform across months and were significantly unimodal with a peak occurring

in mid-April (e-h, Table 5.2). Species dispersed by birds had a peak of fruiting at the end

of April, lemur seed-dispersed species had a peak of fruiting at early May, and species

with mixed seed-dispersal syndrome had a peak of fruiting in early April.

Overall Bird-dispersed Lemur-dispersed Mixed

(a) (b) (c) (d) Jul Jul Jul Jul Jun 40 Aug Jun 20 Aug Jun 20 Aug Jun 20 Aug 30 15 15 15 May 20 Sep May 10 Sep May 10 Sep May 10 Sep 10 5 5 5 Apr 0 Oct Apr 0 Oct Apr 0 Oct Apr 0 Oct

Mar Nov Mar Nov Mar Nov Mar Nov Feb Dec Feb Dec Feb Dec Feb Dec Jan Jan Jan Jan

# species # with fruits

(e) (f) (g) (h) Jul Jul Jul Jul Jun600 Aug 400 Aug Jun400 Aug Jun400 Aug 400 300 300 300 May Sep May 200 Sep May 200 Sep May 200 Sep 200 100 100 100 Apr 0 Oct Apr 0 Oct Apr 0 Oct Apr 0 Oct

Mar Nov Mar Nov Mar Nov Mar Nov Feb Dec Feb Dec Feb Dec Feb Dec Jan Jan Jan Jan # trees with trees # fruits

Figure 5 .2 – Absolute frequency distribution of fruiting species (a-d) and individual trees (e-h) (averaged across years), considering all species (a) and all individuals (e), and by dispersal syndromes (b,f: bird-dispersed; c,g: lemur-dispersed; and d,h: mixed).

Table 5.2 – Results of circular statistics on fruiting phenology of different dispersal syndromes. The mean vector α represents the peak date of fruiting, r indicates the concentration of fruiting around mean (close to 1 means high concentration), and the values of Z and p-value of Rayleigh test, which tests the null hypothesis that fruiting is uniformly distributed around year Syndrome Mean Standard vector r Rayleigh p-value (α, in degrees) deviation Z all 278.3384 78.97 0.05 0.835 0.434 Based on bird 273.7807 73.33 0.18 1.557 0.212 number of species fruiting lemur 82.36926 80.86 0.00 0.003 0.997 mixed 284.468 76.01 0.12 1.225 0.295

all 257.6127 64.38 0.37 359.275 < 0.001 Based on bird 283.1601 59.14 0.47 70.520 < 0.001 number of trees fruiting lemur 292.9844 74.04 0.17 23.252 < 0.001 mixed 247.4557 55.97 0.52 385.301 < 0.001

5.3.3. Phylogenetic signal

We used principal component scores resulting from PCA of each trait type to examine the phylogenetic signal for each trait; variations explained by each retained principal scores are detailed in Table 5.3. For fruit color, PC1 explained 43.23% of the variance and was most correlated with hue (load: 0.854) and lightness (load: -0.729); PC2 explained 38.45% of the variance, with Chroma as the largest loading (0.905). For fruit size, both length and width had the same loadings (0.936). Likewise, seed length and seed width had the same loadings (0.964). For fruit reward, PC-1 explained 38.18% of the variance with percentage of lipid as the largest loading (0.732) followed by nitrogen

(0.643) and sugar (-0.628); for PC-2 the largest loading was protein (0.810) followed by nitrogen (-0.525).

We found no significant phylogenetic signal in fruiting phenology (Table 5.3), indicating that fruiting phenology is not phylogenetically conserved. For fruit traits, closely related species tended to be more similar than expected by chance in color 80

chroma (color PC2), seed size and fruit rewards (both principal components), but not in color PC1, fruit size and defense (Table 5.3). For dispersal mode, we also found significant phylogenetic signals of being dispersed by birds (D = 0.179, p < 0.0001), lemurs (D = 0.529, p < 0.0001), and both (D = 0.708, p = 0.01) compared to random expectations. When compared with Brownian model, these phylogenetic signals remained significant for lemur-dispersal (p = 0.014) and mixed mode (p = 0.001), but not for bird-dispersed (p = 0.37).

Component for each PCA results Phylogenetic signal trait type Eigenvalue % variance λ p-value Phenology PC-1 1.717 57.22 0.089 0.41 Fruit color PC-1 1.297 43.22 0.040 0.80 Fruit color PC-2 1.154 38.45 0.201 0.03 Fruit size PC-1 1.751 87.57 < 0.0001 1.00 Seed size PC-1 1.860 92.98 0.885 < 0.001 Fruit reward PC-1 1.488 37.20 0.355 < 0.001 Fruit reward PC-2 1.126 28.15 0.340 0.002 Fruit defense PC-1 2.209 44.18 0.112 0.27 Fruit defense PC-2 1.472 29.44 < 0.0001 1.00

5.3.4. Patterns of trait dispersion of co-fruiting assemblages

The patterns of trait dispersion of co-fruiting plant assemblages were observed to fluctuate over time, but differed between the two temporal scales of observation (Figure

5.3). At the time scale of our phenological sampling, which recorded co-fruiting species every 2 weeks, we found a predominant pattern of clustering of traits for each trait 81

category (Figure 5.3A). In all cases, the distribution of the values of ZFDis was significantly non-random with respect to time. However, looking at a larger timescale, in which all species fruiting within the same month were pooled, we found that the tendencies toward overall clustering of traits were lost, and overtime trait distributions fluctuated from clustered to even (Figure 5.3Error! Reference source not found.B). All time-series patterns of trait dispersion values at this scale were indistinguishable from a random series, however, tests for the dispersion of fruit size, seed size and biochemical defense were each marginally non-significant at the monthly time scale.

5.3.5. Phylogenetic structure of co-fruiting assemblages

The patterns of phylogenetic structure of co-fruiting plant assemblages were observed to fluctuate over time from clustered to overdispersed for both temporal scales.

At the smallest time scale (Figure 5.4A), we rejected the null hypothesis that the distribution of NRI values over time was random but not for NTI, which measures phylogenetic dispersion at the tips of the phylogeny. When examining monthly sets of co- fruiting species (Figure 5.4B), we found a marginally non-significant difference from random for the distribution pattern of NRI while the NTI series did not differ from random. 82

Fruit color Fruit size Seed size Nutrient reward Biochemical defense Dispersal mode

(A)

2 ZFDis

0 Values of Values -2

p < 0.0001 p = 0.001 p = 0.001 p < 0.0001 p < 0.0001 p < 0.0001

(B)

0 ZFDis

-2 Values of Values -4 p = 0835 p = 0.060 p = 0.060 p = 0.287 p = 0.060 p = 0.287

Figure 5.3 – Trait dispersion over time of co-fruiting assemblages, as represented by the standardized effect size of functional diversity (ZFDis) for each fruit trait type, from clustered (positive values) to even (negative values). (A) 2-week sampling period (B) monthly fruiting assemblages. P-values correspond to runs test used to assess temporal independence of the trait-dispersion index. Observations were from July 2012-June 2014. 83

NRI NTI

(A) 3.0

1.5

0.0

Index values Index -1.5

-3.0

p < 0.0001 p = 0.890

(B) 3.0

1.5

0.0

Index values Index -1.5

-3.0 J A S O N D J F M A M J J A S O N D J F M A M J J A S O N D J F M A M J J A S O N D J F M A M J 2012 2013 2014 2012 2013 2014

p = 0.060 p = 0.297

5.4. Discussion

Fruit and seed traits often play an important role as communicative signals for the

attraction of animal mutualists, and may mediate competition or facilitation for seed

dispersal services among co-fruiting plant species within a community (Schaefer et al.

2004, Whitney 2009). While some studies have examined how fruit and seed traits are 84

distributed among co-fruiting species within a community, we know relatively little about trait dispersion in a temporal framework. A temporal context, in which we examine co- fruiting within certain time windows, can help us identify mechanisms and processes underlying the structure of co-fruiting assemblages within a community given that interacting species vary over time. Examining a community of plant species dispersed by frugivores in the rainforest of Ranomafana National Park, Madagascar, we found that the patterns of trait distribution and phylogenetic structure of co-fruiting assemblages varied over time, and were influenced by the temporal scales used. Our findings suggest that co- fruiting species may be benefit from sharing similar fruit and seed traits, probably as a result of facilitation or environmental filtering. Results also suggest that the mechanisms and processes driving species interactions and coexistence and affecting patterns of biodiversity do not depend only on spatial scales, but depend on temporal scales as well.

At the timescale of our fortnightly phenological sampling, we found overall clustered patterns of fruit and seed trait dispersion among co-fruiting plant species. This suggests that species with similar traits would set fruits in synchrony to attract the same seed dispersers, which may facilitate and increase fruit removal (Burns 2002, Saracco et al. 2005). Alternatively, such clustered patterns may also be a result of environmental filtering, such that co-fruiting species share tolerances to abiotic factors (Sargent and

Ackerly 2008, Cavender-Bares et al. 2009). Interactions with antagonists could also drive trait similarities among co-fruiting species, because fruit traits can also serve as cues to attract or defenses against natural enemies (Whitney and Stanton 2004, Schaefer et al.

2004), and future work is needed to better understand how enemies might structure fruit traits. 85

The presence of strong phylogenetic signal in seed-dispersal mode and in some fruit and seed traits may also result in clustering of trait dispersion, despite the absence of phylogenetic signal in the timing of fruiting. Similar to other studies, we detected significant phylogenetic signals in seed size, and nutritional reward (e.g., Lord et al.

1995, Cazetta et al. 2012), and low phylogenetic signal for color hue, color lightness, fruit size, or biochemical defense(e.g., Lomáscolo and Schaefer 2010, Cazetta et al. 2012,

Harrison et al. 2012, Stournaras et al. 2013). Our results also support other findings suggesting that phylogenetic constraints on fruit color hue are weak, suggesting that labile color traits may have evolved in response to the selective pressures of seed dispersers (Voigt et al. 2004, Lomáscolo et al. 2010, Lomáscolo and Schaefer 2010,

Cazetta et al. 2012). However, we did not find significant differences between the color traits of fruits dispersed by birds and lemurs, despite the fact that lemurs are dichromatic and can differentiate only blues and greens but not reds (Jacobs and Deegan 2003), whereas birds have trichromatic vision (Emmerton and Delhis 1980). This suggests that color hue may also be linked to environmental filters and/or with defense against natural enemies (Whitney and Stanton 2004, Schaefer et al. 2004).

Examining co-fruiting within monthly time windows (i.e., all species fruiting within the same month were pooled), we found that the clustered patterns of trait dispersion of co-fruiting plant assemblages were lost; but over time, the patterns fluctuated idiosyncratically from clustered (i.e., more similar than expected by chance) to even (i.e., less similar than expected by chance). These fluctuations may be a result of different seasonal biotic and abiotic pressures or stochastic processes in a community that has variable fruiting across years (Wright et al. 2005) and extreme variability in weather 86

patterns (Dunham et al. 2011, 2013). These fluctuating patterns of trait dispersion among co-fruiting species may have important implications for seed dispersal services received by heterospecific co-fruiting neighbors. Thus, fruiting at the same time as another species may have different fitness consequence than fruiting at different times of the year (Carlo et al. 2007). For instance, frugivores have a tendency to disperse seeds beneath fruiting trees upon which they fed (Chap 4, Clark et al. 2004, Carlo et al. 2007); and thus, co- fruiting species sharing dispersers may influence each other’s seed dispersal services.

We also found that the phylogenetic patterns of co-fruiting fluctuated over time, but varied between the phylogenetic indices used. When examining the tips of the phylogeny (NTI), we found that relationships of co-fruiting assemblages across time tend to not differ from random expectation, probably as a result of the lack of phylogenetic signal in fruiting phenology. However, when looking at the base of the tree (NRI), we found non-random pattern of phylogenetic structures over time. These results suggest that the families that have taxa fruiting are either more similar or more divergent than expected by chance, but that within those families, the pattern does not differ from random. Using a temporally explicit approach, the phylogenetic structures fluctuated over an annual cycle ranging from clustered (i.e., co-fruiting species were more closely related species than expected by chance) to overdispersed (i.e., co-fruiting species were more distantly related species than expected by chance) for both metrics. These fluctuating patterns of phylogenetic structure of co-fruiting assemblages can have critical consequences for plant recruitment if closely related species tend to share environmental tolerances related to germination and early survival (Burns and Strauss 2011). For example, if seeds are dispersed near closely related adult-tree species during a certain 87

time window, in which co-fruiting assemblages are constituted of more closely related species than expected by chance (i.e., clustered phylogenetic structure), the recruitment probability may be low (Liu et al. 2012).

The lack of phylogenetic signal in fruiting phenology we observed for plant species in Ranomafana is consistent with previous studies on evolution of fruiting phenology (Staggemeier et al. 2010, Silva et al. 2011). This finding supports recent theoretical work that predicts evolutionary lability of phenological traits in the tropics because of the broad temporal niche axis afforded by less extreme seasonality (Pau et al.

2011). Competition for seed disperser services (Wheelwright 1985) or factors affecting post dispersal mortality such as natural enemies (Whitney 2009) could then drive divergence in fruiting times. It has also been suggested that closely related species sharing dispersers will benefit from divergent fruiting times because it would help maintain disperser populations (Staggermeier et al 2010); however, since selection acts on individuals, this hypothesis remains controversial.

Conclusions and implications

Our findings suggest that phenology can be an important component for structuring communities. Studies examining the phylogenetic and trait structures related to fruiting have not historically considered how these patterns might change in a temporal context (e.g., Staggemeier et al. 2010, Silva et al. 2011). The patterns of community structure may vary with the timing and temporal scale of observations. This could explain some of the variation in patterns of community structure across studies. The temporal context is particularly important in tropical forests, where fruiting occurs year 88

round (van Schaik et al. 1993, Oberrath and Bohning-Gaese 2002). Our findings also open new venues for further studies to address the combined importance of ecological and evolutionary factors for plant interactions, recruitment and community structure.

References

Albert, A., A. Hambuckers, L. Culot, T. Savini, and M.-C. Huynen. 2013. Frugivory and

seed dispersal by Northern Pigtailed Macaques (Macaca leonina), in Thailand -

Springer. International Journal of Primatology 34:170–193.

Anderson, M. J. 2001. A new method for non-parametric multivariate analysis of

variance. Austral Ecology 26:32–46.

Baak, E. 2005. To succeed globally, disperse locally: effects of local pollen and seed

dispersal on tetraploid establishment. Heredity 94:538–546.

Baden, A. L. 2011. Communal breeding in ruffed lemurs. PhD dissertation, Stony Brook

University.

Baden, A. L., R. A. Brenneman, and E. E. Louis. 2008. Morphometrics of wild black-

and-white ruffed lemurs [Varecia variegata; Kerr, 1792]. American Journal of

Primatology 70:913–926.

Bagchi, R., M. C. Press, and J. D. Scholes. 2010. Evolutionary history and distance

dependence control survival of dipterocarp seedlings. Ecology Letters 13:51–59.

Barrett, M. A., and J. Ratsimbazafy. 2009. Luxury bushmeat trade threatens lemur

conservation. Nature 461:470–470.

Bascompte, J., and P. Jordano. 2007. Plant-animal mutualistic networks: The architecture

of biodiversity. Annual Review of Ecology, Evolution, and Systematics 38:567–

593. 90

Beckman, N. G., and H. S. Rogers. 2013. Consequences of seed dispersal for plant

recruitment in tropical forests: interactions within the seedscape. Biotropica

45:666–681.

Birkinshaw, C. 2001. Fruit characteristics of species dispersed by the black Lemur

(Eulemur macaco) in the Lokobe Forest, Madagascar. Biotropica 33:478–486.

Bleher, B., and K. Böhning-Gaese. 2001. Consequences of frugivore diversity for seed

dispersal, seedling establishment and the spatial pattern of seedlings and trees.

Oecologia 129:385–394.

Blendinger, P., J. Blake, and B. Loiselle. 2011. Composition and clumping of seeds

deposited by frugivorous birds varies between forest microsites. Oikos 120:463–

471.

Bollen, A., G. Donati, J. Fietz, D. Schwab, J.-B. Ramanamanjato, L. Randrihasipara, L.

van Elsacker, and J. Ganzhorn. 2005. An intersite comparison of fruit

characteristics in Madagascar: evidence for selection pressure through abiotic

constraints rather than through co-evolution. Pages 93–119 in J. L. Dew and J. P.

Boubli, editors. Tropical Fruits and Frugivores. Springer Netherlands.

Bollen, A., L. Van Elsacker, and J. U. Ganzhorn. 2004. Tree dispersal strategies in the

littoral forest of Sainte Luce (SE-Madagascar). Oecologia 139:604–616.

Boyer, D., G. Ramos-Fernández, O. Miramontes, J. L. Mateos, G. Cocho, H. Larralde, H.

Ramos, and F. Rojas. 2006. Scale-free foraging by primates emerges from their

interaction with a complex environment. Proceedings of the Royal Society B:

Biological Sciences 273:1743–1750. 91

Brodie, J. F., O. E. Helmy, W. Y. Brockelman, and J. L. Maron. 2009. Functional

differences within a guild of tropical mammalian frugivores. Ecology 90:666–

681.

Burns, J. H., and S. Y. Strauss. 2011. More closely related species are more ecologically

similar in an experimental test. Proceedings of the National Academy of Sciences

108:5302–5307.

Cabre-Vert, N., and A. Feistner. 1995. Comparative gut passage time in captive lemurs.

Dodo Journal of Wildlife Preservation Trusts 31:76–81.

Cain, M. L., B. G. Milligan, and A. E. Strand. 2000. Long-distance seed dispersal in plant

populations. American Journal of Botany 87:1217–1227.

Carlo, T. A. 2005. Interspecific neighbors change seed dispersal pattern of an avian-

dispersed plant. Ecology 86:2440–2449.

Carlo, T. A., and J. E. Aukema. 2005. Female-directed dispersal and facilitation between

a tropical mistletoe and a dioecious host. Ecology 86:3245–3251.

Carlo, T. A., J. E. Aukema, and J. M. Morales. 2007. Plant-frugivore interactions as

spatially explicit networks: integrating frugivore foraging with fruiting plant

spatial patterns. Pages 369–390 in A. J. Dennis, R. J. Green, E. W. Schupp, and

D. A. Westcott, editors. Seed dispersal: theory and its application in a changing

world. CABI International, Oxford, UK.

Cavender-Bares, J., K. H. Kozak, P. V. A. Fine, and S. W. Kembel. 2009. The merging of

community ecology and phylogenetic biology. Ecology Letters 12:693–715.

Chapman, C. A. 1995. Primate seed dispersal: coevolution and conservation implications.

Evolutionary Anthropology: Issues, News, and Reviews 4:74–82. 92

Chapman, C. A., and S. E. Russo. 2006. Primate seed dispersal: Linking behavioral

ecology with forest community structure. Pages 510–525 in J. C. Campbell, A. F.

Fuentes, K. C. MacKinnon, M. Panger, and S. Bearder, editors. Primates in

perspective. Oxford University Press, Oxford.

Clark, C. J., J. R. Poulsen, E. F. Connor, and V. T. Parker. 2004. Fruiting trees as

dispersal foci in a semi-deciduous tropical forest. Oecologia 139:66–75.

Clarke, D. R., and R. Gorley. 2015. PRIMER 7. Plymouth Marine Laboratory, Plymouth,

UK.

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 P. J. Den Boer

and G. Gradwell, editors. Dynamics of populations. PUDOC, Wageningen.

Connor, E. F., and D. Simberloff. 1979. The assembly of species communities: chance or

competition? Ecology 60:1132–1140.

Cordeiro, N. J., and H. F. Howe. 2003. Forest fragmentation severs mutualism between

seed dispersers and an endemic African tree. Proceedings of the National

Academy of Sciences 100:14052–14056.

Cottam, G., and J. T. Curtis. 1956. The use of distance measures in phytosociological

sampling. Ecology:451–460.

Cousens, R. D., J. Hill, K. French, and I. D. Bishop. 2010. Towards better prediction of

seed dispersal by animals. Functional Ecology 24:1163–1170.

Crowley, B. E., L. R. Godfrey, and M. T. Irwin. 2011. A glance to the past: subfossils,

stable isotopes, seed dispersal, and lemur species loss in southern Madagascar.

American Journal of Primatology 73:25–37. 93

Dewar, R. E., and A. F. Richard. 2012. Madagascar: A History of Arrivals, What

Happened, and Will Happen Next. Annual Review of Anthropology 41:495–517.

Dew, J. L., and P. Wright. 1998. Frugivory and seed dispersal by four species of primates

in Madagascar’s eastern rain forest. Biotropica 30:425–437.

Donatti, C. I., P. R. Guimarães, M. Galetti, M. A. Pizo, F. M. D. Marquitti, and R. Dirzo.

2011. Analysis of a hyper-diverse seed dispersal network: modularity and

underlying mechanisms. Ecology Letters 14:773–781.

Dunham, A. E., E. M. Erhart, D. J. Overdorff, and P. C. Wright. 2008. Evaluating effects

of deforestation, hunting, and El Nino events on a threatened lemur. Biological

Conservation 141:287–297.

Dunham, A. E., E. M. Erhart, and P. C. Wright. 2011. Global climate cycles and

cyclones: consequences for rainfall patterns and lemur reproduction in

southeastern Madagascar. Global Change Biology 17:219–227.

Eisenberg, J. F., and R. W. Thorington Jr. 1973. A preliminary analysis of a Neotropical

mammal fauna. Biotropica 5:150–161.

Elton, C. 1946. Competition and the structure of ecological communities. Journal of

Animal Ecology 15:54–68.

Erhart, E. M., and D. J. Overdorff. 1999. Female coordination of group travel in wild

Propithecus and Eulemur. International Journal of Primatology 20:927–940.

Erhart, E. M., and D. J. Overdorff. 2008a. Spatial Memory during Foraging in Prosimian

Primates: Propithecus edwardsi and Eulemur fulvus rufus. Folia Primatologica

79:185–196. 94

Erhart, E. M., and D. J. Overdorff. 2008b. Population demography and social structure

changes in Eulemur fulvus rufus from 1988 to 2003. American Journal of Physical

Anthropology 136:183–193.

Estrada, A., and R. Coates-Estrada. 1984. Fruit eating and seed dispersal by howling

monkeys (Alouatta palliata) in the tropical rain forest of Los Tuxtlas, Mexico.

American Journal of Primatology 6:77–91.

Farwig, N., and D. G. Berens. 2012. Imagine a world without seed dispersers: A review

of threats, consequences and future directions. Basic and Applied Ecology

13:109–115.

Di Fiore, A., and S. A. Suarez. 2007. Route-based travel and shared routes in sympatric

spider and woolly monkeys: cognitive and evolutionary implications. Animal

Cognition 10:317–329.

Fleming, T. H. 1979. Do tropical frugivores compete for food? Integrative and

Comparative Biology 19:1157–1172.

Forget, P., and L. Cuijpers. 2008. Survival and scatterhoarding of frugivores-dispersed

seeds as a function of forest disturbance. Biotropica 40:380–385.

Fort, K. P., and J. H. Richards. 1998. Does seed dispersal limit initiation of primary

succession in desert playas? American Journal of Botany 85:1722–1731.

Freckleton, R. P., and O. T. Lewis. 2006. Pathogens, density dependence and the

coexistence of tropical trees. Proceedings of the Royal Society of London B:

Biological Sciences 273:2909–2916. 95

Fritz, S. A., and A. Purvis. 2010. Selectivity in mammalian extinction risk and threat

types: a new measure of phylogenetic signal strength in binary traits.

Conservation Biology 24:1042–1051.

Ganzhorn, J. U., J. Fietz, E. Rakotovao, D. Schwab, and D. Zinner. 1999. Lemurs and the

regeneration of dry deciduous forest in Madagascar. Conservation Biology

13:794–804.

Garber, P. A. 1986. The ecology of seed dispersal in two species of callitrichid primates

(Saguinus mystax and Saguinus fuscicollis). American Journal of Primatology

10:155–170.

Garber, P. A. 1989. Role of spatial memory in primate foraging patterns: Saguinus

mystax and Saguinus fuscicollis. American Journal of Primatology 19:203–216.

Garber, P. A. 2000. Evidence for the use of spatial, temporal, and social information by

some primate foragers. Pages 261–298 in S. Boinski and P. A. Garber, editors. On

the move: how and why animals travel in groups. University of Chicago Press,

Chicago.

Garber, P. A., and P. E. Jelinek. 2006. Travel Patterns and Spatial Mapping in

Nicaraguan Mantled Howler Monkeys (Alouatta palliata). Pages 287–309 in A.

Estrada, P. A. Garber, M. S. M. Pavelka, and L. Luecke, editors. New

Perspectives in the Study of Mesoamerican Primates. Springer US.

Garcia, D., and D. Martinez. 2012. Species richness matters for the quality of ecosystem

services: a test using seed dispersal by frugivorous birds. Proceedings of the

Royal Society B: Biological Sciences 279:3106–3113. 96

Gilbert, G. S., and C. O. Webb. 2007. Phylogenetic signal in plant pathogen-host range.

Proceedings of the National Academy of Sciences 104:4979–4983.

Glander, K. E., P. C. Wright, P. S. Daniels, and A. M. Merenlender. 1992.

Morphometrics and testicle size of rain forest lemur species from southeastern

Madagascar. Journal of Human Evolution 22:1–17.

Godfrey, L. R., J. M. Winchester, S. J. King, D. M. Boyer, and J. Jernvall. 2012. Dental

topography indicates ecological contraction of lemur communities. American

Journal of Physical Anthropology 148:215–227.

Green, A. K., D. Ward, and M. E. Griffiths. 2009. Directed dispersal of mistletoe

(Plicosepalus acaciae) by Yellow-vented Bulbuls (Pycnonotus xanthopygos).

Journal of Ornithology 150:167–173.

Harvey, P. H., and T. H. Clutton-Brock. 1981. Primate home-range size and metabolic

needs. Behavioral Ecology and Sociobiology 8:151–155.

Hawkins, A. F. A., and S. M. Goodman. 2003. Introduction to the birds. Pages 1019–

1044 in S. M. Goodman and J. P. Benstead, editors. The natural history of

Madagascar. The University of Chicago Press, Chicago.

Herrera, C. M. 1982. Seasonal variation in the quality of fruits and diffuse coevolution

between plants and avian dispersers. Ecology 63:773–785.

Herrera, C. M. 2002. Seed dispersal by vertebrates. Pages 185–208 in C. M. Herrera and

O. Pellmyr, editors. Plant-animal interactions: an evolutionary approach.

Blackwell Science Ltd, USA. 97

Herrera, J. P., P. C. Wright, E. Lauterbur, L. Ratovonjanahary, and L. L. Taylor. 2011.

The effects of habitat disturbance on lemurs at Ranomafana National Park,

Madagascar. International Journal of Primatology 32:1091–1108.

Higgins, S., R. Nathan, and M. Cain. 2003. Are long-distance dispersal events in plants

usually caused by nonstandard means of dispersal? Ecology 84:1945–1956.

Hirsch, B. T., R. Kays, V. E. Pereira, and P. A. Jansen. 2012. Directed seed dispersal

towards areas with low conspecific tree density by a scatter‐hoarding rodent.

Ecology Letters 15:1423–1429.

Howe, H. F., and J. Smallwood. 1982. Ecology of seed dispersal. Annual Review of

Ecology and Systematics 13:201–228.

Hubbell, S. P. 1979. Tree dispersion, abundance, and diversity in a tropical dry forest.

Science 203:1299–1309.

Hubbell, S. P. 2001. The unified neutral theory of biodiversity and biogeography.

Princeton University Press.

Hubbell, S. P., J. A. Ahumada, R. Condit, and R. B. Foster. 2001. Local neighborhood

effects on long-term survival of individual trees in a Neotropical forest.

Ecological Research 16:859–875.

Hubbell, S. P., and R. B. Foster. 1986. Biology, chance, and history and the structure of

tropical rain forest tree communities. Pages 314–329 in M. L. Cody and J.

Diamond, editors. Community ecology. Harper and Row, New York.

IUCN. 2012. IUCN Red List of threatened species. IUCN, the World Conservation

Union. 98

Janmaat, K. R. L., R. W. Byrne, and K. Zuberbühler. 2006. Evidence for a spatial

memory of fruiting states of rainforest trees in wild mangabeys. Animal

Behaviour 72:797–807.

Janzen, D. H. 1970. Herbivores and the number of tree species in tropical forests.

American Naturalist 104:501–528.

Johnson, D. J., W. T. Beaulieu, J. D. Bever, and K. Clay. 2012. Conspecific negative

density dependence and forest diversity. Science 336:904–907.

Julliot, C. 1996. Seed dispersal by red howling monkeys (Alouatta seniculus) in the

tropical rain forest of French Guiana. International Journal of Primatology

17:239–258.

Karubian, J., and R. Duraes. 2009. Effects of seed disperser social behavior on patterns of

seed movement and deposition. Oecologica Autralis 13:45–57.

King, S. J., S. J. Arrigo-Nelson, S. T. Pochron, G. M. Semprebon, L. R. Godfrey, P. C.

Wright, and J. Jernvall. 2005. Dental senescence in a long-lived primate links

infant survival to rainfall. Proceedings of the National Academy of Sciences of

the United States of America 102:16579–16583.

Kwit, C., D. J. Levey, and C. H. Greenberg. 2004. Contagious seed dispersal beneath

heterospecific fruiting trees and its consequences. Oikos 107:303–308.

Kwit, C., Levey, D.J., Turner, S.A., Clark, C.J., and Poulsen, J.R. 2007. Out of one

shadow into another: causes and consequences of spatially contagious seed

dispersal by frugivores. Pages 427–444 in Dennis, A.J., Green, R.J., Schupp,

E.W., and Westcott, D., editors. Seed dispersal: theory and its application in a

changing world. CABI, Orxford, U.K. 99

Lahann, P. 2007. Feeding ecology and seed dispersal of sympatric cheirogaleid lemurs

(Microcebus murinus, Cheirogaleus medius, Cheirogaleus major) in the littoral

rainforest of south-east Madagascar. Journal of Zoology 271:88–98.

Langrand, O. 1990. Guide to the Birds of Madagascar. Yale Univ Press.

Lapenta, M. J., and P. Procópio-de-Oliveira. 2008. Some aspects of seed dispersal

effectiveness of golden lion tamarins (Leontopithecus rosalia) in a Brazilian

Atlantic forest. Tropical Conservation Science 1:122–139.

Lebrija-Trejos, E., S. J. Wright, A. Hernández, and P. B. Reich. 2014. Does relatedness

matter? Phylogenetic density-dependent survival of seedlings in a tropical forest.

Ecology 95:940–951.

Levine, J. M., and D. J. Murrell. 2003. The community-level consequences of seed

dispersal patterns. Annual Review of Ecology, Evolution, and Systematics

34:549–574.

Levi, T., and C. A. Peres. 2013. Dispersal vacuum in the seedling recruitment of a

primate-dispersed Amazonian tree. Biological Conservation 163:99–106.

Link, A., and A. Di Fiore. 2006. Seed dispersal by spider monkeys and its importance in

the maintenance of neotropical rain-forest diversity. Journal of Tropical Ecology

22:235–246.

Liu, X., M. Liang, R. S. Etienne, Y. Wang, C. Staehelin, and S. Yu. 2012. Experimental

evidence for a phylogenetic Janzen-Connell effect in a subtropical forest. Ecology

Letters 15:111–118. 100

Loayza, A. P., and T. Knight. 2010. Seed dispersal by pulp consumers, not “legitimate”

seed dispersers, increases Guettarda viburnoides population growth. Ecology

91:2684–2695.

McArdle, B. H., and M. J. Anderson. 2001. Fitting multivariate models to community

data: a comment on distance-based redundancy analysis. Ecology 82:290–297.

McConkey, K. R. 2000. Primary seed shadow generated by gibbons in the rain forests of

Barito Ulu, Central Borneo. American Journal of Primatology 52:13–29.

McConkey, K. R., and W. Y. Brockelman. 2011. Nonredundancy in the dispersal

network of a generalist tropical forest tree. Ecology 92:1492–1502.

McConkey, K. R., and D. J. Chivers. 2007. Influence of gibbon ranging patterns on seed

dispersal distance and deposition site in a Bornean forest. Journal of Tropical

Ecology 23:269–275.

Morellato, L. P. C., D. C. Talora, A. Takahasi, C. C. Bencke, E. C. Romera, and V. B.

Zipparro. 2000. Phenology of Atlantic rainforest trees: a comparative study.

Biotropica 32:811–823.

Moses, K. L., and S. Semple. 2011. Primary seed dispersal by the black-and-white ruffed

lemur (Varecia variegata) in the Manombo forest, south-east Madagascar. Journal

of Tropical Ecology 27:529–538.

Muller-Landau, H. C., and B. D. Hardesty. 2005. Seed dispersal of woody plants in

tropical forests: concepts, examples and future directions. Pages 267–309 in D.

Burslem, M. Pinard, and S. Hartley, editors. Biotic interactions in the tropics:

Their role in the maintenance of species diversity. Cambridge University Press.

Nathan, R. 2006. Long-distance dispersal of plants. Science 313:786–788. 101

Nathan, R., and H. C. Muller-Landau. 2000. Spatial patterns of seed dispersal, their

determinants and consequences for recruitment. Trends in Ecology & Evolution

15:278–285.

Norconk, M. A., S. Boinski, and P. M. Forget. 2011. Primates in 21st century

ecosystems: does primate conservation promote ecosystem conservation?

American Journal of Primatology 73:3–8.

Overdorff, D. 1988. Preliminary report on the activity cycle and diet of the red‐bellied

Lemur (Lemur rubriventer) in Madagascar. American Journal of Primatology

16:143–153.

Overdorff, D. J. 1993. Similarities, differences, and seasonal patterns in the diets of

Eulemur rubriventer and Eulemur fulvus rufus in the Ranomafana National Park,

Madagascar. International Journal of Primatology 14:721–753.

Overdorff, D. J. 1996. Ecological correlates to activity and habitat use of two prosimian

primates: Eulemur rubriventer and Eulemur fulvus rufus in Madagascar.

American Journal of Primatology 40:327–342.

Overdorff, D. J., A. M. Merenlender, P. Talata, A. Telo, and Z. A. Forward. 1999. Life

history of Eulemur fulvus rufus from 1988–1998 in southeastern Madagascar.

American Journal of Physical Anthropology 108:295–310.

Overdorff, D. J., and S. G. Strait. 1998. Seed handling by three prosimian primates in

southeastern Madagascar: implications for seed dispersal. American Journal of

Primatology 45:69–82. 102

Overdorff, D. J., and S. R. Tecot. 2006. Social pair-bonding and resource defense in wild

red-bellied lemurs (Eulemur rubriventer). Pages 235–254 in L. Gould and M. L.

Sauther, editors. Lemurs: Ecology and Adaptation. Springer US.

Overdorff, D., and M. Rasmussen. 1995. Determinants of nighttime activity in “diurnal”

lemurid primates. Pages 61–74 in L. Alterman, editor. Creatures of the dark: the

nocturnal prosimians. Plenum, New York.

Parga, J. A., M. L. Sauther, F. P. Cuozzo, I. A. Y. Jacky, and R. R. Lawler. 2012.

Evaluating ring-tailed lemurs (Lemur catta) from southwestern Madagascar for a

genetic population bottleneck. American Journal of Physical Anthropology

147:21–29.

Peres-Neto, P. R., D. A. Jackson, and K. M. Somers. 2005. How many principal

components? Stopping rules for determining the number of non-trivial axes

revisited. Computational Statistics & Data Analysis 49:974–997.

Pratt, T. K., and E. W. Stiles. 1983. How long fruit-eating birds stay in the plants where

they feed: implications for seed dispersal. American Naturalist 122:797–805.

Quinn, G. P., and M. J. Keough. 2002. Experimental design and data analysis for

biologists. Cambridge University Press.

Rakotomanana, H., T. Hino, M. Kanzaki, and H. Morioka. 2003. The role of the Velvet

Asity Philepitta castanea in regeneration of understory shrubs in Madagascan

rainforest. Ornithological Science 2:49–58.

Ratsimbazafy, J. H., S. J. Arrigo-Nelson, L. Dollar, C. M. Holmes, M. T. Irwin, S. E.

Johnson, N. J. Stevens, and P. C. Wright. 2012. Conservation of Malagasy 103

prosimians: a view from the Great Red Island. Pages 387–396 in J. Masters, M.

Gamba, and F. Génin, editors. Leaping Ahead. Springer New York.

Razafindratsima, O. H., and A. E. Dunham. 2015. Assessing the impacts of nonrandom

seed dispersal by multiple frugivore partners on plant recruitment. Ecology

96:24–30.

Razafindratsima, O. H., T. A. Jones, and A. E. Dunham. 2014. Patterns of movement and

seed dispersal by three lemur species. American Journal of Primatology 76:84–96.

Razafindratsima, O. H., and B. T. Martinez. 2012. Seed dispersal by red-ruffed lemurs:

seed size, viability, and beneficial effect on seedling growth. Ecotropica 18:15–

26.

Razafindratsima, O. H., S. Mehtani, and A. E. Dunham. 2013. Extinctions, traits and

phylogenetic community structure: insights from primate assemblages in

Madagascar. Ecography 36:47–56.

R Core Team. 2012. R: A language and environment for statistical computing. R

Foundation for Statistical Computing, Vienna, Austria, 2012. ISBN 3-900051-07-

Rodríguez-Pérez, J., T. Wiegand, and L. Santamaria. 2012. Frugivore behaviour

determines plant distribution: a spatially-explicit analysis of a plant-disperser

interaction. Ecography 35:113–123.

Rosindell, J., S. P. Hubbell, and R. S. Etienne. 2011. The unified neutral theory of

biodiversity and biogeography at age ten. Trends in Ecology & Evolution 26:340–

348. 104

Ross, C. 1992. Basal metabolic rate, body weight and diet in primates: an evaluation of

the evidence. Folia Primatologica 58:7–23.

Rowell, T. E., and B. J. Mitchell. 1991. Comparison of seed dispersal by guenons in

Kenya and capuchins in Panama. Journal of Tropical Ecology 7:269–274.

Russo, S. E., and C. K. Augspurger. 2004. Aggregated seed dispersal by spider monkeys

limits recruitment to clumped patterns in Virola calophylla. Ecology Letters

7:1058–1067.

Russo, S. E., S. Portnoy, and C. K. Augspurger. 2006. Incorporating animal behavior into

seed dispersal models: implications for seed shadows. Ecology 87:3160–3174.

Scharfe, F., and W. Schlund. 1996. Seed removal by lemurs in a dry deciduous forest of

western Madagascar. Primate report 46:295–304.

Schupp, E. W., T. Milleron, and S. E. Russo. 2002. Dissemination limitation and the

origin and maintenance of species-rich tropical forests. Pages 19–33 in D. J.

Levey, W. R. Silva, and M. Galetti, editors. Seed dispersal and frugivory:

ecology, evolution, and conservation. CABI, New York.

Schwitzer, C., R. A. Mittermeier, S. E. Johnson, G. Donati, M. Irwin, H. Peacock, J.

Ratsimbazafy, J. Razafindramanana, E. E. Louis, L. Chikhi, I. C. Colquhoun, J.

Tinsman, R. Dolch, M. LaFleur, S. Nash, E. Patel, B. Randrianambinina, T.

Rasolofoharivelo, and P. C. Wright. 2014. Averting lemur extinctions amid

Madagascar’s political crisis. Science 343:842–843.

Snyder, R. E., and P. Chesson. 2003. Local dispersal can facilitate coexistence in the

presence of permanent spatial heterogeneity. Ecology Letters 6:301–309. 105

Sparrow, W. A., and K. M. Newell. 1998. Metabolic energy expenditure and the

regulation of movement economy. Psychonomic Bulletin & Review 5:173–196.

Spehn, S., and J. U. Ganzhorn. 2000. Influence of seed dispersal by brown lemurs on

removal rates of three Grewia species (Tiliaceae) in the dry deciduous forest of

Madagascar. Ecotropica 6:13–21.

Spiegel, O., and R. Nathan. 2007. Incorporating dispersal distance into the disperser

effectiveness framework: frugivorous birds provide complementary dispersal to

plants in a patchy environment. Ecology Letters 10:718–728.

Stevenson, P. R. 2000. Seed dispersal by woolly monkeys (Lagothrix lagothricha) at

Tinigua National Park, Colombia: dispersal distance, germination rates, and

dispersal quantity. American Journal of Primatology 50:275–289.

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 39:385–392.

Terborgh, J., N. Pitman, M. Silman, H. Schichter, and P. Nunez. 2002. Maintenance of

tree diversity in tropical forests. Pages 1–17 in A. J. Dennis, R. J. Green, E. W.

Schupp, and D. A. Westcott, editors. Seed dispersal: theory and its application in

a changing world. CABI, Wallingford, UK.

Tilman, D. 1994. Competition and biodiversity in spatially structured habitats. Ecology

75:2–16.

Valenta, K., and L. M. Fedigan. 2010. Spatial patterns of seed dispersal by white‐faced

capuchins in Costa Rica: evaluating distant‐dependent seed mortality. Biotropica

42:223–228. 106

Vamosi, S., S. Heard, J. Vamosi, and C. Webb. 2009. Emerging patterns in the

comparative analysis of phylogenetic community structure. Molecular Ecology

18:572–592.

Vidal, M. M., M. M. Pires, and P. R. Guimarães. 2013. Large vertebrates as the missing

components of seed-dispersal networks. Biological Conservation 163:42–48.

Wang, B. ., and T. B. Smith. 2002. Closing the seed dispersal loop. Trends in Ecology

and Evolution 17:379–385.

Webb, C. O., D. D. Ackerly, and S. W. Kembel. 2008. Phylocom: software for the

analysis of phylogenetic community structure and trait evolution. Bioinformatics

24:2098–2100.

Webb, C. O., D. D. Ackerly, M. A. McPeek, and M. J. Donoghue. 2002. Phylogenies and

community ecology. Annual Review of Ecology and Systematics 33:475–505.

Webb, C. O., and M. J. Donoghue. 2005. Phylomatic: tree assembly for applied

phylogenetics. Molecular Ecology Notes 5:181–183.

Webb, C. O., G. S. Gilbert, and M. J. Donoghue. 2006. Phylodiversity-dependent

seedling mortality, size structure, and disease in a Bornean rain forest. Ecology

87:123–131.

Wehncke, E. V., and C. A. Dominguez. 2007. Seed dispersal ecology of non-restricted

frugivores, capuchin monkeys in three neotropical forests. Journal of Tropical

Ecology 23:519–528.

Wehncke, E. V., S. P. Hubbell, R. B. Foster, and J. W. Dalling. 2003. Seed dispersal

patterns produced by white-faced monkeys: implications for the dispersal

limitation of neotropical tree species. Journal of Ecology 91:677–685. 107

Wenny, D. G. 2001. Advantages of seed dispersal: a re-evaluation of directed dispersal.

Evolutionary Ecology Research 3:51–74.

Wenny, D. G., and D. J. Levey. 1998. Directed seed dispersal by bellbirds in a tropical

cloud forest. Proceedings of the National Academy of Sciences 95:6204–6207.

Westcott, D. A., J. Bentrupperbäumer, M. G. Bradford, and A. McKeown. 2005.

Incorporating patterns of disperser behaviour into models of seed dispersal and its

effects on estimated dispersal curves. Oecologia 146:57–67.

Westcott, D. A., and D. L. Graham. 2000. Patterns of movement and seed dispersal of a

tropical frugivore. Oecologia 122:249–257.

White, F. J., D. J. Overdorff, E. A. Balko, and P. C. Wright. 1995. Distribution of ruffed

lemurs (Varecia variegata) in Ranomafana National Park, Madagascar. Folia

Primatologica 64:124–131.

Wikström, N., V. Savolainen, and M. W. Chase. 2001. Evolution of the angiosperms:

calibrating the family tree. Proceedings of the Royal Society of London. Series B:

Biological Sciences 268:2211–2220.

Will, H., and O. Tackenberg. 2008. A mechanistic simulation model of seed dispersal by

animals. Journal of Applied Ecology 96:1011–1022.

Wright, J. S. 2002. Plant diversity in tropical forests: a review of mechanisms of species

coexistence. Oecologia 130:1–14.

Wright, P. C. 1999. Lemur traits and Madagascar ecology: coping with an island

environment. Yearbook of Physical Anthropology 42:31–72.

Wright, P. C., and B. Andriamihaja. 2002. Making a rain forest national park work in

Madagascar: Ranomafana National Park and its long-term research commitment. 108

Pages 112–136 in J. Terborgh, C. P. van Schaik, L. Davenport, and R. Madhu,

editors. Making parks work: strategies for preserving tropical nature. Island Press,

Washington, DC.

Wright, P. C., E. M. Erhart, S. Tecot, A. L. Baden, S. J. Arrigo-Nelson, J. Herrera, T. L.

Morelli, M. B. Blanco, A. Deppe, and S. Atsalis. 2012. Long-Term Lemur

Research at Centre Valbio, Ranomafana National Park, Madagascar. Pages 67–

100 in P. Kappeler and D. P. Watts, editors. Long-term field studies of primates.

Springer-Verlag, Berlin Heidelberg.

Wright, P. C., S. R. Tecot, E. M. Erhart, A. L. Baden, S. J. King, and C. Grassi. 2011.

Frugivory in four sympatric lemurs: implications for the future of Madagascar’s

forests. American Journal of Primatology 73:585–602.

Zhang, S.-Y., and L.-X. Wang. 1995. Fruit Consumption and seed dispersal of Ziziphus

cinnamomum (Rhamnaceae) by two sympatric primates (Cebus apella and Ateles

paniscus) in French Guiana. Biotropica 27:397–401.

Zimmerman, M. 1979. Optimal foraging: a case for random movement. Oecologia

43:261–267. 109

Appendices

Appendix A – Fruit species consumed and dispersed by E. rufifrons, E. rubriventer and V. v. editorum in Ranomafana National Park from June 2010 to June 2011: presence and relative proportion of intact seed species in defecations relative to total defecated seeds during study period. n: dispersed-seeds of < 1 mm in length (count not recorded). *: fruit species found in defecation but not observed being consumed. x: consumed plant species but not dispersed

Eulemur Eulemur Varecia variegata Family Species names Local names rubriventer rufiffrons editorum Anacardiaceae Abrahamia thouvenotii Sandramy fotsy x

Abrahamia turkii Sandramy 1.018 0.106 11.360

Anisophylleaceae Anisophyllea fallax Hazoharaka x

Annonaceae Ambavia capuronii Ramiavontoloho x

Xylopia buxifolia Ramiavona x

Aphloiaceae Aphloia theaformis Fandramanana x x

Apocynaceae Carissa edulis Fantsy x x

Craspidospermum verticillatum Vandrika 0.157 x

Plectaneia sp Vahikondro x x

Arecaceae Dypsis decipiens Sihara 7.410

Dypsis nodifera Sirahazo x

Ravenea robustior Lafa 1.188

Asteraceae Mikania sp Vahivahia x x x Burseraceae Canarium madagascariensis Ramy 1.328

Burseraceae Canarium madagascariensis Ramy boribory x 0.140

Canellaceae Cinnamosma madagascariensis Fanalamangidy x

Celastraceae Brexiella sp Voamasoandro 0.053*

Clusiaceae Garcinia goudotiana Kimbaletaka x x

Garcinia sp Voamalambotaho x

Mammea bongo Natojabo x

Mammea vatoensis Natovoraka x 1.783

Psorospermum androsaemifolium Fanerandahy x

Connaraceae Agelaea pentagyna Vahiherotra x x

Cucurbitaceae Raphidocystis sp Vahimbarongy x

Cunoniaceae Weinmannia bojeriana Maka 2.741 1.856

Weinmannia rutenbergii Lalona x

Dichapetalaceae Dichapetalum chlorinum Vahindavenona x

Dichapetalum sp Vahimavo 1.175 x

Ebenaceae Diospyros gracilipes var. lecomtei Hazomainty 2.174

Fabaceae Abrus precatorius Vahimboamena x

110

Eulemur Eulemur Varecia variegata Family Species names Local names rubriventer rufiffrons editorum Gentianaceae Anthocleista amplexicaulis Dendemy x x

Lamiaceae Clerodendrum petunioides Voalatakakohoala x

Premna corymbosa Odimamo x

Lauraceae Cryptocarya crassifolia Tavolomalady 2.741 3.924 1.538 Cryptocarya crassifolia Tavolomanitra 17.933 8.643 3.530 Cryptocarya parareolata Tavolomaintso 0.140 Tavolomalady Lauraceae Cryptocarya sp fotsy 0.078* 0.530* 0.629 Cryptocarya thouvenotii Tavolopina 10.180 13.786 2.377 Ocotea nervosa Varongy 0.078 x x Ocotea racemosa Varongy fotsy x

Potameia rubra Sary 0.035

Loranthaceae Bakerella sp Tongoalahy x

Malvaceae Grewia sp Hafipotsy 4.507

Melastomataceae Medinilla sp Kalamasimbarika x

Monimiaceae Ephippiandra madagascariensis Tambonetra x

Tambourissa perrieri Ambora lahy 1.488* x

Moraceae Ficus botryoides Voararano x x x Ficus lutea Amontana x x x Ficus politoria Famakilela x x

Ficus reflexa Nonoka n n n Ficus rubra Nonoka vaventy x

Ficus tiliifolia Voara n n

Streblus dimepate Mahanoro 7.204 2.863

Myrtaceae Eugenia louvelii Voabe 3.759 x x Psidium cattleianum Goavy tsinahy n n

Psidium guajava Goavy vaventy n n

Syzygium emirnense Robary x x

Syzygium emirnense Rotrafotsy 0.235 x 3.251 Myrtaceae Syzygium parkeri Rotramena 2.036* 0.106 4.124 Oleaceae Noronhia incurvifolius Tsilaitra fotsy x

Noronhia introversa Tsilaitra x x

Pandanaceae Pandanus sp Tsirika 3.915 x

Passifloraceae Deidamia sp Kilelakomby 1.273

Primulaceae Oncostemum botryoides Kalafana x

Oncostemum botryoides Kalafana lg x x 0.070 Oncostemum leprosum Kalafana sm x

Oncostemum nervosum Kalafambakaka 0.157 0.053 x Rubiaceae Antirhea borbonica Fatsikahitra x x x Rubiaceae Danais sp Tamborimantsina x x

Gaertnera brevipedicellata Ranjopody x

111

Eulemur Eulemur Varecia variegata Family Species names Local names rubriventer rufiffrons editorum Gaertnera phyllostachya Bararata x

Gyrostipula foveolata valotra x

Gyrostipula foveolata Valotra tenany x

Mussaenda arcuata Anambahy x x

Mussaenda erectiloba Fatora 7.048 3.287 x Psychotria mandrarensis Fanorafa x x

Psychotria reducta Fohaninasity 18.246 11.612

Saldinia sp Tongely x x

Rutaceae Toddalia asiatica Anakatsimba x x

Vepris sp Apody lg x

Rutaceae Zanthoxylum sp Voangy x

Zanthoxylum tsihanimposa Tsitongamposa x x

Salicaceae Scolopia madagascariensis Faritraty 11.511 32.185

Sapindaceae Allophylus cobbe Dikana 3.994 2.015 x Deinbollia neglecta Lanary madinika x 0.424 x Zanha sp Zahana 0.318 0.489

Sapotaceae Chrysophyllum boivinianum Rahiaka 4.229 10.127 59.560 Sideroxylon betsimisarakum Nato 0.105

Smilacaceae Smilax anceps Rohindambo x 0.159

Torricelliaceae Melanophylla crenata Vavaporetaka x

Unidentified Lonjo x

Unidentified Vahimberana x

Unidentified Vahitamboro x

Unidentified Voatakaboka 0.078 x 0.944

112

Appendix B – Published accounts of mean observed seed dispersal distances of several lemur and non-lemur primates. Data on body mass were recorded from “All The World’s Primates” database ()

female body mean dispersal reference for dispersal data Primate species mass (g) distance (m) Non-lemur species Alouatta palliata 4855 - (Estrada and Coates-Estrada 1984) Alouatta seniculus 5310 260 (Julliot 1996) Ateles belzebuth 7996 443 (Link and Di Fiore 2006) Ateles paniscus 8440 202.5 (Zhang and Wang 1995) Cebus apella 2300 390 (Zhang and Wang 1995) (Rowell and Mitchell 1991, Wehncke et al. Cebus capucinus 2666 225.8 2003, Valenta and Fedigan 2010) Hylobates muelleri 5250 385 (McConkey and Chivers 2007) Hylobates muelleri x agilis 5375 220 (McConkey 2000) Lagothrix lagotricha 5740 300 (Stevenson 2000) Leontopithecus rosalia 595 105 (Lapenta and Procópio-de-Oliveira 2008) Macaca leonina 6480 - (Albert et al. 2013) Saguinus fuscicollis 377 - (Garber 1986) Saguinus mystax 586 - (Garber 1986)

Lemur species Eulemur fulvus rufus 1775 128 (Spehn and Ganzhorn 2000) Eulemur rubriventer 1960 119.81 this study Eulemur rufifrons 1775 95.95 this study Varecia variegata 3350 180 (Moses and Semple 2011) Varecia variegata editorum 3350 116.86 this study

113

Appendix C – Distribution of the dispersal events sampled across species and sites

Average daily Frugivore species Group ID Site # of dispersal events group size E. rubriventer 1 3.00 Talatakely 29 E. rubriventer 2 3.40 Talatakely 48 E. rubriventer 3 3.92 Talatakely 51 E. rubriventer 4 3.00 Talatakely 19 E. rubriventer 5 3.00 Talatakely 58 E. rubriventer 6 3.00 Valohoaka 45 E. rubriventer 7 2.00 Valohoaka 16 E. rubriventer 8 3.00 Valohoaka 10 E. rufifrons 1 8.42 Talatakely 91 E. rufifrons 2 3.30 Talatakely 129 E. rufifrons 3 9.84 Talatakely 51 E. rufifrons 4 10.00 Talatakely 27 E. rufifrons 5 3.56 Talatakely 27 E. rufifrons 6 1.89 Talatakely 30 E. rufifrons 7 12.00 Valohoaka 20 E. rufifrons 8 6.05 Mangevo 38 V. v. editorum 1 3.98 Vatoharanana 152 V. v. editorum 2 3.79 Mangevo 94 V. v. editorum 3 4.00 Mangevo 92 V. v. editorum 4 2.25 Mangevo 41 V. v. editorum 5 2.31 Valohoaka 26 V. v. editorum 6 4.15 Mangevo 63 V. v. editorum 7 1.62 Mangevo 55 V. v. editorum 8 2.00 Mangevo 128 Total defecation collected 1340

114

Appendix D – Description of the four study sites where data on dispersal events were collected

Proportion of each canopy # tree Cryptocarya mean ± sd Site #stems/ha cover category species stems/ha canopy cover < 55% 55 – 75% > 75% Mangevo 172 2582 7.78 ± 12.50 73.52 ± 15.01 % 0.11 0.39 0.50

Talatakely 119 3288 411.67 ± 207.82 74.90 ± 11.16 % 0.05 0.41 0.54

Valohoaka 120 3522 30.00 ± 11.06 66.37 ± 7.20 % 0.04 0.86 0.10

Vatoharanana 102 3292 181.67 ± 48.82 70.97 ± 9.84 % 0.06 0.57 0.37

115

Appendix E – Estimation of dispersal parameters and demographic rates, and creation of probability recruitment models

Quantity of dispersed seeds

We estimated the quantity of Cryptocarya seeds dispersed per tree for each individual lemur group during the tree species’ fruiting season (March-October). These estimates were based on the average observed hourly rates of defecation of Cryptocarya seeds per lemur relative to the density of adult Cryptocarya trees in their respective territory (from transect data). We extrapolated from hourly dispersal rates, the per capita number of seeds per tree dispersed per fruiting season (240 days), assuming a 12 h day for each frugivore species. It is important to note that while this allowed us to standardize our comparison between the dispersers, all three lemur species have been observed to exhibit some level of cathemerality in Ranomafana (Eulemur spp.: Overdorff 1988, 1996,

Overdorff and Rasmussen 1995; V. v. editorum: Johnson, personal communication, and

Balko, personal communication in Wright 1999), introducing uncertainty to our estimates since potential nighttime feeding rates are unknown. We then multiplied our per capita

(frugivore) estimates of dispersal rate by the disperser’s average density in the park (E. rubriventer: 5.46 individuals / km2 ; E. rufifrons: 6.70 individuals / km2 ; V. v. editorum :

2.23 individuals / km2 , Wright et al. 2012) and divided by the estimated density of adult

Cryptocarya trees in each species’ territory. Seed dispersal rates per frugivore species used in our probability recruitment models were calculated using averages across groups.

Dispersal bias into different microhabitats

For each frugivore species, we determined the pattern of seed deposition into the different microhabitat categories (gap: <55%; medium-shaded: 55-75%; shaded: >75%; 116

and under a conspecific adult) with data from direct observations of dispersal events. We calculated the relative availability of each microhabitat category at each study area as the proportion of occurrence from the transect data. Canopy cover categories were calculated separately from the availability of microhabitats under conspecific adults. The

“preference” of each frugivore species for depositing seeds into each microhabitat category was estimated as the proportion of dispersal events in a given microhabitat category (i.e. under each of the different canopy covers described above) divided by the relative availability of that category in the study site where the lemur groups were observed. Results were then subtracted from one such that positive numbers represented

“preference” and negative numbers represented “avoidance”.

Demographic rates

Relative germination rates were calculated as the proportion of seeds that germinated in the exclosure plots during the 3-month initial experiment relative to the number of seeds planted. We estimated removal/predation rates as the proportion of seeds that were being preyed upon or removed in the open subplots during the 3-month experiment. We estimated survival from the initial set-up to 3 months as the number of intact seeds and seedlings at 3 months in open subplots relative to the number of seeds planted. The one year seedling transition probabilities were estimated as the mean proportion of surviving seedlings from the paired subplots after a year of experiment initiation relative to the total number of intact seeds and seedlings at 3 months. And finally, transition probabilities from one year seedling to three year old sapling were estimated from the number of saplings at year 3 relative to seedlings at year 1. 117

Three-year recruitment model

We assessed the relative contribution of each disperser to the 3-year recruitment of their host plant with a simulated recruitment model performed in MATLAB R2011a

(The MathWorks Inc., Natick, MA) parameterized by our observational and experimental data (Figure E1). To do this, we used the following model:

Sapling recruitment = Σ (Qi x Si x TP1i x TP2i)

For each microhabitat “i” (i = 1 … n), Qi represents the average number of

Cryptocarya seeds dispersed per adult tree into microhabitat i during the species’ fruiting season. Si represents the 3 month survival of seeds and seedlings in each microhabitat “i”, and TP1i and TP2i respectively, represent the 3 month - 1 year and 1 year - 3 year transition probabilities of seedlings in microhabitat “i”. The simulation modelled the quantity of seeds dispersed per tree per fruiting season by each disperser with a Poisson distribution, which is appropriate for counting the number of occurrences of a random event in a given time (Quinn and Keough 2002). We used a multinomial distribution to model the probabilities of seed deposition into the different microhabitats and a binomial function to model the vital rates (germination and transition rates). A multinomial distribution is often used to model the probability of success from multiple mutually exclusive outcomes (Quinn and Keough 2002). Simulations were run 1000 times.

We used our model to estimate the contribution of each disperser to per capita seed recruitment by dividing the number of recruited saplings with the number of seeds dispersed as estimated in the model. “No dispersal” assumes that all seeds fall under the 118

parent tree. “Random dispersal” assumes that seeds fall into different microhabitat categories with probabilities based on availability in the habitat. We calculated “random dispersal” separately for each species such that estimates of availability of Cryptocarya trees and each microhabitat category were based on the same distribution used in the individual species models.

Figure E1: Schematic representation of Cryptocarya’s recruitment dynamics in each microhabitat category in this study

seedlings and dispersed seedlings saplings seeds seeds (Q) 1 year 3 year survival (S) 3 months transition transition probabilities probabilities (TP1) (TP2)

119

Appendix F – Dispersal events by the three frugivore species under heterospecific trees with different microhabitat categories (gap: <55% cover, medium-shaded: 55-75%, shaded: > 75%) relative to availability in their habitats. : proportional dispersal; > 1: more seeds deposited in the microhabitat category than expected based on availability; < 1: less seeds dispersed in the microhabitat than expected. Error bars represent standard errors at lemur group levels. All dispersal events are considered including those under conspecific trees.

3.00 □ E. rubriventer

E. rufifrons 2.50 V. v. editorum

2.00 Dispersal Dispersal bias 1.50

1.00

0.50

0.00 gap (< 55 %) medium-shaded (55 - shade (> 75 %) 75 %) Canopy cover categories

120

Appendix G – List of tree species that have crowns overhead of the seed traps, and their dispersal syndromes. Data on dispersal syndromes were based on frugivore diet, and fruit/seed traits

Dispersal Family Species name Local name Data source syndrome Anacardiaceae Abrahamia turkii Sandramy lemurs Overdorff 1993, Razafindratsima et al. 2014 Anacardiaceae Micronychia macrophylla Sehana lemurs Overdorff 1993, White et al. 1995 Annonaceae Xylopia buxifolia Ramiavona lemurs Razafindratsima et al. 2014 Aphloiaceae Aphloia theaformis Fandramanana lemurs Overdorff 1993, Razafindratsima et al. 2014 Apocynaceae Cabucala cryptophlebia Kabokala lemurs Overdorff 1993 Aquifoliaceae Ilex mitis Hazondrano lemurs and birds Razafindratsima unpbl. data Araliaceae Polyscias ariadnes Maniny birds Razafindratsima unpbl. data Araliaceae Polyscias tripinnata Vatsilana birds Razafindratsima unpbl. data Arecaceae Dypsis nodifera Sirahazo lemurs Overdorff 1993, Razafindratsima et al. 2014 Asparagaceae Dracaena sp Hasindavaravina lemurs and birds Razafindratsima unpbl. data Asparagaceae Dracaena sp Taviavola lemurs and birds Razafindratsima unpbl. data Asteraceae Vernonia exserta Tavilona abiotic Razafindratsima unpbl. data Bignoniaceae Colea aff lantziana bail Malemy Fo unknown Clusiaceae Calophyllum paniculatum Vitanina lemurs Razafindratsima unpbl. data Clusiaceae Garcinia goudotiana Kimbaletaka lemurs Overdorff 1993, Razafindratsima et al. 2014 Clusiaceae Harungana madagascariensis Harongana lemurs and birds Overdorff 1993, Razafindratsima unpbl. data Clusiaceae Mammea bongo Natojabo lemurs Razafindratsima et al. 2014 Clusiaceae Mammea vatoensis Natovoraka lemurs Dew and Wright 1998, Razafindratsima et al. 2014 Clusiaceae Symphonia gymnoclada Kimba lemurs Overdorff 1993, Razafindratsima unpbl. data Cunoniaceae Weinmannia humblotii Sisitra abiotic Razafindratsima unpbl. data Cyatheaceae Cyathea sp Fahofaho abiotic Razafindratsima unpbl. data Diospyros gracilipes var. Ebenaceae lecomtei Hazomainty lemurs Razafindratsima et al. 2014 Elaeocarpaceae Sloanea rhodantha Vanana abiotic Overdorff 1993 Erythroxylaceae Erythroxylum sp Malambovony lemurs and birds Razafindratsima unpbl. data Euphorbiaceae Bridelia tulasneana Harina lemurs Razafindratsima unpbl. data 121

Dispersal Family Species name Local name Data source syndrome Croton antanosiensis var. Euphorbiaceae fianarantsoae Sily Fotsy unknown Euphorbiaceae Drypetes madagascariensis Tsivalandrano lemurs Razafindratsima unpbl. data Euphorbiaceae Macaranga boutonioides Mokaranana birds Razafindratsima unpbl. data Fabaceae Albizia gummifera Volomborona abiotic Razafindratsima unpbl. data Fabaceae Dalbergia baronii Voamboana abiotic Razafindratsima unpbl. data Gentianaceae Anthocleista amplexicaulis Dendemy lemurs Razafindratsima et al. 2014 Lauraceae Cryptocarya crassifolia 2 Tavolomalady lemurs Razafindratsima et al. 2014 Lauraceae Cryptocarya crassifolia 3 Tavolomanitra lemurs Razafindratsima et al. 2014 Lauraceae Cryptocarya parareolata Tavolomaintso lemurs Dew and Wright 1998, Razafindratsima et al. 2014 Lauraceae Cryptocarya thouvenotii Tavolopina lemurs Razafindratsima et al. 2014 Overdorff 1993; White et al. 1995; Dew & Wright Lauraceae Ocotea nervosa Varongy lemurs 1998; Razafindratsima et al. 2014 Lauraceae Potameia rubra Sary lemurs Razafindratsima et al. 2014 Malvaceae Dombeya dichotomopsis Merika abiotic Razafindratsima unpbl. data Malvaceae Dombeya erythroclada Hafitra Diavorona abiotic Razafindratsima unpbl. data Malvaceae Dombeya lucida Hafidahy abiotic Razafindratsima unpbl. data Malvaceae Grewia brideliifolia Hafitra Taikalalao lemurs and birds Razafindratsima unpbl. data Malvaceae Grewia sprima Hafipotsy lemurs Overdorff 1993, Razafindratsima et al. 2014 Melastomataceae Dichaetanthera oblongifolia Tsingotroka abiotic Razafindratsima unpbl. data Meliaceae Malleastrum sp Tongobivy lemurs and birds Razafindratsima unpbl. data Monimiaceae Ephippiandra madagascariensis Tambonetra lemurs Razafindratsima et al. 2014 Monimiaceae Tambourissa perrieri Ambora Lahy lemurs Overdorff 1993, Razafindratsima et al. 2014 Monimiaceae Tambourissa thouvenotii Ambora Vavy lemurs Overdorff 1993 Moraceae Ficus pachyclada Apana lemurs and birds Razafindratsima unpbl. data Moraceae Ficus politoria Famakilela lemurs and birds Dew and Wright 1998, Razafindratsima et al. 2014 Moraceae Streblus dimepate Mahanoro lemurs Razafindratsima et al. 2014 Overdorff 1993, White et al. 1995, Razafindratsima et Myrtaceae Syzygium emirnense Rotrafotsy lemurs al. 2014 122

Dispersal Family Species name Local name Data source syndrome Myrtaceae Syzygium parkeri Rotra lemurs Dew and Wright 1998, Razafindratsima et al. 2014 Myrtaceae Syzygium parkeri Rotramena lemurs Dew and Wright 1998, Razafindratsima et al. 2014 Myrtaceae Syzygium parkeri 1 Rotramena Lg lemurs Razafindratsima unpbl. data Overdorff 1993, Dew and Wright 1998, Razafindratsima et Oleaceae Noronhia incurvifolius Solaitra Fotsy lemurs al. 2014 Oleaceae Noronhia introversa Solaitra lemurs Razafindratsima et al. 2014 Pandanaceae Pandanus sp 1 Tsirika lemurs Dew and Wright 1998, Razafindratsima et al. 2014 Overdorff 1993, White et al. 1995, Dew and Wright Primulaceae Oncostemum botryoides Kalafana Lg lemurs 1998, Razafindratsima et al. 2014 Primulaceae Oncostemum nervosum Kalafambakaka lemurs and birds Overdorff 1993, Razafindratsima et al. 2014 Primulaceae Oncostemum palmiforme Hazotoho lemurs and birds Razafindratsima, unpbl. data White et al. 1995, Dew and Wright 1998, Rubiaceae Antirhea borbonica Fatsikahitra lemurs Razafindratsima et al. 2014 Rubiaceae Gaertnera brevipedicellata Ranjopody lemurs Razafindratsima et al. 2014 Gaertnera obovata var. Rubiaceae sphaerocarpa Bararata Lahy birds Razafindratsima unpbl. data White et al. 1995, Dew and Wright 1998, Rubiaceae Mussaenda erectiloba Fatora lemurs Razafindratsima et al. 2014 Fohaninasity Rubiaceae Psychotria parkeri Mena lemurs and birds Razafindratsima unpbl. data Rubiaceae Psychotria reducta Fohaninasity lemurs and birds Overdorff 1993, Razafindratsima et al. 2014 Rubiaceae Saldinia sp1 Tongely lemurs Razafindratsima et al. 2014 Rutaceae Zanthoxylum madagascariensis Fahavalonkazo lemurs and birds Razafindratsima unpbl. data Sapindaceae Allophylus cobbe Dikana lemurs and birds Overdorff 1993, Razafindratsima et al. 2014 Sapindaceae Plagioscyphus louvelii Lanary Mainty lemurs Razafindratsima unpbl. data White et al. 1995, Dew and Wright 1998, Sapotaceae Chrysophyllum boivinianum Rahiaka lemurs Razafindratsima et al. 2014 Scrophulariaceae Nuxia coricea Lambinanala lemurs Razafindratsima unpbl. data Torricelliaceae Melanophylla crenata Vavaporetaka lemurs Razafindratsima et al. 2014

123

Appendix H – Seed dispersal by each lemur species near conspecific and heterospecific fruiting trees upon which they fed

25.00 conspecific

heterospecific

20.00

15.00

feeding tree feeding 10.00

5.00 proportion of dispersal events under fruiting under events dispersal of proportion

0.00 E. rubriventer E. rufifrons V. v. editorum

editorum ooe 1 1 1 1 1 1 1 1 1 1 1 1 — 1 — I 1 §ÿ 1 CO l-j 1— CD CD 1 — HÿH 124 I 'SL H-Cj-QO '

Appendix I – When dispersal under conspecifics were included in the analysis of the 1

pattern of relatedness (NTI) between seeds dispersed by lemurs and the adult neighbor 1 1 trees they fell under, the phylogenetic relationships of dispersed seeds with nearest adult trees showed a pattern of phylogenetic clustering for those dispersed by E. rufifrons ho]

(mean = 0.49, t = 3.25, P = 0.01) and V. v. editorum1 (mean = 0.31, t = 3.36, P = 0.01), 1 1 1 1 1 7 7 1 l—

and a pattern no different than random expectation for those dispersed by E rubriventer00 — CO I— (mean = 0.28, t7 = 1.35, P = 0.22). [ CO

Taking a temporally explicit approach, we found that this pattern varied over time,; I— * ~z. idiosyncratically for each species, ranging from patternsK3H of overdispersed (negative -tk

values) in some months to clustered (positive values) patterns in others. Values near zero ! >— indicate a random relatedness. 5 o- —

* bCH CO h+H adult-tree ISO CO — — v. i ► 1 i ►H 1 N> + adult-tree dispersed 1 1 1 ► l between dispersed ► between □ ► neighbor neighbor ij ► fO ■ □ □ ► (c) □ H ■n relatedness CO relatedness ► □ heterospecific and heterospecific I pi phylognetic pi and _ seed 0|H 2 seed — DH phylognetic — N> *ÿ V. 1 § } > b b § D-O- N) — 3 3 > > adult-tree dispersed I o o b • c_ o HMMIII > O o between i neighbor o c_ c_ O I • o o o o o b I • o I o b I relatedness CO • o l heterospecific o CO b h«H b and seed phylognetic ♦ - ♦ l»l i i

| > > o i 3 c_ c_ c_ - - N) i i o o O b o i - - b ------

------] II II 125

Appendix J – List of plant species for which we monitored fruiting phenology in Ranomafana National Park, Madagascar

Family Plant spp Anacardiaceae Abrahamia ditimena 1 Abrahamia oblongifolia Abrahamia sp Abrahamia thouvenotii Abrahamia tsiramiramy Abrahamia turkii Micronychia macrophylla Micronychia tsiramiramy var. minutiflora Anisophylleaceae Anisophyllea fallax Annonaceae Ambavia capuronii Uvaria acuminata Xylopia buxifolia Apocynaceae cabucala cryptophlebia Carissa edulis Craspidospermum verticillatum Tabernaemontana retusa Aquifoliaceae Ilex mitis Araliaceae Polyscias ariadnes Polyscias sp polyscias tripinnata Schefflera longipedicellata Arecaceae Dypsis linearis Dypsis nodifera ravenea robustior Asparagaceae Dracaena reflexa Dracaena sp1 Dracaena sp2 Asteraceae Brachylaena ramiflora Vernonia exserta Bignoniaceae Colea aff lantziana bail Burseraceae Canarium madagascariensis Canellaceae Cinnamosma madagascariensis Cinnamosma sp Clusiaceae Calophyllum paniculatum Garcinia aphanophlebia Garcinia goudotiana Garcinia megistophylla Garcinia sp 126

Family Plant spp Harungana madagascariensis Clusiaceae Mammea angustifolia Mammea bongo Mammea vatoensis Symphonia gymnoclada Combretaceae Terminalia tetrandra Cunoniaceae Weinmannia bojeriana Weinmannia humblotii Weinmannia rutenbergii Ebenaceae Diospyros gracilipes var. lecomtei Elaeocarpaceae Sloanea rhodantha Ericaceae Agauria polyphylla Erythroxylaceae Erythroxylum nitidulum Erythroxylum sp Erythroxylum sphaeranthum Euphorbiaceae Antidesma petiolare Bridelia tulasneana Croton sp Drypetes madagascariensis Macaranga boutonioides Macaranga myriolepida Fabaceae Albizia gummifera Cynometra lyalii Dalbergia baronii Entada louvelii Viguieranthus subauriculatus Gentianaceae Anthocleista amplexicaulis Anthocleista madagascariensis Lamiaceae Clerodendrum petunioides Lauraceae Beilschmiedia pedicellata Cryptocarya crassifolia 2 Cryptocarya crassifolia 3 Cryptocarya crassifolia 5 Cryptocarya parareolata Cryptocarya sp 3 Cryptocarya thouvenotii Ocotea auriculiformis Ocotea grayi Ocotea nervosa Ocotea racemosa Ocotea sp 127

Family Plant spp Lauraceae Potameia rubra Malvaceae Dombeya acerifolia Dombeya antsihanakensis Dombeya dichotomopsis Dombeya erythroclada Dombeya hilsembergii Dombeya lucida Dombeya seyrigii Dombeya sp Grewia brideliifolia Grewia sp Grewia sprima Melastomataceae Dichaetanthera oblongifolia Meliaceae Malleastrum sp Menispermaceae Burasaia madagascariensis Monimiaceae Ephippiandra madagascariensis Tambourissa perrieri Tambourissa thouvenotti Moraceae Ficus botryoides Ficus pachyclada Ficus politoria Streblus dimepate Streblus mauritianus Treculia africana madagascarica Trilepisium madagascariensis Myrtaceae Eugenia bernieri Eugenia sp Syzygium emirnense Syzygium parkeri 1 Syzygium parkeri 2 Oleaceae Noronhia incurvifolius Noronhia introversa Pandanaceae Pandanus sp 1 Pandanus sp 2 Primulaceae Oncostemum botryoides Oncostemum leprosum Oncostemum nervosum Oncostemum palmiforme Proteaceae Dilobeia thouarsii Rubiaceae Antirhea borbonica Chassalia sp

128

Family Plant spp Rubiaceae Chassalia ternifolia Gaertnera brevipedicellata Gaertnera obovata var. sphaerocarpa 1 Gaertnera obovata var. sphaerocarpa 2 Lemyrea krugii Mussaenda erectiloba Psychotria mandrarensis Psychotria parkeri Psychotria reducta Pyrostria sp Saldinia sp1 Rutaceae Vepris ampody Vepris sp 1 Zanthoxylum madagascariensis Salicaceae Aphloia theaformis Homalium parkeri Scolopia madagascariensis Scolopia sp Sapindaceae Allophylus cobbe Beguea apetala Deinbollia neglecta Macphersonia gracilis Plagioscyphus louvelii Plagioscyphus sp Tina striata Sapotaceae Chrysophyllum boivinianum Sideroxylon betsimisarakum Scrophulariaceae Nuxia capitata Nuxia coricea Nuxia pachyphylla Solanaceae Solanum mauritianum Torricelliaceae Melanophylla crenata Unknown Unknown 1 (Fandramanandahy) Unknown 2 (Fanerambavy)