Open Collections

UBC Theses and Dissertations

UBC Theses Logo

UBC Theses and Dissertations

Ecological effects of disrupting plant-animal interactions Granados, Alys 2017

Your browser doesn't seem to have a PDF viewer, please download the PDF to view this item.

Item Metadata

Download

Media
24-ubc_2017_november_granados_alys.pdf [ 2.67MB ]
Metadata
JSON: 24-1.0355399.json
JSON-LD: 24-1.0355399-ld.json
RDF/XML (Pretty): 24-1.0355399-rdf.xml
RDF/JSON: 24-1.0355399-rdf.json
Turtle: 24-1.0355399-turtle.txt
N-Triples: 24-1.0355399-rdf-ntriples.txt
Original Record: 24-1.0355399-source.json
Full Text
24-1.0355399-fulltext.txt
Citation
24-1.0355399.ris

Full Text

 ECOLOGICAL EFFECTS OF DISRUPTING PLANT-ANIMAL INTERACTIONS by  Alys Granados  B.Sc. (Hons), University of Guelph, 2007 M.Sc., Concordia University, 2011  A THESIS SUBMITTED IN PARTIAL FULFILLMENT OF THE REQUIREMENTS FOR THE DEGREE OF  DOCTOR OF PHILOSOPHY in THE FACULTY OF GRADUATE AND POSTDOCTORAL STUDIES (Zoology)  THE UNIVERSITY OF BRITISH COLUMBIA (Vancouver)  August 2017  © Alys Granados, 2017 ii  Abstract Human disturbances threaten tropical forests, potentially disrupting plant-animal interactions. Though many studies have assessed the effects of habitat disturbance on either plants or animals, we have a limited understanding of how disrupted interactions will affect both groups of interacting species and whether impacts vary with spatial scale. This hinders our ability to predict how logging, a widespread threat to tropical forests, affects plant and animal communities, particularly in combination with overhunting, a common co-occurring threat. I excluded small to large-bodied mammals from research plots in logged and unlogged forests in Borneo to quantify the effects of logging and hunting on seed predation and establishment in five dominant tree species (Chapter 2). Granivore-induced seed mortality was higher in logged forest. Defaunation did not affect seed mortality. Seedling establishment was highest when small to large-bodied mammals were excluded in logged forest, suggesting that the effects of logging and hunting interact to alter seedling recruitment. I used the same animal exclusion plots to assess impacts on diversity and the distribution of morphological traits in tagged seedling communities over four years (Chapter 3). Excluding small to large vertebrates did not affect overall seedling diversity but animals disproportionately killed seedlings from large-fruited genera in logged forest, reducing community fruit size. Animals altered plant traits, though impacts may be underestimated by focusing only on changes to taxonomic diversity. Selective logging can also intensify the patchiness of food availability across the landscape. Spatiotemporal variability in food abundance is especially pronounced during mast fruiting events, yet the consequences for animal habitat use at different spatial scales in faunally intact systems is unclear. I used camera traps to estimate the site use of several vertebrate species across two dipterocarp mast years and a non-mast year at large and small spatial scales (Chapter 4). Site use was positively associated iii  with masting for several taxa, though strong response were mostly limited to intact forest. Even key consumers (bearded pigs) showed reduced responses in the logged forest at the small scale. Overall, my work demonstrates that anthropogenic disturbances disrupt plant-animal interactions by altering plant recruitment and limiting vertebrate responses to resource availability.   iv  Lay Summary Logging and hunting are major threats to tropical biodiversity, potentially disrupting plant-animal interactions, though the consequences for plant and animal communities are not well understood. I excluded mammals from research plots in Malaysian Borneo to quantify the effects of logging and hunting on plants. Seeding establishment was higher when large mammals were absent in logged forest, while small to large mammals disproportionately killed seedlings from species that would produce large fruits as trees. Next, I used camera traps to estimate how extreme increases in fruit abundance affect animal habitat use and whether logging affects animals’ ability to track food. In unlogged forest, several species were positively associated with fruit availability at camera stations, whereas habitat use in logged forest was mostly unaffected by spatio-temporal variation in fruit abundance. Logging therefore disrupts plant-animal interactions by altering seedling establishment and by interfering with animal responses to changes in food abundance.  v  Preface Chapter 2 is co-authored with Jedediah Brodie, Henry Bernard, and Michael J. O’Brien. H. Bernard provided guidance for field work. M.J. O’Brien, J. Brodie and A. Granados developed the experimental design for this chapter. Data were collected by A. Granados, M.J. O’Brien and field assistants (Musa Markus and Egli Lee). Statistical analyses were performed by A. Granados and M.J O’Brien. The manuscript was written by A. Granados with assistance from M.J. O’Brien and J. Brodie and has been accepted for publication in Ecological Applications (EAP17-0201).   Chapters 3 and 4 are co-authored with J. Brodie and H. Bernard. Experimental design for both chapters were developed by A. Granados and J. Brodie. Data were collected by A. Granados and several field assistants: Anton Sorokin, M. Markus, E. Lee, Tamar Ferrel, Megan Young, Remmy Murus, Philip Ulok, Shari Mang, Sabina Havalic, Sean Espinola, Théa Rachinski, and Mona Schmidtt. H. Bernard provided guidance with field work. A. Granados performed all statistical analyses with help from J. Brodie. The manuscripts were written by A. Granados with guidance from J. Brodie.  vi  Table of Contents  Abstract .......................................................................................................................................... ii	Lay Summary ............................................................................................................................... iv	Preface .............................................................................................................................................v	Table of Contents ......................................................................................................................... vi	List of Tables ............................................................................................................................... vii	List of Figures ............................................................................................................................. viii	Acknowledgements ...................................................................................................................... ix	Chapter 1: Introduction ................................................................................................................1	Chapter 2: Defaunation and habitat disturbance interact synergistically to alter seedling recruitment .....................................................................................................................................8	Chapter 3: Experimental defaunation alters seedling functional traits but not diversity ....29	Chapter 4: The importance of mast-fruiting for vertebrates in a faunally intact ecosystem ......................................................................................................................................42	Chapter 5: Conclusion .................................................................................................................71	Bibliography .................................................................................................................................76	Appendix A - Supplementary material for Chapter 3 ............................................................... 93	Appendix B - Supplementary material for Chapter 4 ............................................................. 103	 vii  List of Tables Table 2-1 Model parameters and AIC weights for seed fate…………………………… 22   Table 4-1 Model coefficient estimates for fruit biomass……………………………….. 57   Table 4-2 Model selection results for large-scale animal site use……….…………….. 58   Table 4-3 Model coefficient results for small-scale animal site use……………….….. 60   Table A-1 Mammal herbivore species in study sites …………………………………... 93   Table A-2 Plant families in logged and unlogged forest exclosure plots ……………… 94   Table B-1 Plant families in fruit collections …………………………………………… 103   Table B-2 Model selection results for small-scale animal site use …………………….. 106  viii  List of Figures Figure 2-1 Habitat use of mammal granivores ………………………………………… 19   Figure 2-2 Seed mortality and germination in logged and unlogged forest exclosures………………………………………………………………………………..  20   Figure 2-3 Seedling establishment in logged and unlogged forest exclosure plots…….. 21   Figure 2-4 Percent of seedlings killed, germinated, established in logged and unlogged forest exclosures………………………………………………………………………...  27   Figure 3-1 Rarefied seedling genus richness in logged and unlogged forest exclosures……………………………………………………………………………......  37   Figure 3-2 Seedling community fruit length in logged and unlogged forest exclosures………………………………………………………………………………..  38   Figure 4-1 Map of study sites ………………………………………………………….. 52   Figure 4-2 Mean fruit biomass at camera stations ……………………………………... 62   Figure 4-3 Mean animal detections in non-mast year and mast years …………………. 63   Figure 4-4 Bearded pig site use as a function of ripe dipterocarp fruit biomass……….. 64   Figure 4-5 Bearded piglet site use as a function of dipterocarp fruit biomass……........ 65   Figure A-1 Seedling mortality in logged and unlogged forest exclosure plots………… 101   Figure A-2 Seedling growth in logged and unlogged forest exclosure plots…………… 102   Figure B-1 Plant families in fruit biomass collections ………………………………… 111   Figure B-2 Yellow muntjac site use relative to dipterocarp fruit biomass…..……........ 112   Figure B-3 Malayan porcupine site use relative to ripe dipterocarp fruit biomass……... 113   Figure B-4 Argus pheasant site use relative to dipterocarp fruit biomass…..………….. 114   Figure B-5 Fireback pheasant site use relative to total fruit biomass…………………... 115   ix  Acknowledgements First and foremost, I am extremely grateful and appreciative for the guidance and support from my supervisor Jedediah Brodie over the last five years. Thanks to Jill Jankowski for stepping in as my co-supervisor. I am thankful for Jill and the rest of my committee members, past and present for their encouragement and constructive comments: Chris Harley, Roy Turkington, Jennifer Williams, and Greg Crustinger. In Malaysia, I am grateful for the support from Henry Bernard and Glen Reynolds. I am especially thankful to Michael J. O’Brien for being a great collaborator and mentor. I was lucky to have an amazing crew of field volunteers. Thanks for your hard work and enthusiasm: Anton Sorokin, Shari Mang, Théa Rachinski, Sabina Havalic, Tamar Ferrel, Sean Espinola, Megan Young, Lauren Nerfa, Fanny Tricone, Ryan Chiong, Mari Kondoh, and Mona Schmidt.  Thanks to past and present lab mates: Adrienne Contasti, Matt Strimas-Mackey, Sam Yue, Cheng Chen, Mairin Deith, and Patrick Burke. Muchas gracias a mis padres por siempre animando me. Akhirnya, saya sangat bersyukur ke semua orang dari South East Asian Rainforest Research Program (SEARRP). Saya sangat bersyukur persahabatan anda punya. Terima kasih untuk berkongsi pengetahuan sekitar hutan. Terima kasih ke semua kawan dalam Sabah: Philip Ulok, Mike Berandus, Musa Markus, Egli Lee, Udin Jaga, Mohd Fauzi Osman, Remmy Murus, dan Adrian Karolus. Terima kasih kepada hutan. 1  Chapter 1: Introduction  Despite extensive research on the ecology and evolution of species interactions (e.g. Hart 1992, Tylianakis et al. 2008), we still lack a general understanding of how interacting organisms may be affected when species interactions are disrupted by disturbances that alter interaction outcomes (Wootton 1993). Animals interact with plants as consumers of seeds, foliage, and fruit, with direct consequences for the individual fitness of both plants and animals (Loiselle and Blake 1991, Strauss and Agrawal 1999). The net effects of these interactions are often difficult to predict because outcomes can be context-dependent, varying with the local environment as well as the interacting species involved (Chamberlain et al. 2014). The same animal can have both positive (e.g. via seed dispersal) and negative (e.g. via herbivory, granivory) impacts on plants and the net outcomes of interactions can vary with species traits or across life stages. Increasingly, ecosystems are affected by human activity, resulting in habitat degradation or loss and declines in animal abundance (Guariguata et al. 2002, Wright 2003, Peres and Palacios 2007, Harrison et al. 2013, Rosin and Poulsen 2016). The direct effects of these anthropogenic disturbances could indirectly affect interaction outcomes between groups of interacting species (Wootton 1993, Klanderud 2005). Habitat disturbances can modify the outcome of species interactions but the consequences are not well understood. Altered plant-animal interactions due to human activity have received increasing interest from ecologists (e.g. Galetti and Dirzo 2013, Harrison et al. 2013, Kurten 2013), particularly with respect to interactions involving large-bodied vertebrates. Globally, large-bodied vertebrates are disproportionality at a higher risk of extinction than smaller species (Cardillo et al. 2005), yet they play important roles as seed dispersers, predators, 2  and herbivores (Curran and Leighton 2000, Wright et al. 2007, Goheen et al. 2010). The loss of large seed dispersers can lead to reduced recruitment of large-seeded species (Wright et al. 2000, Peres and Palacios 2007, Galetti and Dirzo 2013), while declines in vertebrate seed predators may be associated with increased seed survival (Hautier et al. 2010) and germination rates (Bricker et al. 2010, Beckman et al. 2011, Harrison et al. 2013). Mammal herbivores consume foliage and trample young plants but disturbances that alter plant-herbivores interactions can lead to increased (Paine and Beck 2007, Theimer et al. 2011) or unchanged seedling survival (Roldán and Simonetti 2001, Brocardo et al. 2013). Animals may also alter plant community composition and ecosystem function if they disproportionately kill plant taxa associated with certain plant traits (Camargo-Sanabria et al. 2015). Most studies examining the effects of altered plant-frugivore, -granivore, or -herbivore interactions take place in areas previously determined to have reduced abundances of interacting vertebrates with a focus on the consequences for plant recruitment. Fewer studies attempt to assess how the impacts of animals on early plant life stages may translate to consequences for plant communities. The effects on animals may also be studied separately from the effects on plants or may be entirely ignored. We currently have a limited understanding of how human-related disturbances that disrupt plant-animal interactions affect both groups of interacting species in the same system and the consequences for ecosystem processes are also poorly understood. Further, interactions may be scale-dependent, yet the effects of disrupting plant-animal interactions at multiple spatial scales are unclear. In spite of a large body of existing research on how anthropogenic disturbances affect plant-animal interactions (Harrison 2011, Harrison et al. 2013), the conditions under which interaction outcomes are disrupted at different spatial scales are not well known (but see Garcia et al. 2011). Food resources (e.g. fruits, seeds, vegetation) are patchily distributed across 3  landscapes and the level of patchiness varies with spatial scale. At small scales, food patches (i.e. plant communities) are nested within patches at larger scales (i.e. forest patches, landscapes) (Senft et al. 1987, Wiens 1989, Kotliar and Wiens 1990). Outcomes of plant-animal interactions could therefore hinge upon on the ability of animals to reach areas of high food productivity, requiring them to spatially and temporally track asynchronous fruiting patterns across the forest at multiple spatial scales. Changes to habitat structure due to human activity could further modify the spatial distribution of food resources. This could affect how animals respond to food availability, subsequently affecting interacting species (Curran and Webb 2000, Lehouck et al. 2009).  Tropical forests are highly biodiverse systems with complex webs of interactions and the effects of altered species interactions could impact multiple trophic levels (Morris 2010). The loss of primary forest due to timber extraction is a major threat to tropical biodiversity. It is estimated that 20 - 50% of tropical forests were subject to logging between 2000 to 2005 (Asner et al. 2009). The rate of tropical forest loss is highest in Southeast Asia and selective logging is rampant in this region (Achard et al. 2002, Sodhi et al. 2004). Most forests in this region are classified as production forest and are open to logging (Johns 1997, Edwards et al. 2010). Selective logging targets specific trees based on their taxonomy or size but incidental damage to surrounding vegetation can be substantial, creating large areas of degraded forest (Johns 1988, Corlett and Primack 2011). Besides altering forest structure, logging facilitates access into the forest for hunters via the creation of roads (Wilkie et al. 2000). Vertebrates, especially large-bodied taxa, often show reduced abundances as a result of hunting in disturbed areas (Laurance and Laurance 1996, Poulsen and Clark 2011); this might have cascading impacts on interacting species (Wang et al. 2007). 4  Study region Southeast Asia is an ideal location to study the consequences of anthropogenic disturbance on species interactions. Forests in this region are incredibly biodiverse (Sodhi et al. 2010) with correspondingly complex webs of species interactions. However, this region has the fastest rate of deforestation and forest degradation in the tropics (Achard et al. 2002, Mayaux et al. 2005).  My study took place in eastern Sabah, Malaysian Borneo from 2013 to 2016. Data were collected in unlogged forest (Danum Valley Conservation Area (DVCA, N5.10189°/ E117.688°) as well as in a selectively logged forest (Sabah Biodiversity Experiment, SBE and Malua Forest reserve, MFR, N5.16727°/E117.564°). This region is characterized by lowland forest dominated by trees from the family Dipterocarpaceae. Dipterocarp trees participate in regional synchronized mast fruiting events every two to ten years (Appanah 1985, Whitmore 1998), with little to no seed production between mast events (Cannon et al. 2007). These masting periods tend to coincide with El Niño-Southern Oscillation events (Curran et al. 1999). Selective logging operations in the region target dipterocarp trees. The logged forest study site was logged for dipterocarps >60 cm diameter at breast height (dbh) in the late 1980s with the use of tractors and high lead cables (Berry et al. 2008, Tuck et al. 2016). The SBE itself is found within that area that was previously logged. From 1999 to 2007, parts of the Malua Forest Reserve were re-logged, although the SBE itself was not (Tuck et al. 2016). In this second logging cycle, a reduced impact logging regime was adopted and trees of 40 cm dbh or greater were targeted (Anon. 2008, Ang et al. 2016).     5  Thesis objectives  The objective of this dissertation is to assess the impacts of habitat disturbance on plant and animal communities to better understand the conditions under which species interactions are altered by habitat disturbance. Specifically, I aim to answer the following questions: 1. How do animals affect the survival of early plant life stages in dominant canopy tree species? 2. Do animals alter the distribution of morphological traits in tree seedling communities? Are altered traits accompanied by changes to tree seedling diversity? 3. How do terrestrial vertebrate assemblages respond to resource pulse events of varying intensity at broad and small spatial scales?   In Chapter 2, I assess the joint effects of habitat disturbance and defaunation on seed mortality, germination, and seedling establishment in five dominant tree species in disturbed and undisturbed tropical rainforest. Seed predation by vertebrates can lead to seed mortality, but whether animal-induced seed mortality alters plant recruitment can vary with habitat context, seed traits, and among granivore species. An incomplete understanding of seed predation makes it difficult to predict how widespread extirpations of vertebrate granivores in tropical forests might affect trees, especially in the face of habitat disturbance. I used wire cages to exclude large (elephants), medium (sambar deer, bearded pigs, muntjac deer), and small (porcupines, chevrotains) ground-dwelling mammalian granivores in logged and unlogged forests in 2014.  6  In Chapter 3, I examine the combined impacts of habitat disturbance and defaunation on tree seedling communities over four years (2013 to 2016) with the use of experimental exclusion plots. Though logging and overhunting often occur together, most studies document their effects on plant recruitment separately or additively. Compared to their isolated impacts, less is known about how logging and defauntion might interact to influence seedling communities. Medium- to large-bodied vertebrates may alter seedling diversity in some cases (Camargo-Sanabria et al. 2015) but not others (Roldán and Simonetti 2001, Brocardo et al. 2013), and could potentially alter the distribution of morphological plant traits (Kurten et al. 2015). I examine how the loss of vertebrate herbivores in disturbed and undisturbed forest affects both the taxonomic diversity and the representation of key morphological traits among tree seedlings known to be important to ecosystem function (fruit size, wood density, and specific leaf area).   In Chapter 4, I quantify vertebrate responses to resource pulses in disturbed and undisturbed forest at large and small spatial scales. Resource pulses, such as mast fruiting, can have dramatic consequences for animal populations and communities (Ostfeld et al. 1996, Ostfeld and Keesing 2000) but most studies of masting take place where key consumers of mast have already been greatly reduced or extirpated (Ellsworth and McComb 2003, Kelly et al. 2008). Our knowledge of whether animals in systems with intact vertebrate communities show strong responses to masting is limited, as is our understanding of how responses may vary with spatial scale, if at all. I used passive infrared cameras to estimate large-scale site use of terrestrial vertebrates in two mast years (2014, 2015) and in a non-mast year (2013) between habitats varying in masting intensity (logged and unlogged forest). I also quantify animal responses to masting at the small 7  scale, by assessing whether animals’ use of camera trap stations in mast years varied with spatio-temporal patterns in fruit availability.  8  Chapter 2: Defaunation and habitat disturbance interact synergistically to alter seedling recruitment  Defaunation, the loss or functional loss of large animals from natural communities, is receiving increasing interest from ecologists because it is globally widespread and can potentially disrupt important species interactions such as seed predation (Harrison et al. 2013, Dirzo et al. 2014). But the conditions under which disrupted plant–animal interactions alter the abundance, distribution, or persistence of plant species are not well understood (McConkey et al. 2012, Brodie et al. 2014). The impacts of animal declines on plant populations may be wide-ranging in highly biodiverse systems with complex webs of interactions, such as tropical forests (Canale et al. 2012, Kurten 2013, Morris 2010, Terborgh 2013, Wright 2003), implying that dramatic changes in abundance in certain species could indirectly affect numerous other taxa (Wright 2003). Our lack of knowledge on the extent to which animals influence plant recruitment and persistence via interactions like seed predation makes it difficult to predict the ecological consequences of defaunation in these systems.  Animal mutualists and antagonists can affect seed survival (Ostfeld et al. 1997, Fragoso and Huffman 2000) and germination (Asquith et al. 1997, Nunez-Iturri et al. 2008, Sherry 2008, Sethi and Howe 2009, Harrison et al. 2013). However, reduced germination does not necessarily lead to reduced seedling establishment or plant recruitment (Paine and Beck 2007) and therefore, might not influence trees at the population level (Brodie et al. 2009, Harrison et al. 2013). The plant vital rate altered by granivores is seed germination, and this demographic transition often has little impact on population growth in long-lived plants (Pfister 1998, Howe and Miriti 2004). 9  In some cases, reduced granivory due to the loss or exclusion of vertebrates can increase germination rates and seedling emergence (Guariguata and Pinard 1998, Bricker et al. 2010, Beckman et al. 2011, Harrison et al. 2013). In other cases, the loss of vertebrates decreases seedling and sapling diversity (Harrison et al. 2013, Camargo-Sanabria et al. 2014) or leads to compensatory seedling consumption by other plant consumers such as invertebrates (Itoh et al. 1995). Impacts of defaunation may also be obscured if small-bodied granivores increase in abundance following the loss of larger species, resulting in unchanged or even increased levels of seed predation (Wright and Duber 2001, Mendoza and Dirzo 2007, Maclean et al. 2011, Galetti et al. 2015).   In addition to defaunation, tropical forests worldwide are often simultaneously threatened by human-induced habitat disturbances (Wright 2005, Dirzo et al. 2014). However, the impacts on plants are usually assessed in isolation or additively and the combined effects of habitat disturbance and defaunation on plant recruitment are not well understood. Selective logging, a common and widespread activity in tropical forests, alters forest structure and may reduce seed production, seedling establishment (Hautier et al. 2010, Bagchi et al. 2011), and animal abundance (Laurance and Laurance 1996). The decline or absence of large-bodied animals, in particular, can indirectly reduce the recruitment of certain trees (Terborgh et al. 2008, Melo et al. 2010), with variable consequences for tree population dynamics and forest regeneration (Forget and Jansen 2007, Paine and Beck 2007, Harrison et al. 2013). Although logging and defaunation often occur together, we have little understanding of their interacting effects on plant recruitment in tropical forests (Peres 2000, Guariguata et al. 2002, Poulsen and Clark 2011, Brodie et al. 2014). 	10   Our knowledge of the conditions under which seed dynamics can influence long-lived tree abundance is limited because most studies focus on one or a few plant species (often short-lived taxa with similar ecological traits) or on a single demographic transition (usually seed germination) (Comita et al. 2014, Visser et al. 2016). Here, I sought to assess the joint effects of defaunation and habitat disturbance on seed mortality, germination, and seedling establishment of five dominant tree species in disturbed and undisturbed tropical rainforest. I excluded large, medium, and small terrestrial mammals from three experimental exclosure blocks in logged and unlogged forest in Malaysian Borneo in 2014. Each block consisted of three exclusion treatments and one control. I also used camera traps to quantify the presence of large and small-bodied vertebrates (i.e., those highly vulnerable to anthropogenic defaunation and those not, respectively) in logged and unlogged forest study sites. I expected a higher occurrence of vertebrate taxa in unlogged forest (Barlow et al. 2007, Poulsen and Clark 2011) and higher levels of seed mortality to occur in areas where more vertebrates were detected (Curran and Webb 2000, Beckman and Muller-Landau 2007). The predation of large-seeded species can also be lower where large mammals are absent versus where they are present (Mendoza and Dirzo 2007) so I expected reduced seed mortality among large seeded species within exclusion plots. Finally, I predicted that seed mortality for all tree species would be higher in control plots.  Materials and methods Study system My experiment was carried out at the Danum Valley Conservation Area (DVCA, N5.10189°/ E117.688°) and the Sabah Biodiversity Experiment (SBE, N5.16727°/E117.564°) in Sabah, Malaysian Borneo in 2014; both areas are part of the Yayasan Sabah Forest Management 11  Area. This region is characterized by lowland forest dominated by trees in the family Dipterocarpaceae. The DVCA (438 km2) is unlogged forest that has remained relatively undisturbed by human activity, with no accounts of logging or agriculture (Hazebroek et al. 2012). It is the largest expanse of intact primary lowland dipterocarp forest in Sabah (Hazebroek et al. 2012). The logged forest study site is the Sabah Biodiversity Experiment and adjacent forest (Malua Forest Reserve, 335 km2, total area). This area is located 25 km north of the DVCA and was logged for dipterocarp trees greater than 60 cm dbh in the late 1980s with the use of tractors and high lead cables (Berry et al. 2008, Tuck et al. 2016). The SBE itself is found within that area that was previously logged and consists of 124 plots; each plot is 200 x 200 m. From 1999 to 2007, parts of the Malua Forest Reserve were re-logged, although the SBE itself was not (Tuck et al. 2016). In this second logging cycle, a reduced impact logging regime was adopted and trees of 40 cm dbh or greater were targeted (Anon. 2008; Ang et al. 2017). A 500 m unlogged area was left around the SBE during the logging in the early 2000s.  Most dipterocarp reproduction occurs in mast fruiting events every two to ten years, with minor fruit production that results in little to no seedling recruitment occurring in between (Appanah 1985, Ashton et al. 1988). Dipterocarp seeds are wind- and gravity-dispersed and several bird and mammal species consume their fruit and seeds (Curran and Leighton 2000). Dipterocarp seeds germinate very rapidly or else perish – delayed germination and seed banks are unknown. 	Bearded pigs (Sus barbatus) are the largest major predator of dipterocarp seeds in my study system (Curran and Leighton 2000, Curran and Webb 2000). Rodents (e.g. Murids: Rattus, Maxomys spp.) are important seed predators throughout the tropics, though their importance to plants can be habitat- and species-dependent (Blate et al. 1998, Fleury and Galetti 2006, Wells et 12  al. 2009, Wells and Bagchi 2005). Bornean pygmy elephants (Elephas maximus) occur in the study area. They are generalist herbivores and the extent to which they consume dipterocarp seeds, if at all, is not known (M. Bernadus, personal communication). Other locally common herbivore-granivores include sambar deer (Rusa unicolor), two species each of chevrotains (Tragulus spp.), and muntjac deer (Muntiacus spp.), all of which eat fruit and seeds to varying degrees (Hazebroek et al. 2012). Pig-tailed macaques (Macaca nemestrina) occur in the study area and while mostly frugivorous, may consume dipterocarp seeds on occasion (Curran and Leighton 2000). Omnivores that may eat dipterocarp seeds include Malay civet (Viverra tangalunga) and porcupine species (Malayan, thick-spined, and long-tailed porcupine Hystrix brachyura, H. crassispinis, and Trichys fasciculata, respectively). Finally, climbing animals such as squirrels and treeshrews (Sciuridae and Tupaiidae) occur in this system, with the former known to be granivorous (Phillips and Phillips 2016).  Habitat use by granivorous mammals in logged and unlogged forest	I quantified animal communities in my study area with camera traps to determine whether differences in occurrence among animals exhibiting varying levels of granivory might explain patterns in seed mortality, germination or seedling establishment. Passive infrared Reconyx HC500 camera traps were deployed from May to November 2014, with 22 in the DVCA and 20 in the SBE. Cameras were active for an average of 100 days in DVCA and 90 days in SBE. Camera trap photos were date and time stamped, allowing me to estimate the average number of independent detections (photos) per 100 camera trap days — the metric I used for camera site usage. Detections of the same species at the same site were deemed to be independent if they were non-consecutive (i.e., another species was photographed in between) or if they occurred at 13  least one hour apart — such methods are increasingly common in camera trap studies (e.g. Carter et al. 2012, Brodie and Giordano 2013). I used generalized linear models (GLM) to determine if the number of independent animal detections at each camera trap station differed between forest types. Data were fit using a quasipoisson distribution to account for overdispersion. Detections at each camera trap station were pooled for the duration of the field season (May to November 2014). I focused these analyses on mammal taxa that may consume seeds to varying degrees. These taxa included bearded pigs, murid rodents, pygmy elephants, sambar, muntjac, chevrotains, pig-tailed macaques, Malay civets porcupines (data pooled for common, thick-spined, and long-tailed porcupine), small climbing mammals (Sciuridae and Tupaiidae spp. pooled), and sun bears (Helarctos malayanus). Tupaiidae spp. are insectivorous and frugivorous (Phillips and Phillips 2016) but data for this group were pooled with Sciuridae spp. because of difficulties in differentiating between taxonomic groups in photos.  I determined differences in occurrence between forest types for a given mammal species to be significant if the 95% confidence intervals (CIs) of the regression coefficient for “forest type” did not include zero.  Influence of forest type, exclusion treatment, and seed size on seed fate	Seeds from five species of dipterocarp trees were collected during a mast-fruiting event in August 2014: Shorea macrophylla (mean seed mass 13.80 g), Parashorea tomentella (2.95 g), Dryobalanops lanceolata (3.27 g), Shorea leprosula (0.74 g) and Hopea nervosa (0.65 g). These were selected to span a gradient in seed size, allowing us to assess seed fate across a range of values of this important plant reproductive trait (Moles and Westoby 2004), as well as across 14  different scenarios of vertebrate exclusion. Seeds with evidence of insect damage or decay were not used.   I established three experimental blocks in the unlogged forest and three in the logged forest. Blocks were at least 1 km apart and their locations were selected to minimize microsite differences and environmental variation among blocks. Each block consisted of four treatments that excluded: 1) large (elephant), medium (deer, bearded pigs), and small (porcupines, chevrotains) mammals, 2) elephants only, 3) small mammals only, or 4) no mammal species (control). The experimental design was partly factorial because deer and pigs could not be excluded without simultaneously excluding elephants. The arrangement of treatments in each block was not consistent between blocks or forest types as treatment orders were randomly assigned at the time of exclosure construction. Different construction materials were used for experimental exclusion treatments based on which ground-dwelling vertebrates I intended to exclude. To simultaneously exclude large, medium, and small mammals, 2 m tall metal fencing with 2.2 cm mesh was used, coupled with two layers of barbed wire 2 m high around the perimeter of plots to deter elephants. Barbed wire was also placed 1 m off the ground around the perimeter of the metal fencing to further deter bearded pigs and deer from entering. To exclude only small mammals, 50 cm tall fencing was placed around the plot perimeter. To exclude only elephants, two layers of barbed wire 2 m high were placed around the plot perimeter with no other fencing. In the control plots, the corners were marked with flagging tape to delimit plot boundaries. During the experiment no breach of fencing or barbed wire was detected, confirming that the animals meant to be excluded from plots were actually excluded. However, the exclosures were not covered so it is possible that climbing or arboreal mammals entered the plots. 15  Seeds were placed in 1 × 1 m quadrats in the southeast corner of 7 × 7 m plots used for a concurrent study of seedling survival; quadrats were 1 m from the plot perimeter to minimize potential effects of metal leaching from the fencing. Five seeds of each tree species were placed within each quadrat, resulting in 25 seeds in each treatment and 20 seeds per species in each experimental block. One of the blocks in the logged forest did not have seeds placed in the control plot, resulting in an uneven number of seeds between the logged and unlogged forest, but this was accounted for in the analysis (below). Seeds were placed 10 cm apart, subject to the constraint that two seeds of the same species were never adjacent. Each seed was numbered with a unique ID in order to track its fate over time. A 1 m long string was attached to each seed and secured to 10 cm nail, which was then placed into the soil to locate seeds that had been moved.   Plots were visited approximately every 7 days for 60 days. These species germinate in less than 30 days (O’Brien et al. 2013) so the 60-day period allowed for germination and seedling establishment. Dipterocarp seeds do not form seed banks because of their sensitivity to desiccation and fast decomposition (O’Brien et al. 2013) so it was assumed that seeds that had not germinated at the end of the study would not germinate at all. At each census, seed status was recorded as dead from granivory (i.e. browsed and radical damaged), eaten but alive (i.e. browsed with only cotyledon damage), removed or disappeared (i.e. no longer attached to the string and absent from the plot), germinated (i.e. radical emergence), or established as a seedling (i.e. established root and standing stem). Post-germination seed mortality caused by seed predation can occur (Terborgh et al. 1993), so more than one response was possible for an individual seed during a single visit. Also, granivory — whether leading to mortality or not — was possible both before and after germination. I did not quantify the mortality of seeds predated upon after removal from treatment plots. While rodents may transport seeds to burrows, scatter-16  hoarding in this system is unlikely due to the rapid germination of dipterocarp seeds (Wells and Bagchi 2005). Responses of each species were aggregated from all censuses in each treatment per block to calculate final proportions of seeds exhibiting each response. Therefore, all results represent the final fate observed for seeds exhibited at the conclusion of the experiment. Generalized linear mixed effect models (GLMM) were used to assess the final proportions of seeds killed by granivores, eaten but alive, removed, germinated, or that established as seedlings as functions of exclusion treatment (a fixed factor with 4 levels), forest type (a fixed factor with 2 levels; logged and unlogged) and species identity of the seed (a fixed factor with 5 levels). I also tested for forest × exclusion treatment, tree species × exclusion treatment, and tree species × forest interactions. I performed these GLMM analyses on final proportions using a binomial distribution weighted by the number of seeds of each species initially planted inside each treatment of each block. I included random effects for experimental block nested in forest site (a random term with six levels) and used model selection based on Akaike Information Criteria (AICc) to determine which model best explained each response variable. I used R (version 2.15.2) for all analyses (http://r-project.org). GLMM models were run using the lme4 package in R (Bates et al. 2015). Post-hoc pairwise tests for fixed effects were performed for each response variable using the multcomp package (Hothorn et al. 2008) for models with the lowest AICc. 	 Results Habitat use by granivorous mammals in logged and unlogged forest Bearded pigs, muntjac, and chevrotains were the most frequently detected granivores while elephants, sun bear, and arboreal granivores (squirrels and treeshrews) were among the least 17  detected (Fig 2-1). There were significantly more detections of bearded pigs in unlogged forest than in logged forest (regression slope = 0.9; 95% CI: 0.1 to 1.6).  	Influence of forest type, exclusion treatment, and seed size on seed mortality, germination, and establishment The proportions of seeds removed in unlogged and logged forest were 0.10 and 0.08, respectively. Removal was statistically unrelated to forest type (P = 0.34) and, overall, was higher in control plots than in exclude-large, medium, and small- mammal treatments (F3,111 = 2.91, P = 0.05). No species-related trends in seed removal were detected (P = 0.41). Granivore-induced seed mortality was only affected by seed species, treatment and forest site as main effects, so the interaction terms were dropped.  The overall proportion of seeds killed by granivores was greater in the logged forest (0.77) than in the unlogged forest (0.66; F1,113 = 8.25, P = 0.03; Fig 2-2). Seed mortality was not significantly different between exclosure treatments (P = 0.99) and mortality was similar among large-seeded species (P. tomentella compared to D. lanceolata: P = 0.44; S. macrophylla compared to D. lanceolata: P = 0.99, S. macrophylla compared to P. tomentella: P = 0.54). The only size-related differences in mortality were between the smallest species (H. nervosa and S. leprosula) and P. tomentella, the species with the median seed weight. Mortality was less likely among P. tomentella seeds compared to the smaller species. (F4,110 = 2.58; H. nervosa: P = 0.03; S. leprosula: P = 0.05). Seeds of the largest-seeded species (S. macrophylla) were more likely to survive granivore attack than either S. leprosula or H. nervosa (F4,110 = 9.58, P < 0.01). Also, both H. nervosa and S. leprosula were less likely to survive granivore attack than the second largest seeded-species, D. lanceolata (P < 0.01).  18  A smaller proportion of seeds germinated in the logged forest (0.54 of all seeds) than in unlogged forest (0.79 of all seeds; F1,113 = 5.01, P < 0.01). Germination was also lower in control plots than in exclude-large, medium, and small mammal treatments (proportion germinated in control = 0.60 and proportion germinated in exclude- large, medium and small mammal treatments = 0.73; F3, 111 = 5.34, P < 0.01; Figs 2e & h). There was a significant species × forest interaction (F4, 110 = 2.72) for some of the pairwise comparisons: germination was greater for small-seeded species in unlogged forest (proportion of H. nervosa that germinated = 0.63 and S. leprosula = 0.78; P = 0.04) and was lowest for S. macrophylla in logged forest (S. macrophylla vs. S. leprosula in logged forest: P < 0.01). Overall seedling establishment was greater in logged forest but this difference was not significant (proportion of seeds that established in logged = 0.24, proportion that established in unlogged = 0.13; P = 0.88). No significant differences in establishment were found among treatments (P > 0.76 for all pairwise treatment comparisons). The best model for seedling establishment included a forest × treatment interaction (Table 2-1). Within logged forest, fewer seeds established in control plots than in treatments simultaneously excluding large, medium, and small herbivores (F3,111= 7.06, P = 0.02). In the logged forest, establishment levels were higher in the absence of large, medium, and small mammals relative to treatments excluding only elephants (P < 0.01, Fig 2-3). Small-herbivore exclusion plots experienced more seedling establishment relative to treatments excluding only elephants within logged forest (P < 0.01). No treatment differences in seedling establishment were found within unlogged forest. Species-related differences in overall establishment were not found, with the exception for P. tomentella, the species with the median seed weight, which had the highest proportion of established seeds (F4,111 = 0.64, P < 0.01).  19      Figure 2-1 Habitat use of mammal granivores Habitat use (average number of independent detections per 100 camera trap days) of mammalian granivore species in Danum Valley (unlogged forest, black) and the Sabah Biodiversity Experiment (logged forest, grey), Malaysian Borneo from May to November 2014. Standard error bars are shown. Asterisks denote significant differences in relative site usage of animals between forest types based on model coefficients whose 95% confidence intervals do not include zero.    20   Figure 2-2 Seed mortality and germination in logged and unlogged forest exclosures  Overall proportions (± SE) of dipterocarp seeds killed by granivores in experimental plots excluding A) large, medium, and small mammals, B) elephants only, C) small mammals only, and D) no species, in logged (triangles) and unlogged (circles) forest in Malaysian Borneo for five dipterocarp species (Hopea nervosa, Shorea leprosula, Dryobalanops lanceolata, Parashorea tomentella and Shorea macrophylla). Also shown are the overall proportions of seeds that germinated in plots excluding E) large, medium, and small mammal herbivores, F) elephants only, G) small mammals only, and H) no species. Dipterocarp seeds on the X-axes are ordered from smallest (left) to largest (right) seed mass. The proportion of seeds killed by granivores was greater in logged forest, but no significant differences among exclosure treatments were detected. Seed germination was higher in unlogged forest, in control plots, and for small-seeded species.  21     Figure 2-3 Seedling establishment in logged and unlogged forest exclosures Overall proportion of seeds (± SE) that established as seedlings out of initial seeds planted in experimental plots excluding large, medium, and small mammals, elephants only, small mammals only, and no species, in logged (triangles) and unlogged (circles) forest in Malaysian Borneo for five dipterocarp species: Hopea nervosa, Shorea leprosula, Dryobalanops lanceolata, Parashorea tomentella, and S. macrophylla. Data for these species were pooled because significant differences in establishment between species were not found, with the exception of P. tomentella seeds, which were more likely to establish.      22  Table 2-1 Model parameters and AIC weights for seed fate  Number of model parameters, Akaike weights (wi), and delta AIC (∆AIC) for binomial general linear mixed effect models. We tested forest (“forest”; logged or unlogged), treatment (“treat”; large, medium and small- mammals excluded elephant-excluded small-excluded, and exclude none), and dipterocarp seed species (“spp”; H. nervosa, S. leprosula, P. tomentella, D. lanceolata, and S. macrophylla) on each response variable. I also tested for forest × treatment (“forest × treat”), species × treatment (“spp × treat”), and species × forest (“spp × forest”) interactions. All models included the same random effects of (1 | site: block). Akaike weights in bold represent the best model for each response variable.                                                   Response                               No. model  Removed or   Dead from  Eaten but  Germinated  Established  parameters  missing  granivory  alive         as seedling                  Model   wi ∆AIC  wi ∆AIC  wi ∆AIC  wi ∆AIC  wi ∆AIC                                                    Forest + treat + spp + spp × treat + 28  0.00 38.82  0.00 25.93  0.00 28.34  0.00 14.20  0.06 5.35 forest × treat + spp × forest                                  Forest + treat + spp + forest × treat  12  0.04 5.99  0.28 1.38  0.09 0.10  0.26 1.03  0.93 0.00                  Forest + treat + forest+ spp × treat  21  0.00 21.62  0.00 14.72  0.00 20.19  0.00 9.21  0.00 13.03                  Forest + treat + spp+ spp × forest  13  0.08 4.79  0.16 2.46  0.47 0.00  0.44 0.00  0.00 27.09                  Forest + treat + spp 9  0.88 0.00  0.56 0.00  0.45 11.59  0.30 0.75  0.01 20.68 23  Discussion  Most tropical forests are simultaneously affected by multiple human activities (Wright 2005), but considering only their additive effects could hinder our ability to understand consequences of these disturbances for forest regeneration. The combined impacts of habitat disturbance and altered animal abundances could affect plant recruitment in ways that are not apparent when studied in isolation (Peres 2000, Guariguata et al. 2002, Poulsen and Clark 2011, Brodie et al. 2014). Though the effects of defaunation on plant recruitment may vary (Kurten 2013), my findings suggest that the interactive effects of defaunation and logging could influence forest regeneration by increasing the probability of seedling establishment and recruitment, in contrast to findings elsewhere (Asquith et al. 1997, Galetti et al. 2015, Nunez-Iturri et al. 2008, Rosin and Poulsen 2016). My assessment of these interacting factors suggests two distinct patterns to seedling establishment that are logging history-dependent. Logged forests had higher seed mortality and lower germination but higher survival of germinated seeds that turned into seedlings (Fig. 2-4). Unlogged forests had lower mortality and higher germination but lower survival of germinated seeds (i.e. high mortality of young seedlings; Fig 2-4). At the seed stage, mortality was higher in the logged forest and the effects of logging outweighed those of experimental defaunation. However, lower seed mortality in the unlogged forest did not necessarily improve the chances of survival at future life stages, as seed establishment rates were statistically similar between forest types (Fig 2-4). Herbivory by invertebrates or pathogen infection (which were not excluded from exclosure plots) after seeds germinated may have contributed to overall mortality in the unlogged forest, compensating for the reduced mortality of ungerminated seeds. Herbivory by insects can increase following the seed germination stage in mature forest (Toy and Toy 1992, Lyal and Curran 2000, Notman and 24  Gorchov 2001, Nakagawa et al. 2005), whereas the removal of reproductive trees for logging can reduce invertebrate and pathogen populations (Intachat et al. 1997). Forest gaps created by tree removal in logged forest may also increase light availability to the understory, promoting seedling growth in disturbed forest (Ashton 2010), contributing to the similar establishment levels I observed between forest types. Seed mortality was greater in logged forest despite lower usage of that habitat by bearded pigs. Differences in seed mortality between logged and unlogged forest could be related to changes in food availability and animal foraging patterns driven by logging rather than by hunting-induced changes in seed predator abundance (Barnes et al. 1991, Chapman et al. 2000, Farwig et al. 2006). Hunting is limited in my study area and no hunters were detected by camera traps in 2014 (nor in 2013 or 2015); this is partly due to the restricted access and remoteness of this forest area. Seed production tends to be lower in selectively logged forest because of the reduced abundance of reproductive adult trees (Ghazoul and McLeish 2001), so the chance of large seed predators becoming satiated may be lower in logged forest. Where dipterocarp seeds are available, seed predators (e.g. bearded pigs or pig-tailed macaques) would therefore spend more time foraging in an area of logged forest than they would in an equivalent portion of unlogged forest, causing higher localized mortality in the former habitat (Curran and Leighton 2000).  Although seed removal was highest in control plots, excluding large, medium, and small-bodied mammals was not associated with reduced seed mortality, contrary to my initial predictions. The absence of large vertebrates could lead to increased seed predation by small taxa (e.g. Rattus spp.) if the latter is released from competition in the absence of large animal taxa (McCauley et al. 2006, Galetti et al. 2015, Rosin and Poulsen 2016). However, I found similar 25  levels of seed mortality across treatments suggesting that small ground-dwelling animals were not released from competition in the absence of larger mammal taxa. Logged areas may contain higher rodent abundances compared to unlogged forest (Malcolm and Ray 2000) but camera traps consistently captured fewer vertebrates — both small and large —  in the logged forest, and rodent occurrence did not differ between forest types.  Logging reduced germination in seeds that were not predated, possibly by altering forest structure in ways that impacted the seedling microhabitat environment. Forest gaps created by logging, for example, increase the amount of light reaching the forest floor, thereby increasing the likelihood of seed desiccation (Itoh et al. 1995). Germination was also higher among small-seeded species (H. nervosa and S. leprosula) in the logged forest, illustrating that reduced mortality among large-seeded species does not necessarily lead to higher germination levels. In contrast to logging, experimental defaunation did not reduce seed survival, though reduced seed predation (Beckman and Muller-Landau 2007, Dirzo et al. 2007, Hautier et al. 2010) and increased germination in the absence of medium to large-bodied terrestrial mammals have been documented elsewhere (Paine and Beck 2007).  Unsustainable hunting of animals and habitat disturbance in the form of selective logging are two important threats facing wildlife throughout the tropics (Corlett and Primack 2011). Many tropical forests are simultaneously affected by both, but the consequences of each for forest regeneration are usually studied separately or additively, and usually on a single plant demographic stage (Wang and Smith 2002, Comita et al. 2014). I found that seed mortality was higher in logged forest, but the absence of medium and large mammals had little influence on seed mortality or on seedling recruitment. Instead, seedling recruitment was more likely where logging and animal loss occurred together. My findings suggest that defaunation could have 26  important consequences for forest regeneration and community composition in logged areas — a particularly striking result given that the overexploitation of both large mammals and trees continue to be widespread in tropical regions.    27     Fig 2-4. Percent of seeds killed, germinated, established in unlogged and logged forest exclosures Percentage of dipterocarp seeds killed and eaten by granivores (% dead), germinated, removed, and/or that established as seedlings in animal exclusion treatments in a) unlogged and b) logged forest in Malaysian Borneo. Treatments excluded large, medium, and small mammals, elephants only, small herbivores only, or no species (control). Granivore-induced mortality and seed removal were possible both before and after germination. Data for five dipterocarp species 28  (Hopea nervosa, Shorea leprosula, Dryobalanops lanceolata, Parashorea tomentella and Shorea macrophylla) were pooled and percentages shown are out of the initial pool of planted seeds in each treatment within logged and unlogged forest. Overall percentages do not add up to 100% because more than one seed fate was possible for individual seeds (e.g. germination and establishment). Also, not all unpredated seeds successfully germinated.29  Chapter 3: Experimental defaunation alters seedling functional traits but not diversity   Defaunation, the decline or loss of medium- and large-bodied animal taxa, mainly through overhunting, threatens ecosystems around the world (Peres 2000, Dirzo et al. 2014). There is widespread concern that the decline of large animals could have cascading impacts on plant communities (Harrison et al. 2013, Dirzo et al. 2014). But such cascading effects remain difficult to predict because the same animal taxa can have both negative (e.g. herbivory, granivory) and positive (e.g. seed dispersal, pollination) effects on plants (Beck 2006). Moreover, in some cases, the loss of large animals could have limited effect on plant communities if rates of herbivory by smaller animals increase (Wright 2003, Goheen et al. 2004). Indeed, the loss of large mammals has led to reduced plant recruitment and diversity in some systems (Asquith et al. 1997, Goheen et al. 2004, 2010) but not others (Brocardo et al. 2013).  Defaunation seldom occurs in isolation –habitat disturbances such as logging are widespread in forest ecosystems around the world and can induce dramatic changes in tree communities. Species interactions are known to vary strongly with ecological context (Bronstein 1994, Chamberlain et al. 2014), suggesting that the impacts of defaunation on plant communities could be different in forests where the abiotic conditions have been altered by logging than in undisturbed forests. While the impacts of defaunation (Camargo-Sanabria et al. 2015) and logging (Curran et al. 1999) on seedling dynamics have been assessed separately, we have little understanding of whether or how they might interact. 30  An assessment of how defaunation and habitat disturbance affects seedling communities by altering diversity patterns or the distribution of plant traits is necessary to better understand how these threats might affect related ecosystem processes. Three important plant traits that could be affected by defaunation and logging are fruit size, wood density, and specific leaf area. Declines in large frugivores have led to reduced dispersal and recruitment of large-fruited plants in some areas, altering community composition by favoring the survival of small-seeded species (Fadini et al. 2009, Galetti et al. 2013). Wood density is positively associated with aboveground carbon storage but reduced seed dispersal of large-seeded taxa in defaunated areas may decrease average wood density across the tree community; this could affect carbon cycling and climate change (Bello et al. 2015, Osuri et al. 2016). Specific leaf area (SLA) is commonly used as an indicator of overall plant life-history strategy (Poorter and Bongers 2006). Defaunation could increase mean SLA by promoting the growth of fast-growing, abiotically dispersed taxa (with high SLA) at the expense of slower, longer-lived, vertebrate-dispersed taxa (low SLA) (Coley et al. 1985, Grime et al. 1996). Understanding the potentially complex cascading impacts of defaunation on plant traits and diversity requires the use of experimental manipulations. Recently, the use of vertebrate exclusion plots has demonstrated that the loss of ground-dwelling mammals in Mexico reduced seedling diversity and differentially affected plant taxa in Panama based on their morphological traits (wood density and seed mass) (Camargo-Sanabria et al. 2015). But our limited knowledge precludes a general understanding of the community-wide effects of defaunation, particularly in terms of incorporating the concurrent impacts of habitat disturbance (Curran et al. 1999, Fragoso et al. 2003, Camargo-Sanabria et al. 2015).	31  Here, I assess the combined impacts of defaunation and habitat disturbance on tree seedling community composition. We used vertebrate exclosures in logged and unlogged forest in Sabah, Malaysian Borneo, from 2013 to 2016 to examine how the loss of different animal groups affected both the taxonomic diversity and the distribution of key morphological traits among tree seedlings. Camargo-Sanabria et al. (2015) used similar exclosures in Mexico to test the hypothesis that animal loss reduced tree seedling diversity. We extend this to also assess whether experimental defaunation alters the distribution of three plant traits known to be important to ecosystem function, fruit size, wood density, and specific leaf area (SLA), and to compare these changes in logged versus unlogged forests. Our study provides a unique experimental assessment of how the direct effects of two widespread human impacts, logging and hunting, might individually and jointly affect the structure and function of tropical plant communities. 	Methods Study system Data were collected in the Danum Valley Conservation Area (DVCA, N5.10189°/ E117.688°) and the Sabah Biodiversity Experiment (SBE, N5.16727°/E117.564°) in Sabah, Malaysian Borneo; both areas are part of the Yayasan Sabah Forest Management Area (YFSMA). This region is characterized by lowland forest dominated by trees in the family Dipterocarpaceae.  The unlogged forest study site is the DVCA (438 km2); itis the largest expanse of primary dipterocarp forest in Sabah and is classified as Class 1 (“Totally protected”) reserve protected from resourced extraction(Hazebroek et al. 2012). The Sabah Biodiversity Experiment (SBE) is the logged forest study site and is located 25 km north of the DVCA. In 1984 and 1986, this area 32  was logged for dipterocarp trees greater than 60 cm dbh with the use of tractors and high lead cables (Berry et al. 2008, Tuck et al. 2016). From 1999 to 2007, the forest adjacent to the SBE (Malua Forest Reserve, 335 km2) was re-logged for dipterocarps, although the SBE itself was not (Tuck et al. 2016). In this second logging cycle, dipterocarp trees 40 cm dbh or more were targeted for logging (Anon. 2008; Ang et al. 2017). In the early 2000s, a 500 m area of unlogged forest was left around the SBE. Large-bodied vertebrates in the study area include the Bornean pygmy elephant (Elephas maximus borneensis), sambar deer (Rusa unicolor), two species of muntjac (red: Muntiacus muntjak; yellow: M. atherodes), and bearded pig (Sus barbatus). Small-bodied herbivores include two species of chevrotain (lesser: Tragulus napu and greater: T. javanicus) and three porcupine species (Malayan: Hystrix brachyura; long-tailed: Trichys fasciculata; thick-spined: Thecurus crassipinis, Appendix Table A-1). 	 Field experiments We used experimental exclosure plots in logged and unlogged forest to assess animal-related impacts on seedling communities. Three experimental blocks were established in the unlogged forest and three in the logged forest (within the SBE) (see Chapter 2 Methods for additional details regarding construction of exclusion plots). Blocks were within 1.5 m of the field stations. Treatments excluded: 1) all terrestrial mammalian herbivores (elephant, deer, bearded pigs, porcupines, chevrotains), 2) elephants only, 3) small mammals only (porcupines, chevrotains), or 4) no mammal species (control). The arrangement of treatments in each block was not consistent between blocks or forest types, as treatment orders were randomly assigned at the time of exclosure construction. 	33  In 2013, 1489 seedlings (<1 m high at breast height and <1 cm circumference) of tree species were marked using aluminum tags containing unique identification numbers. Seedling height was measured to the apical meristem. Tagged seedlings were identified to genus; distinguishing among congeners was considered extremely difficult even for an expert botanist (M. Bernadus, pers. comm.). In 2014 and 2015, seedlings were re-surveyed twice a year, with surveys approximately 3 months apart. Seedling plots were revisited in 2016, approximately 10 months after the second 2015 survey period, resulting in a total of six surveys. At each survey period after 2013, seedlings were recorded as dead, alive, or missing. Seedlings were recorded as dead if the stem was broken or if the tag was found unattached to any seedling. Seedling heights were re-measured at each survey. Seedlings were recorded as missing if the tagged seedling could not be located. Efforts were made to locate previously missing seedlings in subsequent surveys.  To determine whether plant traits inside our experimental plots changed over time or in association with logging or mammal exclusion, I obtained trait values for seedling genera represented in our plots from published and unpublished sources and the Dryad Database (http://www.datadryiad.org, Appendix A-3). Focus was on three plant traits: wood density (g/cm3), specific leaf area (SLA) (cm2/g), and fruit length (mm). I determined community-weighted mean (CWM) values of traits in each treatment of each experimental block at each sample period. CWMs are the mean trait values across genera, weighted by the relative abundance of each genus in the community (Garnier et al. 2004). The use of weighted means incorporates information on the distribution of a given trait in communities and has been shown to predict ecosystem function better than functional diversity indices alone (Laughlin 2011, Roscher et al. 2012, Cohen et al. 2014). 34  Data analyses I used linear mixed effects models (LMM) to determine the influence of exclusion treatment (a factor with 4 levels), forest type (a factor with 2 levels; logged and unlogged), and sample period (an ordinal variable) on means of Shannon diversity and community weighted plant traits (wood density, fruit length, and SLA). All continuous variables were log-transformed and all models included a random effect for experimental block nested in forest (a random term with six levels). Individual seedlings recorded as “dead” at one sample period were removed from analysis at the next period, such that the number of living seedlings decreased with each subsequent sample. LMM were used to assess how seedling growth varied between exclusion treatments and forest type. General linear mixed effect models (GLMM) were used to model overall seedling mortality as a function of treatment and forest type, with mortality modeled as a binary response (alive or dead). Shannon diversity indices were generated using the vegan package (Dixon 2016). Rarefaction curves for genus richness were calculated using the iNEXT package (Chao et al. 2014, Hsieh et al. 2016) to compare richness at the onset and conclusion of the study in each treatment and forest type.  I ran separate models for each plant trait (fruit length, wood density, and SLA), with community trait values modeled as a function of forest type, exclosure treatment, and sample period. Because seedling communities in our experimental plots may have shown initial variation in these traits (i.e., unrelated to experimental treatment), I determined treatment effects to be significant if treatment × sample period interaction terms were significant; this would indicate that exclosure treatments induced changes in a given community plant trait. I also assessed whether mean trait values and taxonomic diversity (the number of genera) differed between logged and unlogged forest at the onset of our study. All analyses were performed in R 35  (version 3.2.4, http://r-project.org). Mixed effect models were run using the lme4 package (Bates et al. 2015) in R.   Results In 2013, 1489 seedlings from 82 genera in 42 families were tagged (Appendix Table A-2). Overall mortality was lower where all herbivores were excluded than in control plots (β	= -1.30; 95% CI: -2.09 to -0.50) and where only small herbivores were excluded relative to control plots (β	= -1.22; 95% CI: -1.90 to -0.54, Appendix Fig A-1). In 2013, the diversity of seedling genera, as measured by the Shannon index (Shannon 1948) was already higher in logged forest than in unlogged forest (β	in 2013 = -0.35; 95% CI:     -0.66 to -0.04), and forest-related differences were still present at the conclusion of the study (β in 2016 = -0.55; 95% CI: -1.08 to -0.02). Seedling diversity declined over subsequent sampling periods in both forests (β	= -0.10; 95% CI: -0.22 to -0.04) but no forest × treatment × sample period interactions were significant. Rarefaction curves for genus richness showed that within each type of exclosure treatment, richness was higher in the logged forest. As with the Shannon diversity indices, rarefaction-based richness estimates declined from 2013 to 2016, but no significant differences were detected among types of exclosures in rarefied genus richness (Fig 3-1). At the onset of the study, community weighted mean fruit length was higher in unlogged than in logged forest (β = 0.14; 95% CI: 0.05 to 0.23). Within control plots, community weighted mean fruit length decreased over subsequent sampling periods (β = -0.05; 95% CI: -0.09 to  -0.01), and while only marginally significant, this effect was greater in the logged forest (β	=     36  -0.06; 95% CI: -0.12 to -0.003). Excluding small herbivores in the unlogged forest decreased community fruit length over time (β	= -0.06; 95% CI: -0.11 to -0.002). The exclusion of elephants (β	= -0.04; 95% CI: -0.10 to 0.02) or all mammal herbivores (β	= -0.01; 95% CI: -0.11 to 0.01; Fig 3-2) did not significantly alter mean fruit length in either forest type. At the conclusion of the study, fruit length was statistically similar between forest types (β = 0.14; 95% CI: to -0.06 to 0.34). For community wood density and specific leaf area, no forest × treatment × sample period interactions were significant. Mean wood density did not differ significantly between logged and unlogged forest either at the beginning (β	= -0.42; 95% CI: -0.25 to 0.17) or end (β	= 0.06; 95% CI: -0.27 to 0.22) of the study. Also, wood density did not change over time within forests or within treatments. Similarly, we detected no significant difference in community weighted SLA between logged and unlogged forest in 2013 (β = 0.01; 95% CI: -0.06 to 0.09) or in 2016 (β = -0.02; 95% CI: -0.10 to 0.07). SLA did not change over time within either forest type or within any of the treatments by the end of the study.    37   Figure 3-1 Rarefied genera richness in logged and unlogged forest exclosures Rarefaction curves for richness of seedling genera present initially (2013) and at the end (2016) of my study in logged and unlogged forest. Solid lines represent richness estimates based on interpolation and dashed lines represent richness estimated based on extrapolation to 300 individuals. Genera richness was not significantly different at the conclusion of the study relative to richness at the onset of the study within any of the exclosure treatments. Richness of genera was already higher in logged forest in 2013 and this was still the case in 2016. 95% confidence bands are shown for each time point.  38   Figure 3-2 Distribution of fruit length for seedling genera in logged and unlogged forest exclosures Change in log10-transformed community fruit length (mm) for seedling genera in exclusion treatments in unlogged and logged forest. Standard error bars are shown.  Exclusion plots were visited six times over the course of my study (July 2013, June 2014, August 2014, June 2015, June 2015, and April 2016) during which individual seedlings were recorded as alive, dead, or missing.   39  Discussion Increasing evidence points to impacts of hunting (Kurten et al. 2015) and anthropogenic habitat disturbance (Harrison et al. 2013), in isolation, on plant communities, but we have very little understanding of the potential interactions among these factors. Our knowledge is also limited with respect to whether animal groups differ in driving seedling mortality or the distribution of seedling traits. Treatment-related differences in mortality did not affect genera richness. However, animals were more likely to kill seedlings from large-fruited genera, which led to a reduction in mean fruit length for tree seedlings exposed to herbivory. This could be attributed to small to medium-bodied herbivores, such as ungulates which may forage selectively and disproportionately consume large-fruited seedlings (Augustine and McNaughton 1998, Paine and Beck 2007). Large-fruited, non-dipterocarp seedling genera were more prevalent in logged forest and may have been more vulnerable to predation or trampling than dipterocarps, which are better chemically defended (Ashton et al. 1988). In my study area, ungulates also act as important dispersal agents for seeds from large fruits (Corlett 1998, Phillips and Phillips 2016). My results suggest that the future benefits of seed dispersal (e.g. germination of seeds in new patches and reduced likelihood of negative density-dependent morality) may be offset by immediate increased herbivory or trampling at the seedling stage. This illustrates how animals can affect forest regeneration in opposing ways, and highlights the importance of assessing species interactions across multiple stages of the plant life-cycle.  Logging had a stronger effect on seedling diversity than experimental defaunation. I found that genus richness was consistently higher in logged forest (Berry et al. 2010) which could be attributable to habitat heterogeneity or competitive release created by the removal of dipterocarp trees (Cannon et al. 1994). Changes to forest structure, such as soil disturbances and 40  the creation of canopy gaps, facilitate the establishment of fast growing, non-dipterocarp taxa (Cannon et al. 1994), which can persist decades after logging (Pinard and Cropper 2000).  Seedling survival was lower in the presence of large, medium, and small-bodied mammals than where these species were excluded, but decreased survival did not lead to differences in seedling diversity, at least in the short term (4 years). My findings are consistent with those of Roldan and Simonetti (2001) from Bolivia and Brocardo et al. (2013) from Brazil, but are in contrast to results from Mexico and Peru (Paine and Beck 2007, Camargo-Sanabria et al. 2015). These varied results from different study systems could be due to idiosyncratic responses to herbivory among the dominant trees at each site. In my study system, for example, the most common trees were dipterocarps, and mammals were not more likely to kill seedlings from this family than from any other. In contrast, disproportionate vertebrate impacts on dominant taxa in other systems may have altered diversity (Roldán and Simonetti 2001).  In contrast to mean fruit length, animal-induced seedling mortality did not significantly alter the distribution of wood density and SLA. Wood density affects carbon storage in forests, which is a critical pool in the global carbon cycle (Pan et al. 2011). Disturbance (Bunker et al. 2005) or defaunation (Bello et al. 2015, Kurten et al. 2015) can lead to reductions in mean wood density in systems where dominant and carbon-rich trees rely on animals for seed dispersal, such as in the Neotropics (Kurten et al. 2015, Osuri et al. 2016). In contrast, tropical forests in Borneo are dominated by dipterocarps that are targeted for selective logging. The loss of mammal frugivores may have little impact on the average wood density of the tree assemblage because dipterocarp seeds are abiotically dispersed (Harrison et al. 2013, Osuri et al. 2016).  Leaf traits can be good predictors of plant growth and survival (Poorter and Bongers 2006), but whether animals actually alter the distribution of leaf traits in a community is unclear 41  (Kurten 2013). I expected SLA to be higher in the absence of mammal herbivores due to enhanced survival of poorly defended plant taxa (i.e. those with high SLA) but similar to findings in Panama (Kurten et al. 2015), animals had no significant effect on seedling SLA in our study.  Changes in forest structure caused by logging that occurred thirty years prior led to elevated seedling diversity and richness. The combined influence of vertebrate herbivory and habitat disturbance affected community fruit length, as animals disproportionately increased the mortality of seedlings from large-fruited genera in logged forest. However, in isolation (i.e. without accounting for interactions with logging), short-term experimental defaunation did not affect seedling diversity in my study area and did not cause shifts in seedling trait distribution. Terrestrial vertebrates may therefore, alter plant traits even without changing plant diversity, with potentially important long-term implications for ecosystem function in many tropical forests.   	42  Chapter 4:  The importance of mast-fruiting for vertebrates in a faunally intact ecosystem   The manner in which animals respond to spatio-temporal variation in resource availability can have important consequences for individual fitness, population dynamics, and species interactions (Loiselle and Blake 1991, Ostfeld and Keesing 2000, Yang et al. 2008). Mast fruiting, one of the most well studied resource pulses (Kelly and Sork 2002, Kelly et al. 2008, Yang et al. 2008, 2010), can dramatically impact animal populations and communities (Curran and Leighton 2000, Ostfeld and Keesing 2000). In eastern North America, for example, the masting of oak trees leads to large increases in abundance and shifts in distribution of rodents, deer, bears, and birds (Ostfeld et al. 1996, Koenig and Knops 2005). Increased vertebrate abundance can boost tick populations and, in turn, Lyme disease prevalence (Ostfeld et al. 1996). Individual mast years can therefore have long-lasting and cascading effects across multiple trophic levels (Koenig and Knops 2005, Yang et al. 2010). Nearly all documented cases of the population- and community-level repercussions of masting are from temperate systems where certain key consumers of mast have been greatly reduced or extirpated (Ellsworth and McComb 2003, Kelly et al. 2008). In eastern North America, the once mega-abundant passenger pigeon (Ectopistes migratorius) was a highly mobile specialist on acorn (Quercus spp.), beechnut (Fagus grandifolia), and chestnut (Castanea dentata) seeds (Schorger 1955, Bucher 1992). Pigeon movements across the landscape were tightly linked with mast production, and the large number of foraging individuals could have left few resources for other animals, particularly terrestrial and less-mobile taxa. The strong 43  magnitude of community-wide effects of masting that have been documented in oak forests (Ostfeld et al. 1996, Koenig and Knops 2005) could therefore be partially attributed to the extinction of passenger pigeons (Kelly et al. 2008). Studies from oak systems highlight that masting can have important effects on species interactions, such as resource competition and also on species abundance via population growth. However, the extent to which multiple trophic levels are affected by masting in systems with intact vertebrate communities is not clear, and we know little about whether similar population- and community-wide ripple effects are common to masting systems.    In Southeast Asia, Dipterocarpaceae trees may comprise 80% of the upper canopy layer and reproduce approximately every two- to ten-years in some of the largest masting events on the planet (Appanah 1985, Whitmore 1998). Masting events can vary in magnitude (Curran and Leighton 2000) but synchronous mast fruiting among dipterocarp (and many non-dipterocarp) species at regional scales clearly defines mast versus non-mast years, with fruit production being minimal to non-existent in the latter (Kelly 1994). However, the importance of dipterocarp masting for tropical forest vertebrate communities is not well studied. Curran and Leighton (2000) quantified responses of nomadic and resident animal consumers to masting, with a focus on granivorous species. Two highly mobile vertebrate granivores, the bearded pig (Sus barbatus) and the long-tailed parakeet (Psittacula longicauda), track masting dipterocarps while resident species such as pheasants (Lophura spp.) and crested partridges (Rollulus roulroul) may increase reproductive effort during mast years in response to increased food availability (Curran and Leighton 2000). Some animals may respond to mast events more strongly than others (Ostfeld et al. 1996) and while species responses to masting have been documented in Southeast Asia, the 44  community-level effects on vertebrates via altered species interactions or abundances remain unclear.   Vertebrate responses to mast fruiting could potentially be shaped by habitat disturbance. Human alterations to forest structure, for example via logging, can reduce the magnitude and intensity of resource pulses (Curran et al. 1999), possibly changing how animals use the forest (Grindal and Brigham 1999, Curran and Webb 2000). From 2000 to 2005, 20 - 50% of tropical forest area was subject to logging (Asner et al. 2009). In equatorial Asia, the removal of adult dipterocarp trees reduced the magnitude of masting (i.e., seed production) and could increase the overall proportion of animal-induced seed mortality (Curran et al. 1999, Curran and Webb 2000) without necessarily affecting seedling establishment. Animals may track resources across large (e.g. landscapes) (Rey 1995, Whitney and Smith 1998, Curran and Leighton 2000) and small spatial scales (e.g. within patches, plant communities) (Senft et al. 1987, Levey 1988, Kotliar and Wiens 1990, Tellería and Pérez-Tris 2003, Tellería and Pérez-tris 2007). However, the influence of food abundance on animal habitat use may differ at large versus small scales (Garcia et al. 2011). Similarly, the relative influence of biotic (e.g. species interactions), intrinsic (e.g. morphological constraints), or extrinsic factors (e.g. habitat structure) on animal responses to food resource heterogeneity may vary with spatial scale (Grindal and Brigham 1999, Curran and Leighton 2000, Johnson et al. 2001, Valeix et al. 2009, Garcia et al. 2011, Schleuning et al. 2011). For example, at large spatial scales, resource tracking may be more likely among mobile species (i.e. bearded pigs and parakeets) compared to less mobile species (i.e. porcupines).  Here, I examine the impacts of mast fruiting on a suite of species of vertebrate consumers in a faunally intact ecosystem across two spatial scales. This study took place in Malaysian 45  Borneo in both logged and unlogged forest. I assessed whether animal use of habitat patches varied between a non-mast year (2013), a major mast event (2014), and a minor mast event (2015). Although I could not experimentally manipulate large-scale masting, several aspects of our study design allowed me to move beyond pure correlations to potentially assess the causality in the variation in animal occurrence rates. First, I measured site use in both frugivorous (including granivorous) and non-frugivorous taxa – this helped to assign any observed changes in local site usage to patterns in fruit availability rather than to other factors (e.g. climate) that might affect all feeding guilds. Second, I used a natural experiment to estimate the effects of large-scale masting by comparing temporal changes in animal occurrence in a forest logged for dipterocarp trees thirty years prior (reduced mast availability) and an unlogged forest (greater mast availability). This allowed me to assess how large-scale differences in masting may affect a range of vertebrates, while holding constant other large-scale confounding factors.  Methods Study Sites Data were collected in the Danum Valley Conservation Area (DVCA, N 5.10189° / E 117.688°) and the Sabah Biodiversity Experiment (SBE, N 5.16727° / E 117.564°) in Sabah, Malaysian Borneo from 2013– 2015 (Fig 4-1A). DVCA and SBE are part of the Yayasan Sabah Forest Management Area (YFSMA). The DVCA (438 km2) is unlogged forest and is classified as a Class I ("Totally protected") reserve protected from resource extraction. It is one of the largest expanses of primary, lowland dipterocarp forest remaining in Southeast Asia and has no history of logging or agriculture (Hazebroek et al. 2012). The SBE lies within the Malua Forest Reserve (MFR) (335 km2 total area) and both are located approximately 25 km north of the DVCA. 46  These areas were logged for dipterocarps >60 cm diameter at breast height (dbh) in the late 1980s with the use of tractors and high-lead cables (Berry et al. 2008, Hector et al. 2011). The SBE is found within the area of the MFR that was previously logged. From 1999 to 2007, parts of the MFR were re-logged, although the SBE was not (Hector et al. 2011). During the second logging cycle, a reduced impact logging regime was adopted and trees >40 cm dbh were removed (Anon. 2008, Ang et al. 2017). Dipterocarpaceae trees are the dominant family in the region and during mast events, single-seeded fruits are produced over a period of about four to five months. The seeds are enclosed in a woody (i.e., non-fleshy) endocarp. Dipterocarp trees produce little to no fruit between masting periods and seeds that do not germinate will not survive; seed banks are not known in this taxon (Ashton et al. 1988, Curran and Leighton 2000). Two mast fruiting events took place in the study area, in 2014 and 2015. Within each year, the period of fruiting began in mid- May to early June and started to decline by the end of August or early September. Dipterocarp trees did not produce fruit during the 2013 field season. Prior to the study, the most recent masting event was in 2010. The most widely accepted ecological explanation for dipterocarp mast fruiting is the predator satiation hypothesis. According to this hypothesis, overabundant fruit produced at irregular intervals prevents can increase in seed predator populations, thereby increasing overall seed recruitment (Janzen 1971, Silvertown 1980). Dipterocarp fruits and seeds are consumed by several vertebrate species, but given the irregular frequency of fruiting events, few animals can specialize on this resource (Curran and Leighton 2000). Dipterocarp seeds are wind and gravity dispersed and therefore do not require animals as dispersal agents. Bearded pigs (Sus barbatus) are the major predator of dipterocarp seeds (Kawanishi et al 2008; Curran and Leighton 2000, 47  Curran and Webb 2000). Movement patterns of bearded pigs are thought to closely follow spatial patterns in dipterocarp fruit production (Curran and Leighton 2000). The cue for mating is believed to be the onset of dipterocarp flowering, and bearded pigs may respond to masting with increased reproductive output (Curran and Leighton 2000). Gestation is approximately 90 - 120 days which roughly corresponds to the time between flowering and fruit fall (~ 3 months; Curran and Leighton 2000, Curran and Webb 2000, Kawanishi et al. 2008).   In addition to bearded pigs, several vertebrate species that consume dipterocarp fruit occur in DVCA and SBE. Locally common, herbivorous and granivorous mammals include muntjac deer (Muntiacus atherodes and M. muntjak) and two species each of chevrotains (Tragulus napu and T. kancil) both of which eat fruit and seeds to varying degrees (Hazebroek et al. 2012). Murid rodents also occur and consume dipterocarps, as do larger rodents (thick-spined porcupine, Hystrix crassispinis and Malayan porcupine, H. brachyura). Largely-terrestrial primates include pig-tailed macaques (Macaca nemestrina). Several carnivorous mammals occur in the area, including Sunda clouded leopard (Neofelis diardi), leopard cat (Pionailurus bengalensis), Bornean bay cat (Pardofelis badia), yellow-throated marten (Martes flavigula), collared mongoose (Herpestes semitorquatus), and short-tailed mongoose (H. brachyurus). Frugivorous carnivores included Malay civet (Viverra tangalunga), banded civet (Hemigalus derbyanus), sun bear (Helarctos malayanus), and Sunda stink badger (Mydaus javanensis). Several ground-dwelling birds are found in the study area, including the great Argus (Argusianus argus) and the Bornean crested fireback pheasant (Lophura ignita), both of which consume dipterocarp seeds (Corlett 1998, Curran and Leighton 2000, BirdLife International 2016).     48  Data collection I used passive infrared Reconyx HC500 camera traps in 2013, 2014, and 2015 to estimate site use of different animals at our two field sites. In each year, 22 camera trap stations were established in DVCA and 20 in SBE (Fig 4-1). Cameras remained in place from late May to early September (2013) or late September (2014, 2015). Field seasons within each year captured periods of pre-mast fruiting, peak mast fruiting, and declines in fruit production. Camera traps were placed at the same locations each year.  At the unlogged forest site (DVCA), UTM coordinates were obtained for potential camera trap stations near the Research Centre (Danum Valley Field Centre) on a 1 × 1 km grid created using Garmin Basecamp (Tobler et al. 2008). Camera trap stations were placed within 100 m of each coordinate, taking into account the terrain and ease of access, as camera trap stations would be revisited every few weeks to assess temporal patterns in fruiting. At the logged forest site, nine cameras were placed within the SBE. The remaining 11 were placed in the adjacent forest (MFR), at least 100 m from forestry roads. Camera trap stations in “logged forest” refer to cameras in either the SBE or those around the MFR. Directions from the road were determined by randomized compass bearings and were approximately 1 km apart. Each camera trap was attached to a tree approximately 30 cm above the ground. Cameras were set to high trigger sensitivity and were programmed to take three photographs in rapid-fire succession with a 30 second delay before a subsequent trigger. Photos were date and time stamped, and camera traps were active 24 hours/day. No baits or lures were used.  In 2014 and 2015, repeated visits to camera stations were made to collect fallen fruit, which allowed me to examine temporal patterns in animals’ use of camera trap stations relative to changes in fruit availability. Camera traps were checked on a rotating basis and visits were 49  approximately two to three weeks apart (7 surveys in 2014, 8 in 2015). During each visit, four locations around the camera trap were surveyed (Fig 4-1B); each location consisted of a circular area (2 m radius) that was searched for fallen fruit. Three of these locations were approximately 20 m, in different directions, from the tree that the camera trap was attached to and were also    20 m apart from each other. The fourth location was an area 2 m in radius around the tree with the camera trap. Fruits that were visible on top of leaf litter and underneath the leaf litter were collected. If a large number of the same fruit was present, five to ten specimens were collected and the total number in the area searched was visually estimated. Samples were taken to the lab, identified to family- or genus-level, and were dried and weighed to determine dry fruit biomass at each camera station.   Differences in animal site use between years and forests Animals’ use of camera sites were analyzed as the total number of independent detections in each forest type and year for a given species. Detections of species at the same site were deemed to be independent if they were non-consecutive (i.e., another species was photographed in between) or if they occurred at least one hour apart (Carter et al. 2012, Brodie and Giordano 2013).  To quantify animal responses to masting at large spatial scales (i.e. within logged versus unlogged forest), I used general linear models (GLM) with a Poisson error distribution to determine whether relative site usage for 11 vertebrate groups varied with year (a factor with three levels: 2013, 2014, 2015), forest type (a factor with two levels: logged and unlogged), or a forest × year interaction. I also tested an intercept-only model, which resulted in a total of five models tested for each species (models are listed Appendix Table B-2). 50  Bearded pigs are thought to produce offspring only during mast events, so I sought to assess the relationship between masting and pig reproduction. To do this, I separated bearded pigs into two age classes for all analyses: individuals at least 3 months in age (hereafter “pigs”) and those less than 3 months in age (hereafter “piglets”). Piglets were distinguishable by having stripes on their backs and small body size.  I used AIC-based model selection to determine which model best predicted site usage of animals across years and forest type (Burnham and Anderson 2002). Models were nested, so I used model averaging to assess the relative importance of forest and year coefficients on animal site use and to account for model-selection uncertainty. Unconditional standard errors were used to calculate 95% confidence intervals (CIs), except in cases where the best-fit model had an AICc weight > 0.9, in which case, conditional standard error estimates were used (Burnham and Anderson 2002).   Animal site use at camera stations relative to fruit biomass  In addition to overall differences in masting intensity between forest types I tested whether fine-scale patterns in fruit availability influenced animals’ use of camera trap stations. Here, animal site use was analyzed as the total number of independent detections for a given species in the 2 - 3-week period between visits to camera stations. At a given camera station, animal detections during each period corresponded to fruit collections made at the end of the 2 - 3 weeks. Animal detection data from 2013 were excluded from this analysis because it was a non-mast year so there were no fallen dipterocarp fruits at the camera traps during our field season. I included the following animal taxa in our analysis, as they exhibited varying levels of granivory and frugivory or were among the most commonly detected by camera traps: pigs, piglets, yellow muntjac, 51  murid rodents, chevrotains, great argus pheasant, crested fireback pheasant, thick-spined porcupine, and Malayan porcupine.  To account for the presence of excess zeros in animal detections, I used Zero-Inflated Poisson (ZIP) generalized linear regression (Lambert 1992, Welsh et al. 1996). For these analyses, I estimated the effects of fruit biomass (g/m2), forest type (logged, unlogged), and year (2014, 2015) on animals’ use of camera trap stations. Camera trap station (a factor with 42 levels) was included as a random effect. Fruit biomass data were standardized to have a mean of 0 and standard deviation of 1.  For each species or age class, I tested a total of 18 models to assess whether variation in fruit availability at camera sites affected local animal site use (see Appendix Table B-2) for complete model list). Models with ∆AIC < 2 and were selected as the best-fit models for each species (Burnham and Anderson 2002).  I used these models to test whether fruit biomass from one of four categories (all fruit taxa, dipterocarp, ripe dipterocarp, non-dipterocarp taxa) was significantly associated with animal site use at camera trap stations. I also tested for significant forest × fruit, year × fruit, and fruit × year × forest interactions to determine whether the influence of masting varied with forest type or year.  All statistical analyses were run using R (version 3.3.2) for all analyses (http://r-project.org). Generalized linear models were run using the lme4 package (Bates and Maechler 2011). ZIP regression models were run with the glmmADMB package (Bolker et al. 2012). The MuMIn package was used to perform model selection and model averaging (Barton 2016).   52   Figure 4-1 Map of study sites Map of study area and camera trap lay out in Sabah, Malaysian Borneo. A) Forty-two camera trap stations at the Danum Valley Field Centre (unlogged forest) and Sabah Biodiversity Experiment (logged forest). B) Camera trap and fruit collection location at each station. See Methods for details.  53  Results Fruit biomass during mast years  In all, 50.6 kg (dry weight) of fruit from 66 plant families were collected from camera trap stations, most of which was collected in the unlogged forest (33.9 kg). By weight, Dipterocarpaceae fruits were the most dominant family in our collections, except for in the logged forest in 2015 where the most prominent family was Fagaceae (see Appendix Fig B-1, Table B-1 for complete list of families represented in the fruit collections). Fruit biomass for all fruit categories tested (total fruit biomass, all dipterocarp, ripe dipterocarp, and non-dipterocarp taxa) was highest in 2014 (major mast year). We found fewer fruits at camera trap stations in 2015. However, more fruit, particularly from non-dipterocarp taxa, were collected in the logged forest in that year (Table 4-1, Fig 4-2).  Differences in animal site use between years and forests Though the magnitude and significance of responses varied among species, our findings suggest that several vertebrate taxa were positively associated by large-scale mast availability. Overall site use by pigs, piglets, chevrotains, yellow muntjac, thick-spined porcupines, and murid rodents was greater in the mast years (2014, 2015) than in the non-mast year (2013; Fig 4-3), The model of best fit for bearded pigs, chevrotains, thick-spined porcupine, and murid rodents consisted of a forest × year interaction (w > 0.9 for each species, Table 4-2) whereby occurrence rates were highest in the unlogged forest in 2015. Detections of pigs, chevrotains, and murid rodents were higher in 2015 than in 2014 (pigs: difference between years in model-averaged estimates β = 1.07, 95% CI: 0.90 to 1.24; chevrotains: β = 0.91, 95% CI: 0.61 to 1.12; murid rodents: β = 2.42, 95% CI: 1.78 to 3.07). Also, more pigs were detected in 2015 than in 2013 (β = 2.03, 95% CI: 54  1.69 to 2.37) and site use was significantly higher in 2014 than 2013 (difference between years β = 0.97, 95% CI: 0.60 to 1.38, Fig 4-3). No piglets were detected at camera trap stations in 2013, and piglet site use was significantly greater in 2014 (difference between years: β = 1.93, 95% CI: 0.49 to 3.37).  Within years, animal site use differed between logged and unlogged forest, illustrating that large-scale differences in masting intensity likely affected animal habitat use. Bearded pigs > 3 months old were more prevalent in the unlogged forest within the mast years (forest × year model-averaged β for 2014 = 1.14, 95% CI: 0.64 to 1.65; forest × year model-averaged β for 2015 = 0.62, 95% CI: 0.14 to 1.11). Similarly, chevrotains and murid rodents were more prevalent in the unlogged forest in 2014 (chevrotains: β = 1.14, 95% CI: 0.64 to 1.65; murids: β = 0.78, 95% CI: 0.29 to 1.27). This was also the case for chevrotains in 2015 (β = 0.34, 95% CI: 0.09 to 0.59).  Forest differences in overall site use were detected among porcupine and pheasants; both groups were more prevalent in unlogged forest (thick-spined porcupine: β = 1.97, 95% CI: 0.87 to 3.06; Malayan porcupines: β = 0.95, 95% CI: 0.28 to 1.64; argus: β = 1.68, 95% CI: 1.09 to 2.28, crested fireback pheasants: β = 0.25, 95% CI: 0.01 to 0.48). Also, model-averaged forest coefficient estimates for yellow muntjac revealed that muntjac were more prevalent in unlogged forest (β = 0.72, 95% CI: 0.04 to 1.11), though the AIC weight for the model of best fit was <0.5.  Mast fruiting did not strongly impact all species included in our analyses. For both frugivorous (civets, stink badger, sun bear) and non-frugivorous (felids, yellow-throated marten, mongoose spp.) carnivores, site use was best predicted by a forest × year interaction and both exhibited differences in site use between forests but only in certain years (Table 4-2, Fig 4-3). Frugivorous carnivores were more likely to use the unlogged forest, but only in 2015 (model 55  averaged β = 0.88, 95% CI: 0.14 to 1.61). Model averaged estimates for forest and year were not significantly associated with the prevalence of pig-tailed macaques and non-frugivorous carnivores.   Animal site use at camera stations relative to fruit biomass  Spatio-temporal variability in fallen fruit biomass at camera trap stations affected animal site use at the small-spatial scale, with bearded pigs exhibiting the most pronounced responses to local fruiting patterns of dipterocarps (Fig 4-4). Models of best fit (i.e. those with ∆ AIC < 2) for bearded pigs > 3 months old included a ripe dipterocarp × year × forest interaction (Table 4-3, Appendix Table B-2), suggesting that pigs were more likely to use camera trap sites with relatively higher ripe dipterocarp fruit biomass, particularly in unlogged forest and in 2015 (Fig 4-4). Piglet site use was also best predicted by dipterocarp fruits (ripe and unripe dipterocarp fruits pooled, w = 0.42) and by ripe dipterocarps (w = 0.41; Fig 4-5), though individual parameter estimates for dipterocarp fruits were not significant.   Models of best fit for yellow muntjac and Malayan porcupines both included dipterocarp fruit biomass as covariates, though parameter estimates within these models were not significant for either species (Table 4-3). For yellow muntjac these models included dipterocarp fruit ×year × forest (w= 0.42) and forest × year (w= 0.30) interactions suggesting higher site use in 2014, and higher site use in unlogged forest in particular, in 2015 (Appendix Fig B-3). The best model for Malayan porcupines included a 3-way interaction between ripe dipterocarp fruit, year, and forest (w = 0.42); again, no parameter estimates were significant, indicative of poor overall model fit (Table 4-3, Appendix Fig B-4).  56  The best fit models for argus pheasants included total fruits × forest × year and dipterocarp fruit × forest interactions. Model coefficient estimates suggested argus were more prevalent at camera trap stations with relatively higher dipterocarp fruit biomass and also at camera sites in unlogged forest (Table 4-3). Argus detections were positively associated with total fruit biomass in 2014, and also with dipterocarp fruits in particular in unlogged forest in that year. However, the positive association between argus site use and dipterocarp fruit biomass was generally stronger in logged forest than in unlogged forest (Appendix Fig B-5).  Crested fireback pheasants were less affected by masting than argus pheasants, and site use showed no significant relationship with fruit biomass of any kind, though detections were higher in unlogged forest than in logged forest in 2015 (Appendix Fig B-6). For several taxa, fruit was not a significant predictor of their use of camera sites, and none of the fruit categories we tested appeared in the models of best fit (Table 4-3). For chevrotains, murid rodents, and thick-spined porcupines, the best performing models included only a forest × year interactions. Chevrotains and murid rodents were more likely to use camera traps stations in unlogged forest in 2015 (forest × year w for chevrotains = 0.55; w for murids = 0.72), while thick-spined porcupines were generally more prevalent in 2015 (w  = 0.81).      57  Table 4-1 Model coefficient estimates for fruit biomass Models tested to determine the impact of year (2014, 2015) and forest type (unlogged, logged) on patterns of all plant taxa, all dipterocarp, ripe dipterocarp, and non-dipterocarp fruit biomass (g/m2) collected at camera trap stations in Sabah, Malaysian Borneo. Upper and lower 95% confidence limits (CL) are shown.           lower upper Fruit type   Model coefficient    β CL CL       All taxa  forest 4.63 2.73 6.53   year 0.91 -0.72 2.53   forest × year -5.14 -7.35 -2.93       All dipterocarp  forest 2.97 1.65 4.29   year 0.08 -1.09 1.24   forest × year -2.97 -4.56 -1.38       Ripe dipterocarp   forest 2.83 1.56 4.09   year -0.02 -1.17 1.14   forest × year -2.83 -4.41 -1.26       Non-dipterocarp   forest 1.66 0.43 2.89   year 0.83 -0.21 1.87     forest x year -2.17 -3.59 -0.75   58  Table 4-2 Model selection results for large-scale animal site use List of models used to test for differences in overall animal site use as a function of forest type (logged, unlogged) and year (2013, 2014, 2015) for different vertebrate taxa in Sabah, Malaysian Borneo. Model selection results are based on general linear models (GLM). The model of best fit was determined as having the lowest AICc value and highest AIC weight (w). ∆AIC values relative to the best fit model are also shown (∆AIC).       Species  Model  AICc w ∆AIC           Bearded pigs > 3 months Forest × year 5799.6 0.98 0 Forest + year 5830.2 0.01 30.55  Forest  6677.6 0.00 878.02  Year 5928.3 0.01 878.02  Intercept-only 6767.0 0 967.38      Bearded pigs < 3 months Forest × year 1550.3 0.97 0 Forest + year 1567.9 0.02 17.60 Forest  1759.9 0 209.64  Year 1631.4 0.01 81.13  Intercept-only 1822.0 0 271.71      Yellow muntjac Forest × year 1838.3 0.48 0 Forest + year 1839.1 0.31 0.86  Forest  1839.9 0.21 1.61  Year 1877.3 0 39.06  Intercept-only 1877.6 0 39.31      Chevrotains Forest × year 4161.9 0.99 0  Forest + year 4193.1 0.01 31.18  Forest  4197.9 0 35.93  Year 4246.9 0 85.02  Intercept-only 4251.1 0 89.15      Argus pheasant Forest × year 921.7 0.30 1.60 Forest + year 920.1 0.66 0  Forest  925.5 0.04 5.42  Year 1011.4 0 91.34  Intercept-only 1016.1 0 96.06      Crested Fireback Forest × year 1823.7 0.24 2.03 Forest + year 1821.7 0.67 0  Forest  1825.2 0.00 13.46  Year 1825.9 0.08 4.18  Intercept-only 1839.2 0 17.52      Forest × year 641.1 0.94 0 59       Species  Model  AICc w ∆AIC           Thick-spined porcupine Forest + year 647.6 0.04 6.42  Forest  648.4 0.03 7.21  Year 712.0 0.00 70.85  Intercept-only 712.3 0.00 71.19      Malayan  Forest × year 964.1 0.71 0 Forest + year 965.9 0.29 1.79 Porcupine Forest  980.2 0 16.15  Year 999.0 0 34.94  Intercept-only 1012.5 0 48.48      Pig-tailed macaque Forest × year 1189.8 0.49 0 Forest + year 1189.9 0.42 0.11  Forest  1199.9 0.003 10.17  Year 1192.3 0.13 2.55  Intercept-only 1202.5 0.001 12.78      Murid rodents Forest × year 2086.6 0.99 0 Forest + year 2151.5 0.00 64.98  Forest  2154.8 0.00 68.28  Year 2200.5 0.00 113.95  Intercept-only 2204.8 0.00 118.27      Frugivorous carnivores Forest × year 1122.7 0.32 0 Forest + year 1125.3 0.09 2.64  Forest  1124.4 0.13 1.72  Year 1123.8 0.18 1.14  Intercept-only 112.9 0.29 0.19      Non-frugivorous carnivores Forest × year 558.1 0.68 0 Forest + year 562.5 0.08 4.34 Forest  564.0 0.04 5.90 Year 560.6 0.20 2.45  Intercept-only 565.9 0.01 7.77          60  Table 4-3 Model coefficient estimates for small-scale animal site use Model coefficient estimates from best fit models testing animals’ use of camera trap stations as a function of forest type (unlogged, logged), mast year (major mast, 2014; minor mast, 2015), and fruit biomass [all fruit taxa (all), all dipterocarp (dipt), ripe dipterocarp (ripe), and non-dipterocarp (nd) fruit biomass (g/m2)]. Models of best fit were determined as those with ∆AIC< 2. Only significant model coefficients are shown. Significance is based on 95% confidence intervals (CI) not overlapping with zero. ∆AIC values and AIC weights for all models tested are given in Appendix Table B-2.  Species Model(s) < 2 ∆AICc Coefficient β estimate Lower CI Upper CI        Bearded pigs ripe × forest × year   ripe × forest × year   1.19 0.08 2.3 > 3 months  forest × year   0.45 0.12 0.7    year 0.84 0.57 1.07  ripe × year   year 0.04 0.06 0.98        Bearded pigs  ripe  ripe ns ns ns < 3 months  dipt  dipt ns ns ns        Yellow  dipt × forest × year   dipt × forest × year   -2.69 -8.47 3.09 muntjac forest × year   forest × year 0.65 0.12 1.19    year -0.56 -1.02 -0.1        Chevrotains forest × year   forest × year 0.35 0.07 0.64  dipt × forest × year   year -0.32 -0.59 -0.05    forest × year 0.4 0.08 0.73        Great Argus all × forest × year   forest 2.06 0.085 4.04 pheasant dipt × forest dipt × forest -1.08 -1.83 -0.32    dipt 0.89 0.24 1.54    forest 1.97 1.08 2.86  all × forest   all × forest -1.27 -2.12 -0.33    forest 1.98 1.06 2.91    all 1.29 0.38 2.2 61   Species Model(s) < 2 ∆AICc Coefficient β estimate Lower CI Upper CI         Crested forest × year forest × year -0.72 -1.36 -0.09 fireback   year 0.49 0.09 0.97 pheasant nd  nd ns ns ns  all  all ns ns ns  dipt  ns ns ns ns  ripe  ns ns ns ns        Malayan  ripe × forest × year   ripe × forest × year   ns ns ns porcupine              Thick-spined  forest × year   year 2.13 -0.05 4.31 porcupine              Murid spp. forest × year   forest × year   1.23 0.5 1.96    forest 1.29 0.09 2.48    year 1.27 -2.14 -0.86  62   Figure 4-2 Mean fruit biomass at camera stations Mean (+SE) fruit biomass (g/m2) collected from camera trap stations in A) 2014 and B) 2015 in unlogged and logged forest. Note the difference in y-axis scaling between years. Mean biomass is shown for fruits from all plant taxa (all fruits), all taxa excluding Dipterocarpaceae (non-dipterocarp), only fruit from the family Dipterocarpaceae (Dipterocarp), and ripe dipterocarp fruits only (Ripe dipterocarp).   63     Figure 4-3 Animal detections in non-mast year and mast years Mean number of independent detections per day for eleven vertebrate species at camera trap stations in unlogged and logged forest in a non-mast year for dipterocarp trees (2013), a major dipterocarp mast year (2014) and a minor dipterocarp mast year (2015). Two age classes are shown for bearded pigs: >3 months old (adults and sub-adults) and < 3 months old (piglets). Animal detections at camera stations were deemed to be independent if they were non-consecutive (i.e., another species was photographed in between) or if they occurred at least one hour apart. 64      Figure 4-4 Bearded pig site use as a function of ripe dipterocarp fruit biomass Partial residual plots showing site use of bearded pigs > 3 months old (adults and sub-adults) as a function of ripe dipterocarp fruit biomass (standardized dry weight, [g/m2]) in unlogged and logged forest in A) major mast (2014) and b) minor mast (2015). Site use is modeled as the number of independent detections of animals at camera trap stations. Detections were deemed to be independent if they were non-consecutive (i.e., another species was photographed in between) or if they occurred at least one hour apart. Fruit biomass data are from fallen fruit collected during repeated visits to camera trap sites within each year. Confidence bands shown around the predicted fit line from the models of best fit selected by AICc comparison. Note that y-axes vary between years.    65    Figure 4-5 Piglet site use as a function of overall dipterocarp fruit biomass Partial residual plots showing the relationship between site use of piglets (bearded pigs < 3 months old) as a function of overall A) dipterocarp fruit (ripe and unripe pooled) and B) ripe dipterocarp fruit biomass [(standardized dry weight, [g/m2]). Site use is modeled as the number of independent detections of animals at camera trap stations. Detections were deemed to be independent if they were non-consecutive (i.e., another species was photographed in between) or if they occurred at least one hour apart. Fruit biomass data are from fallen fruit collected during repeated visits to camera trap sites in 2014 and 2015. No differences in site use between logged and unlogged forest were found nor did site use differ between mast years. Confidence bands shown around the predicted fit line from the models of best fit selected by AICc comparison.      66  Discussion Food resources are patchily distributed across landscapes. Asynchronous fruiting at fine scales between similar or surrounding plant species reduces the overlap of fruiting phenologies (Chivers 1980, Leighton and Leighton 1983, Ashton et al. 1988, van Schaik et al. 1993). Overall fruit abundance in our study varied at the landscape scale (between forests) and temporally (across mast years), and camera stations also showed spatio-temporal variation in fallen fruit biomass. Animals must cope with these changing patterns in availability in order to locate food. Our findings are indicative of community-wide responses to mast fruiting at the large spatial scale, though the strength of animal responses to dipterocarp fruit abundance, and therefore the relative importance of masting, varied between species. Bearded pigs showed the most pronounced and consistent responses to mast fruiting. Consistent with findings in Indonesia, bearded pigs appear to be the primary consumers of dipterocarp fruit in our study area (Curran and Leighton 2000, Kawanishi et al. 2008). Animal responses to food availability may additionally vary with spatial scale (Wiens 1989, Kotliar and Wiens 1990, Levin 1992, Garcia et al. 2011) though the consequences of resource pulse events for animals at multiple spatial scales are not well understood. Patchiness in food distribution that varies with scale requires animals to adjust foraging strategies to detect areas of high food productivity (Wiens 1989, Kotliar and Wiens 1990). At large spatial scales, animals may search across the landscape for areas of increased food abundance (Senft et al. 1987, Kotliar and Wiens 1990, Bucher 1992). Such large-scale movements are more likely among large-bodied, mobile taxa, such as bearded pigs (Curran and Leighton 2000, Ostfeld and Keesing 2000, Yang et al. 2010).  67  Differences in site use between years and forest types may also point to increased reproductive output as a response to overall fruit abundance (Yang et al. 2010). Relative to the non-mast year, ungulates and murid rodents showed higher site use in the mast years, especially in the year following the major mast, suggesting that several taxa responded to masting with increased reproduction. By assessing site use by bearded pigs of different age classes (piglets versus adults), I was also able to estimate the effects of masting on local pig abundance. Increased piglet site use in the major mast year likely contributed to the greater prevalence of adult pigs the following year, consistent with previous findings (Curran and Leighton 2000). Pig-tailed macaques and non-frugivorous carnivores, in contrast were less affected at the large scale by masting over our study period. This is indicative that increases in more frugivorous taxa were likely driven by enhanced food availability rather than some other factor (e.g. climate) that might have affected all feeding guilds similarly. However, vertebrates that exhibited increased site use at the forest level did not necessarily show similar increases at the small spatial scale, suggesting that animal responses to masting can be scale-dependent (Johnson et al. 2001).  At fine spatial scales (e.g. within particular communities), animals may track fruiting patterns of particular tree taxa (Senft et al. 1987, Kotliar and Wiens 1990, Bucher 1992), but the forces behind community-wide resource tracking in the tropics are not clear. Foraging decisions made at fine spatial scales can influence animal distribution within habitat patches; which could subsequently affect species interactions and population growth (Levin 1992, Mayor et al. 2009). Large herbivores may track forage availability at fine-spatial scales (Senft et al. 1987, Wilmshurst et al. 1999), but most examples of fine-scale resource tracking in the literature concern avian taxa (Tellería and Pérez-Tris 2003, García and Ortiz-Pulido 2004, Saracco et al. 2004) and resource tracking is generally not well-studied among mammals. Bearded pigs are 68  social, mobile and nomadic, particularly during mast years, which could explain their ability to locate dipterocarp fruit patches at small spatial scales (Curran and Leighton 2000). Yellow muntjac can be important predators of dipterocarp seedlings (Curran and Webb 2000) and dipterocarp fruits might also be an important food source during mast years. In contrast, less mobile taxa such as porcupines and crested fireback were not significantly associated with dipterocarp fruit at the small scale. Similarly, chevrotain responses to masting were limited compared to those of bearded pigs. Chevrotains are mostly frugivorous with small home ranges (around 7 - 9 ha; Ahmad 1994, Heydon 1994) and may only search for dipterocarp fruits within those areas, if at all (Spiegel and Nathan 2007). The effects of masting thus affected multiple trophic levels to varying degrees, though food abundance in combination with altered food distribution could have a greater influence on whether animals track fruiting patterns. I also assessed the impacts of masting on vertebrates using a natural experiment: habitat disturbance via selective logging. At the large scale, logging directly reduces dipterocarp fruit abundance, and only bearded pigs showed increased site use in the logged forest compared to the non-mast year. At the small spatial scale, canopy gaps created by logging alter forest structure by facilitating liana growth, which may impede movement (Broadbent et al. 2006, 2008). Only argus pheasants were positively associated with dipterocarp fruit in both logged and unlogged forest. For all other taxa in our study, animal responses to masting were generally limited at the small spatial scale in logged forest. In some cases, fewer individuals were detected in logged forest even when fruit biomass was comparable to the unlogged forest. Bearded pig site use, for example, increased as a function of overall fruit biomass in both forest types, yet site use was largely unchanged as a function of dipterocarp fruit abundance in logged forest. Animals in logged forest may track non-dipterocarp food resources that we did not measure or their 69  movements could be influenced by other factors such as interactions with predators (Frair et al. 2005). On the other hand, altered forest structure could indirectly affect animal foraging behavior, which could have affected their ability to arrive at areas of high dipterocarp fruit biomass. The chances of encountering dipterocarp food patches may be lower in logged forest due to reduced seed production and increased patchiness of dipterocarp fruit crops (Curran et al. 1999, Curran and Webb 2000). Upon encountering dipterocarp-rich food patches, pigs might spend more time foraging in one place than they would in an equivalent patch in unlogged forest (Curran and Leighton 2000), consistent with the marginal value theorem (Charnov 1976). This could, in turn, affect plant recruitment as a lack of continuous foraging within logged forest could lead to increased localized seed mortality (Curran and Webb 2000), suggesting that logging reduces the chance of seed predators becoming satiated at fine scales. The effects of logging on animal responses to masting may therefore, have important implications for plant recruitment. The relevance of spatial scale for animal responses to resource pulses is not well understood, particularly when combined with habitat disturbance (but see Lehouck et al. 2009). Strong responses to masting at the forest-scale among taxa from multiple trophic levels suggest that consequences of mast fruiting can ripple through intact forest communities (Curran and Leighton 2000, Clotfelter et al. 2007) and that strong, community-wide responses to masting are not limited to systems lacking their key mast consumers (Ostfeld et al. 1996). By using logged forest as a natural experiment, we were able assess how large-scale differences in masting affected a range of vertebrates at large and small spatial scales. At the large scale, animal detections in logged forest showed little variation between years, despite pronounced responses in unlogged forest. At the small scale, site use of bearded pigs, the key consumer in our study 70  area, was less influenced by masting in logged forest than in unlogged forest. Even though animals in faunally intact communities showed strong responses to mast fruiting, reduced masting intensity associated with logging could interfere with animal-dipterocarp interactions, potentially affecting animal abundances and fine-scale resource tracking. Given the widespread threat of habitat disturbance to forests in Southeast Asia, continued selective logging could have important repercussions for animal populations and dipterocarp regeneration.   71  Chapter 5: Conclusion In this dissertation, I examined the consequences of habitat disturbance for both groups of interacting species in plant-animal interactions and the conditions under which such interactions outcomes may be altered. In logged and unlogged forest in Malaysian Borneo, I quantified the effects of small- to large-bodied vertebrate granivore-herbivores on plant regeneration (Chapter 2) as well as taxonomic seedling diversity and morphological trait distribution (Chapter 3). I also assessed whether animal responses to spatio-temporal variability in fruit abundance varied with spatial scale, and whether logging disrupted resource-tracking strategies (Chapter 4).  The key findings of this thesis are:  1. Experimental defaunation and selective logging can synergistically interact to increase seedling establishment of dipterocarp trees and may also affect seedling survival.  2. Mammal herbivores may alter the distribution of tree seedling traits (fruit size) in logged forest without altering tree taxonomic diversity.  3. Despite strong responses to mast fruiting in nearby unlogged forest, masting intensity in logged forest was reduced and vertebrate site in logged forest was not affected by spatio-temporal variation in dipterocarp fruit biomass.  Together, my results suggest that while habitat disturbance in the form of selective logging can alter the outcome of plant-animal interactions, the consequences for plants and animals are variable. Logged forests in Borneo may show reduced dipterocarp seed survival, while the loss of large vertebrates may have no effect on seed fate. However, defaunation in logged forest can actually increase seedling establishment, contrary to most findings in the literature, where defaunation had the opposite effect on recruitment (Asquith et al. 1997, Galetti et al. 2015, Nunez-Iturri et al. 2008, Rosin and Poulsen 2016). On the other hand, the presence of vertebrate 72  herbivores in logged forest could alter the distribution of plant traits in seedling communities, with animals disproportionately killing seedlings from large-fruited genera. Logging may also affect animal responses to resource pulse events. Dipterocarp masting intensity and vertebrate responses to masting were generally weaker in logged forest. Altered forest structure may disrupt resource tracking ability at fine spatial scales, which might also affect large-scale responses to masting.   Limitations of research The intent of this thesis was to investigate the effects of selective logging on plant recruitment and animal habitat use; this was done via data collection in a logged forest (the Sabah Biodiversity Experiment and adjacent forest, Malua Forest Reserve) and an unlogged study site (Danum Valley Field Centre). Having only one of each forest type limits me from making general conclusions about the effects of logging, but my findings from this study region provide valuable insight into how selective logging, which is rampant throughout Southeast Asia, might affect forest communities.   A limitation of my experimental design is the low number of treatment replicates within exclosure blocks in each forest site (Chapters 2 and 3), such that the exclosure design may have been pseudoreplicated. Pseudoreplication occurs when the number of replicates is small or when replicates may not be independent, and spatial variation cannot be disentangled from variation related to treatment effects (Hurlbert 1984, Ramage et al. 2013). Pseudoreplication is a common issue in ecological field experiments and in some cases, logistical and feasibility issues makes pseudoreplication unavoidable. Ecologists must, therefore, acknowledge the potential lack of independence between replicates, which may be dealt with by reporting inferential statistics and 73  with the inclusion of nesting or random effects (Kozlov and Hurlbert 2006, Davies and Gray 2015), as I report in Chapters 2 and 3. Finally, my study of the effects of logging and experimental defaunation on seedling traits and diversity included four years of data and my conclusions of whether these disturbances affect seedling communities are based on a relatively short time period. Animal impacts on plant community composition may not be detected until years or even decades after defaunation. Kurten (2010) similarly used experimental exclusion plots to estimate defaunation impacts on seedling traits. Consistent with my findings in Chapter 3, experimental defaunation had no significant effect on leaf traits in seedling communities (Kurten 2010). However, Kurten (2010) additionally examined animal impacts on leaf traits at the sapling stage, where vertebrate defaunation was associated with reduced mean leaf toughness in sapling communities. It is possible that defaunation can alter the distribution of leaf traits and wood density in my study area, but that these effects are not detected at the seedling stage.   Anthropogenic disturbance can have cascading impacts on plant and animal communities by altering the nature of species interactions, but most studies assess the consequences on only plants or only animals, potentially underestimating how disturbances affect tropical systems. Further, many studies are carried out soon after areas are logged and the long-term consequences for plant and animal communities in regenerating forest are not well understood (Edwards et al. 2014). Using multi-year replicated field experiments in addition to natural experiments in Malaysian Borneo, I assessed the effects of logging that occurred 30 years prior on both plants and animals. I found that even decades after the initial logging cycle, animal-induced mortality to seeds and seedlings may still differ between logged and unlogged forest, with repercussions for plant recruitment and seedling community composition in the former. Logging can also affect 74  animal responses to spatio-temporal patterns in resource availability, which may in turn, affect population growth and resource tracking ability.  Tropical forests are often faced with multiple threats and my work emphasizes the need for long-term studies that disentangle the effects of logging and other co-occuring disturbances because in isolation, the effects of experimental defaunation and hunting on plant recruitment differed from their combined effects. I further recommend that when assessing the impacts of human activity on plant community composition, there be an increased focus on morphological plant traits, due to the relevance of morphological traits for ecosystem processes (Díaz et al. 2007). Finally, logging can alter the patchiness of resource distribution along with resource abundance at multiple spatial scales, subsequently affecting animal habitat use and whether they track resource at large and small spatial scales; I therefore, recommend that research into the effects of altered habitat structure on animal habitat use be performed at more than one spatial scale to better understand the role of resource heterogeneity on animal responses to resource pulses events. Our understanding of the importance of plant-animal interactions in structuring communities is limited, especially in diverse systems affected by human activity. Selective logging in tropical forests is widespread and is known to alter forest structure (Johns 1988). My findings demonstrate that logging-induced changes in forest structure can affect species interactions at multiple plant life stages, spatial scales, and in combination with co-occurring disturbances. This adds to our understanding of the community-wide effects of habitat disturbance in the tropics, as well as in systems elsewhere facing similar levels of disturbance. A greater focus towards multi-species studies to quantify how human disturbances alter the outcomes of plant-animal interactions is essential. This is particularly important in highly 75  biodiverse systems where habitat loss or degradation, and defauntion put countless species at risk.  76  Bibliography   Achard, F., H. D. Eva, H.-J. Stibig, P. Mayaux, J. Gallego, T. Richards, and J.-P. Malingreau. 2002. Determination of deforestation rates of the world’s humid tropical forests. Science 297:999–1002. Ahmad, A. H. 1994. The ecology of mousedeer (Tragulus sp.) in a Bornean rain forest, Sabah Malaysia. University of Aberdeen. Ang, C. C., M. J. O’Brien, K. K. S. Ng, P. C. Lee, A. Hector, B. Schmid, and K. K. Shimizu. 2017. Genetic diversity of two tropical tree species of the Dipterocarpaceae following logging and restoration in Borneo: high genetic diversity in plots with high species diversity. Plant Ecology and Diversity:1755–1768. Appanah, S. 1985. General flowering in the climax rain forests of South-east Asia. Journal of Tropical Ecology 1:225. Ashton, P. S., T. J. Givnish, and S. Appanah. 1988. Staggered flowering in the Dipterocarpaceae: new insights into floral induction and evolution of mast fruiting in the aseasonal tropics. The American Naturalist 132:44–66. Asner, G. P., T. K. Rudel, T. M. Aide, R. Defries, and R. Emerson. 2009. A contemporary assessment of change in humid tropical forests. Conservation Biology 23:1386–1395. Asquith, N. M. M., S. J. Wright, M. J. Clauss., and M. J. Clauss. 1997. Does mammal community composition control recruitment in neotropical forest? Evidence from Panama. Ecology 78:941–946. Augustine, D. J., and S. J. McNaughton. 1998. Ungulate effects on the functional species composition of plant communities: herbivore selectivity and plant tolerance. The Journal of 77  Wildlife Management 62:1165–1183. Barton, K. 2016. MuMIn: Multi-Model Inference. R package. Bates, D. M., and M. Maechler. 2011. lme4: Linear mixed-effects models using S4 classes. R package version 1.1-7. Bates, D., M. Maechler, B. M. Bolker, and S. Walker. 2015. Fitting linear mixed-effects models using lme4. Journal of Statistical Software 67:1–48. Beck, H. 2006. A review of peccary–palm interactions and their ecological ramifications across the Neotropics. Journal of Mammalogy 87:519–530. Beckman, N. G., H. C. Muller-Landau, N. O. G. B. Eckman, and H. E. C. M. U. Andau. 2011. Linking fruit traits to variation in predispersal vertebrate seed predation, insect seed predation, and pathogen attack. Ecology 92:2131–2140. Bello, C., M. Galetti, M. A. Pizo, L. F. S. Magnago, M. F. Rocha, R. A. F. Lima, C. A. Peres, O. Ovaskainen, and P. Jordano. 2015. Defaunation affects carbon storage in tropical forests. Science Advances 1:1–11. Berry, N. J., O. L. Phillips, S. L. Lewis, J. K. Hill, D. P. Edwards, N. B. Tawatao, N. Ahmad, D. Magintan, C. V. Khen, M. Maryati, R. C. Ong, and K. C. Hamer. 2010. The high value of logged tropical forests: Lessons from northern Borneo. Biodiversity and Conservation 19:985–997. Berry, N. J., O. L. Phillips, R. C. Ong, and K. C. Hamer. 2008. Impacts of selective logging on tree diversity across a rainforest landscape: The importance of spatial scale. Landscape Ecology 23:915–929. BirdLife International. 2016. Argusianus argus. The IUCN Red List of Threatened Species 2016: e.T22725006A94883506. 78  Bolker, B., H. Skaug, A. Magnusson, and A. Nielsen. 2012. Getting started with the glmmADMB package.  Bricker, M., D. Pearson, and J. Maron. 2010. Small-mammal seed predation limits the recruitment and abundance of two perennial grassland forbs. Ecology 91:85–92. Broadbent, E. N., G. P. Asner, M. Keller, D. E. Knapp, P. J. C. Oliveira, and J. N. Silva. 2008. Forest fragmentation and edge effects from deforestation and selective logging in the Brazilian Amazon. Biological Conservation 141:1745–1757. Broadbent, E. N., D. J. Zarin, G. P. Asner, M. Penña-Claros, A. Cooper, and R. Littell. 2006. Recovery of forest structure and spectral properties after selective logging in lowland Bolivia. Ecological Applications 16:1148–1163. Brocardo, C. R., V. B. Zipparro, R. A. F. de Lima, R. Guevara, and M. Galetti. 2013. No changes in seedling recruitment when terrestrial mammals are excluded in a partially defaunated Atlantic rainforest. Biological Conservation 163:107–114. Brodie, J. F., and A. Giordano. 2013. Lack of trophic release with large mammal predators and prey in Borneo. Biological Conservation 163:58–67. Bronstein, J. L. 1994. Conditional outcomes in mutualistic interactions. Trends in Ecology & Evolution 9:214–217. Bucher, E. H. 1992. The causes of extinction of the Passenger Pigeon. Current Ornithology. Springer. Bunker, D. E., F. Declerck, J. C. Bradford, R. K. Colwell, I. Perfecto, O. L. Phillips, M. Sankaran, and S. Naeem. 2005. Species loss and aboveground carbon storage in a tropical forest. Science 310:1029–2031. Burnham, K. P., and D. R. Anderson. 2002. Model selection and multimodel inference. Springer, 79  New York. Camargo-Sanabria, A. A., E. Mendoza, R. Guevara, M. Martinez-Ramos, and R. Dirzo. 2015. Experimental defaunation of terrestrial mammalian herbivores alters tropical rainforest understorey diversity. Proceedings of the Royal Society B: Biological Sciences 282:20142580–20142580. Cannon, C. H., L. M. Curran, A. J. Marshall, and M. Leighton. 2007. Long-term reproductive behaviour of woody plants across seven Bornean forest types in the Gunung Palung National Park (Indonesia): suprannual synchrony, temporal productivity and fruiting diversity. Ecology Letters 10:956–969. Cannon, C., K. Kartawinata, M. Leighton, and R. Peart David. 1994. The structure of lowland rainforest after selective logging in West Kalimantan, Indonesia. Forest Ecology and Management 67:49–68. Cardillo, M., G. M. Mace, K. E. Jones, J. Bielby, O. R. P. Bininda-Emonds, W. Sechrest, C. D. L. Orme, and A. Purvis. 2005. Multiple causes of high extinction risk in large mammal species. Science 309:1239–1241. Carter, N. H., B. K. Shrestha, J. B. Karki, N. M. B. Pradhan, and J. Liu. 2012. Coexistence between wildlife and humans at fine spatial scales. Proceedings of the National Academy of Sciences 109:15360–15365. Chamberlain, S. A., J. L. Bronstein, and J. A. Rudgers. 2014. How context dependent are species interactions? Ecology Letters 17:881–890. Chao, A., T. Hseih, E. Sander, K. Ma, R. Colwell, and A. Ellison. 2014. Rarefaction and extrapolation with Hill numbers: a framework for sampling and estimation in species diversity studies. Ecological Monographs 84:45–67. 80  Charnov, E. L. 1976. Optimal foraging, the marginal value theorem. Theoretical population biology 9:129–136. Chivers, D. J. 1980. Malayan forest primates. Ten years’ study in tropical rain forest. Plenum Press, New York, NY. Clotfelter, E. D., A. B. Pedersen, J. A. Cranford, N. Ram, E. A. Snajdr, V. Nolan, and E. D. Ketterson. 2007. Acorn mast drives long-term dynamics of rodent and songbird populations. Oecologia 154:493–503. Cohen, J. S., S. K. D. Rainford, and B. Blossey. 2014. Community-weighted mean functional effect traits determine larval amphibian responses to litter mixtures. Oecologia 174:1359–1366. Coley, P. D., J. P. Bryant, and F. S. Chapin. 1985. Resource availability and plant antiherbivore defense. Science 230:895–899. Corlett, R. T. 1998. Frugivory and seed dispersal by vertebrates in the Oriental (Indomalayan) Region. Biological reviews of the Cambridge Philosophical Society 73:413–448. Corlett, R. T., and R. B. Primack. 2011. Tropical rain forests: An ecological and biogeographical comparison. 2nd edition. Wiley-Blackwell, West Sussex, UK. Curran, L., and M. Leighton. 2000. Vertebrate responses to spatiotemporal variation in seed-production of mast fruiting Dipterocarpaceae. Ecological Monographs 70:101–128. Curran, L. M., I. Caniago, G. D. G. D. Paoli, D. Astianti, M. Kusneti, M. Leighton, C. E. Nirarita, and H. Haeruman. 1999. Impact of El Niño and logging on canopy tree recruitment in Borneo. Science 286:2184–2188. Curran, L. M., and C. O. Webb. 2000. Experimental tests of the spatiotemporal scale of seed predation in mast-fruiting Dipterocarpaceae. Ecological Monographs 70:129–148. 81  Davies, G. M., and A. Gray. 2015. Don’t let spurious accusations of pseudoreplication limit our ability to learn from natural experiments (and other messy kinds of ecological monitoring). Ecology and Evolution 5:5295–5304. Dirzo, R., H. S. Young, M. Galetti, G. Ceballos, N. J. B. Isaac, and B. Collen. 2014. Defaunation in the Anthropocene. Science 345:401–406. Dixon, P. 2016. VEGAN, a package of R functions for community ecology. Journal of Vegetation Science 14:927–930. Edwards, D. P., T. H. Larsen, T. D. S. Docherty, F. A. Ansell, W. W. Hsu, M. A. Derhé, K. C. Hamer, and D. S. Wilcove. 2010. Degraded lands worth protecting: the biological importance of Southeast Asia’s repeatedly logged forests. Proceedings of the Royal Society B: Biological Sciences 278:82 LP-90. Ellsworth, J. W., and B. C. McComb. 2003. Potential effects of passenger pigeon flocks on the structure and composition of presettlement forests of eastern North America. Conservation Biology 17:1548–1558. Fadini, R. F., M. Fleury, C. I. Donatti, and M. Galetti. 2009. Effects of frugivore impoverishment and seed predators on the recruitment of a keystone palm. Acta Oecologica 35:188–196. Fragoso, J. M. V, K. M. Silvius, and J. A. Correa. 2003. Long-distance seed dispersal by tapirs increases seed survival and aggregates tropical trees. Ecology 84:1998–2006. Frair, J. L., E. H. Merrill, D. R. Visscher, D. Fortin, H. L. Beyer, and J. M. Morales. 2005. Scales of movement by elk (Cervus elaphus) in response to heterogeneity in forage resources and predation risk. Landscape ecology 20:273–287. Galetti, M., and R. Dirzo. 2013. Ecological and evolutionary consequences of living in a defaunated world. Biological Conservation 163:1–6. 82  Galetti, M., R. Guevara, M. C. Côrtes, R. Fadini, S. Von Matter, Leite, A. B., F. Labecca, T. Ribeiro, C. S. Carvalho, R. G. Collevatti, P. R. G. J. Mathias M. Pires, P. H. Brancalion, M. C. Ribeiro, and P. Jordano. 2013. Functional extinction of birds drives rapid evolutionary changes in seed size. Science 340:1086–1090. García, D., and R. Ortiz-Pulido. 2004. Patterns of resource tracking by avian frugivores at multiple spatial scales: two case studies on discordance among scales. Ecography 27:187–196. Garcia, D., R. Zamora, and G. C. Amico. 2011. The spatial scale of plant – animal interactions : effects of resource availability and habitat structure. Ecological Monographs 81:103–121. Garnier, E., J. Cortez, G. Billès, M. L. Navas, C. Roumet, M. Debussche, G. Laurent, A. Blanchard, D. Aubry, A. Bellmann, and C. Neill. 2004. Plant functional markers capture ecosystem properties during secondary succession. Ecology 85:2630–2637. Goheen, J. R., F. Keesing, B. F. Allan, D. Ogada, S. Richard, S. Ecology, and N. Jun. 2004. Net effects of large mammals on Acacia seedling survival in an African Savanna. Ecology 85:1555–1561. Goheen, J. R., T. M. Palmer, F. Keesing, C. Riginos, and T. P. Young. 2010. Large herbivores facilitate savanna tree establishment via diverse and indirect pathways. Journal of Animal Ecology 79:372–382. Grime, J. P., J. H. C. Cornelissen, K. Thompson, and J. G. Hodgson. 1996. Evidence of a causal connection between anti-herbivore defence and the decomposition rate of leaves. Oikos:489–494. Grindal, S. D., and R. M. Brigham. 1999. Impacts of forest harvesting on habitat use by foraging insectivorous bats at different spatial scales. Ecoscience 6:25–34. 83  Guariguata, M. R., H. A.-L. Claire, and G. Jones. 2002. Tree seed fate in a logged and fragmented forest landscape, northeastern Costa Rica. Biotropica 34:405–415. Harrison, R. D. 2011. Emptying the forest: hunting and the extirpation of wildlife from tropical nature reserves. BioScience 61:920–924. Harrison, R. D., S. Tan, J. B. Plotkin, F. Slik, M. Detto, T. Brenes, A. Itoh, and S. J. Davies. 2013. Consequences of defaunation for a tropical tree community. Ecology Letters 16:687–694. Hart, D. D. 1992. Community organization in streams: the importance of species interactions, physical factors, and chance. Oecologia 91:220–228. Hautier, Y., P. Saner, C. Philipson, R. Bagchi, R. C. Ong, and A. Hector. 2010. Effects of seed predators of different body size on seed mortality in Bornean logged forest. PLoS ONE 5:e11651. Hazebroek, H., T. Adlin, and W. Sinun. 2012. Danum Valley the rainforest. Natural History Publications (Borneo), Kota Kinabalu. Hector, A., C. Philipson, P. Saner, J. Chamagne, D. Dzulkifli, M. J. O’Brien, J. L. Snaddon, P. Ulok, M. Weilenmann, G. Reynolds, and H. C. J. Godfray. 2011. The Sabah Biodiversity Experiment: a long-term test of the role of tree diversity in restoring tropical forest structure and functioning. Philosophical Transactions of the Royal Society B: Biological Sciences 366:3303–3315. Heydon, M. J. 1994. The ecology and management of rainforest ungulates in Sabah , Malaysia : implications of forest disturbance. Aberdeen, Scotland. Hsieh, T. C., K. H. Ma, and A. Chao. 2016. iNEXT: An R package for rarefaction and extrapolation of species diversity (Hill numbers). Methods in Ecology and Evolution 84  7:1451–1456. Hurlbert, S. H. 1984. Pseudoreplication and the design of ecological field experiments. Ecological monographs 54:187–211. Janzen, D. H. 1971. Seed predation by animals. Annual Review of Ecology and Systematics 2:465–492. Johns, A. D. 1988. Effects of “selective” timber extraction on rain forest structure and composition and some consequences for frugivores and folivores. Biotropica 20:31–37. Johns, A. G. 1997. Timber production and biodiversity conservation in tropical rain forests. Cambridge University Press., New York. Johnson, C. J., K. L. Parker, and D. C. Heard. 2001. Foraging across a variable landscape: behavioral decisions made by woodland caribou at multiple spatial scales. Oecologia 127:590–602. Jones, K. E., J. Bielby, M. Cardillo, S. A. Fritz, J. O’Dell, C. D. L., K. S. Orme, W. Sechrest, E. H. Boakes, C. Carbone, C. Connolly, M. J. Cutts, J. K. Foster, R. Grenyer, M. Habib, C. A. Plaster, A. Purvis, E. A. R. Samantha A. Price, J. Rist, A. Teacher, O. R. P. Bininda-Emonds, J. L. Gittleman, G. M. Mace, G. M. Mace, and A. Purvis. 2009. PanTHERIA: a species-level database of life history, ecology, and geography of extant and recently extinct mammals. Ecology 90:2648. Kawanishi, K., M. Gumal, and W. Oliver. 2008. Sus barbatus. The IUCN Red List of Threatened Species 2008: e.T41772A10559190. Kelly, D. 1994. The evolutionary ecology of mast seeding. Trends in Ecology and Evolution 9:465–470. Kelly, D., W. D. Koenig, and A. M. Liebhold. 2008. An intercontinental comparison of the 85  dynamic behavior of mast seeding communities. Population Ecology 50:329–342. Kelly, D., and V. L. Sork. 2002. Mast seeding in perrenial plants: why, how, where? Annual Review of Ecology and Systematics 33:427–447. Khan, J. A. 1994. Food habits of ungulates in dry tropical forests of Gir Lion Sanctuary, Gujarat, India. Acta theriologica 39. Klanderud, K. 2005. Climate change effects on species interactions in an alpine plant community. Journal of Ecology 93:127–137. Koenig, W. D., and J. M. H. Knops. 2005. The mystery of masting in trees. American Scientist 93:340–347. Kotliar, N. B., and J. A. Wiens. 1990. Multiple scales of patchiness and patch structure : a hierarchical framework for the study of heterogeneity. Oikos 59:253–260. Kozlov, M. V, and S. H. Hurlbert. 2006. Pseudoreplication, chatter, and the international nature of science: A response to DV Tatarnikov. Journal of Fundamental Biology (Moscow) 67:145–152. Kurten, E. L. 2010. Functional trait mediation of plant-animal interactions: effects of defaunation on plant functional diversity in a Neotropical forest. Stanford University. Kurten, E. L. 2013. Cascading effects of contemporaneous defaunation on tropical forest communities. Biological Conservation 163:22–32. Kurten, E. L., S. J. Wright, W. P. Carson, and T. M. Palmer. 2015. Hunting alters seedling functional trait composition in a Neotropical forest. Ecology 96:1923–1932. Lambert, D. 1992. Zero-Inflated Poisson regression, with an application to defects in manufacturing. Technometrics 34:1–14. Laughlin, D. C. 2011. Nitrification is linked to dominant leaf traits rather than functional 86  diversity. Journal of Ecology 99:1091–1099. Laurance, W., and S. Laurance. 1996. Responses of five arboreal marsupials to recent selective logging in tropical Australia. Biotropica 28:310–322. Lehouck, V., T. Spanhove, C. Vangestel, N. J. Cordeiro, and L. Lens. 2009. Does landscape structure affect resource tracking by avian frugivores in a fragmented Afrotropical forest? Ecography 32:789–799. Leighton, M., and D. R. Leighton. 1983. Vertebrate responses to fruiting seasonality within a Bornean rain forest. Special publications series of the British Ecological Society:181–196. Levey, J. D. 1988. Spatial and temporal variation in Costa Rican fruit and fruit-eating bird abundance. Ecological Monographs 58:251–269. Levin, S. A. 1992. The problem of pattern and scale in ecology. Ecology 73:1943–1967. Loiselle, B. A., and J. G. Blake. 1991. Temporal variation in birds and fruits along an elevational gradient in Costa Rica. Ecology 72:180–193. Mayaux, P., P. Holmgren, F. Achard, H. Eva, H.-J. Stibig, and A. Branthomme. 2005. Tropical forest cover change in the 1990s and options for future monitoring. Philosophical Transactions of the Royal Society B: Biological Sciences 360:373 LP-384. Mayor, S. J., D. C. Schneider, J. A. Schaefer, and S. P. Mahoney. 2009. Habitat selection at multiple scales. Ecoscience 16:238–247. Morris, R. J. 2010. Anthropogenic impacts on tropical forest biodiversity: a network structure and ecosystem functioning perspective. Philosophical transactions of the Royal Society of London. Series B, Biological sciences 365:3709–3718. Ostfeld, R. S., C. G. Jones, and J. O. Wolff. 1996. Of mice and mast. BioScience 46:323–330. Ostfeld, R. S., and F. Keesing. 2000. Pulsed resources and community dynamics of consumers in 87  terrestrial ecosystems. Trends in Ecology & Evolution 15:232–237. Osuri, A. M., J. Ratnam, V. Varma, P. Alvarez-loayza, J. H. Astaiza, M. Bradford, C. Fletcher, M. Ndoundou-hockemba, P. A. Jansen, D. Kenfack, A. R. Marshall, B. R. Ramesh, F. Rovero, and M. Sankaran. 2016. Contrasting effect of defaunation on aboveground carbon storage across the global tropics. Nature Communications 7:1–7. Paine, C. E. T., and H. Beck. 2007. Seed predation by neotropical rain forest mammals increases diversity in seedling recruitment. Ecology 88:3076–3087. Pan, Y., R. A. Birdsey, J. Fang, R. Houghton, P. E. Kauppi, W. A. Kurz, O. L. Phillips, A. Shvidenko, S. L. Lewis, J. G. Canadell, P. Ciais, R. B. Jackson, S. W. Pacala, A. D. McGuire, S. Piao, A. Rautiainen, S. Sitch, and D. Hayes. 2011. A large and persistent carbon sink in the world’s forests. Science 333:988–93. Peres, C. A. 2000. Effects of subsistence hunting on vertebrate community structure in Amazonian forests. Conservation Biology 14:240–253. Peres, C. A., and E. Palacios. 2007. Basin wide effects of game harvest on vertebrate population densities in Amazonian forests: implications for animal mediated seed dispersal. Biotropica 39:304–315. Phillips, Q., and K. Phillips. 2016. Phillips field guide to the mammals of Borneo and their ecology: Sabah, Sarawak, Brunei, and Kalimantan. Princeton University Press, Princeton, NJ. Pinard, M. A., and W. P. Cropper. 2000. Simulated effects of logging on carbon storage in dipterocarp forest. Journal of Applied Ecology 37:267–284. Poorter, L., and F. Bongers. 2006. Leaf traits are good predictors of plant performance across 53 rain forest species. Ecology 87:1733–1743. 88  Poulsen, J. R., and C. J. Clark. 2011. Decoupling the effects of logging and hunting on an Afrotropical animal community. Ecological Applications 21:1819–1836. R Core Team. 2016. R: A language and environment for statistical computing. R Foundation for Statistical Computing, Vienna. Ramage, B. S., D. Sheil, H. M. W. Salim, C. Fletcher, N. A. MUSTAFA, J. C. Luruthusamay, R. D. Harrison, E. Butod, A. D. Dzulkiply, and A. R. Kassim. 2013. Pseudoreplication in tropical forests and the resulting effects on biodiversity conservation. Conservation Biology 27:364–372. Rey, P. J. 1995. Spatio-temporal variation in fruit and frugivorous bird abundance in olive orchards. Ecology 76:1625–1635. Roldán, A. I., and J. a. Simonetti. 2001. Plant-mammal interactions in tropical Bolivian forests with different hunting pressures. Conservation Biology 15:617–623. Roscher, C., J. Schumacher, M. Gubsch, A. Lipowsky, A. Weigelt, N. Buchmann, B. Schmid, and E. D. Schulze. 2012. Using plant functional traits to explain diversity-productivity relationships. PLoS ONE 7:1–11. Rosin, C., and J. R. Poulsen. 2016. Hunting-induced defaunation drives increased seed predation and decreased seedling establishment of commercially important tree species in an Afrotropical forest. Forest Ecology and Management 382:206–213. Saracco, J. F., J. A. Collazo, and M. J. Groom. 2004. How do frugivores track resources? Insights from spatial analyses of bird foraging in a tropical forest. Oecologia 139:235–245. van Schaik, C. P., J. W. Terborgh, and S. J. Wright. 1993. The phenology of tropical forests: adaptive significance and consequences for primary consumers. Annual Review of ecology and Systematics 24:353–377. 89  Schleuning, M., N. Farwig, M. K. Peters, T. Bergsdorf, B. Bleher, R. Brandl, H. Dalitz, G. Fischer, W. Freund, M. W. Gikungu, M. Hagen, F. H. Garcia, G. H. Kagezi, M. Kaib, M. Kraemer, T. Lung, C. M. Naumann, G. Schaab, M. Templin, D. Uster, J. W. Wägele, and K. Böhning-Gaese. 2011. Forest fragmentation and selective logging have inconsistent effects on multiple animal-mediated ecosystem processes in a tropical forest. PLoS ONE 6. Schorger, A. 1955. The Passenger Pigeon. Its natural history and extinction. Blackburn Press, Caldwell, NJ [Reprint]. Senft, R. L., M. B. Coughenour, D. W. Bailey, L. R. Rittenhouse, O. E. Sala, and M. Swift, D. 1987. Large herbivore foraging and ecological hierarchies. BioScience 37:789–795. Shannon, C. E. 1948. A mathematical theory of communication. Bell System Technical Journal 27:379–423. Silvertown, J. . 1980. The evolutionary ecology of mast seeding in trees. Biological Journal of the Linnean society 14:235–250. Sodhi, N. S., L. P. Koh, B. W. Brook, and P. K. L. Ng. 2004. Southeast Asian biodiversity: An impending disaster. Trends in Ecology and Evolution 19:654–660. Sodhi, N. S., M. R. C. Posa, T. M. Lee, D. Bickford, L. P. Koh, and B. W. Brook. 2010. The state and conservation of Southeast Asian biodiversity. Biodiversity and Conservation 19:317–328. 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. Strauss, S. Y., and A. A. Agrawal. 1999. The ecology and evolution of plant tolerance to herbivory. Trends in Ecology & Evolution 14:179–185. 90  Tellería, J. L., and J. Pérez-tris. 2007. Habitat effects on resource tracking ability: do wintering Blackcaps Sylvia atricapilla track fruit availability? Ibis 149:18–25. Tellería, J. L., and J. Pérez-Tris. 2003. Seasonal distribution of a migratory bird: effects of local and regional resource tracking. Journal of Biogeography 30:1583–1591. Theimer, T. C., C. A. Gehring, P. T. Green, and J. H. Connell. 2011. Terrestrial vertebrates alter seedling composition and richness but not diversity in an Australian tropical rain forest. Ecology 92:1637–1647. Tobler, M. W., S. E. Carrillo-Percastegui, R. Leite Pitman, R. Mares, and G. Powell. 2008. Further notes on the analysis of mammal inventory data collected with camera traps. Animal Conservation 11:187–189. Tuck, S. L., M. J. O’Brien, C. D. Philipson, P. Saner, M. Tanadini, D. Dzulkifli, H. C. J. Godfray, E. Godoong, R. Nilus, and R. C. Ong. 2016. The value of biodiversity for the functioning of tropical forests: insurance effects during the first decade of the Sabah biodiversity experiment. Proceedings of the Royal Society B: Biological Sciences 283:20161451. Tylianakis, J. M., R. K. Didham, J. Bascompte, and D. A. Wardle. 2008. Global change and species interactions in terrestrial ecosystems. Ecology Letters 11:1351–1363. Valeix, M., A. J. Loveridge, S. Chamaillé-Jammes, Z. Davidson, F. Murindagomo, H. Fritz, and D. W. Macdonald. 2009. Behavioral adjustments of African herbivores to predation risk by lions: spatiotemporal variations influence habitat use. Ecology 90:23–30. Wang, B. C., V. L. Sork, M. T. Leong, and T. B. Smith. 2007. Hunting of mammals reduces seed removal and dispersal of the afrotropical tree Antrocaryon klaineanum (Anacardiaceae). Biotropica 39:340–347. 91  Welsh, A. H., R. B. Cunningham, C. F. Donnelly, and D. B. Lindenmayer. 1996. Modelling the abundance of rare species: statistical models for counts with extra zeros. Ecological Modelling 88:297–308. Whitmore, T. C. 1998. An introduction to tropical rain forests. Oxford University Press, New York, NY. Whitney, K. D., and T. B. Smith. 1998. Habitat use and resource tracking by African Ceratogymna hornbills : implications for seed dispersal and forest conservation:107–117. Wiens, J. A. 1989. Spatial scaling in ecology. Functional ecology 3:385–397. Wilkie, D., E. Shaw, F. Rotberg, G. Morelli, and P. Auzel. 2000. Roads, development, and conservation in the Congo basin. Conservation Biology 14:1614–1622. Wilmshurst, J. F., J. M. Fryxell, B. P. Farm, A. R. E. Sinclair, and C. P. Henschel. 1999. Spatial distribution of Serengeti wildebeest in relation to resources 1232:1223–1232. Wootton, J. T. 1993. Indirect effects and habitat use in an intertidal community: interaction chains and interaction modifications. The American Naturalist 141:71–89. Wright, S. J. 2003. The myriad consequences of hunting for vertebrates and plants in tropical forests. Perspectives in Plant Ecology, Evolution and Systematics 6:73–86. Wright, S. J., K. E. Stoner, N. Beckman, R. T. Corlett, R. Dirzo, H. C. Muller-Landau, G. Nuñez-Iturri, C. a. Peres, and B. C. Wang. 2007. The plight of large animals in tropical forests and the consequences for plant regeneration. Biotropica 39:289–291. Wright, S. J., H. Zeballos, I. Dominguez, M. M. Gallardo, M. C. Moreno, and R. Ibáñez. 2000. Poachers alter mammal abundance, seed dispersal, and seed predation in a neotropical forest. Conservation Biology 14:227–239. Yang, L., J. Bastow, K. Spence, and A. Wright. 2008. What can we learn from resource pulses. 92  Ecology 89:251–256. Yang, L., K. Edwards, J. Byrnes, J. Bastow, A. Wright, and K. Spence. 2010. A meta-analysis of resource pulse - consumer interactions. Ecological Monographs 80:125–151.    93  Appendices Appendix A  - Supplementary material for Chapter 3   Table A-1. Mammal herbivore species in study sites  Body mass (kg) and diet of mammal species excluded from our treatment plots in logged and unlogged forest in Sabah, Malaysian Borneo.       Latin name Common name Body mass (kg) Diet References      Elephas maximum borneensis Borneo pygmy elephant 2500 browse, fruit Phillips and Phillips (2016) Rusa unicolor Sambar deer 200 grass, browse, fruit Khan (1994), Phillips and Phillips (2016) Muntiacus atheroides Yellow muntjac 16-20 browse, fruit Phillips and Phillips (2016) Muntiacus muntjak Red muntjac 20-28 browse, fruit Phillips and Phillips (2016)  Tragulus napu Greater chevrotain 4.25 fruit Heydon (1994), Phillips and Phillips (2016)  Tragulus javanicus Lesser chevrotain 2.25 fruit Phillips and Phillips (2016)  Sus barbatus Bearded pig 75 - 200 seeds, fruits, invertebrates Phillips and Phillips (2016)  Hystrix brachyura Malayan porcupine 8 seeds, fruits, bark Jones et al. (2009) Trichys fasciculata Long tailed porcupine  1.75 seeds, fruits, bark Jones et al. (2009) Hystrix crassispinis Thick-spined porcupine 4.68 seeds, fruits, bark Jones et al. (2009)      94  Table A-2. Plant families in logged and unlogged forest  Taxonomic information for seedlings in exclosure and control plots logged (L) and unlogged (UL) forest in Malaysian Borneo. Family Genus Forest type Achariaceae Ryparosa L,UL  Hydnocarpus L Cornaceae Alangium L,UL Anacardiaceae Buchanania L  Polyalthia L,UL Annonaceae Neouvaria   Popowia L,UL  Xylopia L  onionthalamus UL Apocynaceae Kopsia L,UL Bombaceae Durio L Burseraceae Santiria L  Dacryodes L  Canarium L Calophyllaceae Calophyllum L Celastraceae Lophopetalum L Clusiaceae Garcinia UL Dillenaceae Dillenia L Dipterocarpaceae Dipterocarpus L  Dryobalanops L  Shorea L,UL  Parashorea L,UL Ebenaceae Diospyros L,UL Elaeocarpaceae Elaeocarpus L Euphorbiaceae Mallotus L,UL  Croton UL Escalloniaceae Polyosma UL  Neoscortechinia UL Fabaceae Dialium L,UL  Cynometra L  Saraca UL  Crudia UL Fagaceae Castanopsis L,UL  Lithocarpus L,UL Hypericaceae Cratoxylum L Leguminosae Fordia L  Koompassia L,UL Lauraceae Litsea L,UL 95  Family Genus Forest type Lauraceae Beilschmiedia L,UL  Dehaasia UL  Eusideroxylon UL Leeaceae Leea L Malvaceae Brownlowia L  Microcos L,UL  Pentace L,UL  Pterospermum L Melastomataceae Pternandra L Meliaceae Reinwardtiodendron UL  Dysoxylum L  Chisocheton L,UL  Aglaia L,UL  Lansium L Moraceae Ficus L Myristicaceae Knema L,UL Myrsinaceae Ardisia L Myrtaceae Syzygium L,UL Olacaceae Ochanostachys L  Chionanthus UL Pentaphylacaceae Adrinandra L Phyllanthaceae Glochidion L,UL  Antidesma L  Baccaurea L,UL  Aporusa L,UL  Cleistanthus L Polygalaceae Xanthophyllum L,UL Putranjivaceae Drypetes L,UL Rhamnaceae Ziziphus UL Rhizophoraceae Carallia L Rubiaceae Urophyllum UL  Praravinia L Rutaceae Glylcosmis L,UL  Diplospora L,UL  Clausena L,UL Sapindaceae Lepisanthes L  Paranephelium L  Dimocarpus L,UL  Pometia L  Nephelium UL Stemonuraceae Stemonurus L,UL Symplocaceae Symplocos L Urticaceae Pipturus L 96  Appendix A-3. Supplementary Methods & Results Trait information I obtained trait values for wood density, specific leaf area, and fruit size for each seedling genus represented in our treatment plots. These values were then used to calculate community weighted means of each trait in each treatment at each survey period. Trait values were obtained from the following sources:  Wood density  Chave, J., Coomes, D., Jansen, S., Lewis, S.L., Swenson, N.G. & Zanne, A.E. 2009 Towards a  worldwide wood economics spectrum Ecol. Lett. 12, 351-366.  Zanne A.E., Lopez-Gonzalez G., Coomes D.A., Ilic J., Jansen S, Lewis SL, Miller R.B.,  Swenson N.G., Wiemann M.C. & Chave J. 2009 Data from: Towards a worldwide wood economics spectrum. Dryad Digital Repository. http://dx.doi.org/10.5061/dryad.234   Specific Leaf Area  Burslem, D., Grubb, P. J. & Turner, I. M. 1996 Responses to simulated drought and elevated  nutrient supply among shade-tolerant tree seedlings of lowland tropical forest in Singapore. Biotropica 4, 636–648.   Kurokawa, H. & Nakashizuka, T. 2008 Leaf herbivory and decomposability in a Malaysian  tropical rain forest. Ecology 89, 2946–2656.  McConkey, K. R. & Chivers, D. J. 2004 Low mammal and hornbill abundance in the forests of  Barito Ulu, Central Kalimantan, Indonesia. Oryx 38.   Peñuelas, J., Sardans, J., Llusia, J., Silva, J., Owen, S. M., Bala-Ola, B., Linatoc, A. C., Dalimin, M. N. & Niinemets, Ü. 2013 Foliar chemistry and standing folivory of early and late-successional species in a Bornean rainforest. Plant Ecol. Divers. 6, 245–256.   Prentice, I. C., Meng, T., Wang, H., Harrison, S. P., Ni, J. & Wang, G. 2011 Evidence of a  universal scaling relationship for leaf CO2 drawdown along an aridity gradient. New Phytol. 190, 169–180.   Shipley, B. 2002 Trade-offs between net assimilation rate and specific leaf area in determining  relative growth rate: relationship with daily irradiance. Funct. Ecol. 16, 682–689.   Wright, I. J. et al. 2004 The worldwide leaf economics spectrum. Nature 428, 821–827.  97  Wright, J. Pasoh Leaf Data. Unpublished Data.  Fruit Length  Adema, F., Leenhouts, P. W. & Welzen, P. C. van 1994 Sapindaceae Flora Malesiana. Ser 1,  Spermatophyta 11/3, 419–768.   Birkinshaw, C. 2001 Fruit characteristics of species dispersed by the Black Lemur (Eulemur  macaco) in the Lokobe Forest, Madagascar. Biotropica 33, 478–486.  Brown, K.A., Flynn, D.F., Abram, N.K., Ingram, J.C., Johnson, S.E. and Wright, P.  2011. Assessing natural resource use by forest-reliant communities in Madagascar using functional diversity and functional redundancy metrics. PloS One 6, e24107.   Granados et al. Unpublished Data.  Jacobs, M. 1976 The study of lianas. Flora Malesiana Bull. 29, 2610–2618.  Leenhouts, P. W., Kalkman, C. & Lam, H. J. 1955 Burseraceae. Flora Malesiana - Ser. 1, Spermatophyta 5, 209–296.  Mabberley, D. J., Pannell, C. M. & Sing, A. M. 1995 Meliaceae. Flora Malesiana- Ser. I.  Spermatophyta 12, ii + 407 pp.  Nielsen, I. C. 1992 Mimosaceae (Leguminosae-Mimosoideae). Flora Malesiana - Ser. 1,  Spermatophyta 11, 1–226. Steenis,  van C. G. G. J. & van Steenis-Kruseman, M. J. 1950 Flora Malesiana Ser. I,  Spermatophyta Jakarta, Indonesia: Noordhoff-Kolff N.V.  van der Meijden, R. 1984 Polygalaceae. Flora Malesiana - Ser. 1, Spermatophyta 10, 455–539.   Analyses of seedling mortality and growth I used general linear mixed effects models (GLMM) to determine the impacts of exclusion treatment (a factor with 4 levels) and forest type (a factor with 2 levels; logged and unlogged) on each of our response variables: overall seedling mortality, average yearly growth (cm/year). All models included a random effect for experimental block nested in forest (a random term with six levels). Overall seedling mortality was modeled as a binary response with two levels (dead or alive). Aside from testing the effects of forest type and treatment, I also tested for a significant 98  forest × treatment interaction. Estimates are presented on a logit scale and errors are assumed to be binomially distributed.  I evaluated seedling height (cm) at the onset of our study as a function of forest type (logged and unlogged) to test for differences prior to treatment effects. I then used data from all sample periods to determine the average annual change in height for each seedling and modeled this growth response as a function of forest type, exclusion treatment, and a forest × treatment interaction.   Results for seedling growth & mortality  Seedling mortality was lower where all herbivores were excluded than in control plots (difference between treatments = -1.30; 95% CI: -2.09 to -0.50; S4). Fewer seedlings also died where only small herbivores were excluded than in control plots (difference between treatments = -1.22; 95% CI: -1.90 to -0.54). Within unlogged forest, mortality was higher where only small herbivores were excluded (parameter estimate for forest × treatment interaction = 0.94; 95% CI: 0.13 to 1.75). For elephant-excluded treatments, mortality was higher in logged than in unlogged forest (parameter estimate for forest × treatment interaction = -1.03 95% CI: 1.94 to -0.13) (figure S5.1). Shorter seedlings were more likely to die, suggesting they were more vulnerable to herbivory (parameter estimate = -3.01; 95% CI:  -3.59 to -2.07) whereas taller seedlings were more likely to die in control plots (parameter estimate for treatment × height interaction: 0.38; 95% CI: 0.0246 to 0.74). Mean seedling height did not differ between logged and unlogged forest in 2013 nor did mean height differ between areas in 2016. Seedlings with smaller changes in growth exhibited 99  greater mortality than seedlings with larger changes in growth (parameter estimate = -0.60; 95% CI: -0.75 to -0.44). Average annual changes in growth were higher where all herbivores were excluded than in control plots (difference between treatments = 0.85; 95% CI: 0.04 to 1.66; figure S5.2). Growth was significantly lower in control plots than in treatments excluding only elephants (difference between treatments = -1.19; 95% CI: -1.98 to -0.40). There was a significant forest × treatment interaction as seedlings grew more with the exclusion of all herbivores in logged forest than it did in unlogged forest (difference between forests = 1.44; 95% CI: 0.04 to 2.85).    100  A-4. References for Appendix A  Heydon, M. J. 1994 The ecology and management of rainforest ungulates in Sabah , Malaysia : implications of forest disturbance. (doi:10.13140/2.1.2587.4245) Khan, J. A. 1994 Food habits of ungulates in dry tropical forests of Gir Lion Sanctuary, Gujarat, India. Acta Theriol. (Warsz). 39.  Jones, K. E. et al. 2009 PanTHERIA: a species-level database of life history, ecology, and geography of extant and recently extinct mammals. Ecology 90, 2648. Phillips, Q. & Phillips, K. 2016 Phillips field guide to the mammals of Borneo and their ecology:  Sabah, Sarawak, Brunei, and Kalimantan. Princeton, NJ: Princeton University Press.     101   Figure A-1 Seedling mortality in logged and unlogged forest exclosure plots Seedling mortality across exclusion treatments in unlogged (black) and logged (grey) forest. Standard error bars are shown. Letters denote significant differences in mortality between exclosure treatments based on non-overlapping 95% confidence intervals.   102   Figure A-2 Seedling growth in logged and unlogged forest exclosure plots Seedling growth across exclusion treatments in unlogged (black) and logged (grey) forest. Standard error bars are shown. Letters denote significant differences in mortality between exclosure treatments based on non-overlapping 95% confidence intervals.   103  Appendix B  - Supplementary material for Chapter 4 Supplementary Tables  Table B-1 Plant families in fruit collections List of plant families represented in fruit collections from camera trap stations in 2014 and 2015 in unlogged (UL) and logged (L) forest in Sabah, Malaysian Borneo.         Year   Family    2014  2015 Achariaceae UL L  UL  Actinidiaceae    UL L Alangiaceae UL L    Anacardiaceae UL L  UL L Anisophylleaceae UL   UL  Annonaceae UL L  UL L Apocynaceae UL   UL  Asteraceae    UL L Barringtoniaceae UL     Burseraceae UL L  UL L Caesalpinaceae UL L  UL  Cardiopteridaceae UL     Clusiaceae UL L    Combretaceae UL L  UL L Cucurbitaceae UL     Connoraceae  L  UL L Cornaceae      Dilleniaceae UL   UL  Dipterocarpaceae UL L  UL L Ebenaceae UL L  UL  Elaeocarpaceae UL L  UL L Euphorbiaceae UL L  UL L Fabaceae  UL L  UL L Fagaceae  UL L  UL L Gentianaceae UL     Icadinaceae UL     Lamiaceae UL L  UL L 104          Year   Family    2014  2015 Lauraceae UL   UL L Lecythidaceae    UL  Leeaceae  UL L  UL  Loganiaceae UL L  UL L Loranthaceae    UL L Lythraceae UL L  UL L Magnoliaceae  L  UL  Malvaceae UL L  UL L Marathaceae    UL  Melastomataceae UL L  UL L Meliaceae UL L  UL L Melisticaceae UL     Moraceae UL L  UL L Myristicaceae UL   UL L Myrsinaceae    UL L Myrtaceae UL L  UL L Olacaceae UL   UL  Orchidaceae UL     Pandanaceae  L    Passifloraceae UL     Pentaphylacaceae    UL L Phyllanthaceae UL L  UL L Poaceae  UL     Polygalaceae    UL L Rhamnaceae UL   UL L Rosaceae   L  UL L Rubiaceae UL L  UL L Rutaceae  UL   UL L Salicaceae    UL L Sapindaceae UL L  UL L Sapotaceae UL   UL L Simaroubaceae    UL L Smilacaceae    UL L Sterculiaceae    UL L Tetramelaceae    UL L Uritcaceae    UL  Vitaceae  UL   UL L 105          Year   Family    2014  2015 Zingiberaceae       UL L          106  Table B-2 Model selection results for small-scale animal site use  List of models used in model selection to test for differences in animal site use as a function of fruit biomass (g/m2), forest type (logged, unlogged) and year (2013, 2014, 2015) at camera trap stations for vertebrate taxa in Sabah, Malaysian Borneo.  Four categories of fruit biomass were tested: fruit from all taxa (all), fruit from non-dipterocarp taxa (nd), dipterocarp fruits (ripe and unripe fruits pooled), and ripe dipterocarp fruits. Model selection results are based on zero-inflated Poisson regression models with camera trap station modeled as the random effect. The model of best fit was determined as having the lowest AICc value and highest AIC weight (w). ∆AIC values relative to the best fit model are also shown (∆AIC). Species Model df logLik AICc ∆AIC wi 	       Bearded pigs ripe × forest × year   10 -1337.78 2696 0 0.630 > 3 months ripe × year   6 -1342.84 2697.9 1.83 0.253  all × year   6 -1344.2 2700.6 4.55 0.065  dipt × forest × year   10 -1340.92 2702.3 6.28 0.027  all × forest × year   10 -1341.3 2703.1 7.04 0.019  dipt × year   6 -1346.5 2705.2 9.15 0.007  nd × year   6 -1354.05 2720.3 24.25 0  nd × forest × year   10 -1353.52 2727.5 31.48 0  forest × year   6 -1363.04 2738.3 42.23 0  ripe × forest  6 -1440.5 2893.2 197.15 0  ripe  4 -1442.97 2894 197.99 0  all 4 -1444.5 2897.1 201.05 0  dipt × forest 6 -1443.55 2899.3 203.25 0  nd × forest 4 -1445.83 2899.7 203.71 0  all × forest   6 -1444.02 2900.2 204.19 0  dipt 4 -1446.14 2900.4 204.33 0  nd × forest 6 -1445.61 2903.4 207.37 0  Intercept-only 2 -1572.32 3148.7 452.63 0        Bearded pigs ripe 4 -198.854 405.8 0 0.424 < 3 months dipt 4 -198.882 405.8 0.06 0.412  nd × forest 4 -200.805 409.7 3.9 0.06 107  Species Model df logLik AICc ∆AIC wi Bearded pigs dipt × year   6 -198.828 409.8 4.04 0.056 < 3 months all × year   6 -199.023 410.2 4.43 0.046  ripe × forest × year   10 -198.889 418.3 12.46 0.001  dipt × forest × year   10 -200.562 421.6 15.8 0  nd × year   6 -205.671 423.5 17.73 0  all × forest × year   10 -201.649 423.8 17.98 0  all 4 -208.66 425.4 19.61 0  forest × year   6 -206.872 425.9 20.13 0  ripe × forest  6 -207.778 427.7 21.94 0  dipt × forest 6 -207.793 427.8 21.97 0  ripe × year   6 -207.983 428.1 22.35 0  all × forest   6 -208.373 428.9 23.13 0  Intercept-only 2 -214.516 433.1 27.26 0  nd × forest × year   10 -213.684 447.8 42.05 0        Yellow dipt × forest × year   10 -571.778 1164 0 0.416 muntjac forest × year   6 -576.269 1164.7 0.69 0.295  ripe × forest × year   10 -573.502 1167.5 3.45 0.074  all × forest × year   10 -573.768 1168 3.98 0.057  dipt × forest 6 -578.384 1168.9 4.92 0.036  ripe × forest  6 -578.876 1169.9 5.9 0.022  nd × forest 6 -579.054 1170.3 6.26 0.018  all × forest   6 -579.067 1170.3 6.28 0.018  dipt × year   6 -579.23 1170.6 6.61 0.015  nd × forest × year   10 -575.513 1171.5 7.47 0.01  dipt 4 -581.736 1171.6 7.53 0.01  ripe 4 -581.738 1171.6 7.53 0.01  all 4 -581.912 1171.9 7.88 0.008  nd 4 -582.37 1172.8 8.8 0.005  ripe × year   6 -580.353 1172.9 8.86 0.005  all × year   6 -581.556 1175.3 11.26 0.001  nd × year   6 -582.166 1176.5 12.48 0.001  Intercept-only 2 -612.405 1228.8 64.81 0        Chevrotains forest × year   6 -944.216 1900.6 0 0.548  dipt × forest × year   10 -941.828 1904.1 3.52 0.094  ripe × forest × year   10 -941.969 1904.4 3.8 0.082  nd × forest × year   10 -942.987 1906.4 5.84 0.03  nd × year   6 -947.175 1906.5 5.92 0.028  all 4 -949.253 1906.6 5.98 0.028 108  Species Model df logLik AICc ∆AIC wi Chevrotains nd 4 -949.35 1906.8 6.17 0.025  ripe 4 -949.37 1906.8 6.21 0.025  dipt 4 -949.436 1907 6.35 0.023  ripe × year   6 -947.437 1907.1 6.44 0.022  all × year   6 -947.496 1907.2 6.56 0.021  all × forest × year   10 -943.487 1907.4 6.84 0.018  dipt × year   6 -947.657 1907.5 6.88 0.018  ripe × forest  6 -947.912 1908 7.39 0.014  dipt × forest 6 -948.019 1908.2 7.61 0.012  all × forest   6 -948.421 1909 8.41 0.008  nd × forest 6 -948.766 1909.7 9.1 0.006  Intercept-only 2 -1118.89 2241.8 341.19 0        Great Argus all × forest × year   10 -296.637 613.7 0 0.314 pheasant dipt × forest 6 -300.89 614 0.21 0.282  all × forest   6 -301.036 614.3 0.5 0.244  nd × forest × year   10 -298.437 617.3 3.6 0.052  ripe × forest  6 -302.717 617.6 3.87 0.045  nd × forest 6 -303.366 618.9 5.16 0.024  dipt × forest × year   10 -299.251 619 5.23 0.023  forest × year   6 -303.888 620 6.21 0.014  ripe × forest × year   10 -301.426 623.3 9.58 0.003  nd × year   6 -309.282 630.7 17 0  dipt × year   6 -310.59 633.4 19.61 0  nd 4 -312.707 633.5 19.75 0  all 4 -312.772 633.6 19.88 0  ripe × year   6 -311.093 634.4 20.62 0  dipt 4 -313.2 634.5 20.74 0  ripe 4 -313.201 634.5 20.74 0  all × year   6 -312.001 636.2 22.43 0  Intercept-only 2 -334.044 672.1 58.37 0        Crested fireback forest × year   6 -474.671 961.5 0 0.204 pheasant nd 4 -476.738 961.6 0.04 0.2  all 4 -477.037 962.2 0.64 0.148  dipt 4 -477.297 962.7 1.16 0.114  ripe 4 -477.336 962.8 1.24 0.11  nd × forest 6 -476.212 964.6 3.08 0.044  nd × year   6 -476.483 965.1 3.62 0.033  all × forest   6 -476.708 965.6 4.07 0.027 109  Species Model df logLik AICc ∆AIC wi Crested fireback all × year   6 -476.781 965.7 4.22 0.025 pheasant ripe × year   6 -477.039 966.3 4.74 0.019  dipt × year   6 -477.099 966.4 4.86 0.018  dipt × forest 6 -477.27 966.7 5.2 0.015  ripe × forest  6 -477.301 966.8 5.26 0.015  ripe × forest × year   10 -473.286 967 5.52 0.013  dipt × forest × year   10 -473.995 968.5 6.94 0.006  nd × forest × year   10 -474.052 968.6 7.06 0.006  all × forest × year   10 -474.455 969.4 7.86 0.004  Intercept-only 2 -491.104 986.2 24.71 0        Malayan ripe × forest × year   10 -303.797 628.1 0 0.423 porcupine dipt × forest × year   10 -304.967 630.4 2.34 0.131  ripe × year   6 -309.524 631.2 3.16 0.087  dipt × forest 6 -309.609 631.4 3.33 0.08  ripe × forest  6 -309.686 631.6 3.48 0.074  forest × year   6 -309.863 631.9 3.84 0.062  all × forest   6 -310.402 633 4.92 0.036  nd × forest 6 -310.553 633.3 5.22 0.031  dipt × year   6 -311.301 634.8 6.71 0.015  ripe 4 -313.617 635.3 7.25 0.011  all 4 -313.627 635.3 7.27 0.011  nd 4 -313.627 635.3 7.27 0.011  dipt 4 -313.691 635.5 7.4 0.01  all × forest × year   10 -307.882 636.2 8.17 0.007  all × year   6 -312.635 637.4 9.38 0.004  nd × year   6 -312.956 638.1 10.02 0.003  nd × forest × year   10 -308.911 638.3 10.23 0.003  Intercept-only 2 -331.798 667.6 39.55 0        Thick-spined forest × year   6 -191.462 395.1 0 0.809 porcupine all × forest × year   10 -190.184 400.8 5.74 0.046  ripe × forest × year   10 -190.337 401.1 6.04 0.039  nd × forest × year   10 -190.449 401.4 6.27 0.035  nd × forest 6 -195.378 402.9 7.83 0.016  ripe × forest  6 -195.722 403.6 8.52 0.011  all × forest   6 -195.756 403.7 8.59 0.011  dipt × forest 6 -195.79 403.8 8.66 0.011  all × year   6 -196.733 405.6 10.54 0.004  nd × year   6 -196.849 405.9 10.77 0.004 110  Species Model df logLik AICc ∆AIC wi Thick-spined nd 4 -199.172 406.4 11.33 0.003 porcupine all 4 -199.275 406.6 11.53 0.003  ripe 4 -199.394 406.9 11.77 0.002  ripe × year   6 -197.362 406.9 11.8 0.002  dipt 4 -199.422 406.9 11.83 0.002  dipt × year   6 -197.776 407.7 12.63 0.001  Intercept-only 2 -226.577 457.2 62.08 0  dipt × forest × year   10 -304.967 630.4 235.3 0        Murid rodent forest × year   6 -431.007 874.2 0 0.722 spp. nd × forest × year   10 -428.687 877.8 3.65 0.116  all × forest × year   10 -428.759 878 3.8 0.108  dipt × forest × year   10 -429.994 880.5 6.27 0.031  ripe × forest × year   10 -430.353 881.2 6.99 0.022  nd × year   6 -439.734 891.6 17.45 0  all × year   6 -440.71 893.6 19.41 0  ripe × year   6 -440.904 894 19.79 0  dipt × year   6 -440.972 894.1 19.93 0  nd × forest 6 -443.537 899.3 25.06 0  all × forest   6 -444.346 900.9 26.68 0  dipt × forest 6 -445.319 902.8 28.62 0  ripe × forest  6 -445.336 902.9 28.66 0  nd 4 -448.919 905.9 31.73 0  all 4 -449.247 906.6 32.39 0  dipt 4 -449.291 906.7 32.47 0  ripe 4 -449.31 906.7 32.51 0  Intercept-only  2 997.5 123.33 0  111  Supplementary Figures  Figure B-1 Plant families in fruit biomass collections Proportion of total fruit biomass (g/m2) from the 15 most common plant families collected from ground surveys at camera trap stations in A) 2014 and B) 2015 in unlogged (black) and logged (grey) forest in Sabah, Malaysian Borneo.  112     Figure B-2 Yellow muntjac site use relative to dipterocarp fruit biomass Partial residual plots showing site use of yellow muntjac as a function of dipterocarp fruit biomass (ripe and unripe fruits pooled, standardized dry weight, [g/m2]) in unlogged and logged forest in A) 2014 (major mast) and b) 2015 (minor mast). Site use is modeled as the number of independent detections of animals at camera trap stations. Detections were deemed to be independent if they were non-consecutive (i.e., another species was photographed in between) or if they occurred at least one hour apart. Fruit biomass data are from fallen fruit collected during repeated visits to camera trap sites within each year. Confidence bands shown around the predicted fit line from the models of best fit selected by AICc comparison.    113     Figure B-3 Malayan porcupine site use relative to ripe dipterocarp fruit biomass Partial residual plots showing site use of Malayan porcupine as a function of ripe dipterocarp fruit biomass (standardized dry weight, [g/m2]) in unlogged and logged forest in A) 2014 (major mast) and b) 2015 (minor mast). Site use is modeled as the number of independent detections of animals at camera trap stations. Detections were deemed to be independent if they were non-consecutive (i.e., another species was photographed in between) or if they occurred at least one hour apart. Fruit biomass data are from fallen fruit collected during repeated visits to camera trap sites within each year. Confidence bands shown around the predicted fit line from the models of best fit selected by AICc comparison.    114     Figure B-4 Argus pheasant site use relative to dipterocarp fruit biomass Partial residual plots showing site use of Argus pheasant as a function of dipterocarp fruit biomass (ripe and unripe pooled, standardized dry weight, [g/m2]) in unlogged and logged forest in A) 2014 (major mast) and b) 2015 (minor mast). Site use is modeled as the number of independent detections of animals at camera trap stations. Detections were deemed to be independent if they were non-consecutive (i.e., another species was photographed in between) or if they occurred at least one hour apart. Fruit biomass data are from fallen fruit collected during repeated visits to camera trap sites within each year. Confidence bands shown around the predicted fit line from the models of best fit selected by AICc comparison.  115     Figure B-5 Fireback pheasant site use relative to total fruit biomass Partial residual plots showing site use of Crested fireback pheasant as a function of total fruit biomass (standardized dry weight, [g/m2]) in unlogged and logged forest in A) 2014 (major mast) and b) 2015 (minor mast). Site use is modeled as the number of independent detections of animals at camera trap stations. Detections were deemed to be independent if they were non-consecutive (i.e., another species was photographed in between) or if they occurred at least one hour apart. Fruit biomass data are from fallen fruit collected during repeated visits to camera trap sites within each year. Confidence bands shown around the predicted fit line from the models of best fit selected by AICc comparison.    

Cite

Citation Scheme:

        

Citations by CSL (citeproc-js)

Usage Statistics

Share

Embed

Customize your widget with the following options, then copy and paste the code below into the HTML of your page to embed this item in your website.
                        
                            <div id="ubcOpenCollectionsWidgetDisplay">
                            <script id="ubcOpenCollectionsWidget"
                            src="{[{embed.src}]}"
                            data-item="{[{embed.item}]}"
                            data-collection="{[{embed.collection}]}"
                            data-metadata="{[{embed.showMetadata}]}"
                            data-width="{[{embed.width}]}"
                            async >
                            </script>
                            </div>
                        
                    
IIIF logo Our image viewer uses the IIIF 2.0 standard. To load this item in other compatible viewers, use this url:
https://iiif.library.ubc.ca/presentation/dsp.24.1-0355399/manifest

Comment

Related Items